Rethinking Power Sector Reform in the Developing World Vivien Foster and Anshul Rana Rethinking Power Sector Reform in the Developing World SUSTAINABLE INFRASTRUCTURE SERIES Rethinking Power Sector Reform in the Developing World Vivien Foster and Anshul Rana © 2020 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 22 21 20 19 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. ­ Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http:// creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Foster, Vivien, and Anshul Rana. 2020. Rethinking Power Sector Reform in the Developing World. Sustainable Infrastructure Series. Washington, DC: World Bank. doi:10.1596/978​ -1-4648-1442-6. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISBN (paper): 978-1-4648-1442-6 ISBN (electronic): 978-1-4648-1443-3 DOI: 10/1596/978-1-4648-1442-6 Cover design: Bill Pragluski, Critical Stages, LLC. Library of Congress Control Number: 2019915533. Contents Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi Background Papers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxvii Key Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxix Overview: Key Findings and Policy Implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Policy implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Part I:  Setting the Stage Chapter 1  What Do We Mean by Power Sector Reform?. . . . . . . . . . . . . . . . . . . . . . . . . . 41 Motivation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 A brief history of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 The 1990s power sector reform model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 A theory of change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 An uncertain future. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Chapter 2  How Far Did Power Sector Reform Spread in the Developing World? . . . . 57 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Note . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 v vi Contents Chapter 3  How Did Political Economy Affect the Uptake of Power Sector Reform?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Looking ahead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Annex 3A. Global Power Sector Reform Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Annex 3B. Chi-squared contingency tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Annex 3C. World Bank support for Power Sector Reform Observatory countries and states. . . . 104 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Part II:  Building Blocks of Reform Chapter 4  What Has Been Done to Restructure Utilities and Improve Governance?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Looking ahead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 Annex 4A. Utility Restructuring Index, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 Annex 4B. Planning and Procurement Index, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 Annex 4C. Utility Governance Index, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Annex 4D. Utility Classification, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 Annex 4E. Utility Restructuring Index, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 Chapter 5  What Has the Private Sector Contributed? . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 Annex 5­ A. Private Sector Participation Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Annex 5­ B. Private Sector Participation Index, 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Chapter 6  Did Countries Establish Meaningful Power Sector Regulation?. . . . . . . . . 171 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 Looking ahead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 Annex 6A. Formal (de jure) scores on the Regulatory Performance Index. . . . . . . . . . . . . . . 203 Annex 6B. Perceived scores on the Regulatory Performance Index. . . . . . . . . . . . . . . . . . . . . 204 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Chapter 7  What Progress Has Been Made with Wholesale Power Markets?. . . . . . . . 207 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 Contents vii Looking ahead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Part III:  Gauging Impact Chapter 8  Did Power Sector Reforms Improve Efficiency and Cost Recovery?. . . . . 241 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 Annex 8A. Major studies of cost recovery and financial viability in the power sector in developing countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Annex 8B. Coverage of quantitative cost recovery analysis undertaken for this chapter. . . . . . 269 Annex 8C. Indicators of cost recovery and financial viability of power sectors and utilities in 17 case studies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 Annex 8D. Indicators of efficiency of utilities in 17 case studies . . . . . . . . . . . . . . . . . . . . . . . 273 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 Chapter 9  Did Power Sector Reform Deliver Better Sector Outcomes?. . . . . . . . . . . . 277 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 Key findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 Annex 9A. Econometric analysis of power sector reform impacts based on large sample (88 countries). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 Annex 9B. Results of econometric analysis of power sector reform impacts based on large sample (88 countries). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 Annex 9C. Cross-sectional regression analysis on the impact of power sector reform based on small sample (17 economies). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 Boxes 1.1 The World Bank and three decades of power sector reforms . . . . . . . . . . . . . . . . . . . . . . 42 1.2 The Rethinking Power Sector Reform Observatory: An introduction. . . . . . . . . . . . . . . . 52 2.1 Defining the Power Sector Reform Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.1 Methodology for political economy stakeholder analysis. . . . . . . . . . . . . . . . . . . . . . . . . 73 3.2 Importance of public communication strategy to support reforms. . . . . . . . . . . . . . . . . . 96 4.1 Selected power sector structures around the world. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 6.1 Introducing the Regulatory Performance Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 7.1 How big must a power system be to support a wholesale power market?. . . . . . . . . . . 213 8.1 Methodology used in cost recovery analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 9.1 The methodological challenges of inferring the impact of power sector reforms. . . . . . 280 Figures O.1 The 1990s model was based on an underlying theory of change. . . . . . . . . . . . . . . . . . . . 2 O.2 The trajectory of power sector reform followed different paths across countries. . . . . . . . 6 O.3 For some countries, the gap between reform announcement and implementation has been considerable. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 viii Contents O.4 The private sector’s contribution to new generation capacity was steady across income groups but heavily skewed by generation technology. . . . . . . . . . . . . . . . . . . . . . 8 O.5 The bulk of private investment in generation came from foreign sponsors. . . . . . . . . . . . 8 O.6 Direct negotiation of IPPs remains widespread in South Asia and Sub-Saharan Africa . . . . 9 O.7 Electricity spot prices have shown wide variation across developing country markets. . . . 10 O.8 Certain aspects of corporate governance are strongly associated with improved efficiency performance for distribution utilities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.9 Private sector participation in distribution peaked in the late 1990s before declining. . . 14 O.10 Private sector participation is associated with much higher levels of cost recovery, while performance on efficiency is within the range observed for public utilities. . . . . . 15 O.11 Significant divergence exists between regulation on paper and regulation in practice . . 17 O.12 Regulatory tariff recommendations are not always respected in practice. . . . . . . . . . . . . 18 O.13 More countries made progress on efficiency than on cost recovery, 1990–2015. . . . . . . 19 O.14 Progress on twenty-first-century policy objectives for electrification and decarbonization, 1990–2015, countries ranked in descending order of reform effort. . . 35 1.1 1990s model sector reforms: Inferred simple theory of change . . . . . . . . . . . . . . . . . . . . 50 2.1 OECD countries score systematically higher on the Power Sector Reform Index . . . . . . 59 2.2 OECD countries are more likely to have adopted restructuring and liberalization reforms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.3 The uptake of power sector reform measures has been slowing since 2005. . . . . . . . . . . 60 2.4 A slowdown is evident across a wide range of different reform measures. . . . . . . . . . . . 61 2.5 Latin America and the Caribbean led the way on power sector reform with many other regions lagging behind. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 2.6 Reform uptake has been strongly associated with income group and scale of system. . . . . 64 2.7 Some reform measures proved a lot more popular than others. . . . . . . . . . . . . . . . . . . . 65 2.8 Some types of reforms diffused more rapidly than others. . . . . . . . . . . . . . . . . . . . . . . . . 65 2.9 Barely a dozen developing countries managed to implement the full 1990s reform package. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 2.10 Countries with high PSRI scores undertook reforms at differing speeds. . . . . . . . . . . . . .67 2.11 Power sector reform was conceived as a coherent package of measures implemented according to a logical sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 2.12 A significant ­percentage of countries adopted reforms in unorthodox ways. . . . . . . . . . 69 B3.1.1 No strong relationship between income group and quality of governance . . . . . . . . . . . 73 3.1 Power sector reform process: Announcement, implementation, delivery, and sustainability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.2 The contrasting reform trajectories of a bold reformer, Ukraine, and an incremental one, Vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3.3 Contrast between reforms announced and reforms delivered in observatory countries and states. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.4 Some countries experienced reversals of reform once implemented. . . . . . . . . . . . . . . . 76 3.5 Nested model of political influences in power sector reform . . . . . . . . . . . . . . . . . . . . . . 78 3.6 Dependency on foreign aid was substantial at the time of power sector reform . . . . . . . 82 B4.1.1 Vertically integrated utility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 B4.1.2 Single-buyer model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 B4.1.3 Wholesale-buyer model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 4.1 Plans must be mandatory in implementation and combined with transparent and competitive procurement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.2 Sector planning remains necessary and calls for adequate institutional capacity and sound processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.3 Overview of utility governance performance indicators. . . . . . . . . . . . . . . . . . . . . . . . . 120 Contents ix 4.4 Private utilities implement more governance best practices, but government-owned utilities improve governance if there is some private competition. . . . . . . . . . . . . . . . . 121 4.5 Private utilities far outpace their public counterparts when it comes to making decisions independently. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.6 Boards in state-owned utilities are less accountable than their private counterparts, which must answer to various shareholders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 4.7 Public utilities have little independence when it comes to raising capital and tend to follow national rather than international accounting standards . . . . . . . . . . . . 123 4.8 Public utilities have little freedom in making staffing decisions and have less transparency in hiring as compared to their private counterparts . . . . . . . . . . . . . . . . . 124 4.9 Both public and private utilities have adopted the latest information and technology solutions and are mostly at par. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 4.10 Close to 60 percent of developing countries still operate with a vertically integrated national monopoly utility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 4.11 Most countries deploy some version of the single-buyer model in their power sectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 4.12 Horizontal unbundling of the distribution sector in Peru created a couple of large profitable metropolitan utilities and a number of small regional utilities with limited scope for profits, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5.1 Private investment in electricity came predominantly from foreign sources and mostly in the generation sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 5.2 Private investment in independent power producers has been substantial though subject to fluctuations and concentrated in a few countries, 1990–2017 . . . . . 149 5.3 Most countries still depend on a combination of public and private investment for the development of new power generation capacity . . . . . . . . . . . . . . . . . . . . . . . . 150 5.4 Private investment shares reveal different drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 5.5 Independent power producer investment is moving toward cleaner sources of energy, although capacity expansion lags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.6 Competitive procurement of independent power producers varies widely between regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 5.7 In the early years, most private sector participation in distribution was through concessions, primarily in Latin America and the Caribbean. . . . . . . . . . . . . . . . . . . . . . 157 5.8 The rise and fall of private sector participation in electricity distribution, 1990–2017. . . . 159 5.9 Private sector participation in distribution suffers from premature contract cancellation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 6.1 Overview of Regulatory Performance Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 6.2 Latin American countries were early adopters of regulatory reform, but countries in Africa and Asia caught up fast. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 6.3 Tariff and quality regulation are the core responsibilities of regulatory institutions. . . . 176 6.4 Countries with higher levels of private sector participation in distribution have stronger perceived regulatory performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 6.5 Regulatory frameworks are not always perceived to function as written on paper. . . . 179 6.6 Gaps between paper and practice are particularly large on some aspects. . . . . . . . . . . . 180 6.7 Countries with higher private sector participation have smaller gaps between de jure and perceived regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 6.8 Quality of utility governance seems to be related to positive regulatory performance. . . . 181 6.9 Achievement of autonomous regulators remains a significant challenge even as regulators are more accountable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 6.10 Certain accountability measures have been universally adopted, whereas others are practiced in only about half of the countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 x Contents 6.11 Regulator’s autonomy to make legally binding decisions on key issues is perceived to be significantly lower in practice than it looks de jure. . . . . . . . . . . . . . . . 184 6.12 Most regulatory regimes are closer to rate-of-return regulation, with some incentive-based elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186 6.13 Two-thirds of the observatory countries practice automatic indexation of tariffs . . . . . 189 6.14 Substantial divergence exists between tariff regulation as it appears on paper and as it is actually practiced . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 6.15 Tariff recommendations made by regulatory entities are not necessarily respected or applied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 6.16 In India, authorized tariff increases fall well short of those approved by state regulators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 6.17 Tanzania and Uganda represent the extremes of the range of experiences of African countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 6.18 Quality of service regulation has some of the largest gaps between rules on paper and in practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 6.19 Quality-of-service regulations are not widely implemented, often for lack of technical capability within utilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 6.20 Most countries with well-developed formal regulations for market entry lag far behind in terms of practice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 6.21 Serious deficiencies relate to the revocation of licenses and the award of IPPs . . . . . . . 198 6.22 RIIO framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 7.1 Only one in five developing countries has established a wholesale market. . . . . . . . . . 209 7.2 Addressing market power is a critical component of establishing efficient power markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 7.3 Significant interannual price variations across power markets in the observatory countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 7.4 Power markets have helped countries attract adequate investment into generation . . . . 227 7.5 Inflexible contracts of long duration can limit a market’s ability to adapt to changing conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 8.1 Although most achieved operating cost recovery, few attained full cost recovery. . . . . 246 8.2 About half of cases cover cash needs for limited capital costs. . . . . . . . . . . . . . . . . . . . . 247 8.3 Below-cost tariffs remain an issue whether underlying costs are high or low. . . . . . . . 248 8.4 Inefficiencies can account for as much as 10 to 20 percent of utility revenues . . . . . . . 249 8.5 Utilities often borrow at rates well below commercial benchmarks. . . . . . . . . . . . . . . . 254 8.6 International financial institutions are typically the cheapest source of borrowing. . . . 255 8.7 Investment levels do not bear much relationship to cost recovery. . . . . . . . . . . . . . . . . 256 8.8 Underpricing is the largest driver of quasi-fiscal deficits in the power sector. . . . . . . . . 257 8.9 Many countries seem able to reconcile cost recovery and affordability . . . . . . . . . . . . . 258 8.10 Cross-subsidies are pronounced in some countries and states . . . . . . . . . . . . . . . . . . . . 259 8.11 Higher levels of cross-subsidy are associated with lower levels of cost recovery . . . . . . 260 8.12 Only a handful of countries and states had achieved full cost recovery prior to reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 8.13 Prereform cost recovery was not necessarily sustained over time. . . . . . . . . . . . . . . . . . 262 8.14 Postreform cost recovery improved in about half of cases and deteriorated in half. . . . 263 8.15 Nominal tariff increases were largely eroded by inflationary pressures . . . . . . . . . . . . . 264 8.16 In most cases, system losses declined substantially over time. . . . . . . . . . . . . . . . . . . . . 265 8.17 Improvements in system losses were often outweighed by other cost movements. . . . 266 9.1 The theory of change underpinning the 1990s power sector reform model. . . . . . . . . 278 9.2 Utilities with stronger financial discipline show better performance on operating cost recovery and distribution efficiency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 Contents xi 9.3 Certain financial discipline practices are much more widely practiced among utilities that perform well in terms of operating cost recovery and distribution efficiency. . . . . . . 286 9.4 Utilities with better human resource management show better performance on operating cost recovery and full capital cost recovery . . . . . . . . . . . . . . . . . . . . . . . . 287 9.5 Certain financial human resource practices are much more widely practiced among utilities that perform well in terms of operating cost recovery and full capital cost recovery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 9.6 Operating cost recovery bears little relation to the quality of tariff regulations on paper, but it is closely associated with the quality of tariff regulations in practice . . . . 288 9.7 Distribution efficiency follows the practice of quality regulation more than the practice of tariff regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 9.8 Private sector participation is strongly associated with full capital cost recovery. . . . . . 291 9.9 Only a few publicly owned utilities are allowed to charge cost recovery tariffs. . . . . . . 292 9.10 A substantial minority of publicly owned utilities performs as efficiently as private ones. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 9.11 Among the least efficient utilities are those that experienced privatization reversals. . . . 294 9.12 Efficient public and private utilities score relatively higher on management and governance good practices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 9.13 Performance on various measures of security of supply with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 9.14 Normalized capacity is strongly associated with income per capita. . . . . . . . . . . . . . . . . 298 9.15 Power sector diversification is significantly related to various reform steps. . . . . . . . . . 299 9.16 Reliability of supply is associated with greater distribution efficiency. . . . . . . . . . . . . . . 299 9.17 Performance on social inclusion, countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan). . . . . . . . . . . . 301 9.18 The strongest driver of electrification has been GDP per capita . . . . . . . . . . . . . . . . . . . 302 9.19 Performance on environmental sustainability, with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304 9.20 Carbon intensity shows strong inverse correlation with operating cost recovery. . . . . . 305 Maps O.1 Power sector reform spread unevenly across the developing world. . . . . . . . . . . . . . . . . . 4 2.1 Outlier countries on power sector reform exist in every region. . . . . . . . . . . . . . . . . . . . 67 Tables O.1 Overview of preconditions among groups of countries at the time of reform . . . . . . . . . 20 O.2 Comparison of country performance according to reform strategy . . . . . . . . . . . . . . . . . 22 O.3 Customizing power sector reforms to country context. . . . . . . . . . . . . . . . . . . . . . . . . . . 25 B1.2.1 Power Sector Reform Observatory countries and selected characteristics . . . . . . . . . . . . 52 B2.1.1 Power Sector Reform Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 3.1 The pace of power sector reform in the observatory countries and states . . . . . . . . . . . . 77 3.2 Crisis at the time of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 3.3 Crisis severity and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3.4 Donor influence on patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 3.5 Observatory countries and states and their ideological orientation. . . . . . . . . . . . . . . . . .85 3.6 Ideology and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 3.7 Political systems of observatory countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 3.8 Type of political system and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . 88 xii Contents 3.9 Overview of reform champions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.10 Reform champions and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . 92 3.11 Stakeholder friction in power sector reforms: An overview. . . . . . . . . . . . . . . . . . . . . . . 94 3.12 Stakeholder alignment and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . 97 3.13 Legislation and power sector reform: An overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 3.14 Legislation and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 3.15 Political economy analysis: How key stakeholders respond to technological disruption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .100 3.16 Political drivers of each stage of the reform process: An overview. . . . . . . . . . . . . . . . . 101 3A.1 Power Sector Reform Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 3B.1 Crisis and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 3B.2 Donor impact on patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 3B.3 Ideology and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 3B.4 Political system and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 3B.5 Reform champions and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . . . . 103 3B.6 Stakeholder alignment and patterns of power sector reform . . . . . . . . . . . . . . . . . . . . . 104 3B.7 Legislation and patterns of power sector reform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.1 Unbundling is a means to remove conflict of interest and encourage competition. . . . 116 4.2 Overview of restructuring and competition reforms in 17 jurisdictions, 2015. . . . . . . . 127 5.1 Forms of private sector participation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 5.2 Preferred private sector participation modality varies across regions/electricity supply chain (global) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 5.3 Private sector engagement across the 15 observatory countries. . . . . . . . . . . . . . . . . . . 147 5.4 Transmission: A few observatory countries have embraced private participation . . . . . 156 5.5 Management contracts for power utilities in Africa: A difficult history . . . . . . . . . . . . . 164 6.1 Design characteristics of tariff regulatory regimes: An overview. . . . . . . . . . . . . . . . . . . 188 B7.1.1 Classification of Observatory countries, by size threshold . . . . . . . . . . . . . . . . . . . . . . . 213 7.1 Immediate or transitional implementation of a power market is possible; each path has benefits and challenges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 7.2 A country-specific process for monitoring and overseeing the electricity market is required from the start of reforms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 7.3 Overview of power market governance across developing countries. . . . . . . . . . . . . . . 220 7.4 Comparison of short-term power market design across developing countries. . . . . . . . 222 7.5 Comparison of mechanisms for generation investment across developing countries. . . . 225 8.1 The cost recovery ladder from financial, fiscal, and economic perspectives. . . . . . . . . . 242 8.2 Full cost recovery is a moving target because of exogenous cost shocks . . . . . . . . . . . . 251 8.3 Utilities absorb financial shortfalls in various ways. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252 8C.1 Full cost recovery for power sectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 8C.2 Full financial cost recovery for power sectors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 8C.3 Financial operating cost recovery for power sectors, based on cash collected . . . . . . . . 271 8C.4 Full cost recovery for power utilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272 8D.1 Revenue lost due to undercollection for power utilities. . . . . . . . . . . . . . . . . . . . . . . . . 273 8D.2 Revenues lost due to excessive system losses for power utilities . . . . . . . . . . . . . . . . . . 274 9.1 Mapping of data availability across small and large country samples. . . . . . . . . . . . . . . 281 9.2 Framework for evaluating impact on final sector performance outcomes. . . . . . . . . . . 282 9.3 Impact of reform measures on intermediate outcomes for the large sample. . . . . . . . . 283 9.4 Impact of utility governance and intermediate outcomes for the 15-country sample. . 284 9.5 Impact of regulatory reforms on intermediate outcomes for small sample. . . . . . . . . . . 288 9.6 Impact of private sector participation on intermediate outcomes for small sample. . . . 291 Contents xiii 9.7 Impact of reform measures on final outcomes for large sample. . . . . . . . . . . . . . . . . . . 296 9.8 Qualitative analysis of underlying causality for security of supply trends . . . . . . . . . . . 300 9.9 Qualitative analysis of underlying causality for electrification trends. . . . . . . . . . . . . . . 303 9.10 Qualitative analysis of underlying causality for decarbonization trends. . . . . . . . . . . . . 306 9.11 Overview of cross-sectional performance on final outcomes as of 2015 with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 9.12 Overview of performance on differences in final outcomes from 1990 to 2015 with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 9B.1 Impact of reforms on system losses (intermediate outcome), 1995–2015 . . . . . . . . . . . 311 9B.2 Impact of reforms on generation efficiency (intermediate outcome), 1995–2015. . . . . 312 9B.3 Impact of reforms on modern renewable share in TFEC (final outcome), 1995–2015. . . 313 9B.4 Impact of reforms on carbon intensity (final outcome), 1995–2015. . . . . . . . . . . . . . . 314 9B.5 Impact of reforms on normalized capacity (final outcome), 1995–2015 . . . . . . . . . . . . 315 9B.6 Impact of reforms on access (final outcome), 1995–2015. . . . . . . . . . . . . . . . . . . . . . . . 316 9B.7 Correlation between reform steps and final outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . 317 9C.1 Correlation between reform steps and outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 9C.2 Regression analysis: reforms and intermediate outcomes. . . . . . . . . . . . . . . . . . . . . . . . 319 9C.3 Regression analysis: utility governance reforms and intermediate outcomes. . . . . . . . . 319 9C.4 Regression analysis: regulatory reforms and intermediate outcomes. . . . . . . . . . . . . . . 320 9C.5 Regression analysis: reform steps and final outcomes. . . . . . . . . . . . . . . . . . . . . . . . . . . 321 9C.6 Regression analysis: regulatory reforms and final outcomes. . . . . . . . . . . . . . . . . . . . . . 322 9C.7 Regression analysis: utility governance reforms and final outcomes . . . . . . . . . . . . . . . 323 Foreword Thirty years ago, a new paradigm emerged to Rethinking Power Sector Reform in the fundamentally alter power sector organization. Developing World comes at a crucial time. The It aimed to improve the financial and oper- world is changing—and so must the power ational performance of utilities, ensure reliable sector. The principles that guided policy mak- power supply, and attract private sector partici- ers and stakeholders in the 1990s remain pation and fair market forces while setting up strong today. Financial sustainability and the public sector to take on a regulatory role. good institutional governance in the power Yet, after almost three decades, only about a sector are still just as critical, even as the dozen developing countries have been able to scope of private sector participation is increas- fully implement the 1990s model. For many ing and technological disruptions and the countries, the model simply did not fit the eco- benefits of competition energize the sector. nomic preconditions of their power sectors; for It is o nly natur al that the r efo r m many others, the approach encountered politi- approaches will need to be updated to sup- cal challenges in implementation. Many of port these changes. those who have adopted the reforms have done This report offers a fresh frame of reference so selectively, leading to a situation in which shaped by context, driven by outcomes, and elements of market orientation coexist with a informed by alternatives. It has three clear strong state presence—something the designers messages for policy makers and industry practi- of the 1990s model did not anticipate. tioners. First, reform approaches must be Moreover, since the turn of the twenty-first shaped by the political and economic contexts century, the power sector has been overtaken of individual countries. Second, reform by important policy shifts and momentous approaches should be tailored to achieve technological changes. In recent years, the desired policy outcomes. Finally, multiple insti- world has embraced the Sustainable tutional pathways to achieve the desired Development Goal on Energy (SDG7), which ­ outcomes must be possible. There is no one- aims to achieve universal access to sustainable, size-fits-all framework, and the particular affordable, and modern energy by 2030. We needs and challenges of low-income and fragile are also witnessing a swift global transition to environments deserve special consideration. low-carbon and renewable energy sources in Our hope is that this report can refresh the line with the Paris Accord’s commitment to thinking on power sector reform in the devel- fight climate change. Technological disruption oping world; help deliver electricity access to is ushering new, decentralized actors into the those who need it most; and ultimately result sector and reshaping business models. in a clean, green, and financially sustainable However, the various reform approaches power sector. based on the 1990s model alone will not be sufficient to deliver on global energy objec- tives. We also need complementary, targeted Riccardo Puliti policies to reach the 840 million people who Global Director, Energy and live without access to electricity today and to Extractive Industries and rapidly increase the share of clean energy in Regional Director, Infrastructure (Africa) the global energy mix. The World Bank xv Acknowledgments This report would not have been possible • For the regulation theme, Katharina without generous funding provided by the Gassner and Joseph Kapika led the team, Energy Sector Management Assistance which comprised Martin Rodriguez Pardina Program (ESMAP) and the Public-Private and Julieta Schiro of Macroconsulting Infrastructure Advisory Facility (PPIAF). Argentina, as well as Kagaba Paul Mukibi. Valuable guidance was received from the • For the power market theme, original report’s peer reviewers: Pierre Audinet, Clive research was conducted by a team led by Harris, Dejan Ostojic, Sheoli Pargal, Mike Debabrata Chattopadhyay and comprising Toman, and Maria Vagliasindi (World Bank) Hugh Rudnick and Constantin Velasquez and Tooraj Jamasb (Durham University). The from the University of Chile, as well as report also benefited from the comments and Tatyana Kramskaya and Martin Schroder. suggestions of Sudeshna Ghosh Banerjee • For the utility restructuring and gover- (World Bank) throughout the process. nance theme, the team was led by Vivien The wide scope of the report meant that a Foster and comprised Joeri de Wit, Victor very large team of sector experts was involved Loksha, and Anshul Rana, with advi- in preparing each chapter, including many sor y suppor t f rom Ped ro A ntma n n, World Bank specialists complemented by Elvira Morella, Mariano Salto, and Pedro external consultants.1 These were based on the Sanchez. main themes of the project, namely, political • For the cost recover y theme, or ig i- economy of power sector reform, economic nal research was conducted by a team regulation of the power sector, wholesale led by Ani Balabanyan and comprising power markets, utility reform, cost recovery, Joern Huenteler, Arthur Kochnakyan, and disruptive technology. Tu Chi Nguyen, Arun Singh, and Denzel Hankinson and Nicole Rosenthal of DH • For the political economy theme, original Infrastructure. The analysis is based on a research was done by a team led by Ashish methodologically consistent set of finan- Khanna and comprising Anton Eberhard, cial models prepared for 25 utilities across Catrina Godinho, Alan David Lee, Brian 14 countries and 3 Indian states. The finan- Levy, Zainab Usman, and Jonathan Walters. cial analysis was led by Arthur Kochnakyan,   1 All names listed are World Bank staff unless otherwise stated. xvii xviii Acknowledgments supported by a team of independent consul- • For India, Deloitte India was responsible for tants, including Emiliano Lafalla, Adrian the data collection and conducting stake- Ratner, Vazgen Sargsyan, and Martin holder interviews in the country. The team Tarzyan. The chapter was drafted by Joern benefited from the expert guidance and Huenteler and Tu Chi Nguyen. support of Mani Khurana. • For the disruptive technologies theme, • For Kenya, data collection and stake- original research was done by Kelli Joseph holder interviews were conducted by David and Jonathan Walters, both indepen- Mwangi, an independent consultant, dent consultants, under the guidance of under the guidance of Anton Eberhard Gabriela Elizondo Azuela. The team also and Catrina Godinho and the University benefited from feedback provided by Pierre of Cape Town. Laurencia Karimi Njagi and Audinet. Zubair K. M. Sadeque provided valuable support and feedback. Econometric modeling for the report was • For Morocco, Tayeb Amegroud, an inde- done by Nisan Gorgulu of The George pendent consultant, was responsible for Washington University under the guidance of collecting data and conducting stakeholder Jevgenjis Steinbucks. interviews. The project benefited greatly Although the report’s scope is global, it ben- from the guidance provided by Moez Cherif efited immensely from detailed analysis under- and Manaf Touati. taken for the 15 countries in the Rethinking • For Pakistan, data collection and stake- Power Sector Reform Observatory. The unprec- holder interviews were carried out by Fariel edented data collection and analysis would not Salahuddin, an independent consultant. have been possible without a combination of The team benefited from the guidance independent local consultants and World Bank of Anjum Ahmad, Defne Gencer, Rikard colleagues from the respective country teams Liden, Anh Nguyen Pham, Mohammad covering the energy sector. Saqib, and Richard Spencer. • For Colombia, data collection and stake- • For Peru, Eduardo Zolezzi, an independent holder interviews were done by Edison consultant, was responsible for collecting Giraldo, Jorge Giraldo, and Isaac Dyner data and conducting stakeholder inter- Rezonzew, all of whom worked as indepen- views. Valuable guidance and support were dent consultants. Gabriela Elizondo Azuela, provided by Janina Franco. Janina Franco, Elvira Morella, and David • For the Philippines, the data collection and Reinstein provided valuable support and stakeholder interviews were undertaken guidance. by Rauf Tan, an independent consultant. • For the Dominican Republic, George Yuriy Myroshnychenko provided valu- Reinoso was responsible for collecting data able support and feedback, while Wali del and conducting stakeholder interviews as Mundo acted as a reviewer. an independent consultant. Valuable sup- • For Senegal, data collection and stake- port and feedback were provided by Pedro holder interviews were conducted by Antmann and Elvira Morella. Assane Diouf, an independent consultant, • For the Arab Republic of Egypt, data col- under the guidance of Anton Eberhard, lection and stakeholder interviews were Catrina Godinho, and Celine Paton of the conducted by Hafez el Salmawy as an inde- University of Cape Town. Valuable support pendent consultant. Marwa Mostafa Khalil and guidance were provided by Manuel and Ashish Khanna provided valuable Berlengiero, Alioune Fall, Manuel Luengo, guidance and feedback. and Chris Trimble. Acknowledgments xix • For Tajikistan, data collection and stake- guidance and support of Fabrice Karl holder interviews were conducted by Bertholet and Dmytro Glazkov. Jamshed Vazirov, an independent consul- • For Vietnam, Nguyen Trinh Hoang Anh, tant. The team benefited from the guidance Adam Fforde, Eric Groom, and Tran Dinh and support of Arthur Kochnakyan and Long, all independent consultants, were Takhmina Mukhamedova. responsible for data collection and con- • For Tanzania, Anastas Mbawala, an inde- ducting stakeholder interviews in the pendent consultant, was responsible for country. Franz Gerner and Tran Hong collecting data and conducting stakeholder Ky provided valuable guidance, while interviews under the guidance of Anton Sebastian Eckardt and Madhu Raghunath Eberhard and Catrina Godinho of the acted as reviewers. University of Cape Town. Joern Huenteler and Vadislav Vucetic provided valuable The report also benefited immensely guidance. from global databases including the World Bank • For Uganda, data collection and stakeholder Private Participation in Infrastructure Database interviews were conducted by Geofrey (PPI 2018), the S&P Global Platts World Electric Bakkabulindi, independent consultant, Power Plants Database, Regulatory Indicators for under the guidance of Anton Eberhard and Sustainable Energy (RISE), Tracking SDG7: The Catrina Godinho of the University of Cape Energy Progress Report, and other databases man- Town. Raihan Elahi and Mbuso Gwafila aged by the International Energy Agency (IEA) provided valuable support and feedback. and the U.S. Energy Information Administration • For Ukraine, Svetlana Golikova, indepen- (EIA). dent consultant, was responsible for col- The report was edited by Steven Kennedy. lecting data and conducting stakeholder Final design was provided by Critical Stages, interviews. The team benefited from the and typesetting was undertaken by Datapage. About the Authors Vivien Foster is the Chief Economist for the PhD from University College London, both in Infrastructure Vice Presidency of the World economics. Bank. Throughout her 20 years at the World Bank, she has played a variety of leadership Anshul Rana is a consultant in the Office of roles and contributed to client dialogue, as well the Chief Economist for the Infrastructure Vice as advisory and lending engagements, in more Presidency at the World Bank. He ­ specializes than 30 countries across Africa, Asia, Europe, in institutional reform in the power sector. He Latin America, and the Middle East. She has has also worked on the global dashboard to spearheaded several major policy research ini- track progress toward SDG7 goals and on RISE, tiatives, including: Water, Electricity, and the Poor an Energy Sector Management Assistance (2005), examining the distributional impact of Program (ESMAP) product used to monitor utility subsidies; Africa’s Infrastructure (2009), policy frameworks to support sustainable analyzing the continent’s network infrastruc- energy globally. Prior to joining the World ture challenges; Building Bridges (2009), detail- Bank, he taught at the School of Advanced ing China’s growing role as infrastructure International Studies at Johns Hopkins financier for Africa; Tracking SDG7: The Energy University, focusing on the political economy Progress Report (2013–18), a global dashboard for of infrastructure development and energy tracking progress toward the achievement of policy in the developing world. He has also ­ SDG7 goals for energy; and Regulatory Indicators worked as a reporter for major newspapers and for Sustainable Energy (RISE) (2016, 2018), moni- television networks in India and the United toring worldwide adoption of good-practice pol- States. He holds a master’s degree in interna- icies to support sustainable energy. She is a tional economics from the School of Advanced graduate of Oxford University; she holds a mas- International Studies at Johns Hopkins ter’s degree from Stanford University and a University. xxi Background Papers The report draws from 27 background papers • Bacon, R. W. 2018. “Taking Stock of the that are being published on a rolling basis in Impact of Power Utility Reform in the World Bank’s Policy Research Working Developing Countries: A Literature Review.” Paper series and can be accessed on the ­ project Policy Research Working Paper No. 8460, website at http://www.esmap.org/­ rethinking​ World Bank, Washington, DC. _power_sector_reform. • Huenteler, J., I. Dobozi, A. Balabanyan, and The background papers include some that S. Ghosh Banerjee. 2017. “Cost Recovery aim to paint a broad-brush picture of power and Financial Viability of the Power Sector sector reform across the developing world: in Developing Countries: A Literature Review.” Policy Research Working Paper • Foster, V., S. Witte, S. Ghosh Banerjee, and No. 8287, World Bank, Washington, DC. A. Moreno. 2017. “Charting the Diffusion of • Lee, A. D., and Z. Usman. 2018. “Taking Power Sector Reforms across the Developing Stock of the Political Economy of Power World.” Policy Research Working Paper No. Sector Reform in Developing Countries: 8235, World Bank, Washington, DC. A Literature Review.” Policy Research • Foster, V., and S. Witte. Forthcoming. Working Paper No. 8518, World Bank, “Evaluating Electricity Tariff Structure Washington, DC. Design in the Developing World.” Policy • Pardina, M. R., and J. Schiro. 2018. Research Working Paper, World Bank, “Taking Stock of Economic Regulation of Washington, DC. Power Utilities in the Developing World: Considerable emphasis was placed on con- A Literature Review.” Policy Research ducting thorough reviews of existing literature Working Paper No. 8461, World Bank, covering the main themes of this report, Washington, DC. namely, utility reform, political economy of • Rudnick, H., and C. Velasquez. 2018. power sector reform, economic regulation of “Taking Stock of Wholesale Power Markets the power sector, wholesale power markets, in Developing Countries: A Literature and cost recovery. These reviews are published Review.” Policy Research Working Paper in the following set of background papers. No. 8519, World Bank, Washington, DC. xxiii xxiv Background Papers Detailed power market case studies were Policy Research Working Paper, World prepared for four countries from the Bank, Washington, DC. Rethinking Power Sector Reform Observatory • Godinho, C., and A. Eberhard. 2019. that were most advanced in the creation of “Learning from Power Sector Reform: wholesale power markets, namely, Colombia, The Case of Kenya.” Polic y Research India, Peru, and the Philippines. Working Paper No. 8819, World Bank, Washington, DC. • Rudnick, H., and C. Velasquez. 2019. • Godinho, C., and A. Eberhard. 2019. “Learning from Developing Country “Learning from Power Sector Reform: Power Market Experiences: The Case of The Case of Uganda.” Policy Research the Philippines.” Policy Research Working Paper No. 8820, World Bank, Working Paper No. 8721, World Bank, Washington, DC. Washington, DC. • Godinho, C., and A. Eberhard. Forthcoming. • Rudnick, H., and C. Velasquez. 2019. “Learning from Power Sector Reform: The “Learn i ng from Developi ng Countr y Case of Tanzania.” Policy Research Working Power Market Experiences: The Case of Paper, World Bank, Washington, DC. Colombia.” Policy Research Working Paper • Khurana, M. Forthcoming. “Learning from No. 8771, World Bank, Washington, DC. Power Sector Reform: The Case of Andhra • Rudnick, H., and C. Velasquez. 2019. Pradesh.” Policy Research Working Paper, “Learning from Developing Country Power World Bank, Washington, DC. Market Experiences: The Case of Peru.” • Khurana, M. Forthcoming. “Learning Policy Research Working Paper No. 8772, from Power Sector Reform: The Case of World Bank, Washington, DC. Rajasthan.” Policy Research Working • R u d n i c k , H . , a n d C . Ve l a s q u e z . Paper, World Bank, Washington, DC. Forthcoming. “Learning from Developing • Khurana, M. Forthcoming. “Learning from Country Power Market Experiences: The Power Sector Reform: The Case of Odisha.” Case of India.” Policy Research Working Policy Research Working Paper, World Paper, World Bank, Washington, DC. Bank, Washington, DC. The background papers also include detailed • Lee, A. D., and F. Gerner. 2019. “Learning country case studies that provide a narrative of from Power Sector Reform: The Case of the reform dynamics for each country in the Vietnam.” Policy Research Working Paper, Rethinking Observatory and evaluate the World Bank, Washington, DC. impact of reforms on key dimensions of sector • Paton, C., C. Godinho, and A. Eberhard. performance. Forthcom ing. “Learning from Power Sector Reform: The Case of Senegal.” Policy • Bacon, R. W. 2019. “Learning from Power Research Working Paper, World Bank, Sector Reform: The Case of Pakistan.” Policy Washington, DC. Research Working Paper No. 8842, • Rana, A. 2019. “Learning from Power World Bank, Washington, DC.  Sector Reform: The Case of the Dominican • Bacon, R. W. 2019. “Learning from Power Republic.” Policy Research Working Paper, Sector Reform: The Case of the Philippines.” World Bank, Washington, DC. Policy Research Working Paper No. 8853, • Rana, A., and A. Khanna. 2019. “Learning World Bank, Washington, DC. from Power Sector Reform: The Case of the • Bacon, R. W. Forthcoming. “Learning from Arab Republic of Egypt.” Policy Research Power Sector Reform: The Case of Ukraine.” Working Paper, World Bank, Washington, DC. Background Papers xxv • Rudnick, H., and C. Velasquez. Forthcoming. Case of Colombia.” Policy Research Working “Learning from Power Sector Reform: The Paper, World Bank, Washington, DC. Case of Peru.” Policy Research Working • Usman, Z., and Amegroud, T. 2019. “Lessons Paper, World Bank, Washington, DC. from Power Sector Reforms: The Case of • Rudnick, H., and C. Velasquez. Forthcoming. Morocco.” Policy Research Working Paper “Learning from Power Sector Reform: The No. 8969, World Bank, Washington, DC. Abbreviations BOO build, own, operate BOOT build, own, operate, transfer BOT build, operate, transfer DER distributed energy resources EE energy efficiency EIA Energy Information Administration ESMAP Energy Sector Management Assistance Program EVN (national utility of Vietnam) FY fiscal year GNI gross national income GW gigawatt HHI Herfindahl-Hirschman Index IEA International Energy Agency IPP independent power producers ISO independent system operator IT information technology km kilometer KPI key performance indicator kWh kilowatt-hour LACE levelized avoided cost of energy LCOE levelized cost of energy MW megawatt MWh megawatt-hour OECD Organisation for Economic Co-operation and Development PJM Pennsylvania–New Jersey–Maryland xxvii xxviii Abbreviations PPA power purchase agreement PPI Private Participation in Infrastructure database PPIAF Public-Private Infrastructure Advisory Facility PSP private sector participation PSRI Power Sector Reform Index QFD quasi-fiscal deficit RE renewable energy REV Reforming the Energy Vision (New York) RIIO revenue using incentives to deliver innovation and outputs (United Kingdom) RISE Regulatory Indicators for Sustainable Energy SAIDI System Average Interruption Duration Index SAIFI System Average Interruption Frequency Index SDG7 Sustainable Development Goal 7 SO system operator SOE state-owned enterprise T&D transmission and distribution TSO transmission system operator Key Messages During the 1990s, a new paradigm for power be matched by well-­ governed public utilities. sector reform was put forward that empha- Restructuring and liberalization have been sized the restructuring of utilities, the creation beneficial in a handful of larger middle-­ of regulators, the participation of the private income nations but have proved too complex sector, and the establishment of competitive for most countries to implement. power markets. Twenty-five years later, only a Based on these findings, the report points handful of developing countries have fully to three major policy implications. implemented these Washington Consensus policies. Across the developing world, reforms • Context dependence. First, reform efforts were adopted rather selectively, resulting in a need to be shaped by both the political and hybrid model in which elements of market economic context of the host country. The orientation coexist with continued state dom- 1990s reform model was most successful in inance of the sector. countries that had reached certain mini- This book aims to revisit and refresh the mum conditions of power sector develop- thinking on power sector reform approaches ment and offered a supportive political for developing countries. The approach relies environment. When these same reforms heavily on evidence from the past, drawing were adopted in more challenging environ- both on broad global trends and deep case ments, the risk of policy reversal was high, material from 15 developing countries. It is while successful outcomes were by no also forward looking, considering the implica- means guaranteed. The 1990s approach to tions of new social and environmental policy power sector reform is more compatible goals, as well as emerging technological with political systems that are based on a disruptions. market-oriented ideology and contestable A nuanced picture emerges. Regulation has power structures. Economic preconditions been widely adopted, but practice often falls include a relatively large power system at a well short of theory, and cost recovery high level of electrification with good oper- remains an elusive goal. The private sector ational and financial data and a well-­ has financed a substantial expansion of gener- functioning framework of tariff regulation. ation capacity. Yet, its contribution to power The report proposes a two-track distribution has been much more limited, and approach, with countries in more challeng- its performance on efficiency can sometimes ing environments focusing on governance xxix xxx Key Messages reforms and and the achievement of finan- t wenty-first-century policy not deliver on ­ cial viability, while more ambitious struc- objectives; they need to be complemented tural reforms are deferred until systems are by more targeted policy measures. at a more mature stage of development. • Pluralist approaches . Third, countries • Outcome orientation . Second, reform found alternative institutional pathways efforts should be driven and tailored to to achieving good power sector outcomes. desired policy outcomes and less preoc- Among the best-performing power sec- cupied with following a predetermined tors in the developing world are some that process. Since the 1990s, countries’ policy decisively implemented the 1990s reform objectives for the power sector have wid- model and others that retained a domi- ened beyond security of supply and fiscal nant and competent state-owned utility, sustainability to encompass important guided by strong policy objectives, and social and environmental goals, such as combined this with a more gradualist and universal access and power sector decar- targeted role for the private sector. This bonization. The evidence indicates that evidence makes a case for greater plural- Washington Consensus reforms alone will ism of approaches going forward. Overview: Key Findings and Policy Implications INTRODUCTION Behavior changes when private management During the 1990s, a new paradigm for is introduced. Private management reorients power sector organization grew out of enterprises from bureaucratic and political the wider “Washington Consensus” on incentives to profit-seeking, cost-control, and development and was spearheaded by customer orientation. Market pressure or regu- multilateral institutions. The new para- latory incentives would discipline any potential digm came on the heels of growing dissatisfac- for private management abuses. The private tion with state-owned utilities (Bacon and sector and the regulator would prevent day-to- Besant-Jones 2001). These vertically inte- day political interference. The combination of grated monopolies had successfully supported stronger commercial incentives, competitive the rollout of national infrastructure networks pressures, and regulatory oversight was in many countries during the 1960s–80s but expected to improve the efficiency and cost had begun to show limitations in the form of recovery of power utilities. The resulting inefficient operations, burgeoning subsidies, decline in state subsidies and increase in finan- and financing constraints. The 1990s power cial viability would make possible the major sector reform model comprised a package of investment programs needed to achieve secu- four structural reforms: rity of supply in fast-growing power systems (World Bank 1993). This thought process is • Regulation (through the creation of an presented as a theory of change underpinning autonomous regulatory entity) the 1990s reform model in figure O.1. The the- • Restructuring (entailing corporatization ory of change is used as a conceptual frame- and full vertical and horizontal unbundling work for evaluating the model’s efficacy in this of the utility) study. By 2015, the adoption of Sustainable • Private sector participation (particularly in Development Goal 7 (SDG7) and the Paris generation and distribution) Climate Accord had broadened the policy • Competition (ultimately in the form of a objectives for the power sector, bringing a new wholesale power market). focus on electrification and decarbonization, The 1990s reform model was based on the goals that had not been envisaged in the 1990s. idea that reforms would lead to beneficial The aim of this study is to revisit, behavior change among the key sector actors, refresh, and update thinking on power resulting in improved sector performance. sector reform in developing countries in 1 2 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE O.1 The 1990s model was based on an underlying theory of change Reforms Intermediate outcomes Outcomes Liberalized, regulated markets for Post 2000s wholesale and retail electricity outcomes • 1st best: Competitive market • Access and Cost recovery • 2nd best: Regulated by independent agency a ordability • Reasonable return • Environmental on new investment Attracts sustainability Cost recovery prices • Cost recovery in ongoing operations Least-cost Sector outcomes investment • Enables Security of supply Cost optimization/ e cient operation • Generation and quality of • Transmission service Incentivizes E ciency • Distribution • Fiscal a ordability • E cient operations Economically e cient • Optimal resource Guides decision allocation Commercially operating utilities • 1 best: Restructured and privatized utilities st • 2nd best: Restructured, corporatized SOEs as o -takers of private investment Source: World Bank elaboration. Note: SOEs = state-owned enterprises. the light of historical evidence and future efforts and performance in 88 developing trends. The prescriptions of the 1990s reform countries, complemented by a Power Sector model were primarily derived from economic Reform Observatory that provides deep-dive theory and principles. By the early 2000s, it studies of 15 countries.1 Countries are not had become clear that the model was not uni- judged for the reforms they have undertaken versally applicable in practice (Besant-Jones but rather for the results they have delivered. 2006). We now have a quarter-century of Sector outcomes are evaluated along multiple empirical evidence against which to evaluate dimensions, including traditional objectives the approach. The case for such an evaluation such as security of supply, as well as the new hinges both on the practical difficulties encoun- policy agenda focusing on electrification and tered with the application of the model in the decarbonization. developing world, as well as on the significant At the same time, the study looks changes in policy objectives. At the same time, ahead to the technological disruptions the emergence of disruptive technologies raises sweeping the power sector, developments ­ questions about how the recommendations of that are challenging conventional wisdom the 1990s model may need to be adapted going about sector organization and structure. forward. Traditionally, power systems have developed Relying on a rich new evidence base, around centralized infrastructure designed to the study looks back over 25 years of reap economies of scale and achieve simulta- experience with power sector reform neous balancing of supply and demand across the developing world. The approach through the one-way flow of power to passive is heavily evidence-based, drawing on reform consumers. However, the current wave of Overview: Key Findings and Policy Implications 3 innovations—including decentralized renew- consideration when determining the appropri- able energy, battery storage, and ­digitalization— ate reform path for each country. Reform empowers consumers and other decentralized choices also need to be guided by desired sector actors to participate in the production of elec- outcomes, notably, with respect to decarbon- tricity and in so-called demand-response ser- ization and universal access objectives. vices,2 generating reverse flows along power Fortunately, good sector outcomes can be networks and introducing the possibility of achieved in a variety of institutional settings, as trade at the retail level. Moreover, as large- the experience of developing countries around scale battery storage becomes increasingly the world has shown. Those settings will be flexible and cost-effective, the need for power ­ tested, as new business models emerge in systems to simultaneously match supply and response to the technological disruptions that demand will recede. are reshaping the economic logic of the sector. The purpose of this overview is to sum- marize the lessons from the study and KEY FINDINGS reflect upon their implications for future This section summarizes the most relevant practice. Ten key findings are followed by the and interesting results of the study in the form policy implications of those findings. The com- of 10 key findings. prehensive analysis contained in the main report begins with a survey of the uptake of the • Finding #1. Uptake of power sector reform in 1990s power sector model by developing coun- the developing world did not evolve accord- tries, considering both the economic and politi- ing to the textbook model. cal drivers of reform. Attention then turns to • Finding #2. Power sector reforms were more the implementation of each of the fundamental likely to gain traction if they were consis- building blocks of the reform model: sector tent with the country’s political system and restructuring and governance; private sector ideology and led by champions enjoying participation; regulation; and market liberaliza- broad stakeholder support. tion. Thereafter, reform measures are evaluated • Finding #3. The private sector made an in terms of their impacts both on intermediate important contribution to expanding sector outcomes (such as ­ efficiency and cost power generation capacity in the develop- recovery) and on final s­ ector outcomes (such as ing world, albeit with significant challenges security of supply, access and affordability, envi- along the way. ronmental sustainability). • Finding # 4. Wholesale power markets The study suggests that future reforms helped improve efficiency in the minority should be shaped by context, driven by of countries that was ready for them; many outcomes, and informed by alternatives. others found themselves stuck in transition. The 1990s reform model is sometimes miscon- • Finding #5. Good corporate practices, partic- strued as a universally applicable policy ularly with respect to human resources and prescription. However, the findings reported ­ financial discipline, were associated with here suggest instead that the 1990s model con- better utility performance; these were more tains valuable insights that can support prevalent among privatized utilities. improvements in efficiency, cost recovery, and • Finding #6. Private sector participation in security of supply when deployed in the right power transmission and distribution deliv- circumstances and for the right reasons. ered good outcomes in favorable settings; However, economic and political preconditions elsewhere, it was susceptible to reversal. are found to be important determinants of • Finding #7. Regulatory frameworks have the success of reforms; these deserve closer been widely adopted, but implementation 4 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD has often fallen far short of design, partic- power sector reform model spread rapidly ularly when utilities remained under state across both the developed and developing ownership. worlds. A quarter-century later, however, the • Finding # 8. Cost recovery has proved patterns of adoption are quite different. remarkably difficult to achieve and sustain; Organisation for Economic Co-operation and the limited progress made owes more to effi- Development (OECD) countries have adopted ciency improvements than to tariff hikes. (on average) close to 80 percent of the 1990s • Finding #9. The outcomes of power sec- policy prescriptions, although with some tor reform were heavily influenced by the notable exceptions. The degree of adoption in starting conditions in each country. the developing world is much lower at under • Finding #10. Good sector outcomes were 40 percent. The level of uptake differs system- achieved by countries adopting a variety of atically according to the geographical, eco- different institutional patterns of organiza- nomic, and technical characteristics of tion for the sector. countries (map O.1). 3 Specifically, reform adoption is twice as high in Latin America rel- ative to the Middle East, in middle-income Finding #1: Uptake of power sector relative to low-income groups, and in coun- reform in the developing world did not tries with larger power systems relative to evolve according to the textbook model smaller ones. Moreover, the momentum of The diffusion of power sector reform in reform slowed markedly over time, with the developing world was strongly uptake more limited during the decade from affected by contextual factors. The 1990s 2005–15 than from 1995–2005. MAP O.1 Power sector reform spread unevenly across the developing world Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015; Regulatory Indicators for Sustainable Energy 2016. Note: PSRI score based on existing legislation (as of 2015), which may be different from practice. Overview: Key Findings and Policy Implications 5 As a result, reform implementation communities—are a natural focus for patron- diverged from the theoretical paradigm. age politics. Moreover, the cost and quality of Overall, barely a dozen developing countries electricity supply has the potential to become were able to implement the 1990s model in its an electoral issue that can mobilize public entirety. Instead, most are stuck at an interme- unrest and topple governments. diate stage of implementation, sometimes Power sector reforms almost always referred to as the “hybrid model” (Eberhard and take place in the context of a crisis Gratwick 2008). A further quarter of developing and often as part of a wider national countries—including many small, low-income, transformation process. There are few and fragile states—have barely begun to reform examples of countries that reformed in the their power systems. Underlying this partial absence of a crisis or of countries that failed to implementation has been a tendency to cherry-​ reform when beset by crisis. The triggering pick components of the 1990s model that were events sometimes originated within the power easier to implement, while leaving others aside. sector, such as a drought or oil price shock Creation of a regulatory entity and private or a situation of unsustainable utility debt. sector participation in generation through inde- ­ However, in many cases, the power sector was pendent power projects (IPPs) were, by far, the implicated in a wider national crisis, either two most popular reforms, adopted by more linked to fiscal stabilization (such as tariff than 70 percent of developing countries; the reforms in the Arab Republic of Egypt) or to uptake of other reforms was much lower. This socioeconomic transition (such as privatization à la carte approach to reform does not sit well in Ukraine). This finding underscores the fact with the original conception of the 1990s model that power sector reform does not take place in as a coherent package of mutually supportive a vacuum; it needs to be understood in terms reform measures. It meant that countries ended of the wider political and economic context. up with contradictory reform combinations, The trajectory of reform varies sub- such as private sector participation in stantially across countries, with reform distribution without a regulator—or, more fre- ­ announcements providing no guaran- quently, the other way around. tee of sustained implementation. The reform process typically begins with the pub- lic announcement of a reform program. Some Finding #2: Power sector reforms were countries then move rapidly toward imple- more likely to gain traction if they were menting the full suite of reforms announced, as consistent with the country’s political in the case of Peru (figure O.2a). In other cases, system and ideology and led by reforms rapidly lose momentum, with delivery champions enjoying broad stakeholder falling well short of original aspirations and support even being susceptible to reversal over time, as The 1990s reform model drew heavily on in the case of Senegal (­ figure O.2b). Overall, economic first principles, with no explicit the gap between reform announcements and attention to the political dynamics of the implementation can be quite considerable reform process. Yet, the reality is that the (­figure O.3). power sector is highly politicized across much Reform trajectories reflect the politi- of the developing world. Power utilities—with cal dynamics around the power sector in their significant employment rolls and con- each country, as well as the strategy tracting volumes, as well as their ability to adopted for reform implementation. direct valued electricity services to different Although reforms are announced by countries 6 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE O.2 The trajectory of power sector reform followed different paths across countries a. Peru 100 90 80 70 60 Percent 50 40 30 20 10 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 b. Senegal 100 90 80 70 60 Percent 50 40 30 20 10 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 Competition PSP Regulation Restructuring Announced competition Announced PSP Announced regulation Announced restructuring Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: PSP = private sector participation. across the ideological spectrum, evidence sug- delegating responsibility to regulators and pri- gests that those with a stronger market orien- vate operators, and allowing new entry to tation are more likely to make meaningful competitive markets. The strategy for reform progress with implementation. Similarly, implementation at the sector level is also reforms tend to proceed further in countries important. Countries that can mobilize a that have contestable or multipolar political strong reform champion, ideally supported by systems, as opposed to those where power is a stable and competent bureaucracy, generally more centralized. This is consistent with the go further with sector reform. However, observation that the reform process typically unless wider stakeholder alignment is involves the delegation and decentralization achieved through outreach efforts and ulti- of power by breaking up national monopolies, mately legislative support, reforms may prove Overview: Key Findings and Policy Implications 7 FIGURE O.3 For some countries, the gap between reform announcement and implementation has been considerable 100 80 60 Percent 40 20 0 an ia co p. l m a h n ic an a a da e ru s ga ne ny sh bi in es ha bl an Re na Pe oc ist st an ra ne m di ad pi pu Ke st nz ki et jik or Ug lo Uk ab O ilip ja Se Pa Pr Re Vi Ta Co M Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg In m – Do a di In Cumulative reform delivery, 2015 Cumulative reform announced, 2015 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. difficult to sustain and vulnerable to reversals energy technologies—now in the ascendancy— of various kinds. Finally, while donors play an the share was almost twice as high, at around important role in introducing reform ideas 70–80 percent (figure O.4b). Nevertheless, only and supporting their implementation, they do a handful of countries was able to rely exclu- not seem to have much influence on a coun- sively on the private sector for almost all new try’s overall reform trajectory, which is rather generation capacity. Foreign sponsors have been shaped by local political factors. a major source of private investment in power generation, particularly in the Middle East and Sub-Saharan Africa (­ figure O.5). South Asia Finding #3: The private sector stands out as the only region where the major- made an important contribution to ity of private investment has been domestically expanding power generation capacity sourced. in the developing world, albeit with Nevertheless, private investments in significant challenges along the way generation have not always been guided The private sector has contributed just by principles of least-cost planning. over 40 percent of new generation capacity During the 1990s, little attention was paid to in the developing world since 1990, a share power system planning, leaving many devel- that has been remarkably consistent across oping countries without strong technical country income groups. The absolute amount of capacity in this critical area. This was unfor- private investment in Africa has been relatively tunate at a time when demand for electricity low, but it still represents about 40 percent of was growing so quickly across the developing total investment, similar to other regions. Across world that the scale of the system had to income groups, the share of private sector double every decade in many countries. Even investment in capacity additions hovers around when plans were made, they were seldom 40 percent (­figure O.4a). For modern renewable enforced. Only one in five countries makes 8 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE O.4 The private sector’s contribution to new generation capacity was steady across income groups but heavily skewed by generation technology a. By income group b. By generation technology 100 100 Percent of capacity additions Percent of capacity additions 80 80 60 60 40 40 20 20 0 0 Low Upper- Lower- Upper-middle Hydro Coal Gas Geo- Oil Wind Solar Biomass income middle middle income thermal income income without China Private Public Source: World Bank elaboration based on World Bank–PPIAF Private Participation in Infrastructure Database 2018; UDI World Electric Power Plants ­database 2017. FIGURE O.5 The bulk of private investment in generation came from foreign sponsors 100 90 80 Percent of investment 70 60 50 40 30 20 10 0 Middle East and Sub-Saharan Latin America and East Asia and Europe and South Asia North Africa Africa the Caribbean Pacific Central Asia Domestic Mixed Foreign Unknown Source: World Bank elaboration based on World Bank–PPIAF Private Participation in Infrastructure Database 2018 ; UDI World Electric Power Plants database 2017. ­ power system plans mandatory, which often deals for IPPs remains widespread across Sub- leaves important decisions about plant capac- Saharan Africa and South Asia (­ figure O.6). ity vulnerable to the vagaries of political Such nontransparent procurement processes interference or unsolicited bids. In contrast to jeopardize value for money in generation and Latin America and the Middle East, where invite allegations of corruption, which has competitive tendering is more established bedeviled IPP programs in some countries— (although the number of deals in the latter Tanzania being a prominent example. region is limited), direct negotiation of Countries with strong planning and Overview: Key Findings and Policy Implications 9 FIGURE O.6 Direct negotiation of IPPs remains widespread in South Asia and Sub-Saharan Africa 100 90 Share of private power 80 70 plants (%) 60 50 40 30 20 10 0 East Asia and Europe and Latin America Middle East South Asia Sub-Saharan Overall Pacific Central Asia and the and North Africa Caribbean Africa Competetive IPPs Merchant plants Direct negotiation of IPPs Not available Source: World Bank elaboration based on World Bank–PPIAF Private Participation in Infrastructure Database 2018. Note: IPPs = independent power producers. procurement frameworks were more likely to termination. At one end of the spectrum, IPP be able to expand generation capacity to keep programs have sometimes stalled when private pace with growth in peak demand. The avail- sector demands for risk mitigation were not able evidence suggests that the features of matched by the willingness of governments to the planning and procurement framework provide them. Examples include Egypt’s first IPP most closely associated with good outcomes program in the early 2000s and Vietnam’s pro- for security of supply are the existence of gram in the 2010s. At the other end of the spec- institutional capacity for planning, the use of trum, when governments have assumed exces- a transparent and participatory process for sive risk, IPP programs have sometimes triggered developing plans, and the adoption of com- financial crises. Large-scale IPP programs have petitive bidding for new generation. left governments exposed to currency or oil Striking the right balance of risk price risks, as happened during the Asian finan- between the public and private sector in cial crisis of the late 1990s in Pakistan and the power generation has proved challenging. Philippines, where the power sector became a IPPs face a plethora of risks, including demand major contributor to public debt. risk, fuel price risk, exchange rate risk, and ter- mination risk. All can weaken investor interest, Finding #4: Wholesale power markets particularly in untested markets, until a reliable helped improve efficiency in the track record has been established. In response, minority of countries that were governments provide contractual protections of ready for them; many others found various kinds. Oil price and currency fluctua- themselves stuck in transition tions are often passed through directly to the tariff specified in the power purchase agreement Only one in five developing countries has (PPA). “Take-or-pay” clauses prevalent in many introduced a wholesale power market, African IPP contracts guarantee purchase of reflecting the formidable list of precondi- power even in the absence of demand; else- tions that must be met before such where, capacity charges at least ensure that markets become possible or meaningful. ­ fixed capital costs can be covered. Power markets are for the most part found in Sovereign guarantees often need to be provided middle-income countries whose power sys- to compensate investors in case of p ­ remature tems are relatively large, financially viable, and 10 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD unbundled (both vertically and h­ orizontally)— Effective functioning of wholesale and where regulatory governance is sound. markets requires a high-resolution, However, regional power markets at varying short-term pricing mechanism, as well as stages of development are also allowing smaller a sound and adaptive governance struc- countries in Africa, Central America, and ture. The main function of wholesale power South Asia to capture some of the benefits of markets is to provide efficient short-term power trade. price signals to guide dispatch and inform Close to half of the developing coun- investment. Prices across developing country tries have adopted the single buyer spot markets have varied widely, rang- model as a (sometimes indefinite) step ing between US$20 to US$200 per mega- toward wholesale competition. After some watt-hour, with price trends conveying the vertical and horizontal unbundling of the sec- evolution of local market conditions, such as tor, IPPs compete alongside incumbent gener- expanding investment in India or drought ators to supply power to the publicly owned conditions in Colombia (figure O.7). High spa- single buyer, which is typically the transmis- tial resolution of prices—such as the nodal sion (and sometimes also distribution) utility. prices used in Peru—is important to signal Although often conceived as a transitional transmission constraints. Close monitoring of model toward a competitive market, in prac- market prices and performance by an inde- tice most countries have remained stuck at pendent watchdog, such as the system opera- this stage. A key concern is that the long-term tor or regulator, has proved important to take-or-pay arrangements that are often detect abuses of market power often attribut- required to induce IPP investments in emerg- able to inadequate restructuring of generation ing markets can introduce distortions into assets prior to the launch of the market power dispatch and build contractual rigidity (Jamasb, Newberry, and Pollitt 2005; Jamasb, into the power system—both of which signifi- Nepal, and Timilsina 2015; Nepal and Jamasb cantly limit the scope for competition when a 2012). This has been particularly challenging wholesale market is eventually introduced. in the Philippines, but it has improved over time owing to new entries and the intercon- nection of segmented markets, reflected FIGURE O.7 Electricity spot prices have shown wide variation in tumbling wholesale market prices across developing country markets (­figure O.7). Good governance of the system 200 operator is critical for the impartial and effec- 180 tive dispatch practices that underpin price 160 ­ formation. Some countries have chosen to Prices (2015 US$/MWh) 140 combine this function with that of transmis- 120 sion system operator, which is a viable option 100 as long as conflicts of interest can be avoided. 80 The functions of system and market operator 60 have also proved possible to combine. 40 Despite expectations, spot market 20 0 prices have not provided adequate incen- tives for investment in new generation 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 capacity across the developing world. India Philippines Colombia Peru There has been relatively little entry by mer- Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. chant plants4 and limited willingness of regula- Note: MWh = megawatt-hours. tors to allow spot market prices to spike during Overview: Key Findings and Policy Implications 11 scarcity periods to the levels needed to incen- the system as needed, yet many markets lack tivize new investment. Accordingly, several mechanisms for appropriately incentivizing countries have adopted regulated capacity pay- such ancillary services. ments, which, although effective in incentiviz- ing new investment, have led to concerns Finding #5: Good corporate practices, about excess capacity—for example, in Chile. particularly with respect to human Capacity markets have also been tried, though resources and financial discipline, without success, in Colombia. Increasingly, were associated with better utility supply auctions are proving to be an effective performance; these were more model for ensuring security of supply across prevalent among privatized utilities several Latin American countries. In supply auctions, potential generators compete for the Corporatization of public utilities was right to supply power to distribution compa- conceived as a way to put the power sec- nies on a long-term basis, but they do so with- tor on a more commercial footing. Prior to out take-or-pay provisions. 1990, many public power utilities operated as More recently, decarbonization of the administrative departments of their respective generation mix has emerged as a new line ministries without any separate corporate policy objective to be pursued, creating existence. Doing so left them subject to the further challenges for wholesale power vicissitudes of public administration and unable markets. With few exceptions, decarboniza- to adopt a commercial orientation. For this rea- tion was not a major policy objective pursued son, the first step to power sector reform in through least-cost generation plans during the many countries was to separate out the opera- period under study. Generation investments tional functions associated with ­ service provi- were largely driven by concerns over security sion into a distinct state-owned corporation, of supply, which coincidentally pushed typically operating under company law. In hydro-dominated countries toward greater doing so, many important decisions were made carbon intensity and oil-dominated countries regarding the governance of the company and toward lower carbon intensity. Nevertheless, the establishment of management processes. these experiences illustrate that such There is a significant governance gap p olicy-directed investment decisions can ­ between corporatized public utilities materially move the dial on carbon intensity and privatized ones. A well-established once that becomes the objective. More ­ literature on corporate governance of state- recently, some Latin American countries, as owned enterprises provides a clear frame of well as India, have adapted their supply auc- reference for good practice in this domain. tions to explicitly support the transition to For those jurisdictions where power utilities renewable energy by targeting certain genera- are entirely state-owned, corporate gover- tion technologies. The growing share of nance tends to reflect about 55 percent of ­ variable renewable energy has created even good-­ practice measures, suggesting consider- further challenges for capital cost recovery in able room for improvement. 5 Governance the generation segment, since the presence of scores tend to be systematically higher for resources such as wind and solar—which are private utilities, falling in the 60–90 percent characterized by zero marginal cost—can lead range, a level only occasionally matched by to periods of zero and even negative spot public utilities. Boards of private utilities prices in some markets. Also, the variability of enjoy almost complete decision-making wind and solar resources increases the need autonomy, whereas those of public utilities for fast-ramping flexible resources to balance have limited freedom on critical matters of 12 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD finance and human resources—particularly service obligations, the ability to fire employees with respect to raising capital and appointing for nonperformance, the use of transparent the chief executive officer. Public utilities hiring processes for selecting employees, the also suffer considerable interference in the adoption of modern information technologies, appointment and removal of board members. and the board’s freedom to appoint and remove Overall, public utilities tend to be less rigor- the chief executive officer. ous in staff hiring, with more limited use of standard good practices, such as advertising, Finding #6: Private sector participation shortlisting, interviewing, and checking of in power transmission and distribution references. Owing to public sector employ- delivered good outcomes in favorable ment restrictions, they also have less ability settings; elsewhere, it was susceptible to reward employees through performance to reversal bonuses or to fire those who perform poorly. Public utilities also tend to fall particularly Private sector participation in transmis- short with respect to basic accounting prac- sion has not been widespread, but some tices that are universal in the private sector. successful examples exist in Latin When it comes to adoption of information America and Asia. The reform model of the technology, by contrast, there seems to be 1990s was primarily concerned with establish- relatively little difference between public and ing private sector participation in generation private utilities. and distribution. The transmission segment Good practice on corporate governance was regarded as a natural monopoly, exercising is strongly correlated with good utility system-coordination functions best handled performance in terms of cost recovery under public ownership. Nevertheless, the and distribution efficiency—irrespective experience of some countries in Latin America of public or private management. has illustrated that new transmission lines can Surprisingly little has been documented to date readily be bid out under build-operate-transfer regarding the extent to which corporatized structures where the investment climate is power utilities pursued good governance prac- adequate. These contracts are similar to those tices and the resulting performance impact. used for IPPs, but more straightforward, to the New evidence presented in this study suggests extent that there are no fuel costs or dispatch that the quality of managerial practices related issues to consider, and remuneration is reduced to human resources and financial discipline are to a simple annuity payment covering capital strongly associated with better performance on and operating costs over the life cycle. Cases of distribution efficiency and operating cost system-wide transmission concessions or recovery (figure O.8). The correlation holds even divestiture are much rarer. irrespective of whether utilities are publicly or Some of the early-reforming countries privately managed, since the best-performing introduced widespread private sector public utilities exhibit somewhat better man- participation in their distribution sectors. agement practices than their peers. Board The financial health and operational strength autonomy and accountability, however, are of distribution utilities is a key driver of overall not so clearly linked to performance. Some of power sector performance. A financially pre- the dimensions of corporate governance that carious distribution utility can undermine the are most strongly associated with efficient util- entire payment chain, while operational weak- ity performance are the publication of accounts nesses in the local grid can prevent power from consistent with international financial report- reaching customers even when it is available. ing standards, the explicit definition of public For precisely these reasons, the 1990s model Overview: Key Findings and Policy Implications 13 FIGURE O.8 Certain aspects of corporate governance are strongly associated with improved efficiency performance for distribution utilities a. Board autonomy and accountability b. Human resources Board final decision making 27 Employees can be fired 50 body on appointing a CEO 79 for poor performance 100 Board members cannot 8 Recruitment involves 67 be removed at will 44 interviewing candidates 94 Good practices Good practices Private or public shareholders 17 Recruitment involves 75 appoint board 50 short-listing candidates 100 Minority shareholders’ 25 Sta training policy 75 rights are protected 50 exists 94 58 Government employment 17 Audit committee of the board 81 33 regulations do not apply 0 20 40 60 80 100 0 20 40 60 80 100 Percent Percent c. Financial discipline d. Information technology Financial accounts meet 25 IT system to support 62 international standards 81 distribution management 93 Public service obligations 17 IT system to support 62 are explicitly defined 69 incidence resolution 86 Good practices Good practices Financial accounts are 50 Geographic 69 publicly disclosed 100 Information System 92 Utility pays dividends 0 Advanced Metering 38 to shareholders 50 Infrastructure 62 25 Customer satisfaction 46 Utility can issue new bonds 44 69 regularly monitored 0 20 40 60 80 100 0 20 40 60 80 100 Percent Percent Poor performers Good performers Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: IT = information technology. prescribed private sector participation in the utilities have coexisted within the same coun- distribution tier as one of the first measures to try, with private operators often serving capital be taken to turn around an ailing power sector. cities or larger commercial centers. The deci- This is reflected in the surge of private sector sion to privatize only some distribution utilities participation in distribution that took place may reflect differences in the commercial via- during the 1990s (figure O.9). Divestiture of bility of the service areas, or variations in the distribution utilities was prevalent among local political environment, particularly in ­ early-reforming countries in Latin America, countries where electricity distribution remains Central Asia, and Eastern and Central Europe, a subnational responsibility. although it was comparatively rare in Africa Private sector participation in distribu- and in East and South Asia. Nevertheless, even tion has proved susceptible to reversals, among countries undertaking privatization of and appetite for the reform subsided in power distribution utilities, relatively few pri- the 2000s. Overall, 32 distribution transac- vatized the entire distribution sector. More tions in 15 developing countries have been typically, public and private distribution reversed (in the case of divestitures) or 14 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE O.9 Private sector participation in distribution peaked in the late 1990s before declining 300 250 Investment (US$, millions) 200 150 100 50 0 00 08 09 10 92 98 02 90 93 94 95 96 99 04 06 13 14 91 97 01 03 05 07 11 12 15 16 17 20 20 20 20 20 20 20 19 20 19 20 19 20 20 20 19 19 20 19 19 19 19 19 20 20 20 20 20 Concession (lease) Divestiture (full, partial) Greenfield (BOT, BOO) Management contract Source: World Bank elaboration based on World Bank–PPIAF Private Participation in Infrastructure Database 2018. Note: BOO = build, own, operate; BOT = build, operate, transfer. prematurely terminated (in the case of conces- or Uganda). Such concerns led to a dramatic sions and other contractual instruments), par- tail-off in private sector participation in ticularly during the first decade of reform. The electricity distribution after the early 2000s ­ probability of reversal was particularly high (figure O.9). in Sub-Saharan Africa, affecting more than Private sector participation in distribu- 20 percent of transactions. Sub-Saharan tion is strongly associated with full cost Africa’s experiments with utility management recovery. Private sector participation is the contracts, in particular, have been checkered, only reform that is associated with higher lev- encountering difficulties in recruiting and els of full capital cost recovery, as opposed to retaining qualified managers and suffering recovery of operating costs alone. Among the from tense labor relations and inadequate countries reviewed that have undertaken sig- transfer of skills to local staff. Privatization nificant and sustainable privatization of the reversals were most often associated with distribution segment, it is exceedingly rare for defective operational data (for example, seri- tariffs to fall below full cost recovery levels. ous underestimation of system losses) that led This partly reflects the fact that countries to unsustainable bids (for example, in the achieving higher levels of cost recovery are Indian state of Odisha), or with the govern- more likely to attract private sector participa- ment’s unwillingness to apply tariff regulation tion; it also indicates that the presence of the as laid down in the legal framework (as in the private sector obliges the government to follow Dominican Republic). Stakeholder opposition through on tariff regulations that call for cost has also been a serious issue in some cases (as recovery pricing. in Senegal, where the labor unions vehemently With respect to efficiency, the perfor- opposed utility privatization). Customers, in mance of privatized distribution utilities is particular, often bear the brunt of tariff hikes on par with the top half of performers associated with privatization, without always among public utilities. Many of the ­ privatized seeing an immediate impact on the quality of utilities studied perform to a high degree of service, and this can sometimes lead to public ­ owever, a operational efficiency (figure O.10). H disaffection (as in the Pakistani city of Karachi group of publicly owned utilities (in the Indian Overview: Key Findings and Policy Implications 15 FIGURE O.10 Private sector participation is associated with much higher levels of cost recovery, while performance on efficiency is within the range observed for public utilities a. E ciency b. Cost recovery NPC (Vietnam) HCMCPC (Vietnam) Meralco (Philippines) 100% 120% EPM (Colombia) Dniproblenergo (Ukraine) Hidrandina (Peru) Luz del Sur (Peru) EVN (Vietnam) ONEE (Morocco) Codensa (Colombia) APSPDCL (Andhra Pradesh) Dniproblenergo (Ukraine) CESU (Odisha) Codensa (Colombia) APEPDCL (Andhra Pradesh) Beneco (Philippines) LESCO (Pakistan) 100% Hidrandina (Peru) Kenya Power Luz del Sur (Peru) Khmelnitskoblenergo (Ukraine) LESCO (Pakistan) Meralco (Philippines) Distribution e ciency as percent of revenue UMEME (Uganda) ONEE (Morocco) 90% 80% WESCO (Odisha) TANESCO (Tanzania) UMEME (Uganda) Full cost recovery ratio SENELEC (Senegal) Senelec (Senegal) Barki Tojik Kenya Power Tanesco (Tanzania) Dominican 60% Republic Barki Tojik (Tajikistan) Beneco (Philippines) (Egypt, Arab Rep.) K-Electric Alexandria (Egypt, Arab Rep.) (Pakistan) JDVVNL (Rajasthan) JVVNL (Rajasthan) 80% JVVNL (Rajasthan) 40% Edesur (DR) Edenorte (DR) JDVVNL (Rajasthan) WESCO (Odisha) CESU (Odisha) 20% APEPDCL (Andhra Pradesh) K-Electric (Pakistan) Cairo North (Egypt, Arab Rep.) APSPDCL (Andhra Pradesh) 70% 0% Public Private Public Private Public vs. private utilities Public vs. private utilities Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: Red boxes indicate utilities that have seen privatization rollback. state of Andhra Pradesh, Morocco, and that private sector participation has a signifi- Vietnam) performs as efficiently as the privat- cant positive impact on generation capacity ized utilities. There are also some privatized util- and electricity access in low-income countries ities facing difficult operating environments and that it supports the expansion of renew- (such as in the Pakistani city of Karachi or the able energy in middle-income countries. Indian state of Odisha) that perform no better However, by far the strongest driver of than some of the worst public utilities. At the electrification is income per capita, same time, some of the worst-performing public rather than any structural reform. The utilities are cases of failed privatization (as in the substantial progress on electrification made in Dominican Republic and Senegal). many countries approaching middle-income There is also evidence that private sec- status from 1990–2015 primarily took the tor participation is associated with good form of utility-driven, grid-based electrifica- sector outcomes. Ultimately, the impact of tion programs backed by clear political targets reform is best evaluated in terms of results. and public investment. In some cases (such as Analysis undertaken for this study suggests India, Morocco, and Vietnam), these efforts 16 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD predated the sector reform process in the the efficacy of regulation. Regulatory frame- country. In other cases (such as Kenya, works are to varying degrees overlooked or con- Tanzania, and Uganda), they were adopted tradicted in practice (Andres, Guasch, and Diop long after the power sector reform, usually in 2007; Gilardi and Maggetti 2011). Whereas, on response to the limited dynamism of electrifi- average, countries meet about 47 percent of cation in the post-reform period. Grid electri- good practice regulatory standards on paper, this fication may be loss-making for the utility at score drops to 30 percent in practice.6 The gap the margin, meaning that it cannot be left to between regulation on paper and regulation in commercial incentives alone. With the advent practice can be relatively narrow (as in Peru and of solar technology, decentralized private sec- Uganda, where the gap is less than 10 percent- tor actors are playing an increasingly import- age points) or extremely wide (as in the ant role in the electrification process, although Dominican Republic and several Indian states, the jury is still out on the question of whether where the gap can be 30–50 percentage points) the most remote populations can be served on (figure O.11a). One critical area is the authority a purely commercial basis. of regulators to determine electricity tariffs, which is legally granted in 94 percent of coun- tries but actually honored in only 65 ­ percent— Finding #7: Regulatory frameworks with a lot of caveats. Not surprisingly, the have been widely adopted, but achievement of operating cost recovery is sig- implementation has often fallen far short nificantly related to the quality of regulation as of design, particularly when utilities practiced rather than as written. remained under state ownership Although originally conceived as an The creation of regulatory agencies was enabler of privatization and competition, widely embraced and supported by regulation was often introduced into sec- sound regulatory frameworks in many tors still dominated by monopolistic countries. As of 2015, over 70 percent of state-owned actors. Many countries that fit developing countries had created a power this description adopted legal frameworks ­ s ector regulator. On paper, the associated based on the principles of incentive regulation, ­ r egulatory frameworks were relatively according to which the regulator harnesses the well-designed, incorporating provisions to utility’s profit motive to incentivize efficient balance the autonomy and accountability of delivery of high-quality services. Such incen- the regulatory framework. In addition to the tives are not typically effective unless regulated central functions of all such entities—­ utilities operate according to strong commer- regulation of tariffs and service quality (based cial principles, making them responsive to on detailed methodologies laid down in the incentives. Regulation does seem to have regulatory framework), regulators are widely worked quite well, however, in countries responsible for licensing market entry, includ- with largely privatized distribution sectors. ing negotiation of the terms of PPAs Moreover, evidence indicates that the presence (85 ­percent) and competitive procurement of private actors in the sector is associated with (60 percent). They may also play a role in much closer adherence by regulators to the other important policy areas, such as clean established legal framework. The reason may energy (80 percent), power market design simply be that it is more difficult for the gov- (65 percent), and electrification (55 percent). ernment to deviate from enacted regulations In practice, however, it has proved very when third-party private actors are involved. difficult to apply regulatory frameworks Where utilities remain in public hands, as written, and this has adversely affected the Ministry of Finance can become an Overview: Key Findings and Policy Implications 17 FIGURE O.11 Significant divergence exists between regulation on paper and regulation in practice a. By country 80 70 70 66 Regulatory framework index (%) 58 60 60 60 54 54 52 52 48 50 45 46 46 46 43 37 39 40 36 36 36 34 27 28 30 25 19 21 21 20 11 9 10 3 3 0 an a a h ia an da m n a e ru s ic p. l ny sh ga es ne bi in ha an bl na Pe Re ist st an ra m di ad Ke ne pi st pu nz ki et jik Ug lo Uk O ab ilip ja Pa Pr Se Vi Ta Re Co Ta Ra – Ar Ph ra a an di – dh t, ic a yp In An di in Eg In m – Do a di In De jure score Perceived score b. By aspect of regulation 77 Market entry 56 75 Quality 51 77 Tari s 65 71 Autonomy 65 83 Accountability 73 0 10 20 30 40 50 60 70 80 90 100 Percent Average de jure Average perceived Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. important player in the tariff-setting the frame. Several countries, such as Egypt process. Countries where utilities remain and Senegal, have explicitly recognized this in publicly owned are often characterized by their tariff-setting frameworks, committing to weak regulatory authority over tariff-setting fiscal transfers that exactly compensate for and a soft budget constraint overall. When any shortfall in cost recovery from tariffs. This tariffs are not allowed to keep pace with costs approach acknowledges that sector costs (figure O.12), a degree of fiscal liability is must ultimately be covered by a combination ­ created bringing the Ministry of Finance into of taxes and user charges and provides a 18 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD coherent framework for making such trade- While regulators have struggled with offs. Nevertheless, the Senegalese experience tariff-setting challenges, quality-of-­ service i llustrates the challenges of meeting such ­ regulation has not received the attention commitments during periods of fiscal stress. that it deserves and is too often observed in the breach. The shortfall in practice is partic- FIGURE O.12 Regulatory tariff recommendations are not ularly large for regulations pertaining to quality always respected in practice of service and market entry (figure O.11b). a. Andhra Pradesh, India Indeed, few countries were found to have a 800 meaningful system in place for regulating quality of service. (Colombia and Peru are ­ 600 among the few that do.) On the one hand, legal Tari adjustment (%) requirements to develop and monitor quali- 452 ty-of-service standards and penalize noncompli- 400 ance are not always observed by regulators. On the other hand, utilities may lack the informa- 275 200 tion systems to fully comply with such a frame- work and to manage r ­eliability issues ade- 0 quately. This is a serious deficiency, given the 2010 2011 2012 2013 2014 2015 importance of service ­reliability for customers. b. Odisha, India 800 Finding #8: Cost recovery has proved remarkably difficult to achieve and 600 sustain; the limited progress made Tari adjustment (%) owes more to efficiency improvements 400 446 than to tariff hikes Full cost recovery has been a challenge for 200 power utilities. Only about half of them can be 183 considered financially viable. Over the 25-year 0 period under review, the extent to which end- 2010 2011 2012 2013 2014 2015 user tariffs covered the full capital cost of supply- ing electricity increased from 69 percent to 79 c. Rajasthan, India 800 percent, and about as many countries saw their performance on cost recovery deteriorate as 683 improve (figure O.13a). Strikingly, even coun- 600 Tari adjustment (%) tries with relatively low cost of service some- times struggle to achieve full capital cost 400 recovery. In fact, full capital cost recovery is almost exclusively confined to utilities that have 200 been privatized. Experience shows that progress 220 toward cost recovery is subject to sudden erosion by exogenous f ­actors, such as droughts, devalu- 0 2010 2011 2012 2013 2014 2015 ations, and oil price shocks. Although full capital Proposed adjustment Actual adjustment cost recovery has proved difficult to attain, Source: World Bank elaboration based on Rethinking Power Sector Reform almost all of the utilities have achieved operating utility database 2015. cost recovery. Moreover, about half of the Overview: Key Findings and Policy Implications 19 FIGURE O.13 More countries made progress on efficiency than on cost recovery, 1990–2015 a. Cost recovery b. E ciency Ukraine India – Andhra Pradesh Uganda Vietnam Colombia India – Rajasthan Tajikistan Peru India – Odisha Colombia Egypt, Arab Rep. Dominican Republic Peru Philippines Vietnam Uganda India – Rajasthan Senegal Pakistan Kenya India – Andhra Pradesh Morocco Kenya Tajikistan Senegal Pakistan Tanzania Tanzania Morocco Ukraine Philippines Egypt, Arab Rep. Dominican Republic India – Odisha –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 –40 –30 –20 –10 0 10 20 Change in full cost recovery Change in average system losses compared to prereform (%) compared to prereform (%) Improvement Deterioration Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. utilities can be ­considered financially viable in recommended adjustments aggressively scaled the sense of covering both operating costs and back or even completely overturned by the historic debt service and repayment obligations, political authorities. albeit without providing a full rate of return on Utilities with revenue shortfalls are sel- their asset base. dom fiscally compensated. The quasifiscal Where progress on full cost recovery deficit across the study sample remains high, was made over time, cost reductions averaging close to one percentage point of gross played a greater role than tariff adjust- domestic product, with underpricing being the ments in bringing utilities closer to this major contributor in most cases. Financial anal- goal. Specifically, average system losses across ysis of the utilities showed that such shortfalls the study sample of countries fell from 24 to are not typically compensated by fiscal transfers 17 percent between 1990 and 2015, and from the state. Instead, utilities are forced to improvements were observed in more than adopt a range of suboptimal coping strategies 80 percent of jurisdictions (figure O.13b). that often include taking on high-cost short- Indeed, some countries would already be able term commercial debt to cover cashflow short- to fully recover costs based on current tariffs falls or simply falling into arrears with upstream if they could raise their commercial and opera- suppliers of bulk fuel or electricity. tional efficiency to industry benchmarks. Tariff Cross-subsidies among customer adjustments, however, have proved hard to groups and across consumption levels apply as some regulators have seen their have long been the norm for electricity 20 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD tariff structures and may further under- ­ reconditions—both economic and political— p mine cost recovery. About three-quarters of have emerged as important in shaping the developing countries practice cross-subsidies applicability of the approach. Across the between commercial and residential custom- developing world, systematic differences can ers, with the former paying on average more be observed in the uptake of the 1990s reform than twice as much as the latter for each unit model across countries, based on factors such of electricity purchased. A similar share of as income group, system size, and political countries makes use of increasing block tariffs system. Contextual factors also seem to have for residential customers, which typically pro- played a role in shaping the outcome of vide sizable discounts at low or even average reforms (table O.1). The analysis distinguishes consumption levels and then step up tariffs for among “comprehensive reformers,” which higher levels of consumption without ever applied at least 70 percent of the prescriptions reaching full cost recovery even in the highest of the 1990s model; “limited reformers”; consumption brackets. Deeper analysis shows “stronger performers,” which scored above that while modest amounts of cross-subsidy average on outcome variables capturing prog- have been accommodated historically without ress on security of supply, electrification, and seriously prejudicing the achievement of cost decarbonization; and the remaining “weaker recovery, cross-subsidization can seriously performers.” undermine the financial equation of the utility A first group of countries largely if even the highest-paying customers are not applied the full policy prescriptions of the paying at the cost recovery level. 1990s reform model and went on to see a range of positive outcomes as a result, experiencing improved operational Finding #9: The outcomes of power efficiency and cost recovery, as well as ­ sector reform were heavily influenced by enhanced security of supply. Foremost the starting conditions in each country among these were Colombia, Peru, and the The 1990s power sector reform model Philippines. In all these cases, the reform was largely derived from principles package was adopted comprehensively and ­ believed to apply universally, indepen- ­ relatively rapidly during the 1990s without dent of context. In practice, numerous major implementation setbacks. A continuous TABLE O.1 Overview of preconditions among groups of countries at the time of reform Sector preconditions Country preconditions Cost of Full cost System Access to Electricity System Income Quality of electricity recovery losses electricity consumption size level governance ($/kWh) (%) (%) (%) (kWh pc pa) (GW) (GDP pc) (index) Comprehensive reformers Stronger performers 0.15 69 19 82 1,413 20 1,405 –0.43 Weaker performers 0.17 70 30 53 315 15 756 –0.49 Limited reformers Stronger performers 0.13 55 21 77 804 22 737 –0.55 Weaker performers 0.23 84 27 27 172 2 428 –0.40 Source: World Bank data. Note: The Rethinking Power Sector Reform observatory countries are assigned their categories, specified in the table. All figures relate to the pivotal reform year for each country during the decade of the 1990s or the nearest data point available in some cases. India and Ukraine are excluded from system size calculations. GDP = gross domestic product; GW = gigawatt; kWh = kilowatt-hour; pa = per annum; pc = per capita. ­ Overview: Key Findings and Policy Implications 21 process of second-generation reforms fine- sector context, the strong performers started tuned the operation of the model. Each of out with much better operational performance these countries faced its own challenges, but in terms of system losses (19 percent versus these could be accommodated, by and large, 30 ­percent), much higher levels of electrifica- within the parameters of the new institutional tion (82 percent versus 53 percent), and a framework. much more developed energy system with A second group of countries also significantly higher installed capacity (20 giga- ­ adopted comprehensive reforms but did watts versus 15 gigawatts). Their per capita not experience the same comparatively electricity consumption was about four times smooth implementation and positive out- as high. Even among the group of countries comes. In Pakistan, for example, the unbun- that made only limited reforms, the stronger dled power sector has been plagued by a performers enjoyed significantly better precon- chronic circular-debt crisis that undermines the ditions than those with weaker performance. payment chain; the only privatization in the distribution sector continues to be disputed in Finding #10: Good sector outcomes the courts after more than a decade of litiga- were achieved by countries adopting tion. In the meantime, the country struggles to a variety of different institutional achieve security of supply and universal access patterns of organization for the sector to electricity. Other illustrative cases include the Dominican Republic and the Indian state of Although the 1990s reform model started Odisha, where an extensive power reform was out with a unified reform blueprint, that undertaken, including privatization of distribu- blueprint was adapted to widely varying tion utilities. However, in both cases, it proved degrees. A significant minority of countries difficult in practice to apply the prescribed remains with a traditional vertically integrated framework of tariff regulation, leading to a national utility model, while the majority finds subsequent renationalization and persisting itself under an assortment of hybrid models. concerns about security of supply, as well as Countries where adoption of reforms has weak performance on both intermediate and been slower or more limited have, in some final outcomes. cases, performed as well, in terms of sector out- Some insight into these disparate expe- comes, as those that went further with the riences can be gained by comparing the reform agenda. Comparing across a wide range preconditions that existed in these two of postreform outcomes covering security of groups of comprehensive reformers at the supply, social inclusion, and environmental time of the reform in the 1990s (compare sustainability shows that the stronger perform- the first two rows of table O.1). In particular, ers divide into two equal groups comprising those countries where reforms proved to be both comprehensive and limited reformers successful started out from a much more (table O.2). The performance differences are advantageous national and sectoral position remarkably small between these two groups than the others. In terms of country context, of countries; the limited reformers do slightly the strong performers had already achieved an better on reliability, access, and affordability, income level (around US$1,400 per capita) and slightly worse on overall adequacy of that was approximately double that of the capacity and carbon intensity. In a similar fash- weak performers, and they also enjoyed a bet- ion, the weaker performers are also evenly ter institutional environment, as captured by split between countries that took a more com- the World Bank Governance Index. In terms of prehensive or limited approach to reform. 22 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE O.2 Comparison of country performance according to reform strategy Security of supply Social inclusion Normalized capacity Capacity Meeting Access to Carbon (in MWs per million diversification demand electricity Affordability intensity SAIFI population) (HHI) (ratio) (%) (% of GNI) (gCO2/kWh) Stronger performers Comprehensive reformers 8.8 551 0.4 1.1 94 4.6 357 Limited reformers 4.3 429 0.6 1.0 99 1.4 387 Weaker performers Comprehensive reformers 61.3 166 0.2 0.9 62 4.7 330 Limited reformers 30.3 45 0.5 1.0 35 13.0 419 Sources: IEA data; World Bank data. Note: All figures relate to postreform performance as measured in 2015. The outcomes are judged based on a detailed framework provided in table 9.2 in chapter 9. Green signifies good outcome, yellow signifies moderate outcome, and red signifies poor outcome. GNI = gross national income; Herfindahl-Hirschman Index; MWs = ­ HHI = ­ megawatts; SAIFI = System Average Interruption Frequency Index. Of particular interest, then, are the institu- spearheaded by the national utility of Vietnam tional paths taken by limited reformers that (EVN). The country is moving toward the achieved stronger performance outcomes. staged implementation of a wholesale power Salient in this group are countries like Morocco market, in which a minority of privately and Vietnam, as well as the Indian state of owned generators competes alongside pub- Andhra Pradesh. What these cases appear to licly owned subsidiaries of EVN. have in common is a continued role for a com- In the Indian state of Andhra Pradesh, the petent state-owned utility, with a more tar- state government completed unbundling and geted role for the private sector. regulatory reforms but stopped short of privatiz- Morocco kept a vertically integrated, pub- ing the distribution segment. Instead, consider- licly owned monopoly at the core of the sector, able efforts were made to sharpen incentives for while opening to the private sector for certain managerial performance through the establish- generation plants and city-level distribution ment of clear performance indicators r ­ elating concessions. Rather than focusing on structural to revenue collection, combined with frequent reform and the creation of regulatory capacity, monitoring by senior management and finan- Morocco’s energy policy was characterized by cial reward for good outcomes. This approach the articulation of clear and ambitious social was combined with legal reforms to make and environmental objectives at the highest power theft a prosecutable criminal offense. political level. Those objectives were accompa- Finally, although Kenya does not feature nied by clear institutional responsibility and among the stronger performers globally, it accountability for delivery and supported by does present the best overall range of sector adequate investment finance, capturing both outcomes among the Sub-Saharan African public and private sources as appropriate. case studies considered. Kenya’s approach to In Vietnam, the sector continues to be reform was also incremental and distinctive. dominated by the incumbent utility operating In particular, majority public ownership was as an unbundled public sector holding retained in the distribution sector, but an ­ c ompany with weak regulatory oversight. almost equal share of equity floated on the Vietnam’s power sector journey prioritized Nairobi Stock Exchange provided an addi- the achievement of universal access through tional discipline on corporate governance of a sustained and well-financed program Kenya Power. Overview: Key Findings and Policy Implications 23 POLICY IMPLICATIONS financial distress resulting from grid The 10 policy implications that follow draw on defection. the review of historical evidence provided and Second, the speed and coherence of the on a forward look at disruptive technology technological transition will depend critically trends in the power ­ sector. The momentous on the design of the regulatory framework, technological changes underway—­ notably, which shapes the incentives for innovation. increasingly cost-effective decentralized Incentives for utilities to innovate depend on ­ technologies—are posing fundamental ques- the regulatory regime under which they oper- tions about the viability of the traditional ate, since it is this that determines whether centralized utility and promising to change ­ and how investments and operational savings the structure of the power sector. In some can be turned into profits. Incentives for cus- frontier markets, the wave of change takes tomers to innovate will depend on how much the form of distribution utilities splitting into freedom they are given by the regulatory a wires business and a distribution system framework to engage in decentralized energy operator, whose primary role is to provide a production and storage activities, as well as platform that consumers and businesses can the associated impact on tariffs. Incentives for use to trade energy both within the distribu- new players to enter the market and innovate tion segment and into the wholesale power will similarly depend on the flexibility of the ­ market. In other cases, the new technologies regulatory licensing regime. In view of this, it are seen primarily as opportunities to improve is clear that the design of the regulatory the efficiency and effectiveness of the tradi- framework will give countries a certain tional utility. amount of discretion to accelerate or impede As these debates play out into an uncertain the uptake of disruptive technologies. future, at least two things seem clear. The following policy implications draw on First, power consumers will no longer be lessons learned from past experience and also captive to underperforming utilities. The identify how disruptive technologies are likely to technological disruption in OECD member affect aspects of the power sector reform agenda. countries is taking place against a backdrop of universal access to a relatively high-­ • Policy implication #1. The design of power quality and reasonably priced grid service. In sector reforms should be informed by the contrast, across the developing world, many enabling conditions of each country and ori- utility customers are faced with a costly and ented primarily toward achieving better sec- unreliable supply. Historically, the only alter- tor outcomes. native for unsatisfied customers was to sup- • Policy implication #2. The design of power sec- ply their own electricity using expensive tor reform needs to be thoroughly grounded ­ d iesel generators. As rooftop solar power in the political realities of each country. becomes cheaper and approaches grid parity, • Policy implication #3 . Greater emphasis self-generation will become increasingly should be placed on building institutional attractive where utility service is deficient, capacity for power sector planning and particularly once battery storage becomes associated implementation. more cost-effective. This development will • Policy implication # 4 . Generation plants start to contest the monopoly power of the should be procured through a transparent incumbent utility, potentially providing and competitive process, with as much con- incentives for improved performance. At the tractual flexibility as the context allows. same time, there is the risk that already pre- • Policy implication #5. Unbundling should carious utilities may be exposed to further not be the highest priority where more 24 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD f ­undamental financial and governance seem to have played a role in shaping the challenges persist; it should be undertaken outcome of reforms. primarily to facilitate deeper reforms. Experience suggests that it may be helpful • Policy implication #6. Wholesale power mar- to think about power sector reform engage- kets remain a viable option for countries ments in two phases, depending on the nature that have put in place all the foundational of the country environment. This overall measures; others may derive greater benefit framework is depicted in table O.3, which pres- from regional trade. ents the reform measures likely to be applica- • Policy implication #7. Greater efforts should ble in more challenging versus more mature be made to strengthen the corporate gov- environments, as well as the enabling condi- ernance and managerial practices of state- tions that signal a country’s readiness for vari- owned utilities. ous aspects of the reform package. • Policy implication #8. The regulatory frame- In more challenging environments, a basic work needs to be adapted to reflect the set of preliminary reform measures is proposed. institutional context and to accommodate This applies to countries that may be challenged emerging technological trends. by low incomes, fragile settings, small scale, or • Policy implication #9. Private sector partici- other limiting factors. The priority in these pation in distribution should be considered environments should be to work toward a only when enabling conditions are met. foundation of good sector g ­ overnance and basic • Policy implication #10. Delivering on the financial viability, without embarking on overly twenty-first century agenda of universal complex structural reforms. access and decarbonization calls for addi- tional reform measures targeted explicitly The policy implications are as follows: at these objectives. • Regulation. Critical at this juncture is to adopt a transparent and well-founded tariff-setting methodology and to apply Policy implication #1: The design it each year. This could be done by a reg- of power sector reforms should be ulatory agency or, at this stage, by a com- informed by the enabling conditions petent unit within the Ministry of Energy of each country and oriented primarily or the Ministry of Finance. An adequate toward achieving better sector i nitial aspiration for tariff-setting would ­ outcomes be to ensure financial viability through The 1990s power sector reform model recovery of enough capital costs to service was derived from economic first princi- and repay existing debt. Equally important ples believed to apply universally, inde- would be for the Ministry of Energy to lay pendent of context. As a result, it lacks a the foundations for monitoring the quality framework for customizing reform to the of service. The process of tariff and qual- country context. In practice, numerous ity regulation should be integrated with enabling conditions—both economic and other processes for overseeing state-owned political—have emerged as important in enterprises (relating, for example, to per- shaping its applicability. Across the develop- formance contracts or fiscal transfers). ing world, systematic differences can be • Restructuring. This is unlikely to be a high observed in the uptake of the model across priority at this stage. A vertically integrated countries, depending on their income group, power system may be easiest to manage system size, political system, and other fac- while putting in place strong foundations tors. Drawing on the case studies that have for the sector. However, the entry of the informed this study, contextual factors also private sector into generation—through Overview: Key Findings and Policy Implications 25 TABLE O.3 Customizing power sector reforms to country context More challenging environments Enabling conditions More mature environments Regulation Establish clear tariff-setting methodology • Cost recovery ratio exceeds Create separate regulatory entity. with oversight from ministry of energy or 70 percent. finance. Aim for full capital cost recovery. • Revenue collection ratio Aim for achievement of limited capital cost exceeds 90 percent and is Ensure enforcement of quality-of- recovery (that is, financial viability). enforced by disconnection. service regulation. • System losses are below Establish clear quality-of-service framework 15 percent. with oversight from Ministry of Energy. • Electrification rate exceeds 80 percent. Restructuring Retain vertically integrated utility, and Restructure the power sector to selectively introduce private investment • Regular audited financial separate out the transmission system for new plants. accounts are compliant operator and ensure adequate degree with international financial of competition in generation. reporting standards. Privatization Focus on establishing sound corporate • Modern IT systems are in Strengthen commercial incentives in governance arrangements and good place and deliver good distribution through measures such as: managerial practices for power operational data. credit-rating and bond issues; stock distribution, with special focus on • Regular tariff adjustments market listing; and/or private sector human resource management and are in line with regulatory participation. measures to promote financial discipline. methodology. Prioritize electrification through carefully • The political context is planned parallel efforts with reach of the supportive, in terms of grid and off-grid, backed up by strong ideology, leadership, and political commitment and adequate public stakeholders. funding. • Generation capacity reaches 1–3 GW. Competition Ensure adequate technical capacity for • No major bottlenecks exist Open the grid to third-party access and power system planning directly linked to on the transmission grid or in allow bilateral contracting between competitive procurement of generation. fuel supply. generators and large customers. Introduce economic dispatch of Create wholesale power market. generation plants administered by utility. Conduct supply auctions for investment in new plant. Source: World Bank. Note: GW = gigawatts; IT = information technology. supply contracts with the utility—can play procurement process. Furthermore, some of a valuable role in expanding capacity. the benefits of a competitive market can be • Private sector participation. It may be best at mimicked through the administrative prac- this stage to limit private involvement to tice of economic dispatch. generation. For the distribution segment, the emphasis should be on building good In more mature environments, it becomes governance and managerial practices, par- feasible to contemplate a more sophisticated ticularly with respect to financial disci- package of reforms, as long as these improve pline and human resource management. sector outcomes. This applies particularly to • Competition. The only relevant form of compe- middle-income countries with stable political tition at this stage is likely to be competition environments and large power systems, where for the right to build new generation plants. progress has been made toward good gover- Particularly critical is the development of the nance and financial viability for the sector. technical capacity required to conduct least- Given that reform is a means to an end, the cost planning to determine what plants to ­ priority in these environments should be to build, with mandatory links to a competitive identify where power sector performance 26 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD continues to fall short of expectations and to The transition from challenging to mature pursue more advanced reform measures geared environments can be gauged in terms of certain to delivering results in these specific areas. key enabling conditions. In practice, it may not be necessary or feasible for countries to meet The policy implications are as follows: every one of these enabling conditions; how- • Regulation. Thought should be given to ever, the more conditions that are met, the bet- establishing a separate regulatory entity if ter are the prospects for implementation of the one does not already exist. It now becomes more sophisticated reforms. Most of these more important to set tariffs to achieve full enabling conditions are related to readiness for capital cost recovery, as well as to tighten the introduction of private participation in dis- enforcement of quality-of-service regula- tribution. This is more likely to succeed when tion. Strengthening the regulatory frame- certain minimum thresholds of financial viabil- work is particularly critical if the policy ity and commercial efficiency have been passed, objective is for the sector to repay invest- and when the challenge of electrification is at a ment finance at market rates. reasonably advanced stage. Good financial and • Restructuring. This is the right juncture at operational data systems will also help to which to consider vertical unbundling reduce information asymmetries and increase to create a separate transmission system confidence among private participants, as will a operator that will support impartial third- good track record of regulatory tariff-setting party access to the grid. At the same time, it and a conducive political environment. Other becomes important to break up generation enabling conditions are more directly related to assets to provide for sufficient competitive the establishment of wholesale power markets. pressure among market players. In particular, the power system should be large • Private sector participation. Countries moving enough to support at least five competing gen- toward a wholesale power market should eration firms (at least 3 gigawatts) and to gen- ideally divest at least part of their generation erate enough turnover to justify the fixed costs assets to the private sector to ensure some of establishing market platforms (at least US$1 diversity of ownership among ­ competing billion in annual revenues). companies. In the distribution tier, countries experiencing operational inefficiencies may Policy implication #2: The design wish to consider private sector participation. of power sector reform needs to be Where public utilities are performing effi- thoroughly grounded in the political ciently, the case for private sector participa- realities of each country tion is weaker; the need to raise additional capital, however, may make it necessary for Commitments to power sector reform the utility to obtain a credit rating to support should reflect a sober assessment of the access to bond finance, or a minority stock country’s political economy. The 1990s exchange listing, both of which will also reform model drew heavily on economic first have the desirable effect of tightening the principles, with no explicit attention to the utility’s financial discipline. political dynamics of the reform process. • Competition. Countries at this stage are ready Yet, the reality is that the power sector is to consider the transition to a wholesale highly politicized across much of the develop- power market. This should be accompanied ing world. Understanding a country’s political by parallel supply auctions or an equivalent d ynam ics and ho w they im p inge o n measure to ensure timely development of ­ stakeholder interactions in the power sector adequate new generation capacity. should be the starting point for any power Overview: Key Findings and Policy Implications 27 sector reform. Rather than overlooking the was that the advent of a wholesale power political dimension, a smart reform process market would somehow circumvent the need should be adapted to fit the political context, for planning. The ultimate goal of the 1990s harnessing potential reform champions and model was to create a competitive market. At explicitly engaging in consensus-building with the time, it was assumed that private invest- contrarian groups. ments in power generation would be ade- quately guided by price signals. The role of the The policy implications are as follows: state was seen primarily as the regulator of a • Undertake a political economy analysis before privately owned and operated competitive engaging in reform. The analysis should aim sector, and great emphasis was placed on the at discovering how the power sector touches creation of a capable regulatory institution upon the country’s vested interests and and associated legal framework. Central plan- political groupings to identify potential win- ning functions were overlooked or down- ners and losers from reform. It should also played. Indeed, in some countries, the consider whether the proposed direction planning function traditionally housed in of reform is compatible with the country’s national power utilities or line ministries fell ideological orientation and broader political through the cracks as power sector reform system. The findings of the political econ- processes worked to unbundle the incumbent omy analysis should explicitly guide the utilities and to build technical capacity in reg- design of the reform program to be adopted. ulatory agencies operating outside of line • Integrate outreach and communication efforts ministries. In practice, power markets proved to engage all relevant stakeholders. The com- difficult to establish in all but a handful of munications campaign should be based on developing countries; even there, price signals messages that can be used by the reform have not provided an adequate basis for champions to articulate the value propo- investment decisions. sition associated with the reform. Those ­ messages can be disseminated through The policy implications are as follows: a variety of channels. Communications • Create strong technical capacity for planning and should be complemented by outreach that empower the planning function. The develop- directly engages with all stakeholders, ment of a strong planning capacity for the particularly those most threatened by the development of new generation and trans- reform process. In addition to an inten- mission infrastructure should be prioritized sive effort at the outset of a reform process, as a critical component of power sector there is a need to monitor the state of pub- reform. Various alternative institutional lic opinion throughout implementation, as models have been successfully used around sudden changes in the political environ- the world to locate the planning function, ment can easily lead to reform reversals. including the line ministry, the transmis- sion utility, the system operator, or a dedi- cated technical agency. Regulators can play Policy implication #3: Greater a valuable role in the technical review of emphasis should be placed on investment plans as part of the process of building institutional capacity for setting revenue requirements for capital power sector planning and associated expenditure. implementation • Make sure the power system plan is actually The 1990s model had little to say on the implemented. As important as the planning issue of planning. The implicit assumption process itself is a strong link between the 28 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD power system plan and the procurement need to mitigate risk to reassure investors of new generation and transmission plant, entering uncharted waters has left many so that procurement is aligned with the countries with rigid take-or-pay contracts and plan and contracted in a timely and cost-­ extensive guarantee clauses that both con- effective manner that keeps pace with strain the efficiency of dispatch and saddle the demand. Without such a clear linkage, utility and the government with onerous governments are vulnerable to unsolicited liabilities. proposals that may not represent the most cost-effective option for the power system. The policy implications are as follows: • Incorporate new technologies in power system • Mandate the use of competitive procurement planning. Technologies such as distributed for generation projects. Competitive bidding energy resources, together with storage of new generation plants should be the and demand response, have the potential default modality for procurement. If unso- to reduce the costs of reaching supply-­ licited proposals are considered—only in demand balance. However, the incorpora- clearly defined and exceptional cases and tion of such resources is not considered in when their prefeasibility and compatibil- traditional power system planning, in part ity with existing investment plans can be because they introduce significant complex- established—they should also be subjected ­ ity into standard planning ­ methodologies, to a competitive process. but also because they would not necessar- • Maximize the flexibility of contractual provisions. ily be undertaken by the incumbent utility. Risk-mitigation mechanisms will inevitably Storage—in particular—can play multiple be needed in unproven environments, but roles in the power system, potentially sub- these should be carefully scrutinized and stituting for conventional investments in limited to the minimum required to meet generation, transmission, and distribution investors’ legitimate expectations of return. assets. There is a need to modernize plan- Doing this could mean, for instance, scaling ning tools and techniques to integrate such back the volume or duration of take-or-pay considerations. clauses or making use of two-part pricing mechanisms that separate capacity and energy charges. Policy implication #4: Generation • Consider the adoption of supply auctions wher- plants should be procured through a ever possible. The foregoing challenges have transparent and competitive process, been successfully addressed by countries with as much contractual flexibility as that have moved toward the adoption of the context allows supply auctions, ensuring a pipeline of reg- Although IPPs have proved a popular ular, well-structured offerings of batches and effective means of bringing private of new generation plant. These are linked capital into power generation, much to long-term contracts with distribution room for improvement remains in the utilities that give generators first right of way such projects are implemented. supply without committing to take-or-pay Direct negotiation of projects, often in arrangements. A growing number of coun- response to unsolicited proposals, remains tries are adopting such mechanisms to pro- widespread across Africa and Asia, raising cure variable renewable energy, and these concerns about value for money and the could readily be extended to cover other potential for corruption. At the same time, the technologies. Overview: Key Findings and Policy Implications 29 Policy implication #5: Unbundling or to introduce private sector participation should not be the highest priority in a specific segment of the industry but not where more fundamental financial and elsewhere. The enabling conditions would governance challenges persist; it should include (1) a minimum system size of at be undertaken primarily to facilitate least 1 gigawatt to avoid the loss of econo- deeper reforms mies of scale and (2) adequate institutional governance, including strong payment dis- In the past, power sector restructuring cipline and technical coordination along has, at times, been treated as a panacea for the supply chain. reform and prioritized as an early reform measure. However, in and of itself, power sec- tor restructuring does little to tackle the funda- Policy implication #6: Wholesale mental issues of weak governance and financial power markets remain a viable option fragility that plague the power sector in many for countries that have put in place all developing countries. Moreover, restructuring a the foundational measures; others may sector that suffers from weak governance and derive greater benefit from regional financial fragility may only exacerbate the chal- trade lenges of technical coordination and financial The 1990s power sector reform model held payment along the supply chain. up a competitive power market as the In reality, unbundling was never endpoint of reform. The aspiration intended as an isolated reform measure remains legitimate, but it has proved to be but rather as a necessary precursor for a farther out than originally envisaged. The competitive market. Unless the latter is a difficulty of fulfilling the many enabling condi- realistic possibility in the medium term, tions that a wholesale power market requires restructuring the sector may not be a pressing has deferred indefinitely the introduction of matter. Unbundling entails significant trans­ such markets across much of the developing action costs, as well as the potential loss of world. Nevertheless, their attainment remains a economies of scale and scope, which should valuable and legitimate aspiration, provided that not be underestimated (Pollitt 2008; Vagliasindi the enabling conditions can be met. Indeed, the 2012). For these reasons, the relevance of present wave of technological disruption only unbundling to smaller power systems is partic- increases the value of wholesale power markets, ularly questionable. There is a well-established which, when properly designed, can support minimum size threshold of 1 gigawatt before the discovery of rapidly evolving costs and foster countries should even consider embarking on the integration into the power system of vari- sector restructuring, and a further threshold of able renewables, ancillary services, battery stor- 3 gigawatts before they definitely need to age, and demand response. unbundle should they be preparing for the establishment of a wholesale power market. The policy implications are as follows: • Ensure that the enabling conditions for a whole- The policy implications are as follows: sale power market are in place. Countries • Consider unbundling when there is a clear pur- should not consider developing such a pose for doing so and where enabling conditions market until a wide range of preconditions ­ are in place. The purpose behind unbun- have been met. These include the following: dling might be to establish a wholesale (1) a fully restructured power sector that has power market in the not-too-distant future created at least five competing g ­ enerators 30 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD with diversified ownership, (2) an absence forms of energy (with associated storage) of significant constraints in transmission and can increasingly be used to c ­ ontract for or fuel availability, (3) a financially viable adequate ancillary services to balance vari- sector with a solid payment chain, (4) solid able renewable energy. regulatory practices, and (5) sufficient sys- ­ • Modernize wholesale power markets to accom- tem size. A wholesale power market entails modate new resources. Conventional power certain fixed costs that are unlikely to be market designs are not adapted for the justified by the potential efficiency gains presence of variable renewable energy until the market is large enough. As a rule resources, battery storage, or increasingly of thumb, power markets are not likely sophisticated demand response. I nte- to become very interesting until a coun- grating them calls for the development of try reaches a national market turnover of new pricing mechanisms that are able to around US$1 billion, which is equivalent to remunerate the ancillary services required a power system size of some 3 gigawatts. for the successful integration of variable • Avoid getting locked into transitional arrange- renewable energy, provide suitable price ments. Countries that are ready to move signals to incentivize efficient investment to a competitive market should consider in utility-scale battery storage, and allow carefully whether transition mechanisms demand-response aggregators to participate are really needed, since experience sug- in the process of dispatch. gests there is a relatively high risk of getting • Participate in regional and cross-border trad- stuck in intermediate stages, in particular, ing arrangements wherever possible. Regional the single-buyer model. power markets also offer significant benefits • Establish a strong transmission system operator. for arbitrage based on differential generation The transmission utility plays a critical role costs and load profiles among neighboring in a competitive power market, ensuring countries. Other benefits include shared equitable access of third parties to the grid reserve margins and greater flexibility to infrastructure, and potentially also playing a accommodate variable renewable energy. leading role in power sector planning, system For countries not yet ready to develop planning, and sometimes market operation. wholesale power markets domestically, • Monitor and adapt the design of the wholesale regional markets can provide an important power market based on implementation expe- first step. Nevertheless, even regional mar- rience. Wholesale power markets may not kets entail certain basic minimum enabling always function according to design. Proac- conditions that cannot always be taken for tive monitoring for potential abuses of mar- granted—in particular, creditworthiness on ket power is very important, particularly in the part of power importers and security of the early stages, as is the flexibility to learn supply on the part of power exporters. from this experience and adapt market • Move toward economic dispatch of power plants. design accordingly. Deviations from principles of economic • Provide a parallel mechanism for incentivizing dispatch are widespread in the develop- investment in generation. Short-term market ing world, leading to major generation price signals alone are not always adequate inefficiencies. Countries not yet ready to to provide incentives for investment in new develop wholesale power markets should capacity. Parallel capacity mechanisms are consider having their system operator needed, with supply auctions proving to move toward the practice of economic be particularly efficient and effective. Such ­ d ispatch based on the marginal costs of auctions can be adapted to target low-­ carbon operating different plants. Overview: Key Findings and Policy Implications 31 Policy implication #7: Greater efforts utility performance. Again, these comprise should be made to strengthen the standard measures, such as the publication of corporate governance and managerial externally audited financial accounts that are practices of state-owned utilities prepared in conformity with i ­nternational financial reporting standards. Another good The 1990s reform model focused on pri- practice is the explicit identification and cost- vatization of distribution utilities, but ing of public service ­ obligations that cannot the reality is that most remain publicly be justified on commercial grounds. owned. The creation of corporatized public utilities out of traditional ministerial depart- ments was viewed as a short transitional Policy implication #8: The regulatory measure toward eventual privatization, framework needs to be adapted to which would lead to a full overhaul of man- reflect the institutional context and to agerial practices. However, given the rela- accommodate emerging technological tively limited uptake of privatization in the trends distribution segment, it has become very The creation of sector regulators has important to address enduring weaknesses in been a popular reform, but many of these the corporate governance of public utilities. entities find themselves regulating pub- The evidence shows that there is wide varia- lic rather than private utilities. The power tion in the performance of public utilities; a sector reform model of the 1990s envisaged substantial minority reaches efficiency levels the creation of a regulatory entity as a prereq- comparable to private utilities, while the uisite for introducing private sector participa- majority continues to flag. Better-performing tion, particularly in power distribution. The public utilities share many aspects of good regulator was supposed to play the dual role corporate governance with each other and of protecting private investors from opportu- with private utilities. nistic government meddling, while also pro- tecting consumers from abuses of privately The policy implications are as follows: held monopoly power. The evidence suggests • Improve human resource management of that regulation has functioned much more ­public utilities. Public utilities should take effectively where the private sector entered care to apply aspects of human resource power distribution than where utilities management that are strongly associated ­ remained state-owned. with improved performance. These relate Moreover, the regulatory regimes of the primarily to the quality of the selection 1990s did not anticipate the current wave process for hiring employees—in ­ ­ particular, of technological disruption in the power the application of standard good practices, sector. The power sector has seen momentous such as advertising vacancies, shortlisting technological change since the development of and interviewing candidates, and conducting the 1990s power sector reform model. The reference checks. The liberty to fire employ- changes are challenging the traditional ees for underperformance is also found to be approach to tariff regulation, which is based on important, although this is often difficult to ensuring that the utility collects enough reve- enforce in public sector environments. nue to enable it to roll out new infrastructure. • Strengthen financial discipline of public utili- It also raises questions about the traditional ties. Similarly, public utilities should adopt design of tariff structures that were often moti- certain aspects of financial discipline that ­ vated by social policy concerns in a context are strongly associated with improved where consumers were largely captive. 32 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD The policy implications are as follows: be held accountable through performance • Ensure that the instruments of price regulation contracts with the Ministry of Energy, for are consistent with the governance of the utility. example, while tariff-setting is inextricably There is little value in applying the instru- linked with financial oversight and subsidy ments of incentive regulation—designed decisions that lie in the hands of the Minis- to harness the profit motive of private try of Finance. Rather than creating paral- utilities—to state-owned utilities that are ­ lel tracks, regulation should build upon and not driven by profit maximization and may integrate these complementary processes. not even operate under hard budget con- Quality-of-service regulation should be straints. In these cases, it makes more sense reflected in the key performance indicators to use traditional cost-of-service regulation determined under performance contracts. and focus on creating supportive manage- Tariff and subsidy decisions should be taken rial performance incentives. Even the cre- simultaneously in a coordinated manner, ation of a separate regulatory entity may be ensuring that the overall revenue require- less of a priority when the sector remains ments of public utilities are met through a state-owned, because, in practice, both the combination of both sources. utility and the regulator are likely to be • Give greater attention to creating a credible reg- closely overseen by the line ministry, mak- ulatory framework for quality of service. With ing regulatory independence somewhat regulatory attention focused primarily on illusory. Nevertheless, irrespective of which tariff-setting, efforts to provide a credible institution is responsible for regulation, framework for monitoring quality of ser- a clear, well-grounded methodology for vice and enforcing the achievement of the tariff-setting, applied on an annual basis, is ­ prescribed standards have been inadequate. of tantamount importance. Such a framework is of critical importance • Aim for limited capital cost recovery initially. to ensure that regulatory reforms yield tan- Most regulatory tariff methodologies are gible benefits for electricity consumers. based on principles of full capital cost • Test the “future-readiness” of the regulatory recovery, including remuneration of the framework. The regulatory pricing regime for full asset base at the market cost of capital. power utilities can affect the incentives for Where utilities have been privatized, this adoption of new technologies. For instance, principle is critical for financial sustain- traditional cost-of-service regulation will ability. However, in the case of state-owned not encourage a utility to adopt technologies utilities, which often benefit from signif- that may reduce demand for energy or meet icant capital grants, it is not essential to demand at a lower investment cost. The reg- remunerate the full asset base at the market ulatory licensing regime may also create cost of capital. Rather, the concern should barriers to the entry of new actors, such as be to ensure that the utility is able to cover providers of distributed energy resources the costs associated with the loans that are or demand aggregators. There is therefore carried on its books. This limited capital a need to review existing regulatory frame- cost recovery, which ensures the financial works to evaluate whether they offer ade- viability of the enterprise, is a reasonable quate incentives for innovation. interim tariff-setting objective. • Ensure that the economics of decentralized elec- • Integrate regulation with other key public sector tricity supply are reflected in tariff structures. processes for state-owned utilities. In some coun- Electricity tariff structures have tradition- tries, regulatory frameworks coexist with ally been designed under the premise that other forms of state oversight. Utilities may consumers have limited alternatives to grid Overview: Key Findings and Policy Implications 33 electricity, so pricing can be guided primar- Private sector participation is more likely ily by considerations of fairness and equity to be successful in circumstances where rather than economic efficiency. This prac- (1) there is reasonably accurate information tice has led to tariff structures under which about the operating performance of the util- costs are recovered primarily through vol- ity and the condition of its assets; (2) retail umetric charges, with extensive embedded tariffs are relatively close to full (capital) cross-subsidies across consumption bands cost recovery (at least 70 percent); (3) it is and consumer groups. Because such tariff accepted that customers can be disconnected structures fail to recognize the fixed-cost for nonpayment of bills; and (4) a competent nature of the power grid, they overreward regulator possesses the power to adjust tar- customers choosing to self-supply and fail to iffs as needed and the technical competence convey time-of-use price signals that would to monitor quality of service. incentivize customers to participate more • Evaluate whether the political preconditions for actively in demand response. Future tariff privatization of distribution are in place. Even structures will have to give greater weight when the economic preconditions for pri- to fixed charges that take into account cus- vate sector participation are in place, politi- tomer load. Volumetric charges will have to cal impediments may remain. Private sector reflect time of use and be designed in com- participation is more likely to be politically bination with structures to remunerate pro- feasible in circumstances where (1) there sumers injecting power into the grid.7 is a broad, established tradition of private sector–led economic activity; (2) domestic actors can be involved in the privatization; Policy implication #9: Private sector (3) the value of private sector participation participation in distribution should is clear; and (4) positive outcomes can be be considered only when enabling arranged for key stakeholder groups.  conditions are met • Explore alternative modalities for engaging the Privatization of distribution utilities has private sector. The 1990s model considered delivered good outcomes in suitable private sector participation primarily in environments, but it has proved risky terms of private ownership, or at least man- where conditions were not right. Private agement, of the utility. However, finan- sector participation in power distribution was cial markets can provide another channel widely adopted in Latin America and parts of through which private sector discipline can Europe and Central Asia, with outcomes that be introduced into power distribution. This were quite encouraging. Nevertheless, it has can be done through mechanisms such as also been associated with disappointing per- listing minority shares of a state-owned formance and dramatic reversals in cases utility on a local stock exchange or having where the utility was not yet functioning at a the utility secure a credit rating and raise basic level or the authorizing environment its own bond finance. was weak. Some countries that eschewed util- • Maintain a proper focus on energy access. ity privatization found other ways to incorpo- Strengthening the utility’s commercial rate the benefit of private sector discipline orientation should sharpen its incentive to through financial market channels. expand its market through electrification. However, in many developing countries, The policy implications are as follows: unserved customers are unprofitable owing • Determine whether the economic precondi- to high incremental costs and relatively low tions for distribution privatization are in place. consumption. This underscores the need 34 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD to complement distribution reforms with a but additional policy and planning mea- sound electrification planning process com- sures must be taken to direct investors prising clear targets, an associated public toward cleaner energy options. Private sec- funding program, and a suitable monitor- tor investment in generation can make a signifi- ing framework. At the same time, off-grid cant contribution to expanding renewable rural electrification can be advanced by energy capacity. In addition, a wholesale power creating a suitable enabling environment market, particularly when complemented by for private provision of off-grid solar power. supply auctions, can provide a useful mecha- nism for price discovery related to new technol- ogies, as well as a solid economic framework for Policy implication #10: Delivering pricing services ancillary to variable renewable on the twenty-first century energy and for remunerating demand response. agenda of universal access and Nevertheless, the evidence suggests that signifi- decarbonization calls for additional cant progress toward decarbonization over the reform measures targeted explicitly past 25 years has been primarily driven by at these objectives p olicy targets rather than by institutional ­ Universal electrification eventually comes reforms per se (figure O.14b). For most coun- into conflict with a utility’s commercial tries over this period, the overriding policy goal incentives and requires parallel policy and for generation was security of supply rather financial supports. Strengthening ­ utilities’ than decarbonization, leading oil-dependent commercial orientation through private sec- countries to become less carbon-intensive as tor participation or other means can drive they diversified into gas, and hydro-dependent a rapid expansion of connections in urban areas. countries to become more carbon-intensive as However, extending access to electricity to the they diversified into fossil fuels. periurban and rural periphery often leads a util- ity into diminishing and even negative marginal The policy implications are as follows: returns on investment, particularly if the power • Advance electrification on multiple fronts. Coun- consumption of poor households remains very tries making the most rapid progress toward low. Thus, universal electrification cannot be electrification have done so by making simul- achieved purely by allowing a utility to pursue taneous progress on- and off-grid, based on commercial incentives. It requires complemen- an integrated spatial master plan. They typi- tary policy action to set access targets, provide cally make long-term commitments to ambi- sustained public subsidies to offset the associated tious electrification targets, supporting them financial losses, and exploit the opportunities with public and donor finance and providing offered by solar technology for off-grid electri- a suitable enabling environment. A critical fication. Looking back over the past 25 years, issue is to ensure that both the upfront and progress on electrification was not typically ongoing costs of electricity are affordable for synchronized with power sector reform (figure the target populations. O.14a); rather, it reflected policy commitments • Determine explicit policy targets for decarbon- that became increasingly likely as a country’s per ization. Achieving decarbonization goals capita income grew. In some countries, the big requires explicit government direction of push on electrification preceded sector reform; investment decisions in power generation, in others, it came more as an afterthought. as well as incentives for the adoption of Power sector reform provides certain low-carbon technologies and more efficient enabling conditions for decarbonization, consumption of energy. Overview: Key Findings and Policy Implications 35 FIGURE O.14 Progress on twenty-first-century policy objectives for electrification and decarbonization, 1990–2015, countries ranked in descending order of reform effort a. Electrification b. Decarbonization Philippines Philippines Peru Peru Ukraine Ukraine Uganda Colombia Colombia Dominican Republic Dominican Republic India India Pakistan Pakistan Kenya Kenya Vietnam Vietnam Senegal Senegal Egypt, Arab Rep. Egypt, Arab Rep. Morocco Morocco Tanzania Tanzania Tajikistan Tajikistan 0 20 40 60 80 100 –400 –200 0 200 400 600 800 1,000 Electrification rate (%) Carbon intensity (gCO2/kWh) Source: Based on data from Tracking SDG7 report and IEA. Source: Based on data from Tracking SDG7 report and IEA. Note: Dark shaded bars represent prereform electrification; the light Note: Dark shaded bars represent average value in 2010–15; light shaded shaded bars represent the change since then. For all panels the color of the bars represent the change in values from prereform era. For all panels bar follows the evaluation framework set up in table 9.2. IEA = International the color of the bar follows the evaluation framework set up in table 9.2. Energy Agency; SDG7 = Sustainable Development Goal 7. gCO2/kWh = grams of carbon dioxide produced per kWh; IEA = Inter­ national Energy Agency; SDG7 = Sustainable Development Goal 7. CONCLUSIONS relatively sophisticated environments. As a Overall, it is recommended that future result, it lacks a framework for adapting reform reforms be increasingly shaped by con- to the country context. In practice, numerous text, driven by outcomes, and informed preconditions—both economic and political— by alternatives. have emerged as important in shaping its applic- First, there is a need to shift from a con- ability. A more structured approach to mapping text-neutral approach to reform to one out such prerequisites should figure promin- that is shaped by context. An overarching ently in future efforts along the lines offered in message is that the design of reforms should be this report. sensitive to country conditions. The 1990s Second, there is a need to shift from power sector reform model was largely derived process-oriented reform to outcome-­ from economic first principles and first tested in oriented reform. The 1990s model focused 36 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD primarily on a particular package of institutional 4. Merchant plants are typically nonutility reforms, which, it was argued, would lead in power generation plants that compete to time to better overall sector outcomes. Rather, it sell power. They usually do not have long term power purchase agreements and are is important to design a reform process by iden- mostly found in competitive wholesale tifying the most critical outcomes and working power market places. backward from there to identify the measures 5. A Utility Governance Index measures the most likely to remove key bottlenecks and road- extent to which specific utilities conform to blocks preventing achievement of the desired good practices. It is difficult to say exactly when and how good governance and man- outcomes. agement practices have been adopted over Third, there is a need to shift to a more time, because such measures are usually pluralistic range of institutional models. implemented within institutions and do Although the 1990s power sector reform blue- not necessarily involve major legal or struc- print has demonstrated its ability to deliver in tural changes that can readily be tracked in certain country contexts, the results have been the public record. Nevertheless, it is possi- ble to measure the current rate of adoption quite disappointing in other settings. Moreover, of such practices. Based on a sample of 19 some countries that adopted only limited state-owned and 9 privatized utilities from reforms have achieved outcomes at least as the 15 observatory countries, the Utility good as those achieved by countries that went Governance Index measures the existence of further with the reform agenda. These findings best practices in utility rules and regulations. For example, a utility may, on paper, allow make the case for a more pluralistic approach to managers to hire and fire employees based power sector reform going forward, recognizing on performance—and the index captures that there is more than one route to success. this—however it is unable to tell whether the manager actually does so. This index is NOTES described in greater detail in chapter 4 and 1. The Rethinking Power Sector Reform the full technical definitions are provided in Observatory includes Colombia, Dominican the annex of the chapter. Republic, the Arab Republic of Egypt, India 6. The survey conducted in each of the 15 (states of Andhra Pradesh, Odisha, and Observatory countries included 355 categorical Rajasthan), Kenya, Morocco, Pakistan, Peru, and quantitative questions on the regulatory the Philippines, Senegal, Tajikistan, Tanzania, system. The questions were both descriptive Uganda, Ukraine, and Vietnam. and normative. Normative questions aimed 2. Demand response is defined as when the to capture regulatory best practices based on end user changes their electricity usage the literature. To synthesize the normative in response to price signals or incentives data in a convenient and intelligible format, ­payments. a Regulatory Performance Index was created. 3. A simple Power Sector Reform Index was Two versions of the same index were calcu- constructed to aggregate data across the four lated for each country. First, a de jure index dimensions of power sector reform con- derives from the country’s regulatory frame- sidered in this study. The index gives each work as captured on paper in laws, regula- country a score in the range 0–100 on each tions, and administrative procedures. Second, dimension of reform. The scores give equal a perception index determines whether the weight to each step of each dimension on paper provisions are applied in practice. The the reform continuum. The simple average local consultant in each country provided of the four 0–100 scores is used to summa- the perception index; his or her professional rize the extent of reform. The index is purely opinion was informed by some 20 interviews descriptive and has no normative value. with key stakeholders in the reform process. This index is described in greater detail in The perception index was also reviewed by Chapter 2, and full technical definitions are the World Bank country energy team knowl- provided in the annex of the chapter. edgeable about local context. Despite best Overview: Key Findings and Policy Implications 37 efforts, this second index is more subjective by D. Levi-Faur, 201–14. Edward Elgar than the first. This index is described in further Publishing. detail in chapter 6 with technical definitions in IEA (International Energy Agency), IRENA the annex of the ­ chapter. (International Renewable Energy Agency), UN 7. Prosumers are entities that consume as well as (United Nations), WBG (World Bank Group), produce electricity. and WHO (World Health Organization). 2018. Tracking SDG7: The Energy Progress Report. REFERENCES Washington, DC: World Bank Group. Andres, L., J. Guasch, and M. Diop. 2007. Jamasb, T., R. Nepal, and G. R. Timilsina. Assessing the Governance of Electricity Regulatory 2015. “A Quarter Century Effort Yet To Agencies in the Latin American and Caribbean Come of Age: A Survey of Power Sector Region: A Benchmarking Analysis. Washington, Reforms in Developing Countries.” Policy DC: World Bank. Research Working Paper 7330, World Bank, Bacon, R. W., and J. Besant-Jones. 2001. “Global Washington, DC. Electric Power Reform, Privatization and Jamasb, T., D. Newberry, and M. Pollitt. 2005. “Core Liberalisation of the Electric Power Industry in Indicators for Determinants and Performance of Developing Countries.” Annual Review of Energy the Electricity Sector in Developing Countries.” and the Environment 26 (November): 331–59. Policy Research Working Paper No 3599, World Banerjee, S. G., A. Moreno, J. Sinton, T. Primiani, Bank, Washington, DC. and J. Seong. 2017. Regulatory Indicators for Jayarajah, C., and W. Branson. 1995. Structural Sustainable Energy: A Global Scorecard for Policy and Sectoral Adjustment: World Bank Experience, Makers. Washington, DC: World Bank. 1980–92. Washington, DC: World Bank. Besant-Jones, J. 2006. “Reforming Power Nepal, R., and T. Jamasb. 2012. “Reforming the Markets in Developing Countries: What Have Power Sector in Transition: Do Institutions We Learned?” Energy and Mining Sector Matter?” Energy Economics 34 (5): 1675–82. Board Discussion Paper No. 19, World Bank, Pollitt, M. 2008. “The Arguments for and Washington, DC. against Ownership Unbundling of Energy Eberhard, A., and K. Gratwick. 2008. “Demise of Transmission.” Energy Policy 36 (2): 704–71. the Standard Model of Power Sector Reform PPI Database. 2018. Private Particicipation in and the Emergence of Hybrid Power Markets.” Infrastructure Database. https://ppi.world​ Energy Policy 36 (10): 3948–60. bank.org. Foster, V., S. Witte, S. G. Banerjee, and A. Moreno. Vagliasindi, M. 2012. “Power Market Structure 2017. “Charting the Diffusion of Power Sector and Performance.” Policy Research Working Reforms across the Developing World.” Policy Paper 6123, World Bank, Washington, DC. Research Working Paper 8235, World Bank, Williams, J., and R. Ghanadan. 2006. “Electricity Washington, DC. Reform in Developing and Transition Gavin, M., and D. Rodrik. 1995. “The World Bank Countries: A Reappraisal.” Energy 31: 815–44. in Historical Perspective.” American Economic World Bank. 1993. “The World Bank’s Role Review 85 (2): 329–34. in the Electric Power Sector: Policies for Gilardi, F., and M. Maggetti. 2011. “The Effective Institutional, Regulatory, and Independence on Regulatory Authorities.” Financial Reform.” Policy Paper, World Bank, In Handbook on the Politics of Regulation, edited Washington, DC. Setting the Stage PART I 39 What Do We Mean by Power Sector Reform? 1 MOTIVATION working off the landmark World Development During the 1990s, a new paradigm for power Report of 1994 (Gavin and Rodrik 1995; sector organization emerged from the wider World Bank 1994). “Washington Consensus,” a term coined in By the early 2000s it had become clear, 1989. Multilateral institutions spearheaded however, that power sector reform was not the new paradigm across the world, and it universally applicable in the developing w ­ orld. rapidly took hold just as the Soviet Union was Despite widespread uptake in Latin America unraveling. Marked by 10 neoliberal policy ­ and Eastern Europe, implementation proved recommendations, the era featured two poli- more complicated than a ­ nticipated. The cies that were particularly relevant to the reforms required many distinct implementa- power sector—namely, the privatization of tion phases that often called for second-­ state-owned enterprises (SOEs) and the generation reforms (as in Brazil and ­ Turkey). abolition of regulations restricting competi- ­ In addition, the associated political and social tion. International financial institutions, challenges could be s­ ignificant. Sometimes, the notably the World Bank (box 1.1) and the ­ announced reforms could not be implemented, International Monetary Fund, played promi- as, for example, in Lebanon or ­ Zambia. At nent roles in diffusing market ideas through- other times, reforms were implemented only out the developing ­ world. They encouraged to be ultimately reversed, as with the renation- structural and sectoral adjustment programs alization of privatized utilities in Bolivia and based on market-oriented macroeconomic Kazakhstan. Many developing countries were ­ and fiscal policies (Jayarajah and Branson selective about the 1990s policy recommenda- 1995). The World Bank explicitly recom- tions, an approach that produced various mended enforcing conditions to make gov- hybrid models and outcomes (Gratwick and ernments commit to market reforms in the Eberhard 2008). It became evident that the power sector (World Bank 1993). Other mul- 1990s reform model had been developed tilateral development banks and bilateral largely in the context of member countries of donors adopted similar approaches (Williams the Organisation for Economic Co-operation and Ghanadan 2006). Financial support was and Development (OECD), where, with uni- accompanied by extensive research, advocacy, versal access achieved and demand stagnant, technical assistance, and capacity building, boosting efficiency was the primary c ­ oncern. 41 42 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 1.1 The World Bank and three decades of power sector reforms Because the World Bank is often associated with the 1990s reform model, it is important to understand how the insti- tution’s own understanding and implementation of power sector reform have evolved, through the lens of official guid- ance documents published since the 1990­s. 1993 policy paper A World Bank policy paper issued in 1993 stands out as an early articulation of the 1990s reform ­ model. The policy paper clearly articulates a shift in World Bank policy, which had once “largely supported the state-owned monopoly power utilities,” (World Bank 1993, 11) toward a new lending focus that would “aggressively pursue the commercial- ization and corporatization of, and private sector participation in, developing-country power sectors” ( ­ World Bank 1993, 16). The paper further clarified that “a requirement for all power lending will be an explicit country movement toward the establishment of a legal framework and regulatory process satisfactory to the Bank” ( ­ World Bank 1993, 59). Thus, the paper clearly advocates the establishment of an independent regulator and a movement toward more commercial and corporate principles governing the sector, with emphasis on increased private sector ­ participation. Unbundling or restructuring was suggested as a means to increase efficiency and reduce ­ costs. Although this was the main thrust of the paper, there was also some passing recognition that “there can be no one standard approach … (World Bank 1993, 56). Rather, “individual countries should be encouraged to review and select the for all countries” ­ options, mechanisms, and pace of reform most appropriate to their needs and circumstances” ­ (World Bank 1993, 22). 1994 World Development Report The World Development Report 1994: Infrastructure for Development further underlines the shift in policy toward the 1990s reform model across all infrastructure sectors—including power—entailing greater competition, stronger com- mercialization, and a more market-oriented ­ approach. The report stresses that infrastructure should be managed “like a business not a bureaucracy” and with more private sector involvement in management, financing, or ownership to ensure a commercial orientation (World Bank 1994, 2). The report calls for introducing competition directly (open entry) ­ ccountability. It views unbundling as promoting or indirectly (through competitive bidding) to increase efficiency and a “new entry and competition in segments that are potentially competitive” but cautions against higher transaction costs ­ (World Bank 1994, 53). 2003 World Bank Group review A 2003 review of the World Bank Group’s experience in promoting private sector participation in the power sector states that the Bank “underestimated the complexity and time required” for reforms to mature and achieve lasting and equitable sector outcomes (World Bank 2003, 26). It goes on to underline that “no single blueprint” is suitable for all sector reforms and private sector participation and that “an evolving menu of options for the combinations and sequences of reforms” had developed over time ( ­ World Bank 2003). The report mentions that the 1993 poli- cy’s emphasis on “learning by doing” had led to inadequate weight being placed on the political economy of reform ­ (World Bank 2003, v ­ i). The report mentions that reforms to date were “complex, time consuming, resource intensive ­ World Bank 2003, ­ and require[d] sequencing,” although it provides few details on their timing and sequencing ( ix). 2004 guidance note Within a decade of the 1993 policy paper, the World Bank’s operations were moving toward a more nuanced application of the 1990s reform model, captured in the World Bank’s 2004 guidance note for “Public and Private Sector Roles in the Services.” This note clearly states that staff should consider the full range of reform options, from Supply of Electricity ­ pure public interventions to public–private arrangements and pure private arrangements, and “assess the credibility and realism of proposed Government strategies while developing interventions” (World Bank 2004, 4). It also “cautions against prescriptive, one-size-fits-all recommendations” ­ (World Bank 2004, 1). 2006 board discussion paper More than ten years after the publication of the 1993 policy paper, one of its authors, John Besant-Jones (2006), compiled “lessons learned” from reforms’ implementation in developing ­ countries. His paper served as a follow-up to the 2004 guidance ­note. Entitled “Reforming Power Markets in Developing Countries: What Have We Learned?” the paper emphasizes that a one-size-fits-all approach was “ruled out by the extensive range of economic and insti- tutional endowments of these countries” ­ (Besant-Jones 2006, 1). It highlights the importance of starting conditions forth). The paper recognizes that “the economic case for breaking up and country context (political, social, and so ­ (Box continued next page) What Do We Mean by Power Sector Reform? 43 BOX 1.1 The World Bank and three decades of power sector reforms (Continued) a vertically integrated utility rests on various factors” and may be worthwhile when the benefits exceed the costs ­ (Besant-Jones 2006, 3). It advocates intermediate reform options (such as power-trade areas or the single-buyer model) for countries too small for a competitive power m ­ arket. The paper argues for a strategy for sequencing reforms as being less risky and more sustainable, though it recognizes other ­ a pproaches. It reasons that most developing countries whose power sector reforms had made substantial progress followed a logical sequence and “passed primary legislation for power market reform, established sector regulation, transacted with [independent power producers], and privatized some of the power supply industry” ­ (Besant-Jones 2006, 111). 2016 Independent Evaluation Group evaluation More than two decades after the landmark 1993 policy, the emphasis came to rest explicitly on the context and timing economy. The Independent Evaluation Group’s 2016 evaluation of World Bank of reforms, particularly the political ­ interventions in the power sector emphasizes the need to address the political economy of reforms by “aligning pro- gram timelines with government reform programs” (World Bank 2016, 23) and matching the “scale of the [World Bank ­ World Bank 2016, 24). It cautions against “overambitious Group] support to the scope of reforms and political risk” ( agendas and excessive conditionalities” ­(World Bank 2016, 25). As a result, the model did not pay heed to the figure into the 1990s paradigm of power sec- ­ policy objectives most critical to developing tor ­reform. ­ countries—namely, meeting growing demand More recently, the power sector has faced for electricity and completing the electrifica- profound technological c ­ hanges. The plum- tion ­process. meting cost of solar photovoltaic power and The policy objectives of the power sector new developments in battery storage, com- have since evolved and expanded to encom- bined with digitized power grids, are creating pass environmental and social ­ c oncerns. possibilities for the decentralization of energy Climate change emerged as a major global services, with power provided by a wider vari- concern in the 2000s, with the concomitant ety of a­ ctors. These technological trends are need to decarbonize the energy ­ sector. This disrupting frontier markets where some are need led to extensive public policy interven- even calling into question the need for a tradi- tions promoting the scale-up of new utility. Again, these technol- tional, centralized ­ ­ r enewable energy technologies, while also ogies were absent from the 1990s reform providing stronger incentives for energy ­landscape. e fficiency. The climate change debate also ­ In light of the historical evidence and likely surfaced the need to balance environmental future trends, this study aims to revisit, refresh, action with social ­objectives. This imperative and update the thinking on power sector was enshrined in 2015 as Sustainable reform in developing c ­ ountries. The prescrip- Development Goal (SDG) 7, committing tions of the 1990s reform model derived international financial institutions to univer- ­ primarily from economic theory and p ­ rinciples. sal access to energy that is modern, afford- A quarter century of experience with applica- able, reliable, and s ­ ustainable. Yet some tions of the model allows us to reevaluate this 1 billion of the world’s people continue to ­ approach. The case for reevaluation hinges live without access to electricity (IEA and both on the practical difficulties encountered others 2018), and many more make do with in the developing world and on the significant inadequate and unreliable ­ supply. These key changes in policy objectives and technological environmental and social objectives did not opportunities that have unfolded during the 44 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD intervening p ­ eriod. At the same time, the companies serving local ­ systems. As demand emergence of disruptive technologies raises for the new service increased, the need arose questions as to how the recommendations of for larger, integrated supply systems to capture the 1990s model might need to be adapted economies of scale and scope; costs and prices going ­forward. declined. Many governments came to consider ­ This study presents the findings of a the entire sector as a natural monopoly, research exercise that documents and evalu- whereby integration would minimize the costs ates power sector reforms across the develop- of coordination between supply chain func- ing ­w orld. Over a three-year period, the tions and finance (IEA 1999). States could also Rethinking Power Sector Reform initiative capture economies of scale by funding large took stock of accumulated knowledge, compil- projects whose high capital costs were less eas- ing the latest research evidence and data on ily financed by private ­investors. State control global power sector reform ­ trends.1 In particu- in the sector was thus justified on grounds of lar, deep-dive case studies across 15 develop- economic efficiency, in addition to the public ing countries examine both the political policy objectives of consumer welfare, national dynamics and technical evidence regarding the security, and industrial growth (Besant-Jones process of power sector reform and its implica- 2006; Brown and Mobarak 2009). To avoid a tions for sector p ­ erformance. This introductory monopoly’s negative outcomes, such as exces- chapter provides a brief history of power sector sive profits, solutions included public owner- reform, outlines the 1990s model, and sketches ship and r­ egulation. This second phase contin- out the theory of change used to analyze and ued through the 20th ­ century. evaluate the impacts of r ­ eform. Finally, it artic- From the 1940s to the 1960s, developing ulates the nature of technological disruption in countries established state-owned monopoly the sector and its potential relevance for future utilities in a wave of consolidation and ­reforms. ­ nationalization. These efforts received exter- nal support, including from the World ­ Bank.3 A BRIEF HISTORY OF POWER Public monopolies in the power sector were SECTOR REFORM considered “generally satisfactory in most From the 1870s to 1920s, electric power ser- developing countries, in an environment of vices were largely unregulated, elite, and low inflation and low debt levels, and with ­private.2 This period is regarded as the first of governments allowing utilities a significant four institutionally normative phases in the degree of managerial autonomy” (World power sector (Bhattacharyya 2011). In most Bank 1993, 34). More broadly, public invest- countries, supply began as a fragmented mar- ment in infrastructure and management of ket of local power providers owned by decen- markets for economic stability were consis- tralized private companies or municipal tent with the Keynesian economics that governments (Besant-Jones 2006). Beyond ­ dominated many developed countries from those benefiting from public street lighting, the 1940s to 1970s (Jahan, Mahmud, and users of electricity were mostly private—firms Papageorgiou 2014). In parallel, public and privileged ­ households. monopolies aligned well with the socialist and From the 1920s onward, as electricity nationalist ideologies that prevailed in many became a mass public good, governments took newly independent developing ­ c ountries. increasing control over the sector, and utilities The United States’ own model of regulated, grew from oligopolies to ­ monopolies. The elec- ­ i nvestor-owned monopoly utilities p rivate-​ tricity sector started out in the form of private was “widely admired and exported abroad in What Do We Mean by Power Sector Reform? 45 postwar years, though few developing Market-oriented power sector reforms countries had the capacity to duplicate the ­ began as experiments based on economic the- ­ public-private checks and balances inherent leaders. Beginning in ory, taken up by political ­ in the American system” (Williams and 1978, Chile was the first to pursue comprehen- Dubash 2004). sive market reforms in its power sector (Bacon By the 1970s and 1980s, however, various 1995), fusing elements from existing arrange- economic and political factors came together to ments in Belgium, France, and the United trigger a shift away from the paradigm of state Kingdom (Pollitt 2004). The ideological foun- ­ control. Countries exhausted economies of dations, however, can be traced to the United scale and scope in the power sector, depending States, in particular, to Milton Friedman and on the fuel and technology used and the legacy Friedrich Hayek at the University of C ­ hicago. of prior policies (Victor and Heller 2007). The A generation of Chileans known as the 1970s oil crises made countries aware of their “Chicago Boys” studied economics there in the vulnerability to fuel i ­mports. This awareness 1950­ s. When Augusto Pinochet came to power contributed to growing consciousness of the in Chile in 1975, he empowered newly benefits of energy conservation, especially in appointed officials from this group to pursue a the United States, where experience with “revolutionary market society,” including in nuclear power eroded trust in u ­ tilities. The oil the power sector (Clarke 2017). By the late crises contributed to global economic recession; 1960s, Friedman had asserted the success of Latin America’s debt crisis in the 1980s also free market ideas as a counterrevolution to played a ­role. Indebted countries turned to the Keynesian ­ economics. The influence of these International Monetary Fund and other for- ideas reached new heights when Prime eign sources for f ­inance. At the same time, Minister Margaret Thatcher and President technological innovation opened up new Ronald Reagan took office in 1979 and 1981, ­ possibilities. Newly developed combined cycle ­ respectively. Thatcher’s pursuit of economy-​ gas turbine (CCGT) power plants were more wide market reforms included enacting full efficient, at a smaller scale, than fossil fuel reform of the ­ U.K. power sector by 1989, with power plants, dramatically reducing the capital a wholesale market of unprecedented com- requirements as well as the marginal cost of plexity (Erdogdu 2014). In the United States, power generated from new ­ plants. Advances in the 1978 Public Utilities Regulatory Policy Act information and communication technology allowed relatively efficient independent power made it easier to coordinate grid operation and producers to serve the grid and so conserve integrate independent ­ plants. These gains, energy in response to the oil ­crises. By facilitat- however, were offset by several ­ challenges. ing competition with incumbent monopolies, Average costs of power generation reflected however, the act paved the way for subsequent sunk capital investment in aging assets that broader restructuring in line with the liberal became uneconomic before the end of their economic agenda of President Reagan and his expected l ­ife. As high-income countries’ successor, President George ­ W. ­ H. ­ Bush. demand growth slowed, their power compa- By the 1990s, market-oriented reforms in nies expanded business to developing country the power sector had crystalized into a global markets. Subsequent poorly managed interna- ­ norm, albeit with alternative f ­ormulations. tional capital flows contributed to the 1997 The market-oriented experiences of early Asian financial crisis, prompting further sup- reformers involved a loose set of ­ ideas. The port from the International Monetary Fund reform ­ literature assembled them into what and the World ­ Bank. was s ­ ubsequently called a “blueprint for 46 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD action” (Bacon 1995, 124), a “standard ­ ecades. energy sector likely to be sustained for d model” (Littlechild 2001, 1), a “standard Concurrently, technology is rapidly advancing prescription” (Hunt 2002, 8), or “textbook ­ in solar photovoltaics, wind turbines, energy architecture” (Joskow 2008, 11) of sector storage, microgrids, distributed resources, elec- norms. Other characterizations in the litera- ­ tric vehicles, and related information and com- ture are normative, including those in munication ­technologies. Conway and Nicoletti (2006), EBRD (2010), and ESMAP (1999). The core elements of the THE 1990s POWER SECTOR 1990s reform model (described in ­ f urther REFORM MODEL detail in the next section) entailed restructur- Power sector reforms in the 1990s were ing the incumbent utility; creating an inde- driven by a range of internal and external fac- pendent regulatory entity; introducing private tors that varied from country to country sector ownership (or at least commercial ori- (Bacon 1995; Bacon and Besant-Jones 2001; entation); and opening to competition where ESMAP 1999; Jamasb, Nepal, and Timilsina ­ relevant. This full suite of structural reforms 2015; Joskow 2008; Wamukonya 2003). was further underpinned by pricing reforms In developing countries, the principal driver aimed at achieving cost-reflective tariffs, to be of reform was the poor performance of SOEs guaranteed by the regulatory entity, and upon and the associated fiscal consequences (Bacon which private sector investment and market 2018). Developing countries faced surging competition were essentially ­ premised. demand for electricity, as well as low levels of Nevertheless, by the 2000s, government ­ electrification. Marked by high costs and low intervention in the power sector reemerged operational efficiency, SOEs proved unequal to amid evolving policy concerns and observa- demand, leading to a yawning supply–demand tions about the perceived limitations of deficit, static electrification, and widespread market-oriented r ­ ­ eforms. This renewed inter- blackouts and ­ brownouts. Subsidies to the sec- vention may be considered the fourth phase of tor mushroomed as governments held tariffs power sector reform, with a “new debate ­ ... below cost-recovery levels to avoid politically about the need for intervention in the market” unpopular i ­ncreases. The mounting fiscal (Bhattacharyya 2011, 720). One concern burden starved the sector of the capital needed ­ involved the security of supply, including in to invest in generation capacity and grid high-income countries such as the United extension. To put this situation in perspective, ­ Kingdom, as old plants reached the end of their the quasi-fiscal deficit of the power sector as a l ife. California’s 2000–01 electricity crisis ­ result of inefficient operations and underpric- revealed striking failures of both market and ing of electricity became macroeconomically sector regulation (Hunt 2002). Although these significant; it has been estimated to have a failures are not intrinsic to model reforms, the median value of about 1 percent of gross crisis nevertheless contributed to broad con- domestic product across a range of developing cern, first, over the influence of private actors countries in Africa and the Middle East in the sector and, second, about the global (Alleyne 2013; Briceño-Garmendia, Smits, and slowdown in the pace of subsequent market Foster 2009; Camos and others 2017; Trimble reforms (Bhattacharyya 2011). More recently, and others 2016). Governments’ willingness to international interest in universal access and permit the worsening of financial performance clean energy has reached new heights, as provided no incentive for utilities to cut costs reflected in the SDGs and the Paris ­Agreement. and improve ­ efficiency. The further worsening These accords set a global policy agenda for the of the fiscal situation through an external cost What Do We Mean by Power Sector Reform? 47 shock, such as an oil price hike or currency consequence of energy deregulation in devaluation, sometimes provided the final ­ member countries of the OECD was the cre- impetus for r­ eform. In other cases, the immedi- multinational. ation of a new breed of energy ­ ate catalysts were broader debt crises and ensu- This new kind of multinational would eventu- ing power sector reforms as a precondition for ally go looking for investment opportunities access to multilateral ­ finance. abroad as power sector reform took root in the In transition economies, power sector emerging ­economies. reform was part of a broader effort to create a The 1990s power sector reform model com- market economy (Bacon 2018). The sector prises a package of four structural reforms faced different challenges in these countries, (Foster and others 2017): which tended to have excess generation capac- • The first of the structural actions is regula- ity and high rates of ­ access. Nevertheless, they tion, which requires an autonomous regula- still charged prices far below replacement costs tory entity able both to provide a degree of and struggled to maintain their aging political independence and to hold utilities ­ infrastructure. Over and above these sector accountable for their operational and finan- issues was the general reform strategy of creat- cial ­performance. ing a market economy based on private owner- • The second is restructuring, which involves ship and ­ competition. A further incentive was steps toward the eventual full vertical and provided by the desire of certain countries to horizontal unbundling of the incumbent join the European Union, membership in state-owned ­monopoly. which would require them to bring the power • The third is private sector participation, which sector in line with European Union ­ directives. brings private management and capital into In developed countries, power sector the sector to boost operational efficiency reform was seen as a means to reduce prices and investment (which often involved a for consumers while raising proceeds for the preceding step of corporatizing ­ SOEs). national treasury (Bacon 2018). Developed • The fourth element is competition, which countries did not typically suffer from many of initially allows generators to compete to the severe operational and financial challenges supply a monopoly utility and eventu- facing the power sector ­ elsewhere. ally allows customers to negotiate their Accordingly, the motivation for reform was supply contracts directly with power pro- different. The advent of new technologies, par- ­ ducers and traders supported by a power ticularly CCGT, that could lower generation exchange; in some countries, retail com- costs and did not require very large units in petition for small customers was facilitated order to reap the benefits of economies of through suppliers of alternative ­ energy. scale, reduced the strength of the case for allowing a vertically integrated m ­ onopoly. Regulatory reform is defined as the estab- Without economies of scale, plants and firms lishment of an autonomous entity with could be smaller, opening the field for multiple responsibility for regulatory oversight and with players and competition, once private capital some role in decision making (Foster and could enter the i ­ndustry. The hope was that o thers 2017). The power sector provides ­ competition would deliver lower prices, better ­ policy-making, regulatory, and service provi- quality, and greater choice for consumers, sion ­functions. Policy making charts the sec- while the associated ­ privatization of state- direction. Regulation oversees tor’s strategic ­ owned assets would strengthen the financial the sector to ensure that it follows and enforces position of the ­ s tate. An unintended the strategic direction, while service provision 48 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD is the implementation of that strategic removing any conflict of interest that may ­ d irection. Traditionally, all three functions arise when a single utility has more than one have been combined within the line ministry; function along the electricity supply ­ chain. however, with widespread corporatization of For example, a transmission company that the service provider to form an enterprise dis- also engages in generation may have the tinct from the line ministry, as well as the incentive to prioritize grid access for its own growing delegation of these activities to the generation capacity as opposed to that of private sector, the need for a clearer regulatory ­ competitors. It is relevant to distinguish partial function has been f ­elt. A widely adopted vertical unbundling—in which, for example, model, drawing heavily on experience in the generation is broken out but transmission and United Kingdom and the United States, has distribution remain joined—from full vertical been to create an entity dedicated to r­ egulation. unbundling, whereby separate entities under- Independence is considered i ­mportant. It iso- take all three segments of the electricity sup- lates the regulator (and, ultimately, the service ply ­ c hain. Second, horizontal unbundling provider) from short-term, opportunistic aims at diluting market power, which is par- political interference, and provides balance ­ generation. For example, ticularly relevant for ­ between the investor’s right to a fair return on a country with five or six similar generation capital and the consumer’s right to ­ v alue. companies will likely experience stronger Independence, although always relative, has competitive pressure than a country with only been defined in terms of an institutional exis- two companies—a large one and a small o ­ ne. tence, governance structure, and budget line Although full unbundling entails separate that are separate from the line ministry ­ itself. ownership of the different entities created in In practice, however, genuine independence the process, in practice unbundling may pro- has proved difficult in many political systems ceed gradually—beginning with defining the and c ­ ultures. Governments are generally reluc- distinct management units and then separat- tant to relinquish their discretionary powers ing accounts and constituting separate entities over political patronage such as electricity tar- under distinct ­ ownership. iffs, power sector investment plans, and utility Competition is defined as the coexistence of employment. The roles of the regulator include ­ multiple service providers in the same market setting tariffs to recover efficient costs, moni- (Foster and others 2017). Competition among toring and enforcing service standards, and service providers promotes efficiency and overseeing market ­ entry. innovation. When multiple companies com- ­ Restructuring reform is defined as move- pete head to head for consumers, a market dis- ment along a spectrum toward full vertical cipline emerges, along with pressure to keep and horizontal unbundling of the sector costs down to efficient levels and to improve (Foster and others 2017). The starting point service ­ quality. The large economies of scale for restructuring reform is typically a vertically in the power sector mean that key activities integrated national monopoly utility, and its (for example, transmission) are traditionally theoretical endpoint is a fully restructured ­ c onsidered natural monopolies, making it sector, with restructuring entailing both the ­ inefficient to have more than one ­ ­ supplier. vertical unbundling of generation, transmis- Even under a natural monopoly, however, it is sion, and distribution, and the horizontal still possible to have different companies com- unbundling of the generation and distribution pete for the right to supply the market on a tiers to create multiple companies operating in monopoly basis for a certain period of t ­ime. p arallel. Such a process paves the way for ­ The liberalization of the power sector therefore competition, according to the theoretical often proceeds in incremental stages, begin- ­ rationale. First, vertical unbundling aims at ning with the opening up of generation to What Do We Mean by Power Sector Reform? 49 independent power producers that compete for participation may range from, for example, a the m­ arket. Eventually, it may transition to a management contract, where the private sec- full single-buyer model where generation is tor has neither responsibility for investment fully divested from the incumbent utility, with nor any exposure to commercial risk; to a the latter acting as the single buyer of genera- concession, where the private sector has time- ­ tion on behalf of end c ­ onsumers. The next bound responsibility for investment and expo- stage—once the transmission segment has sure to commercial risk; to a divestiture, where been fully unbundled—is to allow third-party the private sector permanently takes over access to the power grid so large customers can all responsibilities and risks associated with purchase power directly from generators on a ­ upply. Although the 1990s reform electricity s bilateral negotiated ­basis. In due course, it may model encouraged private sector participation evolve into a wholesale power market, with a in generation and distribution, no concomitant centralized price-setting mechanism and recommendation existed for the transmission a variety of contracts and products being segment, whose strong natural monopoly exchanged. In some instances, a final step ­ characteristics and strategic character seemed would unbundle the distribution and retail to justify continued public ­ownership. functions of the utility, allowing the latter to be It is important to note that the various open to competition for energy ­ supply. measures constituting the 1990s power sector ­ Private sector participation is defined as the reform model were seen as mutually support- introduction of private sector management and ive and intended for implementation as a investment, whether through temporary con- ­ package (Bacon 2018). Some policies have a tractual arrangements or permanent asset sales direct effect, whereas others act as facilitators, (Foster and others 2017). The 1990s power without which the direct policies cannot be sector reform model presupposed that the effective. For example, privatizing the SOE ­ power utility had already been corporatized— without prior restructuring would simply cre- that is, separated from a ministerial depart- ate a private monopoly in lieu of a public o ­ ne. ment and constituted as a free-standing SOE In the absence of regulation and competition, operating with a commercial orientation and private monopolies tend to use their monopoly governed by company l ­aw. Thereafter, private power to maximize profits by restricting output sector participation can be implemented to via excessive ­ prices. Hence, they absorb much varying degrees along a number of d ­ imensions. of the gain from efficiency improvement First, the scope of participation varies according and produce little benefit for consumers even to the extent of the electricity supply chain though production costs have been c ­ ut. affected. Private sector participation may ini- ­ Competition serves to discipline abusive behav- tially be undertaken in one segment of the ior and is made possible by a prior unbundling supply chain, often generation, but not neces- of the sector; for natural monopolies, regula- sarily in another, such as ­ distribution. Second, tion can redistribute the gains between con- it may affect some, but not all, companies in a sumers and the ­ producer. Sector unbundling particular supply ­ segment. For example, a without privatization simply multiplies the country may privatize some of its generation number of SOEs in the sector without funda- plants, but leave the others under public mentally affecting their ­performance. Or again, ­o wnership. The coverage of private sector regulation without privatization essentially ­ participation may be gauged for generation duplicates the control of whichever govern- according to the percentage of capacity under ment department was responsible for the private control, and for distribution by the per- actions of the ­SOE. In short, all the elements of centage of distribution companies under pri- the reform package are needed for the full per- vate ­control. Third, the depth of private sector formance impacts to ­ materialize. 50 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD A THEORY OF CHANGE adoption of the 1990s reform package was This study uses the theory of change under- expected to improve cost recovery and opera- pinning the 1990s reform model as a concep- tional e ­ fficiency. Over time, the utility tual framework for evaluating the model’s becomes more financially viable and permits efficacy. This theory of change proposes that ­ greater ­investment. These intermediate out- reforms would lead to beneficial behavior comes would then feed through into greater change among the key sector actors, resulting security of ­ supply. Security of supply was the in improved sector ­ p erformance. Behavior main outcome envisaged in the 1990s para- changes when private management is digm (figure 1.1). ­ i ntroduced. Private management reorients More recently, however, the range of out- enterprises from bureaucratic and political comes sought for the sector has been incentives toward profit seeking, cost control, expanded to include social and environmen- and customer ­ orientation. Potential for pri- benefits. These outcomes are enshrined in tal ­ vate management abuses would be disciplined SDG7 and comprise universal access to elec- either by market pressures in competitive seg- tricity as well as renewable energy and ments of the supply chain or by regulatory energy efficiency, both important contribu- incentives in natural monopoly segments tors to a clean and low-carbon energy future, of the supply ­chain. The presence of the regu- helping to reduce global warming and lator, as well as the engagement of the private improve local air ­ q uality. These objectives sector, would meanwhile prevent opportunis- were never explicitly addressed by the 1990s tic interference in the day-to-day operation of model; however, to the extent that they rely the sector for political e ­ nds. Thus, overall, the on increased investment for their FIGURE 1.1 1990s model sector reforms: Inferred simple theory of change Reforms Intermediate outcomes Outcomes Liberalized, regulated markets for Post 2000s wholesale and retail electricity outcomes • 1st best: Competitive market • Access and Cost recovery • 2nd best: Regulated by independent agency a ordability • Reasonable return • Environmental on new investment Attracts sustainability Cost recovery prices • Cost recovery in ongoing operations Least-cost Sector outcomes investment Enables • Security of supply Cost optimization/ e cient operation • Generation and quality of • Transmission service Incentivizes E ciency • Distribution • Fiscal a ordability • E cient operations Economically e cient • Optimal resource Guides decision allocation Commercially operating utilities • 1st best: Restructured and privatized utilities • 2nd best: Restructured, corporatized SOEs as o -takers of private investment elaboration. Source: World Bank ­ enterprise. Note: SOE = state-owned ­ What Do We Mean by Power Sector Reform? 51 achievement, they could (at least in allowing cross-country comparisons to be illus- principle) benefit from the kinds of reforms ­ trative of wider relationships between power supported by the 1990s ­ model. On the one sector measures and outcomes (box 1.2). hand, it is unfair to judge the performance of Because power sector reform is but a the 1990s model against environmental and means to an end, the important thing is social policy objectives it was never explicitly whether countries are able to achieve good designed to ­ a chieve. On the other hand, performance ­ outcomes. As noted, the 1990s given the importance of these policy objec- power sector reform model is motivated by tives, it is relevant to ask if the 1990s model the desire to obtain good sector performance was incidentally helpful in advancing these outcomes; it offers one particular theory of newer ­agendas. change regarding how these outcomes can be This study uses a hybrid methodology— ­ reached. This report is based on the premise quantitative cross-country analysis combined that the implementation of the 1990s sector with in-depth qualitative country case ­ studies. reform model—or, for that matter, of any Much of the academic literature on power sec- other institutional model for the power tor reform focuses on econometric analysis of sector—is valuable insofar as it improves per- ­ large sample cross-country panel data (Erdogdu formance ­ outcomes. This report will focus on 2011; Nepal and Jamasb 2012; Sen, Nepal, and evaluating the degree to which countries Jamasb 2016; Zhang, Parker, and Kirkpatrick improved their performance outcomes in their 2005), with a view to estimating whether vari- power sectors and understanding the extent ous reforms affect sector and economic perfor- to which the reform measures may have mance outcomes (Bacon 2018). Although ­contributed. wide-ranging and precise, these studies are Power sector performance can be evaluated constrained by the availability of panel data for along three key dimensions: security of suitable indicators and cannot provide insights supply, social inclusion, and environmental ­ on the process and dynamics of r ­ eforms. Much ­ sustainability. For the power sector to fulfill its of the policy literature on power sector reform role as an enabler of economic activity, it must focuses instead on in-depth case studies of expand generation capacity to keep pace with country-level reforms (Newberry and Pollitt peak demand so consumers have a reliable 1997; Pineau 2009). Although those studies supply of e ­ lectricity. For the power sector to are rich in content, their narrow scope does fulfill its role as a platform for social inclusion, not allow for wider ­ generalizations. Following a country must attain universal electrification the approach of Vagliasindi and Besant-Jones and provide basic service that is affordable (2013), this study adopts a hybrid approach— across all population ­ tiers. For the power sector power sector reforms in 15 countries are to contribute to environmental sustainability, it studied in depth (with India providing three ­ must reduce the carbon intensity of electricity state-level case s ­ tudies). These case studies can generation and diminish the impact on local air then be both tapped for qualitative insights and ­quality. pooled to provide cross-country data ­ patterns. Whereas the 1990s reform model focused It is true, however, that a sample of 15 countries primarily on the economic dimension of the remains relatively small, and cross-­ country power sector, since 2010 the social and envi- analysis cannot be considered representative in ronmental dimensions of the power sector any statistical s­ ense. Nevertheless, the sample have grown in ­ significance. The three dimen- contains a range of geographies, income groups, sions of power sector reform described repre- political systems, and power sector conditions, sent a 21st-century understanding of the ­ sector. 52 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 1.2 The Rethinking Power Sector Reform Observatory: An introduction At the heart of this study is a Power Sector Reform Observatory, comprising data from 15 c ­ ountries. Although not representative in the statistical sense, this sample illustrates a range of country and power sector ­ settings. As illus- trated in table B1.2.1, the observatory countries are spread evenly across the six developing regions of the world and include a range of income levels, political systems, system scales, and degrees of power system d ­ evelopment. Furthermore, the sample was designed to include countries at an advanced reform stage (for example, Peru and the Philippines) in addition to countries that have made slower progress (for example, Tajikistan and T ­ anzania). It also aimed to include countries with strong and weak power sector ­ performances. Of particular interest were countries at an advanced stage of reform yet with relatively weak performance (for example, the Dominican Republic) and coun- tries with more modest reforms yet relatively strong performance (for example, ­ Vietnam). In each of the 15 observatory countries, a local consultant was hired to undertake an ambitious data collection exercise. The exercise detailed 90 quantitative performance indicators for the power system overall and for the one ­ or two major distribution utilities in that jurisdiction; a questionnaire seeking qualitative data across 460 categorical variables produced an exhaustive description of power sector reform measures affecting policy, planning, regulation, ­ overnance. A series of in-depth open-ended interviews with some 20 key stakeholders documented their and utility g experiences and perspectives on the country’s power sector reform j ­ourney. Following the collection phase, data were validated by the project team with input from the World Bank energy team for each c ­ ountry. The information collected across all 15 countries was consolidated into a single cross-country power sector d ­ atabase. Quantitative data were cleaned and analyzed according to a consistent r ­ ubric. Qualitative data were classified between descriptive variables and those that captured normative features of the institutional framework. Researchers developed a series of normative indexes to aggregate this wealth of information into a more ­ tractable form; these indexes cover good practices in power system planning, power sector regulation, and utility ­ governance. The indexes will be introduced later, in their corresponding ­ chapters. Interview transcripts were also compared across countries and summarized according to a qualitative framework seeking to identify any differences and ­commonalities. TABLE B1.2.1 Power Sector Reform Observatory countries and selected characteristics Political Installed Electrification Country Region Income group systema capacity (GW)b rate (%)c Colombia Latin America and Caribbean Upper middle Multipolar 17.3 99 Dominican Latin America and Caribbean Upper middle Multipolar 3.8 100 Republic Egypt, Arab Middle East and North Africa Lower middle Unipolar 42.1 100 ­Rep. India South Asia Lower middle Multipolar 291.3 85 Kenya Sub-Saharan Africa Lower middle Transition 2.4 75d Morocco Middle East and North Africa Lower middle Unipolar 7.8 100 Pakistan South Asia Lower middle Transition 29.3 72d Peru Latin America and Caribbean Upper middle Unipolar 12.2 95 Philippines East Asia and Pacific Lower middle Transition 21.7 91 Senegal Sub-Saharan Africa Lower middle Transition 0.9 65 Tajikistan Europe and Central Asia Lower middle Unipolar 5.4 100 Tanzania Sub-Saharan Africa Low Transition 1.5 33 Uganda Sub-Saharan Africa Low Unipolar 0.9 27 Ukraine Europe and Central Asia Lower middle Multipolar 49.9 100 Vietnam East Asia and Pacific Lower middle Unipolar 37.9 100 Project. Source: World Bank elaboration based on data collected for Rethinking Power Sector Reform ­ Note: GW = ­ gigawatt. implementation. a. At time of reform ­ ­ apacity. b. Platts database up to 2017 for operational installed c c. Access data from IEA, IRENA, UN Statistics Division, WBG, and WHO 2018. ­ d. Based on individual country ­statistics. What Do We Mean by Power Sector Reform? 53 They are particularly informed by recent politi- have slightly affected institutions—for exam- cal agreements such as SDG7, which articulates ple, modular CCGTs made the unbundling and a target for universal access to affordable elec- privatization of generation a bit easier, and pre- tricity by 2030 alongside substantial progress payment meters helped alleviate financial dis- toward a cleaner power s ­ector. The Paris tress along the electricity supply chain—but Climate Accord, too, calls for energy sector not much could really be described as truly dis- measures that cap global warming at 1.5 ruptive of ­institutions. It was against this rela- degrees C­ elsius. In the 1990s, the key concern tively stable technological backdrop that the was to achieve conditions necessary to support 1990s power sector reform model emerged, investment in new infrastructure aimed at developing under the implicit assumption that achieving security of ­ s upply. Because the technology was largely a ­ given. urgency of environmental and social concerns In contrast, the electricity sector in the 21st was not as evident in the 1990s when the century has found itself at the center of power sector reform model was developed, it momentous technological ­ change. Like cars would be anachronistic to evaluate the out- displacing horses, or cell phones displacing comes of power sector reforms against these landlines, or photography going digital, a new more recently articulated ­ goals. Nevertheless, tipping point—that is, the point at which a given their prominence, it is relevant to ask new technology becomes so inexpensive it is whether the 1990s power sector reform model nearly ubiquitous—may be ­ here. In the power might also play a role in advancing these sector, this means that almost all new invest- ­agendas. ments in generation will soon be renewable, with implications for grid operations, system AN UNCERTAIN FUTURE planning, and market d ­ esign. It means that The traditional power sector reform agenda battery storage will become increasingly eco- is already being affected by technological nomic and prevalent, breaking the long-­ ­ d isruptions sweeping through the ­ s ector. standing constraint that supply and demand Technological change—particularly when it is for electricity must match precisely and instan- rapid—fundamentally affects the cost struc- taneously at every m ­ oment. It means that ture of industries and may reshape a sector’s power grids are becoming increasingly digitized institutional ­ organization. The electricity sec- and intelligent, enabling more sophisticated tor is undergoing precisely such changes, coordination of supply, demand, and storage which are increasingly calling into question activities across a growing number of ­ actors. It many of the traditional paradigms and reshap- means electricity consumers in many countries ing both the generation and distribution ­ tiers. are playing roles that are much different from The power sector reform paradigm of the those they played in the ­ past. Exceeding the 1990s emerged during a period of relative tech- old passive consumption roles, they can engage nological ­ stability. The technical principles gov- in their own decentralized generation, sell sur- erning the electricity industry did not change plus energy or storage capacity back into the fundamentally during its first century of grid, participate in remunerated demand-­ ­ existence. For many years, technology change reduction measures, and even trade electricity in the power sector has been largely on the distribution ­ grid. incremental. Open cycle gas turbine plants ­ These new technologies also challenge gradually gave way to CCGT plants, meters got existing models of planning, regulation, and progressively smarter, and renewable energy institutional s ­ tructure. The integration of costs gradually f ­ell. A few technologies may renewables, storage, and consumer-sited 54 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD d ­ istributed energy resources has profound clear is that the technological disruption could implications for patterns of electricity con- have implications for the prescriptions of the sumption, approaches to power sector reform, 1990s reform model, affecting their degree of and pathways toward climate g ­ oals. When relevance and ease of ­ implementation. Each power no longer flows one way, but instead key chapter of this report ends with some pre- needs to be coordinated across multiple liminary reflections on how technological dis- consumer-sited locations on the distribution ­ ruption will affect the various dimensions of system, then regulatory models designed for reform covered by the ­ study. one-way power flow need to be reconsidered, This report is organized as ­ follows. Part I sector planning and grid operations become (chapters 1–3) sets the stage for power sector much more complicated, and traditional busi- reform. Chapter 2 outlines the uptake, diffu- ­ ness models are ­ c hallenged. As consumers sion, packaging, and sequencing of power sec- have the option to produce some of their own tor reforms around the ­ world. These reforms’ energy or reduce some of their own consump- underlying political economy dynamics are tion, they alter the total system peak demand, examined in chapter 3, which also considers and may even alter how the grid system is (or variations in ­ implementation. should be) ­ planned. Part II (chapters 4–7) examines the four Politicians and technocrats, policy makers building blocks of power sector reform. and system planners, investors and civil society, Chapter 4 looks at utilities and the extent to and financiers and regulators are all attempting which unbundling and governance reforms to understand the implications of that ­ change. were able to turn around operational perfor- They are trying to figure out what will happen mance and what strategies were adopted to investment decision making and to sector to address financial ­ s hortfalls. Chapter 5 institutions in the disruptive technology mael- investigates the role of the private sector in ­ strom, and that outcome may be unclear for determining the performance of the power some ­ time. As these new disruptive technolo- ­ sector. Chapter 6 considers the practice of regu- gies are adopted, there will be big winners and lation, evaluating the quality of regulatory gov- big losers, and the inevitable tensions between ernance and practice and the extent to which them will affect the likely pace at which new they have been successful in insulating the sec- technologies are a ­ dopted. Which energy future tor from political interference and helping utili- countries adopt has a lot to do with the choices ties achieve cost ­ recovery. Chapter 7 considers they have already made to enable electricity developing country experience with the cre- ­ production. Thus, policy makers and practi- ation of wholesale power ­ markets, clarifying tioners will have to balance their act delicately the conditions in which they may be helpful to adapt to these coming ­ changes. and elucidating a range of market design and Finally, these technological disruptions will transition ­issues. clearly have implications for the future applica- Part III (chapters 8–9) looks at the impact of tion of the 1990s reform m ­ odel. The techno- power sector reforms both on improving utility logical disruption is still at an early stage, and it performance and on enhancing outcomes for would be foolhardy to attempt to predict where ­ consumers. Chapter 8 examines the extent of it will ultimately l ­ead. Moreover, because of utilities’ progress toward cost recovery and large differences in starting conditions, policy operational e ­ fficiency. Intermediate outcomes objectives, and institutional capacities, the are seen to improve utilities’ financial viability direction taken by technological disruption and permit greater ­ investment. Finally, chapter could be expected to differ substantially across 9 evaluates the extent to which power sector developed and developing ­ countries. What is reforms have been able to deliver on the What Do We Mean by Power Sector Reform? 55 outcomes they were intended to performance ­ in the Middle East and North Africa: Insights from a ­achieve. Performance ­Diagnostic. Washington, DC: World ­Bank. Clarke, Timothy ­ D avid. 2017. “Rethinking NOTES Chile’s Chicago Boys: Neoliberal Technocrats 1. The 20 background papers for this project or Revolutionary Vanguard?” Third World are being gradually released and posted on Quarterly 38 (6): 1350–65. the Energy Sector Management Assistance Collier, H ­ ugh. 1984. Developing Electric Power: Thirty Program website, h ­ ttps://www.esmap.org​ Years of World Bank E ­ xperience. Washington, /­rethinking_power_sector_reform. DC: World ­ Bank. 2. This section is based on Lee and Usman (2018). Conway, Paul, and Giuseppe ­ Nicoletti. 2006. 3. See Kapur, Lewis, and Webb (1997) for a “Product Market Regulation in the Non- ­ fascinating account of the World Bank’s Manufacturing Sectors of OECD Countries: history, and Collier (1984) for its early, prom- ­ Measurement and ­ H ighlights.” Economics inent support of the power sector of develop- Department Working Paper N ­ o. 530, ing ­countries. Organisation for Economic Co-operation and Development, ­Paris. EBRD (European Bank for Reconstruction and REFERENCES ­Development). 2010. Transition Report 2010: Alleyne, ­ Trevor. 2013. Energy Subsidy Reform in Recovery and ­ Reform. London: ­ EBRD. Sub-Saharan Africa: Experiences and ­ Lessons. Erdogdu, E ­ rkan. 2011. “The Impact of Power Washington, DC: International Monetary ­ Fund. Market Reforms on Electricity Price-Cost Bacon, Robert ­ W. 1995. “Privatization and Reform Margins and Cross-Subsidy Levels: A Cross- in the Global Electricity Supply I ­ndustry.“ Country Panel Data A ­ nalysis.” Energy Policy Annual Review of Energy and the Environment 39 (3): 1080–92­ . 20 (1): 119–43­ . ———. 2014. “The Political Economy of Electricity ———. 2018. “Taking Stock of the Impact of Market Liberalisation: A Cross-Country Power Utility Reform in Developing Countries: ­Approach.” Energy Journal 35 (3): 91–128. ­ eview.” Policy Research Working A Literature R ESMAP (Energy Sector Management Assistance Paper 8460, World Bank, Washington, ­ DC. ­Program). 1999. Global Energy Sector Reform In Bacon, Robert ­ W., and John ­ Besant-Jones. 2001. Developing Countries: A ­ Scorecard. Washington, “Global Electric Power Reform, Privatization, DC: World ­ Bank. and Liberalization of the Electric Power Industry Foster, Vivien, Samantha Witte, Sudeshna Ghosh in Developing C ­ ountries.” Annual Review of Banerjee, and Alejandro M ­ oreno. 2017. Energy and the Environment 26 (1): 331–59. “Charting the Diffusion of Power Sector Reforms Besant-Jones, John ­ E. 2006. “Reforming Power Across the Developing ­ World.” Policy Research Markets in Developing Countries: What Have Working Paper 8235, Rethinking Power Sector We Learned?” Energy and Mining Sector Reform, World Bank, Washington, ­ DC.  Board Discussion Paper N ­ o. 19, World Bank, Gavin, Michael, and Dani R ­ odrik. 1995. “The Washington, ­DC. World Bank in Historical ­ P erspective.” Bhattacharyya, Subhes ­ C. 2011. Energy Economics: American Economic Review 85 (2): 329–34. Concepts, Issues, Markets, and G ­ overnance. Gratwick, Katharine N ­ ., and Anton E ­ berhard. London: ­Springer. 2008. “Demise of the Standard Model of Briceño-Garmendia, ­ K. Smits, and ­ C., ­ Foster. V. ­ Power Sector Reform and the Emergence 2009. Financing Public Infrastructure in Sub- of Hybrid Power M ­ arkets.” Energy Policy Saharan Africa: Patterns and Emerging I ­ssues. 36 (10): 3948–60. Washington DC: World Bank ­ Group. Hunt, S ­ ally. 2002. Making Competition Work in Brown, David ­ S., and Ahmed ­ Mobarak. 2009. ­Electricity. New York: ­ Wiley. “The Transforming Power of Democracy: IEA (International Energy A ­ gency). 1999. Regime Type and the Distribution of Electricity Market Reform: An IEA H ­ andbook. ­Electricity.” American Political Science Review Paris: Organisation for Economic Co-operation 103 (2): 193–213. and ­Development. Camos, ­ R. Bacon, ­ D., ­ A. Estache, and ­ M. ­ M. IEA (International Energy Agency), IRENA ­Hamid. 2017. Shedding Light on Electricity Utilities (International Renewable Energy Agency), 56 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD UN (United Nations) Statistics Division, WBG OIES Paper: EL 18, Oxford Institute for Energy (World Bank Group), and WHO (World Studies, Oxford, ­ UK. Health O ­ rganization). 2018. Tracking SDG7: Tr i m b l e , C h r i s t o p h e r P h i l i p , M a s a m i The Energy Progress ­ Report. Washington, DC: Kojima, Ines Perez Arroyo, and Farah World Bank ­ Group. Mohammadzadeh. 2016. “Financial Viability Jahan, Sarwat, Ahmed Saber Mahmud, and of Electricity Sectors in Sub-Saharan Africa: Chris P­ apageorgiou. 2014. “What Is Keynesian Quasi-Fiscal Deficits and Hidden ­ C osts.” Economics?” Finance and Development 51 (3): Policy Research Working Paper 7788, World 53–54. Bank, Washington, ­ DC. Jamasb, Tooraj, Rabindra Nepal, and Govinda R ­. Vagliasindi, Maria, and John ­ B esant-Jones. Timilsina. 2015. “A Quarter-Century Effort ­ 2013. Power Market Structure: Revisiting Policy Yet to Come of Age: A Survey of Power Sector ­Options. Directions in Development S ­ eries. Reforms in Developing C ­ ountries.” Policy Washington, DC: World ­ Bank. Research Working Paper 7330, World Bank, Victor, David ­ G., and Thomas ­ C. ­Heller. 2007. Washington, ­DC. The Political Economy of Power Sector Reform: The Jayarajah, Carl, and William B ­ ranson. 1995. Structural Experiences of Five Major Developing C ­ ountries. and Sectoral Adjustment: World Bank Experience, New York: Cambridge University ­ Press. 1980–92. A World Bank operations evaluation Wamukonya, ­ Njeri. 2003. “Power Sector Reform ­ study. Washington, DC: World Bank ­ Group. in Developing Countries: Mismatched Joskow, P ­. ­L. 2008. “Lessons Learned from the ­Agendas.” Energy Policy 31 (12): 1273–89. Electricity Market ­ L iberalization.” Energy Williams, James, and Navroz ­ K. ­Dubash. 2004. Journal 29 (special issue): 9–42. “Asian Electricity Reform in Historical Kapur, Devesh, John P ­ . Lewis, and Richard ­Perspective.” Pacific Affairs 77 (3): 411–36. ­C. ­Webb. 1997. The World Bank: Its First Half Williams, James, and Rebecca ­ G hanadan. ­Century. Volumes 1 and 2. Washington, DC: 2006. “Electricity Reform in Developing and World ­Bank. Transition Countries: A R ­ eappraisal.” Energy Lee, Alan David, and Zainab ­ Usman. 2018. “Taking 31 (6–7): 815–44. Stock of the Political Economy of Power Sector World ­ B ank. 1993. The World Bank’s Role in Reforms in Developing Countries: A Literature the Electric Power Sector: Policies for Effective ­ Review.” Policy Research Working Paper 8518, Institutional, Regulatory, and Financial R ­ eform. Rethinking Power Sector Reform, World Bank, Washington, DC: World ­ Bank. Washington, ­DC. ———. 1994. World Development Report 1994: Littlechild, S ­. ­ C. 2001. “Electricity: Regulatory Infrastructure and ­ D evelopment . New York: Developments from Around the ­ World.” Oxford University ­ Press. Beesley Lectures on Regulation series XI, ———. 2003. “Private Sector Development in Institute of Economic Affairs/London Business the Electric Power Sector: A Joint OED/OEG/ School, London, October 9. OEU Review of the World Bank Group’s Nepal, Rabindra, and Tooraj ­ J amasb. 2012. Assistance in the 1990­ s .” World Bank, “Reforming the Power Sector in Transition: Washington, ­DC. Do Institutions Matter?” Energy Economics 34 ———. 2004. “Public and Private Sector Roles in (5): 1675–82. the Supply of Electricity S ­ ervices.” Operational Newberry, David ­ M ., and Michael ­ G. ­ P ollitt. Guidance for World Bank Group Staff, 1997. “The Restructuring and Privatisation of Energy and Water Department, World Bank, Britain’s CEGB—Was It Worth It?” Journal of Washington, ­DC. Industrial Economics 45 (3): 269–303. ———. 2016. “Financial Viability of the Pineau, P ­ ierre-Olivier. 2009. “Electricity Sector Electricity Sector in Developing Countries: Reform in Cameroon: Is Privatization the Recent Trends and Effectivenness of Solution?” Energy Policy 30 (11–12): 2249–61. World Bank ­ I ntervention.” World Bank, Pollitt, Michael G ­ . 2004. “Electricity Reform in Washington, ­DC. Chile: Lessons for Developing C ­ ountries.” Zhang, Yinfang, David Parker, and Colin Competition and Regulation in Network Industries ­ Kirkpatrick. 2005. “Competition, Regulation, 5 (3–4): 221–62. and Privatization of Electricity Generation in Sen, Anupama, Rabindra Nepal, and Tooraj Developing Countries: Does the Sequencing Jamasb. 2016. “Reforming Electricity Reforms? ­ of the Reforms Matter?” Quarterly Review of Empirical Evidence from Asian ­ Economies.” Economics and Finance 45 (2–3): 358–79. How Far Did Power Sector Reform Spread in the Developing World? 2 Guiding Questions • What factors have affected the uptake of power sector reform over time? • Was power sector reform implemented as originally envisaged? Summary • The uptake of power sector reform in the developing world lags well behind that in the member countries of the Organisation for Economic Co-operation and Development. • Following the rapid diffusion of power sector reform from 1995 to 2005, uptake notably slowed in the subsequent decade, 2005–15. • A country’s geographical, economic, and power system characteristics strongly affect the uptake of reform. • Power sector reform packages show clear signs of cherry-picking of recommended measures: easier ones were implemented more frequently than more difficult ones. • Many countries are settled at an intermediate stage of power sector reform. • Power sector reforms were often packaged and sequenced in ways unrelated to the original reform logic. This chapter examines how extensively researchers to characterize how the reform developing countries adopted the policy pre- ­ measures were sequenced and packaged. The scriptions of the 1990s power sector reform guiding questions for this chapter are the fol- model.1 Building on a global survey of reform lowing: How rapidly did the 1990s model of measures and their diffusion and uptake, the power sector reform spread globally, and has chapter tracks the model’s four components: the initial momentum been sustained? Did regulation, restructuring, private sector partic- the model adopted in developing countries go ipation (PSP), and competition. Considering as far it did in developed countries? Were each of these measures separately allows reforms sequenced and packaged in line with 57 58 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD the original policy recommendations? Finally, evidence regarding global patterns of power how was the adoption of power sector reforms sector reform. Six main findings are described influenced by the particular characteristics of below. countries’ economies and power sectors? A Power Sector Reform Index (PSRI) was Finding #1: The uptake of power developed as part of this study to measure the sector reform in the developing world extent to which countries adopted the four lags substantially behind that in the structural measures of the 1990s model member countries of the Organisation (box 2.1). The index aims only to describe a for Economic Co-operation and country’s reform efforts; it makes no judgment Development on whether more reform is better or worse than less. By quantifying the extent of imple- The 1990s power sector reform model has mentation in every country and every year in been more widely adopted among the sample, however, the index is useful for Organisation for Economic Co-operation and characterizing and analyzing patterns. The Development (OECD) countries than in the main findings are presented as a series of developing world. The median PSRI score is observations in the next section of this 78 for OECD countries; it is 37 for developing chapter. countries. This disparity is evident in com- parisons of the PSRI frequency distribution KEY FINDINGS for both groups of countries (figure 2.1). For The analysis set out in this chapter is based on the OECD country group, the values are a survey of 22 developed and 88 developing strongly skewed toward scores in the 80–90 countries and their implementation of power range; and only about 1 in 10 of these coun- sector reform measures. Undertaken for the tries scores below 50. For the developing Regulatory Indicators for Sustainable Energy country group, the distribution is much flat- (RISE) project for the period from 1990 to ter and concentrated in the lower range of 2015, this survey provides a rich source of the PSRI, with two-thirds of countries BOX 2.1 Defining the Power Sector Reform Index A simple Power Sector Reform Index was constructed to aggregate data across the four dimensions of power sector reform considered in this study. The index gives each country a score in the range 0–100 on each dimension of reform. The scores give equal weight to each step of each dimension on the reform continuum (see table B2.1.1). The simple average of the four 0–100 scores is used to summarize the extent of reform, although one could also weight the four according to their relative difficulty and importance. The index is purely descriptive and has no normative value. TABLE B2.1.1 Power Sector Reform Index Regulation No regulator = 0 Regulator = 100 Restructuring Vertically Partial vertical Full vertical Vertical and horizontal integrated = 0 unbundling = 33 unbundling = 67 unbundling = 100 Competition/ Monopoly = 0 Independent power Single buyer Bilateral Competitive liberalization producers = 25 model = 50 contracts = 75 market = 100 Private sector 0.5 × (percentage of generation capacity with private sector participation) participation + 0.5 × (percentage of distribution utilities with private sector participation) How Far Did Power Sector Reform Spread in the Developing World? 59 FIGURE 2.1 OECD countries score systematically higher on the Power Sector Reform Index PSRI Scores: OECD vs. developing countries 40 38 35 30 29 Share of countries (%) 25 20 19 16 15 13 9 10 10 10 10 10 7 7 5 5 5 4 5 5 0 0 0 0 0–10 10–20 20–30 30–40 40–50 50–60 60–70 70–80 80–90 >90 PSRI score OECD countries Developing countries Sources: Based on Regulatory Indicators for Sustainable Energy 2017; World Bank Private Participation in Infrastructure Database 2016 (https://ppi.worldbank.org/); and industry data. Note: OECD = Organisation for Economic Co-operation and Development; PSRI = Power Sector Reform Index. scoring below 50. Yet adoptions of the full set FIGURE 2.2 OECD countries are more likely to have adopted of measures envisaged by the 1990s model restructuring and liberalization reforms have been far from universal, even in the Advancement on four reform elements: OECD vs. developing countries OECD country group. For example, Japan and the Republic of Korea both retain verti- 100 95 90 87 90 86 86 cally integrated utilities, and France’s elec- Share of countries (%) 80 69 tricity supply chain remains almost entirely 70 65 60 state owned. Overall, about 85 ­ p ercent of 50 38 OECD countries has fully unbundled power 40 sectors, vertically and horizontally; however, 30 20 only about half of them (40 ­ p ercent) has 10 introduced PSP in generation and 0 Restructuring Regulation Liberalization Privatization distribution. Reform element The gap in reform uptake between OECD OECD countries Developing countries and developing countries is particularly Source: World Bank elaboration based on Regulatory Indicators for Sustainable p ronounced for some elements of reform. ­ Energy 2017; World Bank Private Participation in Infrastructure Database 2016 Although the aggregate reform scores differ (https://ppi.worldbank.org/); and industry data. Note: OECD = Organisation for Economic Co-operation and Development. significantly, this disparity conceals important variations in the elements of reform that are implemented (figure 2.2). For instance, the power sectors. The depth of privatization is prevalence of PSP is relatively similar greater, meanwhile, in the OECD countries, between OECD and the developing world affecting on average 50 ­ percent of the sector; (about 85 to 95 ­ percent). Both groups have the developing world has managed to privat- introduced at least a degree of PSP in their ize only 33 percent of its power sector. A large 60 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD gap exists, however, between the two groups Kingdom began spreading to the developing on the prevalence of regulatory agencies world. From 1990 to 1995, reformers experi- (65 ­percent in developing versus 86 ­ percent mented primarily with contract-based forms in OECD countries), and liberalization of PSP in generation and, to a lesser extent, (69 percent in developing and 90 ­ percent in distribution. Divestitures were relatively rare, OECD countries). The most striking disparity and less than 10 ­ p ercent of countries had in reform uptake is seen in restructuring. The established a regulatory entity. By 1995, the share of OECD countries embarking on these global average PSRI score for developing reforms is about 50 ­ p ercentage points countries stood at just 12 (figure 2.3). The higher than it is for developing countries. subsequent decade saw rapid diffusion of The discrepancy is even greater in the depth power sector reforms, boosting the global of reform. Compared with developing coun- index almost threefold to a PSRI score of tries, OECD countries are twice as likely to 35 in 2005. These measures established regu- have fully unbundled their power s ­ ectors latory entities and introduced PSP or the (vertically and horizontally); they are divestiture of generation and distribution. 10 times more likely to have introduced a Between 2005 and 2015, the uptake of these wholesale power market. measures slowed considerably, with the global score climbing only 8 points (from 35 to 43) in 10 years. Finding #2: The rapid diffusion of Furthermore, power sector reform since power sector reform from 1995 to 2005 2005 has been limited to certain relatively was followed by a notable slowdown straightforward measures. Reform uptake is from 2005 to 2015 slowing down across a range of key measures, The diffusion of power sector reforms across with adoption rates varying by a factor of two the developing world has slowed since 2005. or three across the two decades (figure 2.4). The year 1990 marks the baseline—when Most of the reform from 2005 to 2015 centered early reforms in Chile and the United on the creation of regulatory entities and par- tial vertical unbundling—seen in about 15–20 ­ percent of surveyed countries. Major FIGURE 2.3 The uptake of power sector reform measures has reform measures—for example, full vertical been slowing since 2005 unbundling, privatization of generation or dis- Uptake of power sector reform measures in the developing tribution, and creation of the single-buyer world, by type of measure, 1995–2015 model—were adopted by only about 5 ­ percent 100 of the surveyed countries. 90 80 The 1990s power sector reform model has 70 been promoted tirelessly for 25 years. Still, the developing world is less than halfway to PSRI score 60 50 full adoption. The slowdown since 2005 does 40 not, however, reflect saturation. We know 30 that about 70 ­percent of the developing coun- 20 10 tries analyzed has cherry-picked the most 0 prevalent reform measures; meanwhile, other countries have yet to even embark on power 20 9 00 10 20 1 02 20 4 20 5 06 20 8 09 20 1 15 12 14 20 3 20 7 13 19 5 96 19 7 98 0 1 9 0 0 0 20 0 0 9 9 20 20 20 20 20 19 20 20 19 19 sector reforms. The rate and extent of reform Regulation Restructuring Liberalization Privatization suggest that several factors may be inhibiting Source: Foster and others 2017. Note: PSRI = Power Sector Reform Index. uptake. How Far Did Power Sector Reform Spread in the Developing World? 61 FIGURE 2.4 A slowdown is evident across a wide range of different reform measures Reform measures undertaken, 1995–2005 vs. 2005–15 50 Share of countries adopting reform (%) 45 40 35 30 25 20 15 10 5 0 Regulation PSP in At least some Full PSP in Divestiture Divestiture Single generation degree of vertical distribution in in buyer restructuring unbundling generation distribution model Reform measures 1995–2005 2005–15 Source: Foster and others 2017. Note: PSP = private sector participation. Finding #3: Geographical, economic, With regulatory reform measures, the and power system characteristics all c ­ ountry-specific effects are negligible, perhaps strongly affect the uptake of power because these measures are relatively easy to sector reform adopt. The strongest and only statistically sig- nificant differences observed in the uptake of The uptake of various reform measures is regulation are seen across income groups. closely related to a country’s geography, Specifically, the prevalence of regulatory agen- income level, power system size, and politi- cies is highest for the upper-middle-income cal economy. As a general rule, reform group. Furthermore, the leading geographic uptake is much stronger in Latin America group, Latin America and the Caribbean, was and the Caribbean and in Europe and 48 percent more likely to have introduced reg- Central Asia compared to other regions. ulatory reforms than the lagging group, the A country’s income group typically has a Middle East and North Africa; overall results by strong positive effect on the prevalence of geographic region were not, however, statisti- reform; the physical size of the power system cally significant. has an even stronger positive effect. There is Large differences were seen in the ­ adoption weaker evidence that oil importers go fur- of restructuring measures, particularly across ther with their reforms than do oil exporters. geographic regions. For unbundling, the gap Uptake is more prevalent in multipolar (with between the leading region, Europe and competing power centers across jurisdic- Central Asia, and the lagging region, Sub- tions) political systems relative to unipolar Saharan Africa, is as much as 60 ­ percentage (where political power is concentrated) ones. points and statistically significant. Moreover, a The magnitude and statistical significance of high-income country, an oil-­ importing state, these effects vary according to the dimension or a country with a power system producing of reform. more than 10 gigawatts (GW) can each add 62 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD between 14 and 46 ­ percentage points to the Regarding PSP, differences persist across probability of unbundling its power sector. country groupings, with little prospect of con- With respect to competition, the scale of a vergence. The gap between the uptake of PSP country’s power system proves critical for the in the leading region, Latin America and the adoption of wholesale power markets. Caribbean, and the lagging region, Sub- Looking at liberalization overall, we see the Saharan Africa, is 43 ­ p ercentage points. most statistically significant differential across Countries in higher-­ i ncome groupings are regions. The leading region, Latin America more likely to have PSP by a differential of and the Caribbean, and the lagging region, 23 ­percentage points. Again, a country with a Middle East and North Africa, are separated large power system is about 30 ­ percent more by a gap of up to 74 ­ percentage points. Also likely to undertake PSP in its system. By and striking is that countries with multipolar large, countries in the ­different size groupings political systems are not necessarily more show no signs of converging. Differences in inclined to adopt competitive power markets. uptake are therefore becoming more pro- Countries with a large power system are nounced over time. Countries ­ producing more about 50 ­ p ercent more likely to introduce than 1 GW are twice as likely to have PSP in competition into their power systems. The generation as those with smaller systems. In effect of size is particularly stark for wholesale the case of distribution, the threshold is higher, power ­ markets. No country with a national and only in systems with more than 5 GW of power system of less than 1 GW of capacity capacity does the probability of PSP double has adopted a wholesale power market. Their relative to countries with smaller systems. systems are likely too small to support enough The persistence of such pronounced differ- generators for a competitive market to be ences in reform uptake over two decades and meaningful. Among countries with systems in across country groupings suggests that coun- excess of 10 GW, about 50 ­ p ercent has a try characteristics constrain reform. The wholesale power market—more than twice 1990s reform model would appear more the ratio for countries with smaller s ­ ystems in readily applicable in certain environments. the 1–5 GW range. The fact that regional variations are the larg- est of any observed suggests a couple of FIGURE 2.5 Latin America and the Caribbean led the way on things. First, some of the other effects are power sector reform with many other regions lagging behind aggregated and accentuated by geography— Reform progress, by region, 1995–2015 for example, by the concentration of small, 100 low-income, fragile countries in Africa. 90 Second, region-specific drivers might be at 80 70 work, including perhaps an intraregional Reform score 60 bandwagon or domino effect. 50 On virtually every type of reform mea- 40 30 sure, Latin America and the Caribbean has 20 been a pioneer, whereas the Middle East and 10 0 North Africa region has lagged. Chile’s pio- neering reforms of the 1980s spread rapidly 19 5 96 97 98 20 9 00 20 1 02 20 3 04 20 5 06 20 7 08 20 9 10 20 1 12 13 14 15 0 1 0 0 9 0 9 0 20 20 20 20 20 19 19 19 19 20 20 20 20 across Latin America and the Caribbean, East Asia and Pacific Europe and Central Asia Latin America and Caribbean Middle East and North Africa doubling the average reform score from South Asia Sub-Saharan Africa about 30 to 60 in the five years from 1995 to Source: Foster and others 2017. 2000 (figure 2.5). Europe and Central Asia How Far Did Power Sector Reform Spread in the Developing World? 63 also reformed at a brisk pace, reaching a broad groups according to their popularity score of about 60 in 2005, taking an addi- (figure 2.7). tional five years to get there. In other devel- The most popular reform, undertaken by oping regions, reforms have been adopted two-thirds of developing countries, is the intro- gradually, reaching an average score of just duction of regulatory agencies and indepen- 30 to 40 by 2015. Noteworthy, nonetheless, dent power producers (IPPs). Creating a is Sub-Saharan Africa’s reform journey, regulatory entity is politically ­ ­ straightforward— which showed the largest increase in the the reform simply adds to the sector rather PSRI from 1995–2005, eventually outstrip- than changes it. A regulatory entity may, in ping the Middle East and North Africa and principle, represent a challenge to the sector’s registering scores similar to those observed political hegemony. In practice, h ­ owever, regu- in South Asia. lators don’t always possess the power to pro- Overall reforms have had relatively little vide such challenges. PSP in generation is the traction in lower-income countries and those s econd-most-widespread reform measure, ­ with smaller power systems. Country charac- ­ particularly through build, operate, transfer teristics such as geography, income group, arrangements (called BOTs) or IPPs, although power system size, and political system seem divestiture is also common. PSP in generation to have had a statistically significant influence has proved less c ­ hallenging to implement than on uptake (figure 2.6). One of the greatest other measures in part because it can often be influences has been system size: countries applied to new plants. This application avoids with installed capacity above 10 GW scored the need to alter existing institutions, involves more than twice as high on the PSRI as those few employees, is relatively straightforward to with systems below 1 GW. Similarly, countries manage, and is remote from the customer in the middle-income bracket and those with interface. Furthermore, it lends itself to a relatively competitive political dynamics simple ring-fenced power purchase arrange- ­ scored much higher than others. Countries ment, either with a creditworthy utility or one with unipolar political systems also score whose ­ creditworthiness can be enhanced with about 15 ­ percentage points behind those with a state guarantee. Meanwhile, about one-third multipolar systems. Equally striking is the fact of developing countries was unwilling or that reform uptake is about the same for frag- unable to undertake even these relatively trac- ile states and nonfragile states (although coun- table measures over 20 years of tireless promo- tries that have not implemented any reform tion in policy circles. are almost all fragile), and between countries About half of developing countries made with stronger and weaker rules of law. progress in the vertical unbundling of their utilities and in introducing PSP in distribution. Both measures ­ d isrupt the status quo. Finding #4: Countries cherry-pick their Unbundling means that the incumbent utility power sector reform measures has to be restructured into multiple entities, The 1990s power sector reform model was creating governance and human resource chal- originally conceived of as a package of mea- lenges. When the private sector enters into dis- sures. In practice, developing countries have tribution, it changes the point of interface with been selective in what measures they adopt. the public—a politically sensitive change that Some elements of the package have proved likely imposes tougher payment discipline, much more popular than others. The different higher tariffs, and more stringent oversight of reform measures can be divided into three labor practices in a larger workforce. 64 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 2.6 Reform uptake has been strongly associated with income group and scale of system Reform uptake by various country characteristics, 1995–2015 a. By income group b. By region 100 100 80 80 Reform score (%) Reform score (%) 60 60 40 40 20 20 0 0 Low Lower- Upper- Middle East Sub- East Asia South Europe Latin income middle middle and North Saharan and Asia and America income income Africa Africa Pacific Central and Asia Caribbean c. By system size d. By rule of law 100 100 Reform score (%) 80 80 Reform score (%) 60 60 40 40 20 20 0 0 <1 GW 1–10 GW >10 GW Low High e. By fragility status f. By political economy 100 100 80 Reform score (%) Reform score (%) 80 60 60 40 40 20 20 0 0 Nonfragile Fragile Dominant Intermediate Competitive Source: Foster and others 2017. Note: All panels are based on data from developing countries and do not include any high income countries. GW = gigawatt. percent level. For panel f, differences are statistically significant For panels a, b, and c, differences are statistically significant at the 5 ­ percent level. at the 10 ­ Only one in five developing countries because their establishment is relevant only achieved full vertical and horizontal unbun- in larger power systems, remains contingent dling of the sector and created a wholesale on the prior adoption of many other reform power market. It makes sense that the cre- measures, and asks states to relinquish much ation of power markets would be limited, o f t h e i r i n f l u e n c e o v e r t h e s e c t o r. How Far Did Power Sector Reform Spread in the Developing World? 65 In addition, the comparative rarity of PSP in FIGURE 2.7 Some reform measures proved a lot more popular transmission partly reflects that this element than others was not a strong policy prescription of the Relative uptake of various reform measures, by type, 1995–2015 reform model, because of the many public good functions provided by the power grid. A regulatory entity Finally, management contracts are not widely adopted in the power sector, except in BOT/BOO contracts Sub-Saharan Africa, where experience has been mixed. Although these contracts are Reform measure At least partial vertical unbundling relatively easy to introduce from a political and technical standpoint, the impacts have Some form of PSP in distribution been limited. Whereas momentum is slowing for reforms Full vertical and horizontal unbundling aimed at restructuring and competition, it remains robust for regulation and PSP A competitive power market ­(figure 2.8). The adoption of utility unbun- dling and creation of wholesale power 0 10 20 30 40 50 60 70 80 90 100 ­ markets has been almost flat since the late Share of countries (%) 1990s. By contrast, the creation of regulatory Source: Foster and others 2017. agencies and adoption of PSP have continued Note: BOO = build, own, operate; BOT = build, operate, transfer; PSP = private to enjoy some uptake throughout the period. ­sector participation. FIGURE 2.8 Some types of reforms diffused more rapidly than others Reform year, by share of countries, 1995–2015 100 80 Share of countries (%) 60 40 20 0 00 15 5 96 97 98 99 01 02 03 04 05 06 07 08 09 10 11 12 13 14 9 20 20 20 20 20 20 20 19 20 19 19 19 19 20 20 20 20 20 20 20 20 Full degree of competition (retail electricity market) Full degree of restructuring (vertical and horizontal unbundling) Full degree of private sector participation (PSP in generation and distribution) Establishment of regulator Source: Foster and others 2017. Note: PSP = private sector participation. 66 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 2.9 Barely a dozen developing countries managed to Finding #5: Many countries find implement the full 1990s reform package themselves stuck at an intermediate Reform package adoption, by country and reform element, stage of power sector reform 1995–2015 Argentina Barely a dozen developing countries have Turkey Philippines managed to implement the full 1990s reform Guatemala Peru package, with most countries stuck at an India Nicaragua intermediate stage of reform. Overall, the Romania Ukraine Jordan median score on the PSRI is 43, and Dominican Republic Nigeria three-quarters of developing countries score Brazil Uganda below 72 on the index (figure 2.9). This Poland Ecuador Colombia score illustrates clearly how most developing China Chile countries find themselves at the early or Armenia Ghana Bangladesh intermediate stage of the power sector Russian Federation Vanuatu reform agenda. Not a single developing Senegal Vietnam country has a PSRI score of 100, and only Malaysia Kazakhstan Angola about a dozen have scores above 80. The Côte d'Ivoire Cameroon most aggressive reformers are found in Latin Kenya Mexico America and the Caribbean (Argentina, Pakistan Algeria Honduras Brazil, Guatemala, Nicaragua, and Peru) as Venezuela, RB Cambodia well as Europe and Central Asia (Romania Thailand Mozambique and Turkey). Also, outliers are found in Sierra Leone Egypt, Arab Rep. Mongolia every region (map 2.1). Jordan, Nigeria, and Ethiopia Kyrgyz Republic Median the Philippines stand out as aggressive Zimbabwe Zambia reformers in regions where bold reform has South Africa Togo Nepal not been the norm; India also ­ figures in this Bolivia Madagascar group. At the other extreme, only about a Rwanda Tanzania dozen countries have taken no steps toward Morocco Mali Burkina Faso implementation of the 1990s reform model, Myanmar Sudan scoring close to zero on the index. They are Burundi Mauritania primarily fragile (Afghanistan, Democratic Belarus Uzbekistan Niger Republic of Congo, Eritrea, Lebanon, Liberia, Malawi Congo, Rep. South Sudan, and Republic of Yemen) or Central African Republic Benin small island states (Maldives and Solomon Lao PDR Iran, Islamic Rep. Somalia Islands). One striking feature is that the clear Tajikistan Chad majority of developing countries has taken Indonesia Haiti partial power sector reform measures but Tunisia Sri Lanka Korea, Rep. remains stuck at an intermediate stage: about Afghanistan Lebanon half of countries score between 20 and 60 on Yemen, Rep. Maldives the index. Congo, Dem. Rep. South Sudan Solomon Islands When countries have adopted holistic Liberia Guinea power sector reforms, the time it took for Eritrea 0 20 40 60 80 100 implementation ranged from 3 to 18 years, Reform element (%) with a median value of 12 years. Countries Regulation Restructuring Liberalization Private sector participation with high PSRI scores undertook reforms at Source: Foster and others 2017. differing speeds (figure 2.10). To understand How Far Did Power Sector Reform Spread in the Developing World? 67 MAP 2.1 Outlier countries on power sector reform exist in every region Power sector reforms, by country, 1995–2015 Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015; Regulatory Indicators for Sustainable Energy 2016. Note: PSRI score based on existing legislation (as of 2015), which may be different from practice. the time periods needed to accomplish major FIGURE 2.10 Countries with high PSRI scores undertook reforms, we confined our attention to those reforms at differing speeds countries that completed the bulk of the reform Time elapsed between first and most recent reform step (in years) agenda by introducing a regulatory entity, unbundling vertically and horizontally, intro- Ukraine ducing a wholesale power market, and privat- Romania percent of their generation and izing at least 50 ­ Philippines distribution sectors. Eight countries fall in this category. In Latin America countries enacted India reforms relatively rapidly in under five years. Turkey Peru took three years, Argentina four, and Guatemala Guatemala five. Reforms in Europe and Asia Argentina took substantially longer, in the 10–20 year Peru range: Turkey 12 years, India 13 years, the Philippines 15 years, and Romania 16 years. 0 5 10 15 20 Ukraine took 18 years. These long implemen- Number of years tation time frames indicate that considerable Source: Based on Regulatory Indicators for Sustainable Energy 2017; World Bank Private Participation in Infrastructure Database 2016 (https://ppi.worldbank.org/); time is needed to adopt a complex reform and industry data. package. Note: PSRI = Power Sector Reform Index. 68 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Finding #6: Countries often packaged Most countries reversed the order of and sequenced power sector reform implementation, contravening the reforms in ways unrelated to original sequence originally envisaged. There is no reform logic common sequence for the adoption of reform measures across countries; rather, all possible Power sector reform measures were originally permutations can be found (figure 2.12). conceived as a coherent package, and it was Overall, 34 percent of countries introduced envisaged that they would be implemented regulation before undertaking any PSP in according to a certain logical sequence generation or distribution. This p ­ attern is (­figure 2.11). Regulation would typically be overshadowed by the 52 ­ percent of countries the starting point, so that the rules and incen- that did things the other way around. tives of the sector would be specified before Similarly, some 20 ­ percent of countries intro- any changes were made to the institutional duced PSP into distribution ahead of genera- actors. This step would be followed by unbun- tion; they were greatly outnumbered by the dling to create the desired sector architecture, percent of countries that began with gen- 40 ­ which, once completed, would pave the way eration instead. This practice may reflect the for potential changes in ownership. Given fact that, although it is more logical, from an that the distribution segment represents the economic perspective, to start PSP in distribu- beginning of the payment chain for electricity, tion, it is politically and practically more feasi- it was generally thought that it made sense to ble to introduce PSP to the generation seg- begin PSP with distribution to improve the ment, where employment, affordability, and sector’s operational performance and com- customer interface are lesser issues. mercial viability. Improvements here would About one-third of developing countries then generate a solid revenue stream able to ended up with partial and incoherent combi- provide the basis for remunerating PSP in nations of the reform package. We have generation. Only after all these elements were seen how countries have been selective in in place would it make sense to move toward their implementation of reforms, and their a wholesale power market, which relies on cherry-picking was not always informed by ­ multiple actors and a strong set of commercial the synergies produced among the different incentives—both of which would need to be reform combinations. By examining the achieved through prior unbundling and PSP reform combinations that countries ended up ­ measures. Some competition could be intro- with in 2015, we were able to identify imbal- duced at earlier stages, however. IPPs could ances; for example, one dimension of reform precede unbundling, for example, whereas proceeded further than was warranted by single-buyer or third-party access could ­follow achievements along a related dimension. unbundling. FIGURE 2.11 Power sector reform was conceived as a coherent package of measures implemented according to a logical sequence Introduction Introduction Establishment of Establishment of Unbundling of of PSP in of PSP in wholesale or regulatory entity utility distribution generation retail power market Source: World Bank elaboration. Note: PSP = Private sector participation. How Far Did Power Sector Reform Spread in the Developing World? 69 percentage of countries adopted reforms in unorthodox ways FIGURE 2.12 A significant ­ Variations on the initially recommended reform sequence, by share of countries 16 14 12 Share of countries (%) 10 8 6 4 2 0 Adoption of Regulator PSP without Some Wholesale Bilateral competition without regulates public regulator competition competition contracting relevant unbundling vertically without without partial without PSP integrated regulator unbundling monopolist Reform adopted Source: Foster and others 2017. Note: PSP = private sector participation. The measures to restructure and privatize ­ istribution yet has not introduced any d utilities are strongly linked with those that competition beyond allowing IPPs. Another ­ introduce competition, and these measures anomalous situation is found in countries have proved challenging to progress that is (11 ­percent) that have introduced PSP in a balanced. Vertical unbundling distinguishes ­ utility that remains a vertically integrated natural monopolies from activities that are national monopoly. potentially competitive; PSP then places poten- Another area of incoherence is seen in the tially competitive ­ segments on a commercial creation of regulation entities. On the one footing, and these segments are allowed to hand, strictly speaking, a regulatory entity is interact through ­measures that introduce com- not essential when the sector is organized as a petition. Over 30 ­percent of countries demon- publicly owned vertically integrated utility, strates imbalances among these three elements because the whole rationale for state owner- of structural reform. For example, almost ship is to act in the public interest, and no 10 ­percent of countries introduced power mar- wider market relations require regulation. kets before fully unbundling their power sec- Moreover, when the regulator and the public tors. Conversely, just over 10 percent of coun- utility report directly to the line ministry, the tries completed full vertical and horizontal regulator has difficulty exercising authority. unbundling of the sector, but they still practice Nonetheless, 13 ­ percent of developing coun- a single-buyer model when they look structur- tries created regulators when their utility ally prepared for a power market. Furthermore, remained an integrated state-owned monop- almost 17 ­percent of countries has been able to oly. On the other hand, a regulator is critical introduce PSP in both generation and once private ownership or operation has been 70 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD introduced to any part of the sector, and as group of countries i ­mplemented reforms the reforms of competition continue. Still, selectively (typically emphasizing regula- 13 ­ percent of developing countries lacks any tion and IPPs) and often ended up at an regulatory entities despite having introduced intermediate stage of reform without much PSP while 12 ­percent of developing countries impetus to introduce further reform mea- lacks regulatory entities despite having com- sures. In addition, a significant minority of menced the process of increasing competition. countries has barely started on reform and does not appear to present suitable condi- tions for doing so. CONCLUSION In summary, these patterns in the uptake of NOTE power sector reform illustrate that, although 1. This chapter is based on a background paper the Washington Consensus has been com- by Foster and others (2017). monplace across the OECD, implementing recommended reforms in their entirety REFERENCES remains challenging to developing Banerjee, Sudeshna Ghosh, Francisco Alejandro c ountries. Policy prescriptions for power ­ Moreno, Jonathan Edwards Sinton, Tanya Primiani, and Joonkyung Seong. 2017. sector reform diffused rapidly around the “Regulatory Indicators for Sustainable Energy: world during the decade 1995–2005; uptake A Global Scorecard for Policy Makers.” World in stable, m­ iddle-income countries with Bank, Washington, DC. large power systems was significant during Foster, Vivien, Samantha Helen Witte, Sudeshna this period. Even among this group, how- Ghosh Banerjee, and Alejandro Vega Moreno. 2017. “Charting the Diffusion of ever, only a handful of countries was able Power Sector Reforms across the Developing to fully implement the reform model as World.” Policy Research Working Paper 8235, originally envisaged and t­ ypically managed Rethinking Power Sector Reform, World Bank, to do so over a long period. A much larger Washington, DC. How Did Political Economy Affect the Uptake of Power Sector Reform? 3 Guiding questions • How rapidly did countries implement reforms, and how long did they manage to sustain them? • How did the nature of a country’s political system affect its experience of reform? • How important were reform champions and stakeholder alignment in delivering and sustaining reform? • How will technological disruption affect the political economy of power sector reform? Summary • Crises typically provide political windows of opportunity for reform, yet the ambition and implementation of the reform have little to do with the depth of the crisis. ­ • Donor influence can be a catalyst and provide support for reform, but local ownership is what ­ determines the boldness of reform and its ultimate sustainability. • Countries espousing market-oriented ideologies are more likely to implement and sustain bold reforms. • Reforms can be introduced across a range of political systems but are more likely to progress in coun- tries where power is decentralized and contestable. • Reform champions are often pivotal figures who help to ensure reform momentum, particularly when they are supported by stable bureaucracies. • Unless champions orchestrate durable stakeholder alignments, reforms can become unsustainable. • Legislation enshrines a level of political commitment to reform that supports longer-term sustainability. • Looking ahead, the pace of technological disruption will reflect the interplay between existing vested interests in the power sector and new decentralized actors entering the sector. 71 72 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION by initially corporatizing and eventually privat- This chapter examines the extent to which izing service providers, taking business deci- power sector reform in the developing world sions out of the hands of bureaucrats. At the has been shaped by the political dynamics of same time, the creation of an autonomous reg- each country.1 It first lays out the implemen- ulatory entity—operating at arm’s length from tation stages for power sector reform. It then the executive branch of government with its charts the progress made by different coun- own source of financing and independent tries through these stages and explores the ­ leadership—was intended to insulate tariff and influence of diverse factors on the evolution licensing decisions from direct political inter- of reforms. The guiding questions for this ference. The global diffusion of reforms from chapter are the following: To what extent did the 1990s to 2002 was for the most part countries implement and then sustain reform? unprecedented, and so could not be accompa- What factors affected the speed of implemen- nied by evidence-based forecasts of the possible tation? How was the reform process affected impacts on efficiency or the political feasibility by the ideology and structure of a country’s of implementation. Geopolitical factors made political system? Finally, how important were policy makers willing to pursue reforms despite reform champions and stakeholder alignment risks and costs, especially in response to crisis. in delivering and sustaining a reform agenda? For reform advocates of the period, the intrin- The 1990s model of power sector reform was sically political objectives of equitable access to born in an age of relative political innocence. modern and clean energy were not generally The Cold War was ending, political left and high on the agenda, but least-cost “efficient” right seemed to be converging, political solutions to power supply were most certainly and economic liberalism felt itself triumphant, paramount. and some even declared the “end of history” In practice, it proved impossible to contain was nigh. Former socialist countries were join- political influence over the sector. Various ing the “transition to market economies,” and commentators have underscored the power developing countries were downplaying the sector’s inherently political nature. As noted by economic role of the state as they entered peri- Kofi Annan (APP 2015, 16), “Governments ods of “structural adjustment.” Many countries often view [electricity] utilities primarily as were moving toward greater political pluralism. sites of political patronage and vehicles for cor- From this historical context emerged the 1990s ruption.” Similarly, “Power market reform is model of power sector reform, which empha- an inherently political process … often an sized accountability through the separation of arena of conflict between competing interests powers and the transparent interplay of politi- that are of fundamental importance to society” cal forces. A technocratic and avowedly apoliti- (Besant-Jones 2006, 14). The experience of cal vision of the future power sector emerged. power sector reform has reflected this reality, In theory, the 1990s model aimed to with political factors determining the extent to improve sector efficiency and attract invest- which reform is feasible—beyond the original ment, in part by depoliticizing key decisions. technocratic vision. This experience has led to It envisaged a power sector that would pursue the rise of explicitly political interpretations of public purpose through the motivator of pri- power sector reform dynamics, and eventually vate profit, with competition or antimonopoly to approaches that explicitly take the political regulation providing checks and balances. economy into account. Though still emerging, A key objective of reform measures was to such approaches seem to offer a sturdy frame- depoliticize sector decision making (Lee and work for understanding reform as it takes Usman 2018). This objective was to be achieved shape on the ground (Lee and Usman 2018). How Did Political Economy Affect the Uptake of Power Sector Reform? 73 This chapter examines power sector reform examines how people interact politically, in through a political economy lens; it aims to pursuit of specific interests, within the context understand how political forces shaped the of their ideas, history, and perceptions. reforms. Drawing on extensive stakeholder The narrative considers how the means of interviews (see box 3.1), the discussion influence matter to the formulation and BOX 3.1 Methodology for political economy stakeholder analysis Using original, in-depth qualitative analysis of 15 countries constituting the Power Sector Reform Observatory, this study examines the role of the political economy in power sector reform (discussed in chapter 1). The countries were drawn from across the ideological spectrum—from state- to market-oriented political economies and hybrids of both. They are unipolar (where political power is concentrated) to multipolar (with competing power centers across jurisdictions). During the period under consideration, some of the countries evolved from unipolar to multipolar (for example, Peru) or moved among different political systems (for example, Pakistan). The countries perform differently on the relevant World Governance Indicators (for example, on rule of law, corruption, regulatory quality, and government effectiveness). The Latin American countries, whose annual income per capita is above US$5,000, perform relatively well on governance. But governance scores vary wildly among countries with an annual income per capita under US$3,000 (see figure B3.1.1). In each of the 15 observatory countries, local experts conducted interviews with some 20 or more stakeholders involved in the reform process, often over long periods or during key episodes.a Some of the stakeholders had occupied positions of authority, whereas others were close observers. Interviews were structured partly around specific ques- tions about reform episodes and processes in that country, but they were also allowed to flow in an open-ended way in order to be more informative and frank. Interviews were written up individually, and also summarized into a n ­ arrative of events supported by a timeline of reforms and a stakeholder mapping exercise. The interview-driven methodology can support deeper analysis of political economy processes in particular countries at particular times. Interviews make it possible, first, to get behind the scenes to see what happened and why, and, second, to obtain a range of perspectives and interpretations. It is essentially like triangulating history, providing considerable country-specific detail to illustrate and support broad global generalizations. Apparent trends across the case studies were systematically verified through tabulation. This in turn allowed for qualitative comparisons of political factors and their impact on reform outcomes. FIGURE B3.1.1 No strong relationship between income group and quality of governance World Governance Indicator, by income group and GDP per capita, 2010–15 0 Average world governance indicators (2010–15) Peru –0.2 Senegal Colombia Morocco India Tanzania Philippines Dominican –0.4 Republic Vietnam Uganda –0.6 Kenya Ukraine –0.8 Egypt, Arab Rep. –1.0 Pakistan Tajikistan –1.2 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 GDP per capita, 2010–15 (US$) Source: Based on Rethinking Power Sector Reform utility database 2015 and World Bank World Development Indicators database 2018. Note: GDP = gross domestic product. a. All interviewees gave generously of their time and experience; most were candid and insightful. Interviews were confidential, so speakers are identified only generically (for example, as a senior power sector official or manager of a private power company). 74 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD implementation of policy, and how real-life reform agenda. On average, across the 15 institutions shape events and outcomes. Power observatory countries, reform announce- sector reform is a complex political process that ments made over a three-year period men- may take many years to implement. It is tioned 68 percent of the measures set out dependent on unfolding circumstances and in the 1990s reform agenda. often nonlinear in nature. Political history The second stage of reform is implementa- matters as much as economic concepts, but tion. Countries implement reform measures history in pursuit of economic objectives can, with varying speeds. Some countries not only to a greater or lesser extent, be managed announced ambitious reforms but also politically. went ahead to implement them boldly, com- For the purpose of this analysis, the stages pleting at least 60 percent of the 1990s reform of the power sector reform process involve agenda within three years. The three boldest steps forward and potentially backward. reformers—Colombia, Peru, and Ukraine— ­ The stages of the reform process can be each implemented as much as 70 percent characterized as, first, the announcement of ­ of the 1990s reform package within three the reform; second, the process of implemen- years. Other countries, even those that had tation; third, the delivery endpoint; and, expressed similar ambition, were more incre- term sustainability (figure 3.1). finally, longer-​ mental in their reform process. Figure 3.2 The Power Sector Reform Index (PSRI) illustrates the reform trajectories of a bold developed for the study2 (see annex 3A) can ­ reformer, like Ukraine, versus an incremental be used to track the extent of reform that reformer, like Vietnam. A third and smaller each country announced and implemented group of countries—including Morocco and year on year from 1990 to 2015. Tajikistan—took only limited action toward The first stage of reform is usually a public the implementation of reform. On average, announcement of the intention to reform. across the 15 observatory countries, countries Announcements generally feature a sector were able to implement 50 percent of the strategy, road map, or policy paper. In some 1990s reform agenda in the space of three countries, such announcements (made over a consecutive years. consecutive three-year period) were ambi- The third stage of reform is delivery, which tious, signaling an intention to undertake denotes the maximum extent of reform that more than 60 percent of the measures entailed the country was able to implement. Irrespective by the 1990s reform model. Four countries— of the speed at which they reform, countries Colombia, Pakistan, Senegal, and Uganda— differ in the extent to which they were finally pledged to undertake more than 80 percent of able to follow through on the implementation the 1990s reforms. A minority of countries— of their overall reform announcement. 3 Morocco, Tajikistan, and Tanzania—made Although none of the countries succeeded in modest announcements committing to fully implementing the reforms they implement less than 50 percent of the 1990s announced over 1990–2015, the degree of FIGURE 3.1 Power sector reform process: Announcement, implementation, delivery, and sustainability Extensive Rapid Complete Sustained Announcement Moderate Implementation Incremental Delivery Partial Sustainability Minimal Limited Negligible Nonsustained Source: World Bank 2019. How Did Political Economy Affect the Uptake of Power Sector Reform? 75 FIGURE 3.2 The contrasting reform trajectories of a bold reformer, Ukraine, and an incremental one, Vietnam PSRI, by reform measures, 1990–2015 a. Ukraine 100 80 60 Percent 40 20 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 b. Vietnam 100 80 60 Percent 40 20 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 Competition PSP Regulation Restructuring Announced competition Announced PSP Announced regulation Announced restructuring Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PSP = private sector participation; PSRI = Power Sector Reform Index. implementation varied from those that substan- 67 percent of the 1990s reform agenda, or tially achieved their announced reform goals, about 20 percentage points short of what was to those that did so partially, or those whose announced cumulatively from 1990 to 2015 progress was only negligible. Two groups of (figure 3.3). countries stand out. The first group includes The final stage of reform is sustaining the those whose reform delivery closely aligned measures implemented. Reforms can always be with their announcements: Peru, the undone by contrary political forces. Although Philippines, Uganda, and Ukraine. The second most countries were able to sustain the struc- group implemented reforms that fell short of tural reforms they implemented during the their reform announcement—short by as study period, a significant minority reversed much as 40 percentage points. This group enacted privatization reforms, as in the included the Arab Republic of Egypt, Senegal, Dominican Republic, the Indian state of Tajikistan, and Tanzania (figure 3.3). On aver- Odisha, and Senegal (figure 3.4). These rever- age, across the 15 observatory countries, sals reduced their reform index scores by reform delivery as of 2015 amounted to 11 percentage points on average. 76 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 3.3 Contrast between reforms announced and reforms delivered in observatory countries and states PSRI, announced vs. delivered, 2015 100 80 60 Percent 40 20 0 an ia co p. l m a an h n a ic a da e ru s ga ne ny sh bi in es ha bl an Re na Pe oc ist st an ra ne m di ad pi pu Ke st nz ki et jik or Ug lo Uk ab O ilip ja Se Pa Pr Re Vi Ta Co M Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg In m – Do a di In Cumulative reform delivery, 2015 Cumulative reform announced, 2015 Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PSRI = Power Sector Reform Index. FIGURE 3.4 Some countries experienced reversals of reform once implemented Reform reversals: Dominican Republic and Senegal a. Dominican Republic 100 80 60 Percent 40 20 0 90 1 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 9 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 b. Senegal 100 80 60 Percent 40 20 0 0 1 92 3 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 9 9 9 20 20 20 20 19 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 Competition PSP Regulation Restructuring Announced competition Announced PSP Announced regulation Announced restructuring Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PSP = private sector participation. How Did Political Economy Affect the Uptake of Power Sector Reform? 77 Countries display very different patterns in correlation coefficient of minus 0.52). their reform processes (table 3.1). Only two Moreover, the correlation between the bold countries (Peru and Ukraine) announced reform and likelihood of reversal is low (with a ambitious reforms, implemented them boldly, correlation coefficient of 0.18). In fact, the substantially delivered on them, and sustained probability of reform reversal at 29 percent was them over time. Four entities (the Indian states slightly higher among incremental reformers of Andhra Pradesh and Rajasthan, as well as than at 25 percent for bold reformers. the Philippines and Uganda) announced Characterizations of a country’s reform pro- ambitious reforms, implemented them incre- ­ cess are purely descriptive and not in the least mentally, substantially delivered them, and normative. As described in chapter 2, the subsequently sustained them. At the other end reform measure index developed for the study of the spectrum were countries that both does not intend to pass judgment on whether promised and delivered little. The most more reform is better or worse. It does, how- extreme case is Tajikistan, which delivered ever, prove helpful in measuring how much only 13 percent of the 1990s reform agenda. reform took place in a standardized way; it Overall, bold reform implementation correlates allows meaningful comparisons of the extent with the extent of reform ultimately delivered of reform across countries and over time. (with a correlation coefficient of 0.87). Put dif- Many factors shape the dynamics of a coun- ferently, countries approaching reform boldly try’s power sector reform. To understand them, tended to have less of a shortfall between we needed a qualitative comparison of the announcement and implementation (with a political economy narratives across the TABLE 3.1 The pace of power sector reform in the observatory countries and states Announcement Implementation Delivery Sustainability Peru Extensive Rapid Complete Sustained Ukraine Extensive Rapid Complete Sustained Colombia Extensive Rapid Partial Sustained India – Odisha Extensive Rapid Partial Nonsustained India – Andhra Pradesh Extensive Incremental Complete Sustained India – Rajasthan Extensive Incremental Complete Sustained Philippines Extensive Incremental Complete Sustained Uganda Extensive Incremental Complete Sustained Pakistan Extensive Incremental Partial Sustained Dominican Republic Extensive Incremental Partial Nonsustained Senegal Extensive Incremental Negligible Nonsustained Vietnam Extensive Limited Negligible Sustained Kenya Cautious Incremental Partial Sustained Egypt, Arab Rep. Cautious Limited Negligible Sustained Morocco Minimal Limited Partial Sustained Tajikistan Minimal Limited Negligible Sustained Source: Based on Rethinking Power Sector Reform utility database 2015. Note: All announcements made over a period of three years with an intention to implement more than 60 percent of the 1990s reform model are cumulatively classified as “ambitious.” Those between 50 and 60 percent are “cautious,” and those below percent are classified as “minimal.” Economies managing to implement more than 60 percent of the 1990s model reforms over a 50 ­ period of three years are classified as “rapid,” those implementing between 60 and 50 percent are “incremental,” and those below 50 percent are called “limited.” In economies where the difference between total announced and total implemented reforms is less than 15 percent, delivery is classified as “complete.” Where the difference is between 16 and 25 percent, it is classified as “partial.” Where the difference is greater than 25, the delivery of reform is classified as “negligible.” 78 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 3.5 Nested model of political influences in power desperate situation with no investments. The sector reform sector had been managed very poorly, politi- cally speaking, with illogical prices, and so Global: External influences forth. During the 1990s we had no choice— • Macro shocks changes had to be made. These changes were • Donor influence not necessarily well planned, but we had to act Country: Internal influences in the urgency of the moment to cover the defi- • Ideology Sector: Reform approach • Political system cit. The first reaction was to bring back private • Stakeholder alignment investments: it was logical and correct. Now, • Reform champion • Legislation how it was done and why it was done the way it was, was simply due to desperation. We looked at what had been done in other Source: Based on Rethinking Power Sector Reform utility database 2015. countries, did a quick balance, and got some ideas. 15 observatory countries. This exercise surfaced —Energy sector official several factors that recurred with regularity What were the goals of the reform then? It across all cases. Conceptually, they belong in cannot address efficiency. It cannot address three nested categories (figure 3.5), the broad- additional capacity. It cannot address higher est being exogenous factors, such as the erup- cost of electricity. It cannot. Only [thing] to tion of a crisis or influence exerted by foreign address was: stop the bleeding at that time. So donors. Factors pertaining to the national polit- it was purely reactional—and then you put ical sphere fall into the middle nest and include all that into the law: long-term objectives. Oh a country’s political system and ideology. my gosh. Finally, factors affecting reform at the level of —Energy sector official the power sector include reform champions, stakeholder coalitions, and legislation. The initial impetus for power sector reform among the 15 country case studies almost KEY FINDINGS always sprang out of economic crisis. Crisis Analysis in this chapter draws on the power certainly presents opportunities for action. It sector reform process characterized in table 3.1 reduces the strength of opposition to reform and on the nested model of political influences and highlights the benefits of reform. These on power sector reform in fi ­ gure 3.5. Drawing opportunities create critical junctures in the on the extensive qualitative material culled history of a country so that things that looked from stakeholder interviews, we examined the politically unlikely suddenly become emi- relevance of each political influence at each nently possible. Some sort of crisis is, in fact, stage of reform. Seven key findings are set out almost a prerequisite for reform. A glance at below. the observatory country cases is informative. It is unusual for a country to pursue power sector reform outside of an economic crisis. Finding #1: Crises typically provide Vietnam is the only possible exception, political windows of opportunity although it too struggled to meet the rapid for reform, yet the ambition and growth in demand for electricity. At the other implementation of the reform have extreme, Tajikistan, alone among the coun- little to do with the depth of the crisis tries, introduced no significant reform What we must understand is that the story m easures despite finding itself in a deep ­ started in the 1990s, when we were living in a ­sector crisis. How Did Political Economy Affect the Uptake of Power Sector Reform? 79 The triggering crisis can be macroeconomic TABLE 3.2 Crisis at the time of power sector reform or specific to the power sector (table 3.2). Macroeconomic Utility Power Macroeconomic crises take a variety of forms. Economy crisis financial crisis supply crisis Government budget pressures on the power Colombia S S S sector can create such a crisis. Another kind of Dominican Republic S S S crisis might take the form of accumulated non- Egypt, Arab Rep. M M — performing loans made to the power sector by India – Andhra Pradesh — M — the financial sector. Crises could emerge from India – Odisha M M — India – Rajasthan M S — the external balance of payments, as external Kenya M — — power sector debt accumulates. A depreciating Morocco — M M currency increases costs such that shortages of Pakistan — S S foreign exchange make fuel and spare parts Peru S S S unaffordable. Macroeconomic triggers like Philippines — M M these were seen in power sector reform epi- Senegal S S S sodes in Egypt, Kenya, Peru, and Tanzania. In Tajikistan — S S the case of Ukraine, although power sector Tanzania S S S reform was triggered by macroeconomic crisis, Uganda S S S it was also part of a wider structural shift Ukraine S — — toward a market economy. Alternatively, the Vietnam No Crisis crisis can manifest itself more directly in the Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Vietnam had no sectoral crisis at reform time. M = moderate crisis; power sector through blackouts and power S = severe crisis; — = not available. rationing, as seen in Colombia, Pakistan, or the Philippines. Or the impending threat of elec- tricity shortages might stem from the financial Electricity Board was in chronic operational fragility of the sector, as happened in the and financial crisis. By this stage, the public Dominican Republic and India. Electricity was exhausted by the failures of the previous shortages may themselves signal an underlying regimes and open to almost any option— financial crisis in the sector or, in hydro-domi- including neoliberal ones—to fix both the nated systems, occur because of drought. In economy and the power sector. This pragma- other cases, crises are brought on after coun- tism made Uganda’s thoroughgoing reforms tries emerge from periods of armed conflict or possible—more so than in most other African political turmoil, as with Uganda. countries at that time. By the early 1990s, the The case of Uganda illustrates how a wider government was implementing bold reforms of economic and political crisis wore down oppo- its power sector: a fully unbundled utility, pri- sition to power sector reform. When Yoweri vate sector participation in generation and dis- Museveni took power in Uganda in 1986, the tribution, a competent and relatively indepen- economy was barely functioning after years of dent regulator, and tariffs adjusted to reflect economic mismanagement. There was political costs. The implementation of so many struc- oppression, untrammeled corruption, and tural changes speaks to the power of crisis as a war. Inflation was rampant, production was political catalyst for reform. minimal, the exchange rate was under extreme Colombia’s power sector contributed greatly pressure, and the government budget was to the country’s macroeconomic crisis, and it in disarray. The power sector was likewise benefited from the wider political and institu- shattered—losses were running to at least tional reforms introduced in response. 40 percent, generation was down to 60 mega- By 1990, Colombia was in a deep macroeco- watts with poor system reliability. The Uganda nomic crisis. The power sector was responsible 80 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD for 30 percent of total external debt and and ultimate power sector reform outcomes in 33 percent of the public sector deficit—an the wider literature; the limited evidence is unsustainable burden for the state. Support equivocal about the reform value of a crisis from financing agencies was drying up. The (Lee and Usman 2018). These reservations are ­ situation was exacerbated by drought in 1992. borne out by data on the 15 observatory coun- A year of blackouts and power rationing tries, where we found no statistically signifi- revealed the sector’s deep structural failings cant relationship between the severity of the and threatened Colombia’s economic recovery. crisis, the ambition of the reform, and the tra- The government declared a social and eco- jectory of its implementation, including the nomic emergency. The new 1991 constitution boldness of implementation and how substan- incorporated institutional and governance tially reforms were delivered and ultimately reforms emphasizing decentralization and sustained (table 3.3). Indeed, in some cases, transparency. Building on this legal founda- reversals of model reforms have occurred tion, the government began to restructure the during crisis conditions, such as when utilities energy sector and its regulatory institutions. By were renationalized in the Dominican adopting new public service and electricity Republic, the Indian state of Odisha, and laws, the government was able to reform tariff Senegal. Crisis can therefore be understood regulation, open the sector to private invest- as creating a space for response; the choice of ment, and initiate a competitive wholesale response and its ultimate sustainability depend electricity market. on other factors. Nevertheless, although countries mostly In the Indian state of Rajasthan, a severe responded to crisis with ambitious reform crisis prompted the announcement of a deep announcements, subsequent implementation reform of the power sector, but political resis- and delivery bear little relationship to the tance prevented full adoption and sowed the severity of the crisis. Crisis is the standard back- seeds for further crisis. In 1998, the year before drop for power sector reform, but there is little reforms began, system losses were at 42 ­ percent evidence on the relationship between crisis and collection rates had fallen to 76 percent. TABLE 3.3 Crisis severity and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Severe Crisis Moderate None Statistical significance — — — — Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows the extent to which the severity of the crisis affects the subsequent process of reform across the sample 15 country cases. The darker the shading of the box, the more countries from the sample fall into each combination of circumstances. The Chi-squared test was performed to reveal any statistically significant relationship between the severity of the crisis and the ­ process of reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 p significance ­ ercent ­ (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low ­ significance); — = not significant. Full statistical results are reported in table 3B.1 in annex 3B. How Did Political Economy Affect the Uptake of Power Sector Reform? 81 The state electricity board was in critical condi- had done so much to help the utility, but it is tion, so the state enacted the Rajasthan Power still not working. These reforms are not due to Sector Act in 1999 to restructure the sector and external pressure but come from within improve operational and financial perfor- government. mance. The goal was to carry out tariff reform, —Academic introduce private sector participation in For policy reform to succeed, governments distribution, and make the sector more effi- ­ need to exhibit leadership and ownership of cient. Although the unbundling succeeded, the process. Local ownership is sometimes opposition to privatization emerged early in questioned when external development agen- the reform implementation, halting any cies exert influence, or where there are other attempts to move ahead with the measure. Left external influences. Donors have sometimes with the option of raising tariffs or reducing applied policy-based loan conditions or pro- system losses, the government focused on the vided technical assistance. Donor-supported latter and managed to keep losses in check for knowledge exchange and reform advocacy several years. Success was short-lived, how- can also be influential, particularly if there ever, and losses soon mounted. Combined with is a dominant global reform blueprint. a tariff freeze between 2004 and 2011, these Development partners aim to provide advice losses left the sector in a deep financial crisis that is technocratic and nonpolitical, but their with cumulative losses of over US$3 billion by counterparts in government do not necessarily 2008–09. perceive their purported neutrality. Moreover, political opinion in some countries is shaped by Finding #2: Donor influence is often adverse historical experience with foreign a catalyst for reform and provides investment or debt dependence. ongoing support, yet local ownership During the 1990s, many countries were shapes the boldness of the reform and heavily dependent on donors for financial ultimately determines its sustainability assistance and subject to donor influence on power sector reform. Almost all the 15 obser- As an economy, we were the strongest in the vatory countries were dependent on foreign region. I think because of that the donors … assistance at the time of reform, with aid pushed us a lot harder than they pushed other accounting for about 5 percent of gross domes- countries. Sometimes our political leaders … tic product (GDP) on average (figure 3.6). Even could not understand why the country was for countries with historically low reliance on being pushed so hard…. The push from out- official development assistance, donor influ- side, over a long period of time, made the lead- ence could be high when macroeconomic crisis ers realize it was the only way to go and occurred, as in Colombia, the Dominican decided to see if it would work. Over time, we Republic, and Peru. International financial found that it did work for us, and we have institutions were instrumental in diffusing embraced it since. market ideas during the 1980s and 1990s, —Utility manager reflecting the resurgence of neoclassical eco- nomic thinking. A notable feature of World The current Electricity Sector Reform Strategy Bank and International Monetary Fund lend- is a result of the poor performance of the ing in the 1980s was structural and sectoral power sector and perhaps an indication that adjustment programs based on market-­ government has given up on supporting an oriented macroeconomic and fiscal policies inefficient utility. Government thought they (Jayarajah and Branson 1995). The term 82 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 3.6 Dependency on foreign aid was substantial at the time of power sector reform Official development assistance at the time of power sector reform 25 20.73 20 Net aid as percent of GNI 15 12.34 11.06 10 8.09 4.89 5 3.79 3.83 2.61 1.38 1.49 1.53 No 0.14 0.38 0.44 0.66 data 0 ic ru a s e co p. an m a l ia e a da an ga ne di ny ag in bi bl an Re na Pe oc st ist an ra In m ne pi pu Ke er nz ki et jik or Ug Uk lo ab ilip Se av Pa Re Vi Ta Co M Ta Ar Ph ry an t, to ic yp va in Eg er m Do bs O Source: Based on World Development Indicators database 2018. Note: The net assistance as a percentage of gross national income (GNI) is calculated as a three-year moving average for the year in which substantial reform was first announced. “Washington Consensus” was coined in 1989, are relatively easy to observe and enforce com- as noted at the outset of chapter 1, to describe pared with other reforms such as liberalization 10 such policies supported by the United and independent regulatory practice, and thus States, International Monetary Fund, and are more tractable as targets of foreign aid. World Bank (Williamson 2005). Among these Although donor influence over the adoption of policies, two in particular are relevant to the reform agendas is evident in many countries, it power sector: the privatization of state-owned is also clear that countries that do not feel enterprises and the abolition of regulations ownership for such reforms can be adept at restricting competition. The World Bank paper reforms (through a policy paper or even (1993) explicitly recommended enforcing con- legislation) that are never enacted, or take ditions to make governments commit to power symbolic measures (such as the creation of a sector market reforms. Other multilateral regulator) (Pritchett, Woolcock, and Andrews development banks and bilateral donors 2010) that may have a lower political cost but adopted similar approaches (Williams and do not address the sector’s underlying prob- Ghanadan 2006). In addition, reforming coun- lems (Levy 2014). tries continued to rely on international finan- These kinds of cosmetic reform measures cial institutions for significant investment proj- have been termed “isomorphic mimicry” ects. World Bank support for the 15 observatory whereby countries mimic reform activity countries throughout the reform process is ­ without implementing any measures (Pritchett, shown in annex 3C. Woolcock, and Andrews 2010).4 In Tajikistan, Governments motivated by the need for for example, the country’s vertically integrated finance are more likely to follow the norms of utility, Barki Tojik, has a history of poor opera- their financiers. One study (Henisz, Zelner, and tional and financial performance. Finally, the Guillén 2005) suggests that privatization and government’s need to lessen its financial the establishment of regulatory agencies burden together with donor pressure led in ­ How Did Political Economy Affect the Uptake of Power Sector Reform? 83 2004 to the formation of a state commission. leadership gave reform its deep roots and The commission recommended a cautious momentum. restructuring, unbundling the utility over a In Ukraine, repeated reform efforts mirror long time period. In 2008 the country finally the influences of competing outside adopted a resolution to restructure, and several influences—principally the Russian Federation ­ subsidiaries and companies, all controlled by and the European Union. Since Ukraine’s inde- Barki Tojik, were formed. No meaningful pendence in 1991, following the Soviet Union’s unbundling occurred, however, and Barki breakup, the country’s geographic location has Tojik remains a fully state-controlled, vertically meant significant and countervailing political integrated utility. Prior to the 2008 resolution, dynamics. Ukraine tried early on, in the 1990s, Tajikistan passed several laws and resolutions to establish a wholesale competitive spot market referring to privatizing state-owned enter- in electricity. Although broadly reflective of prises, including Barki Tojik, but no steps were Ukraine’s movement toward a market econ- taken in that direction either. omy, this liberalization of the power sector Kenya’s power sector had two successive remained deeply at odds with the inherently waves of sector reform. What began in 1996 as oligarchic economy lacking in any kind a perfunctory, donor-driven reform process of competition. The reform failed. Twenty gave way, in 2006, to a second wave of deeper, years later, the same tensions are in place. home-grown reform. An aid embargo was Nevertheless, the plan to introduce a wholesale introduced in 1991 in response to Kenya’s electricity market resurfaced. This time the weak macroeconomic performance and gover- agreements are tied to integration with nance failures. Donor organizations insisted on the European Union, which in turn provides sector reform. Reluctantly, Kenya complied. momentum and incentive. Ukraine’s electricity The measures taken at that time nevertheless market and regulatory reform can converge laid important foundations—policy and regula- with the European Union’s acquis communau- tory functions were separated from commer- taire.5 The reform seems to be moving ahead. cial activities, and generation was unbundled Although donors have been important cata- from transmission and distribution. The lysts of reform and providers of ongoing sup- government introduced cost-reflective tariffs, ­ port, the dynamics of the reform process are liberalizing generation. The sector reform pro- largely driven by domestic political factors cess also benefited from significant technical (table 3.4). A glance at the experience of the 15 assistance and capacity building. Key sector observatory countries shows that, although officials were sent abroad on study tours to donor influence has been ubiquitous, it has not observe progress in other countries. In time, definitively shaped a country’s pace of reform or Kenya’s political and technocratic leadership ­ subsequent delivery. No statistically significant was persuaded of the relevance of power sector relationship can be found between aid depen- reform. A second wave of reform, initiated in dency and the depth of reforms undertaken. 2006, was led entirely by these converts to This suggests that reform implementation may domestic reform, champions who were able to be driven primarily by factors internal to the build on ­ foundations laid down in the 1990s. country, something we consider in the follow- They strengthened independent regulation, ing sections. Still, once a country begins moving partially privatized the national generation ahead with reform, donor support remains company, and created agencies to address critical to implementation. For example, the ­ renewable energy and energy access goals. World Bank has remained involved in nearly all Donor conditionality may have helped trigger of the reforms undertaken in the observatory power sector reform in Kenya, but local countries (see annex 3C). 84 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.4 Donor influence on patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Complete Sustained Extensive Cautious Minimal Limited Partial Rapid influence High Donor Moderate Statistical significance — — — — Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows the extent of donor influence on the process of reform across the sample of 15 country cases. The darker the shading, the higher the number of countries from the sample that fall into each combination of circumstances. The Chi-squared test was ­ applied to uncover statistically significant relationships between ideology and reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less significant); ** = 85–90 percent significance (somewhat ­ significant); * = 75–85 percent significance (low significance); — = not significant. Full statistical results are reported in table 3B.2 in annex 3B. Finding #3: Countries espousing market- The relationship between power sector oriented ideologies are more likely to reform and political ideology is not easy to implement and sustain bold reforms define. The 1990s power sector reform model, grounded as it is in neoliberal thinking, appears The donors recommended that we undertake to sit more easily within a market-oriented ide- reforms…. There was this idea that govern- ology. This tendency, however, is not clear cut. ment should not be involved in the business…. Some left-wing or even communist govern- A neighboring country came to us and said ments espouse the role of the market, and the that we should not accept the donor conditions private sector, in power sector reform (Senegal … the most that should be done was unbun- and Vietnam in the late 1990s), whereas some dling. We listened, but we kept quiet. Our right-wing governments rely on the state neighbor had a different background to us. (Rajasthan in India, and Senegal, in the 2000s). They had an interest in government providing The academic literature presents mixed evi- services such as electricity because of their dence on ideology as a driver in power sector socialist history. reform (Erdogdu 2013, 2014). The same mixed —Utility official results can be found in the case study litera- ture, which finds that ideology has shaped Ideology plays an important role in a politi- power sector reform in Africa (Gore and others cal economy. Some countries had a longstand- 2018), but not among Indian states (Cheng ing orientation toward either state dominance and others 2016). The reform narratives of the (Tajikistan, Tanzania, and Vietnam) or a ­ market 15 observatory countries likewise show that a economy (the Philippines) (table 3.5). Others subtle blend of ideology, historical legacy, and might lean more heavily in one direction or political perceptions explains power sector another over time and are best characterized reform, albeit not in a deterministic way or in as hybrids (India and Senegal). Despite the ways that fit into a simple left–right spectrum. importance of ideology, political parties across Tanzania’s strong brand of African the developing world do not necessarily divide ­ s ocialism has asserted itself by limiting along ideological lines. power ­ s ector reforms, notwithstanding its How Did Political Economy Affect the Uptake of Power Sector Reform? 85 TABLE 3.5 Observatory countries and states and their ideological orientation Ideological leaning Country/State Market-oriented Colombia, Peru, Philippines, Uganda Hybrid Dominican Republic, India (Andhra Pradesh, Odisha, and Rajasthan), Kenya, Morocco, Pakistan, Senegal, Ukraine State-oriented Arab Republic of Egypt, Tajikistan, Tanzania, Vietnam Source: World Bank 2019. official policy pronouncements. The spirit of reform. The state still plays a leading role in the President Julius Nyerere’s Ujamaa—Tanzanian economy in general and the power sector in African ­ socialism—and the Arusha Declaration particular, under communist party guidance. of 1967 remain powerful influences. The 1967 The introduction of the role of markets has been declaration outlined a vision of economic jus- deliberately steady, with the 2004 Electricity tice, a socialist command economy, nationaliza- Law, the basis for sector reform, taking more tion, self-reliance, and independence from for- than eight years to draft. The 2006 road map set eign private investment or foreign aid. Those out a careful rollout of power sector reform— principles set the foundation for a degree of designing each stage of the process as a pilot to Tanzania’s postindependence national unity, test, improve, and learn from, followed by full ethnic cohesion, and political stability, not uni- implementation. The approach has been grad- versal at the time in Sub-Saharan Africa. The ual and consensus-driven (within the frame- country’s lackluster economic performance, work of the political and sector leadership), as however, led to dwindling donor support in the well as sensitive to social harmony. Nevertheless, 1980s, prompting loan conditionality around this approach has meant that reform in Vietnam the liberalization of key economic sectors that, has come slowly. In essence, markets are a tech- by the 1990s, extended to the power sector. nocratic device rather than part of any ideologi- Public commitments were made in 1997 to pri- cal or political transformation. Implementation vatize the national utility, Tanzania Electric has been slow and much delayed. Supply Company Limited (TANESCO), and in In short, ideology affects both the pace at 1999 to fully unbundle the power sector. which reform is implemented and the extent to Nevertheless, 20 years later, the only major which announcements are followed through to reforms enacted in Tanzania are opening the full implementation. This observation is some- market to independent power producers (IPPs) what at variance with the academic literature and creating an independent regulator. The cited earlier in the chapter. A systematic look at power utility, TANESCO, remains a vertically the 15 observatory countries suggests, however, integrated state-owned monopoly, characterized that ideology is only weakly significant (at by weak financial and technical performance 20 percent)6 in explaining reform announce- and tariffs that are not cost-reflective. The suc- ments but becomes more important as the cessive governments have manifested varying reform process unfolds (table 3.6). Market- approaches to the liberalization and restructur- oriented ideology emerges as a strong driver for ing of the power sector. Without momentum or bold reform implementation (significant at the policy c­ ontinuity, public support for power sec- 1 percent level): half of the market-​ oriented tor reform appears neither deep nor wide. countries implemented bold reforms compared In Vietnam, the country’s longstanding with none of the state-­ oriented countries. state-orientation has shaped a more cautious Ideology also appears to be strongly related and incremental approach to power sector (again at the 1 percent significance level) 86 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.6 Ideology and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Complete Sustained Extensive Cautious Minimal Limited Partial Rapid Market Ideology Hybrid State Statistical significance * **** **** * Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows the extent to which a country’s prevailing ideology influences the subsequent process of reform across the sample of 15 country cases. The darker the shading of the box, the more countries from the sample that fall into each combination of circumstances. The Chi-squared test was applied to uncover statistically significant relationships between ideology and reform, with significant); the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less ­ ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low significance); — = not significant. Full results are reported in table 3B.3 in annex 3B. ­ riginal to whether a country fully delivers on its o We haven’t really done any power sector reform announcements: two-thirds of reforms at all even since 1998. All that hap- ­ market-oriented countries substantially followed pened in 1998 was that they unbundled the through on promised measures compared with national utility. So they unbundled it and none in the case of state-oriented countries. made several generation and distribution com- Ideology does not seem to be closely related, panies, but they were all still populated by however, to the sustainability of reforms. people from the national utility and still owned by the government. They were treated as one organization. Finding #4: Reforms can be introduced —Government official across a range of political systems but are more likely to progress in countries The nature of a country’s political system is where power is decentralized and relevant to power sector reform. Countries vary contestable according to the extent to which political power is sustained by a centralized decision maker There were huge fights, but they managed to (a unipolar system) or oscillates among a more make way for reforms because of the presi- dispersed set of actors (multipolar) (table 3.7). dent’s mode of work. He would engage people The 15 observatory countries present a range of for long hours, talk to the opposition to explain political systems. Some have had sustained uni- why liberalization and privatization made polar leadership of varying kinds—for example, sense—the meetings could run for a whole day the king in Morocco, President Museveni in and a whole night. In this way, he wore the Uganda, and the communist party in Vietnam. opposition down to accepting the approach. Others have been consistently multipolar, such The president would go through rounds and as the democracies of Colombia or India. Still rounds of discussions—with industry, mem- others went through major transitions in their bers of parliament, and others. The success of political s­ ystems during the reform process. the reform was probably because the president Peru transitioned from a unipolar system under took this role upon himself. President Alberto Fujimori to a multiparty  —Former senior official democracy. The Philippines followed a similar How Did Political Economy Affect the Uptake of Power Sector Reform? 87 TABLE 3.7 Political systems of observatory political and investment environment has not countries been conducive to further privatization of elec- Concentration of political tricity distribution; subsequent governments power have not been able to make progress on this First significant issue. Country reform 2015 Following major structural reforms under Arab Republic of Egypt, Unipolar Unipolar centralized rule in Peru, the subsequent Morocco, Tajikistan, Uganda, Vietnam d emocratic government focused efforts on ­ Transition Multipolar ­ second-generation reforms. In the late 1980s, Kenya, Pakistan, a Senegal, Tanzania Peru faced a severe macroeconomic crisis that Colombia, Dominican Multipolar Multipolar was mirrored by a power sector crisis. The Republic, India, Fujimori government that came to power in Philippines, Ukraine 1990 initiated a drastic program of stabilization Peru Unipolar Multipolar and structural reform; it encompassed a Source: Based on Rethinking Power Sector Reform utility ­database 2015. thoroughgoing reform of the power sector, ­ a. Pakistan transitioned through military and democratic including profound unbundling, privatization, ­ governments during this period. regulation, and liberalization measures. Following promulgation of a sector law in path after Ferdinand Marcos was ushered from 1991, 100 percent of transmission, 70 percent power. Pakistan, in contrast, has periods of of generation, and 45 percent of distribution democratic government punctuated with epi- were transferred to private ownership, man- sodes of military rule. The reform stories of all agement, and operations in the space of a few of these countries illustrate how shifts in politi- years. When democratic government returned cal systems affect implementation of power in 2002, attempts to privatize the remaining ­sector reform. distribution companies in the provinces met In Pakistan, distribution was privatized with civil unrest and were eventually dropped. during a period of martial law, but the legiti- Attention turned to several second-generation macy of this action remains in question. reforms designed to revive flagging investment Pakistan’s reform journey began in 1992 with levels in the sector. A new sector law in 2006 the publication of a Power Sector Strategic Plan reorganized the system operator, reformed that envisaged opening generation, unbun- planning and regulation functions, improved dling the national utility, privatization, and administration of the electricity market, and regulation. Almost all these measures were introduced regular supply auctions. This proac- implemented between 1994 and 1997 by a tive management of incremental reforms democratically elected government; privatiza- has proved successful in sustaining sector tion proved a sticking point. After the intro- performance. duction of military rule in 1999, a decision was In Senegal, a change in the political regime taken in 2005 to privatize Karachi Electric. The led to a reversal of the utility privatization pro- emergency conditions in force at the time put cess. It was the socialist regime (in power for all privatization rules and processes into abey- 40 years, from independence to 2000) that first ance, and the process therefore happened opened the power sector to private investment without prior deliberation and approval by the and decided in 1997 to begin privatizing the Council of Common Interest. The unions, vertically integrated utility, Senelec. The opposition political parties, and consumer rep- Senelec privatization eventually resulted in the resentatives have contested the privatization of award of a 25-year concession to a consortium Karachi Electric ever since. As a result, the of foreign private companies, but this decision 88 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD was followed by widespread labor unrest, per- power market reform is to depoliticize the sec- vasive blackouts, and the arrests of some labor tor by ceding control to autonomous bodies leaders. Elections held in the backdrop of pri- (the regulator and nongovernment actors such vatization and the consequent unrest led to the as the private sector), this course of action first-ever defeat of the Socialist party and might be inherently less attractive to a unipolar brought the liberal Senegalese Democratic leader. party to power. One of the first actions of this In practice, countries with unipolar systems new dispensation was to cancel Senelec’s con- seem more likely to drag their feet on imple- cession contract. A second attempt at privatiza- menting reform. The evidence from the tion subsequently failed, in part because of 15 observatory countries suggests that foot-­ weak private sector interest but also because dragging is dominant among unipolar systems political will was lacking. The rights of labor (table 3.8). The nature of the political system and public perceptions about the role of the appears to have no material impact on a foreign private sector in providing a public ser- country’s announced reform ambitions. There ­ vice played important roles, producing intense is, however, a strong relationship between ideological cleavages that themselves deter- multipolar political systems and bold reform ­ mined Senegal’s reform path. implementation (statistically significant at The manner in which a country’s political 6 percent), with multipolar countries almost system might shape its power sector reform twice as likely to implement bold reforms com- process is conceptually ambiguous. On the one pared with unipolar countries. Furthermore, hand, unipolar systems may find it more countries with multipolar systems are almost straightforward to implement power sector twice as likely to fully deliver on their original reforms should the central decision maker be reform announcements relative to unipolar convinced of the merits of this course of action. countries (statistically significant at the 15 per- On the other hand, given that the nature of cent level). These findings highlight nuances TABLE 3.8 Type of political system and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Complete Sustained Extensive Cautious Minimal Limited Partial Rapid Unipolar Political system Transition Multipolar Statistical significance — *** ** — Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows to what extent a country’s political system influences the process of reform across the sample of 15 country cases. The darker the shading of the box, the higher the number of countries from the sample that fall into each combination of ­ circumstances. The Chi-squared test was applied to uncover statistically significant relationships between ideology and reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low significance); — = not significant. statistical results are reported in table 3B.4 in annex 3B. Full ­ How Did Political Economy Affect the Uptake of Power Sector Reform? 89 beyond the results of studies involving larger The political authorities are very susceptible sample sizes, which find no correlation [to public opinion]. So when the street com- between competitive political systems and plains, the minister or DG gets sacrificed. power sector market reform elements in devel- Another one comes without that translating oping countries (Erdogdu 2014; Foster and into any significant improvements. In reality, others 2017). it’s the political authorities at the highest level, who should have refrained from inter- fering in the management of the company, by Finding #5: Reform champions putting in place the people that were really often emerge as pivotal figures who needed. help to ensure momentum toward —Senior government official implementation, particularly when In 2004, the relationship between the position supported by stable bureaucracies of PS [permanent secretary] and president The president has an interest in the power sec- changed. Previously they had been appointed tor since he came in a wake of a shortage in by the minister but, from 2004, the president supply, which turned into a political problem took on the responsibility of these appoint- to the former president. He does care about ments. In the past, the minister could have tariff reform, as a part of the subsidy reform worked with the minister of public service to policy. However, [on] the other reform dimen- “fire” the PS. This created a stronger and pro- sions and details the president has less role. He tected relationship between the technical has less clear impact on these aspects as he experts in the ministries and the president, may look to them as details which may not which the president used to get the technical attract his attention. Regarding the cabinet … advice he needed. they deal with the ­ program case-by-case and —Electricity company official not in a comprehensive plan with an inte- Reform champions sometimes emerge and grated vision. play a pivotal role in directing the reform pro- —Government official cess (see table 3.9). They are able to articulate Political class was not much involved; reform a clear vision regarding the endpoint of the was essentially a one-man decision. It did not reform, motivate action by critical players, trickle down to political class and civil ser- and use their political capital to address vari- vants. Once the political decision was taken at ous roadblocks that may emerge along the a high level, civil servants formed groups to way. Few studies are available on the role of take care of reforms. Two years after the individual leaders in power sector reforms reform agenda was adopted, the leader lost across a large number of countries.7 About elections. Even then, top political class was half of the 15 countries in the observatory committed but at bureaucracy level no per- had a clearly identifiable reform champion. sonal commitment was there owing to changes The nature of these champions varies from made there (by the new g ­ overnment) like the highest authority—the king in Morocco, energy secretary, etc. president in Uganda, or chief minister in some of the Indian states to technocrats in  —Former regulator the line ministries responsible for the sector. When a director general [(DG) of the utility] The reform narratives illustrate that champi- is there for 18 months, or a maximum of two ons are most ­ effective when they can rely on years, he can’t really do anything. strong and ­ s table leadership at senior 90 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.9 Overview of reform champions Country/State Champion(s) Role Dominican Republic President Fernandez With reform legislation stuck in congress, carried out almost the entire reform process through decrees and ministerial orders. India – Andhra Pradesh Chief Minister Naidu Despite strong political opposition, pushed through with reforms, driving adoption and implementation. India – Odisha Chief Minister Patnaik First political leader to buy in to 1990s model reforms for power sector in India. Implemented full spectrum of the reform including private sector participation in distribution, which several later reformers did not. Kenya Senior ministry and Convinced other actors and orchestrated their actions on basis of utility officials their conviction, technical credibility, and closeness to the president. Morocco Monarch Articulated a vision of reform and directed efforts of key institutions. Pakistan PML(N) and Party enacted comprehensive reform law in 1992 but lost power bureaucrats following year. On coming back in 1997, moved fast to implement various aspects of the law—regulatory, restructuring WAPDA. Peru President Fujimori Bought into the 1990s model reforms and implemented all aspects of restructuring, competition, and private sector participation in a span of a few years. Philippines President Arroyo Put her weight behind the EPIRA legislation to ensure it would get the votes necessary to be adopted as law. Uganda President Museveni Bought into reform, drove implementation, and persuaded skeptical stakeholders to collaborate. Source: Based on Rethinking Power Sector Reform utility database 2015. Note: EPIRA = Electric Power Industry Reform Act; PML(N) = Pakistan Muslim League (Nawaz); WAPDA = Water and Power ­Development Authority. technical levels—­ p ermanent secretaries, reform process in the country. As institutional directors general, chief executive officers, leadership of power sector reform passed more regulators, and sometimes ministers. In some to the Ministry of Energy and Mineral instances, such as Kenya, the technocrats Development, the role of the top bureaucrat themselves have championed reform. The there (who had been in office for 20 years) power sector reform is, in other words, highly was pivotal. The long-term close relationships technical and highly political at the same between such technocrats and the president time. clearly determined both the delivery of In Uganda, reform momentum was main- Uganda’s power sector reform and the way in tained by strong technocrats in key ministry which it progressively deepened over the years positions enjoying strong support from and adjusted to evolving circumstances. ­ political leadership. Circa 1990, the Ministry In the Philippines, President Gloria Arroyo of F inance, Planning, and E conomic used her scant political capital to ensure that Development laid the legal and policy ground- the power sector reform legislation would pass work for public enterprise reform and divesti- the congress. When Arroyo took over in ture (which formed the basis for later power January 2001, the country was still reeling sector reform), because that was deemed to be under the effects of the Asian financial crisis of critical to macroeconomic stabilization and 1997–98, with a depreciating currency and recovery. Top bureaucrats in the ministry, who exodus of capital. On the political side, her remained in office for more than two decades party leader and former president Joseph (1992–2013), were critical to the overall Estrada had been impeached, leaving her with How Did Political Economy Affect the Uptake of Power Sector Reform? 91 limited space to maneuver. The Estrada gov- 10 years, and has enjoyed strong political sup- ernment had presented a comprehensive port in removing obstacles to implementation power sector reform bill in the congress, but of the strategy. the changing political environment meant The Indian civil service, in contrast, sees it was going nowhere. Arroyo prioritized the much more rapid turnover of critical positions, bill when she entered office, using all her making it more challenging to follow through ­ political capital to get it passed in the congress on reform implementation at the state level. despite several attempts to destabilize her gov- Senior bureaucrats at the state level are part of ernment and impeach her. As a result, the a civil service; members are recruited nation- Electric Power Industry Reform Act (EPIRA) ally and can be seen as a permanent executive was signed into law and became effective in branch of the government. These officers are July 2001. generalists and move from one ministry to In Morocco, the king provides the political another, separate from the political system, but vision for the sector as well as ensuring the all bureaucratic appointments, from the power stability of the technical cadre. Morocco is an secretary and the chief executive officer of the active albeit constitutional monarchy that state-owned utility to the top police officers in provides a degree of autonomy for the prime every city, are made by the state government. minister, cabinet of ministers, and parliament. Consequently, officers are usually moved from The monarchy typically exercises ­ s trategic one ministry to the other after a period of two leadership and has played a central role in to three years. Each new political dispensation articulating a long-term vision for the power shuffles the top bureaucrats at the ministries. sector. In general, the most senior leaders of This discontinuity limits the benefits of long- the power sector are appointed by the king term technical staff who can buy into and push and require his full support to remain in reforms. office. As such, the average ­ tendency is for The impact of reform champions is evident key technical leaders and staff to have long in the pace of reform implementation and tenures (even if some individuals can be rap- helps to ensure that reform announcements idly removed if judged ineffective), and to are fully delivered (table 3.10). The presence enjoy high-level political ­ support. Key initia- of a champion does not affect a country’s pub- tives in the power sector—such as its initial licly avowed level of ambition on reform. It opening to the private s ­ ector, a drive to matters significantly, however, for the success achieve universal electricity access, and a stra- of implementation. Countries lacking a reform tegic shift to renewables—are initiated at the champion were four times as likely to stall on highest level. Follow-up implementation is the implementation of reforms (significant at the responsibility of the technical leaders, but the 15 percent level). The impact is even royal oversight helps remove ­ roadblocks and starker when it comes to delivery. More than keeps the overall process on schedule. The half the countries lacking a reform champion implementation of Morocco’s renewable largely failed to deliver on any of the reform energy strategy by the Moroccan Agency for announcements, whereas all those with Sustainable Energy (MASEN) is a classic reform champions made considerable prog- example of that process. The president of ress (significant at the 2 percent level). MASEN has a strong ­ p rofessional back- Champions do not seem to be so helpful, ground in financial sector leadership; he was however, when it comes to the longer-term appointed by the king to get financing for sustainability; perhaps because they are not ­ challenging projects, has held office for almost around. 92 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.10 Reform champions and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Complete Sustained Extensive Cautious Minimal Limited Partial Rapid champion Strong Reform Weak Statistical significance — ** **** — Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows to what extent the strength of the reform champion influences the subsequent process of reform across the sample of 15 country cases. The darker the shading of the box, the higher the number of countries from the sample that fall into each combination of circumstances. The Chi-squared test was applied to uncover statistically significant relationships between ideology and reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low significance); — = not significant. Full results are reported in table 3B.5 in annex 3B. Finding #6: Unless champions can owned by the State and more than that, it [is] orchestrate wider stakeholder being subsidized by the State. On the other alignment, the sustainability of reforms hand, it does not have sufficient information to may be called into question address the reform deeply. In general, and due to its inability to convey the message of reform There are vested interests that take advantage they are more tending to status quo, which is of the situation in the power industry. When more understood to them. The reform is not a you talk of reforms, you mean change. priority to the media. Also, it shows the lack of However, there are people who benefit from capacity of the state to communicate to the the status quo, especially the utility’s situation public regarding the reforms. (such as p­ oliticians, suppliers—especially fuel suppliers). These vested interests do not sup-  —Government advisor port reform. Unfortunately, they are also Reform champions alone cannot ensure the placed in positions of influence. success of a reform process. Power sector  —Former utility official reforms are often initiated by a political leader or technical-level reform champion, or by The entity responsible to do the reform hesitated one particular institution; and they can due to the fear of public/employees’ reaction. achieve quite a lot and impart momentum to Furthermore, they are not convinced about the the reform program. To deepen and widen timeframe for the reform. Therefore, they are the reforms, and to sustain the momentum, a ­ normally on the side of postponing the reform. wider range of stakeholders and interests needs leadership level from There is reluctance on the ­ to be engaged; compromises reached between information ­disclosure. contesting views, a coalition formed, and levels  —Government advisor of communication heightened to directly Regarding the media, on one side it is loyal to involve stakeholders and often also the public. the government since a substantial part is The degree of interaction between people and How Did Political Economy Affect the Uptake of Power Sector Reform? 93 institutions needs to intensify to keep the their power and influence. Labor unions fear reforms moving (Lee and Usman 2018). The loss of jobs, competing ministries fight for public may come to oppose electricity privat- influence, consumer groups may oppose ization or foreign private investment because any tariff hikes, and regional utilities or of various negative associations (Sen, Nepal, subnational leaders want to avoid losing con- ­ and Jamasb 2016; Zelner, Henisz, and Holburn trol (table 3.11). How these frictions are 2009;). It is also evident, however, that public resolved may determine the direction of the participation in power sector policy processes reform. In some cases, such as Colombia and has been limited in many recent developing Peru, local governments wielded enough country cases, even where civil society is active, power to shape the national reform process by such as in India (Dubash 2002; Nakhooda, forcing the central government to compromise Dixit, and Dubash 2007). on some measures. In other cases, such as the Power sector reform touches on many com- Indian states of Andhra Pradesh and Rajasthan, peting interests and inevitably creates losers as consumers were powerful enough to stop tariff well as winners. Regulatory tariff adjustments reform undermining utility financials. designed to achieve cost recovery will make In Tajikistan, vested interests among key electricity less affordable for key constituencies. stakeholders have consistently blocked the Gains in the commercial and operational path of reform. One key political economy ­ efficiency of utilities improve their financial issue has involved the national utility, Barki performance but may also threaten jobs or pre- Tojik, and its inability to recover costs from two vent pilferers from availing themselves of critical and energy-intensive state-owned enti- “free” electricity through network theft. ties, the aluminum company, TALCO, and the Reform of power markets may spell the loss of Agency for Land Reclamation and Irrigation, business for those supplying fuel to private which provides irrigation services to farmers. generators. Privatization of power utilities may Neither entity is charged a cost-recovery tariff create new commercial opportunities, but for electricity, and their substantial accumu- those new opportunities may also spell the loss lated debts are repeatedly forgiven. There are of political patronage and bureaucratic influ- at least two political economy explanations for ence and ultimately benefit foreign commercial these large implicit subsidies. One is that the interests. It follows that friction can and does government gains legitimacy through support- emerge among different stakeholder groups, ing employment in these entities, because any which can sometimes threaten to derail reform downsizing might expose them to market con- attempts. These problems may be at least par- ditions. The other explanation is that those tially mitigated by aligning stakeholder inter- entities, although nominally state-owned, ests through consensus-building dialogue and exhibit beneficial ownership by the ruling elite potentially compensatory measures, such as and their political and business allies. As such, labor protection or generous redundancy pack- key political interests would be threatened by ages, or safety net subsidies to protect the any deep structural reform in the power sector ­ poorest from tariff increases, or by adjusting that would seek financial discipline, cost recov- the design of privatization programs to create ery, operational autonomy for unbundled greater opportunities for the domestic private sector entities, or private participation. If deep ­ sector. power sector reform involved the state giving As a result, an integral part of power sector up control over the levers that provide the reform is managing friction among stakeholder large subsidies, the beneficial owners would groups. Stakeholder groups can have different lose the influence that ensures their very impacts on the reform process depending on ­ benefits. This dynamic in Tajikistan’s political 94 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.11 Stakeholder friction in power sector reforms: An overview Country/State Stakeholders Nature of conflict Colombia Municipal utilities Successfully opposed restructuring and struck a compromise to perform under new guidelines of market power. Result is a mixed ownership sector. Dominican Republic Political actors Reluctance in implementing reforms due to loss of patronage opportunities led to prolonged delays in implementing reform legislation. Egypt, Arab Rep. Political actors Reluctance to implement tariff reforms associated with power sector reform due to fear of civil unrest. India – Andhra Pradesh Farmers Strong political force, successfully lobbied for free electricity undermining sector financials. India – Odisha Utility employees State electricity board employees opposed privatization but were won over with employment guarantees. India – Rajasthan Consumers Successfully lobbied for a tariff freeze undermining utility financials. Kenya Political actors Exerted influence over regulator to control tariffs during politically sensitive times such as elections. Morocco Municipal utilities Successfully opposed reform legislation and a full market model because they use revenue from electricity businesses to balance losses from sales of water. Pakistan PEPCO The body established to implement restructuring and privatization in sector in 1996 still exists and wants to maintain its role, thereby slowing the reform process. Peru Provincial governments Regional utilities and politicians were opposed to privatization and successfully blocked attempts to privatize provincial power utilities. Philippines Oligarchs Oligarch industry players with congressional affiliations exercised preference for hybrid state/market approach with restricted competition. Senegal Labor unions Actively opposed privatization of the power utility. Tajikistan Economic elite Oppose any reform linked to tariff subsidy because it hits their economic interests in industry and agriculture. Tanzania Political actors Large number opposed to private sector involvement in power sector. Uganda Incumbent utility Incumbent staff and senior executives were opposed to restructuring and privatization but failed to stop process. Ukraine Economic elite Economic and political elite acquired controlling stakes of utilities through the privatization process. Vietnam Party members Serious disagreement between factions in party on reform path led to prolonged debate and delays in implementation. Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PEPCO = Pakistan Electric Power Company. economy has so far been sufficient motivation stakeholder engagement was undertaken with to block deep power sector reform in the unions and employees but did not extend to country. electricity consumers. Farmers, in particular, Across a number of Indian states, reforms retain a deep sense of entitlement to free promoted by influential chief ministers often power. The issue of electricity subsidies for flounder at later stages under their successors, agriculture became politically charged, because of a lack of wider support. This pattern overriding the reform process and leading to ­ repeats itself in the states of Andhra Pradesh the ­government’s electoral defeat in 2004–05. and Rajasthan. The policy of free power for farmers persists to In Andhra Pradesh, the then–chief minister this day and has debilitated the financial stand- eschewed privatization but pursued tariff ing of power utilities in the state. reforms and information technology invest- In Rajasthan, reforms of the power sector ments aimed at enhancing the efficiency and were initiated at the turn of the century. The financial viability of the utilities. Extensive objective was the financial turnaround of the How Did Political Economy Affect the Uptake of Power Sector Reform? 95 sector using tariff reforms and privatization of contracting for short-term rental power, the distribution business. The reforms were importing liquefied natural gas, and building driven by a chief minister motivated by fiscal new gas-fired plants. Once these fundamentals concerns, who nevertheless did not lay the were in place, the government restored the necessary groundwork of stakeholder engage- financial equilibrium of the power sector ment. After a new government came to power, through a five-year rising tariff trajectory. The the reforms stalled in 2004–05. Tariffs were initial price increases of electricity and major frozen for the next eight years, with an accu- categories of fuels (40 to 78 percent) were mulated debt of up to US$14 billion by 2014. intended to reduce its subsidy burden. The Colombia’s experience shows how reforms magnitude of the reform and consumer can be adjusted to accommodate the compet- sensitivity to energy price hikes meant the ­ ing interests of national and municipal govern- government had to develop an effective com- ­ ments, thereby achieving the necessary con- munication strategy. The campaign highlighted sensus to progress. Colombia’s power sector the inherently inequitable nature of the subsi- has long combined national-level power utili- dies, explaining how they benefited the rich ties, controlled by central government along disproportionately (Moerenhout 2018). At the with powerful municipal utilities—notably, same time, the government communicated Empresa de Energia Electrica de Bogota (EEEB) that fiscal savings from energy subsidy reform in Bogotá and Empresas Públicas de Medellín would be at least partially redirected to social (EPM) in Medellín—that together accounted programs, which up until that time received for about 40 percent of electricity distribution. fewer resources than energy subsidies. The There had long been tension over the alloca- emphasis on social spending was in line with tion of roles between central government and the country’s new constitution. This strategy municipal actors in the sector, and these came helped to forge a social consensus around tariff to a head when the sector reform laws of 1994 reform and staved off unrest. Other countries’ called for unbundling and privatizing the utili- experience with communication campaigns is ties. The municipal governments objected. summarized in box 3.2. A compromise allowed EPM and EEEB to Finally, in some instances, stakeholders can remain vertically integrated public utilities as lend support to reforms that provide opportu- long as they separated the accounting of their nities to advance their own interests, as hap- generation and distribution activities and pened in Ukraine. The period of 2000–02 saw abided by new restrictions on market shares in several dubious privatization efforts in Ukraine these activities. EPM flourished under the new stall the entire reform process. Large business regulatory framework and remains a vertically interests used their clientelist relationship with integrated publicly owned utility and one of senior government officials to secure owner- the main actors in the power sector. EEEB, in ship of generating assets and distribution com- contrast, was unable to turn its performance panies. In a sector plagued by nonpayment by around and eventually underwent vertical sep- distributors in the wholesale electricity market, aration and privatization. the issue of dues was huge at every level. Egypt employed an effective public commu- The wholesale electricity market owed money nications campaign to explain the need for to generators, which owed money to fuel electricity tariff reforms. It preempted con- suppliers. Meanwhile, the government’s own ­ sumer opposition. In 2014 Egypt experienced fuel-supply companies took the state-owned major electricity supply outages that led to civic generation companies to court, and the court unrest. The government moved swiftly to promptly forced the sale of company assets to address concerns about security of supply by settle the debt. These assets were sold in a 96 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 3.2 Importance of public communication strategy to support reforms International experience has shown that communication is critical to the success of major economic reforms. If an effec- tive communication program is not implemented before, during, and after reform measures go into effect, it is difficult to earn the public’s trust and foster understanding of the political decisions that underpin the reform. A well-researched communication program with informational, attitudinal, and behavioral objectives can enhance the effectiveness of reform. An effective communication campaign involves mapping key stakeholders, using outreach and two-way dialogue with citizens, conducting opinion research, consulting with stakeholders, creating and testing compelling messages that build awareness of reform benefits, assigning credible messengers, identifying good channels of communica- tion, coordinating within government, setting strategic goals, and communicating consistently with evidence-based messages. Public reactions to reform programs are highly contextual and dynamic. A well-informed public understands the rationale for reform and greatly improves the likelihood of success. Some successful examples are listed below. In Vietnam , public acceptance of an electricity tariff reform was facilitated by a communication strategy that focused on raising awareness about the rationale for tariff hikes. A capacity gap analysis was conducted to identify ways to improve Vietnam Electricity’s (EVN) communication efforts, and opinion research was conducted to under- stand public sentiments. Iraq undertook a qualitative assessment of key consumer perceptions and awareness levels of tariffs and subsi- dies. The assessment mapped stakeholders and analyzed the audience; surveys and focus groups gathered informa- tion on citizens’ views. The government strengthened its internal coordination to deliver consistent and convincing messages. Ukraine rolled out a communication strategy in support of stiff tariff hikes. The rollout included a 30-second public service announcement that aired on 19 TV stations across Ukraine and appeared on 15 government websites. The announcement was rooted in a detailed understanding of public perceptions; the key messages reflected the public’s concern. The effort included involving local media in major cities and making them aware of the energy sector status. In Belarus, focus group discussions and a stakeholder mapping found that opposition to tariff reforms stemmed from the lack of knowledge consumers had about tariff-setting policy and reform processes. A communica- s trategy was designed to support efforts to better engage consumers in the governance of district heating tion ­ providers. Workshops were held with local service providers to build capacity for improved public communication ­ with consumers; well-designed graphics helped to explain reforms and the benefits of energy efficiency. These efforts—and clearer and more transparent heating bills—helped mitigate resistance to higher tariffs. Source: Worley, Pasquier, and Canpolat 2018. nontransparent manner, leading to allegations on the initial level of reform ambition, or on of corruption in a process termed “asset strip- the pace of reform implementation, or on ping.” Auctions were announced in random the extent of delivery of reform measures. fashion. In one case in 2001, about 4,000 Where it appears to make a difference is in megawatts of thermal units were sold for as lit- the sustainability of reform. Among coun- tle as US$38 million. The same interests that tries without stakeholder alignment, about had acquired the generation assets went on to one-third undergo privatization reversals, take controlling stakes when the distribution compared with none among countries that companies were privatized—the result of weak achieved stronger alignment (significant at corporate laws that accorded private sector the 11 percent level). control of the companies with shareholding as low as 26 percent. This corrupt process became Finding #7: Legislation is usually known as shadow privatization. a necessary statement of political Achieving stakeholder alignment helps commitment to reform that helps to safeguard the sustainability of reform mea- support longer-term sustainability sures (table 3.12). Stakeholder alignment does not appear to be critical at the early There was the Sector Policy Paper closely fol- stages of reform. It has no material impact lowed by the Energy Act, with clear guidelines How Did Political Economy Affect the Uptake of Power Sector Reform? 97 TABLE 3.12 Stakeholder alignment and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Complete Sustained Extensive Cautious Minimal Limited Partial Rapid Stakeholder Strong alignment Weak Statistical significance — — — ** Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows to what extent the degree of stakeholder alignment influences the subsequent process of reform across the sample of 15 country cases. The darker the shading of the box, the higher the number of countries from the sample that fall into each combination of circumstances. The Chi-squared test was applied to uncover statistically significant relationships between ideology and reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low significance); — = not significant. Full results appear in table 3B.6 in annex 3B. on what needed to be implemented—that is Legislation is often a critical step in crystaliz- why things went the way they did. ing political commitment to reform and paving the way for implementation. Drafting a legal —Former senior government official framework is important for working out the The reform has had achievements that can implementation details of a reform process, show that the … general law of electricity was while the process of approving legislation nec- a major accomplishment, it is a good law, essarily engages a range of political actors and although there are some things that need to be becomes one important vehicle for consensus modified; but it is a good law. The country building. Global-level analysis suggests that received good investment through the reform legislation enabling the introduction of IPPs, process, but a new wave of reform has to be for example, helps to attract private investment made; but above all, we must change the cli- in the power sector from domestic and foreign mate of legal security, law enforcement, and sources (Urpelainen and Yang 2017). Among improve the institutional framework for new the observatory countries, most of them intro- investments coming, and let them flow. And duced legislation as a foundation of reform. the state should stop interfering in everything The process was not always straightforward, and stop politicizing everything. If you are however. In some countries, the sector law was investing in new companies, or if you are not enacted until many years after the policy staying with your companies; it has to let insti- ­ commitment, as in the cases of the Dominican tutions work, you cannot be a judge and part Republic and Egypt (see table 3.13). of your institutions; do not obstruct invest- The benefits of legislation are felt primarily ment, because the state may take up to a year in safeguarding the longer-term sustainability to grant a concession to a plant, or two years. of reforms (table 3.14). The presence of strong That’s crazy. sector legislation does not seem to affect the ambition of reform announcements, or  —Sector official the speed or efficacy of implementation 98 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.13 Legislation and power sector reform: An overview Years Country Foundational legislation elapseda Comments Colombia Laws 142 and 143 0 Severity of macroeconomic crisis and power shortages in 1990s prompted immediate legislative action. Dominican Electricity Law 2001 8 Law introduced in congress in 1993 but faced opposition on Republic privatization and languished in the legislature for eight years. Egypt, Arab Rep. Electricity Law 2015 15 Committed to reform in 2000, but political momentum subsided, and full legislation was deferred until 2015. India – Andhra Andhra Pradesh Electricity 4 State government instituted a committee to recommend a Pradesh Reforms Act 1999 reform path in 1995 feeding into eventual legislation. India – Odisha Orissa Electricity Reform 2 Government began reform process in 1993 with assistance Act 1995 from the World Bank and U.K. Department for International Development. India – Rajasthan Rajasthan Power Sector 1 Because other states had drafted similar legislation for Reform Act 1999 reforms, Rajasthan took limited time to do the same. Kenya Electricity Power Act 5, 14 Relatively weak legislation was passed five years after 1997, Energy Act 2006 reform commitment; the legislative foundation took nine years to materialize. Morocco Law No 57-2009, Law n.a. No conventional sector reform laws; focus on role of MASEN No-37-2016 in promoting renewable energy. Peru Law for Power 1 The country committed to overhauling its economic policies Concessions 1992 to come out of the 1991 macro crisis. Philippines Electric Power Industry 5 Original legislation tabled in 1996 did not pass but was Reform Act 2001 adopted after refiling in 1998. Uganda Electricity Act 1999 1 The sector strategy was created in 1998 and in 1999; the government followed up with legislation. Vietnam Electricity Law 2004 8 The original draft law was created in 1996 and went through 25 versions before being enacted as law in 2004. Source: Based on Rethinking Power Sector Reform utility database 2015. Note: MASEN = Moroccan Agency for Sustainable Energy; n.a. = not applicable. a. Elapsed time between announcement of power sector reform and enactment of the legislation on which the reform is based. TABLE 3.14 Legislation and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Legislation Strong Weak Statistical significance — — — * Source: Based on Rethinking Power Sector Reform utility database 2015. Note: This table shows to what extent the strength of the legislative framework influences the subsequent process of reform across the sample of 15 country cases. The darker the shading of the box, the higher the number of countries from the sample that fall into each combination of circumstances. The Chi-squared test was applied to uncover statistically significant relationships between ideology and reform, with the results reported as follows: **** = 95–99 percent significance (significant); *** = 90–95 percent ­ significance (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low ­ significance); — = not significant. Full statistical results are reported in table 3B.7 in annex 3B. How Did Political Economy Affect the Uptake of Power Sector Reform? 99 and delivery. Countries with strong legislation, disagreements within Vietnam’s leadership on however, have a much higher chance of the respective roles of state-owned enterprise s ustaining those reforms (significant at ­ and the party in the 1990s and early 2000s. 21 percent). Although this debate was not fully settled by In the Dominican Republic, difficulties in 2004, the balance of opinion among stakehold- passing new sector legislation meant that major ers had shifted toward more competition in the reforms were undertaken on a fragile legal sector, and the 2004 Electricity Act was passed. basis that was vulnerable to reversal. Following This act resulted in the restructuring of EVN, the decision to reform the power sector, a new increased private sector participation in power reform bill was prepared in 1992 incorporating generation, and the initiation of a path toward the standard reform prescriptions of unbun- a wholesale market. dling, privatization, and liberalization. The bill was tabled in the legislature in 1993 but was LOOKING AHEAD opposed by several members who wanted to Adoption of disruptive technologies will maintain partial state ownership of sector depend on the political interplay of winners assets, and it did not pass the body till 2001. In and losers. As countries prepare themselves for the absence of comprehensive legislation, the an uncertain future in the face of rapid techno- government went ahead with reforms through logical change in the power sector, there are decrees and ministerial resolutions. In 1998, questions about so-called disruptive technolo- decrees were used to unbundle the national gies and their impact on the sector’s institutions utility into seven companies and create a sector and structures. As with historic experience of regulator with an associated regulatory frame- reform, the speed and scope of innovation in work. In 1999, three distribution companies any given country’s power sector will likely be and two generation companies were privatized shaped by political dynamics and lobbying by with 50 percent stakes, with power purchase potential winners and potential losers. agreements extending to 2003; by 2001, the New technology will disrupt not only indus- wholesale electricity market had begun opera- try cost structures, business models, and regu- tion. Owing to weaknesses in the legal frame- latory instruments, but also the political econ- work, political interference remained strong omy of the sector. New winners and new losers and the government refused to allow tariff will emerge from the change process, develop- adjustments under the new regulatory system. ments that go to the heart of political economy All this led to reversing the privatization of two dynamics. A lot will depend on how each of the distribution utilities by 2003 and the player sees its interests affected by disruption, third one in 2009. and whether each makes common cause In Vietnam, the final version of the (forms coalitions) with those it perceives as fel- Electricity Law took years to materialize, but low stakeholders (table 3.15). Some players the long gestation ensured that various stake- will see the future as theirs. They will support holders were on board in the end. Work on the energy transition. These players could drafting a new comprehensive electricity law include storage providers, prosumers, renew- began in 1996; over the next eight years, 25 dif- able energy IPPs, mini/microgrid operators, ferent versions would be prepared. The Ministry and electric vehicle owners/providers. In con- of Industry led the drafting of the law and the trast, conventional IPPs and fuel suppliers team included representatives of Vietnam could oppose the transition (particularly the Electricity (EVN) and various other govern- IPPs) out of fear, anticipating that their assets ment departments and ministries. The lengthy could become stranded and perhaps inade- discussion over the bill reflected deep quately compensated. 100 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3.15 Political economy analysis: How key stakeholders respond to technological disruption Change in Change in influence/ Reasons to resist Reasons to support Possible coalition interests authority change change partners Regulator More defensive Reduced role as competition Erodes power Carve out new role RE IPPs, utilities, increases (seen as out of prosumers, date) fractalists, storage providers Incumbent utility More defensive Diminish somewhat Stranded assets, If opportunity to Conventional IPPs, threat to revenue control profit from fuel suppliers, disruptive techs regulator if threat dominates Conventional IPPs Become less Depends on strategy, pro- or Loss of business Switch to RE, storage, Depends on conventional antichange opportunities, loss fractalism strategy of revenue from dispatch Renewable IPPs Not so interested in Increases as they become Unused to market Bigger opportunities Incumbent utility, pursuing subsidies more valuable/needed risk regulator Renewable Much bigger Changes with market growth None Bigger opportunities Fractalists, minigrids market opportunity telecom, mobile money, desperate fuel suppliers Consumers/ Sell electricity, Gain market clout, more Some might lose Opportunity for Minigrids, prosumers consume for motivated as citizens/voters subsidies or other better/cheaper fractalists, transport, higher privileges electricity, maybe to regulator stake in reliable sell also electricity Source: World Bank 2019. Note: IPP = independent power producer; RE = renewable energy. Particularly intriguing will be the stance of considerably. Much of this will be shaped by incumbent utilities; some might see the energy the existing market and regulatory structures transition as an insuperable threat to be set up to enable the existing electricity produc- resisted, whereas others may see it as a huge tion model. Declining technology costs could opportunity to be pursued. Utilities that are create a viable alternative to the utility struc- able to remain relevant need not necessarily ture, and the ability to integrate renewable be the losers and may even transition out of technologies, while maintaining a reliable their legacy roles into entirely new identities. ­ electricity system, will be an ongoing c­ hallenge. Much of this will depend on existing electricity How countries respond to these challenges production models and the role utilities may or will determine how competitive they will be may not play in owning generation resources. economically, their attractiveness as an invest- Utilities may also resist the adoption of disrup- ment destination, and their credibility as tive technologies, seeing them as a threat to government-managers of technology change ­ their viability. Unless regulators can adapt rate and innovation. One thing seems clear: how a designs to support new forms of cost recovery, country handles its energy transition, or fails utilities are unlikely to innovate on their own. to do so, will be a really big deal. Depending on the pace of adoption of distrib- uted energy resource solutions, utilities may CONCLUSION only be able to resist for so long. By examining relationships among the power Countries are likely to follow different sector actors, one can detect the range of politi- energy transition paths, because their institu- cal factors driving power sector reform tional receptivity to disruption will vary (table 3.16). Power sector reform takes a How Did Political Economy Affect the Uptake of Power Sector Reform? 101 TABLE 3.16 Political drivers of each stage of the reform process: An overview Stage of reform Political drivers Announcement Implementation Delivery Sustainability Crisis — — — — Donors — — — — Ideology * **** **** * Political system — *** ** — Reform champion — ** **** — Stakeholder alignment — — — ** Legislation — — — * Source: Based on Rethinking Power Sector Reform utility database 2015. Note: The Chi-squared test examines whether it is possible to reject the null hypothesis of no statistically significant relationship between the existence of each of the following factors: (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the longer-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. **** = 95–99 percent significance (significant); *** = 90–95 percent significance (less significant); ** = 85–90 percent significance (somewhat significant); * = 75–85 percent significance (low significance); — = not significant. long time. It typically begins with ambitious innovation in the power sector is introducing a announcements of reform intentions. Imple- new set of actors and eroding the position of mentation progresses at varying paces that may institutions that have been the traditional pro- or may not fully deliver on original promises. tagonists of the sector. Although the net effect Delivery is sometimes subject to reform rever- of these changes remains uncertain, it is clear sals later on. Most countries embarked on that these new influences will alter the political power sector reform in response to a crisis and landscape of reform. typically in response to donor pressure. Although these external factors undoubtedly ANNEX 3A. GLOBAL POWER played a catalytic role in the decision to reform, SECTOR REFORM INDEX the subsequent trajectory of reform primarily The standard package of reforms prescribed by reflects internal factors. Of these factors, ideol- international donors in the 1990s included ogy appears to have by far the strongest effect, four principal components: restructuring (ver- particularly on the pace of implementation and tical and horizontal unbundling of power util- the extent of ultimate delivery. The nature of ities), private sector participation, creation of the political system also becomes relevant at the an independent regulator, and competition in implementation stage, in that countries with power generation. centralized power structures tend to make less In order to aggregate across the four dimen- progress with reform. Finally, the approach to sions of power sector reform considered in this reform within the power sector itself also turns study, a simple Power Sector Reform Index is out to be important. Reform champions make a constructed. The index gives each country a significant ­difference in following through on score on an interval of 0 to 100 on each dimen- implementation. Longer-term sustainability sion of power sector reform. The scores are appears to rest, however, on stakeholder align- based on giving equal weight to each step on ment and actual legislation. each dimension of the reform continuum (see Looking ahead, technological disruptions table 3A.1). The average of the four 0–100 will be reshaping the political economy dynam- scores is used to provide an overall summary of ics of the power sector. The recent wave of the extent of reform. 102 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3A.1 Power Sector Reform Index Regulation No regulator Regulator =0 = 100 Restructuring Vertically integrated Partial vertical Full vertical unbundling Vertical and horizontal =0 unbundling = 67 unbundling = 100 = 33 Competition Monopoly Indepent power Single-buyer model Bilateral contracts Competitive market =0 producers = 50 = 75 = 100 = 25 Private sector 0.5 × (percentage of generation capacity with private sector participation) participation + 0.5 × (percentage of distribution utilities with private sector participation) ANNEX 3B. CHI-SQUARED CONTINGENCY TABLES TABLE 3B.1 Crisis and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Severe 0.47 0 0.12 0.18 0.29 0.12 0.24 0.18 0.18 0.47 0.12 Crisis Moderate 0.18 0.12 0.06 0.06 0.18 0.12 0.12 0.18 0.06 0.29 0.06 None 0.06 0 0 0 0 0.06 0 0 0.06 0.06 0 Chi-squared test 34% 55% 52% 88% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. TABLE 3B.2 Donor impact on patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid influence High 0.59 0.12 0.18 0.24 0.35 0.29 0.29 0.29 0.29 0.71 0.18 Donor Moderate 0.12 0 0 0 0.12 0 0.06 0.06 0 0.12 0 Chi-squared test 62% 28% 62% 49% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. How Did Political Economy Affect the Uptake of Power Sector Reform? 103 TABLE 3B.3 Ideology and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Market 0.24 0 0 0.12 0.12 0.00 0.18 0.06 0 0.24 0 Ideology Hybrid 0.41 0.06 0.06 0.12 0.35 0.06 0.18 0.29 0.06 0.35 0.18 State 0.06 0.06 0.12 0 0 0.24 0 0 0.24 0.24 0 Chi-squared test 19% 1% 1% 20% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. TABLE 3B.4 Political system and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Unipolar 0.18 0.06 0.12 0.06 0.06 0.24 0.12 0.06 0.18 0.35 0 Political system Transition 0.12 0.06 0.06 0 0.18 0.06 0 0.12 0.12 0.18 0.06 Multipolar 0.41 0 0 0.18 0.24 0 0.24 0.18 0 0.29 0.12 Chi-squared test 27% 6% 15% 37% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X ­percent level. TABLE 3B.5 Reform champions and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid champion Strong 0.41 0.06 0.06 0.12 0.35 0.06 0.24 0.29 0 0.41 0.12 Reform Weak 0.29 0.06 0.12 0.12 0.12 0.24 0.12 0.06 0.29 0.41 0.06 Chi-squared test 74% 15% 2% 60% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. 104 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 3B.6 Stakeholder alignment and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Stakeholder alignment Strong 0.29 0.06 0.06 0.12 0.18 0.12 0.18 0.18 0.06 0.41 0 Weak 0.41 0.06 0.12 0.12 0.29 0.18 0.18 0.18 0.24 0.41 0.18 Chi-squared test 92% 93% 52% 11% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Notes: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. TABLE 3B.7 Legislation and patterns of power sector reform Announcement Implementation Delivery Sustainability Nonsustained Incremental Negligible Sustained Complete Extensive Cautious Minimal Limited Partial Rapid Legislation Strong 0.47 0.12 0.06 0.18 0.29 0.18 0.29 0.24 0.12 0.59 0.06 Weak 0.24 0 0.12 0.06 0.18 0.12 0.06 0.12 0.18 0.24 0.12 Chi-squared test 30% 88% 32% 21% P-value Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Fractions denote the share of the country sample in each category and sum to 1.00 within each of the boxes. The Chi-squared test examines whether it is possible to reject the null hypothesis of a statistically significant relationship between the severity of the crisis and (1) the extent of the reform announcement; (2) the boldness of the reform implementation; (3) the completeness of the reform delivery; and (4) the long-term sustainability of reform. A value below X percent denotes statistical significance at the X percent level. ANNEX 3C. WORLD BANK SUPPORT FOR POWER SECTOR REFORM OBSERVATORY COUNTRIES AND STATES Country/State Major World Bank loans to power sector Colombia (1994) Regulatory reforms, US$11m (1995) Facilitating operation of wholesale market, US$249m (1997) Regulatory reforms, US$12.5m Dominican Republic (1988) Improve operating efficiency of CDE and build T&D infrastructure, US$105m (2005) Transmission and service expansion, US$150m (2008) Improve cash recovery and quality of supply, US$42m (2015) Financial viability of distribution companies, US$120m Egypt, Arab Rep. (2006–10) Natural gas power plants, US$900m (2010) Transmission for RE, US$70m (2016–20) Pricing and structural energy sector reforms, US$3,000m India – Andhra Pradesh (1999) Comprehensive power sector reforms, US$576m ( Annex continued next page) How Did Political Economy Affect the Uptake of Power Sector Reform? 105 Country/State Major World Bank loans to power sector India – Odisha (1996) Comprehensive power sector reforms, US$350m India – Rajasthan (2001) Comprehensive power sector reforms, US$226m (2016) Financial health of distribution utilities, US$250m (2017) Financial health of distribution utilities, US$250m Kenya (1997) Sector reforms, US$125m (2000) Generation capacity, US$80m (2004) Sector recovery projects, US$80m (2010–11) Generation projects, US$500m (2012) Regional transmission interconnection, US$441m (2015) Modernization of sector, US$450m (2017) Off-grid access, US$150m Morocco (2005–07) Regulatory framework for RE, US$100m (2007–16) Renewable generation US$965m (2013) Improving quality of supply, US$40.5m (2013–15) Supporting renewable expansion, US$600m Pakistan (1994) Support for private power policy, WB: US$475m; IFC: US$378m; MIGA: US$31m (1995–2014) Hydro generation, US$2,688m (2006–11) Supporting private sector in energy projects, US$475m (2016–17) Transmission and distribution modernization, US$680m Peru (1994) Support power sector reform policy, US$150m (2006–11) Rural access, US$110m Philippines (1989, 1993, 1996) Rehabilitate, upgrade transmission and distribution systems, US$425m (1990) Improving generation, transmission, and distribution, US$390m (1992) Rural access, US$91m (1994) Geothermal, US$438m (2003) Rural access, US$40m (2008) Rehabilitating transmission in Bicol, US$13m Senegal (1998) Reforming power sector (private entry), US$100m (2004) Rural access, US$30m (2005) Rehabilitation of transmission and distribution, US$6.9m (2008, 2012, 2016) Improving Senelec financials and efficiency, US$235m (2014) Generation and transmission, US$99m Tajikistan (2005) Loss-reduction program, US$33m Tanzania (2001) Natural gas generation, US$205m (2004) Emergency rentals, US$46m (2007) Access expansion, US$134m (2010) Transmission expansion, US$60m (2013–14) Improve power and gas sector financials and performance, US$198m (2016) Rural access, US$209m Uganda (1991) Rehabilitating infrastructure, US$153m (2000) Support for UEB restructuring and privatization, US$8m (2001, 2009, 2015) Rural access, US$282m (2007) Generation, US$115m (2007) Supply until new generation comes online, US$306m (2011) Improve quality of supply, US$84m (2016) Grid expansion and reinforcement, US$100m Ukraine (1996) Sector reform (restructuring/privatization), US$76m (2001–03) Regulatory framework (economy wide), US$250m (2005, 2009) Hydro rehabilitation, US$197m (2007) Improving transmission, US$194m Vietnam (1990–99) Rural energy; power sector rehabilitation; power development; T&D, US$694m (2002) Improving efficiency, US$230m (2005, 2008) Rural energy (generation, transmission and distribution), US$375m (2006, 2011, 2013, 2015) Improving power grid, US$1,320m (2010, 2012, 2014) Power sector reform, US$712m Source: World Bank. Note: CDE = Corporación Dominicana de Electricidad; IFC = International Finance Corporation; m = million; MIGA = Multilateral Investment Guarantee Agency; RE = renewable energy; T&D = transmission and distribution; UEB = Uganda Electric Board; WB = World Bank. 106 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD NOTES Cheng, Chaoyo, YuJung (Julia) Lee, Galen 1. This chapter was informed in particular by Murray, Yuree Noh, Joseph Van Horn, Lee and Usman (2018). Further original and Johannes Urpelainen. 2016. “Political research was done by a team lead by Ashish Obstacles to Economic Reform: Comparative Khanna and comprising Anton Eberhard, Evidence from the Power Sector in 20 Indian Catrina Godinho, Alan David Lee, Brian States.” Paper presented at the Texas A&M Levy, Zainab Usman, and Jonathan Walters. Conference on Energy, College Station, Texas, The overall work program was coordinated September 26–28. by Vivien Foster and Anshul Rana. Dubash, Navroz K., ed. 2002. Power Politics: Equity 2. In order to aggregate across the four dimen- and Environment in Electricity Reform. June. sions of power sector reform considered in this Washington, DC: World Resources Institute. study, a simple PSRI is constructed. The index Erdogdu, Erkan. 2013. “A Cross-Country Analysis gives each country a score of 0 to 100 on each of Electricity Market Reforms: Potential dimension of power sector reform. The scores Contribution of New Institutional Economics.” are based on giving equal weight to each step Energy Economics 39 (5): 239–51. on each dimension of the reform continuum. ———. 2014. “The Political Economy of Electricity The average of the four 0–100 scores is used to Market Liberalization: A Cross-Country characterize a country’s extent of reform. For Approach.” Energy Journal 35 (3): 91–128. more on the index, see annex 3A. Foster, Vivien, Samantha Witte, Sudeshna Ghosh 3. Overall reform announcement is the sum of Banerjee, and Alejandro Moreno. 2017. all the reforms the country had committed to “Charting the Diffusion of Power Sector implementing from 1990 to 2015. Reforms across the Developing World.” Policy 4. Pritchett, Woolcock, and Andrews (2010) Research Working Paper 8235, World Bank, describe isomorphic mimicry as the “adop- Washington, DC. tion of the forms of other functional states Gore, Christopher D., Jennifer N. Brass, Elizabeth and organizations which camouflages a Baldwin, and Lauren M. MacLean. 2018. persistent lack of function.” It provides the “Political Autonomy and Resistance in “mechanism for avoiding needed reform or Electricity Sector Liberalization In Africa.” innovation while at the same time maintain- World Development 120 (August): 193–209. ing the appearance of legitimate engagement Henisz, Witold J., Bennet A. Zelner, and Mauro with developmental discourses.” F. Guillén. 2005. “The Worldwide Diffusion 5. Acquis communautaire is the accumulated laws of Market-Oriented Infrastructure Reform, and obligations of the European Union from 1977–1999.” American Sociological Review 1958 to present. It includes all European 70 (6): 871–97. Union treaties, laws, declarations, resolu- Jayarajah, Carl, and William Branson. 1995. tions, and international agreements. Structural and Sectoral Adjustment: World Bank 6. Significance is measured by the Chi-squared Experience, 1980–92. A World Bank operations test; details can be found in annex 3B. evaluation study. Washington, DC: World 7. With a sample of 53 diverse countries, Bank Group. Erdogdu (2013) suggests that reforms went Lee, Alan David, and Zainib Usman. 2018. “Taking further when the minister in place at the out- Stock of the Political Economy of Power Sector set of reforms had no previous experience in Reform in Developing Countries: A Literature the power sector. Review.” Policy Research Working Paper 8518, Rethinking Power Sector Reform, World Bank, REFERENCES Washington, DC. http://documents.­ worldbank​ APP (Africa Progress Panel). 2015. Power, People, .org​/curated/en/431981531320704737​ Planet: Seizing Africa’s Energy and Climate /­Taking​-stock-of-the-political-economy​ Opportunities. Geneva: APP. -of-power-sector-reforms-in-developing​-coun- Besant-Jones, John E. 2006. “Reforming Power tries​-a-literature-review. Markets in Developing Countries: What Have Levy, Brian. 2014. Working with the Grain: We Learned?” Energy and Mining Sector Integrating Governance and Growth in Board Discussion Paper no. 19, World Bank, Development Strategies. New York: Oxford Washington, DC. University Press. How Did Political Economy Affect the Uptake of Power Sector Reform? 107 Moerenhout, Tom S. H. 2018. “Reforming Williams, James H., and Reza Ghanadan. 2006. Egypt’s Fossil Fuel Subsidies in the Context “Electricity Reform in Developing and of a Changing Social Contract.” In The Politics Transition Countries: A Reappraisal.” Energy of Fossil Fuel Subsidies and Their Reform, edited 31 (6): 815–44. by Jakob Skovgaard and Harro van Asselt. Williamson, John. 2005. “The Washington Cambridge: Cambridge University Press. Consensus as Policy Description for Nakhooda, Smita, Shantanu Dixit, and Navroz K. Development.” A lecture in the series, Dubash. 2007. Empowering People: A Governance “Practitioners of Development,” ­ d elivered Analysis of Electricity—India, Indonesia, at the World Bank, January 13, 2004. Philippines, Thailand. June. Washington, DC: Peterson Institute for International Economics World Resources Institute. (PIIE). Pritchett, Lant, Michael Woolcock, and Matt World Bank. 1993. The World Bank’s Role in Andrews. 2010. “Capability Traps? The the Electric Power Sector: Policies for Effective Mechanisms of Persistent Implementation Institutional, Regulatory, and Financial Reform. Failure.” CGD working paper 234, Center for Washington, DC: World Bank. Global Development, Washington, DC, December. Worley, Heather, Sara Bryan Pasquier, and Sen, Anupama, Rabindra Nepal, and Tooraj Ezgi Canpolat. 2018. “Good Practice Note Jamasb. 2016. “Reforming Electricity Reforms? 10: Designing Communication Campaigns Empirical Evidence from Asian Economies.” for Energy Subsidy Reform.” World Bank, Oxford Institute for Energy Studies (OIES), Washington, DC. February. Zelner, Bennet A., Witold J. Henisz, and Guy L. F. Urpelainen, Johannes, and Joonseok Yang. 2017. Holburn. 2009. “Contentious Implementation “Policy Reform and the Problem of Private and Retrenchment In Neoliberal Policy Investment: Evidence from the Power Sector.” Reform: The Global Electric Power Industry, Journal of Policy Analysis and Management 36 (1): 1989–2001.” Administrative Science Quarterly 38–64. 54 (3): 379–412. Building Blocks of Reform PART II 109 What Has Been Done to Restructure Utilities and Improve Governance? 4 Guiding questions • How have countries gone about separating the main sectoral functions of policy making, regulation, and service provision? • What measures have been adopted to improve the governance of corporatized power utilities? • To what extent have countries pursued horizontal and vertical unbundling to pave the way for com- petition in the sector? • How does the current wave of technological disruption in the power sector affect utilities’ restructur- ing and management practices? Summary • Reform efforts began with the corporatization of power utilities, many of which had, until then, been operated as departments of energy ministries. This step was followed by the creation of a regulatory entity to provide arm’s-length oversight. • Amid efforts to reallocate responsibility for regulation and service provision, the need to strengthen core ministry functions was too often overlooked. In particular, scant attention was paid to develop- ing capacity for sector planning, which is a critical omission given the vertiginous growth of electric- ity demand and the pressing need for new investments. • A key aspect of the corporatization process was to instill sound management practices in the new utilities. Such practices aim to safeguard the autonomy and accountability of the company’s board of directors, and responsibly manage human resources, financial discipline, and information tech- nology. There is considerable variation in the management quality of state-owned utilities across countries, and by and large management practices fall short of those found in private utilities. • Another major focus of reform efforts was the vertical and horizontal unbundling of corporatized national utilities. Such structural reforms were intended to be stepping-stones to greater private sector participation and competition in the power sector. In many countries, however, these goals have yet to be reached. • Looking ahead, technological disruptions under way in the power sector will make planning increas- ingly complex; further sector restructuring may be needed to facilitate competition at the retail level. 111 112 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION investment program may introduce biases into This chapter examines efforts to restructure the planning process. In view of this, the 1990s the power industry and improve utilities’ gov- reform model recommended clear separation ernance.1 At one level, reforms have called for of these functions under distinct institutions. It a reengineering of the power sector’s institu- was envisaged that policy functions (including tional architecture. This has often involved planning) should be undertaken by the line the breaking up of existing actors, the creation ministry, regulatory functions by a separate of new ones, and the reassignment of respon- regulatory entity, and service provision by the sibilities across them. On a deeper level, utility. reforms have sought to improve the internal Reform efforts sought to ensure that the ser- functioning of all major actors, through vice provider was corporatized and operating encouraging the adoption of stronger gover- under sound governance and management nance and managerial practices. Thus, the arrangements. As of the early 1990s, some guiding questions for this chapter are as fol- power utilities still functioned simply as depart- lows: How did countries go about separating ments of line ministries in the central govern- functions during the reform process? What ment, subject to public sector arrangements for structural models were used? What measures budgeting, employment, and decision making. were adopted to assure improved utility gov- A critical first step was to corporatize the util- ernance? How does the current wave of tech- ity, essentially converting it into a separate nological disruptions in the power ­ s ector state-owned enterprise (SOE) distinct from the affect the restructuring and governance of central government and operating under com- utilities? pany law. Because of the greater managerial Under the 1990s reform model, restructur- autonomy it provided, corporatization was ing the power sector was considered founda- expected to foster improved performance tional for the subsequent implementation of under public sector ownership and was also an deeper reforms meant to foster private sector indispensable precondition for private sector participation and market liberalization. The participation. Depending on the country, the process of restructuring the power sector resulting SOE might remain under the direct entailed two key steps: the separating of key jurisdiction of the line ministry for energy or functions and the unbundling of utilities. Both might find itself reporting to another ministry were essentially directed toward reducing con- or entity charged with exercising public own- flicts of interest. ership and oversight functions for all SOEs The first step involved separating three key across sectors. Corporatization was supposed to functions: policy making, regulation, and ser- create incentives for the efficient management vice provision. Under the arrangements typical of the company, reduce scope for political up to 1990, all three functions fell under the interference in company decisions, and ensure mandate of a single national public utility financial discipline. However, the efficacy of charged with acting in the public interest, often the corporatization process depended on the embedded within the energy ministry and with quality of the governance framework and no distinct institutional identity. Conflicts of internal management arrangements that interest were common. For instance, a utility accompanied it. that self-regulates is unlikely to hold itself The second step of the 1990s reform model accountable to the highest standards. Or, again, was the sector’s vertical and horizontal unbun- a utility whose economic clout and political dling. Unbundling might have begun with sepa- influence are determined by the scale of its rating accounting and managerial business unit What Has Been Done to Restructure Utilities and Improve Governance? 113 segmentation but was supposed to eventually Depending on the depth of the structural lead to full legal unbundling, and companies reforms undertaken, a country might find itself with distinct ownership. Unbundling was not with one of several possible power sector orga- seen as an end in itself but as an enabling mea- nizational models (box 4.1). At one end of the sure that makes further reforms possible (Bacon spectrum, in most countries, the starting point 2018). The purpose of vertical unbundling was was a vertically integrated national monopoly to separate out those elements of the supply power utility. At the other end of the spectrum, chain deemed natural monopolies (notably the endpoint of the 1990s sector reform model transmission and distribution), from those con- was at least a wholesale, if not a retail, power sidered to be competitive (such as generation market. Because of the progressive and partial and retail). This separation was to avoid the application of restructuring reforms, several conflicts of interest that might arise when a gen- intermediate structural models can be found erator operating in a competitive market also (table 4.1). Some countries got no further than controls the transmission network. For exam- opening the market to independent power ple, such a generator might use its dominant producers (IPPs) that competed for the right to position to restrict the participation of other build a new generation plant and supply the generators. In some relatively developed juris- enduring national vertically integrated monop- dictions, vertical unbundling was further oly utility. Others went further, by vertically extended to separate the retail (or commercial unbundling the entire generation tier, and sale) function of power utilities from their phys- allowing new IPPs and divested companies ical distribution function. In principle, this managing existing generation assets to com- makes electricity retailers compete for the same pete for the right to supply a single buyer, end consumers, even as distribution utilities which typically retained a monopoly in both continue to operate as local monopolies. transmission and distribution. An additional Once vertical unbundling was complete, step was to complete the vertical unbundling reformers turned to horizontal unbundling. process to create a separate transmission sys- The purpose of horizontal unbundling was to tem operator, allowing third-party access to the create multiple competing entities in those seg- grid so that generators could compete to ments of the supply chain where competition provide power directly to large industrial cus- ­ was possible. Horizontal unbundling has been tomers, usually through long-term bilateral particularly important in the generation sector, contracts. This situation could, relatively easily, where it has been critical in limiting the mar- evolve into a full-scale competitive power mar- ket shares of individual generators to mitigate ket through the creation of a market operator their abuse of market power. Horizontal to provide a platform for short-term exchanges unbundling has also been relevant in distribu- of electricity in a wholesale power market. tion, though here its effects are more indirect. Although these intermediate models were For the functioning of a competitive wholesale originally intended as transitional phases in the power market, it is necessary to have multiple pursuit of a wholesale power market, in many buyers as well as multiple sellers, with distribu- countries they became quasi-permanent tion utilities being among the major buyers in (Gratwick and Eberhard 2008). This result any market. Furthermore, the existence of occurred either because, in these countries, multiple distribution entities, even if each power systems were not suited to the imple- serves a particular geographic area on a natural mentation of a competitive market, or because monopoly basis, provides comparative infor- political obstacles blocked further steps in the mation that facilitates the task of regulation. restructuring process (Besant-Jones 2006). 114 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 4.1 Selected power sector structures around the world Depending on the extent of the reforms undertaken in recent decades, power sectors around the world may be classi- fied by a variety of types of structures. A few examples are illustrated here. Vertically integrated utility with independent power producers Senegal, Tajikistan, and Tanzania. A vertically integrated utility (VIU) controls generation, transmission, and distribution. Several independent power producers (IPPs) have power purchase agreements with the utility and supply power to it (figure B4.1.1). Variations: In Senegal and Tanzania, the regulator is autonomous; in Tajikistan, it is not. FIGURE B4.1.1 Vertically integrated utility Regulatory entity Generation Independent power producers Transmission Distribution Consumers Public Private Autonomous Source: World Bank elaboration. Note: TSO = transmission system operator. Single-buyer model Arab Republic of Egypt, Kenya, Morocco, Pakistan, Uganda, Ukraine, and Vietnam. In this model, the VIU has been ­ separate restructured. Generation is a mix of IPPs and public generation company(ies). Distribution and transmission are ­ businesses but majority state owned (figure B4.1.2). Variations: Ukraine—distribution companies are private and the transmission system operator is autonomous; Uganda—long-term private concessions exist for the generation and distribution companies; Pakistan—a privately owned VIU also exists in the system; Morocco—a VIU controls transmission and some distribution (as a single buyer); IPPs have a growing share; municipal-level private distributors have long-term concessions. (Box continued next page) What Has Been Done to Restructure Utilities and Improve Governance? 115 BOX 4.1 Selected power sector structures around the world (Continued ) FIGURE B4.1.2 Single-buyer model Regulatory entity Independent power producers Public generation company(s) Transmission operated by TSO Public and private distribution companies Large industrial Consumers buyers Public Private Autonomous Source: World Bank elaboration. Note: TSO = transmission system operator. Wholesale-buyer model Colombia, Dominican Republic, India, Peru, and the Philippines. A wholesale market exists for buyers (large industries and distribution companies) to purchase power from a mix of public and private generators. The wholesale buyers can also purchase directly from generators. Distribution is a mix of public and private ownership. The transmission network is operated by an independent transmission system operator (figure B4.1.3). Variations: Colombia’s power sector has some publicly owned vertically integrated utilities that also participate in the market. (Box continued next page) 116 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 4.1 Selected power sector structures around the world (Continued ) FIGURE B4.1.3 Wholesale-buyer model Regulatory Independent Private Public Public entity power generation generation Private producers companies company Autonomous Transmission operated by TSO Wholesale power market Private distribution Public distribution Large industrial companies companies buyers Consumers Source: World Bank elaboration. Note: TSO = transmission system operator. TABLE 4.1 Unbundling is a means to remove conflict of interest and encourage competition Degrees of vertical and horizontal unbundling, and their effects on competition Vertical unbundling Horizontal unbundling Competition Vertically integrated utility None None None Independent power Incumbent remains vertically New entrants responsible Companies compete for concessions producers integrated for new generation plant to build new generation plant Single-buyer Incumbent generation assets are New entrants responsible All generators supply power under model separated out under a distinct for new generation plant contract to single buyer company Third-party access Incumbent divided into distinct New entrants responsible All generation companies compete to grid companies for generation, for new generation plant to serve large industrial customers transmission, and distribution under bilateral contracts Wholesale power Incumbent divided into distinct Incumbent generator and All generation competes to serve market companies for generation, distributor divided into all large customers including transmission, and distribution distinct companies distributors, with spot market Retail power Incumbent divided into distinct Multiple retailers exist, All generation competes to serve all market companies for generation, allowing new entrants large customers with spot market, transmission, distribution, and retail and retailers compete directly to serve residential customers Source: World Bank elaboration. What Has Been Done to Restructure Utilities and Improve Governance? 117 KEY FINDINGS power utilities or line ministries fell between Drawing on the experience of the 15 c ­ ountries the cracks as power sector reform processes in the Power Sector Reform Observatory, the worked toward the unbundling of the incum- main findings of this chapter can be summa- bent utilities and the creation of technical rized as follows. capacity in regulatory agencies outside of line ministries. In practice, power markets proved difficult to establish in all but a handful of Finding #1: Although reform efforts developing countries, and even among those prioritized the creation of regulatory countries, price signals have not provided an entities, the critical function of sector adequate basis for investment decisions planning was underemphasized (Rudnick and Velasquez 2018) (a subject that Achieving energy security remains a huge plan- will be taken up in chapter 7). ning and procurement challenge for fast-grow- Good sector planning entails technically ing developing economies. With electricity grounded plans for both generation and demand in most developing regions growing at transmission that are fully integrated with ­ rates of 6–7 percent per year since 1990 other relevant plans. At a minimum, plans for (Steinbuks and others 2017), achieving supply– generation and transmission need to be demand balance and associated energy security ­ coordinated and mutually consistent. Ideally, calls for major investments that may entail a generation planning should be aligned with doubling of power system capacity every the country’s broader energy plan (for coun- decade. Keeping up with this exacting pace tries with primary energy resources) and be requires countries to develop sound least-cost compatible with the overall national develop- generation plans that identify the most cost-­ ment plan. Consulting with stakeholders effective path of generation expansion, and to during the planning process helps to bring in implement these in a timely fashion by creating these wider perspectives. In practice, deficien- a strong institutional link between planning and cies can be found in the planning framework associated procurement of generation capacity. for both generation and transmission The 1990s model had little to say on the (­f igure 4.1, panel a). Some countries lag issue of planning, with an implicit assumption ­ further behind on transmission planning (such that the advent of a wholesale power market as the Arab Republic of Egypt, the Philippines, would somehow circumvent the need for it. and Tajikistan), whereas others lag further The end goal of the 1990s model was to create behind on generation planning (­ figure 4.2). a competitive market. The supposition at that The institutional responsibility for planning time was that private investments in power needs to be clearly assigned to an entity with generation could be guided by price signals. adequate technical capacity. A wide range of The role of the state was seen primarily as the institutional arrangements for power systems’ regulator of a privately owned and operated planning can be observed across countries. competitive sector, and great emphasis was Most prevalent are cases where the planning placed on the creation of a capable regulatory function is assigned either to the line ministry institution and associated legal framework or to the power utility; in some cases, it is (Pardina and Schiro 2018) (which will be the assigned to both. For sectors that are unbun- subject of chapter 6). By contrast, central plan- dled, it is the transmission utility that may ning functions were overlooked or down- retain the sector planning function (as in Egypt played. Indeed, in some countries, the plan- and Pakistan), and once competitive markets ning function traditionally housed in national are introduced this responsibility may migrate 118 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 4.1 Plans must be mandatory in implementation and combined with transparent and competitive procurement Best practices observed across 15 countries, by share of total countries a. Best practices for planning b. Best practices for procurement Plan implementation Procurement is mandatory framework exists Planning is transparent and participatory Competetive tendering is allowed Links to other relevant plans Direct negotiations Competent planning are allowed entity exists Projects assigned Master plan exists to incumbents 0 20 40 60 80 100 0 20 40 60 80 100 Share of countries (%) Share of countries (%) Generation Transmission Source: Based on Rethinking Power Sector Reform utility database, 2015. FIGURE 4.2 Sector planning remains necessary and calls for more strategic in nature and the other more adequate institutional capacity and sound processes supervisory. Nevertheless, unifying both func- Generation and transmission planning across 15 countries tions at least takes advantage of the creation of 100 a critical mass of technical expertise among 80 sector regulators, which can also be brought to Planning index (%) bear on planning. Even when planning and 60 regulation are not merged in this way, it is still 40 important for the regulator to review the legit- 20 imacy of investment plans put forward by the utility, which will have a major impact on tariff 0 determinations. Pa lic M tan co Se ru Ta gal Ug ia da yp Uk a ab e ilip p. Ta ines Vi an Co am a a di bi ny Ar ain an Ph Re Pe b oc ist an ne In m n s pu Ke p nz r ki et Ultimately, the impact of good sector plan- jik or lo Re an ning depends on its being linked to timely, t, ic in Eg m competitive, and transparent procurement Do Generation Transmission processes for new capacity. Critical investments Source: Based on Rethinking Power Sector Reform utility database, 2015. in generation and transmission have significant lead times and must be initiated well ahead of to the system operator (as in Ukraine). In a when they are needed, which requires that handful of relatively large countries, where a plans be clearly time bound and that their culture of planning is particularly ingrained, implementation be mandatory. However, only sector planning may be the responsibility of an a few of the 15 observatory countries make independent entity, such as Brazil’s Empresa de the implementation of plans mandatory Pesquisa Energetica or the Central Electricity figure 4.1, panel a). It is also important to (­ Authority of India. An interesting variation is ensure that new generation and transmission found in Kenya and Uganda, where the regu- projects are procured on a competitive basis to lator is given explicit responsibility for power provide value for money. The existence of a system planning. Strictly speaking, planning procurement framework for generation and and regulation are two distinct functions, one transmission projects is far from universal, and What Has Been Done to Restructure Utilities and Improve Governance? 119 in many cases noncompetitive procurement at diversifying the generation mix and has methods—such as direct negotiation—­ been regularly updated every five years continue to be allowed (figure 4.1, panel b). although always by a different international In Tanzania, for example, few plans have firm. Despite a reform process and changes in been realized despite the country’s planning planning roles, expansion plans continue to capacity. The Ministry of Energy and Minerals be ignored. has a well-defined responsibility for develop- ing power generation and transmission expan- Finding #2: The governance of sion plans. The Tanzania Electric Supply corporatized public utilities leaves Company (TANESCO), the state-owned utility, a lot of room for improvement, plays a central role in the development pro- and still falls considerably short cess, which includes experts from various of governance practices in ministries and government agencies. The comparable private utilities resulting power system master plans take their cue from short-, medium-, and long-term In many countries, the first step toward reform national development plans. When it comes to was the corporatization of the public utility. implementation, however, experience shows Before 1990, many public power utilities oper- that master plans are rarely followed, leaving ated as administrative departments of their the country without adequate reserves to respective line ministries without any separate withstand recurring drought situations, and corporate existence. This situation left them leading to emergency procurement of rental subject to the vagaries of public administration plants and direct negotiation of oil-fired plants and unable to adopt a proper commercial orien- at exorbitant costs. To ensure that power sys- tation. For this reason, the first step toward tem master plans are followed and new capac- power sector reform in many countries was to ity is procured in a timely and transparent separate out the operational functions associ- manner, the government plans to establish a ated with service provision into a distinct state- multidisciplinary, interministerial committee owned corporation, typically operating under that would coordinate the procurement of company law. In doing so, countries made many power projects. important decisions regarding the governance of The Dominican Republic has a long history the company and the establishment of its man- of overlooking sector plans, despite institu- agement practices. tional reforms of the sector’s planning func- There is considerable agreement regarding tion. Expansion plans have been developed by the nature of good management and gover- the vertically integrated utility, Corporación nance practices for utilities. All of the 15 obser- Dominicana de Electricidad (CDE), since vatory countries have corporatized their power the 1960s. In 15 years during the 1970s utilities, in most cases during the reform pro- and 1980s, more than 10 expansion plans cess. (In some countries, such as the Philippines, were produced but none were actually fol- the major utility was corporatized well before lowed, contributing to CDE’s deterioration. power sector reforms.) The effectiveness of the Consequently, generation deficits were tack- process, however, depends on the adoption of led by purchasing emergency oil-based sound governance and management practices turbines, which are both inefficient and costly ­ within the corporatized enterprises. Fortunately, to operate. Following power sector reforms, well-established principles of good governance responsibility for long-term planning has exist for SOEs such as power utilities (OECD fallen to the national energy commission. The 2015; World Bank 2014). On this basis, good latest plan covering the period 2011­ –25 aims governance practices can be ranked by several 120 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD criteria, including the autonomy and account- it is possible to measure the current rate of ability of the board of directors, the exercise of adoption of such practices. Using a sample of 19 financial discipline within the company, and the state-owned and 9 privatized utilities from the prevalence of good management practices both 15 observatory countries, the Utility Governance for human resources and for information tech- Index measures the existence of best practices in nology (figure 4.3). utility rules and regulations. For example, a util- A Utility Governance Index measures the ity may, on paper, allow managers to hire and extent to which specific utilities conform to fire employees on the basis of performance— good practices. It is difficult to say exactly when and the index captures this rule but is unable to and how good governance and management tell whether the manager actually follows it. practices have been adopted over time, because The governance gap between corporatized such measures are usually implemented within significant, public utilities and privatized ones is ­ institutions and do not necessarily involve and public utilities practice better governance major legal or structural changes that can read- when they coexist alongside private utilities. It ily be tracked in the public record. Nevertheless, is instructive to look at the variations in utility FIGURE 4.3 Overview of utility governance performance indicators Utility Governance Index Corporate governance Utility management Average scores Average scores Average scores Autonomy Accountability Financial discipline Human resources Information and technology Average scores Average scores Average scores Average scores Average scores 1. Appointment by share 1. Separation of chairperson 1. Utility has a credit rating 1. Conducts annual 1. SCADA system holders or ownership and CEO profiles 2. Liberty to issue new performance reviews 2. IT system to record/ entity (SOE) 2. Existence of company bonds or equity 2. Able to pay bonuses to resolve interruptions 2. Selection is transparent secretary 3. Dividends to award good performance in supply, support and competitive 3. Existence of sub- shareholders 3. Able to fire employees distribution, and energy 3. Removal at term or committees to deal with 4. Explicitly defined public for poor performance management through legal procedure di erent issues service obligations 4. Exemption from public 5. GIS 4. Board is final decision- 4. Existence of an audit 5. PSOs are costed and employment regulations 6. KPIs for supply quality making body on: committee compensated by 5. Exemption from public 7. Advanced metering • Corporate strategy 5. Existence of code of government pay scales 8. Accurate customer • Business plans ethics or conduct 6. Requirement to meet 6. Clear policy for sta database • Performing objectives 6. Required to declare financial targets training 9. Call center for customer • Selection and conflict of interest 7. System of internal 7. Managers have authority complaints monitoring of CEO 7. Regulations to protect financial controls to hire and fire 10. Website for submission • Raising capital from the rights of minority 8. Internal audit function employees, execute and follow up of debt or equity shareholders 9. Utility subject to state budget and implement complaints • Major capital 8. Publishing an annual auditing procedures investment projects 11. Regular monitoring of expenditure report 10. Produces annual 8. Hiring process involves customer satisfaction • Deciding and financial accounts public advertisements, 12. CMS and RMS implementing tari 11. Audited by third party short-listing candidates, 14. KPIs for commercial adjustments 12. Public disclosure of interviews, and reference cycle and corporate • Human resource financial accounts checks resource management decisions 13. Accounting in compliance with national and international standards Source: Based on Rethinking Power Sector Reform utility database, 2015. geographic information system; IT = information technology; KPI = key Note: CEO= chief executive officer; CMS = commercial management system; GIS = ­ performance indicator; PSO = public service obligtion; RMS = resource management system; SCADA = Supervisory Control and Data Acquisition; ­ SOE = state-owned ­enterprise. What Has Been Done to Restructure Utilities and Improve Governance? 121 governance scores across countries, and even FIGURE 4.4 Private utilities implement more governance best more within countries for those cases where practices, but government-owned utilities improve governance both public and private utilities exist side by if there is some private competition side (figure 4.4). For those jurisdictions where Good governance practices followed by public vs. private utilities across 17 jurisdictions, 2015 companies are entirely state owned, utility 100 governance scores tend to be systematically Overall Utility Governance Index (%) low with a median score of 55 percent. Those 80 scores tend to be higher, in the 60–90 percent range, for the private utilities from the obser- 60 vatory group. The gap in scores is particularly striking, for example, between the public util- 40 ity Lahore Electric Supply Company (LESCO) and the privatized Karachi Electric in Pakistan, 20 and the provincial public utility Hidrandina and the metropolitan privatized utility Luz del Sur in Peru. In fact, the highest utility gover- 0 di Ta ep. Vi sha dh Pak m Re cco ab ru ja a Se an lo e ilip ia Ug ya da Pr an jik h ic Mor n ic Ke s Uk gal Ra ni Co ain ne Ta des a Ph mb bl na Ar Pe n h an ra ist ist nance score for a public utility is for the Kenya R – nza di an o ne st pi pu r et a O – Power and Lighting Company (KPLC), which a t, di yp a In An in Eg has 49 percent of its capital floated on the In m – Do a di Nairobi Stock Exchange, with the remainder In held by the Ministry of Finance. Private Public Public utilities are more likely to lag behind Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: The private utility in Kenya has a 49 percent stake floated on the Nairobi Stock with respect to certain areas of governance. In Exchange and 51 percent is owned by government. The Western Electricity Supply this section, the performance of public and pri- Company of Orissa Ltd. (WESCO) in India – Odisha was a private utility during the study period, but the state took over the utility in 2016. The Central Electricity Sup- vate utilities across the sample is averaged out ply Utility of Odisha (CESU) had gone back to the state in 2001 itself and as such is and disaggregated by components of their gov- considered a public utility for the study. ernance frameworks. The results serve to illus- trate that both public and private utilities fall autonomy on almost all of these decisions short of best practice in many areas, but that figure 4.5). By contrast, the autonomy of the (­ their strengths and weaknesses tend to differ. public utilities’ boards is significantly con- Boards of private utilities enjoy almost com- strained on all of these points, particularly the plete decision-making autonomy, whereas raising of finance and the appointment of the public utility boards have limited freedom chief executive, which indicates that govern- on critical matters of finance and human ments continue to be closely involved in the resources. The first area to consider relates to business decisions of state-owned utilities. the autonomy and accountability of the com- Public utilities also suffer considerable inter- pany’s board of directors. The board should in ference in the appointment and removal of principle have the final say on all major busi- board members. To be accountable, boards ness-related decisions, including the definition must have practices that enhance transparency of strategies, plans, and performance ­ objectives; and contain potential conflicts of interest. For important financial decisions on investment example, it is important to have a transparent programs and related financing; and significant process for the appointment of suitably quali- human resource decisions such as the appoint- fied board members, as well as clear and rea- ment of the chief executive officer and the hir- sonable justification for their premature ing and firing of staff. For the most part, the removal from office. To avoid conflicts of inter- private utilities considered have full board est, it is important to separate out the roles of 122 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 4.5 Private utilities far outpace their public counterparts when it comes to making decisions independently Practices to ensure board decision-making autonomy in public vs. private utilities in 17 jurisdictions Board is the final body to make decisions on: Approving business plans Defining corporate strategy Setting and monitoring performing objectives Major capital expenditures Human resource hiring and firing decsions Raising capital from debt Appointing and overseeing the CEO Raising capital from equity Deciding and implementing tari s 0 10 20 30 40 50 60 70 80 90 100 Share of utilities (%) Private Public Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: CEO = chief executive officer. FIGURE 4.6 Boards in state-owned utilities are less accountable than their private counterparts, which must answer to various shareholders Practices to ensure board transparency and limit conflicts of interest in public vs. private utilities in 17 jurisdictions Utility publishes an annual report Function of company secretary exists Requirement to declare conflicts of interest Chairperson and CEO are separate positions Audit committee of the board Board has subcommittees for di erent issues Board code of conduct exists Minority shareholders' rights are protected Transparent process exists for board selection Private or public shareholders appoint board Board members cannot be removed at will 0 10 20 30 40 50 60 70 80 90 100 Share of utilities (%) Private Public Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: CEO = chief executive officer. What Has Been Done to Restructure Utilities and Improve Governance? 123 the board chairperson, chief executive officer, as the extent of financial oversight by owners and company secretary; to create board com- and investors. Among private utilities, financial mittees with clear responsibility for auditing accounts are universally produced and are and other functions; and to apply a code of externally audited. Almost all of the private conduct. Good accountability practices are not utilities (90 percent) prepare accounts in accor- universal among the group of private utilities, dance with international standards and ­ disclose and the gaps between the practices of public them publicly. Among public utilities, financial and private utilities are relatively small on accounts are produced by 95 percent, are some dimensions (figure 4.6). The most strik- externally audited by 89 percent, are publicly ing differences can be found with respect to disclosed by 74 percent, and meet international board appointments and removals: about standards only in the case of 42 percent. In 75 percent of private utilities follow good prac- addition, public utilities are almost twice as tices (at least on paper), compared with less likely to follow national accounting standards than 10 percent of public utilities. as international ones. Private utilities clearly Public utilities tend to fall short of basic good have more freedom when it comes to raising accounting practices, which are universal in various forms of finance and are exposed to the private sector (figure 4.7). Public utilities’ the financial discipline that goes with that free- financial discipline is related to the rigor of dom, but less than half has a credit rating. For their accounting and auditing practices, as well public and private utilities alike, public service FIGURE 4.7 Public utilities have little independence when it comes to raising capital and tend to follow national rather than international accounting standards Financial reporting practices of public vs. private utilities in 17 jurisdictions System of internal financial controls exists Internal audit function exists Financial accounts are produced Financial accounts are audited by external auditor Utility is subject to state auditing procedures Financial accounts meet national standards Financial accounts are publicly disclosed Utility required to meet financial performance targets Financial accounts meet international standards Public service obligations are explicitly defined Utility has a credit rating Utility can issue new bonds Utility pays dividends to shareholders Utility can issue new equity PSOs are costed and compensated by government 0 20 40 60 80 100 Share of utilities (%) Private Public Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: PSO = public service obligation. 124 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD obligations are explicitly defined in less than to be able to fire employees. Although public half of cases, and are never properly costed and and private utilities advertise only about compensated for. Overall, public utilities out- 75 percent of available positions, recruitment perform private ones only in internal auditing in private utilities is more likely to involve and financial controls. short-listing, interviewing, and reference Human resource practices tend to be more checks. Annual performance reviews are prev- rigorous in private than in public utilities, alent in over 80 percent of public utilities and which tend to have far less ability to reward are universal in their private counterparts, but good performers and fire bad ones. Good the latter are more likely to offer perfor- human resource practices entail objective hir- mance-related bonuses. Finally, for private and ing processes, adequate remuneration and per- public companies alike, decisions to hire and formance-related pay, and the ability to fire fire employees can rarely be taken at the man- poor performers. Once again, a comparison agement level but need to involve the board. between private and public utilities has mixed Almost 90 ­percent of private utilities, however, outcomes (figure 4.8). The largest differences can ultimately fire employees for poor perfor- stem from the fact that private utilities are mance as opposed to only about 70 percent of much less likely to be constrained by public public ones. sector wage scales and employment regula- Both public and private utilities are doing tions. They also adopt a more rigorous quite well at adopting information technology approach to hiring and are much more likely to improve internal management practices. FIGURE 4.8 Public utilities have little freedom in making staffing decisions and have less transparency in hiring as compared to their private counterparts Human resource practices of public vs. private utilities in 17 jurisdictions Annual sta performance reviews exist Recruitment involves short-listing candidates Sta training policy exists Recruitment involves interviewing candidates Employees can be fired for poor performance Recruitment involves advertisment of positions Recruitment involves reference checks Employees receive performance related bonuses Wages not based on government pay scales Managers are free to fire employees Managers are free to hire employees Government employment regulations don't apply 0 10 20 30 40 50 60 70 80 90 100 Share of utilities (%) Private Public Source: Based on Rethinking Power Sector Reform utility database, 2015. What Has Been Done to Restructure Utilities and Improve Governance? 125 The continuous advance of digital technolo- sample utilities. In certain areas, notably the gies allows for greater automation and use of online customer interfaces, public utili- remote management of electricity networks. ties seem to have the edge. Information technologies can significantly The example of the Indian state of Andhra enhance a utility’s ability to deliver on many Pradesh shows how governance reforms core areas such as network management and within the public sector can have a material the commercial cycle, but uptake of informa- effect on utility performance. Andhra Pradesh tion technology is far from being universal for unbundled its state electricity board in 1999, either public or private utilities (­ figure 4.9). creating four distribution utilities. Eschewing Among the most widespread applications of privatization of its distribution utilities, the information technology are customer data- state appointed a visionary managing director bases, call centers, and SCADA (Supervisory to the state transmission company, who was Control and Data Acquisition) systems, given unprecedented autonomy to oversee adopted by over 85 percent of both public and the distribution utilities that were subsidiaries private utilities. Among the least prevalent of the transmission company in those days. information technologies are commercial This arrangement was a big change from a management systems, resource management time when the state electricity board was run systems, and advanced metering infrastruc- by the line ministry on a day-to-day basis. The ture, which are adopted by close to half of the new leadership went about implementing a FIGURE 4.9 Both public and private utilities have adopted the latest information and technology solutions and are mostly at par Adoption of information and technology solutions by public vs. private utilities in 17 jurisdictions KPIs are used to monitor quality of supply Accurate customer database SCADA system Call center for dealing with customer complaints KPIs are used to monitor commercial cycle Website for submission of customer complaints IT system to support distribution management Geographic information system (GIS) IT system to support incidence resolution IT system to support energy management KPIs used to monitor corporate resource management Customer satisfaction regularly monitored Advanced metering infrastructure (AMI) Commercial management system (CMS) Resource management system (RMS) 0 10 20 30 40 50 60 70 80 90 100 Share of utilities (%) Private Public Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: IT = information and technology; KPI = key performance indicators ; SCADA = Supervisory Control and Data Acquisition. 126 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD culture change aimed at strengthening Finding #3: The unbundling of vertically accountability by introducing monthly perfor- integrated incumbent power utilities is mance reviews for all officials from the top unlikely to deliver major benefits unless down. Simple and easily calculated key per- it is accompanied by other reforms formance indicators were introduced, and Once utility corporatization is in place, the staff at various levels reported their perfor- next stage of reform is to restructure the mance monthly. The most important of the incumbent utility. As noted above, unbun- key performance indicators was a simple dling is a prerequisite for fostering competi- reporting of cash collected per unit of energy tion in the power sector. It can also be helpful that went into each administrative unit. to separate out elements of the electricity sup- (Before reform, the system had emphasized ply chain that may be suitable for privatiza- technical rather than commercial perfor- tion from those that may not. mance.) In addition, a significant enhance- Full unbundling of the power sector has not ment of employee pay scales increased incen- been widely adopted across the developing tives. In parallel, significant investments were world (figure 4.10). Few developing countries made in meter modernization, as well as in (less than 20 percent) have managed to imple- improving the quality of supply and expand- ment full vertical and horizontal unbundling ing access. The state also enacted legislation of their power sectors, and many (close to that criminalized power theft, making it easier 60 ­percent) continue to operate with a verti- to prosecute felons, and setting an example cally integrated national monopoly utility. The for other customers. The combined effect of remaining 20 percent is at an intermediate these measures was to double utility revenues stage of partial unbundling—that is, those in the space of five years (2002–07). FIGURE 4.10 Close to 60 percent of developing countries still operate with a vertically integrated national monopoly utility The power sector in developing countries, by structure and degree of unbundling, 1995–2015 100 90 80 70 58 64 Share of countries (%) 68 60 77 90 50 40 11 30 9 8 13 20 11 7 13 6 10 18 7 16 10 11 0 2 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 20 20 20 20 20 20 20 19 19 19 19 19 20 20 20 20 20 20 20 20 20 Fully vertically and horizontally unbundled Fully vertically unbundled Partially vertically unbundled Vertically integrated Source: Foster and others 2017. What Has Been Done to Restructure Utilities and Improve Governance? 127 companies have separated out either genera- others unbundled yet retained a single-buyer tion or distribution but not transmission, or model largely dependent on the public sector have completed vertical unbundling without (such as Egypt and Pakistan). Among African further horizontal unbundling of the genera- countries (such as Senegal, Tanzania, and tion or distribution sectors. Among the 15 Uganda), vertical integration of the sector observatory countries, some completed full remains widespread (table 4.2). unbundling as a basis for greater private sector Some variation of the single-buyer model participation and a wholesale power market remains the most widely adopted organiza- (such as Peru and the Philippines), whereas tional model for the power sector in the TABLE 4.2 Overview of restructuring and competition reforms in 17 jurisdictions, 2015 Vertical unbundling Horizontal unbundling Sector model Colombia Yes: but with some vertically integrated utilities Yes: multiple generators and Wholesale power market remaining distributors Dominican Yes: generation, transmission, and distribution Yes: 3 generators and 3 distributors Wholesale power market Republic fully separated Egypt, Arab Yes: generation, transmission, and distribution Yes: 6 generators and 7 distributors Single-buyer model Rep. separated, but under a single holding company under a single holding company India – Andhra Yes: generation, transmission, and distribution Yes: 4 distributors (after bifurcation, 2)a Wholesale power market Pradesh fully separated India – Odisha Yes: generation, transmission, and distribution Yes: 2 generators and 4 distributors Wholesale power market fully separated India – Yes: generation, transmission, and distribution Yes: 3 distributors Wholesale power market Rajasthan fully separated Kenya Yes: generation, transmission, and distribution No: 5 different generation entities Single-buyer model fully separated were all combined into KenGen Morocco No: vertically integrated national utility plus No: but IPPs are closing in on 50% Single-buyer model IPPs and 11 local distribution utilities of generation, and distribution in major cities is handled by city-level distribution companies Pakistan Yes: generation, transmission, and distribution Yes: 4 generators and 8 distributors Single-buyer model fully separated with 1 vertically integrated utility with 1 vertically integrated utility Peru Yes: generation, transmission, and distribution Yes: Multiple generators and Wholesale power market fully separated distributors Philippines Yes: generation, transmission, and distribution Yes: Multiple generators and Wholesale power market fully separated; utilities can own vertical distributors business operations through subsidiaries Senegal No: vertically integrated national No IPPs only; vertically integrated utility plus IPPs utility functions as single buyer Tajikistan No: vertically integrated national utility plus 2 Partial unbundling of distribution in IPPs only; vertically integrated IPPs and 1 regional distribution company one region utility functions as single buyer Tanzania No: vertically integrated national utility plus No IPPs only; vertically integrated IPPs utility functions as single buyer Uganda Yes: generation, transmission, and distribution Partial unbundling in distribution, Single-buyer model fully separated some rural concessions given out Ukraine Yes: generation, transmission, and distribution Yes: 4 generators and 27 distributors Single-buyer model, fully separated transitioning to market Vietnam Yes: generation, transmission, and distribution Yes: 3 generators and 5 distributors Single-buyer model, separated, but under a single holding company under a single holding company transitioning to market Source: Based on Rethinking Power Sector Reform utility database, 2015. Note: IPP = independent power producer. a. In 2014 Andhra Pradesh was divided into two states and the four distribution companies were divided equally between the two. 128 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 4.11 Most countries deploy some version of the single-buyer model in their power sectors The power sector in developing countries, by degree of competition, 1995–2015 100 90 32 36 80 47 50 70 67 Share of countries (%) 60 20 17 50 13 14 40 20 25 30 7 23 24 6 20 5 3 23 15 1 10 10 10 9 2 5 7 7 2 0 00 04 06 09 08 05 03 02 07 01 10 14 96 99 98 15 13 12 95 97 11 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 19 19 19 19 19 Retail market competition Wholesale market competition Bilateral contracting Single-buyer model Independent power producers Monopoly Source: Foster and others 2017. developing world (figure 4.11). Overall, this smaller state-owned companies, that absent represents about 44 percent of all developing other measures may suffer from similar perfor- countries. In half of these cases, the sin- mance problems as the original parent com- gle-buyer model purely consists in an opening pany. Thus unbundling cannot be counted on of generation to IPPs, whereas in the other half in and of itself to improve the performance of the incumbent’s generation assets have been the system overall (Bacon 2018). The one ben- divested to reduce conflicts of interest. Among efit that may result from unbundling per se is the 15 observatory countries, almost all oper- an increase in transparency and managerial ate some form of the single-buyer model, accountability. Whether this benefit has an except for 4 with fully functioning wholesale impact on utility performance is likely to power markets (Colombia, India, Peru, and the depend on the quality of governance and regu- Philippines) (table 4.2). A few others are cur- latory oversight, to ensure that the resulting rently in transition from the single-buyer information is put to good use. model to a wholesale power market (including In the Indian state of Andhra Pradesh, Ukraine and Vietnam). restructuring led to increased transparency and Although unbundling may bring some ben- accountability, affecting utility performance. efits of its own in terms of transparency and By the late 1990s, the power sector in the state accountability, it is debatable whether those was suffering from power shortages and declin- benefits are enough to drive performance ing quality of service, and contributing to a sig- improvements. An unbundling reform con- nificant fiscal crisis. The state embarked on a ducted in isolation simply creates several comprehensive power sector reform plan, What Has Been Done to Restructure Utilities and Improve Governance? 129 which began in 2000 with unbundling the the prospects of developing meaningful com- Andhra Pradesh State Electricity Board petition may be limited. Moreover, although (APSEB) into a generation company, Andhra having multiple distribution companies in a Pradesh Power Generation Corporation country may increase managerial transparency (APGENCO); a transmission company, and accountability, the associated benefits will Transmission Corporation of Andhra Pradesh materialize only if sound governance and regu- Limited (APTRANSCO); and four distribution latory oversight is in place, and even then may companies. The restructuring allowed the not be large enough to compensate for the loss unbundled entities to focus on their core busi- of economies of scale. Often the profitability of ness areas while the government provided the distribution utilities reflects the concentration necessary backing through improved regula- of industrial and commercial customers—​ tion and legislation to outlaw power theft. The as well as relatively affluent residential restructuring had an immediate impact on customers—in metropolitan areas, whereas ­ reducing transmission and distribution losses provincial utilities have a more meager cus- and improving collections. Statewide losses fell tomer base and sometimes higher costs due from 38 percent in 1999 to 20 percent in 2004, to lower population density. This situation is whereas collections rose from 92 percent to illustrated by data from Peru, where the profit 98 percent in the same period. ­ m argin per client in four of the regional The costs of implementing unbundling are ­ distribution utilities (with customer bases of not insignificant (Pollitt 2008; Vagliasindi 2012). 200,000–400,000 connections) is just a fraction Utility restructuring is a complex process entail- of the profit margin of metropolitan area utili- ing significant transaction costs, potentially ties (with customer bases of 700,000–900,000 amounting to tens of millions of dollars, due to connections) (figure 4.12). Hence the breakup the need for an exhaustive inventory of human of a national utility in a small country risks cre- and physical assets; full revision of company ating one or two business units that are com- accounts; installation of meters to monitor mercially attractive and several others that power flows across new company boundaries; may be much less viable. The presence of larger and the legal work associated with the creation customers with a stronger ability to pay within of independent governance structures for each of the subsidiary companies. Unbundling also FIGURE 4.12 Horizontal unbundling of the distribution sector creates higher fixed costs, associated with repli- in Peru created a couple of large profitable metropolitan cating board structures across different compa- utilities and a number of small regional utilities with limited nies, that may prove challenging to implement scope for profits, 2002 in countries with a scarcity of managerial skills. Net profit per client across distribution utilities in Peru after Furthermore, in power systems that lack strong horizontal unbundling, 2002 payment discipline, the unbundling of the sec- Luz del Sur (Lima) 248 tor risks creating a cascade of indebtedness EDELNOR (Lima) 114 across the various resulting entities. Electro Centro 49 The horizontal unbundling of distribution utilities, where no real prospects for competi- Electro Noroeste 32 tion exist, also risks the loss of economies of Electro Norte 26 scale and the creation of unprofitable business Electro Norte Medio 12 units. The breakup of national distribution util- 0 50 100 150 200 250 300 ities into smaller business areas needs to be Net profit per client (S/.) considered with caution, particularly in smaller Source: Based on Rethinking Power Sector Reform utility database, 2015. countries where the loss may be material and Note: S/. = Peruvian nuevos soles. 130 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD the context of a national utility also provides a companies may limit the extent to which the basis for cross-subsidies across service areas, various subsidiary companies can operate inde- which often occur implicitly through national pendently of one another. service areas. Although geographical cross-sub- For example, the Egyptian power sector has sidies could also take place, in principle, across seen various forms of unbundling and restruc- regional utilities through an explicit service turing since the 1960s; however, decision-­ levy of some kind, they may be more difficult making power has remained centralized to achieve politically because they involve a throughout. In the 1960s, private utilities were visible cross-jurisdictional financial transfer. nationalized and the Egyptian Establishment The balance between the costs and benefits for Electricity, under the Ministry of Energy, of unbundling is unlikely to be favorable in was created to oversee the generation, distribu- smaller power systems, where transaction tion, and transmission businesses. In 1976, the costs can weigh quite heavily. For this reason, Egyptian Electricity Authority (EEA), also unbundling is not typically recommended in under the Ministry of Energy, was created to power systems below 1 gigawatt (GW) in size jointly manage both the generation and distri- (Besant-Jones 2006), and may not become a bution businesses, and in 1978 seven regional priority until significant scope for competition distribution companies were established. This starts to materialize in systems of 3 GW and framework continued until 1993, when distri- above. In larger power systems, the benefits of bution utilities were moved from the EEA to unbundling in terms of increased transpar- the Ministry of Public Enterprise. In 1998, dis- ency and accountability are likely to outweigh tribution companies were transferred back to the costs. Overall, although 70 percent of the Ministry of Energy, where the generation countries with power systems smaller than and distribution businesses were bundled into 1 GW have vertically integrated power utili- seven vertically integrated SOEs under ties, that share falls to just over 20 percent for the EEA. In 2000, a further sector reorganiza- countries with power systems larger than tion took place, whereby EEA was again 20 GW. Nevertheless, a significant number of unbundled and turned into a holding larger countries have still not unbundled company—​ ­ the Egyptian Electricity Holding incumbent utilities. Indonesia is the world’s Company (EEHC)—with five generation fourth-most-populous country with over u tilities, a transmission utility, and seven ­ 60 GW of installed capacity (in 2017), and it ­ d istribution utilities. Despite this repeated still has a state-owned vertically integrated restructuring, the underlying governance and national utility, Perusahaan Listrik Negara management arrangements of the distribution (PLN), which has a monopoly on electricity companies were not substantially altered. distribution and transmission while having a In Pakistan, too, old decision-making struc- majority share of generation assets as well. tures continued to prevail after the original The impact of unbundling, when all compa- national utility had been unbundled. Before nies remain under a common state ownership 1994, Pakistan’s entire population was served structure, may be limited. Several countries— by two vertically integrated companies: the such as Egypt and Vietnam—have unbundled Water and Power Development Authority in the context of continued state ownership to (WAPDA) and the Karachi Electric Supply create large state-owned holding company Company (KESC). A 1970–75 plan proposed structures, as part of a transition toward a com- the separation of generation and distribution, petitive market. When unbundling is envis- making the case that WAPDA had become too aged as the foundation of a competitive mar- large and unwieldy to “shoulder the responsi- ket, common ownership of the unbundled bility of retail distribution and generation of What Has Been Done to Restructure Utilities and Improve Governance? 131 power” (Planning Commission 1970). By the On the demand side, net load rather than late 1980s and early 1990s, WAPDA was expe- peak demand becomes the moving target for riencing severe governance issues and declin- planning. Rather than a mostly consistent load ing operational performance, which provided forecast, growing reliance on DERs leads to the final impetus for reform. Following the greater uncertainty about how much consum- WAPDA Amendment Act, WAPDA’s power ers will reduce their total demand, or how wing was unbundled in 1997 to form 15 incor- much consumers might supply back to the porated state-owned entities comprising grid. As a result, the whole profile of expected 3 thermal generating companies, 1 national demand may shift. If, for example, during the transmission and dispatch company, and typical peak2 system load occurring during the 8 regional distribution companies. Hydropower day,3 an increased amount of solar production generation and water management remained is available from customer-sited rooftop solar with WAPDA. However, in 2005, seven years photovoltaic (PV), then the total amount of after the passing of the law, the distribution grid-connected (or utility-scale) resources companies were still not operating in a fully needed to meet consumer demand decreases. autonomous manner. According to Parish This difference between the typical, expected (2006), the finances of the utilities and their peak load and the total amount of customer-­ tariff applications were wholly handled by sited renewable resources4 is known as the net WAPDA, which even had some of the senior load (IEA Wind Task 25 2013). This net load managers of the unbundled companies on its can change throughout the day, for instance, payroll and was closely involved in appoint- shifting the peak to the period when decentral- ments to the board of directors. The resulting ized solar generation dips at the end of the day. institutional confusion prejudiced the perfor- Forecasting the net load—and the resources mance of the unbundled companies. needed to quickly ramp up to the full output needed to meet this net load—are new chal- LOOKING AHEAD lenges for system operators and system plan- Technological disruptions underway in the ners alike. power sector will make planning increasingly On the supply side, planners are increasingly complex and may carry implications for concerned with building in greater system flexi- the sector’s structure. The combination of bility. Uncertainty about the output of renew- increased uncertainty on both the supply side able supply increases the need for fast-ramping and the demand side complicates system plan- resources (that is, resources that can adjust pro- ning and grid operations. When power is gen- duction up or down quickly) that can respond erated from fully dispatchable resources and when the sun is not shining, or the wind is not assumed to flow only one way—from a genera- blowing.5 As renewable penetration increases, it tion source, across wires, to a consumer—both becomes increasingly challenging to maintain supply and demand are relatively predictable. supply/demand balance while ensuring suffi- Day-to-day grid operations and long-term sys- cient voltage and frequency support to preserve tem planning focus on meeting the system reliability of supply. This challenge in turn cre- peak, and ensuring adequate reserve margins. ates the need to build out additional transmis- The increased penetration of both variable sion capacity, both within and across national renewable energy resources and distributed boundaries. Greater transmission capacity energy resources (DERs) however, introduces allows the system to fully absorb variable uncertainty simultaneously to the supply and renewable generation when it is available, demand sides of electricity, respectively, and avoiding curtailment, and provides larger this relatively simple formula begins to change. ­ geographical areas across which to balance 132 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD supply variability. With variable renewable be discharged at any moment in response to energy, supply forecasting also becomes more system needs. This resource makes it easier to complex, and requires better weather forecasts instantaneously match supply with demand, to ensure that the right mix of backup resources because there could always be stored supply is available to meet system needs (including available to meet demand at any time, if stor- ramping, voltage control, frequency regulation, age capacity is available at adequate levels and and black start), bearing in mind that these sys- in appropriate locations. Unlike any other tem needs could change across the day, over resource, storage can also be used for transmis- several hours, or even within the hour. sion. Transmission lines are designed to allow a In addition, planning the distribution sys- certain amount of electricity to flow through tem is getting more challenging. The traditional them, and they must be operated within these system peak is declining, net load consider- limits to avoid damaging the transmission lines ations are more relevant, and the concept of or damaging generation units that are con- what constitutes peak system operations is nected to these lines.7 By shifting when and changing altogether. Indeed, in some European how generation resources are used, storage distribution systems, peak system usage is helps to manage transmission congestion (that already occurring during the times of greatest is, manage the desire to flow more electricity consumer production, not consumer consumption across transmission lines than is technically (Burger and others 2018).6 feasible) or even increase the total hosting So much uncertainty brings a significant capacity of distribution system lines (that is, risk of stranded assets. Assets could get accommodate customer-sited resources, like stranded if the customer demand that grid rooftop solar PV, on an existing system without infrastructure, or generation plants, are requiring additional upgrades). Storage can designed to support is not realized. For exam- also avoid the need for some transmission and ple, large generation assets that are inflexible distribution upgrades altogether. Reducing the will be less helpful to system operators when overall system peak lessens the total system- flexibility is needed. Inflexible, typically ther- wide build-out that grid operators need to plan mal, generation assets are not designed to turn in order to meet system peak demand.8 Finally, on and produce lots of electricity quickly. storage can extend the life of existing transmis- Instead, they are designed to turn on and oper- sion and distribution system assets by allowing ate within a particular range for long periods of for the optimal use of those assets. time. If these units are constantly cycled (that How these different resources, on both the is, turned on and off), thermal efficiency is supply and demand sides, can be used to reduced and the useful life of the plant enable the transition to a low-carbon grid will shortened. be an ongoing challenge. Long-term planning As storage technologies become commer- is already complex even for traditional power cially viable, they could provide a variety of systems based on one-way power flow from grid services to facilitate the integration of centralized grid resources. These challenges are renewable generation and distributed energy compounded in emerging modern systems resources. How storage technologies are used, where consumer demand fluctuates and vari- however, has implications for day-to-day grid able renewable energy introduces the need for operations and long-term system planning. different kinds of system resources. Going for- Storage technologies, in general, enable gener- ward, the changing electricity paradigm will ation supply to be stored and used for con- require decisions about the kinds of resources sumption later. Battery storage, in particular, is that need to be available to grid operators to an example of a fast-ramping resource that can integrate these variable resources, and system What Has Been Done to Restructure Utilities and Improve Governance? 133 planning models will need to evolve to accom- Many unanswered questions remain modate consumer-sited DERs, and the many regarding how DERs should be integrated into possible uses of storage technologies. both transmission and distribution systems. In Furthermore, the advent of low-cost DERs markets where the transmission and distribu- increasingly calls into question the validity of tion segments are clearly separated, with dif- the traditional utility business model. This ferent system operators managing each net- model reflects the historic cost structure of the work, questions about how to manage the power sector, specifically, the presence of integration of DERs with separate system oper- e conomies of scale in power generation, ­ ators are largely unanswered. If DERs partici- ­ transmission, and distribution, as well as large pate in both retail and wholesale markets, economies of scope in the tight coordination of which system operators do these resources energy flows along the electricity supply chain. respond to? Can the different services DERs These elements led to the creation of provide (for example, energy, regulation, ­ national-scale vertically integrated power utili- reserves, and load management) be clearly dis- ties. As technological disruption shifts the tinguished in either system? If so, how should underlying cost structure of the industry, the their market impact on the other system be logic behind the institutional organization of considered? These are complicated questions, the sector also changes. The interplay between whose answers depend on different legal and shifting cost structures and regulatory jurisdictional realities that remain unclear in responses is likely to affect the institutional many of these markets. evolution of the sector going forward. Most regulators have focused instead on the Distribution system operators are moving future role of the distribution system operator, from a passive role to an active one. and what regulatory models might be needed to Previously, they sent electricity from central enable it. Generally, two approaches are being generators directly to consumers, whereas considered. Under the first approach, the distri- now they increasingly manage various kinds bution system operator retains its functions as of DERs that might be consuming, producing, owner and operator of the distribution system, or managing loads. Many of these activities amid changes to the regulatory model to facili- are being undertaken by a range of private tate the integration of DERs. Under the second sector actors, including prosumers as well as approach, an independent entity manages the new entrants that see business opportunities operation and planning of the distribution sys- in aggregating demand or providing decen- tem to ensure that all market activities are sepa- tralized energy resources across the distribu- rated from these functions. In other words, an tion grid. The choice of how to best integrate independent distribution system operator DERs often depends on existing market struc- stripped of retail functions exists and is similar tures and existing regulatory models. Where to some markets at the transmission level. vertically integrated utilities are dominant, The distribution utility becomes a market utilities are likely to continue playing a role in enabler in this platform/network model, integrating various DER technologies. Where whereby various sellers aggregating various competitive markets at the retail level exist, DERs can trade or provide services to the distri- the ongoing role of the utility is less clear. In bution system through the distribution system these markets, utilities do not own genera- platform. New York state is considering several tion. DER integration instead focuses more on utility cost-recovery models that facilitate plat- enabling third-party ownership of DER assets form development and data sharing with third and transitioning the utility into an entirely parties to foster transparent and open markets. different role. More advanced metering and more software 134 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD and controls are needed to enable these kinds where no clear separation exists between the of structures. Regulators may allow utilities to transmission system operator and the distribu- recover the costs associated with infrastructure tion system operator, questions about the that enables such a transactive platform (for ­ utility ownership of DER assets, or about the example, smart metering, or other grid mod- participation of these resources in various sys- ernization efforts), or enable utilities to recover tems, are less relevant. Moreover, a vertically new revenue streams associated with provid- integrated utility that still manages generation ing “value-added” services (for example, pro- may find it easier to deploy DERs where they viding system data for third-party usage) to would provide the most system benefits, as create a more platform-based structure. well as capture the entire value that these These emerging regulatory models, how- resources can provide.10 In places where utili- ever, require regulators to rethink where, ties are largely unbundled and no longer own when, and how utility capital investments are generation assets, and competitive retail mar- made. They also require regulators to consider kets already exist, it may be possible to con- whether utility ownership of various assets (for sider a complicated new market design to fully example, storage or electric vehicle charging integrate DERs. And, as with the challenges infrastructure) that enable the greater integra- associated with integrating renewables into tion of renewable generation could also enable grid operations, if the price signal alone is (or hinder) a more transactive platform—or intended to incentivize new resources, then competitive distribution markets more gener- markets and tariffs must be designed so as ally. The challenge is to determine whether the to accurately value the full services these utility or some other entity altogether is best resources provide to grid operations, including positioned to manage the integration of these both their time and their locational value. resources. In unbundled, competitive distribu- Where the development of a utility service tion markets, there is concern about utility remains incipient and regulatory capacity is ownership of distributed energy resources and weak, there is the real possibility of leapfrog- the functions associated with the management ging toward decentralized models of service of the distribution system that enable a more provision. In many developing countries, nei- active distribution system operator. If regula- ther of the above conditions holds. Utilities may tors want to ensure that competitive distribu- still be incipient, serving only a fraction of the tion markets are maintained, then open potential market, with electrification rates occa- and transparent access to data coming from sionally as low as 10–20 percent, and more typ- consumer meters or electric vehicle charging ically in the 40–60 percent range. Even where stations—data that utilities typically do not service is available, it may be highly unreliable share—will be required (Kufeoglu, Pollitt, and with frequent interruptions that leave consum- Anaya 2018).9 If regulators want to ensure that ers to rely on their own backup generation competitive wholesale or retail markets are options, which have traditionally been diesel maintained, then a utility’s ownership of distri- based. Although explicit or implicit subsidiza- bution system assets (like energy storage or tion of grid electricity is widespread across the electric vehicle charging stations, that could developing world, a significant subset of coun- also provide services to the wholesale market) tries may nonetheless face exceptionally high may not be advisable. tariffs (and even higher costs) for grid electricity The relevance of platform-based distribution due to geographical disadvantages (as in small models very much depends on the legacy mar- islands and landlocked countries) that ket structure in the sector. In places where ver- leave them dependent on small-scale oil-fired tically integrated utilities remain dominant, or generation. Added to this problem may be high ­ What Has Been Done to Restructure Utilities and Improve Governance? 135 levels of inefficiency in the distribution seg- 57 percent. The governance practices seen the ment, and a marked tendency toward least in public utilities include the autonomy of cross-subsidization—to the detriment of non- board decision making, transparency of board residential customers. All of these factors con- appointment and removal processes, and the spire to make DERs increasingly attractive in adoption of good financial accounting and staff many parts of the developing world, both as recruitment practices. off-grid alternatives to grid electrification and as Restructuring utilities was a central focus of backup for inadequate grid supply. As storage reform efforts but proved difficult to imple- technology further improves and becomes ment and may have distracted attention from more cost-­ effective, the possibility of relying on more fundamental issues. In the 1990s, as rec- self-­ generation, rather than the often unreliable ommended by International financial institu- electricity supply from the grid, will become tions, many countries undertook major sector increasingly attractive to individual consumers, restructuring exercises. Barely 20 percent irrespective of whether self-generation is eco- achieved the full vertical and horizontal nomically efficient for the system as a whole. unbundling envisaged. As a result, close to half Mini/microgrids with innovative payment of the world’s developing countries operate structures already play a key role in expanding under some variation of the single-buyer energy access in rural areas and could increas- model. Because unbundling was primarily ingly provide alternative supply options to conceived as a means to deeper reform, the urban customers as well. Grid defection, how- case for vertical and horizontal unbundling is ever, will exacerbate the existing cost-recovery questionable in smaller systems not able to challenges of developing countries’ utilities, and progress toward a wholesale power market. integration of (both centralized and) distributed The benefits of the process need to be balanced renewable energy will exacerbate existing tech- against the potential loss of economies of scale nical challenges in grid operation. The relatively and scope. With their emphasis on regulation weak capacity of regulators in such environ- and restructuring, the sector reforms of the ments will make it challenging to guide this 1990s often underemphasized the critical func- process in an optimal manner. tion of sector planning, which remains inade- quate in many countries. CONCLUSION Looking ahead, technological disruptions In summary, corporatization has led state- currently underway in the power sector will owned utilities to adopt many good governance make planning increasingly complex and may and management practices, even if they still lag carry implications for the sector’s structure. The practice in the private sector. Good governance advent of prosumers and other third-party and management depend on the adoption of a actors (such as demand aggregators) able to range of widely accepted measures that relate to deploy decentralized generation, storage, or the autonomy and accountability of the board demand-response solutions complicates the task of directors, the extent of financial discipline, of planning, which affects both generation and the handling of human resources and infor- expansion and grid development and has mation technology. Overall, only about 55 per- prompted a shift toward concepts of net load. At cent of such practices can be found (on paper at the same time, the presence of these new play- least) in entirely state-owned utilities, compared ers in markets that have already undergone with about 80 percent in private utilities. In extensive restructuring underscores the possibil- countries where public utilities coexist alongside ity of competition in the retail tier, with the util- private utilities, the share of good-governance ity acting as a distribution system operator that practices among the public utilities rises to provides the platform across which other parties 136 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD trade energy on the local grid. Where sectors alternatives to balance demand and supply. remain vertically integrated, the issue is more Finally, where utility networks remain underde- one of incentivizing the utility to consider veloped, new technologies offer the option of the deployment of decentralized off-grid leapfrogging to a decentralized service model. ANNEX 4A. UTILITY RESTRUCTURING INDEX, 2015 Percent Utility restructuring Vertical Horizontal Country/State (overall) unbundling unbundling Colombia 35 70 0 Dominican Republic 67 100 33 Egypt, Arab Rep. 37 40 33 India – Andhra Pradesh 57 80 33 India – Odisha 73 80 67 India – Rajasthan 57 80 33 Kenya 25 50 0 Morocco 0 0 0 Pakistan 73 80 67 Peru 73 80 67 Philippines 100 100 100 Senegal 0 0 0 Tajikistan 0 0 0 Tanzania 0 0 0 Uganda 55 60 50 Ukraine 63 60 67 Vietnam 47 60 33 International 45 55 34 benchmark Source: Based on Rethinking Power Sector Reform utility database, 2015. ANNEX 4B. PLANNING AND PROCUREMENT INDEX, 2015 Percent Planning and procurement Generation Generation Transmission Transmission Country/State (overall) planning procurement planning procurement Colombia 95 86 95 100 100 Dominican Republic 72 29 86 75 100 Egypt, Arab Rep. 82 71 100 75 83 India – Andhra Pradesh 78 57 95 75 83 India – Odisha 78 57 95 75 83 India – Rajasthan 78 57 95 75 83 Kenya 82 86 100 75 67 Morocco 61 43 100 50 50 Pakistan 63 29 100 25 100 Peru 77 43 90 75 100 Philippines 59 71 100 50 17 Senegal 59 43 50 100 42 Tajikistan 64 71 100 50 33 Tanzania 77 43 100 75 92 Uganda 76 43 95 75 92 Ukraine 38 60 0 75 17 Vietnam 59 71 50 100 17 International benchmark 71 56 85 72 68 Source: Based on Rethinking Power Sector Reform utility database, 2015. ANNEX 4C. UTILITY GOVERNANCE INDEX, 2015 Percent Overall Country/ utility Corporate Utility Financial Human Information State Utility governance governance Accountability Autonomy management discipline resource and technology Colombia EPM 80 76 75 78 83 76 86 87 CODENSA 69 96 92 100 43 69 60 0 Dominican EDESUR 51 50 33 67 52 29 50 79 Republic EDENORTE 57 63 58 67 52 21 50 86 Egypt, Arab EEHC 53 44 33 56 61 53 71 60 Rep. EEHC (Discos) 55 49 42 56 61 53 71 60 India – APSPDCL 52 47 50 44 56 53 43 73 Andhra APEPDCL 52 47 50 44 56 53 43 73 Pradesh India – WESCO 68 86 83 89 50 43 70 36 Odisha CESU 26 13 25 0 40 36 40 43 India – JVVNL 63 67 67 67 60 64 36 80 Rajasthan JDDVNL 63 67 67 67 60 64 36 79 Kenya Kenya Power (KPLC) 90 100 100 100 80 76 64 100 Morocco ONEE 53 35 25 44 70 57 79 73 Pakistan LESCO 52 56 67 44 48 50 43 50 KE 90 94 100 89 86 79 86 93 Peru Luz del Sur 85 85 92 78 85 86 90 80 Hidrandina 55 40 58 22 70 65 71 73 Philippines MERALCO 90 100 100 100 81 71 79 93 BENECO 76 83 67 100 68 53 86 67 Senegal Senelec 65 74 58 89 56 57 57 53 Tajikistan Barki Tojik 52 42 17 67 62 64 57 64 Tanzania TANESCO 61 57 58 56 65 64 71 60 Uganda UMEME 80 85 92 78 76 77 71 80 Ukraine Khmelnitskoblenergo 75 82 75 89 67 64 57 80 Dniproblenergo 72 69 50 89 74 65 79 80 Vietnam NPC 37 8 17 0 65 53 50 93 HPCMC 37 8 17 0 65 53 50 93 International benchmark 63 62 60 63 64 59 62 71 Source: Based on Rethinking Power Sector Reform utility database, 2015. 137 138 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD ANNEX 4D. UTILITY CLASSIFICATION, 2015 Utility Country/State Ownership APEPDCL India–Andhra Pradesh Public APSPDCL India–Andhra Pradesh Public Barki Tojik Tajikistan Public BENECO Philippines Private CESU a India–Odisha Public CODENSA Colombia Private Dniproblenergo Ukraine Private EDENORTE Dominican Republic Public EDESUR Dominican Republic Public EEHC Egypt, Arab Rep. Public EEHC (Discos) Egypt, Arab Rep. Public EPM Colombia Public Hidrandina Peru Public HPCMC Vietnam Public JDDVNL India–Rajasthan Public JVVNL India–Rajasthan Public Karach Electric Pakistan Private Kenya Power (KPLC) b Kenya Private Khmelnitskoblenergo Ukraine Public LESCO Pakistan Public Luz del Sur Peru Private MERALCO Philippines Private NPC Vietnam Public ONEE Morocco Public Senelec Senegal Public TANESCO Tanzania Public UMEME Uganda Private WESCOa India–Odisha Private Source: Rethinking Power Sector Reform Observatory. a. The Western Electricity Supply Company of Orissa Ltd. (Wesco) in India–Odisha was a private utility during the study period, but the state took over the utility in 2016. The Central Electricity Supply Utility of Odisha (CESU) had gone back to the state in 2001 itself and as such is considered a public utility for the study. b. The utility in Kenya has a 49 percent stake floated on the Nairobi Stock Exchange and 51 percent is owned by government. However, day-to-day decision making is not in government hands for all intents and purposes. The utility is treated as private for ­ this study. What Has Been Done to Restructure Utilities and Improve Governance? 139 ANNEX 4E. UTILITY RESTRUCTURING INDEX, 2015 Applicable only to the Rethinking Observatory countries Utility restructuring index Average scores Horizontal Vertical unbundling unbundling Average scores Average scores Generation Distribution Generation Distribution Transmission Average scores Average scores Average scores Average scores Average scores 1. Separate 1. Generation split 1. Distribution split 1. Separate 1. Separate transmission entity into separate into separate generation entity distribution entity 2. Separation is: entities entities 2. Separation is: 2. Separation is: Accounting 2. Separation is: 2. Separation is: Accounting Accounting Functional Accounting Accounting Functional Functional Legal Legal Legal Legal Legal Ownership Ownership Ownership Ownership Ownership 3. Open, 3. Generation company 3. Distribution company nondiscriminatory, not allowed to own not allowed to own third party access distribution entities generation entities to transmission. Source: Original figure for this publication. 140 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD NOTES REFERENCES 1. This chapter is based on original research Bacon, R. 2018. “Taking Stock of the Impact ­ conducted by Vivien Foster, Anshul Rana, of Power Utility Reform in Developing Joeri de Wit, and Victor Loksha, supported Countries.” Policy Research Working Paper by an advisory team comprising Pedro 8460, World Bank, Washington, DC. Antmann, Pedro Sanchez, Elvira Morella, Besant-Jones, John E. 2006. “Reforming Power and Mariano Salto. Markets in Developing Countries: What Have 2. The peak could change substantially, depend- We Learned?” Energy and Mining Sector ing on the kind of DERs that are dominant. Board Discussion Paper No. 19, World Bank, A significant amount of distributed rooftop Washington, DC. solar PV could diminish the typical peak, Burger, Scott, Jesse Jenkins, Carlos Batlle, and whereas a significant number of electric Ignacio Perez-Arriaga. 2018. “Restructuring vehicles or their charging stations could ­ Revisited: Competition and Coordination in increase the overall system peak. Electricity Distribution Systems.” Working Paper 3. This example assumes a system load forecast 007, MIT Center for Energy and Environmental with a daytime peak. Although daytime is a Policy Research, Cambridge, MA. common peak period, as we will see, at the Foster, Vivien, Samantha Witte, Sudeshna Ghosh circuit level on distribution systems, the peak Banerjee, and Alejandro Moreno. 2017. may differ. “Charting the Diffusion of Power Sector 4. The net load is typically considered in terms of Reforms across the Developing World.” Policy the difference between the expected peak and Research Working Paper 8235, World Bank, the amount of renewable resources, but it is Washington, DC. in fact the difference between any type of cus- Gratwick, K., and A. Eberhard. 2008. “Demise of tomer-sited DER and the expected peak load. the Standard Model of Power Sector Reform 5. Renewable resources can also provide ramp- and the Emergence of Hybrid Power Markets.” ing capability to the grid, because they can Energy Policy 36 (10): 3948–60. quickly ramp up to full output or down to IEA WIND Task 25. 2013. “Design and Operation zero output (that is, by being curtailed), but of Power Systems with Large Amounts of only when the sun is shining or the wind Wind Power.” Final summary report, phase is blowing. Thermal resources, by contrast, three, VTT Technical Research Centre of require output at some minimum level, and Finland Ltd. so can only be ramped down to a specific Kufeoglu, Sinan, Michael Pollitt, and Karim level, or nuclear units, which cannot provide Anaya. 2018. “Electric Power Distribution in any ramping capability. Although solar and the World: Today and Tomorrow.” Cambridge wind resources can be curtailed, some com- Working Paper in Economics 1846, University mon operating procedures can limit the full of Cambridge. curtailment of wind resources. OECD (Organisation for Economic Co-operation 6. On some distribution circuits, the peak usage and Development). 2015. OECD Guidelines on is driven by consumer production (that is, Corporate Governance of State-Owned Enterprises. from rooftop solar). Vienna: OECD. 7. Transmission lines must be operated within Pardina, M. R., and J. Schiro. 2018. “Taking particular thermal, voltage, and stability limits. Stock of Economic Regulation of Power 8. As we shall see, consumer-sited DERs of Utilities in the Developing World.” Policy all types, including storage, could also play Research Working Paper 8461, World Bank, this role. Washington, DC. 9. For a discussion of some of these challenges, Parish, D. 2006. Evaluation of the Power Sector as well as of existing distribution system oper- Operations in Pakistan. Manila: Asian ator models around the world, see Kufeoglu, Development Bank. Pollitt, and Anaya (2018). Planning Commission, Pakistan. 1970. The Fourth 10. However, there are still integration challenges, Five Year Plan (1970–75). Islamabad: Planning as well as challenges with cost recovery. Commission. What Has Been Done to Restructure Utilities and Improve Governance? 141 Pollitt, M. 2008. “The Arguments for and An Aid for Practitioners.” Live Wire Knowledge against Ownership Unbundling of Energy Note Series No. 73/2017, World Bank, Transmission.” Energy Policy 36 (2): 704–71. Washington, DC. Rudnick, H., and C. Velasquez. 2018. “Taking Stock Vagliasindi, M. 2012. “Power Market Structure of Wholesale Power Markets in Developing and Performance.” Policy Research Working Countries.” Policy Research Working Paper Paper 6123, World Bank, Washington, DC. 8519, World Bank, Washington, DC. World Bank. 2014. Corporate Governance of State- Steinbuks, J., J. de Wit, A. Kochnakyan, and Owned Enterprises: A Toolkit. Washington, DC: V. Foster. 2017. Forecasting Electricity Demand: World Bank Group. What Has the Private Sector Contributed? 5 Guiding questions • How far did private sector participation (PSP) evolve in developing country power sectors? • What form did PSP predominantly take? What were the main implementation challenges? Summary • A broad cross-section of countries introduced some degree of PSP in electricity, using various modalities and relying primarily on foreign ­ ­ investment. • The private sector has contributed greatly to the expansion of power generation capacity in the developing ­ ­ world, even as public investment continues to play an important role. Nevertheless, governance issues have surfaced when capacity is not procured in a competitive ­ ­ fashion. More- over, governments have struggled to strike the right balance in allocating public and private risk in contracts governing independent power p ­ ­ rojects. • Although the reforms of the 1990s did not focus on privatizing transmission, Latin America and Asia had noteworthy and broadly positive experiences with PSP for power grids. • PSP in power distribution was quite widely adopted among the early generation of reformers. Since the early 2000s, however, PSP in distribution has experienced notable setbacks and uptake has largely declined. • Distribution privatization raises delicate challenges because the interests of key stakeholders are involved. Labor unions are anxious about potential layoffs, while customers are sensitive about the ­ prospect of tariff hikes, as well as eager to see improvements in service quality and coverage. INTRODUCTION electricity supply chain, particularly in gener- This chapter evaluates the extent to which the ation and d ­ istribution. Was the introduction private sector has played its envisaged role in of private sector participation (PSP) feasible in power sector reform across the developing developing country power sectors? What ­world.1 It examines how countries went about were the challenges? What form did PSP pre- incorporating the private sector into the dominantly take? Did the experience of 143 144 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD private sector ownership and operation func- time, oversight of output and service quality tion as originally envisaged? would safeguard the pursuit of legitimate PSP was a pillar of the 1990s power sector public ­interests. reform model (World Bank 1993). Many of its The private sector was also touted as a measures were designed to encourage ­ PSP. source of potential investment in the power Power utilities were unbundled in an effort to sector. Investment by SOEs was often con- ­ create an industry structure that isolated the strained by their weak balance sheets—­ natural monopolies (notably transmission and constraints that prevented many of them from distribution [T&D]) in the supply chain, leav- raising financing from commercial banks, capi- ing intact the potentially competitive elements, tal markets, or internal cash ­ generation. State- notably generation and ­ retail. PSP could then owned utilities were therefore reliant on loans be introduced wherever relevant, thereby side- and grants from the central government, often stepping the transfer of national monopolies concessional in nature and originating with from public to private h ­ ands. Similarly, regula- international ­ donors. This dependence was tory entities were considered as important pre- particularly true in times of fiscal austerity, and cursors for PSP because they can build on the it inhibited the timely expansion of the s ­ ector. commercial orientation of private ­ operators. Even where public utility finances were strong, This commercial behavior works in favor of the a state guarantee was generally required to public interest by ensuring adequate control underwrite commercial borrowing, leaving the over market power in monopoly segments— utility exposed to limits on public sector and by licensing market entry and encouraging ­ borrowing. Transfer of responsibility to the pri- ­ competition. At the same time, the endpoint of vate sector was expected to ease these restric- the 1990s reform model—a wholesale compet- tions by allowing finance to be raised directly itive power market—was largely premised on from the markets without fiscal checks, as long the existence of multiple private sector players as private operators were able to improve the sector. in the ­ financial performance of ­ utilities. Although the The private sector was expected to bring a 1990s model was neutral about domestic or more robust commercial orientation, which foreign investment in privatization, the advent would bring a turnaround in sector of foreign participation opened up new reser- p erformance. The state-owned enterprises ­ capital. voirs of international ­ (SOEs) prevalent in the 1990s power sector It was recommended that PSP be introduced lacked clear commercial incentives (Bacon first in distribution and then in ­ generation. 1995). Enterprises tended to focus on social Power distribution and retailing are the cash and ­political objectives, from affordable access cows of the electricity supply chain, capturing to electricity for key electoral constituencies resources directly from customers that are then to furnishing patronage j ­obs. Remuneration used to purchase transmission and generation to directors and managers was not linked to services from upstream ­ providers. The entire o perational ­ ­ e fficiency. Moreover, manage- financial basis of the sector is undermined if ment seldom operated under budget con- these fundamental revenue-capture functions straints, accustomed as it was to periodic fail to function, which is why the power sector bailouts. The advent of private sector manage- ­ reform paradigm has reforms start at the distri- ment and ownership greatly simplified bution ­ end. Once revenue capture was func- matters. Utilities could focus on the bottom ­ tioning efficiently under private management, line, a focus that created strong incentives to the bankability of upstream investments in cut costs and boost ­ r evenues. At the same generation would be greatly ­ enhanced. What Has the Private Sector Contributed? 145 TABLE 5.1 Forms of private sector participation Operations Revenue Investment Ownership (operational risk) (commercial risk) (investment risk) (asset risk) Service/management Private sector Public utility Government Government contract (management fee) Lease contract/ Private sector Private sector Government Government distribution franchise (tariff revenues) Concession/build, operate, Private sector Private sector Private sector Government transfer contract (tariff revenues) Full privatization or Private sector Private sector Private sector Private sector divestiture (tariff revenues) elaboration. Source: World Bank ­ A number of different PSP modalities can be private sector develops new infrastructure placed along a spectrum according to the assets on a project-finance basis, on the basis of extent to which responsibilities are transferred ring-fencing of the resulting revenues for an to the private sector (table 5.1). At one extended period of at least 20 ­ years. The most extreme, a management contract temporarily complete and supposedly permanent form of delegates managerial responsibility for a utility privatization is asset sale or divestiture of either from the public to the private sector for a a minority or majority stake in the ­ company. It period of two to three ­ years. The management may be done through a direct sale or auction, contractor is remunerated directly by the gov- or via a stock market ­ flotation. ernment according to a fixed fee that is some- In the case of the power sector, the deeper times performance related but does not depend forms of PSP were believed to be the most on utility ­revenues. A further step would be a relevant. For many areas of infrastructure, ­ lease contract, where delegation of manage- only some of the weaker forms of PSP proved ment is typically for a longer period of 5–10 to be ­ f easible. Among water utilities, for ­ years. The lease contractor takes full responsi- instance, management, lease, and concession bility for operating the utility and depends on contracts are prevalent, whereas divestiture revenue collections for remuneration, paying a has been relatively ­ unusual. In the transport predetermined share of these revenues to the sector, concessions (for ports, airports, and rail- government as a lease to contribute to the ways) and BOT contracts (for toll roads) are financing of investments, which remain ­ common. In the case of the power sector, the the responsibility of a state-owned holding stronger characteristics of private electric ser- ­ company. A still-further step is a concession, vice, combined with its potentially lucrative whereby the private sector retains the entire nature, meant that deeper forms of PSP, such sector revenue stream and assumes all respon- as divestiture, appeared promising (Bacon sibility for investment, entailing a much longer 1999; Sen 2014). contract duration of at least 20 ­years. Whereas concessions are typically used to take over KEY FINDINGS existing infrastructure, an important variation Drawing on the World Bank’s Private for the case of greenfield infrastructure is the Participation in Infrastructure (PPI) database use of build, operate, transfer (BOT) or equiva- for broad global trends, and on the 15 coun- lent contractual arrangements, whereby the tries of the Power Sector Reform Observatory 146 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD for more detailed case experience, the main divestiture of existing assets and independent findings of this chapter can be summarized as power producers (IPPs) for development of ­follows.2 new ­c apacity. The majority of developing countries opened their markets to IPPs, mak- ing this among the most popular power sec- Finding #1: A broad cross-section of tor reforms and often the first to be countries introduced a degree of PSP undertaken. IPPs for the construction of new ­ in electricity, using various modalities generation plants were often, though not and relying primarily on foreign always, combined with divestiture of existing ­investment generation plants, although minority govern- Although prevalent in generation and distri- ment stakes were often ­ retained. In some of bution, PSP was comparatively rare in the the more advanced markets of the observa- transmission of ­ power. At the global level, tory ­sample—such as Colombia, India, Peru, almost two out of every three developing and the Philippines—there was also some countries implemented at least one entry of merchant plants into the generation t ransaction for PSP in power generation ­ market without the need for a public sector and electricity distribution (table 5.2). By ­offtaker. In the distribution segment, PSP pre- contrast, private participation in the transmis- dominantly took the form of asset s ­ ales. For sion segment occurred in only 15 percent of the most part, distribution privatizations were the observatory ­ countries. Notwithstanding undertaken either in the same time frame as its prevalence in generation and distribution, generation privatizations or (often) some private sector engagement varies greatly years ­later. across the 15 observatory ­ countries. In India, In Sub-Saharan Africa, electricity divesti- Pakistan, Senegal, and Tajikistan, PSP in tures were relatively rare, and PSP was skewed ­ d istribution is far from the ­ n orm. The toward contract-based m ­ odalities. Although Philippines, Uganda, and Ukraine, by con- the overall percentage of countries adopting trast, have privatized nearly all of their distri- any form of PSP in Sub-Saharan Africa was bution utilities (table 5.3). similar to other developing regions, the modal- The preferred PSP modality varies along ities adopted were d ­ ifferent. Divestitures were the electricity supply ­chain. In the generation comparatively rare, with PSP in generation segment, PSP comprised a mixture of p rimarily taking the form of IPPs, and in ­ TABLE 5.2 Preferred private sector participation modality varies across regions/electricity supply chain (global) Countries Sub-Saharan Africa Other developing countries implementing (%) 1990–2016 Generation Transmission Distribution Generation Transmission Distribution Management contract 3 0 34 5 1 10 Lease contract 0 0 3 4 7 3 Concession 20 0 26 29 5 16 BOT contract 46 3 3 59 8 5 Divestiture 14 0 9 38 7 51 Any of the above 63 3 66 75 16 64 ­ https://ppi.worldbank.org/). Source: Based on the Private Participation in Infrastructure database 2018 ( transfer. Note: BOT = build, operate, ­ What Has the Private Sector Contributed? 147 TABLE 5.3 Private sector engagement across the 15 observatory countries Generation Distribution Market Market Country/State Years share (%) Modality Years share (%) Modality Colombia 1993–2017 71 Extensive divestiture, 1996–2010 50 Mainly divestitures, plus one numerous IPPs, and a few concession merchant plants Dominican 1994–2017 61 Extensive divestiture, 1994–99 2 Partial divestitures that have been Republic numerous IPPs, and a few reversed merchant plants Rep. Egypt, Arab ­ 1999–2017 6 Handful of IPPs ­n.a. 0 ­n.a. India – Andhra 1995–2017 39 Numerous IPPs and a few ­n.a. 0 ­n.a. Pradesh merchant plants India – Odisha 1998–2010 66 One partial divestiture and 1999 0 Four instances of full divestiture all numerous IPPs have been reversed India – Rajasthan 2004–17 39 Numerous IPPs 2016 3 Partial divestiture as pilot projects in some areas Kenya 1996–2014 31 One partial divestiture and 2006 50 One partial divestiture with numerous IPPs controlling government stake Morocco 1997–2017 39 Numerous IPPs heading 1997–2001 33 Four urban concessions toward majority power generation share Pakistan 1992–2017 47 Some partial divestiture and 2005–16 16 One partial divestiture and one numerous IPPs concession Peru 1995–2017 83 Extensive partial and one full 1994–2015 50 Several partial and full divestitures divestiture, numerous IPPs, including some reversals, plus and a few merchant plants concessions for transmission Philippines 1991–2017 90 Extensive full divestiture, 1990–2009 67 Extensive partial divestiture in numerous IPPs, and a few distribution, and concession in merchant plants transmission Senegal 1997–2017 46 Several IPPs and a merchant 1999–2010 2 Reversed partial divestiture, and plant several rural concessions Tajikistan 2006–08 17 Several IPPs 2002 5 One regional concession Tanzania 1994–2011 20 Several IPPs 2002 0 ­n.a. Uganda 2003–17 63 Numerous IPPs 2003–05 90 Several concession contracts Ukraine 2002–17 14 Several partial and one full 1998–2010 74 Extensive full and partial divestiture and numerous IPPs divestiture in distribution and transmission Vietnam 1996–2017 34 Several partial divestitures and ­n.a. 0 ­n.a. numerous IPPs ­ ttps://ppi.worldbank.org/) and Rethinking Power Sector Sources: Based on World Bank—PPIAF (Private Participation in Infrastructure database 2018, h Reform utility database 2015. ­ .a. = not a Note: IPP = independent power producer; n ­ pplicable. distribution focusing on concessions and Europe and Central Asia the relative contri- ­management ­contracts. butions of domestic and foreign investors In most regions of the world, foreign spon- look to be evenly balanced ­ ( figure 5.1). sors have been an important source of private South Asia stands out as having about three-­ investment in electricity generation and quarters of private investment domestically ­distribution.3 Across Latin America and the sourced, which largely reflects the develop- Caribbean, East Asia and the Pacific, and ment of the domestic power sector and the 148 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 5.1 Private investment in electricity came predominantly from foreign sources and mostly in the generation sector Private investment in power sector by region and source a. Overall power sector 100 90 80 Investment source (%) 70 60 50 40 30 20 10 0 Middle Sub- Europe South East Asia Latin East and Saharan and Asia and America North Africa Central Pacific and the Africa Asia Caribbean b. Generation c. Distribution and transmission 200 200 180 180 160 160 Investment (US$, billions) Investment (US$, billions) 140 140 120 120 100 100 80 80 60 60 40 40 20 20 0 0 Middle Sub- Europe South Latin East Asia Sub- Middle East Asia South Europe Latin East and Saharan and Asia America and Saharan East and and Asia and America North Africa Central and the Pacific Africa North Pacific Central and the Africa Asia Caribbean Africa Asia Caribbean Domestic Mixed Foreign Unknown ­ https://ppi.worldbank.org/). Source: Based on Private Participation in Infrastructure database 2018 ( depth of local capital markets in ­ I ndia. Finding #2: Working alongside By contrast, in Sub-Saharan Africa, as well as public investment in capacity, the the Middle East and North Africa, about private sector has made substantial three-quarters of private investment in elec- contributions toward the expansion tricity comes from foreign s ­ ources. These dif- of power generation capacity in the ferences have significant implications in developing ­world terms of the political economy of privatiza- IPPs have had a big, albeit geographically tion, as well as sensitivity to foreign exchange ­ oncentrated, impact on developing country c risks. They may also affect government atti- ­ power ­ s ystems. Since 1990, IPPs have tudes toward risk allocation with ­ sponsors. attracted a cumulative total of almost What Has the Private Sector Contributed? 149 US$600 billion of investment to the develop- Overall, IPPs contributed about 42 percent of ing world, 4 leading to the construction of the expansion in generation capacity in the almost 850 gigawatts (GW) of new generation developing world during the period 1990– ­capacity.5 Over 80 percent of IPP investment 2016,6 with the balance continuing to be devel- was captured by just three developing oped as public generation p ­ rojects.7 In just regions—East Asia and the Pacific, Latin about half of the countries, the private sector America and the Caribbean, and South Asia— contributed more than half of the expansion with the top five countries (India, Brazil, over the period ­(figure 5.3). Only in about a China, Turkey, and Indonesia) accounting for quarter of the countries did the private sector nearly 60 percent of all investments in IPPs contribute more than 75 percent of new capac- (figure 5.1). ­ generation. The most salient ity additions in ­ IPP investment has been cyclical, reflecting examples were Cambodia, Georgia, Peru, and movements in the largest ­ markets. Following a the Philippines, where the private sector relatively slow start over the decade from 1992 ­ contribution to capacity additions exceeded to 2002, IPPs underwent a particularly rapid 90 ­percent. expansion during the subsequent decade The contribution of the private sector to (2002–12), peaking at close to US$70 billion in new generation capacity shows some regional 2012, and subsequently settling at a level of and income-group v ­ ariations. Investment was about US$30 billion per year ­ (figure 5.2). The highly skewed across income ­ groups. Of the steep decline witnessed in 2013 is closely new capacity funded by the private sector linked with the cessation of two of the world’s during 1990–2016, almost 80 percent went to largest IPP programs, first in India and later in upper-middle-income countries; less than 1 ­ Brazil. In India, regulatory setbacks were to percent went to low-income c ­ ountries. blame, whereas Brazil was beset with political Nevertheless, taking into account the varying and economic ­ crises. sizes of power systems across these income IPPs typically coexist with continued public groups, the contribution of the private sector to sector investment in power g ­ eneration. capacity expansions in each of these income FIGURE 5.2 Private investment in independent power producers has been substantial though subject to fluctuations and concentrated in a few countries, 1990–2017 Private investment in independent power producers, by region and top five markets a. Investment by region b. Investment in top five markets 80,000 Investments (US$, millions) 70,000 60,000 India, 22% 50,000 40,000 Rest of the 30,000 developing world, 57% Brazil, 17% 43% 20,000 China, 8% 10,000 Turkey, 5% 0 Indonesia, 5% 90 92 94 96 98 00 02 04 06 08 10 12 14 16 20 20 20 20 19 19 19 20 19 20 19 20 20 20 East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa Rest of the developing world India Brazil South Asia Sub-Saharan Africa China Turkey Indonesia ­ https://ppi.worldbank.org/). Source: Based on Private Participation in Infrastructure database 2018 ( 150 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 5.3 Most countries still depend on a combination of renewable energy sources ( ­ figure 5.4, panel c, public and private investment for the development of new and figure 5.5). About 68 percent of the total power generation capacity wind and 78 percent of the solar capacity that Countries with private projects as a percentage of total was added in the period 1990–2016 was funded capacity additions by the private sector compared to 45 percent of 40 thermal c ­ apacity. Nonconventional renewable 35 energy projects particularly lend themselves to private sector investment because of their rela- 30 tively small scale and modular design, as well as Share of countries (%) 25 their comparatively short and low-risk construc- 20 period. A second stronger driver of the share tion ­ of private sector investment in new generation 15 capacity is the country risk rating ( ­ figure 5.4, 10 panel ­d). The share of new capacity contributed by the private sector is almost negligible in coun- 5 tries that are in default, speculative, or ­ unrated. 0 Countries that are rated Ba39 or above do signifi- 0–25 25–50 50–75 75–100 Private share of capacity added, 1990–2016 (%) cantly better at capturing private ­ investment. In some countries, public investments in Source: Based on UDI World Electric Power Plants Database 2017. Note: Considers only greenfield power projects and ignores any divestiture and generation are supported by large Chinese consequent change of ­ownership. financing p ­ ackages. For example, in Pakistan, an intergovernmental financing package as groups was comparatively similar ­ (figure 5.4, part of the China–Pakistan Economic panel ­ a). Differences were somewhat more Corridor project is providing approximately pronounced by g ­ eography. Across the Middle US$35 ­ b illion to support energy projects, East and North Africa, the private sector repre- including c ­ oal-fired, hydro, and solar plants sented just over 20 percent of investment in totaling 17 GW of c ­ apacity. In East Africa, new generation ­ capacity. Interestingly, the also, several projects are being developed share of private investment in new generation with Chinese support, such as the Kinyerezi capacity in Sub-Saharan Africa was almost III and IV gas-to-power projects in Tanzania double that in the Middle East and North and the Karuma and Isimba hydropower ­ Africa. Although the absolute amount of pri- projects in ­ Uganda. vate investment in generation capacity in Sub- Saharan Africa was small, at US$30 billion, it nonetheless represents a substantial share of Finding #3: Governance challenges the region’s investment ­ needs. For other devel- emerge when power generation oping regions, the private investment share capacity is not procured in a was somewhat higher, in the 40–60 percent competitive ­fashion range ­ b).8 (figure 5.4, panel ­ Plant technology and country risk rating IPPs have not always been procured by com- appeared to be much stronger drivers behind pri- petitive means, raising concerns about trans- vate sector investment in g ­ eneration. Whereas parency and value for ­ m oney. IPP projects coal was the dominant technology for IPP invest- are fairly standard, as are the number of ments up to 2010, the balance has subsequently market ­ p articipants. As a result, IPPs lend shifted strongly toward nonconventional themselves to competitive p ­ rocurement. What Has the Private Sector Contributed? 151 FIGURE 5.4 Private investment shares reveal different drivers a. By income group b. By region 100 100 Share of capacity additions (%) Share of capacity additions (%) 80 80 60 60 40 40 20 20 0 0 Low income Upper-middle Lower-middle Upper-middle Middle Sub- East Asia South Europe Latin income income income East and Saharan and Asia and America without China North Africa Africa Pacific Central and the Asia Caribbean Private Public Private Public c. By technology d. By country risk rating 100 100 Share of capacity additions (%) Share of capacity additions (%) 80 80 60 60 40 40 20 20 0 0 Hydro Coal Gas Geo- Oil Wind Solar Biomass Unrated Risky Investment thermal Private Public IPP Public Sources: World Bank elaboration based on Moody’s sovereign ratings and UDI World Electric Power Plants Database 2017. producer. Note: IPP = independent power ­ Competition makes transactions relatively information on procurement methods is open and transparent, while competitive often missing from global databases; how- pressures help to keep costs d ­ own. Latin ever, from 2005 to 2017 direct negotiation America has embraced competitive bidding, was as prevalent as competitive bidding in and several countries—including Brazil, the developing world ­ overall. For both South Chile, and Peru—have established regular and East Asia, though, the number of public auctions for contracting new genera- reported transactions based on direct negoti- tion capacity ­ ( figure 5.6). Nevertheless, ation exceeds those reported as competitive across Africa and Asia, it remains common- ­procurements. place for IPPs to arise as unsolicited bids or The case of Tanzania illustrates the diffi- to be procured through direct ­ n egotiation. culties that can arise when countries rely on Exact figures are hard to come by because direct negotiation of IPP contracts; especially 152 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 5.5 Independent power producer investment is moving toward cleaner sources of energy, although capacity expansion lags IPP investment and capacity expansion, by technology a. IPP investment 80,000 70,000 Investment (US$, millions) 60,000 50,000 40,000 30,000 20,000 10,000 0 00 92 98 99 02 08 09 10 94 95 96 04 06 12 14 16 93 97 90 91 01 03 05 07 11 13 15 20 20 20 20 20 20 20 19 20 19 19 19 20 19 19 20 20 19 19 20 19 20 20 20 20 20 19 b. Capacity expansion 80,000 70,000 Capacity added (MW) 60,000 50,000 40,000 30,000 20,000 10,000 0 90 00 08 10 92 98 99 09 12 96 02 91 94 95 01 04 06 11 16 07 13 14 15 93 03 05 97 20 20 20 19 20 20 20 20 20 19 20 19 19 19 19 20 19 20 20 19 19 20 20 19 20 20 20 Coal Natural gas Oil Biomass Hydro Geothermal Wind Solar ­ https://ppi.worldbank.org/); panel b based on UDI Sources: Panel a based on Private Participation in Infrastructure database 2018 ( World Electric Power Plants Database 2017. Note: IPP = independent power producer; MW = m ­ egawatt. FIGURE 5.6 Competitive procurement of independent power producers varies widely between regions Independent power producer procurement, by region, 2005–15 100 90 Share of private power 80 70 plants (%) 60 50 40 30 20 10 0 East Asia and Europe and Latin America Middle East South Asia Sub-Saharan Overall Pacific Central Asia and the and North Africa Caribbean Africa Competetive IPPs Merchant plants Direct negotiation of IPPs Not available ­ https://ppi.worldbank.org/). Source: Based on Private Participation in Infrastructure database 2018 ( Note: IPP = independent power ­ producer. What Has the Private Sector Contributed? 153 when deals are struck during power debt repayment and given a backloaded ­ s hortages. In particular, the Independent payment structure, portending a substantial ­ Power Tanzania Limited (IPTL) diesel plant hike in the PPA ­ tariff. This hike was avoided negotiated in 1994–95 was exceptionally through a refinancing process that spread the costly. Negotiated just after a drought-related ­ repayments of the final 5 years over a period load shedding created a sense of emergency, of 15. The challenges associated with both the IPTL signed a power purchase agreement Bujagali I and II led the government to adopt ( P PA ) w i t h a t a r i f f o f U S $ 0 . 3 1 p e r a new policy on generation in 2010, requiring ­ k ilowatt-hour ( ­ kWh). Worse, it became the public sector to participate in all embroiled in corruption c ­ harges. Lengthy ­ g eneration projects exceeding 25 MW in court battles ensued, along with arbitration ­ capacity. Uganda awarded two large hydro proceedings between the government and projects—Karuma (600 MW) and Isimba ­ IPTL. During another drought-related crisis (183 MW)— to Chinese developers through not long afterward, in 2006, a contract for direct ­negotiation. another diesel plant was directly awarded to Richmond Development Company, an enter- Finding #4: Governments have found prise lacking any relevant power sector it challenging to strike the right ­ experience. Resulting delays meant the plant balance in allocating risk between was not completed until the power shortage the public and private sectors in was ­ o ver. Meanwhile, the investigation of contracts governing ­IPPs corruption by a select committee appointed by the parliament caused the prime minister The allocation of risk between the public and and the minister of energy and mines to private sectors is a critical issue under IPP resign. Tanzania adopted a new policy that ­ ­ contracts. The key issue in the design of IPP allowed the public utility to play a leading contracts is finding the appropriate allocation role in the development of generation proj- of risk between the government and the pri- ects through a public–private partnership vate ­investor. IPPs face a plethora of risks, ­model. including risks regarding demand, fuel price, Uganda faced repeated challenges while exchange rate, and ­ termination. Such risks implementing the Bujagali hydropower proj- can weaken investor interest, particularly in ect as an IPP, which eventually led to a policy untested markets, until a reliable track record of greater public sector involvement in power has been ­ e stablished. In response, govern- generation. In Uganda, the 1999 legal frame- ­ ments may provide contractual protections of work required all new generation to be pro- various ­kinds. Oil price and currency fluctua- vided by ­ IPPs. The first of these IPPs was the tions, for instance, may be passed through 250 megawatt (MW) Bujagali I hydropower directly in the PPA ­ t ariff. “Take-or-pay” plant in 1998, seen as critical to relieving clauses may guarantee purchase of power power shortages in the c ­ ountry. The associ- even in the absence of demand, or capacity ated direct procurement process lacked charges may at least ensure that fixed capital transparency. Allegations of corruption led to ­ costs can be c ­ overed. Sovereign guarantees the cancellation of the project in 2003. The may be provided to compensate investors in following year, the project was retendered on case of premature ­ termination. a competitive basis as Bujagali I ­ I. Nevertheless, At one end of the spectrum, IPP programs the associated PPA tariff was relatively high at have sometimes stalled when private sector US$0.10­ / kwh. As a result end-user tariffs demands for risk mitigation are not matched were h ­ iked. Bujagali tariffs were indexed to by the willingness of governments to 154 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD provide ­ them. The public sector’s retention of as well as sovereign guarantees to cover risks creates contingent liabilities for the state, offtaker payment risk and the risk of early which needs to be appropriately accounted for ­ termination. The government has been reluc- and monitored by the finance ­ ministry. Where tant to provide this level of protection, con- governments have been reluctant to retain cerned about the contingent liabilities and risk, the private sector has tended to stay ­away. their impact on the achievement of public The experiences of the Arab Republic of Egypt debt limits set as part of macroeconomic and Vietnam are ­ illustrative. ­ policy. Furthermore, without a standardized Following a financial crisis in the early international contract template that meets 2000s, Egypt redesigned its IPP program to standards of bankability, terms must be nego- shift more risk to the private sector, a move tiated on a case-by-case basis for each project, that led to a loss of investor ­ interest. Egypt’s a process that can sometimes take several original IPP program, launched in 1996, had years to ­ conclude. In contrast, Vietnam’s pro- provided incentives that ranged from tax gram for private investment in smaller-scale exemptions and full repatriation of profits to renewable energy projects does not require take-or-pay provisions and a sovereign guar- government support, owing to lower concerns antee for ­payments. The competitive procure- about risk among the domestic investors at ment process attracted a lot of interest, and whom it is ­ targeted. Nevertheless, these fac- three dollar-denominated PPAs were tors have limited the government’s capacity to signed, eliminating currency risk for the support the larger-scale projects the country developer. The 2001–03 currency devaluation, ­ ­needs. however, created significant liabilities for the At the other end of the spectrum, when offtaker, doubling its bill as the Egyptian governments have assumed excessive risk, IPP pound ­ crashed. The experience prompted a programs have occasionally triggered financial major rethinking of the IPP program, followed ­crises. Where IPP programs are large and pri- by a decision to shift more risk to the private vate investors are heavily protected against sector. The new IPP program required devel- ­ risks, governments may find themselves heav- opers to source all foreign currency from for- ily exposed to unpredictable events such as eign banks10 and all local costs to be paid in exchange-rate devaluations, oil price hikes, the local c­ urrency. More important, the new recessions leading to declining demand, or framework favored bids that had larger shares ­ermination. civil strife resulting in contract t of equity financing and more local investment Given the economic weight of the sector in (all earlier IPPs had no local ­ partners). The many developing countries, a power sector proportion of the electricity output to be cov- crisis can rapidly translate into a macrofiscal ered by the take-or-pay clause was also crisis, as happened in Pakistan and the reduced. These changes inhibited further for- ­ ­Philippines. eign investment, and the program was Protections offered to IPP investors in scrapped in 2003 as new funding streams from Pakistan left the power sector highly exposed various donors became ­ available. to an exchange-rate devaluation, combined In Vietnam, further expansion of the 2007–08 with an oil price ­ hike. The opening up of the IPP program has been affected by disagree- market to IPPs in 1994 rapidly attracted 4,500 ments over acceptable risk ­ a llocations. MW of additional power generation ­ capacity. International investors argue that they require Private investors were offered an attractive dollar convertibility of profits for repatriation, package, with a guaranteed fixed rate of return What Has the Private Sector Contributed? 155 and remuneration indexed to the ­ U.S. dollar exchange-rate risks (through pass-through and the ­ U.S. inflation rate, as well as numer- mechanisms), termination risk (through sov- ous tax exemptions and an exoneration from ereign guarantees), and fiscal risks (by offering fuel purchase responsibilities, which remained tax ­ e xemptions). Although the program with the g ­ overnment. The policy fixed PPA expanded capacity and restored supply– tariffs on a “cost plus” basis and gave develop- demand balance in the country, the East Asian ers complete freedom to choose the f ­uel. As a hard. financial crisis of 1997 hit the Philippines ­ result, most IPPs used heavy oil, which made The currency devaluation and a slowdown of Pakistan’s fuel mix one of the most expensive demand left the National Power Corporation in the ­region. Almost immediately the cost of with huge liabilities on the take-or-pay of sur- oil started rising, increasing the cost of pur- plus power as well as compensation for the chasing electricity for the utility o ­ fftaker. The currency d ­ evaluation. By 2001 the sector major currency depreciation in 1998 led to accounted for 25 percent of the national ­ debt. another unaffordable hike in PPA tariffs, lead- The government responded by renegotiating ing to nonpayment by the utility and the accu- 20 of the IPP contracts and canceling 7. mulation of US$1.6 billion of arrears to the Wholesale reform of the power sector IPPs by the end of 1998. The situation was fur- ­followed. ther exacerbated as oil prices continued to rise, doubling by 2004, and increasing the financial Finding #5: Although privatizing burden on the ­ government. Although the IPP transmission was not a focus of the program led to a rapid increase in power 1990s’ reforms, Latin America and s upply—even creating excess supply for a ­ Asia gained a lot of experience in the while—this financial crisis, triggered by default ­segment on IPP contracts, gave birth to the circular debt crisis that bedevils the sector to this d ­ ay. The The most widely used modality for PSP in crisis is described as “circular debt” because transmission has been the BOT contract, anal- customers fall behind on their power bills, and ogous to those widely used in the generation government on its subsidy payments, so distri- ­s ector. 11 Privatization efforts in the 1990s bution utilities cannot honor their power pur- focused primarily on the generation and dis- chase bills with the single buyer, who cannot tribution s ­ ectors. Because it is a public good keep up with payments to generators, who in and has a role in the coordination of electric- turn do not pay their fuel suppliers, almost all ity dispatch, the transmission grid was consid- of whom are government e ­ ntities. Hence, ered less amenable to P ­ SP. Nevertheless, a ­“circular ­debt.” number of countries experimented with this The Philippines’ IPP program initially suc- approach under a variety of modalities ceeded in averting a power supply crisis, but it (table 5.4). Of these, the most widely adopted later amplified a macroeconomic crisis caused has been the BOT contract, which is equiva- by a currency d ­ evaluation. Power shortages in lent to an IPP arrangement for t ­ ransmission. 1993 prompted the government to sign 42 Popular in Latin America, these contracts PPAs with IPPs, mostly through direct entail the payment of a fee for the availability ­ negotiation. In these contracts, the govern- of capacity covering both operating and main- ment provided generous protection to the pri- tenance costs, as well as an annuity for capital vate sector, assuming commercial risks ­ i nvestment. In other instances, the entire (through take-or-pay clauses), fuel price and national transmission grid has been privatized 156 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 5.4 Transmission: A few observatory countries have embraced private participation Country Year Modality Outcomes Colombia 2000–07 Partial divestiture 31% private shareholders; ISA, the new transmission company, owns 70% of national 2015–20 BOOT transmission system 16 projects under a 25-year BOOT system to be awarded with a total investment of US$1.5 billion India 2002–17 BOT Invested US$8.6 billion since 2002 and currently represents 15% of capacity expansion Peru 1998–2015 30-year BOOTs Invested US$2.6 billion; all new capacity since 1998 added through BOOTs 30-year concession Philippines 2009 25-year grid Invested in over 600 circuit kilometers of lines, and consistently met targets for concession availability and losses ­ https://ppi​ Sources: Based on Rethinking Power Sector Reform utility database 2018, and Private Participation in Infrastructure database 2018 ( .worldbank.org/). build, operate, transfer. Note: BOOT = build, own, operate, transfer; BOT = ­ under a concession modality, as in the expenditure and a capital annuity set at a Philippines. Somewhat rarer are permanent ­ 12 percent rate of r ­ eturn. These improved divestitures of the national grid, or merchant arrangements led to an upswing of lines developed at the sole risk of the private US$1.5 ­billion of private investment during the ­investor. period 2006–15. Overall, some 6,000 kilome- Peru is one of the countries that pioneered ters (km) of transmission lines and associated PSP in transmission using the BOT ­ mechanism. transformer stations have been developed in The transmission sector is almost entirely pri- Peru through this ­ modality. vate and operated by 13 different companies, The Philippines took another route: an out- of which the two largest are Red de Energía del right concession as part of the reform process Peru (REP) and Consorcio Transmantaro initiated through the Electric Power Industry (CTM) with 40 and 20 percent of the market, Reform Act (EPIRA) in 2001. But efforts to ­ respectively. The process began in 1992, when attract private investors for the transmission the sector was unbundled into two public network did not begin ­ well. Auctions in 2003 transmission companies, which went on to and early 2007 failed over regulatory uncer- tender 30-year build, own, operate, transfer tainty surrounding the transmission compa- (BOOT) contracts to the private sector for ny’s (TransCo) revenue ­ stream. The Energy extension of the grid, while retaining Regulatory Commission (ERC) managed to set 15 ­percent public sector s ­ takes. Once these up a performance-based regulatory frame- projects had completed the national grid, the work by 2003 but took some time to fine-tune remaining public assets were privatized under the revenue cap methodology that set rates for 30-year ­concessions. The system was strength- the transmission c ­ ompany. By December ened in 2006 with coordinated generation and 2007, however, several private companies bid transmission planning capability housed within for the TransCo, with the National Grid the system o­ perator. In addition, a clear regula- Corporation of Philippines—a group of local tory framework allows the wheeling tariff to be and international companies—putting in determined through the tendering process the winning bid for a 25-year ­ concession. The with subsequent indexation but no periodic group invested about US$4.2 billion in the regulatory ­review. The tender price reflects a sector—US$1.9 ­ billion of which was invested combination of operations and maintenance in physical assets (ESMAP 2015). What Has the Private Sector Contributed? 157 Finding #6: Among the first generation distribution utilities was prevalent among of reformers, PSP in distribution early reformers in Latin America and the was widely adopted and gained Caribbean (such as Bolivia, Brazil, Colombia, considerable ­traction Dominican Republic, El Salvador, Guatemala, Panama, and Peru) and in Europe and Central Some of the early-reforming countries Asia (Georgia, the Russian Federation, and achieved high levels of private sector penetra- ­ U kraine). From Asia, Malaysia and the tion in their power distribution ­ sectors. The Philippines were comparatively unusual in financial health and operational strength of divesting a large share of their distribution distribution utilities is a key driver of overall ­ sector. Privatization of distribution was also power sector p ­ erformance. A financially pre- undertaken in a handful of Indian states, carious distribution utility can undermine the although under a concession ­ modality. entire payment chain, and operational weak- Colombia’s macroeconomic crisis in the late nesses in the local grid can prevent power 1980s was compounded by a blackout in 1994. from reaching customers even when it is These crises led to a complete transformation ­ a vailable. For precisely these reasons, the of the power s­ ystem. Laws 142 and 143 aimed 1990s model prescribed PSP in the distribu- to increase competition and private investment tion tier as one of the first measures that grid. The and provide unrestricted access to the ­ needed to be taken to turn an ailing power thrust was on restructuring the powerful verti- sector a­ round. This recommendation is cally integrated utilities such as the ones in reflected in the surge of PSP in distribution Bogotá (Empresa de Energía de Bogotá, EEB) that took place during the 1990s, with an and Medellín (Empresas Públicas de Medellín, average of 88 transactions per year (1990–99), ­ EPM). The EEB was emblematic of all that primarily concessions (and to a lesser extent) ailed the Colombian power ­ sector. The utility’s divestitures ( ­figure 5.7). Divestiture of FIGURE 5.7 In the early years, most private sector participation in distribution was through concessions, primarily in Latin America and the Caribbean Type of private sector participation in distribution, by region 1990–99 800 700 Number of private transactions 600 500 400 300 200 100 0 East Asia and Europe and Latin America Middle East and South Asia Sub-Saharan Pacific Central Asia and the Caribbean North Africa Africa Concessions Divestiture Greenfield project Management and lease contract ­ https://ppi.worldbank.org/). Source: Based on Private Participation in Infrastructure database 2018 ( 158 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD system losses were at 27 percent (1995), and it sector could not resolve its critical issues, and owed US$900 million to Financiera Eléctrica the privatization process came to a ­halt. Nacional (FEN—a national entity set up to finance the energy s ­ ector). With the utility Finding #7: Since the early 2000s, PSP unable to reduce its losses and its financial situ- in distribution has experienced major ation worsening, the government decided to setbacks and uptake has ­ declined restructure and privatize the u ­ tility. By 1997, EEB became a holding company (81 percent Despite the strategic importance of PSP in national government, 7 percent Bogotá electricity distribution, there has been rela- government, and the rest owned by various ­ (figure 5.8). tively little interest in it since 2000 ­ companies) with a generating company Further uptake of PSP for distribution utilities (EMGESA), a transmission company (EEB), tailed off after 2000, with an average of only and a distribution company ( ­ Codensa). The 15 transactions per year (2000–17) subse- Enel group owns 48.5 percent of the distribu- quently recorded, largely confined to five tion company, whereas the rest is owned by countries (Brazil, Bulgaria, India, Russia, and Grupo Energía Bogotá (GEB—the government Turkey) that together account for 60 percent of Bogotá owns 76.28 percent of G ­ EB). With of the ­transactions. management in private hands, Codensa’s sys- Even among countries undertaking privat- tem losses fell, reaching a low of 8.9 percent in ization of power distribution utilities, few have 2006. The utility’s debt was restructured, and privatized the entire distribution ­ s ector. the company has maintained a healthy profit Cameroon, Côte d’Ivoire, and Uganda privat- margin since 2008. It is considered to be one of ized their national distribution utility; and the better-run utilities ­globally. Argentina, Chile, Jordan, Nigeria, and Turkey Ukraine began its power sector reform in privatized all of their regional distribution 1995 by restructuring the eight vertically inte- ­ companies. More typically, public and private grated utilities in the ­ country. The process cre- distribution utilities coexist within the same ated four regional generation utilities and a country, with private operators often serving single transmission ­ c ompany. Distribution capital cities or larger commercial c ­ enters. For assets, in contrast, were divided into 27 utilities example, in Pakistan, the privatization of (one for each region and one each for Kyiv and Karachi Electric in 2006 was originally ­ Sebastopol). Although a major change in the intended to be the first in a wider privatization system, it was limited to legal and accounting program for distribution utilities, but the plan separation; ownership remained with the s ­ tate. was later shelved because of opposition gener- The privatization process began in 1998, when ated by the Karachi ­ case. shares in five distribution companies were sold The decision to privatize only some distribu- by competitive tender to a group of local and tion utilities may reflect differences in the com- international companies called Investment mercial viability of the service areas or varia- Pool. This process was followed by the sale of ­ tions in the local political authorizing six more distribution companies for US$160 environment. The fact that electricity distribu- ­ million. This early phase of privatization was ­ tion remains a subnational responsibility in largely successful; however, privatization of many countries adds another layer of political utilities undertaken between 2001 and 2004 complexity to the decision to privatize was mired in allegations of ­ corruption. At the distribution. Moreover, with lucrative com- ­ same time, inadequate regulatory frameworks mercial and industrial demand often geograph- and governance structures ensured that the ically concentrated in larger cities, the financial What Has the Private Sector Contributed? 159 FIGURE 5.8 The rise and fall of private sector participation in electricity distribution, 1990–2017 300 250 Investment (US$, millions) 200 150 100 50 0 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 19 20 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 19 20 20 20 20 Concession (lease) Divestiture (full, partial) Greenfield (BOT, BOO) Management contract ­ https://ppi.worldbank.org/). Source: Based on Private Participation in Infrastructure database 2018 ( Note: BOO = build, own, operate; BOT = build, operate, transfer. viability of distribution utilities—and hence Peru completed the privatization of Edelnor their attractiveness to the private sector—may and Luz del Sur, two large distribution utilities vary significantly across ­ j urisdictions. In serving different areas of metropolitan ­ Lima. Colombia, the strong municipal presence in the The success of this process prompted moves to electricity sector led to a hybrid approach, privatize the eight state-owned provincial dis- whereby some major urban distribution utili- tribution ­utilities. In 1998, the Gloria group ties (in Bogotá and on the Caribbean coast) emerged as the successful bidder for a package were privatized, whereas others (such as in of four of the regional companies (Electro Medellín) remained in municipal h ­ ands. In Norte, Electro Norte Medio, Electro Noroeste, Morocco, private players in Casablanca, Rabat, Electro Centro), paying a total of US$145 mil- Tangier, and Tetouan signed 30-year conces- lion for a 30 percent share, or almost twice sion agreements to provide water and electric the reference price for the ­ assets. The privat- services under municipal jurisdiction in these ization process envisaged the transfer of a sec- cities, despite reluctance to unbundle or privat- ond 30 percent tranche of shares after three ize the vertically integrated state-owned ­ y ears. In 2001, however, the regulator national utility, the Office National de l’Elec- approved a new tariff based on the preprivat- tricité et de l’Eau Potable, which still distributes ization reference price for the assets, rather electricity to the rest of the ­ country. In the than on what the company actually p ­ aid. The Philippines, where privatization of urban distri- new tariff levels would not be able to cover bution utilities has been widespread, coopera- the investment made by the Gloria group, so tives continue to be the dominant model of the deal collapsed and the government took provision in rural ­ areas. over. In 2002, under a different a­ dministration, Peru provides a particularly interesting case attempts were made to privatize the utility of the difficulties encountered in attempting in the city of ­ A requipa. That plan was to extend distribution privatization beyond ­ s ubsequently reversed because of popular the capital ­ city. In 1994, the government of ­opposition. 160 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD PSP in distribution has proved susceptible to government and the private operators in the ­ reversals. Overall, 32 distribution transactions early 2000­ s. A significant number of cancella- in 15 developing countries have been sub- tions is also reported for Europe and Central jected to reversal (in the case of divestitures) or Asia—Albania, Kazakhstan, and Russia— premature termination (in the case of conces- which was also privatizing rapidly during this sions and other contractual instruments), par- period. In Sub-Saharan Africa, many of the ­ ticularly during the first decade of ­ reform. early PSPs saw contracts canceled (Comoros, Cancellations have typically taken place about Gambia, Guinea, Mali, Togo) and divestitures five years after the privatization ­ transaction. renationalized (Cabo Verde, ­ Senegal). In South Most documented cancellations were insti- Asia, the earliest Indian PSP in the state of gated by the government, often in the form of Odisha encountered difficulties and was renationalization of divested assets, although a ­ renationalized. No cancellations are reported significant number were also prompted by the in East Asia and the Pacific or in the Middle voluntary departure of private operators dissat- East and North ­ Africa. isfied with contractual c ­ onditions. About half The risk of reversal of PSP in electricity dis- of the recorded cancellations took place in tribution is particularly high in Sub-Saharan Latin America and the Caribbean region, ­ Africa. Globally, premature cancellation of con- which was privatizing very rapidly during this tracts with the private sector affects barely period. These cancellations included a cluster ­ 1 percent of IPPs, but the rate has been about of cases in Argentina following the peso crisis 3 percent for private sector arrangements in in 2003; Bolivia in the form of renationaliza- the distribution ­ sector. Although the largest tion following a change of government in absolute number of cancellations took place in 2010; and Brazil, the Dominican Republic, and Latin America (17 cases, but only 2 percent of Peru, following disputes between the total transactions in the region), the highest rate of cancellation (affecting only 7 cases but FIGURE 5.9 Private sector participation in distribution suffers representing 22 percent of transactions) from premature contract cancellation occurred in Sub-Saharan A ­ frica. Cancellation Private sector contract cancellations in distribution, rates have been particularly high for divesti- 1990–2017 tures (more than 60 percent) and management 70 contracts (more than 30 percent) ­ (figure 5.9). Share of total private transactions (%) 60 From the private sector perspective, it can be challenging to achieve the required ­ financial 50 returns in uncertain operating ­ environments. 40 The financial equation for d ­ istribution utilities is highly sensitive to the condition of network 30 assets and the level of distribution losses, fac- 20 tors that are often poorly understood and 10 poorly documented in many publicly owned ­ companies. At the time of privatization, pro- 0 spective private operators and owners conduct IPPs Total Divestment Concession Management distribution their due diligence and make their bids on the Sub-Saharan Africa Rest of the world basis of the limited information a ­ vailable. Once Source: Based on Private Participation in Infrastructure database 2018 ( ­ https://ppi. private operators take possession of companies worldbank.org/). Note: There was no case of cancellation in greenfield (BOT) distribution ­ projects. and are able to see things at first hand, the sit- BOT = build, operate, transfer; IPP = independent power ­ producer. uation may turn out to be significantly worse What Has the Private Sector Contributed? 161 than originally understood, triggering attempts In the Indian state of Odisha, the privatiza- to renegotiate contractual terms to restore tion of the distribution companies unraveled financial returns to the required ­ level. This sit- after it was discovered that targets set by the uation often leads to renegotiation of the terms regulator were based on faulty ­ data. In 1996, of privatization contracts, or in some cases, the state electricity board began the restruc- where the government may be unwilling to turing process, which by 1999 succeeded in revise initial conditions, it may lead to some privatizing four distribution c ­ ompanies. The form of reversal or cancellation, as took place bids were obtained through a competitive in the Dominican Republic and the Indian state international p ­ rocess. The state regulator of ­Odisha. (Odisha State Electricity Regulatory In the Dominican Republic, repeated oil Commission, OERC) specified T&D loss targets price hikes, combined with an unwillingness to for the privatized utilities on a baseline figure apply tariff regulation principles, made the of 39.6 percent from 1996–97. In reality, how- financial position of the privatized distributors ever, T&D losses just prior to privatization had u ntenable. A 1997 law restructured the ­ been much higher, at about 48.6 ­ percent. The v ertically integrated utility, Corporación ­ faulty baseline meant unrealistic loss-­ Dominicana de Electricidad (CDE), into three reduction targets had been set for the private distribution companies (EDENORTE, EDESUR, distributors. Matters worsened when a cyclone ­ and EDEESTE), three generation companies, later that year wreaked devastation on the and a transmission c ­ ompany. The three distri- T&D n ­ etworks. Despite lobbying from the util- bution utilities were divested in 1998, despite ities, the government refused financial sup- the fact that all of them faced severe financial port to the privatized ­ utilities. The regulator and operational conditions and reported billing agreed to change the performance targets in only 60 percent of energy p ­ urchased. Almost 2001 only after a private player dropped ­ out. as soon as the private utilities began opera- With tariff revisions disallowed between 2000 tions, oil prices escalated, necessitating a tariff and 2010, the utilities’ financial crisis hike, which the government refused to allow w orsened. In addition, the investments ­ because it violated its electoral p ­ romises. needed to stem T&D losses could not be made, Instead, the government promised to pay the and the utilities struggled to meet their t ­ argets. utilities the revenue difference between By 2015 the regulator acted unilaterally and required and actual tariffs, a promise it never canceled the remaining three licenses for ­ kept. Despite management efforts to improve ­nonperformance. commercial and operational performance, the financial situation worsened; PPAs with Finding #8: The privatization of the generators were r ­ enegotiated. In 2002, the distribution raises delicate questions government finally eliminated subsidies and for key stakeholders such as labor passed all operational costs on to the consumer, ­unions providing relief to ­ utilities. The country soon had a banking crisis on its hands, followed by a Privatization of distribution can create signif- currency ­ devaluation. The resulting escalation icant winners and losers among s ­ takeholders. of the costs of predominantly oil-based power These include the government, private generation once again led to a breakdown of i nvestors, utility employees, and utility ­ the electricity payment chain, prompting the customers. Distribution utility privatization ­ renationalization of EDENORTE and EDESUR typically focuses on operational efficiency in 2003, and of EDEESTE in 2009. and financial viability, with concomitant 162 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD emphasis on raising tariffs, improving collec- In Senegal, for example, organized union tions, ­containing distribution losses, and cut- opposition to the privatization of Senelec, the ting ­costs. Such a package of measures can renationalization. The national utility, led to its ­ yield dramatic savings, reducing the need to main trade union, SUTELEC, was always subsidize the sector and bringing tangible fis- opposed to Senelec’s divestiture—opposition cal benefits to the government while making that intensified when the government operations more profitable for ­ investors. The retrenched some 15 percent of company staff picture is different, however, for customers through a voluntary severance p ­ ackage. In and labor ­ u nions. For them, this kind of response, and against a backdrop of heavy load reform agenda does not necessarily generate shedding in the country, SUTELEC organized a tangible benefits and may, in fact, inflict strike. The three-day-long countrywide black- ­ sizable c ­ ­ osts. Stakeholder opposition in these out led to the arrests and imprisonment of cases becomes likely and can disrupt, if not union ­ leaders. By way of compromise, the gov- overturn, ­privatization. ernment altered its plans to sell a majority From the labor union perspective, the threat stake in Senelec, limiting privatization to a of layoffs is r­ eal. Overstaffing is common in dis- minority stake of 26 ­ percent. The transaction tribution utilities in developing ­ countries. In a was completed in 1999, and a consortium study of 24 utilities in the Middle East and comprising Hydro-Quebec International North Africa region, the median value for resi- (Canada) and Elyo (France) won the 25-year dential connections per full-time equivalent concession for Senelec; but the utility’s infra- employee was found to be half that of the util- structure was much worse than determined ities outside the region (Camos and others during the bidding process and clearly required 2018); 23 of the 24 utilities reported overstaff- major ­investment. The consortium submitted a ing, and another study found that 35 out of request for a tariff hike to the regulator, argu- 39 countries in Sub-Saharan Africa had issues ing that the agreed tariff level would no longer of overstaffing (Kojima and Trimble 2016). provide the anticipated rate of r ­ eturn. The Nevertheless, both studies found that the costs request was not well received, given the lack of associated with overstaffing are typically small a clear investment plan or output targets that compared with other types of i ­nefficiencies. might hold the concessionaire accountable for For example, in Sub-Saharan Africa, overstaff- improving the dire power supply situation in ing costs represented only 10 percent of the the ­country. To make matters worse, disputes hidden costs of inefficiency (or “quasi-fiscal arose within the company over the division of deficit”) in the sector, compared with 40 per- responsibilities between foreign management cent attributable to underpricing, 30 percent to and local ­ staff. The two foreign partners then T&D losses, and 20 percent to undercollection clashed over the financing of i ­nvestments. of ­revenues. Similarly, in the Middle East and Tensions with SUTELEC ­ continued. Over the North Africa region, overstaffing accounted for course of 18 months, as the 2000 elections only 5 percent of the hidden costs of ineffi- approached, privatization became drawn into ciency in the sector, with underpricing electoral ­politics. Senelec was renationalized accounting for at least 72 percent of ineffi- and the concession contract canceled following ciency c ­ osts. Nevertheless, for a private opera- the change of ­ administration. tor, staff layoffs cut costs quickly and without Despite such tensions, interactions with requiring any major investment, so labor ten- labor unions can be ­ constructive. Governments sions are common during utility ­ privatizations. can make political choices about labor policy If they are not adequately addressed, they can up front and build these into the terms of a derail ­privatization. ­ p rivatization. Clear communication with What Has the Private Sector Contributed? 163 unions throughout the process, carefully man- in management contracts for African power aged layoffs through attrition, and voluntary ­utilities. Management contracts, with dura- severance programs can all contribute to posi- tions of just two to three years, have been tive ­outcomes. introduced in some 17 African countries For example, in the Indian state of Odisha, (table 5.5). They appear to offer a practical union opposition to privatization was sur- approach to boosting managerial capabilities mounted by binding the private investor to a with a view to reaping quick wins in opera- program of gradual staff ­reductions. At the out- tional performance while building local set, in 1996, the unions had been vehemently capacity. The experience has been checkered, ­ opposed to ­ privatization. In response, the state ­ however. Difficulties may arise even in attract- government built employment guarantees into ing contractors with requisite skills, and there- the privatization contract—terms that assuaged after in retaining qualified managers labor and allowed the transaction to ­ proceed. ­ in-country. In Rwanda, for example, the sole A careful government “staff rationalization” bidder for the management contract tendered exercise identified an excess of 2,800 s ­ taff. in the early 2000s was a company with no Nevertheless, immediate reductions were operational experience with distribution rightly seen as both costly and ­ d isruptive. ­ utilities. Even when good managers can be Instead, private bidders agreed to address over- retained, cultural differences and resentment staffing over a five-year period in line with the over wage differentials can lead to ­ friction. In government’s transition plan, which empha- addition, the private operator may lack the sized attrition, retraining, and ­ redeployment. motivation to transfer skills to a local leader- The surplus staff did not affect the sale price, ship cadre—skills transfers being a key objec- and the deal went through without much tive of a management c ­ ontract. Most critically, disruption from the unions (Ray 2001). ­ in many cases, management contracts lack In Tanzania, labor unions were mollified ­ performance-improvement plans (with targets with clear, proactive communication and a and incentives), making it difficult to channel well-designed severance ­ package. in 2002, the the efforts of contractors whose remuneration private contractor negotiated a new manage- fee. is a fixed service ­ ment contract by engaging with staff and Kenya’s management contract, in force designing a severance package that allowed for from 2006 to 2008, illustrates both the ten- amicable retrenchments of 20 percent of the sions and the benefits contracts can ­ bring. staff. The government-approved labor agree- ­ A Canadian company, Manitoba Hydro ment led to more than 1,200 workers leaving International, won the management contract amicably—about 21 percent of the w ­ orkforce. for the Kenya Power and Lighting Company Tanzania Electric Supply Company Limited (KPLC) in 2006. The reform champions in gov- (TANESCO) financed the voluntary retire- ernment saw a management contract as a way ments from revenues totaling US$21 million to shield KPLC from political interference— (Ghanadan and Eberhard 2007). The managers interference that had been hindering perfor- continued to focus on maintaining good rela- mance improvements even after structural tions with the ­ union. As a result, the union reforms were introduced in the late 1990­ s. was in favor of an extension to the manage- Designed to promote greater managerial ment contract in 2004, and again in 2006, autonomy, the reforms encouraged KPLC staff although the second extension did not take to use the time gained under the management place for other ­reasons. contract to learn, develop, and empower them- The introduction of expatriate managers selves to run KPLC after the contract ­ expired. often spells friction with local staff, particularly Although some skills were transferred to the 164 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 5.5 Management contracts for power utilities in Africa: A difficult history Country/ State Year Renewed Deepened Issues Guinea-Bissau 1991–97 No No — São Tomé and 1993–96 No No — Príncipe Ghana 1994–97 No No — Malia 1995–98 No Yes Differences between board and contractor over ­ ontract. priorities; blackouts in third year of c Namibia 1996–2002 No No National utility wanted to take over distribution ­business. Togo 1997–2000 No No ­ nderdelivery. Contractor won court case regarding u Chad 2000 No (Canceled) No — Malawi 2001–03 No No — Tanzania 2002–06 Yes (once) No period. Political opposition coincided with drought ­ Lesotho 2002–Present Yes No Outright privatization attempts failed and led to a management contract, which led to a t ­ urnaround. Namibia 2002–present Yes No After the previous contract (1996–2002) a new contract was given to new joint venture between the governments. national utility, local, and regional ­ Rwanda 2003 No No Contractor had no prior experience in electricity ­distribution. Madagascar 2005–07 No No The management contractor, mired in corruption blacklisted. scandal elsewhere in Africa, was ­ Kenya 2006–08 No Yes Recurring tensions arose between foreign and local management. Gambia, The 2006–11 No No — Liberia 2010–16 Yes No — Guineaa 2015–19 No No — Sources: Based on Rethinking Power Sector Reform utility database 2015, and Private Participation in Infrastructure (PPI) database 2018 ­(https://ppi.worldbank.org/). Note: — = not ­ available. a. Not in PPI ­ database. local cadre, tensions nevertheless mounted typically fall well below the cost of service in between locals and the foreign leadership many developing countries, tariff hikes help team. Some staff felt that management was ­ utilities recover costs so that private operators overcompensated and disrespectful to local can run the business from r ­ evenues. To the ­staff. extent that investment finance also switches from public and often concessional sources to private sources on commercial terms, the cost Finding #9: Customers are sensitive of capital may also ­ increase. The result? More to the tariff increases that come with upward pressure on ­ tariffs. At the same time, privatization and will have legitimate customers who pilfered electricity or failed to expectations for improved and pay their bills will no longer be able to freeride expanded ­services after a private operator rolls out its revenue From the customer perspective, distribution protection ­ measures. All this adds up to more privatization generally entails higher service expensive electricity for customers, which is payments, with only meager prospects for ser- hard to justify unless the service ­ improves. vice i ­mprovements. Because baseline tariffs Distribution privatization contracts too often What Has the Private Sector Contributed? 165 overlook the quality-of-service dimension by the concession made the regulator hesitant to failing to include and enforce reliability allow more access-related investment by the standards. In some instances, unreliable ser- ­ utility, because it would have implications for vice may have more to do with inadequate tariffs, already among the highest in ­ Africa. generation capacity than with deficiencies in Nonetheless, following the inclusion of access the distribution network; in such cases, targets in 2012, electrification climbed from improvements are beyond the immediate 16 percent in 2012 to 27 percent in 2016 (IEA control of the distribution ­ operator. and others 2018). Access improvements came The privatization of the national electric at a cost to consumers, who have seen cumula- utility in Uganda brought immediate tariff tive tariff hikes of more than 300 percent since i ncreases. Meanwhile, output goals were ­ the start of the reform ­ process. At US$0.17/ ­ deferred. In 2005, the national distribution kWh, Uganda’s average tariff is among the utility was concessioned to a private operator, highest on the ­ continent. UMEME. As the sole bidder for the concession, ­ In Karachi, Pakistan, the privatization of the Eskom-Globeleq consortium was able to K-Electric caused a hike in tariffs that, negotiate comprehensive contractual protec- together with unreliable supply, led to tions as well as a guaranteed fixed rate of ­ consumer ­ disaffection. K-Electric is a large, return on investment of 20 p ­ ercent. The pri- vertically integrated utility that serves nearly vatization took place against the backdrop of one-third of the c ­ ountry. It was divested in serious load ­shedding. A drought had curtailed 2006 under martial law provisions that the supply of hydropower, and the govern- circumvented normal p ­ ­ rocesses. Viewed as ment contracted costly emergency power illegitimate, its privatization was quickly chal- plants to bridge the g­ ap. A renegotiation of the lenged in the c ­ ourts. After company owner- contract took place in 2006, with a relaxation ship changed hands in 2009, the new owners of regulatory performance targets and hikes in improved revenue collection and cut down on two retail tariffs (41 percent and 35 ­ percent, ­ theft. Power sales increased 21 percent and respectively). Tariff hikes and unreliable supply 18 percent in the first two ­ years. Reduction in led to public ­discontent. In time, opposition to nontechnical losses, usually associated with the concession produced two public inquiries, theft, brought losses down to 32 percent in in 2009 and 2011, each calling for the cancella- 2011 from a high of 36 percent in 2009; tion of the ­ concession. Only in 2012, with the but staff who tried collecting on outstanding commissioning of new generation capacity, did bills faced violence in several Karachi the situation start to ­ improve. neighborhoods. Meanwhile, the average tar- ­ The regulator negotiated a new tighter suite iffs increased by almost 90 percent in the years of key performance indicators (KPIs) for the 2009 to 2011, without improvements in sup- distribution concessionaire, which translated ply r ­ eliability. Massive public protests ensued into substantial improvements in operational when the utility began disconnections for efficiency and service ­ coverage. By 2016, reve- nonpayment of outstanding ­ bills. More than a nue collection rates had climbed to 100 per- decade later, the privatization of K-Electric cent, and distribution losses had been halved to remains a prominent political ­ issue. 17 ­percent. UMEME is now ranked as one of Special efforts are needed to ensure that pri- the better-performing utilities in Sub-Saharan vate operators invest in the distribution ­ Africa. Progress in increasing access was under- ­network. Beyond first-order efficiency gains, whelming, however, for two r ­ easons. First, service improvements require major invest- access had only recently been made a K ­ PI. ments in network i ­nfrastructure. Unless other- Second, the high guaranteed rate of return for wise incentivized or obligated, private 166 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD operators will make simple revenue enhance- received 99 percent of its success fees from rev- ment and cost reduction their ­ priorities. These enue improvements and paid only small fines goals involve modest investments and yield for failing to improve reliability, power losses, prompt ­ paybacks. Hefty investments in long- and ­electrification. lived distribution networks to expand electrifi- cation or improve service quality represent a CONCLUSION big risk to the private ­ operator. Such invest- Working alongside public sector investment, ment generally requires explicit contractual the private sector has become a major contrib- terms and reliability targets reinforced by regu- utor to the expansion of power generation latory entities and tariff-based ­remuneration. capacity in the developing ­ world. The adop- Expansion of access to new service areas is a tion of IPPs is one legacy of the standard thorny issue because the incremental costs of reform ­ model. IPPs have contributed greatly expansion tend to outstrip the incremental tar- to power generation capacity since 1990 and iff ­revenues. are linked to the spread of renewable energy TANESCO’s experience with management ­ t echnologies. Nevertheless, private invest- contracts between 2002 and 2006 illustrates ment remains far from the norm—and far the challenges of holding private operators from providing the most funds—in the financ- accountable for output ­ targets. In 2002 the ing of new generation capacity in developing NETGroup signed a management contract for ­ countries. Governance deficiencies made IPPs ­ TANESCO. The contract identified five KPIs— vulnerable to corruption through directly increasing revenue, reducing power losses, negotiated d ­ eals. The resulting scandals have improving customer care, boosting reliability, sometimes lent them a bad ­ name. and expanding electrification—all designed to Governments around the world struggle to turn the utility a­ round. The new management balance investor protections with risk grew revenues straightaway by upping the col- ­transfers. lection rate from a lowly 69 percent in 2001 to The private sector has played an important 89 percent in 2002. They accomplished this role in the power distribution sector for some growth primarily by obliging public institutions large middle-income ­ countries. Its contribu- to pay their ­bills. TANESCO’s revenue grew by tions elsewhere have proved more ­ difficult. An 140 percent over the four-year period early wave of distribution privatizations, pri- (Ghanadan and Eberhard 2007). The other marily in Latin America and Eastern Europe, KPIs—power losses, customer care, reliability, and particularly in larger urban centers, and electrification—were not ­ m et. Power brought sustainable ­ outcomes. Interest in the losses rose to 30 percent and then returned to approach dropped off steeply after the early precontract levels (22 percent) by 2006. Forced 2000s, and about 30 reversals of privatization outages averaged 1,500 incidents and 2,500 have ­occurred. These reversals amounted to hours per month with no ­ improvement. New 3 percent of transactions globally but as much management met only half of the electrifica- as 30 percent in Sub-Saharan A ­ frica. In only tion target (Ghanadan and Eberhard 2007). one-third of developing countries does one find While there were attenuating circumstances, PSP in power ­ distribution. A singular challenge such as drought and high generation costs, the has been to design privatization transactions so contractual incentives to reduce distribution that they balance the benefits across key stake- losses and increase service reliability were very holders, including consumers and labor u ­ nions. weak, amounting to small fines that did not New private entrants in the power sector match the magnitude of the associated ­ efforts. should be able to compete alongside the Despite these deficiencies, the contractor incumbent utility, whether public or ­ private. What Has the Private Sector Contributed? 167 A . PRIVATE SECTOR PARTICIPATION INDEX ANNEX 5­ Specific to the 15 observatory countries Private sector participation index Average function scores PSP in generation PSP in transmission PSP in distribution Multiply by private Multiply by private Multiply by private sector share sector share sector share Average of divestiture and other PSP weighted by number of instances of each Divestiture score Other PSP score Sum (instance of divestiture Sum (instance of PSP × type of × type) and divide by # of contract) and divide by # of divestitures PSP instances Type of divestiture Score Type of contract Score Partial 0.50 Rentals 0.00 Full 1.00 Service, management, 0.33 lease, PPAs for IPPs Concession 0.66 BOT, merchant 1.00 Source: Original figure for this publication. Note: BOT = build, operate, transfer; IPP = independent power producer; PPA = power purchase agreement; PSP = private sector ­participation. 168 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD B. PRIVATE SECTOR PARTICIPATION INDEX, 2015 ANNEX 5­ Percent Private sector Country/State participation (PSP) PSP in generation PSP in distribution PSP in transmission Colombia 49 63 35 50 Dominican Republic 19 52 5 0 Rep. Egypt, Arab ­ 2 7 0 0 India – Andhra Pradesh 13 39 0 0 India – Odisha 21 63 0 0 India – Rajasthan 14 39 1 2 Kenya 16 25 25 0 Morocco 25 53 22 0 Pakistan 19 40 9 8 Peru 61 78 19 88 Philippines 62 84 39 66 Senegal 11 34 1 0 Tajikistan 6 15 3 0 Tanzania 3 10 0 0 Uganda 38 53 63 0 Ukraine 30 13 52 25 Vietnam 10 31 0 0 International benchmark 24 41 16 14 Source: Based on Rethinking Power Sector Reform utility database 2015. NOTES 4. Calculated using the PPI database, includes 1. This chapter is based on original research con- projects that have achieved financial closure. ­ ducted by Vivien Foster, Anshul Rana, Joeri 5. Based on UDI World Electric Power Plants de Wit, and Victor Loksha, supported by an Database 2017. advisory team comprising Pedro Antmann, 6. Based on UDI World Electric Power Plants Pedro Sanchez, Elvira Morella, and Mariano Database 2017. ­Salto. 7. This analysis is limited to capacity addition 2. The Power Sector Reform Index (PSRI) during 1990–2016, focusing only on the plant explained in chapter 2 is used to calculate the ownership at the time of c­ ommissioning. The extent of PSP ­ globally. A more refined index figures cited do not consider existing plants is used to calculate PSP in the rethinking that became private owing to divestiture and observatory ­ countries. This index is detailed without adding to overall ­ capacity. Nor do in annex 5­ A. Scores for the 15 countries from they take into account the change in owner- the observatory can be found in annex 5­ B. ship that may occur from divestiture after the 3. Data on the geographical origin of pri- built. plant is ­ vate investment in the power sector are 8. A somewhat different picture emerges if the incomplete, because project records show analysis is repeated to look at the composi- the nationality of project sponsors but not tion of existing plants in 2016. In regions the magnitude of their c ­ ontribution. On this with notable divestitures of generation plant, basis, it is possible to classify as foreign proj- the private ownership shares for genera- ects where all sponsors were foreign and as tion are much higher than the i ­ncremental domestic projects where all sponsors were capacity additions: 62 percent for South Asia, ­ domestic, with the remainder falling into the 85 ­percent for Europe and Central Asia, mixed ­category. and 87 percent for Latin America and the What Has the Private Sector Contributed? 169 C ­ aribbean. By contrast, in regions marked ESMAP (Energy Sector Management Assistance by few divestitures of generation plant, the Program). 2015. “Private Sector Participation private ownership shares are much lower in Electricity Transmission and Distribution: than the incremental capacity additions: Experiences from Brazil, Peru, the Philippines 11 ­percent for the Middle East and North and Turkey.” ESMAP Knowledge Series Africa, 15 ­percent for East Asia and the Pacific, 023/15, World Bank, Washington, DC. and 22 percent for Sub-Saharan ­ Africa. Ghanadan, R., and A. Eberhard. 2007. “Electricity 9. Under the Moody’s rating scale, “Ba” signi- Utility Management Contracts in Africa: fies obligations judged to have speculative Lessons and Experiences from the TANESCO- elements and that are subject to substantial NET Group Solutions Management Contract credit risk. The number 3 indicates a ranking in Tanzania, 2002–2006.” MIR Working in the lower end of that category. Paper, Management Program in Infrastructure 10. This seems to be a reaction to the immediate Reform and Regulation, University of Cape aftermath of the financial crisis in 2001–03, Town, South Africa. when the country had a shortfall of ­ dollars. IEA (International Energy Agency), International 11. This section on finding #5 draws heavily on Renewable Energy Agency, United Nations ESMAP (2015). Statistics Division, World Bank Group, and World Health Organization. 2018. Tracking REFERENCES SDG7: The Energy Progress Report. Washington, Bacon, R. 1995. “Privatization and Reform in the DC: World Bank Group. Global Electricity Supply Industry.” Annual Kojima, M., and C. Trimble. 2016. “Making Review of Energy and the Environment 20 (1): Power Affordable for Africa and Viable for Its 119–43. Utilities.” World Bank, Washington, DC. ———. 1999. “A Scorecard for Energy Reform in Ray, P. 2001. “HR Issues in Private Participation in Developing Countries.” Viewpoint, Note No. Infrastructure: A Case Study of Orissa Power 175, World Bank, Washington, DC. Reforms.” World Bank, Washington, DC. ———. 2018. “Taking Stock of the Impact Sen, A. 2014. “Divergent Paths to a Common of Power Utility Reform in Developing Goal? An Overview of Challenges to Electricity Countries.” Policy Research Working Paper Sector Reform in Developing versus Developed 8460, World Bank, Washington, DC. Countries.” Paper EL 10, Oxford Institute for Camos, D., R. Bacon, A. Estache, and Energy Studies, Oxford, UK. M. M. Hamid. 2018. Shedding Light on World Bank. 1993. The World Bank’s Role in Electricity Utilities in the Middle East and North the Electric Power Sector: Policies for Effective Africa: Insights from a Performance Diagnostic. Institutional, Regulatory, and Financial Reform. Washington, DC: World Bank. Washington, DC: World Bank. Did Countries Establish Meaningful Power Sector Regulation? 6 Guiding questions • To what extent were developing countries able to introduce regulatory regimes? • What new challenges does technological disruption pose for regulators? • Did the regulatory regimes operate the way they were originally designed? Summary • Creating a sector regulator was one of the most common power sector reforms, owing to the ­ relative ease of implementation. However, there is a big difference between establishing a good regulatory framework on paper and building an effective regulatory system in practice. Quality of service ­ regulation—in particular—too often exists on paper alone. • Although tariff setting is the central function of regulatory agencies, their tariff recommendations are not necessarily respected or applied. Even in such cases, regulators can still play a valuable role in defining the magnitude of subsidies needed to ensure that the sector meets its revenue requirements. • Although many countries have embraced incentive regulation, tariff regimes have often been poorly designed. • While incentive regulation was originally designed with private utilities in mind—due to limited appetite for privatization—many regulators primarily oversee state-owned utilities. • Regulators play a role in licensing market entry but are not always involved in the critical area of power purchase agreements. • Looking ahead, price regulation and tariff structures will need to be significantly overhauled to pro- vide adequate incentives for adoption of emerging technologies. 171 172 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION the rules after irreversible investments had This chapter evaluates the quality of regula- been made. Regulation was intended to bal- tory regimes adopted as part of power sector ance the competing interests of investors, reforms in the developing world.1 It considers consumers, and governments, ensuring that the extent to which these regimes were able investors achieved reliable and fair returns, to meet the good practice standards for gover- consumers received value for money, and nance articulated in the sector reform litera- governments were constrained in their ture. Considering both the design and quality a bility to exploit the sector for political ­ of the regulatory systems put in place, the purposes. chapter probes the feasibility of introducing Practice in the United Kingdom and the regulatory regimes in the power sectors of United States envisioned autonomous regula- developing countries. It explores the modifi- tory entities that are largely separate from the cations introduced during implementation line ministry. A variant of the approach, origi- and asks whether the regimes operated as nating in the French-speaking world, held that designed. regulation could be largely conducted through The power sector reform paradigm of the a legal contract (such as a concession or a lease) 1990s emphasized the separation of powers. between the policy maker and the service pro- Policy making (setting strategic direction), vider. Contracts would be enforced by the service provision (implementing strategic courts, without the need for a specialized regu- direction), and regulation (monitoring and latory institution. In practice, “regulation by overseeing implementation) would be man- agency” proved to be more widespread than aged by separate entities. The previous insti- “regulation by contract”; however, in jurisdic- tutional model for power utilities featured tions that adopted concessions or leases for vertically integrated and state-owned monop- power distribution, the associated contract was olies, where all three functions resided in a undoubtedly an important instrument of single, reliable entity acting in the public regulation. interest. With the separation of powers, policy Regulatory reforms in the power sector making would remain with the line ministry. have focused on economics rather than on An autonomous entity responsible for regula- health and safety, or reforms of technical, tion would help to corporatize service provi- financial, or environmental matters. This focus sion. A privatized and independent company is understood as “the combination of institu- would be guided by a profit motive or take on tions, laws, and processes that, taken together, a commercial orientation (chapter 4). enable a government to exercise formal and Regulation was believed to be particularly informal control over the operating and important in the context of intended private investment decisions of enterprises that supply sector participation in electricity distribu- infrastructure services” (Brown and others tion. The replacement of a public monopoly 2006,5). For analytical purposes, regulation with a private one raised consumer concerns has two dimensions: governance and sub- about the potential abuse of market power stance (Brown and others 2006, 5). either through overpricing or through ero- Regulatory governance is defined by the laws, sion of quality. At the same time, investors processes, and procedures that determine the worried that making large sunk investments enterprises, actions, and parameters that are with lengthy payback periods in a politically regulated; the government entities that make charged sector would leave them vulnerable the regulatory decisions; and the resources and to the risk that governments would rewrite information that are available to them Did Countries Establish Meaningful Power Sector Regulation? 173 (Brown and others 2006). An effective and In addition, regulators set minimum service sustainable system of regulatory governance standards to ensure that consumers receive needs to have adequate quality of service. A third aspect of regulatory substance is entry, meaning control • Credibility, so investors are confident that over whether and how new operators may the system will honor its commitments; enter the regulated market (Rodriguez Pardina • Legitimacy, so consumers trust that the and Schiro 2018). system will protect them from the exercise Many of the principles of tariff and quality of monopoly power whether through high regulation are premised on the commercial ori- prices, poor service, or both; and entation of the power utility. The creation of a • Transparency, so investors and consumers regulatory entity was intended to be just one can see how and why decisions are made. element in a broader power sector reform Above all, however, regulatory governance package that emphasized utility governance must balance autonomy with accountability reforms, private sector participation, and mar- (Rodriguez Pardina and Schiro 2018). With- ket liberalization. These wider reforms were out autonomy, regulators cannot influence intended to increase private sector participa- outcomes. Full autonomy arises from the tion in power utilities, or at least to induce in exercise of a range of powers, including the them a much stronger commercial orientation. freedom to appoint qualified and independent For this reason, much of the substance of regu- decision makers, to take and enforce decisions lation is oriented toward the creation of finan- ­ affecting stakeholders, to capture adequate cial incentives to invest, drive down operating budgetary resources, and to manage internal expenditures, and meet quality standards. In processes. Without accountability, however, ­ each case, the utility’s supposed profit motive is regulators would have no obligation to the harnessed by the regulator to bring about stakeholders they are meant to serve and pro- socially desirable behavior. In the absence of tect. Accountability is secured through gov- these complementary reforms, the rationale for ernment oversight, the right of companies incentive-based regulation methods tends to to legal appeal, and transparent regulatory break down. ­ p rocesses and decisions. Autonomy and These issues are addressed with reference to accountability are meant to work together. It quantitative and qualitative evidence on the is pointless to have a regulator that is highly design and practice of regulatory systems in the accountable but lacking in autonomy. By the 15-country Power Sector Reform Observatory same token, a highly autonomous regulator (introduced in chapter 1). Quantitative evi- that lacks accountability may abuse its power. dence is based on a survey of the categorical Regulatory substance refers to the content of data regarding the regulatory system. Using the regulation—the explicit or implicit decisions conceptual framework outlined earlier, a made by regulators, along with their rationale Regulatory Performance Index was created. It (Brown and others 2006). Substance includes scores countries according to their conformity the setting of tariffs. This complex process with good regulatory practice (box 6.1 and begins with applying a tariff regime (or meth- f igure 6.1). To capture any divergences ­ odology) to determine the average tariff level between regulatory design and regulatory commensurate with efficient cost recovery. practice, two versions of the same index are The process may also seek to determine the calculated for each country. A first score por- tariff structure—the suite of charges customers trays the country’s de jure regulatory frame- pay—which then defines the average tariff. work as it appears in laws, regulations, and 174 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 6.1 Introducing the Regulatory Performance Index The survey instrument applied to the 15 Power Sector Reform Observatory countries included 355 categorical and quantitative questions on the regulatory system. The questions were both descriptive and normative. Descriptive questions collected facts. How many regulatory commissioners are there? How frequently are tariffs revised? They are useful in understanding the nature of practice across the developing world and are summarized where appropriate throughout this chapter. Normative questions aimed to capture regulatory best practices using the literature. To synthesize the normative data in a convenient and intelligible format, a Regulatory Performance Index was created. This index builds on earlier regulatory indexes (Pargal and Banerjee 2014; Vagliasindi and Nellis 2010), but encompasses as many as 74 features of the regulatory framework. The structure of the Regulatory Performance Index is depicted in the flow chart (figure 6.1 in the main text) and closely follows a conceptual framework derived from the literature (Brown and others 2006; Rodriguez Pardina and Schiro 2018). The overall score is the product of two subindexes of regulatory governance and regulatory substance. All indexes and subindexes are scored on an interval of 0 to 100. The subindex on regulatory governance is the product of two further subindexes capturing the degree of auton- omy (based on 17 questions) and accountability (28 questions). Examples of good practices for autonomy include actual powers: if a regulator’s decisions are legally binding; if the regulator can determine the agency’s organizational structure and the use of its budget, and so on. Examples of good practices for accountability include the regulator’s duty to report to another institution, to publicize its decisions and recommendations, and to involve nongovernment stakeholders in its decision-making processes. The subindex on regulatory substance is the simple average of three subindexes capturing performance on tariff regulation (16 questions), quality regulation (12 questions), and entry regulation (7 questions). Some examples of good practices for tariff regulation are that tariff setting is based on a specified regulatory framework; that a written, publicly available formula prescribes how end-user tariff levels are to be set; that the regulator and utility are obliged to adhere to that formula; and that regulatory accounting guidelines are followed. Examples of good practices in regulation of service quality include mandatory and publicly available standards of quality and the existence of fines or penalties for noncompliance. Examples of good practices for entry regulation include the power to monitor compliance with the terms of licenses or permits and the authority to impose penalties for noncompliance. Multiplication of subindexes, rather than averaging, is used where paired variables (autonomy and accountability, governance and substance) are meaningful only when they go together. Averaging of subindexes is adopted as the aggregation method when these are largely independent from each other (tariff and entry regulation). Two versions of the same index were calculated for each country. First, a de jure index derives from the country’s regulatory framework as captured on paper in laws, regulations, and administrative procedures. Second, a perception index determines whether the paper provisions are applied in practice. The local consultant in each country provided the perception index; his or her professional opinion was informed by some 20 interviews with key stakeholders in the reform process. The perception index was also reviewed by the World Bank country energy team knowledgeable about local context. Despite best efforts, this second index is more subjective than the first. administrative processes. A second score Finding #1: Sector regulation was a assesses a perceived score—the country’s regu- popular measure in the package of latory framework as it is practiced according to power sector reforms, but the limited local experts. This information draws on quali- appetite for privatization meant that tative evidence from interviews with some 20 regulators were often overseeing key stakeholders in each country. It forms a state-owned utilities rich source of narratives that shed further light The creation of a regulatory entity was a key on regulatory experience. component of the 1990s power sector reform paradigm. By 2015, more than 70 percent of KEY FINDINGS developing countries had created a power sec- The key findings of the analysis of power sec- tor regulator, making this the most widely tor regulation are based on experiences in the adopted measure among all the reforms. 15 observatory countries. They are summa- Unlike the other reforms, momentum for the rized in the following key findings. regulatory measure remained strong through Did Countries Establish Meaningful Power Sector Regulation? 175 FIGURE 6.1 Overview of Regulatory Performance Index Regulatory Performance Index Regulatory governance Regulatory substance Multiply scores Average scores Multiply scores Accountability Tari Quality Market entry Autonomy regulation regulation regulation Average scores Average scores Average scores Average scores Average scores 1. Regulates end-user 1. Formally states 1. Specifies objectives of 1. Promulgates formal 1. Monitors compliance tari s, quality of supply regulatory objectives tari -setting quality standards with licenses 2. Makes regulatory 2. Requires reporting 2. Establishes tari -setting 2. Legally requires 2. Imposes penalties for decisions legally binding on regulatory principles standards to be met license violations 3. Establishes source of activities 3. Gives regulator 3. Provides specification 3. Removes license in regulatory funding 3. Requires decisions authority to set tari s of quality data cases of noncompliance by law to be made public 4. Defines cost recovery 4. Requires regular 4. Approves PPAs or 4. Allows determination 4. Requires third-party 5. Specifies tari -setting reporting of quality other sales contracts of own budget evaluation of methodology data 5. Follows specific 5. Fixes term and regulator 6. Establishes tari review 5. Requires automated timeline for PPA removal process of 5. Incorporates legal frequency information systems approval commissioners appeals process 7. Excludes illegitimate 6. Conducts validation 6. Holds authority over 6. Prohibits conflicts of 6. Requires annual costs from tari -setting of quality data IPP procurement interest for reports to be 8. Requires tari formula 7. Publishes quality process commissioners published to be public data 7. Allows determination 7. Requires regulatory 9. Requires use of 8. Publishes compliance of own internal decisions to be accounting standards with quality standards structure published 10. Avoids pass-through 9. Applies fines for 8. Obliges regulator 8. Requires stakeholder of ine ciency violation of standards to hire su cient number participation 11. Provides incentives for 10. Applies positive of technical sta renewable energy financial incentives 9. Exempts regulator 12. Specifies treatment to meet standards from public sector of stranded assets pay scale 10. Exempts regulator from public employment rules Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Detailed scores for the rethinking observatory countries are provided in annexes 6A and 6B. IPP = independent power producer; PPA = power ­purchase agreement. the period 1995–2015; every year about In contrast, the diffusion of regulators in 3 percent of countries established a regulator. Europe and Central Asia as well as Middle East In about one-third of developing countries, and North Africa plateaued in the early 2000s; the introduction of a regulatory entity was the the penetration level is about 40 percent. first reform to be implemented. For some it Regulators may undergo governance was the only reform. changes over time—for good or ill. For exam- Latin America initiated the trend in favor of ple, in Kenya, regulation was initially entrusted power sector regulators. Adoption there was to the Electricity Regulatory Board (ERB), rapid—by 1998 nearly 80 percent of countries which started operations in 1998. Almost had established an entity (figure 6.2). In other 10 years later, the 2006 Energy Act led to fur- developing regions, the process lagged. ther restructuring in the sector and the ERB Nevertheless, by 2015, Sub-Saharan Africa and was transformed into the Energy Regulatory much of Asia had caught up with Latin Commission (ERC); it regulates the entire America. These regions can claim regulatory energy sector, including renewable energy and agencies in 60 to 80 percent of their countries. petroleum downstream activities. In Peru, the 176 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.2 Latin American countries were early adopters of regulatory reform, but countries in Africa and Asia caught up fast Creation of regulatory entity for power sector, by region 100 80 Share of developing countries (%) 60 40 20 0 79 92 94 95 96 97 98 99 00 01 02 03 04 05 06 07 09 10 11 14 20 20 20 20 19 19 19 19 19 19 19 20 20 19 20 20 20 20 20 20 East Asia and Pacific Europe and Central Asia Latin America and Caribbean Middle East and North Africa South Asia Sub-Saharan Africa Source: Foster and others 2017. FIGURE 6.3 Tariff and quality regulation are the core original energy regulator, OSINERG, created in responsibilities of regulatory institutions 1996, had its scope of responsibilities expanded Regulatory mandate in the observatory countries, 2015 in 2007 to include the mining sector. It is now called Organismo Supervisor de la Inversión en End-user tari s Energía y Minería (OSINERGMIN). Pakistan’s Quality of supply and service electricity regulator, the National Electric Oversight of regulated utilities Power Regulatory Authority (NEPRA), was originally established as a separate institution Prices or terms of PPA in 1997. In 2016 it was subsumed under the Promotion of RE or EE line ministry in 2016, although the courts reversed that decision in 2017. Licensing trading or supply In addition to being widespread, regulatory Licensing distribution entities have noteworthy commonalities in both governance and methodology, with simi- Licensing transmission lar features seen across at least 80 percent of Licensing generation the observatory countries. Market design The regulation of tariffs and quality is the core of the economic regulation function and Competitive procurement forms part of the regulator’s responsibilities in Electrification or increased all observatory countries (figure 6.3). In almost access to energy 0 20 40 60 80 100 all countries, these functions combine with Share of observatory countries (%) oversight of the utilities to ensure compliance Source: Based on Rethinking Power Sector Reform utility database 2015. with tariff and quality regulation. The role of Note: EE = energy efficiency; PPA = power purchase agreement; RE = renewable energy. the regulator in market entry, though not Did Countries Establish Meaningful Power Sector Regulation? 177 universal, is also prevalent, with at least Potable, ONEE), municipal power distributors, 80 percent of countries giving the regulator and large private power distributors. Tariff responsibilities in licensing entry and negotiat- adjustments are made on recommendation of a ing terms of power purchase agreements tariff study prepared by an independent con- (PPAs). In addition to these core functions, reg- sultant but supervised by ONEE. Quality of ser- ulators at times also play a role in enforcing vice is either regulated by contract—in case of sector policies on clean energy (80 percent), private distributors—or self-regulated in the market design (65 percent), and electrification case of ONEE, as part of its mission statement.2 (55 percent). Despite their roles in negotiating In Tajikistan, regulatory powers are spread the terms of PPAs with independent power across several government institutions. All of producers (IPPs) (85 percent), regulators are these have advisory roles, as opposed to deci- less involved with competitive procurement sion-making authority, in their jurisdictions. (60 percent). Recommendations on tariffs are made by the An interesting exception is Colombia, where general antitrust agency. The national stan- regulatory and supervisory functions, com- dards agency is responsible for regulating qual- bined in most countries, are separated into two ity of service. The Ministry of Energy and distinct institutions. These are the Commission Water Resources issues licenses and is involved for the Regulation of Electricity and Gas in regulating the terms of PPAs, market design, (Comisión de Regulación de Energía y Gas, competitive procurement, and oversight of reg- CREG) and the (multisector) Superintendence ulated utilities. Finally, the nation’s state- for Public Household Services (SSPD). SSPD owned, vertically integrated monopoly power provides oversight for the provision of public utility, Barki Tojik, self-regulates, particularly utility services, as well as protection of con- with regard to tariffs. On the basis of submis- sumers’ rights. This arrangement allows the sions made by Barki Tojik, the Agency for SSPD to take over management of underper- Antimonopoly Services recommends electricity forming companies in extreme situations and tariffs to the government of Tajikistan. imposes discipline on the market. Many regulators continue to oversee state- A significant minority of countries opted not owned power utilities that have limited mana- to create an economic regulator for the sector. gerial incentives to respond to regulatory Among the 15 countries in the Power Sector instruments. As noted earlier, the creation of Reform Observatory, Morocco and Tajikistan regulatory entities was intended to pave the stand out. Their experience shows that regula- way for private sector participation, particu- tory functions remain relevant. In both coun- larly in distribution. Nevertheless, of the 70 tries, the national power utility itself plays a big percent of developing countries that created role in self-regulating, raising concerns about regulatory agencies since 1990, less than half conflict of interest. of them also introduced private sector partici- In the case of Morocco, tariff adjustments pation into power distribution. Most countries are enacted by an interministerial committee left the regulator to engage with largely state- (Commission Interministérielle des Prix). Led owned distribution utilities (Foster and others by the Ministry of General Affairs and 2017). Even countries that introduced some Governance (MAGG) with representatives private sector participation in distribution left from the ministries of the interior, energy, numerous state-owned enterprises (SOEs) finance and economy, trade and industry, and coexisting alongside private ones. agriculture and fisheries. The commission also Regulatory systems appear to function more has members from the national power utility effectively with a sizable private sector pres- (Office National de l’Electricité et de l’Eau ence in the power sector. The presence of 178 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.4 Countries with higher levels of private sector participation in distribution have stronger perceived regulatory performance Private sector participation vs. perceived regulation, 2015 70 Peru 60 Perceived regulatory performance (%) Uganda 50 Senegal Ukraine Philippines 40 Pakistan Tanzania Colombia Egypt, Arab Rep. 30 India – Andhra Pradesh India – Odisha 20 Kenya India – Rajasthan 10 Dominican Republic Tajikistan Correl. Coef. 60% Vietnam 0 10 20 30 40 50 60 70 Private sector participation index (%) Upper-middle income Lower-middle income Lower income Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Correl. Coef. = correlation coefficient. private sector players in power distribution (written in law and regulations) and perceived seems to affect the performance of regulators regulation (what actually happens) is well (figure 6.4). One reason for this effect may be known in the literature and has been empir- that, in the presence of private sector operators ically documented both for Latin America in the sector, it is more difficult for the govern- (Andrés and others 2007; Correa and others ment to circumvent its own regulatory frame- 2006; Rodriguez Pardina, Schlirf Rapti, and work because it can be held accountable by the Groom 2008) and for Western Europe (Gilardi private sector. The magnitude of the gap and Maggetti 2011; Hanretty and Koop 2013), between de jure and de facto regulatory perfor- though the gap is believed to be larger in the mance is smallest in countries where the pri- case of developing countries. Explanations cite vate sector is extensively involved in power the dissonance between imported regulatory distribution. frameworks and the local legal environment (Rodriguez Pardina, Schlirf Rapti, and Groom 2008), or the difficulty of defining a complete Finding #2: There is a difference contract between the political decision makers between establishing a good regulatory (the “principal”) and the regulatory agency framework on paper and building an (the “agent”) (Gilardi and Maggetti 2011). effective regulatory system in practice De jure scores are systematically higher Sound regulatory principles can be enshrined than perceived scores, and the latter are also in the legal framework, but there is no guar- more variable. The Regulatory Performance antee that they will be implemented. The dis- Index has been calculated both for the de jure crepancy between formal, or de jure, regulation and the perceived situation in each observatory Did Countries Establish Meaningful Power Sector Regulation? 179 FIGURE 6.5 Regulatory frameworks are not always perceived to function as written on paper Country performance on the Regulatory Performance Index, formal (de jure) vs. perceived practice 80 70 70 66 Regulatory Performance Index (%) 58 60 60 60 54 54 52 52 48 50 45 46 46 46 43 43 36 37 39 40 36 36 34 27 28 30 25 19 21 21 20 11 9 10 3 3 0 m an ic a a n h p. a ia an s e l da ru ga ne ny sh bi in ha es bl an Re na Pe ist st an ra m ne di ad pi pu Ke st nz ki et jik Ug lo Uk ab O ilip ja Se Pa Pr Re Vi Ta Co Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg m In – Do a di In De jure score Perceived score Source: Based on Rethinking Power Sector Reform utility database 2015. country, making a comparison between the The regulator is responsible for determining tar- two highly informative (figure 6.5). For some iffs in the sector under a price-cap mechanism countries, the difference between de jure and that indexes tariffs to inflation, exchange rate, perceived scores is modest, with both being fuel prices, and so on. Almost as soon as the pri- either consistently high (as in Peru) or consis- vate utilities began operations, oil prices rose, tently low (as in Tajikistan). For about half of so SIE ordered a tariff hike. The government the countries, the divergence can be more than rejected the hike, however, promising instead 20 percentage points. The only area where per- to pay the distribution companies the differ- ceived regulatory practice exceeded de jure ence between cost-reflective and actual tariffs. regulatory requirements was on public consul- The government then reneged on payment. The tations, with some regulators opting to make distribution companies’ worsening finances led use of such channels even when not legally eventually to the renationalization of the three obligated to do so. distribution utilities between 2003 (EDENORTE The Dominican Republic has the most and EDESUR) and 2009 (EDEESTE). As just divergent scores (de jure and perceived), with one government actor among many, SIE strug- a gap of more than 50 percentage points. The gled to implement its decisions regarding tariffs, country’s regulatory entity, Superintendencia quality, and market entry. Successive govern- de Electricidad (SIE), was established in 1998 ments used the utilities for political rather than when the sector was restructured and the three commercial objectives. Thus, in the Dominican distribution utilities were privatized. An impasse Republic, the regulator has no real authority in parliament delayed passage of the new over tariffs. Tariffs are not set according to a reg- energy law, so SIE had to be created in the first ulatory framework, nor are they adjusted in line instance by decree, weakening its legal standing. with legal mandates. 180 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.6 Gaps between paper and practice are particularly large on some aspects Formal (de jure) power vs. perceived practice, 2015 77 Market Entry 56 75 Quality 51 77 Tari s 65 Substance (average of 76 market entry, quality and tari ) 57 71 Autonomy 65 83 Accountability 73 Governance 59 (autonomy × accountability) 50 0 20 40 60 80 100 Regulatory Performance Index (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. There are also systematic differences in the higher levels of private sector participation in gap between de jure and perceived regulatory distribution tend to have both stronger per- scores for different aspects of regulation ceived and de jure regulatory performance. (­ figure 6.6). The average divergence is much Interestingly, they also have smaller gaps greater in regulatory substance, where it between the two (figure 6.7). The quality of reaches some 20 percentage points, than on utility governance seems to be related to posi- regulatory governance, where the gap is only tive regulatory performance (figure 6.8). about 10 percentage points. Despite the fact Because private sector participation and gover- that regulators have legally binding powers to nance are strongly correlated, however, it is set tariffs and regulate quality of service and hard to disentangle the effects. market entry, in practice their decisions may often be delayed or overruled by other govern- Finding #3: Most countries’ regulatory ment entities (see the next section for further regimes have gone further on details). Quality-of-service regulation is even accountability than they have on weaker, with many regulators reluctant to autonomy apply penalties for noncompliance with quality standards (as in the Arab Republic of Egypt, Regulatory autonomy appears to pose greater India, and Pakistan), and others not having political risks than does regulatory account- established any penalties at all (Kenya, ability. Because it involves the delegation of Tajikistan, Uganda, and Vietnam). powers from politicians to regulators, auton- The performance of regulation reflects the omy is less prevalent than accountability, institutional characteristics of the regulated which places regulators under political scru- companies. As noted earlier, countries with tiny. As might be expected, the average Did Countries Establish Meaningful Power Sector Regulation? 181 FIGURE 6.7 Countries with higher private sector participation have smaller gaps between de jure and perceived regulation De jure vs. perceived regulation, 2015 70 Correl. Coef. 49% Peru 60 Philippines Private sector participation index (%) 50 Colombia 40 Uganda 30 Ukraine India – Odisha Kenya 20 Dominican Republic Pakistan India – Rajasthan Vietnam India – Andhra Pradesh Senegal 10 Tajikistan Egypt, Arab Rep. Tanzania 0 –100 –80 –60 –40 –20 0 20 40 De jure and perceived divergence (%) Upper-middle income Lower-middle income Lower income Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Income classification according to 2018 data from the World Bank. Correl. Coef. = correlation coefficient. FIGURE 6.8 Quality of utility governance seems to be related to positive regulatory performance Utility governance vs. perceived regulatory performance 100 90 Kenya Philippines Uganda 80 Colombia Pakistan Ukraine Peru Utility governance index (%) 70 India – Rajasthan Senegal 60 Tanzania Tajikistan Egypt, Arab Rep. 50 Dominican Republic India – Andhra Pradesh India – Odisha 40 Vietnam 30 20 10 Correl. Coef. 57% 0 10 20 30 40 50 60 70 Perceived regulatory performance (%) Upper-middle income Lower-middle income Lower income Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Income classification according to 2018 data from the World Bank. Correl. Coef. = correlation coefficient. 182 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD de jure score across countries for autonomy on wholesale or PPA prices, this requirement is (at 71 percent) is well below the score for complied with in practice for only 73 percent accountability (83 percent). More countries of cases. Regulators are also much less likely to have accountability scores that exceed their involve nongovernment stakeholders in deci- autonomy scores (placing them below the sions to license new generation than the legal 45-degree line) than the other way around framework suggests. Regulated entities are (8), even if scores on autonomy and account- universally allowed to appeal decisions, but ability are also correlated (figure 6.9). appeals are made in only 81 percent of cases. Certain accountability measures are univer- One area where regulatory practice on sal, whereas others are practiced in only about accountability exceeds the legal requirements half of the countries (figure 6.10). Among the is the publication of regulatory recommenda- most ubiquitous practices on paper are legal tions, which are legally required in only appeals, publication of final decisions on tariffs, 33 percent of cases but practiced in 58 percent and production of annual reports. Much less of cases, reflecting voluntary disclosure. widely adopted are measures such as regular Peru’s regulatory framework provides an performance evaluations for regulators and the outstanding example by incorporating many participation of nongovernment stakeholders good-practice measures on accountability. in regulatory decision making. Gaps between Indeed, OSINERGMIN explicitly articulates the de jure and perceived performance in the area “Principle of Transparency” in the regulator’s of regulatory accountability tend to be minor, Action Principles (Decree No. 054-2001-PCM). except in certain areas. For instance, although Consultation drafts of proposed new regula- all regulators are required to publish decisions tions are required to be published for public FIGURE 6.9 Achievement of autonomous regulators remains a significant challenge even as regulators are more accountable Autonomy vs. accountability 1.0 Peru 0.9 Pakistan Dominican Republic Egypt, Arab Rep. 0.8 Uganda India – Andhra Pradesh India – Odisha 0.7 Senegal India – Ukraine Tanzania Rajasthan 0.6 Autonomy (%) Kenya Colombia 0.5 Vietnam Philippines 0.4 Tajikistan 0.3 0.2 Correl. Coef. 21% 0.1 Morocco 0 10 20 30 40 50 60 70 80 90 100 Accountability Upper-middle income Lower-middle income Lower income Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Income classification according to 2018 data from the World Bank. Correl. Coef. = correlation coefficient. Did Countries Establish Meaningful Power Sector Regulation? 183 FIGURE 6.10 Certain accountability measures have been universally adopted, whereas others are practiced in only about half of the countries Formal (de jure) power vs. perceived practice Regulatory oversight Regulator’s objectives formally stated in a law or regulation 100 Regulator required to report on its activities to a government body or other institution 88 Regulator’s performance evaluated by independent, nongovernmental entity 56 50 Legal Appeals Legally established process exists to challenge or appeal regulator’s decisions 100 81 Transparency Regulator provides publicly available annual reports 94 88 33 Regulator’s recommendations required to be made publicly available 58 Body receiving recommendations is required to respond publicly 33 17 Regulator required to publish decisions On end-user tari s 100 On licensing generation or supply 100 On wholesale or PPA prices 100 73 On market design 100 89 On regulated utilities oversight 85 85 Decision-making process legally requires participation of nongovernment stakeholders On end-user tari s 69 69 On licensing generation or supply 69 54 On wholesale or PPA prices and contract terms 38 31 On market design 30 20 On oversight of regulated utilities 38 23 0 20 40 60 80 100 Share of countries (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PPA = power purchase agreement. comment, and all final decisions appear in the appreciably less than their de jure autonomy official government gazette. Stakeholder partic- (figure 6.11). On paper, regulators have ipation in this and all other regulatory decisions authority over tariffs, quality, licensing, and is mandatory and usually takes place through PPAs, but these powers appear weaker in prac- public hearings. All regulatory decisions tice. Whereas 53 percent of regulators have must be justified through clearly articulated legally binding powers over end-user tariffs, principles that help promote predictability. this share drops as low as 20 percent in OSINERGMIN reports annually on its activities ­ practice. Sizable gaps (on the order of 20 per- to the prime minister and also publishes quar- centage points) also exist for the powers to terly updates on its website, allowing progress approve grid access charges and PPA/wholesale to be tracked against a range of performance prices. Although the funding basis for regula- targets and indicators. “Participatory regulatory tory ­entities is almost always determined by impact assessments” are undertaken to evalu- law, coming mainly from levies from consum- ate the costs and benefits of any new regulatory ers or regulated companies, in only about half decision for society as a whole. of cases are regulators free to determine their In practice, the perceived autonomy regula- own budgetary allocations. Many regulators tors have over legally binding decisions is also lack the freedom to determine their own 184 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.11 Regulator’s autonomy to make legally binding decisions on key issues is perceived to be significantly lower in practice than it looks de jure Formal (de jure) power vs. perceived practice End-user tariffs 79 Areas where entity has a mandate 68 Quality of supply and service 100 to regulate include 88 Electrification or increased access to energy 100 Decision-making autonomy On end-user tariffs 53 20 On grid access charges 88 69 On PPA/wholesale prices 87 Regulator’s decision is 67 On quality of supply/service 92 legally binding 77 On market design 87 60 On licensing 50 On utility oversight 85 69 End-user tariffs 71 57 Legal instruments prescribing how regulatory 17 Grid access charges 18 decisions should be taken for Quality of service/supply 94 Body receiving recommendations must provide public explanation for rejecting or modifying them 81 Budgetary autonomy Funding for the regulatory entity established by law or decree 100 94 59 Most of regulator’s budget comes from levies on consumers and regulated companies, or regulation tax 59 Leadership autonomy Legal basis for the entity’s existence 100 Regulator has power to determine its own organizational structure and rules 50 56 Regulator has power to determine the allocation and use of its own budget 44 50 Legal requirements or restrictions regarding the professional profile of regulator's leadership 94 88 Fixed term for regulator's leadership 88 81 Legal provisions by which regulator's head or leadership can be removed from office by government 75 Current leadership with known connection to government, SOE, or other regulated company 25 57 Employees in technical positions 64 Managerial autonomy Legal provision linking regulator's pay scale to private or public sector pay 53 Regulator required to follow the employment regulations applicable to government or civil service 63 47 0 20 40 60 80 100 Share of countries (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. Note: PPA = power purchase agreement; SOE = state-owned enterprise. organizational structures and rules. Regarding entities can be challenging. Egypt provides leadership, multimember commissions are the a particularly striking example, inasmuch as norm, with such commissions having between the minister of energy chairs both the board 3 and 13 members. Commissioners are selected of the regulatory agency, EgyptERA, and mostly by the executive, either the head of that of the national power utility holding state or the ministry of energy, though in a few company, the Egyptian Electricity Holding cases a special independent committee is Company (EEHC). Furthermore, tariff deci- charged with the selection process. In almost sions made by EgyptERA are subject to further all cases, there are legal requirements for the approval by the cabinet. In India, government professional profile of the leadership. The law control is exerted through the appointment of specifies fixed terms for their mandates (any- commissioners to the federal- and state-level where between three and seven years), which ­ regulators—the Central Electricity Regulatory can in most cases be renewed once or at most Commission (CERC) and the State Electricity twice. Regulatory Commissions (SERCs), respectively. Despite these legal provisions, achieving Elsewhere, autonomy of regulatory commis- autonomy for the leadership of regulatory sioners can be revoked through their removal Did Countries Establish Meaningful Power Sector Regulation? 185 from office before they complete their terms. concerns associated with rate-of-return regu- Tanzania appoints its regulatory ­ commissioners lation are pass-through of inefficient or in a transparent and competitive fashion. imprudent costs to the consumer, and poten- Nevertheless, board members have been tial overinvestment in service quality as a removed from office prior to the completion response to the guaranteed rate of return on of their full term. Similarly, in the Dominican investment. Republic, the president has been known to Incentive-based regulation originated in the change the regulatory leadership, and in United Kingdom and seeks to drive down Senegal the head of the regulatory entity was operational inefficiencies. Following an earlier reassigned to another public agency. application to the telecommunications sector, this kind of regulation was first introduced to the power sector by the regulator Stephen Finding #4: Many countries embraced Littlechild as part of the power sector reform of incentive regulation, but tariff regimes the 1990s. Incentive-based regulation normally tend to be poorly designed and the sets a price or revenue cap that is then fixed for incentives questionable a period of several years (Joskow 2014). Under At the heart of tariff regulation is the regula- price cap regulation, the utility is required to tory regime used to adjust tariffs over time. The keep the weighted average increase in its bas- two broad schools of thought on regulatory ket of prices beneath the increase in a specified regimes are known as rate-of-return regulation price index, minus X percent. This means that and incentive-based regulation. prices should decline by X percent per year in Rate-of-return regulation originated in the real terms, where X is an efficiency factor based United States and seeks to create a stable envi- on anticipated productivity improvements in ronment for investment. According to Jamison the sector. Because profits can be made only by (2007): outperforming the price cap over time, this type of regulation creates strong efficiency Rate of return regulation adjusts overall incentives. The four key features of pure price price levels according to the operator’s cap regulation are as follows: (1) infrequent accounting costs and cost of capital. In price reviews give utilities time to respond to most cases, the regulator reviews the the inherent efficiency incentives—in practice, operator’s overall price level in response every four to five years; (2) prices come to be to a claim by the operator that the rate of based on efficient market benchmarks rather return that it is receiving is less than its than the actual costs of the utility; (3) there is cost of capital, or in response to a suspi- no cost pass-through or retrospective adjust- cion of the regulator or claim by a con- ment for historic cost deviations; and (4) regu- sumer group that the actual rate of lators are not involved in setting tariff struc- return is greater than the cost of capital. tures. Incentive-based regulation can create Key features of pure rate-of-return regula- the potential for utilities to underinvest as a tion are that price adjustments can be as fre- means of increasing profits and thereby com- quent as required to maintain the rate of promising quality of service; in addition, return, and in practice are often annual; that investment risks are higher under a regime price adjustments are made in such a way that that permits utilities to make supernormal all costs are passed through to the consumer, losses or profits. including any historic deviations from antici- In practice, the pure forms of both types of pated costs; and that regulators often get regulation define extremes on a spectrum pop- involved in setting tariff structures. The main ulated with many intermediate approaches. 186 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD The practice of either regime is seldom inefficiency. Most developing countries have extreme, because regulators practicing price rapidly growing demand and a major invest- caps will refer to a utility’s actual costs in set- ment backlog; the more stable investment ting the cap, and regulators practicing the rate- incentives associated with rate-of-return regu- of-return model will do some form of bench- lation may be more relevant. Developing coun- marking to ensure that imprudently incurred tries also suffer from inefficiencies, but incen- costs are not passed through to consumers. In tive-based regulation reduces inefficiencies general, it is possible to create a range of hybrid only when the utility operates under a strong or intermediate regimes between the pure profit motive. This motivation is not always the forms by adjusting the length of the review case in countries where utilities continue to be period, using efficiency benchmarking, and state-owned. Furthermore, developing coun- varying the degree of cost pass-through and tries may lack the institutional, legal, and the extent of retrospective carryover of cost financial acumen needed to apply incen- deviations. tive-based regulation. There has been considerable debate regard- The spread of developing countries adopting ing the suitability of rate-of-return and incen- rate-of-return and incentive-based regulation tive-based regulation for developing countries appears even, although most lie somewhere in (Alexander 2014; Kessides 2012; Laffont between. Among the 15 observatory countries, 2005). Because it is supposedly less informa- approximately one-third follow incentive-­ tion intensive, incentive-based regulation is based regulation, whereas one-third f ­ollow said to be more suited to environments where rate-of-­return regulation; the remainder have audited accounts are scarce. Nevertheless, hybrid systems. This self-identified mechanism, incentive-based regulation was developed for c oupled with the length of the regulatory ­ mature markets that do not require major period, gives a sense of the incentive power of investments and where the focus is driving out the regimes in the analyzed countries: FIGURE 6.12 Most regulatory regimes are closer to rate-of-return regulation, with some incentive- based elements Incentive power Price or revenue cap with long review period Rate of return Colombia Dominican Republic Senegal Kenya India Vietnam Egypt, Peru Tanzania Uganda Arab Rep. Philippines Pakistan Source: Based on Rethinking Power Sector Reform utility database 2015. Did Countries Establish Meaningful Power Sector Regulation? 187 incentive-based regimes with relatively long the regulatory regime. Where we have infor- regulatory periods would give greater incen- mation on the tariff formula, it usually tives to efficiency than cost-plus regimes (fig- includes operating expenditure, depreciation, ure 6.12). taxes, and a return on capital. Most countries Nevertheless, the design details of the reg- base depreciation on the straight-line method ulatory regime are often more informative applied to the historic cost–asset valuation. than official descriptions. Countries practic- The allowed weighted average cost of capital ing pure incentive-based regulation would is typically based on a sector-risk premium tend to have long review periods, combined methodology (such as the Capital Asset with little cost pass-through, and full write- Pricing Model) applied to the asset base. off of historic cost deviations. Close analysis A handful of countries work with a preestab- of the design of different regulatory regimes lished fixed numeric value for the cost of shows that almost all of them mix features, capital. For the few cases where it is publicly ­ so that almost no country perfectly corre- available, this value lies in the 12–14 percent sponds to either regime (table 6.1). Some range. Interestingly, in the case of Colombia countries approach the incentive-based end and Peru, investments are converted into of the spectrum (such as Colombia, Peru, annuities. the Philippines, and Senegal), whereas oth- Nevertheless, regulatory accounting stan- ers are closer to the rate-of-return end of the dards have yet to be developed in many coun- spectrum (such as India, Kenya, Pakistan, tries. For regulators to be able to meaningfully Tanzania, Uganda, and Vietnam). In only interpret accounting information in determin- two cases (Peru and Senegal) have regu- ing the cost of service, it is important that this latory regimes fully written off the costs of information be submitted according to regula- over- or underrecovery from previous tariff-­ tory accounting guidelines. Such guidelines control periods rather than considering them have been developed in only about half of the in the next period. In all other cases, those countries considered. Good-practice examples costs are either partially or fully included. In include Colombia, Egypt, Pakistan, Peru, the most cases, end-user tariff-setting regulations Philippines, Tajikistan, Uganda, Ukraine, and incorporate incentives to improve efficiency Vietnam. by using benchmark efficiency or quality Automatic indexation mechanisms are parameters to determine prices, which are prevalent and usually provide protection usually related to transmission and distribu- against oil price shocks, foreign-exchange tion losses. This is the case in nine countries.3 movements, and domestic inflation. Two- In Kenya, for example, the aggregate trans- thirds of the observatory countries practice mission, distribution, and nontechnical losses automatic indexation of tariffs, which is benchmark was 16 percent in the period important in the developing country context 2015–16. For Pakistan utilities, transmission owing to the power sector’s exposure to oil losses were set at 3 percent and distribu- price and foreign-currency shocks that lie tion losses at 15 percent. In the Philippines, beyond the control of utilities. These adjust- these targets vary by utility—in the case of ments are usually done quarterly, though in MERALCO, for example, the system loss tar- some cases (Dominican Republic, Kenya, and get is 9 percent. Finally, maximum technical Pakistan) they are done every month. and commercial losses are 12 percent in the Semiannual inflationary adjustments are Dominican Republic. seen in Kenya and Tanzania. The most com- The tariff-setting methodology is remark- mon elements for cost pass-through are oil ably consistent across countries, regardless of prices, foreign exchange, and domestic 188 TABLE 6.1 Design characteristics of tariff regulatory regimes: An overview Tariff review Cost deviations Regulatory period Use of efficiency Cost pass- Basis of asset Basis for Basis of rate of Country regime (years) benchmarks through OPEX CAPEX valuation depreciation return Colombia Incentive based 5 Yes No Partially carried over Replacement cost Straight line Based on sector risk premium (13–14%) Dominican Incentive based 4 Yes (AT&C 12%) Yes (CPI, PPI, XR, Wholly carried over Market value / Straight line Fixed numeric Republic oil price, PPA) Replacement cost value Egypt, Rate of return No fixed No No Written off Partially carried Net historic cost Straight line Based on sector Arab Rep. period over risk premium India Rate of return 1 No Yes (unspecified) Partially/ wholly carried over Net historic cost Straight line Other plus incentives Kenya Rate of return 3 Yes (AT&C 15.9%) Yes (CPI, XR, oil Wholly carried over Net historic cost Straight line Fixed numeric price, PPA) value Pakistan Rate of return 1 Yes (AT&C 18.25%) Yes (CPI, oil price) Partially — Net historic cost Reducing Based on sector plus incentives carried over balance method risk premium Peru Incentive based 4 Yes Yes (CPI, PPI, XR, Written off — Replacement cost Straight line Annuity oil price, PPA) computed using a 12% discount rate over 30 years Philippines Incentive based 4 Yes Yes (CPI, XR) Partially/ Subject to — Straight line Based on sector wholly regulatory risk premium carried over discretion (12%) Senegal Incentive based 3 No Yes (unspecified) Written off Net historic cost Straight line Based on sector risk premium Tajikistan — — No No Written off — Net historic cost / Sum of the Fixed numeric Market value / years’ digits value (12.5%) Replacement cost Based on sector risk premium Tanzania Rate of return 3 No Yes (PPI, XR, oil Wholly Partially carried Net historic cost Straight line Based on sector price, PPA) carried over over risk premium Uganda Rate of return 1 Yes Yes (CPI, PPI, XR, Wholly carried over Net historic cost Straight line Fixed numeric plus incentives oil price) value Vietnam Rate of return 6 months Yes Yes (CPI, PPI, XR, Partially carried over — Straight line — oil price, PPA, generation mix) Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Information unavailable for Ukraine and inapplicable for Morocco. AT&C = aggregate technical & commercial losses; CAPEX = capital expenditure; CPI = consumer price index; OPEX = operating expenditure; PPA = power purchase agreement; PPI = producer price index; XR= exchange rate; — = not available. Did Countries Establish Meaningful Power Sector Regulation? 189 inflation, which are included in two-thirds of FIGURE 6.13 Two-thirds of the observatory countries practice indexation formulas. Several countries also automatic indexation of tariffs include other elements, such as force By index component majeure (figure 6.13). Oil price 64 Uganda provides an example of a carefully developed automatic indexation mechanism. Exchange rate 64 Base tariffs are adjusted quarterly to account Consumer price inflation 64 for inflation, exchange rate, and fuel costs. These adjustments apply only to the energy Power purchase agreement 45 charge (not to fixed monthly or maximum demand charges or to reconnection fees or the Producer price inflation 45 lifeline end-user tariff) and are capped so as Other 36 not to increase end-user tariffs by more than 2.5 percent in any given quarter. Inflationary 0 20 40 60 80 100 adjustments are applied only to the local cur- Share of observatory countries (%) rency portion of operating expenditure, and Source: Based on Rethinking Power Sector Reform utility database 2015. exchange-rate adjustments are applied only to the foreign currency portion of operating expenditure as well as to the return on capital practice, play only an advisory role in tariff (given that the utility operator is a foreign pri- setting, including the Dominican Republic, vate sector investor). India, Kenya, Senegal, and Tanzania. Discrepancies between de jure and perceived scores for tariff regulation at the country level Finding #5: Although tariff setting can be as large as 20 percentage points—even is the central function of regulatory 50 percentage points in the case of the agencies, their tariff recommendations Dominican Republic (figure 6.15). are not necessarily respected or In India, where a common regulatory applied framework for tariffs exists at the federal level, Tariff regulation on paper diverges substan- the practice of tariff regulation nonetheless tially from the practice of tariff regulation. varies substantially across states, reflecting The gap between de jure and perceived scores local political dynamics. Across three Indian is as large as 20 percentage points for states (Andhra Pradesh, Odisha, and some critical elements of tariff regulation Rajasthan), actual tariff adjustments in (figure 6.14). Particularly striking is the fact local-currency terms have been on the order of that, although most countries (nearly 90 per- 200 percent since 2010, and the corresponding cent) have a clear regulatory framework for regulators had mandated adjustments of tariff setting, barely 60 percent rely on this 400–700 percent over the same period framework in practice. Even more telling, ­(figure 6.16). The politically sensitive nature of although 94 percent of countries give the reg- tariff decisions appears to produce the system- ulatory entity authority over tariff setting, this atic rejection of regulatory decisions. authority is perceived not to prevail in some The state of Rajasthan provides a dramatic 35 percent of cases. Tajikistan and Vietnam illustration of how the regulator’s political are the only two countries where regulatory authorizing environment shifts over time, with tariff setting is not legally binding, but several potentially dire financial consequences for the other countries have regulators that, in utility. The Rajasthan Electricity Regulatory 190 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.14 Substantial divergence exists between tariff regulation as it appears on paper and as it is actually practiced Formal (de jure) power vs. perceived practice Normative framework for Tari s Frequency and schedule of tari revisions is determined by 75 law or regulation 100 End-user tari -setting is based on a clearly specified 88 regulatory framework 59 Legitimacy of costs is used as a basis for end-user tari 88 calculations 82 Regulatory framework includes a clear definition of cost 88 recovery 71 Regulator has authority over the tari level 94 65 Principles of tari -setting for end-user tari s are clearly 88 articulated 65 Existence of explicit policy or legal mandate regarding the 100 objectives in the determination of tari s 94 Determination of Tari s Regulator ensures utilities are compensated for the costs of 25 stranded assets 17 There are regulatory mechanisms to compensate generators 58 for the provision of firm capacity or ancillary services 33 Costs of incentive mechanisms to increase renewable 75 energy generation are borne by users 81 Regulated companies required to submit financial information 53 according to regulatory accounting standards 59 Tari -setting regulations avoid passing-through ine cient 76 costs to customers 65 There is a publicly available written formula prescribing 88 how end-user tari levels are to be set 76 0 20 40 60 80 100 Share of countries (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. Commission (RERC) was established in 2000. freeze caused the utility’s losses to reach a Despite strong legislation that laid out clear cumulative value of US$9 billion by the end of objectives for determining end-user tariffs, the 2014, higher than in any other Indian state. regulator never really managed to take control Meanwhile, distribution losses and quality of of the tariff-setting process. Until 2004, the supply remain problematic. The regulator is state government allowed regulatory tariff reg- unable to implement its orders on the ulations to be partially implemented. Gradually, state-controlled utilities. losses built up to US$50 million. The next Countries across Sub-Saharan Africa have government adopted a more radical position, ­ had a variety of experiences with regulatory refusing to consider any tariff revisions recom- tariff setting. In Uganda, for instance, tariff mended by the regulator. A decade-long tariff adjustments tracked those mandated by the Did Countries Establish Meaningful Power Sector Regulation? 191 FIGURE 6.15 Tariff recommendations made by regulatory entities are not necessarily respected or applied Formal (de jure) power vs. perceived practice 100 100 93 92 92 92 88 88 83 83 83 83 83 79 80 75 75 75 75 75 76 67 67 Tari index (%) 58 60 50 50 50 40 29 29 21 20 17 0 ic co an h a n a p. m ia an l da e a s ru ga ne sh ny bi in es ha bl an Re na Pe oc ist st an ra ne m di ad pi pu Ke st nz ki et jik or Ug Uk lo ab O ilip ja Se Pa Pr Re Vi Ta Co M Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg m In – Do a di In De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. FIGURE 6.16 In India, authorized tariff increases fall well short of those approved by state regulators a. Andhra Pradesh b. Odisha 800 800 Tari adjustment (%) Tari adjustment (%) 600 600 452 446 400 400 275 200 200 183 0 0 2010 2011 2012 2013 2014 2015 2010 2011 2012 2013 2014 2015 c. Rajasthan 800 683 Tari adjustment (%) 600 400 220 200 0 2010 2011 2012 2013 2014 2015 Proposed adjustment Actual adjustment Source: Based on Rethinking Power Sector Reform utility database 2015. 192 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.17 Tanzania and Uganda represent the extremes of the range of experiences of African countries Proposed vs. actual adjustments, 2008–17 a. Tanzania b. Uganda 500 500 404 400 400 Tari adjustment (%) Tari adjustment (%) 300 272 300 200 200 188 100 169 100 100 100 0 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2008 2009 2010 2011 2012 2013 2014 2015 2016 Actual Proposed Source: Based on Rethinking Power Sector Reform utility database 2015. regulator fairly closely from 2008 to 2016. That were permitted: in 1999, 2008, and 2013. a private concessionaire was running the Political sensitivities have led successive gov- national utility partly explains why mandated ernments to prevent full tariff reviews during tariff adjustments were largely honored. electoral periods. Nevertheless, Kenya’s retail In Tanzania, regulatory tariff adjustments were tariffs incorporate a cost pass-through mecha- largely implemented from 2008 to 2012, nism for the monthly indexation of foreign-­ though the escalation of costs associated with exchange fluctuations and fuel costs, as well as the drought period meant that subsequent tar- semiannual indexation of domestic price infla- iff adjustments were kept far below levels tion. Despite the prohibition of regular tariff mandated by the regulator (figure 6.17). Not reviews, this automatic mechanism has been only were regulatory decisions overturned, but allowed to function smoothly since its intro- also the senior leadership and management of duction in 1997; it partially compensates for the regulatory agency were sacked. In Senegal, the absence of regular tariff revisions. This tariffs have been frozen since 2009 by govern- changed in 2017, an election year, when ment fiat, irrespective of regulatory advice. drought conditions would have led to a large Kenya’s experience with tariff regulation indexation adjustment owing to higher fuel illustrates the interaction with the electoral costs, but the indexation was permitted to pass cycle, as well as the role that can be played by after the elections were over. automatic indexation. The ERC replaced the ERB in 2007 and is considered one of the bet- Finding #6: In countries where cost- ter-performing regulators in the region, having recovery tariffs recommended by the greatly improved its technical capacity over the regulator are not implemented, the years. Solid legislation has given the regulator regulator may play a role in ensuring a stronger voice in decision making. The ERC that sector revenue requirements are has moved tariffs closer to cost-recovery levels met through subsidies despite government disregard for the regula- tory process. According to law, ERC is meant to Ideally, regulators should be able to set tariffs conduct a tariff review every 3 years, yet over a at cost-recovery levels and enforce their appli- period of about 20 years only three reviews cation, but this is not always politically Did Countries Establish Meaningful Power Sector Regulation? 193 feasible. Most regulators have the formal legal provides a compensating subsidy to the rele- authority to set tariffs. This authority can be vant utilities. EgyptERA informs the cabinet overridden, however, by concerns about the of the tariff-subsidy combinations that are social impact of higher tariffs during elec- compatible with financial equilibrium for toral periods. As a result, tariff adjustments in the sector. The minister of finance is involved many jurisdictions have been only a fraction in these decisions, because the level of subsidy of what the regulator considered necessary, has to be accounted for in the national budget. leaving utilities with a hefty financial gap. In This approach has been successfully applied principle, the financial viability of the utility since 2014. can be safeguarded if below-cost tariffs are In Senegal, the regulator played a key role offset by fiscal transfers large enough to meet in calculating the magnitude of fiscal subsidies the shortfall. Where cost-recovery tariffs can- needed to maintain the financial equilibrium not be fully implemented, regulators can of the utility, at least for a time. According to counsel a combination of tariff and subsidies the 1998 law, regulation of electricity tariffs in to meet the utility’s revenue requirements. Senegal is based on a revenue-cap system that The choice of the exact tariff and subsidy determines the revenue required for the eco- combination is therefore left to the political nomic and financial viability of the utility, authorities. National Electricity Company of Senegal Regulators have sometimes counseled on (Senelec). In 2008, when the last 18 percent the magnitude of subsidies needed to compen- increase in tariffs occurred, the Senegalese gov- sate for below cost-recovery tariffs, but this ernment decided to freeze tariffs, promising approach is risky. Both Egypt and Senegal used compensation to Senelec. No payment sched- a similar approach, and it met with some tem- ule was specified, however, delaying compen- porary success. The regulator does not really sation and forcing Senelec to take on expensive have the power to hold the ministry of finance commercial bank debt to continue operations. accountable for paying the requisite subsidy, A legal amendment introduced in 2011 aimed leaving the approach vulnerable during periods to tackle this issue by having the regulator cal- of fiscal duress. Moreover, once established, culate Senelec’s maximum authorized revenue such arrangements may be difficult to reverse on a quarterly basis and requiring the ministry and in some specific cases may even lead to of finance to pay compensation accordingly. If sustained inefficiency in utilities as seen in compensation cannot be paid, the government Pakistan. must provide Senelec with a “letter of com- In Egypt, regulations stipulate that the regu- fort,” allowing the company to borrow from lator must inform the Cabinet of Ministers of banks while the government commits to guar- the compensating subsidy owed to the utility if antee all financial fees and principal debt cost-recovery tariffs are not approved. The reg- repayment. The introduction of a more formal, ulatory agency, EgyptERA, has authority over legally based system for compensation pay- the tariff level and structure as well as the fre- ments resulted in more disciplined fiscal trans- quency of tariff revisions; but every tariff order fers for several years, until Senegal faced fiscal requires cabinet approval, and the cabinet does challenges in 2017 and compensation pay- not always endorse the adjustments recom- ments once again ceased. mended by EgyptERA. Following the enact- Pakistan’s experience with so-called tariff ment of a new energy law in 2015, the associ- differential subsidies illustrates the perverse ated regulations stipulate that the cabinet may efficiency incentives that can arise from this choose to set tariffs below the level recom- approach. In Pakistan, for social reasons, the mended by EgyptERA only if the government retail pricing approach involves both a tariff 194 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD determined by the regulator and a tariff noti- adequate capital expenditure allowances in fied by government; the latter is the one actu- the regulatory revenue base so that the invest- ally applied. The regulatory tariff is computed ments needed to meet the standard can be to afford the utility adequate revenues to cover funded. Financial penalties need to be set at a its costs after meeting certain efficiency param- level high enough to affect behavior and eters for revenue collection, system losses, and should ideally reflect the cost of outages to fuel purchases. The difference between the consumers. For example, in the European government’s notified tariff and the regulator’s Union, the economic costs suffered by custom- cost-based tariff, known as the tariff differential ers as a result of power outages (known as subsidy, is paid by the government to the utili- Value of Lost Load) are estimated to range ties. In principle, this subsidy should assure the from €5 to €10 per kilowatt-hour. Quality-of- financial viability of the sector. In practice, service penalties are critical in price cap regu- however, it does not; because utility perfor- latory regimes, where the utility may face mance falls short of the efficiency benchmarks incentives to cut costs by reducing quality of the regulator uses to determine the cost-based service. Beyond financial penalties, regular, tariff, losses are incurred despite the subsidy. transparent reports on the achievement of These losses become circular debt when com- quality standards can also affect a utility’s pub- panies with insufficient revenues fail to pay lic reputation and improve its performance. their suppliers in full. Because the government Quality-of-service regulation is close to uni- is ultimately responsible for this debt through versal in developing countries, but enforce- its ownership of state-owned companies, and ment is weak, even on paper. Near-universal because such circular debts are periodically formal quality standards exist for paid off by the government, managers of the • Product quality (such as frequency or volt- loss-prone utilities have little incentive to age variations), improve the efficiency of their operations. • Service quality (interruptions), and • Customer service (for example, response to Finding #7: Quality-of-service complaints). regulation leaves much to be desired Utilities are legally required to meet these Quality-of-service regulation is an important standards and must submit data periodically complement to tariff regulation, particularly regarding their compliance. Despite the stan- under incentive-based regulatory regimes. dards, however, enforcement remains weak. Regulation of service quality entails first estab- Only about one-third of countries in the lishing a suitable quality standard and then sample attaches positive or negative financial ­ creating a system of incentives to induce incentives to the achievement of quality stan- ­ utilities to meet that standard (Adam 2011). dards, and only half of the countries publishes Quality standards should be industrywide, information on compliance. clearly defined, and long term to provide Moreover, the discrepancy between the ­ regulatory certainty and foster investment. official quality-of-service regulation and actual The desired quality level should be informed regulatory practice is particularly large in some by a balancing of costs to the industry against cases. Whereas the average score for formal benefits to the consumer, because standards (de jure) quality-of-service regulation across can sometimes be set too high. Incentive countries is 75 percent, it drops to 51 ­percent in mechanisms for meeting quality standards will perceived terms. The situation v ­ rastically ­ aries d likely require a combination of financial pen- across countries (figure 6.18). In middle-­ alties for failing to meet the standard, with income countries of Latin America, East Asia, Did Countries Establish Meaningful Power Sector Regulation? 195 FIGURE 6.18 Quality of service regulation has some of the largest gaps between rules on paper and in practice Formal (de jure) power vs. perceived practice 100 100 100 100 92 92 92 83 83 83 79 80 71 71 63 Quality index (%) 58 60 54 43 46 42 42 42 40 38 33 33 29 25 25 20 8 8 0 a n h co a p. l ic ia an da an m s e a ru ga ne sh ny bi in ha es bl an Re na Pe oc ist st an ra ne m di ad pi pu Ke st nz ki et jik or Ug Uk lo ab O ilip ja Se Pa Pr Re Vi Ta Co M Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg In m – Do a di In De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. and Eastern Europe—such as Colombia, Peru, actually publishes such standards. Fines for the Philippines, Ukraine, and Vietnam—a good noncompliance are a legal requirement in system appears to be regulating quality of ser- about 60 percent of cases but have been defined vice, and it also functions effectively in prac- in only 24 percent. Whereas all utilities are tice. For the countries of Sub-Saharan Africa— required to report quality-of-service data to the Kenya, Senegal, Tanzania, and Uganda—the de regulator, only about 70 percent does so, and jure scores for quality of regulation are much little more than half operates an automated lower, at about 40–60 percent, and even these information system. In about 65 percent of partially developed regulatory frameworks countries, quality performance is supposed to are not fully observed. Particularly striking be made public, but this happens in only about is the case of the Indian states, where the de 40 percent of cases. The lack of compliance jure framework for ­ quality-of-service scores with quality-of-service regulation can be are relatively high at 80–90 percent, but the attributed not only to the inefficiency of the perceived scores range from 8 to 25 percent. utilities but also in some cases to standards The situation is similar for the Dominican being set at unrealistically high levels. Many Republic and Egypt. countries report that utilities try to observe Many fundamental components of a quality-​ standards but fail because of the technical chal- of-service regulation system fail to be applied in lenges (figure 6.19). practice. A closer examination reveals numer- Colombia provides a good counterexample ous areas where formal features of the regula- of a country where quality-of-service regula- tory framework for quality of service are over- tion is working effectively. Prior to the power looked in practice. Although all countries have sector reforms of 1994, Colombia barely legal requirements for quality-of-service regu- c onsidered quality-of-service regulations. ­ lation and standards, only about two-thirds The new regulator, CREG, published 196 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.19 Quality-of-service regulations are not widely implemented, often for lack of technical capability within utilities Formal (de jure) power vs. perceived practice Quality-of-service standard Utilities are legally required to meet 100 quality of service standards 65 Existence of formally written and 97 publicly available standards 67 Public disclosure of quality 71 standards achievement 47 59 Existence of fines for non-achievement 24 Quality-of-service enforcement Requirement to report technical 100 data on a periodic basis 71 Specification of how technical performance 71 data should be collected 76 Independent validation of technical 47 performance data 35 Requirement of automated information 71 management system 35 Publicity of measurements of 65 the quality or reliability 41 Existence of financial incentivesto 53 meet customer service standards 47 0 20 40 60 80 100 Share of countries implementing regulation (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. quality-of-service regulations for power distri- Figures are reported monthly to the regulator, bution in 1998, adopting the international which then compiles a quarterly review of standards, SAIFI (System Average Interruption compliance against maximum allowed values Frequency Index) and SAIDI (System Average for each of the four regions. Utilities take Interruption Duration Index), as the two quality-of-service standards seriously and main indicators. A third indicator capturing generally comply, inasmuch as shortfalls the percentage of lost load was added in 2008. result in compensation payments to users. Each of Colombia’s four geographical regions The quality standards, together with the per- had a different standard to meet—variations formance data reported by the utilities and that acknowledged the varying local distribu- information on any resulting penalties paid, tion systems and the magnitude of the associ- are all available to the public. Formal mea- ated technical challenges. Under the current surements of overall customer satisfaction are system, network operators record duration also required and undertaken both by utilities and frequency of outages for their respective and the regulator, both through customer sur- users on each circuit and voltage transformer. veys and website comment forms. Did Countries Establish Meaningful Power Sector Regulation? 197 Finding #8: Regulators play a role in regulation, dropping to 56 percent for per- licensing market entry but are not ceived regulation. Most countries with well-­ always involved in the power purchase developed formal regulations for market entry agreements, a critical area lag on practice (figure 6.20). This finding is true in the Dominican Republic, Kenya, the With most developing countries opening Philippines, Tanzania, and Vietnam. In Egypt, power generation to IPPs, they need a reliable India, and Pakistan, even the formal frame- way to admit new players into their markets. work is not well defined. Colombia and Peru A licensing process for new entrants, one that have advanced regulatory frameworks, but is operated by the regulator, is a good way to their scores on practice are not especially high. go given the sensitive nature of these activi- The most serious deficiencies in the regula- ties. In some cases a government agency, such tory framework for market entry relate to the as the ministry of energy, could oversee revocation of licenses and the award of IPPs. licensing new entrants. Regulators are some- The evidence suggests that, once licenses are times involved with procuring IPPs, or they awarded, regulators fail to monitor compliance review the terms of PPAs prior to their signa- with license conditions. Worse, they do not ture to ensure that the PPAs represent value impose the legally stipulated fines, making it for money, because costs will be passed on difficult to force nonperforming companies directly to consumers through retail tariffs. to relinquish their licenses (figure 6.21). Market-entry regulations either are under- Particularly striking is that, whereas half the developed or fail to be properly implemented. countries empower regulators to conduct IPP Overall, countries scored 77 percent on the de procurement, only about a quarter of them jure regulatory framework for market-entry FIGURE 6.20 Most countries with well-developed formal regulations for market entry lag far behind in terms of practice Formal (de jure) power vs. perceived practice 100 100 100 100 100 100 100 83 80 75 71 71 71 67 67 67 67 67 67 Market entry index (%) 58 60 54 50 50 50 50 50 50 42 42 42 40 29 20 0 an h a n a ic p. u e m an a co s ia da l ga ne sh bi ny in es ha r bl an Re na Pe oc st ist an ra m ne di ad pi pu Ke st nz ki et jik or Ug lo Uk ab O ilip ja Se Pa Pr Re Vi Ta Co M Ta Ra – Ar Ph ra a an – di t, dh ic yp a In An di in Eg In m – Do a di In De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. 198 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 6.21 Serious deficiencies relate to the revocation of licenses and the award of IPPs Formal (de jure) power vs. perceived practice Permitting new entrants Regulator is responsible for monitoring 88 compliance with licenses’ terms 76 Regulator can impose penalties for 100 violating license or permit terms 60 Penalties are formally written 80 and publicly available 75 There are provisions to force companies 100 to relinquish licenses 88 PPA approvals Regulator is legally required to approve 59 PPAs or other power sales contracts 59 There is a legally specified period of time 60 to approve or refuse a proposed PPA 0 Regulator has authority over the process by which 50 utilities can select or procure power from IPPs 27 0 20 40 60 80 100 Share of countries (%) De jure Perceived Source: Based on Rethinking Power Sector Reform utility database 2015. Note: IPPs = independent power producers; PPA = power purchase agreement. actually allow the regulator to do so in practice. are allowed for government comments; in prac- Regulators generally review the terms of PPAs, tice, delays can last a year or more, sometimes but they also, invariably, exceed legal time lim- well past the expiry of the incentive program its for doing so. that originally encouraged the projects. Pakistan has an ineffective system of mar- ket-entry regulation, with damaging results. LOOKING AHEAD Pakistan has allowed IPPs in generation since Disruptive technologies will require new regula- 1994 under a clear framework for market-entry tory frameworks. The historical practice of regu- regulation. In practice, however, new entrants lation reviewed thus far does not consider the (particularly for solar) struggle through the ini- implications of new technologies, such as dis- tial stages of approval. The major roadblocks tributed energy resources, battery storage, and involve land allocation. First, for both wind and smart grids. Emerging experience from frontier solar technologies, the number of letters of markets suggests that such technological disrup- intent issued with associated land-allocation tion is posing new challenges for regulators. rights is greater than the land available and the Traditional practices of tariff regulation may interconnection potential of the grid, which thus need to be overhauled. The traditional means that only private developers with gov- cost-of-service model of utility regulation was ernment connections will have land allocated well suited to advancing the policy objective of and be able to move on to the next stage (feasi- expanding the centralized grid (Graffy and bility studies). Second, the government issues Kihm 2014; Kihm and others 2015). Policy interconnection permits based on the findings goals are now shifting, however, toward encour- of a required interconnection study. Thirty days aging consumers to adopt distributed energy Did Countries Establish Meaningful Power Sector Regulation? 199 resources that serve to accelerate decarboniza- fundamentally change the overall cost-of-ser- tion, while also strengthening distribution net- vice model. Instead, they create mechanisms works (Graffy and Kihm 2014; Kihm and others by which utilities can earn more revenue by 2015). As distributed energy resources become improving operational efficiency, rather than more readily available, the concern becomes by increasing capital expenditures. This has how to adopt a regulatory model that ensures typically been achieved by lengthening the adequate cost recovery for the needed wires and regulatory review period from one year to sev- poles of the distribution network infrastructure, eral years. With a longer review period, utilities while ensuring that there is also an incentive to that spend below the preapproved envelope for embrace emission reduction, resiliency, and operating expenditures can hold on to the sav- energy efficiency goals through increased reli- ings over a longer time period, strengthening ance on decentralized resources. their incentive to cut costs in the first place.5 Traditional cost-of-service approaches to Regulators are already adapting perfor- rate making introduce incentives that run mance-based regulation to incentivize utilities counter to the efficient integration of distrib- to meet energy-efficiency goals and achieve uted energy resources. Perhaps the most specific quality-of-service outcomes. Newer widely adopted regulatory model for determin- forms of performance-based regulation intro- ing the revenue requirement of the d ­ istribution duce specific goals for peak reduction, resil- utilities is the cost-of-service approach, also ience, emissions reduction, and customer known as rate-of-return regulation. According satisfaction. Under performance-based regula- ­ to this methodology, utility tariffs are set to tion, utilities have the freedom to determine earn a guaranteed return on a specified regula- how best to reach those goals while meeting tory rate base of allowed assets plus an allow- their targets for safety, reliability, affordability, ance for legitimate operating expenditures. and accessibility. Revenue caps, rate freezes, This framework embeds a particular set of multiyear rate plans, or earnings adjustments incentives, so utilities benefit financially either based on achieving specific energy-efficiency from increasing their investments or from targets (that is, revenue decoupling) are all decreasing their operating expenditures. 4 examples of performance-based incentives. Moreover, because the focus of regulation is on The new regulatory challenge is to incentiv- the unit price, utilities also benefit financially ize utilities to consider the “nonwires alterna- from expansion in the volume of sales. All of tives” alongside traditional grid investment. The these incentives run counter to the integration challenges posed by incentivizing the adoption of distributed energy resources, which can be of distributed energy resources go beyond expected to reduce the need for capital invest- the examples cited above. The central issue is ments by utilities, potentially increase the need to encourage utilities to consider nonwires for operating expenditures, and reduce the net ­ alternatives—or consumer-sited distributed flow of power from the utility to the customer. energy resources—alongside (or even instead Utilities operating under traditional cost-of-ser- of) traditional grid investment. Customer-sited vice regulation have little incentive to embrace generation, demand response, energy effi- distributed energy resources. ciency, and battery storage could, in combina- Performance-based regulatory approaches tion, serve as an alternative to a traditional provide utilities with more targeted incentives. ­ distribution-grid upgrade, while producing the The long-standing alternative to rate-of-return same result at lower cost. Where adopted, such regulation is incentive- or performance-based nonwires alternatives could create more resil- regulation, as discussed earlier. Existing appli- ient distribution systems while avoiding invest- cations of performance-based regulation do not ments that may prove difficult to recover. 200 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Experiments underway in the United adoption of new technologies and permits third- Kingdom and the U.S. state of New York illus- party delivery of services. The output compo- trate how performance-based regulation can nent focuses on service dimensions of concern be adapted to encourage such technological to consumers. This approach departs from tradi- innovation. The United Kingdom has intro- tional cash-flow analysis, giving more weight to duced the “Revenue using incentives to deliver flexibility and optionality, and incorporating ele- innovation and outputs” (RIIO) model. In the ments of output-based regulation. In order to United States, New York is implementing the ensure sufficient incentive for innovation under “Reforming the Energy Vision” (REV) model. conditions of uncertainty, tariff review periods The British RIIO approach builds up the util- are lengthened beyond the traditional three- ity’s revenue using a series of incentive-based to-five-year cycle to a much longer eight-year components and further lengthens the regula- cycle. According to Ofgem (2015), the first tory review period. Under such a “total expen- implementation of RIIO is proving successful, diture” (TOTEX) approach, capital expenditures fundamentally altering behavior and board and operating expenses are considered together ­ discussions at companies and encouraging rather than as separate costs, as in the past stakeholders to present well-thought-out, (Cave 2016). This approach does not directly detailed, and better-justified business plans. tackle the uncertainty problem, but it dispenses The New York REV model focuses on creat- with the need for a detailed review of cost fore- ing new earnings mechanisms and broadening casts for individual projects and may reduce the the definition of efficiency gains. Under New bias toward capital solutions. Additionally, York’s REV procedures, utilities will provide TOTEX may, in some instances, be more stable distributed system platform services that enable over time and more comparable between com- a market for distributed energy resource pro- panies than capital expenditure alone. The viders. REV’s goal is to integrate distributed overall regulatory revenue requirement has energy resources into utility system planning three components (­ figure 6.22): incentives, and utility operations. By relying on third- innovation, and output (Ofgem 2015). party capital for increased deployment of dis- The incentives component focuses on the tributed energy resources, utilities become less efficient delivery of outputs. The innovation reliant on utility capital investment; however, component provides specific incentives for by relying on a platform-based model, utilities may increase operating expenses associated FIGURE 6.22 RIIO framework with providing services for the platform mar- Constraint set up front to ensure ketplace. To offset this loss of revenue and Timely Network Balance between account for higher operating costs, REV intro- Revenue Transparency and companies costs faced by and duces new revenue sources so utilities can shift e cient are current and predictability delivery financeable future consumers to a platform-based market system. The model = Deliver outputs e ciently over time with introduces market-based earnings (in addition Incentives Eight-year Rewards/penalties Upfront to earnings-impact and earnings-sharing control for delivery e ciency rate mechanisms) that encourage gains based on + Technical and commercial innovation encouraged through performance and outcomes, in line with regu- Innovation Core price Option to give third Innovation lation based on performance instead of on cap- control parties a greater role stimulus incentives in delivery package ital investments alone. These new mechanisms + are designed around the value-added services Outputs set out in clear “compact,” reflecting expectations of Outputs utilities can provide in support of a plat- current and future consumers Source: Nixon 2013. form-based marketplace. Examples include Note: RIIO = revenue using incentives to deliver innovation and outputs. data analysis or creating an online portal. The Did Countries Establish Meaningful Power Sector Regulation? 201 idea is to enable earnings for utilities that customers to receive a credit on their bills for decrease reliance on utility capital investment. any electricity generated by their rooftop solar Furthermore, traditional approaches to set- panels that they feed back into the system, ting consumer-facing tariff structures are charac- essentially by subtracting it from the demand terized by static volumetric charges driven more used to calculate the customer’s bill.6 This by social than economic considerations. Once approach, designed to incentivize the adoption regulators set the overall revenue requirement of rooftop solar, essentially buys back distributed for the utility, the requirement must be met generation from prosumers at the marginal through tariff structures that allocate charges to variable retail charge that they face in the tariff particular consumer groups and consumption structure. Given that almost all the fixed costs of brackets. Traditionally, tariff structures have the grid are being recovered through the vari- involved averaging costs over entire customer able charge, prosumers are being exonerated classes (such as residential, commercial, and from contributing to recovering the fixed costs industrial) without considering the temporal or of the grid while benefiting from the grid as a spatial profiles of consumption, which may in backup source when the electricity they gener- practice be significant cost drivers. Although a ate falls short. From the utility’s perspective, the high share of the costs of electricity production is consumer-supplied energy provides only vari- fixed, at least in the short run, utilities have pre- able supply but is being remunerated at full pro- ferred to recover the bulk of their costs through duction costs. In that sense, the utility is likely to volumetric charges. For residential customers, be financially harmed by distributed generation these charges are often structured as increasing under net metering arrangements. As a recent block tariff structures where the marginal tariff U.S. Department of Energy study observes, rate is higher for higher blocks of consumption, “After a century of utility concerns over whether under the premise of protecting low-income rate increases will be high enough to allow full consumers believed to have lower energy cost recovery, the emergence of elastic demand demand. On the same pretext, cross-subsidies for electricity will shift the focus to whether util- between customer classes are not unusual and ity costs are simply too high to be recoverable” often entail higher variable charges for commer- (Corneli and Kihm 2015). cial than residential customers. The potential perverse incentives created by Existing rate structures, especially for resi- net metering arrangements are further exacer- dential or small commercial consumers, assume bated in the presence of various kinds of that consumers receive power and generate cross-subsidies. The effects described above are none of their own. This fundamental assump- further accentuated where cross-subsidies exist tion is no longer true, so the challenge going between large and small consumers via increas- forward is understanding how to set rates in a ing block tariff structures. Large consumers, on world where consumers may have various which utilities rely disproportionately for reve- options for electricity consumption and produc- nues, will face strong incentives to become tion. Because tariff structures fundamentally prosumers, because their production is remu- ­ affect the consumption and production choices nerated at the higher block tariffs. Doing so of consumers, traditional approaches to pricing might enable them to reduce their use of grid often fail to give the appropriate incentives. electricity to a level compatible with the lower-­ The widespread adoption of net metering as cost consumption blocks. (This incentive may a mechanism for incentivizing distributed gen- widen as consumers acquire electric vehicles, eration illustrates the challenges of creating necessitating higher volumes of domestic elec- appropriate incentives. Net energy metering tricity consumption.) In countries where tariff allows residential and small commercial structures embody cross-subsidies from 202 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD commercial to residential consumers, the lies in extending them to the residential incentive for commercial entities to become segment. prosumers will be particularly strong because own-­ generation may enable them to escape the CONCLUSION cross-­subsidy. Considering that the cost of roof- In conclusion, despite widespread creation of top solar at the larger commercial scale of pro- regulatory entities and the adoption of solid legal duction is more attractive than for smaller frameworks, the practice of independent regula- residential loads, net metering may present ­ tion remains elusive. The establishment of regu- utilities with a significant risk of revenue loss latory frameworks was one of the most popular from the commercial segment. reforms from the 1990s model, and many coun- A simple fix to the problem of net metering tries succeeded in enacting technically sound is to introduce a separate fixed charge for pro- regulatory methodologies. Nevertheless, imple- sumers, although this solution creates prob- mentation has often fallen short. Several lems of its own. The fixed charge would repre- ­ countries in the observatory show sizable dis- sent prosumers’ contribution to the fixed cost crepancies between the quality of formal (de of the network. It is difficult to set the fixed jure) regulatory frameworks and the extent to charge at the right level, however; if set incor- which those frameworks are perceived to oper- rectly, it may discourage the adoption of dis- ate in practice. The discrepancy is particularly tributed energy resources that might otherwise wide when it comes to regulating quality of ser- benefit distribution system operators. vice and market entry. Moreover, although A more sophisticated alternative is to intro- almost all countries grant regulators legal duce a separate charging structure for electricity authority over tariff setting, this authority is sold into the grid, ideally based on time of use. respected in only about two-thirds of cases. Such a rate could reflect the energy, capacity, There is substantial evidence that tariff adjust- environmental, and locational benefits prosum- ments systematically lag behind regulatory rec- ers provide to the system as opposed to credit- ommendations; in some cases, regulators are ing them at the average retail rate. New York being used primarily to determine the subsidy has a rate design for distributed resources, requirement for the sector over and above polit- which is based on a “value stack.” This design ically constrained tariffs. Although many coun- considers the locational value of distributed tries have espoused incentive regulation, the resources in addition to crediting them for their incentive regimes are weak, and the relevance of supply value. Most of these rate designs wrestle incentives may be limited given that many regu- with quantifying locational or time-based val- lators continue to oversee primarily SOEs with ues that adequately compensate consumers for limited commercial orientation. their energy production. Existing rate designs The likely impact of technological disruptions do not take these considerations into account. of the power sector suggests that the regulator’s There is growing interest in time-of-use rates task will become not only more complex but also that vary according to the time band in which critical in driving the pace of innovation. energy consumption or production takes place, Incentive-based regulation will grow in relevance with a view to incentivizing peak-shifting as regulators struggle to encourage utilities to behaviors. Critical peak pricing is a similar con- take on decentralized solutions and more sophis- cept. Load charges that capture the extent to ticated demand management. Regulatory which customers contribute to the system peak involvement in utility tariff structures will also may also be helpful. Many of these structures become more critical to ensure that rooftop solar have already been in place for some time for generation by prosumers is not overincentivized commercial and industrial customers, particu- and that all customers contribute their fair share larly in developed countries, and the novelty to the fixed costs of maintaining the power grid. ANNEX 6A. FORMAL (DE JURE) SCORES ON THE REGULATORY PERFORMANCE INDEX Percent Dominican Egypt, India – India – India – Indicators Colombia Republic Arab Rep. Andhra Pradesh Odisha Rajasthan Kenya Morocco Pakistan Peru Philippines Overall de jure 36 58 60 52 52 54 45 n.a. 60 70 46  Regulatory governance 45 69 77 68 68 68 56 n.a. 72 83 48 Accountability 75 78 87 91 91 91 95 n.a. 79 85 95  Regulatory oversight 67 67 100 100 100 100 100 n.a. 67 67 100  Legal appeals 100 100 100 100 100 100 100 n.a. 100 100 100  Transparency 57 67 62 73 73 73 85 n.a. 70 89 85 Autonomy 60 89 89 75 75 75 59 n.a. 92 98 51  Decision-making autonomy 64 83 86 100 100 100 79 n.a. 92 92 79  Budgetary autonomy 88 97 94 50 50 50 94 n.a. 100 100 50  Leadership autonomy 88 75 75 50 50 50 63 n.a. 75 100 75  Managerial autonomy 0 100 100 100 100 100 0 n.a. 100 100 0  Regulatory substance 81 85 78 76 76 79 81 40 83 83 95 Tariff regulation 92 75 75 75 75 75 88 29 83 100 93  Regulatory framework for tariffs 100 100 100 100 100 100 100 33 100 100 86  Determination of tariffs 83 50 50 50 50 50 75 25 67 100 100 Quality regulation 100 79 83 83 83 92 54 33 100 100 92  Quality of service standards 100 75 100 100 100 100 75 50 100 100 100  Quality of service enforcement 100 83 67 67 67 83 33 17 100 100 83 Market entry regulation 50 100 75 71 71 71 100 58 67 50 100  Permitting new entrants 50 100 100 75 75 75 100 50 100 100 100  PPA approvals 50 n.a. 50 67 67 67 100 67 33 0 100 Source: Based on Rethinking Power Sector Reform utility database 2015. Note: n.a. = not applicable; PPA = power purchase agreement. 203 204 ANNEX 6B. PERCEIVED SCORES ON THE REGULATORY PERFORMANCE INDEX Percent Dominican Egypt, India – India – India – Indicators Colombia Republic Arab Rep. Andhra Pradesh Odisha Rajasthan Kenya Morocco Pakistan Peru Philippines Overall perceived 34 9 28 25 21 21 19 n.a. 37 66 39  Regulatory governance 42 26 56 64 64 64 38 n.a. 61 80 47 Accountability 70 37 77 88 88 88 71 n.a. 85 81 86  Regulatory oversight 67 67 100 100 100 100 67 n.a. 67 67 67  Legal appeals 100 0 100 100 100 100 100 n.a. 100 100 100  Transparency 43 44 31 64 64 64 46 n.a. 89 78 92 Autonomy 60 70 73 73 73 73 53 n.a. 72 98 54  Decision-making autonomy 64 33 86 92 92 92 57 n.a. 62 92 92  Budgetary autonomy 88 97 94 50 50 50 94 n.a. 100 100 50  Leadership autonomy 88 50 63 50 50 50 63 n.a. 75 100 75  Managerial autonomy 0 100 50 100 100 100 0 n.a. 50 100 0  Regulatory substance 81 36 50 39 33 33 51 38 61 83 83 Tariff regulation 92 17 67 50 50 50 58 21 83 100 92  Regulatory framework for tariffs 100 17 83 50 50 50 67 17 100 100 100  Determination of tariffs 83 17 50 50 50 50 50 25 67 100 83 Quality regulation 100 42 33 25 8 8 29 25 71 100 92  Quality of service standards 100 50 50 0 0 0 25 50 75 100 100  Quality of service enforcement 100 33 17 50 17 17 33 0 67 100 83 Market entry regulation 50 50 50 42 42 42 67 67 29 50 67  Permitting new entrants 50 50 100 50 50 50 100 67 25 100 100  PPA approvals 50 n.a. 0 33 33 33 33 67 33 0 33 Source: Based on Rethinking Power Sector Reform utility database 2015. Note: n.a. = not applicable; PPA = power purchase agreement. Did Countries Establish Meaningful Power Sector Regulation? 205 NOTES Cave, M. 2016. “Thoughts on UK Economic 1. This chapter draws on the background paper Regulation, 2016.” Oxera , March. https:// by Rodriguez Pardina and Schiro (2018) www.oxera.com/agenda/thoughts-on-uk​ and original research from a team led by -economic-regulation-2016/. Katharina Gassner and Joseph Kapika. Corneli, S., and S. Kihm. 2015. “Electric Industry Martin Rodriguez Pardina, Julieta Schiro, Structure and Regulatory Responses in a High and Kagaba Paul Mukibi were members of Distributed Energy Resources Future.” Future the team. The work program was coordi- Electric Utility Regulation Report No. 1, nated by Vivien Foster and Anshul Rana. Lawrence Berkeley National Laboratory, 2. Given this particularity, Morocco’s broad University of California. regulatory and governance scores were not Correa, P., C. Pereira, B. Mueller, and M. Melo. computed, nor was Morocco taken into 2006. Regulatory Governance in Infrastructure account when computing average scores for Industries: Assessment and Measurement of broad regulatory performance and gover- Brazilian Operators. Trends and Policy Options nance (and its subareas), or when correlating No. 3. Washington, DC: World Bank. coefficients involving governance or any of Foster, Vivien, Samantha Witte, Sudeshna Ghosh their subareas. Banerjee, and Alejandro Moreno. 2017. 3. These countries are Colombia, the Dominican “Charting the Diffusion of Power Sector Republic, Kenya, Pakistan, Peru, the Reforms across the Developing World.” Policy Philippines, Uganda, Ukraine, and Vietnam. Research Working Paper 8235, World Bank, 4. Various quality-of-service outcomes are mea- Washington, DC. sured to ensure that utilities do not decrease Gilardi, F., and M. Maggetti. 2011. “The Indepen- operating expenses by lowering their quality dence of Regulatory Authorities.” Chapter 14 of service. in Handbook on the Politics of Regulation, edited 5. Most of these savings are refunded to cus- by D. Levi-Faur. Cheltenham, UK: Edward tomers, though utilities can keep some of the Elgar Publishing. savings, which is a key incentive. Graffy, E., and S. Kihm. 2014. “Does Disruptive 6. Note that net metering is one compensation Competetion Mean a Death Spiral for Electric mechanism for rooftop solar. Other mecha- Utilities?” Energy Law Journal 35 (1): 1–44. nisms exist where consumers are not cred- Hanretty, C., and C. Koop. 2013. “Shall the ited at the retail rate at all, or are not credited Law Set Them Free? The Formal and Actual for any self-generation at all, but are instead Independence of Regulatory Agencies.” providing the entire output of rooftop solar Regulation and Governance 7 (2): 195–214. resources directly to the utility. Jamison, M. 2007. “Regulation: Rate of Return.” In Encyclopedia of Energy Engineering and Technology, Vol. 3, edited by B. Capehart, 1252–57. New REFERENCES York: CRC Press, Taylor and Francis. Adam, R. 2011. “Establishing Regulatory Joskow, P. 2014. “Incentive Regulation in Theory Incentives to Raise Service Quality in and Practice: Electricity Distribution and Electricity Networks.” Point of View, Cisco Transmission Networks.” In Economic Regulation Internet Business Solutions Group. and Its Reform: What Have We Learned? edited Alexander, I. 2014. “Developing Countries by N. L. Rose, 291–344. University of Chicago Experience and Outlook: Getting the Press. Framework Right.” Utilities Policy 31 (C): Kessides, I. 2012. “Electricity Reforms—What 184–87. Some Countries Did Right and Others Can Do Andrés, L., J. Guasch, M. Diop, and S. Lopez Better.” Viewpoint, Note No. 332, World Bank, Azumendi. 2007. Assessing the Governance Washington, DC. of Electricity Regulatory Agencies in the Latin Kihm, S., R. Lehr, S. Aggarwal, and E. Burgess. American and Caribbean Region: A Benchmarking 2015. “You Get What You Pay for: Moving Analysis. Washington, DC: World Bank. towards Value in Utility Compensation.” https:// Brown, A., J. Stern, B. Tenenbaum, and D. Gencer. www.seventhwave.org/sites/default/files/you​ 2006. Handbook for Evaluating Infrastructure -get-what-you-pay-for-part-one-2015.pdf. Regulatory Systems. Washington, DC: World Laffont, J.-J. 2005. Regulation and Development. Bank. Cambridge, UK: Cambridge University Press. 206 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Nixon, H. 2013. “RIIO Incentive Framework.” Review.” Policy Research Working Paper 8461, Presentation at the 10th EU-US Energy World Bank, Washington, DC. Regulators Roundtable, The Hague, April 8. Rodriguez Pardina, M., R. Schlirf Rapti, and Ofgem. 2015. RIIO-GD1 Annual Report 2013-14. E. Groom. 2008. Accounting for Infrastructure London: Ofgem. Regulation: An Introduction. Washington, DC: Pargal, S., and S. Banerjee. 2014. More Power to India: World Bank. The Challenge of Electricity Distribution. Directions Vagliasindi, M., and J. Nellis. 2010. “Building in Development Series. Washington, DC: Sound Institutions.” In Africa’s Infrastructure: World Bank. A time for Transformation, edited by Vivien Rodriguez Pardina, M. A., and J. Schiro. 2018. Foster and Cecilia Briceño-Garmendia. Africa “Taking Stock of Economic Regulation of Power Development Forum. Washington DC: World Utilities in the Developing World: A Literature Bank. What Progress Has Been Made with Wholesale Power Markets? 7 Guiding questions • What conditions must countries meet before attempting to create a wholesale power market? • What transition challenges have countries faced in introducing and fine-tuning a competitive market? • What are the emerging best-practice design features for wholesale power markets in the developing world? • What are the effects of the current wave of technological disruption on wholesale power markets? Summary • The rarity of wholesale power markets in the developing world reflects a demanding list of precon- ditions that must be met before competition becomes ­ viable. Until those preconditions are fully met, countries can benefit from wholesale competition by participating in regional power ­ markets. • There are significant risks of getting stuck in the transition to a competitive market. Once in place, constant monitoring is needed so that the market design can be fine-tuned as challenges arise. • An independent and adaptable institutional and regulatory framework is a great aid to market efficiency. The role of system operator can be undertaken by the transmission company or by an ­ independent entity; it may or may not be combined with the market operator ­ role. • Adequate system governance and open access are essential for operational efficiency and for attract- ing new ­ dispatch. entrants. Accurate short-term prices will follow from efficient and secure real-time ­ It is risky, however, to rely on short-term prices alone to incentivize investment in new capacity. Increasingly, incentives for new investment are being provided through the auctioning of long-term power supply contracts. • Looking forward, there is a need to modify market designs to accommodate and incentivize variable renewable energy, battery storage, and demand-side participation. 207 208 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION to increase efficiency of the power sector This chapter reviews the experience of the (Joskow 2008b). few developing countries that have man- The model notwithstanding, mixed experi- aged to implement wholesale power ence tells us that power markets need careful ­markets. 1 After asking why some countries assessment, design, and ­implementation. Over have made more progress than others in the past three decades, power markets have introducing competition, the chapter goes demonstrated their ability to improve perfor- on to examine the complex process of mance through an evolving mix of competi- designing and introducing a power m ­ arket. tion and regulation—for example, in Australia, The guiding questions are as ­ follows. What Chile, Norway, the United Kingdom, and the are the minimum conditions that countries United S ­ tates. However, California’s power must meet before attempting to introduce a market crisis in 2000 revealed the significant wholesale power market? What challenges risks of establishing power markets without have countries faced in introducing and careful design and implementation, even fine-tuning a competitive market? What are in jurisdictions with ample resources and the emerging best practices for wholesale s upportive conditions (Besant-Jones and ­ power markets in the developing world? Tenenbaum 2001). The core principles of Finally, how are wholesale power markets market design, widely agreed to be essential, ­ being affected by the current wave of tech- include open access to the grid, demand-side nological disruption? participation, coordination for short-term The creation of a wholesale power market ­ e fficiency and reliability, and a workable was the endpoint envisaged by the model of framework for supply adequacy (Hogan 2002; power sector reform that prevailed in the Hunt 2002; Joskow 2008b; Rudnick and 1990s. Most of the other reform measures— Velasquez 2018). including restructuring, privatization, and This chapter focuses primarily on wholesale, regulation—were considered stepping-stones ­ rather than retail, power m­ arkets. Wholesale toward full competition in the s ­ ector. According power markets, in which competition is found to the model, competitive forces would effi- in the generation segment, have gained signifi- ciently balance supply and demand by driving cant traction in the developing world owing to investment, operation, and consumption deci- their ability to deliver gains in efficiency and sions in both the short run and the long ­ run. In ­ reliability. Retail competition, entailing the the short run, power markets would improve introduction of competition in the commercial system operations by promoting efficient segment, has evolved slowly, reaching fewer scheduling and dispatch, ensuring reliability, jurisdictions and leading to mixed outcomes and providing appropriate price signals for (Defeuilley 2009; Littlechild 2009). For these operation and ­ investment. In the long run, reasons, the discussion will center on whole- adequately regulated power markets would sale markets for generation, even if some of the incentivize optimal investments at the right issues may carry over to transmission and retail locations and times, attaining desired levels of ­markets. supply security through an efficient mix of generation ­technologies. In contrast to a verti- KEY FINDINGS cally integrated electricity industry, power Relatively little is known about power markets markets shift risks of technology choice, con- in developing c ­ ountries. The literature on struction cost, and operating “mistakes” to sup- power markets in developed countries and on pliers and away from ­ c onsumers. Strong the wider reform process in developing coun- profit-maximizing incentives would thus work ­ extensive. The same cannot be said, tries is ­ What Progress Has Been Made with Wholesale Power Markets? 209 however, for the specific experiences of devel- Out of the 22 percent, only 7 percent has oping countries with the introduction of power also introduced retail c ­ ompetition. This markets. Those experiences are sparsely docu- ­ share compares with about 80 percent of mented (Rudnick and Velasquez 2018). As part Organisation for Economic Co-operation and of the Power Sector Reform Observatory Development (OECD) countries with estab- undertaken for this report, in-depth case stud- lished wholesale power markets, out of which ies were produced for the four countries with 66 percent has also introduced retail relatively mature wholesale power markets in c ompetition. Most of the wholesale power ­ place: Colombia, India, Peru, and the markets in the developing world are in two Philippines (Rudnick and Velasquez 2019a, regions—Latin America and the Caribbean 2019b, 2019c, and ­forthcoming). Drawing on and Europe and Central Asia—where about this new body of knowledge, the developing half of the countries have introduced ­ them. country experience with power markets can be By contrast, not one wholesale power market conveyed through the following key ­ findings. can be found in Africa and the Middle ­ East. Uptake of power markets in the developing size. Just world is strongly related to system ­ Finding #1: The rarity of wholesale 5 percent of countries with systems under power markets in the developing world 1 gigawatt (GW) have them, compared with reflects the demanding preconditions 25 percent of countries with systems above for viable c ­ ompetition 20 ­G W. Nevertheless, many of the largest Only one in five developing countries has power systems in the developing world— established a wholesale power ­ market. The such as those of the Arab Republic of Egypt, share rose gradually from 11 percent in 2000 Indonesia, Pakistan, and South Africa—have to just over 20 percent by 2015 (figure 7.1). not yet introduced wholesale power m ­ arkets. FIGURE 7.1 Only one in five developing countries has established a wholesale market Competition in the power sector, 1995–2015 100 90 32 80 36 50 47 70 Share of countries (%) 67 60 20 17 50 13 14 40 20 25 30 7 23 24 6 20 5 23 3 1 15 10 10 10 9 2 2 5 7 7 0 95 96 97 8 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 9 20 20 20 20 20 20 20 19 19 19 19 19 20 20 20 20 20 20 20 20 20 Retail market competition Wholesale market competition Bilateral contracting Single-buyer model Independent power producers Monopoly Source: Foster and others 2017. 210 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD The preconditions needed to establish a markets (such as those of Chile and the United wholesale power market are relatively Kingdom) started out with highly concen- demanding and not met in much of the devel- trated generation segments; decreasing con- oping ­ world. Starting conditions matter for centration before the market opens can help power m ­ arkets. Those conditions span a wide deter undesirable exertions of market power range, including market structure, transmis- and the resulting public ­ backlash. For exam- sion constraints, financial health, and the ple, Argentina dealt with market power by wider enabling e ­ nvironment. A few subopti- breaking up the generation sector at the out- mal conditions can be addressed by market set of reforms, to a point where none of the design in some very specific situations, and the many generators controlled more than a sin- market must certainly be carefully tailored to plant. Finally, if a country is short of gen- gle ­ local limitations and imperfect starting eration capacity, then all available plants will ­ conditions. Nevertheless, power markets are need to be deployed to meet the system ­ peak. easier to implement and more likely to succeed Without some plant redundancy, competition if starting conditions are supportive of is meaningless, and market power will swell competition. For countries where the risks of ­ during the peak ­ period. developing a market significantly exceed those Second, transmission constraints should be of the status quo, markets are better intro- removed, because they can create temporary duced gradually and only after appropriate concentrations of market ­ power. Adequate attained. conditions are ­ transmission infrastructure is needed to avoid First and foremost, the market structure of congestion, which can create market power on the generation segment needs to be conducive the other side of the bottleneck and prevent competition. The benefits of competition to ­ optimization of available energy resources are unlikely to pass through to end-users if across large a ­ reas.2 Conversely, connecting pre- there is a significant concentration of market viously independent power grids can dramati- power in generation, or if certain actors con- cally increase c ­ ompetition. For example, the trol both generation and transmission assets previously independent wholesale markets of and can strategically manipulate access to the Chile began to integrate in 2017, boosting ­ grid. Accommodating competition under such competition in supply auctions and improving circumstances entails not only vertical unbun- system ­ operations. Interconnection of power dling of generation and transmission but also grids was also successful in the Philippines, horizontal unbundling of generation to make where the Visayas island and its power grid certain that the market includes enough have been integrated with the Luzon market players to ensure competitive ­ ­ pressure. (Five since 2010, thereby lessening the horizontal generators of roughly equivalent size are concentration of the generation segment in the often considered to be a ­ minimum.) Creation combined power ­ grid. The Indian power sys- of a competitive generation segment also tem also went through a major transformation requires careful attention to how the assets starting in the early 1990s, when interconnec- are broken up across players, so as to ensure tion among the states and regions was very that each generation company controls weak. The process concluded in late 2013, ­ roughly equal shares of the total capacity, par- when the southern region synchronized with ticularly during the critical price-setting peak the rest of the nation’s ­grids.3 period, and that none of them can exercise Third, the power sector must be financially market power in the relevant upstream fuel healthy and underpinned by a reliable pay- ­ m arkets. Although some successful power ment ­ chain. Financial ill-health anywhere in What Progress Has Been Made with Wholesale Power Markets? 211 the electricity supply chain lessens investors’ conditions supportive of ­ c ompetition. trust in the market, preventing it from deliver- California’s power market emerged from a ing investment and competition among political reform process that brought many ­ generators. Market participants must have distortions to the market, such as obliging dis- confidence that counterparties will pay for the tribution utilities to supply their customers power they ­ sell. This confidence hinges on the directly from the wholesale market and pre- integrity of payments across the power supply venting them from hedging against wholesale chain, the presence of creditworthy buyers of spot price ­volatility. (The reform required dis- electricity, and cost-recovery tariffs that ensure tributors to sell their generation assets and did the financial viability of the industry as a not allow them to sign bilateral ­ contracts.) whole. The case of Ukraine highlights the chal- ­ These and other market design aspects, along lenges of establishing a power market under with increased demand and diminished weak financial c ­ onditions. Despite consider- hydropower availability in the summer of able scale (25 GW peak demand), attempts to 2000, resulted in acute price spikes (reflecting introduce a wholesale market in Ukraine some abuse of market power) and mandated during the 1990s failed partly because of poor rolling blackouts, severely harming California’s collection from end-users and low levels of economy and end-users (Besant-Jones and cash payments among companies (Besant- Tenenbaum 2001; Wolak 2003­ b). Jones 2006; Krishnaswamy 1999). Other power markets have flourished even Fourth, wider country conditions also mat- under challenging conditions, such as those ter for power markets, including institutions that prevail in ­India. India’s power exchanges and the macroeconomic, political, and social deliver benefits of competition at least to a environment. A lack of government commit- ­ portion of suppliers and customers, despite ­ ment to reform and signals of unwarranted ­ underlying conditions that are far from i ­deal. regulatory intervention or instability can also Indeed, inadequate generation falling short of limit investors’ interest in the m ­ arket. demand persisted in India until 2017, and the Protection of property rights was crucial to the transmission grid remains weak, constraining successful Chilean ­ reform. Argentina, by con- the regional power s ­ ystem. Moreover, the trast, illustrates the negative effects of macro- Indian power sector has many financial prob- economic crises and political interference in lems, with continuing cross-subsidies, tariffs electricity pricing (Pollitt 2004, 2008). The too low to recover costs, and even bankrupt experience of Eastern European countries indi- utilities. Nevertheless, large open-access cus- ­ cates that it is extremely difficult to carry out tomers and other market participants have structural reforms of the sector during eco- benefitted from increased participation in nomic turmoil (Krishnaswamy and Stuggins power e ­ xchanges. Trade in Indian power 2003). Finally, permitting, siting, and social exchanges surged from 1,735 gigawatt-hours approval are persistent obstacles to investment (GWh) in fiscal year 2008 (FY2008) to 41,120 in new generation and transmission infrastruc- GWh in FY2017, while prices fell 83 percent in ture across Asia and Latin ­ America. real terms over the same period as generation Nonetheless, power markets have failed capacity grew to exceed d ­ emand. Delivering even under very advantageous conditions, as the benefits of competition to a wider set of in the ­U.S. state of ­ California. The state’s elec- customers, however, requires tackling the tricity crisis in the summer of 2000 illustrates ­ fundamental problems of the sector, which how power markets can fail even in a jurisdic- include inadequate infrastructure and financial tion with plenty of resources and initial ­nonviability. 212 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Finding #2: Small countries can demand patterns across broad a ­ reas. For benefit from wholesale competition example, demand peaks in the countries of by participating in regional power the South Asia region do not coincide over ­markets the year, implying huge potential savings on the costs of supply and of unserved For small power systems, a competitive market demand—savings derived in both cases from for generation may be neither feasible, owing resource ­ pooling. These benefits have been to size constraints, nor desirable, owing to loss estimated at more than US$9 billion per year of scale e ­ conomies. Caution should be exer- (Timilsina and others 2015). Moreover, cised in moving toward a wholesale power regional power markets can help reduce market if the system falls below certain critical carbon emissions, both by displacing domes- ­ size thresholds (box 7.1). Underdeveloped tic thermal generation with imported renew- power systems in countries with rapidly grow- able generation and by creating larger bal- ing populations and economies—such as many ancing areas that facilitate the integration in Sub-Saharan Africa—may soon reach the of variable renewable energy resources scale where a domestic power market becomes (Chattopadhyay and Fernando 2011; Raineri ­ relevant. In other cases, small scale may be a and others 2013; Wijayatunga, Chattopadhyay, permanent situation dictated by geographic and Fernando 2015. ­constraints. To deliver these benefits, however, suc- Countries that are too small to develop cessful regional power markets require sig- wholesale power markets often have the nificant up-front investments to build infra- option of participating in regional power structure and an institutional framework markets. Small countries for which a domestic ­ capable of delivering efficient outcomes wholesale power market is not viable may be (Oseni and Pollitt 2016). Connecting inde- close to larger countries or regional power pendent power grids presents challenges markets that could allow them to reap the ben- ranging from the purely technical to the efits of competition and increased security of financial and ­ i nstitutional. The challenges ­ supply. Working examples of regional power are worth addressing, because the market markets exist in Africa (Southern Africa Power resulting from integration can rapidly inten- Pool and West African Power Pool) and Central sify ­c ompetition. Sufficient transmission America (Mercado Eléctrico Regional, MER); capacity, both in interconnectors and in the the benefits of electricity integration have also domestic grid of each country, must be been realized in South Asia through India’s financed and developed to enable cross-bor- interconnections with neighboring markets in der ­trading. Institutional challenges include Bangladesh and Nepal (Andrews-Speed 2016; merging independent operators to form a Asian Development Bank 2015; Oseni and single regional one, regulating and oversee- Pollitt 2016). ing cross-border trade, and developing effi- The economic and supply-security benefits cient regional trade arrangements beyond of regional integration and cross-border trade bilateral long-term c ­ ontracts. Such trade can be enormous, for countries large and arrangements require a regional market small. Regional markets can improve resource ­ operator to support short-term trading and optimization across countries by pooling settlement, complemented by financial con- ­ generation resources, sharing reserves, and tracts that allow for risk management while harnessing the seasonal and hourly comple- preserving dispatch efficiency (Rose, Stoner, mentarity of generation availability and of and Pérez-Arriaga 2016). What Progress Has Been Made with Wholesale Power Markets? 213 BOX 7.1 How big must a power system be to support a wholesale power market? The answer to this important policy question depends on both physical and financial ­ considerations. The thresholds indicated here are notional and intended solely as a ­ guide. They are not intended to suggest that power markets are impossible in smaller systems, but the balance between the costs and the benefits of introducing a wholesale power market varies with ­ scale. First, for competition to be meaningful, at least five generation companies should be present in the m ­ arket. Each of them must be large enough to achieve scale e ­ conomies. If the number of companies is too low, there is a collusion. If their size is too small, there is a loss of efficiency in ­ risk of ­ production. Given a minimum efficient scale of 400–1,000 megawatts for thermal generation, and the need for at least five companies, this suggests a minimum system size of 3 gigawatts of peak demand or 20 terawatt-hours of annual energy demand to support a competitive market. The reality may be more nuanced than this simple rule of thumb, because the threshold will also be affected ­ by the extent to which competition exists specifically at the system peak, the conditions of access to fuel supply, and the extent of vertical integration between generation and ­ distribution. Second, the investment and operation costs of a wholesale power market are substantial, and the efficiency gains from competition are proportional to the size of the ­ market. Although the evidence on costs is limited, some figures may be ­ illustrative. Even in Singapore’s small market, the costs of establishing the wholesale power market exceeded US$75 million, and annual running costs are US$15 million–20 million (Ching 2014). For a larger market, such as the United Kingdom, the cost of switching from the power pool market design to the New Electricity Trading Arrangements was estimated to be more than US$1 billion (spread over a five-year period), with annual operating costs approaching US$50 million (Newbery 2005). Ofgem (1999) notes that the up-front investment cost would be justified if the change caused prices to drop by 10 ­ p ercent. The levelized cost of operation of a typical market operator has been estimated at US$0.20/megawatt-hour for a large market like the Pennsylvania–New Jersey–Maryland interconnection or PJM (in the eastern United States) to as much as US$1.00/megawatt-hour for smaller jurisdictions like Ontario, ­ C anada. a Taken in conjunction, this evidence suggests that a country would need to exceed a potential market trading value of US$1 billion before the investment in a market trading platform could be justified; otherwise the value of the potential efficiency gains would be unlikely to surpass the investment and operating costs of the ­ market. Of the 15 countries in the Power Sector Reform Observatory, all of those with functioning wholesale power markets comfortably meet the size thresholds, except the Dominican R ­ epublic. Several countries meet the size thresholds but do not yet have a functioning wholesale power ­ m arket. Of this group, several are in transition toward such a market: the Arab Republic of Egypt by 2023, Ukraine by 2025, and Vietnam by 2024 (table B7.1.1). TABLE B7.1.1 Classification of Observatory countries, by size threshold Meets thresholds Does not meet thresholds Wholesale power market Colombia, India, Peru, Philippines Dominican Republic No wholesale power market Arab Republic of Egypt, Morocco, Kenya, Senegal, Tajikistan, Pakistan, Ukraine, Vietnam Tanzania, Uganda http://www.ieso.ca/corporate-ieso/regulatory-accountability/usage-fees; for PJM, ­ a. For Canada, see, for example, ­ http:// www.pjm.com/-/media/committees-groups/committees/fc/postings/first-quarter-2018-schedule-9 ­-rates.ashx?la=en. Finding #3: A transition toward a Philippines, five years elapsed between the competitive market is required, but passage of legislation for a power market and there are significant risks of becoming its entry into ­ operation. In the case of India, process stuck in the ­ the delay was caused by the absence of a The transition to a competitive market has clear road map for ­ i mplementation. In the taken a variety of f ­orms. Colombia and Peru Philippines, the transition period was delib- opted for immediate implementation of erately aimed at easing in the new arrange- power markets without a transitional phase ments and giving market actors time to (table 7.1). By contrast, in India and the ­adapt. 214 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 7.1 Immediate or transitional implementation of a power market is possible; each path has benefits and challenges   Peru Colombia India Philippines Year transition began a 1992 1994 2003 2001 Year market began 1992 1995 2008 2006 Transition process None None Law established Requirement for principles of distribution utilities reform, with no to source at least clear transition 10 percent of their roadmap power supply from the spot market for first five years Target type of market Wholesale (cost- Wholesale (bid-based) Bilateral, with Wholesale (bid-based) based) competition; competition; regulated power exchanges and retail competition regulated capacity capacity payments payments Source: World Bank elaboration based on data provided by local consultants and independent r ­ esearch. a. The transition is considered to begin with the enactment of the law mandating the future development of a competitive wholesale power ­market. There is no doubt that significant time and over the details of the market during its defi- resources are required to develop and estab- nition and ­ implementation. lish the m­ arket. Once the legal framework Transition processes can create a risk that governing the power sector reform is reforms may never be ­ c ompleted. Several enacted, detailed bylaws, market rules, and countries have opted for a trial or pilot period procedures must be developed; discussed of a few months to allow market participants to with key stakeholders; and a ­ pproved. This prepare for actual market ­ operations. The dan- process can easily take two years or longer if ger of such periods is that they may slow the the tasks are not prioritized and managed momentum of reform or divert it from its orig- p roperly. Moreover, the market requires ­ inal ­objectives. For example, the California development of information systems; optimi- electricity market crisis of 2000 was used as a zation models for planning, scheduling, dis- pretext to stall market reforms not only in the patch, and pricing; and billing and settlement rest of the United States but also in other parts procedures. Even when clear principles have ­ of the world, including Malaysia, which settled been agreed upon, the minutiae of models on a single-buyer market ­ instead. and processes can affect the incomes and The single-buyer model is a risky transition costs of each power company and can measure because rigid contracts with indepen- ­ t herefore be c­ ontentious. Moreover, many dent power producers can deter participation ­ practical details of system operation are not in a subsequent competitive m ­ arket. In a pure formally codified through procedures, but single-buyer model only the integrated rather left to common practice (for example, monopoly is permitted to buy power (includ- criteria for real-time redispatch by the sys- ing energy, capacity, and ancillary services) tem ­operator). Therefore, appropriate time from competing generators or from indepen- and resources should be allocated when lay- dent power producers at regulated prices ing out the timetable for market implemen- (Arizu, Gencer, and Maurer 2006; Hunt 2002). tation, considering the potential for conflicts It is considered by some to be a second-best What Progress Has Been Made with Wholesale Power Markets? 215 alternative to comprehensive restructuring, to incumbent retailers and generators when providing time for a smooth transition the electricity industry is disaggregated, but toward fully competitive wholesale markets before asset divestiture or privatization (Kee (Vagliasindi and Besant-Jones 2013). The 2001). They reduce the incentive of generators single-buyer model can help alleviate supply ­ to bid strategically (and so exercise market shortages by introducing independent power power) by hedging revenues from the spot producers to an industry with limited financ- ­ price. Moreover, vesting contracts can help ing ­capabilities. However, the inflexible long- market development during the initial phases, term power purchase agreements associated when the spot market may not provide enough with the model, usually built around take-or- revenue (or enough revenue certainty) for pay clauses, deter the evolution of competi- generators or ­ retailers. tion, because plants operating under the agree- Cost-based power pools can also be adopted ments may have no incentive to participate in as a transitional measure to deter the exercise a competitive market and cannot be dispatched of market p ­ ower. Cost-based pools are based on a merit-order basis to minimize short-run on the “audited” variable costs of each power production ­ costs. Moreover, high-priced con- plant, whereas more sophisticated markets tracts could be undercut by competition in the allow participants to submit b ­ ids. Bid-based wholesale market, becoming “stranded costs” markets thus allow generation firms to reveal that require a recovery ­ mechanism. Although their full opportunity costs, while also leaving careful design can mitigate some of its prob- room for gaming directly through the submit- lems (Arizu, Gencer, and Maurer 2006), the ted ­bids. Compared to bid-based markets, cost- single-buyer model remains problematic based pools are exposed to fewer opportunities because it transfers risks from generators to for market power abuse and the dramatic price end-users, thus removing incentives for gener- spikes that may result from gaming the mar- ators to manage those risks (Castalia 2013) ket, especially in the presence of transmission and possibly leading to inefficient investment congestion or tight supply–demand c ­ onditions. decisions (Thomas 2012). For example, the Pennsylvania-New Jersey- Incipient power markets remain vulnera- Maryland (PJM) market in the United States ble to abuses of market power even when was administered as a transitional cost-based structural precautions have been t ­aken. The pool during its first year of operation, and physical and economic properties of electricity even now the independent system operator make its markets prone to unilateral requires bids to be cost-based in cases where ­ exercise—and abuse—of market ­ power. Such transmission congestion creates risks of local- abuse can severely harm customers, as well as ized market power (Wolak 2003­ a). the reliability and efficiency of the overall Using cost-based pools as a transitional market. The initial stages of a new market can ­ structure risks leaving the country with incom- be particularly vulnerable to abuse because of plete reforms, unless the target market is also concentration in the ownership of the mar- cost-based. Although cost-based pools can be ­ ginal generation resources that set the market useful transitional measures (as in PJM), they price and because of residual congestion in are not exempt from short and long-term inef- transmission or a tight balance of supply and ficiencies (Munoz and others 2017). In the ­demand. short run, generators can game the parameters Vesting contracts are one device for contain- of the cost-based pool (such as the startup ing market power during the transition p ­ eriod. times or minimum load levels) to maximize Vesting contracts are hedge contracts assigned ­ profits. Also, the audited fuel cost may be very 216 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD different from the true opportunity cost, which Jamasb 2012). A strong, independent, and would reflect take-or-pay clauses in natural effective regulator is an important actor in any gas contracts, as well as the availability of gas wholesale power market (Joskow and storage and a secondary gas ­ market. In the Schmalensee 1983). A tailored process for long run, the generation technology can be monitoring and oversight is needed from the strategically selected by firms to maximize start of reforms; it must be able to account for profits. Hence, initially adopting a cost-based ­ the technical and economic complexities of the market instead of a bid-based one poses the power sector, including the physical laws gov- risk of miring the country in an inefficient erning power flows in the transmission system transition. The Republic of Korea, for example, ­ and the lack of economic large-scale storage has been stuck with a transitional cost-based (Bushnell, Mansur, and Saravia 2008). pool for nearly two decades, failing to evolve The monitoring and oversight process toward the envisioned bid-based market (Kim, should be capable of assessing market out- Kim, and Shin 2013). comes, proposing enhancements to market Nevertheless, in some situations cost-based design, and providing a base for detection of pools may be the preferred long-term market abuse of market p ­ ower. The oversight process design—particularly where market power should assess the performance of submarkets remains concentrated or where the system for energy, capacity, and ancillary ­ services. depends heavily on ­ hydropower. This was the Large sets of market indicators are calculated case in several hydro-dependent Latin and published regularly in some countries American countries (Brazil, Chile, and Peru), (Chile, India, and Peru); other countries ana- where cost-based pools provided a means of lyze the behavior of market participants mitigating short-term market power and (Colombia and the Philippines). Given the hydro-thermal coordination (Rudnick, Varela, complexities of the power sector, indicators are and Hogan 1997). Cost-based power pools in not enough by themselves, but should be used Brazil and Chile have been effective in man- in more detailed analyses to provide meaning- aging multiyear cascading hydrological reser- ful conclusions for regulators, policy makers, voirs, realizing economies from coordination antitrust authorities, and market participants of hydropower resources across broad areas (Stoft 2002). Market monitoring in developing and of hydro and thermal generation p ­ lants. countries often falls short in assessing perfor- These coordination economies could be diffi- mance in terms of outcomes such as security of cult to attain under a more decentralized trad- supply and competitiveness (table 7.2). ing arrangement where each company inde- Adequate monitoring and oversight will pendently managed its own reservoir or occur only if they are formally required and power ­plant. purposefully ­ organized. The definition and allocation of monitoring and oversight ­ functions must be clearly established from the Finding #4: An independent and outset of ­reforms. Depending on the country, adaptable institutional and regulatory those functions might involve the regulator, framework is a great aid to market the ministry, the antitrust authorities, and the ­efficiency system ­ operator. The system or market opera- Power markets require regulation and over- tor can be very effective in compiling and pub- sight by independent and effective institutions lishing performance i ­ndicators. Because these (Jamasb, Nepal, and Timilsina 2015; Jamasb, indicators require deeper analysis before they Newbery, and Pollitt 2005; Nepal and can be meaningfully interpreted, some What Progress Has Been Made with Wholesale Power Markets? 217 TABLE 7.2 A country-specific process for monitoring and overseeing the electricity market is required from the start of reforms India Philippines Colombia Peru Chile Market monitoring Structural Structural and Structural and Market oversight No dedicated entity for approach monitoring by behavioral analysis behavioral monitoring by regulator market monitoring until 2016 regulator (CERC) by regulator, and by regulator (CREG) (OSINERGMIN) law introduced a dedicated independent entity monitoring entity within within LTSO system operator Major reform 2014: 2015: Regulations 2006: Regulated 2006: Centralized 2004–05: Centralized adaptations Improvements in on competitive capacity mechanism auctions to auctions for supplying mechanism for selection procedure is replaced by the supply regulated regulated customers, and balancing the for distribution Firm Energy Market customers; centralized transmission grid (deviation utilities 2009: government improved expansion planning settlement) begins driving independence of technology-specific system operator development of generation (hydro and gas) ­ esearch. Source: World Bank elaboration based on data provided by local consultants and independent r Note: CERC = Central Electricity Regulatory Commission; CREG = Commission for the Regulation of Electricity and Gas; LTSO = legally unbundled transmission system operator; OSINERGMIN = Supervisory Agency for Investment in ­ Mining. jurisdictions have created an independent markets. The entry of new wholesale electricity ­ monitoring entity reporting to the board of the competitors, coupled with wider market inte- system ­ operator. This approach has been suc- gration, has been more effective at lowering cessfully applied in PJM and other power mar- concentration and increasing c ­ ompetition. As kets in the United States; it has also been revealed by the energy concentration index adopted in the ­ Philippines. In the European (Herfhindahl-Hirschmann Index, HHI), 4 the Union, the market monitoring function is per- entry of new competitors gradually drove formed by the Agency for the Cooperation of down concentration in Peru, and the intercon- Energy Regulators (ACER), a regulatory body nection of the two major power grids in the for the entire union that collects electricity Philippines brought about a sharp decline in trading data and refers cases of market manip- concentration in 2010 (figure 7.2). In general, ulation back to the national regulatory the abuse of market power is easier to detect in ­ agencies. Elsewhere, simple reporting is con- bid-based markets, where it manifests itself in ducted by the system operator or the regulator the form of sudden price spikes, than in more (Chile, India, and Peru), along with case-by- tightly regulated cost-based markets; however, case analysis of specific incidents of abuse of cost-based pricing can be selectively introduced market power and of potential market reforms into bid-based markets as a strategy for manag- Colombia). (as in ­ ing market power in the presence of localized A persistent challenge for electricity markets transmission ­constraints. is preventing abuses of market ­ power. Several A balance must be sought between manag- developing countries have introduced power ing the risks of sustained exertions of market markets despite concentration in generation power and restrictive regulations that could (for example, Chile and the ­ Philippines). In cripple the market (Wolak 2005). Market inter- these situations, simple limits on ownership vention (and even suspension) may be justi- and concentration have proven ineffective at fied, for example, if capacity is withheld from deterring the exercise of market power in risks. the market, leading to excessive reliability ­ 218 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 7.2 Addressing market power is a critical component surged during the first months of market of establishing efficient power markets operation; initial analyses found that prices Evolution of market concentration index for generation had been altered by collusion among PSALM’s 4,000 trading ­ t eams. A series of exchanges and 3,500 appeals then ensued, and a case is still pend- Herfhindahl-Hirschmann Index 3,000 ing in the Supreme Court of the Philippines (Abrenica 2009; Roxas and Santiago 2010). 2,500 This conflict illustrates the difficulties of intro- 2,000 ducing a power market where generation 1,500 remains highly concentrated, as well as the 1,000 practical ineffectiveness of many simple miti- 500 gation ­measures. The second conflict occurred when exoge- 0 nous factors led to a period of tight ­ supply. 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 Several large plants were affected by mainte- Philippines Colombia Peru nance activities and forced outages in late Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. 2013, creating tight supply conditions that caused prices to surge (DOE 2014). To prevent further price spikes, the regulator lowered the The intervention must follow a rational and cap on the spot market price and later imposed predictable process, however, ideally specified an even lower cap to be triggered in the event in the form of rules and ­procedures. There is of any sustained period of high ­ p rices. In always a danger that provisions for market response, the major distribution utility, intervention will be used for political purposes, Meralco, attempted to pass these price spikes such as reducing the volatility of spot prices if on to its retail customers to offset its own they are perceived to be too high to be passed ­ exposure. The Supreme Court temporarily through to ­ households. Intervention and sus- restrained Meralco’s price increase while the pension of power markets have been limited to regulator conducted a probe, which found that very specific circumstances in the observatory generators had wielded market power by with- countries; they have generally worked well but holding ­ capacity. The probe led to a suit that is are not exempt from ­ controversy. also still pending in the Supreme C ­ ourt. This Two conflicts in the Philippines’ bid-based case illustrates that, although regulatory inter- power market illustrate the difficulties of deal- ventions (such as price caps) may be effective ing with market power under tight supply in the short-term, they are no substitute for ­ conditions. The first stemmed from a concen- tackling underlying causes—in this case, insuf- tration of market power in the early days of ficient competition in the spot market, plus the ­market. The market was introduced in lack of contracting to supply regulated custom- 2001, with the generation sector still highly ers in the retail m ­ arket. The moral is that, concentrated. At the time, the government ­ although emergency interventions may occa- corporation created to manage privatization of sionally be necessary, more permanent market generation assets, Power Sector Assets and adaptations are required to deal with the Liabilities Management Corporation (PSALM), underlying issues that trigger price spikes and managed 54 percent of installed capacity reliability ­problems. through four trading teams that were required Power markets require regulation to be flex- to act i­ndependently. However, spot prices ible enough to adapt to market changes and What Progress Has Been Made with Wholesale Power Markets? 219 public policy o­ bjectives. The market must cor- electricity, centralized, real-time coordination rect problems and adapt to evolving supply and is required to keep supply and demand in bal- demand conditions by creating new markets or ance; the necessary coordination is made pos- enhancing existing ones, while maintaining sible through the planned and r ­eal-time the confidence of investors by avoiding ex post operation of the transmission system (Hogan adjustments of market rules for purely political 1998­ a). Open and nondiscriminatory access ­ reasons. The observatory countries provide to the transmission network by wholesale plenty of examples of how market designs sellers and buyers is required for efficient pro- have gradually been improved through market duction and ­ e xchange. Moreover, because monitoring efforts (see table 7.2). Several Latin transmission is key for effective wholesale American countries introduced supply auctions competition, transmission operations should for regulated customers as a means of attract- be effectively independent from market par- ing investment in baseload generating c ­ apacity. ticipants such as generation companies, retail- This market-based approach has been success- ers, and distribution utilities (ESMAP 2013). ful in preventing costly rationing during dry Making transmission independent from the hydrological periods, thus reducing the appe- rest of market participants requires restric- tite for direct intervention in the m ­ arket. tions on cross-ownership (Arizu, Dunn, and Evolving public policy objectives also require Tenenbaum 2001). Once structural measures market adaptations, but these should harness are in place, open access should be enshrined the benefits of competition as a means of effi- in law rather than left to bilateral private ciently attaining the desired ­ outcomes. As an negotiations. Separate contracting should be ­ example, India introduced the policy objective allowed for energy and network services (to of increasing renewable energy investment, enable multibuyer, multiseller competition), and this was initially incentivized through and independent ­ system operators should be ­ feed-in-tariffs. Although effective, these have established (Joskow 2008a; Newbery 2002; been gradually replaced by market-based Rudnick and Velasquez 2018). instruments such as renewable auctions and Successful power markets also require an renewable energy credits that can be traded adequate governance structure, especially for through power ­ exchanges. Peru’s energy policy the system o ­ perator. The governance of the objective was to diversify hydrological risk market defines how decisions are made and through greater investment in gas-fired power ­enforced. Effective governance leads to smooth generation. The wholesale power market was ­ and continuous improvements in market rules pushed in the desired direction through the and procedures, and in their practical imple- use of technology-specific auctions and social- mentation, enhancing the benefits of ized market levies to finance the necessary nat- ­ competition. System and market operators ural gas ­ infrastructure. play a central role in any power market, coor- dinating dispatch, determining pricing, and conducting settlement arrangements in the Finding #5: Adequate system wholesale ­ market. Good governance arrange- governance and effective open access ments for the operators enable active involve- are key to ensuring operational ­ nsuring ment by all market participants, while e entrants efficiency and attracting new ­ that no single interest group dominates deci- Open access to the transmission grid is a pre- making. Broader elements of market gov- sion ­ condition for competition in the electricity ernance include expedited dispute resolution, ­ industry. Because of the unique properties of as well as regulatory review and approval of 220 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD procedures (Barker, Tenenbaum, and Woolf market operator with no ownership of trans- 1997). mission assets is not strictly necessary, as long To ensure open access and efficient market as adequate governance and oversight operation, different approaches to transmission arrangements are in p ­ lace. For example, ownership and operation have been adopted in Colombia has a successful system and market the industrialized countries of OECD (Pollitt operator owned by the major transmission 2012). On the one hand, several markets in the company, Interconexion Electrica (ISA) United States have been organized around an (table 7.3). The operator’s functional indepen- independent system operator,5 which operates dence is ensured by appropriate arrangements; the transmission system and the wholesale however, effective transmission system opera- markets without owning any transmission tors have been more difficult to implement in assets. Markets in Europe, on the other hand, ­ other emerging ­ markets. In the case of India, have allowed transmission system operators to the authority of state utilities that are often combine both ownership and operation of vertically integrated and financially distressed ­ transmission. Incentives for cost control have hinders open access and regional market been cited as one of the advantages of trans- integration. State operators have denied open ­ mission system operators, which have been access to large customers, probably to avoid relatively successful in Europe, and particularly financial losses for the state’s distribution in the United Kingdom (Pollitt 2012). Overall, ­ utility. These experiences highlight the impor- however, both integrated and separated mod- tance of both the general structure of the els appear to function effectively in the OECD power sector and the detailed rules governing ­context. open access in enabling the success of power In the developing country context, some markets; the experiences suggest that separa- evidence suggests that integrated transmission tion may be preferable, because it is less prone ownership and operation can pose institu- to concerns that the operator is discriminating tional c ­ hallenges. In principle, unbundling against or in favor of third parties through transmission ownership from operation by opaque and complex operational criteria establishing an independent system and (Arizu, Dunn, and Tenenbaum 2001). TABLE 7.3 Overview of power market governance across developing countries Design element India Philippines Colombia Peru Coordination of 1 national and 5 regional SOs Centralized merit-order, Fully centralized merit-order Fully centralized merit- operations coordinate state SOs though many contracts order are physical System operator Government-owned TSOs at LTSO: Market operator not LTSO: System/market ISO: Private not-for-profit the regional and state ­level. independent (it is chaired operator is functionally organization, independent Often integrated utility at the by the Department of independent, and subsidiary from market participants state level ­Energy). of major government-owned (owns no transmission Transco ­ISA. assets) Open access Regulated, administered by ­Regulated. Distributors ­Regulated. Regulated, established (regulated / SOs. Some problems due ­ may have some conflicts by ­law. negotiated) to conflicts of interest with of interest due to vertical state ­utilities. ­integration. Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: ISO = independent system operator; LTSO = Legally unbundled transmission system operator; SO = system operator; TSO = transmission system ­operator. What Progress Has Been Made with Wholesale Power Markets? 221 Beyond unbundling of transmission own- short-term markets in providing these signals ership from operation, further separation of is important, even though new-generation system and market operation may be orga- capacity in developing countries is largely nized depending on the chosen market d ­ esign. hedged through long-term contracts or other In a centralized model, the functions of mar- mechanisms. Thus, pricing, dispatch, trans- ­ ket and system operation might be assigned to mission capacity allocation, and congestion different coordinated ­ entities. For example, in management are among key components for the Philippines, the Wholesale Electricity Spot efficient short-term wholesale ­ m arkets. Market (WESM) is the independent market Additionally, ancillary services and price vola- operator, and Transco the transmission system tility have become the focus of many market ­ o perator. Other countries with centralized design discussions, given increasing penetra- markets, such as Colombia and Peru, estab- tion of variable renewable energy such as lished an integrated system and market wind and ­ solar. operator. On the basis of their experience, the ­ The ideal pricing mechanism for wholesale strength of governance arrangements seems short-term power markets involves high spatial more important than the separation of system and temporal r ­ esolution. A temporal resolution and market operation per ­ se. For less central- of one hour or less (for example, 5–15 minutes) ized markets, such as India’s, power exchanges is ­ required. The greater the temporal resolution, can be organized to enable voluntary the easier it becomes to integrate variable exchange of standardized products; however, renewable energy and other new technologies voluntary participation probably requires a (IEA 2016). Spatially, a price should be defined large system to ensure sufficient liquidity in for each transmission node (so-called locational the power e ­ xchange. Overall, surveyed expe- marginal prices), reflecting the physical proper- rience does not suggest a preponderance of ties of the transmission network, including either benefits or risks from separating market losses and ­ congestion. Nodal pricing enhances and system ­ operations. It does, however, rein- efficiency and transparency of the market by force the general principle that centralized identifying and managing transmission bottle- coordination is inevitable (whether among necks (Hogan 1998b); it has been implemented multiple operators and balancing authorities, in Peru and the Philippines (table 7.4). Simpler or between the system and market operator) pricing options, such as zonal prices (for exam- and that ultimate operational authority ple, power exchanges with market-splitting in should be given to the (transmission) system Europe and India) or even systemwide prices operator to preserve supply ­ reliability. (for example, Colombia) are also a common choice in both developed and developing ­ countries. Given the pervasiveness of transmis- Finding #6: Efficient, security-oriented sion congestion in these markets, the potential real-time dispatch is critical to getting benefits of nodal pricing and better congestion short-term prices r­ ight management are significant (McRae and Wolak Short-term markets should be designed to ful- 2016; Neuhoff and others 2013; Ryan 2014). fill their primary function of facilitating effi- Moreover, simplicity is not necessarily an cient, liquid, and transparent decisions about advantage of zonal or system ­ prices. Indeed, the dispatch and a ­ djustment. Short-term markets system spot price of the Colombian market is must also produce the right signals to inform arguably more complex and less t ­ransparent. and incentivize investment, with minimum The Colombian market is based on a uni- resort to distorting ­interventions. The role of form-price, bid-based auction, with a single 222 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 7.4 Comparison of short-term power market design across developing countries Design element India Philippines Colombia Peru Centralized or Mostly bilateral; with Centralized; partially Centralized Centralized bilateral market centralized power exchanges bilateral Cost or bid- based Bid in power exchanges Bid-based Bid-based Cost-based dispatch Market for reserves Nascent regional ancillary Reserves are cooptimized Regulation reserves — and ancillary services market aimed at with energy scheduling, at optimized before energy, services restoring frequency, relieving prices agreed by the SO. based on energy-bids by congestion ­Gencos. Mechanism for Priority stack for allocating Nodal spot prices paid Single-node price Nodal spot prices and pricing and transmission capacity to generators, while with up-lift for startup centralized system congestion (with priority for long-term customers pay a zonal and shutdown ­ costs. operation (with power and management contracts). Market-splitting in ­price. ­ Congestion constraints gas constraints suppressed power exchanges dispatch. settle against real ­ from spot prices 2009–16). Settlement Multisettlement in power Single-settlement Single-settlement Single-settlement approach exchanges Source: Based on Rethinking Power Sector Reform utility database 2015. Note: Gencos = generation companies; SO = system operator; — = not applicable. system price that ignores transmission conges- process should support competition and reli- tion and an ideal schedule that disregards the ability, and market outcomes should be as transmission s ­ ystem. A parallel, technically fea- transparent as ­possible. Centralized coordina- sible schedule is also ­ determined. Differences tion of system operations is ultimately neces- between the ideal schedule and the technically sary for reliability, owing to the complexities feasible are settled financially using a reconcilia- of the physical interactions that take place in tion price for each generator, different from the the transmission g ­ rid. The degree of centraliza- spot price, thus resulting in a more complex dis- tion of system decisions varies, however, patch and pricing process than the nodal pricing across power ­ markets. In the fully centralized ­alternative. pool model, bids are submitted to a market Distortions in pricing and dispatch should be run by a market operator that, through an avoided to the extent ­ possible. Direct regula- optimization process known as security-­ tory interventions to deal with undesired mar- constrained economic dispatch, determines ket outcomes should be used as an emergency the market-clearing and technically feasible measure ­ only. Expedients that significantly dis- schedule of production for each power plant to tort economics, such as artificially low price reliably and economically supply d ­ emand. In caps and disregard for the transmission system, the power exchange model, by contrast, mar- are best a ­ voided. For example, price caps ket agents can sign bilateral contracts and then should be high enough to encourage declare their production or consumption ­ investment. The transmission system should be schedules directly to the system operator expanded to reduce congestion and accommo- before gate closure (such as one hour before date a market in liquid contracts (whether real-time operation) or can submit bids to the bilateral or organized futures) as a means of trading platform of the power ­ exchange. After hedging risks and stabilizing retail ­ prices. gate closure, the system operator takes control Scheduling and dispatch can be centralized to perform balancing of real-time operations or d ­ ecentralized. The scheduling and dispatch (Batlle 2013). In either case, procedures and What Progress Has Been Made with Wholesale Power Markets? 223 outcomes should be publicly available in a decentralized short-term power markets have timely manner to all participants and inter- struggled, however, to deliver economic dis- ested participants, including available trans- patch across wide geographic ­ areas. mission capacities, prices, and schedules—all To improve performance of the power sec- without disclosure of commercially sensitive tor, the wholesale market should be complete proprietary ­information. and liquid, with competition covering a signif- Centralized markets based on security-­ icant amount of ­ supply. Competition requires constrained economic dispatch have proven completeness, that is, a full set of forward and successful in delivering transparent market spot markets, in addition to risk-management outcomes reflective of underlying demand and tools (Hunt 2002). Liquidity is particularly supply. That record of success supports the ­ important in voluntary bilateral markets, view that centralized security-constrained eco- where not enough customers or suppliers may nomic dispatch is preferable for wholesale elec- be willing to p­ articipate. Centralized manda- tricity markets (Hogan 2002; Newbery 2005; tory markets such as power pools, by contrast, Rudnick, Varela, and Hogan 1997). Among key are inherently l ­iquid. Organized futures advantages of centralized power pools is the ­ markets have been established only in India, tight integration of operational and economic where power exchanges emerged in 2008 considerations across energy, transmission, and with day-ahead markets, which later incorpo- ancillary services markets to determine the rated week-ahead and renewable energy scheduling and dispatch of power ­ plants. In credit ­m arkets. However, inflexible power theory, such integration enables higher pro- purchase agreements in India relegate the role ductive efficiency (Sioshansi, Oren, and O’Neill of power exchanges to a bare minimum, thus 2008). These advantages are borne out by the limiting the efficiency gains that competition successful experiences of Colombia, Peru, and can provide in both operation and ­ investment. the Philippines, which have all successfully In other developing countries, only single-­ employed some form of centralized short-term settlement centralized m ­ arkets have been markets for dispatch (table 7.4). organized, primarily with day-ahead schedul- Although decentralized bilateral markets ing and monthly ­ settlement. with power exchanges offer some practical Over the past decade, wholesale prices in advantages over centralized pools, they also the observatory countries have come to reflect present significant ­ c hallenges. The main market conditions, regulation, and central advantages of decentralized markets are that planning. There is significant interannual price ­ they are easy to set up and they allow for sim- variation across power markets, with varia- pler trading (resembling market-clearing in tions of three to four times between high-price other commodity ­ markets). This simplicity and low-price years (figure 7.3). The highest comes at the expense of reduced integration sustained annual prices observed are approach- of energy trading and transmission manage- ing US$200 per megawatt-hour (MWh), ment (Wilson 1998). Bilateral markets with whereas the lowest are less than US$20­ /MWh. power exchanges have worked well in Among the key market conditions driving Germany and the Nordic ­ countries. They have prices are the entry of new generation capacity also allowed for some limited efficiency gains (apparent in the downward price trend in in India, which has a bilateral market with a India), fuel availability and prices in interna- regulated mechanism for regional balancing tional and domestic markets, and hydrological complemented by power exchanges function- conditions (especially for Colombia’s hydro-­ ing as organized futures m ­ arkets. India’s dominated ­ system). Evolving regulation is also 224 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 7.3 Significant interannual price variations across capacity and high reliability, many developing power markets in the observatory countries countries pursue power sector reforms to 200 address shortages of generation capacity or to 180 lower the financial burden of large invest- 160 ments in ­ generation. Prices (2015 US$/MWh) 140 Developing countries must not fail to imple- 120 ment mechanisms to ensure the stable reve- 100 nues that investors require before financing 80 ­ infrastructure. There have been few cases of 60 merchant power plants built spontaneously by 40 the private sector at the owner’s risk and with 20 0 full exposure to spot ­ prices. Rather, the state has played an important role, ranging from 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 direct investment in India to indicative plan- India Philippines Colombia Peru ning in ­ C olombia. New capacity additions Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. across developing countries have almost Note: MWh = ­ megawatt-hour. always been driven by medium- to long-term contracts (with maturities ranging from a few visible in Colombian prices, owing to the intro- years to 20 or more ­ years). These contracts, duction of a firm energy market, which con- freely agreed among market participants, can tributed to containing price spikes during the be physical (entailing physical and cash 2009–10 El Niño drought but failed to do so ­ delivery) or purely financial (entailing only during 2015–16 because of design weaknesses cash delivery) (Batlle 2013). In many cases, and unexpected ­ events. Central planning and they are accompanied by special mechanisms government direction of markets are reflected designed to incentivize capacity additions in price trends for India and Peru, where prices (table 7.5). The design of such mechanisms is fell consistently, partly because of surplus challenging and can make a material differ- capacity that resulted from aggressive govern- ence to investment ­ d ecisions. In addition, ment-led investments in g ­ eneration. These investment outcomes also depend on the trends underscore the close interaction of q uality of institutional governance in the ­ short-term efficiency in pricing and dispatch power sector, and on the security of fuel sup- with the longer-term development of the plies and the efficiency of short-term ­ markets. ­market. Regulated capacity payments have proven effective in eliciting investment in g­ eneration— sometimes to the point of creating excess Finding #7: Drawing new entrants capacity. Regulated capacity payments are in into the market requires effective risk place in Chile and Peru; they were also part of management, including the use of the initial approach to ensure adequate capac- ­contracts ity that was adopted and later abandoned in Incentivizing the optimal generation mix and ­ C olombia. These payments, based on the attracting adequate levels of investment is one theory of peak-load pricing (Boiteux 1960), ­ of the most important roles of power markets provide power plants an income based on a in fast-growing developing c ­ountries. regulated price that reflects the regulator’s Whereas developed countries introduced view of the marginal cost of efficient generat- reforms primarily under conditions of surplus ing technology to provide peak power (often What Progress Has Been Made with Wholesale Power Markets? 225 TABLE 7.5 Comparison of mechanisms for generation investment across developing countries Design element India Philippines Colombia Peru Contracts and Physical bilateral market Physical bilateral Financial bilateral contracts Financial bilateral contracts bilateral markets for long and medium contracts (financial term, traders for accounting in WESM) short term Capacity markets None None Firm-energy payments Regulated capacity payments (for generation available during poor hydrology) Auctions Renewable auctions Auctions for privatization Centralized auctions Distribution utility supply of generation assets (by transmission system auctions for regulated and power purchase operator) for firm energy customers (including agreements technology-specific auctions) Investment Primarily driven by Responsibility and Generators bear full risks Generators bear full investment responsibility and government planning risks left entirely to and responsibility of risk in the core market d ­ esign. risk allocation and private ­ investment. private participants investment, incentivized Market has been adjusted to States hold a competitive (government is by short-term power incentivize gas, hydro, and bidding process when a forbidden from investing market and firm-energy other renewable generation demand–supply deficit is or underwriting new ­ payments. (However, costs (extra costs levied directly on ­projected. generation ­capacity). were passed through to ­end-users). customers during the event.) 2015–16 El Niño ­ Renewables Renewable energy Renewable portfolio No specific mechanism in Renewable auctions, with price credits traded in power standards, plus feed-in- place premium if spot price falls exchanges tariffs below auction price Renewable purchase obligations, feed-in- tariffs, and renewable auctions Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: WESM = Wholesale Electricity Spot ­ Market. diesel-fired plants), without necessarily estab- Although in theory an energy-only market lishing a physical target for security of s ­ upply. could provide adequate generating capacity, Regulated capacity mechanisms, which are in practice—at least in Europe and the reflective of the supply side only, are effective United States—a series of market failures (such in attracting peaking capacity, but they often as low price caps on the spot price) have moti- elicit capacity that is perceived as ­ excessive. For vated the development of capacity markets example, the regulated capacity payment in (Newbery 2015). The Philippines has an ener- Chile has been called into question by some gy-only market that has suffered from tight market participants, given continuous capacity capacity margins and spot price volatility, lead- additions in oil-fired peaking power plants, ing to concerns about the adequacy of the despite an apparent shortage of generating ­ system. Colombia has developed a parallel ­ capacity. In this context, some aspects of the market that focuses on firm energy rather administrative procedure for calculating the than capacity per ­ s e. Colombia is highly capacity payments will be revisited by the reg- dependent on hydropower generation, with ulator in the course of 2019. variable availability that can greatly diminish Capacity markets are designed specifically during dry ­seasons. Therefore, meeting peak to ensure that enough generating capacity demand is not the primary issue, as it is in sys- is available to meet demand ( ­adequacy). tems based on thermal ­ generation. Instead, 226 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD the challenge is to ensure adequate energy coal, oil, sugarcane biomass, solar, and wind, supplies during periods of low hydrological as well as cross-border imports (Rudnick and ­ availability. The Colombian firm-energy mar- Velasquez 2018). They can also be purposely ket procures firm energy through auctions, to targeted at specific technologies, as is often ensure availability during such periods of scar- the case when they are used to accelerate city (defined as periods when the spot price uptake of renewable ­ energy. Even without exceeds a regulated strike price called the scar- legal mandates, auctions may still be encour- city ­ price). Generators receive a monthly pay- aged as a means of competitive procurement, ment for firm energy according to the results for example, by facilitating aggregation of of the auctions, in exchange for the obligation smaller customers (for example, in Chile and to supply energy under conditions of ­ scarcity. the ­Philippines). Unfortunately, during the 2015–16 drought, Market outlooks and indicative planning the efficacy of the firm-energy market was of capacity expansion have a role in guiding undermined by unexpected events, including private investment and market p ­ articipants. forced generation outages, reduced fuel avail- The use of capacity markets or auction mech- ability, and falling international oil ­ p rices. anisms requires the government to deter- These events left many generators with oper- mine desired volumes of investment based ating costs that exceeded the regulated scarcity on indicative generation ­ planning. This is the price, leaving them unwilling to provide the case of the firm-energy market in Colombia contracted supplies and necessitating regula- and the capacity markets in the PJM area of tory intervention of the scarcity price to pre- the eastern United S ­ tates. Centralized plan- vent supply ­ shortfalls. ning of transmission capacity has also been Auctions for awarding supply contracts chosen in some markets (for example, Chile) have been successful in fostering competition as the mandatory planning p ­ rocess. Other in contracts markets and in attracting countries provide more comprehensive indic- ­ i nvestment. Various auction mechanisms ative planning for the power and energy have been successfully introduced across the s ector (for example, ­ ­ C olombia). Such out- developing ­ world. For developing countries looks can also help to orient potential new in which financial markets are underdevel- entrants seeking investment opportunities in oped and electricity futures markets are complex ­markets. absent or illiquid, auctions can provide a The evolution of the reserve margin over workable means of harnessing competitive time provides some indication of success in forces (Maurer and Barroso 2011). Peru attracting adequate new investment into gen- accomplished this in 2004, followed by India eration (figure 7.4). The observatory countries in 2006; and the Philippines is currently con- faced different situations with regard to reserve sidering auction ­ a rrangements. These auc- margins. India’s achievement was to reverse its ­ tions can be required by law, for example, for chronic power deficit to reach overall balance, the supply of regulated customers by distri- albeit without any significant reserve m ­ argin. bution utilities years in advance of supply to The Philippines began its wholesale market allow participation of new power ­ plants. That with high reserve margins resulting from is the case in Brazil and Chile, where supply capacity procured under the independent auctions have been of great interest to inter- power producer program of the 1990s, but national investors looking to enter these those margins have been shrinking, in part ­ m arkets. Auctions are often technology-­ because of greater efficiency, but also reflecting neutral, encompassing hydropower, gas, the weaker investment incentives associated What Progress Has Been Made with Wholesale Power Markets? 227 with an energy-only market, compounded by FIGURE 7.4 Power markets have helped countries attract the application of price caps, insufficient com- adequate investment into generation petition in the contracts market, and barriers 100 entry. For Colombia and Peru, reserve mar- to ­ gins are much higher and growing, reflecting 80 the need to diversify hydro-based systems Reserve margin (%) 60 toward natural gas so as to ensure firm energy during ­droughts. 40 It is important to strike a balance between the revenue stability provided by long-term 20 contracts and the need to retain flexibility to adapt to emerging market conditions and 0 ­ opportunities. Contracts of very long maturity –20 may be needed to cope with serious capacity 05 06 07 08 09 10 11 12 13 14 15 16 17 20 shortages, making a wholesale market infeasi- 20 20 20 20 20 20 20 20 20 20 20 20 ble, or to incentivize investment in rapidly India Philippines Colombia Peru growing economies with country-specific Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. risks. A large share of inflexible contracts of ­ Note: The reserve margin is calculated as available generating capacity divided by very long duration (10 or 20 years), however, the system peak demand (maximum hourly metered value for each year), without correcting for low hydropower ­ availability. The reserve margin for the Philippines can curb the ability of the market to adapt to is for the Luzon grid through 2009; from 2010, the Luzon and Visayas grids are changing conditions, such as lower-than-­ ­considered. expected demand growth or technological disruptions like cheaper generation technolo- FIGURE 7.5 Inflexible contracts of long duration can limit a gies (undercutting higher-cost ­ contracts). In market’s ability to adapt to changing conditions some developing country markets, a large Relative size of contract market as share of traded energy share of power is traded under rigid long-term 100 contracts that predate the establishment of a wholesale power market (figure 7.5). When 80 Share of traded energy (%) these contracts expire, power plants can enter into more-competitive contracts or participate 60 in competitive capacity m ­ arkets. Moreover, uncontracted capacity can participate in 40 power exchanges (as in India) or directly in the wholesale market (as in the Philippines), 20 depending on the risk appetite of individual market ­ participants. Competition in contracts 0 is also important for centralized markets such 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 as Peru’s, where dispatch and spot pricing are India Philippines Colombia Peru independent from contractual positions Source: World Bank elaboration based on Rethinking Power Sector Reform utility (which by design fully cover actual ­ demand). database 2015. Variable renewable energy can make a valuable contribution to the overall flexibility opportunity for developing countries, given of the system, if allowed to compete alongside their short lead times and lower scales com- other ­ technologies. In achieving flexibility, pared with large coal-fired and hydropower wind and solar power plants present a unique plants (whether fed by pumped water or 228 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD large ­ r eservoirs). Beyond benefiting from conditions; and (3) greater flexibility in the their contributions to the achievement of pol- power grid’s ­ operation. Organizing demand icy targets for low-carbon investments (for participation requires that distribution utilities example, renewable auctions or renewable and large customers participate in all key sec- portfolio standards), the market can also allow tor processes, such as discussions of market renewables to compete alongside other tech- rules and procedures, regulatory processes, nologies for the supply of regular energy and and legal r ­ eforms. Representatives of major capacity markets, including bilateral contracts consumers also need to be formally incorpo- and centralized supply ­ auctions. In Peru, reg- rated into the governance structure of system ulation effectively bars renewables from par- and market operators, which has not always ticipating in auctions for regulated customers, been the case to ­ date. though discussions are underway to adapt the Deepening demand participation is often a procedure for calculating capacity credits to gradual process that begins with the largest recognize renewables and so allow them customers. The usual starting point is industrial ­ to compete for ­ contracts. to liberate the largest industrial consumers to Renewable generation in developing coun- seek their own power supply directly from tries has been growing faster over the past generators, which creates an important market decade. Growth has been driven by targeted ­ feedback loop and can provide an attractive incentives, support mechanisms, and declining commercial opportunity for new market investment costs for wind and solar photovol- entrants into g ­ eneration. In many countries, taic ­ power. In India, the plunge in solar invest- the size threshold above which customers are ment costs has motivated more competitive freed from the captive or regulated market has market-based forces such as renewable auc- gradually d­ ropped. In the Philippines, custom- tions, instead of the previously established ers above 1 megawatt (MW) have been eligible ­ feed-in-tariffs. In Chile, renewables were ini- since 2013. In Colombia and Peru, where tially incentivized through a quota l ­aw. The reform processes date further back, the thresh- required share of renewable generation has old has progressively fallen to the ­ c urrent been exceeded because of falling investment values of 0.1 MW and 0.2 MW, r ­ ­ espectively. ­costs. Deepening the retail market can increase contestability by lowering barriers to entry, ­ enabling more customers to contract with new Finding #8: Demand-side participation, generators. It is therefore important to allow ­ especially of large customers, is critical regulatory flexibility to revise downward the to several aspects of market efficiency thresholds for customer choice according to and system ­reliability evolving market ­ conditions. Demand-side participation in power markets Demand aggregation is an emerging institu- can greatly improve their ­ p erformance. tional model that allows smaller customers to So-called demand participation has altered a harness the benefits of c ­ ompetition. For exam- previously one-sided market focused on gen- ple, the Philippines is implementing a competi- eration and supply, providing benefits in tive supply procedure after successful aggrega- terms of efficiency and ­ reliability. Key benefits tion experiences with smaller cooperatives that include (1) gradual replacement of price regu- resulted in lower ­ prices. In Brazil and Chile, a lation with market-based forces; (2) limita- few load aggregators have also succeeded by tions on the extent of market power in organizing private auctions for power supply short-term markets, especially during tight from load reductions across a cluster of What Progress Has Been Made with Wholesale Power Markets? 229 ­ ustomers. This experience highlights the need c model works well when all units, including for regulation that allows such innovations and demand response, are dispatchable and have mechanisms wherever market conditions positive marginal cost; variable renewables, make them relevant and ­ feasible. however, are not always dispatchable and Demand participation can also be an typically have zero marginal c ­ osts.6 This cre- important mechanism for efficiently achieving ates problems in terms of providing adequate supply reliability, and the operational flexibil- incentives for investment in new plants and ity required to integrate high levels of variable may even push existing (mainly thermal) renewable ­ energy. Growing shares of wind resources out of the market, raising concerns and solar energy emphasize the need for over the potential for excessive decommis- resources and mechanisms capable of accom- sioning of thermal power plants and associ- modating sudden shifts in production (Strbac ated reliability risks, as well as the financial 2008). Demand response can potentially play challenges associated with the stranding of this r ­ ole. For example, Philippine’s largest dis- ­assets. tribution utility developed a scheme of inter- These challenges associated with variable ruptible load demand to be used under condi- renewable energy further exacerbate long- tions of tight supply as declared by the system standing concerns regarding the adequacy of ­ operator. Such demand response programs investment incentives arising from short-run help to avoid the high economic and political marginal price ­ signals. Longstanding experi- costs associated with rationing, as long as they ence with traditional power markets domi- are voluntary and properly c ­ ompensated. nated by thermal resources has shown that Colombia implemented an emergency demand relying on purely market-based incentives for penalization program for household customers investment in new capacity entails allowing among the packet of measures adopted in the spot prices to spike at high enough levels for wake of the 2015–16 El Niño ­ c risis. These long enough ­ periods. The very high price levels cases illustrate the large potential benefits of that result can be challenging to sustain for demand participation, especially for develop- wider political and economic reasons, to the ing countries such as Colombia and the point that in some competitive markets regula- Philippines, which are periodically subject to tors have set caps that stop prices from going tight supply ­conditions. above some maximum acceptable level, but at the same time prevent investors from receiving LOOKING AHEAD an adequate level of ­ remuneration. Even when The growing penetration of variable renew- complex mechanisms exist to allow prices to able energy is challenging the established par- escalate, grid operators must sometimes take adigm for price-setting in wholesale power actions outside of the market context in order markets. Competitive wholesale markets are ­ to protect system ­ reliability. In both cases, such designed to send a price signal that reflects the measures mute the price signals needed in the marginal cost of the last unit brought on to short-term to ensure longer-term fixed cost meet consumer demand (that is, the short- ­recovery.7 This has led to the development of a run marginal cost of ­ g eneration). In these variety of parallel mechanisms to guide invest- markets, a complex model ensures that gener- ment decisions, such as capacity charges, ators bidding to supply the market are selected capacity markets, and supply ­ auctions. on the basis of least cost, while meeting all In addition, the rise of variable renewable system c ­ onstraints. The last unit selected to energy creates the need for a different com- meet the system demand sets the p ­ rice. This position of investment geared toward flexible 230 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD ­ eneration. Grid operators trying to integrate g operating parameters such as ramping capa- renewables need fast-ramping, flexible bilities (IEA 2016). Although tight pool mod- resources to come online when the sun is not els have worked reasonably well in practice, shining or the wind is not ­ blowing. Whereas there is room for improvement in developing flexible resources such as flexible generators, countries, particularly through cooptimiza- storage, interconnections to neighboring tion of reserves and energy, because the val- grids, and demand response can accommo- ues of energy and reserves are mutually date these variations to some extent, inflexi- dependent (MIT 2016). Chile, for example, is ble generation resources, such as nuclear and expected to introduce auctions for the provi- older coal power plants, are less helpful in sion of some ancillary services by 2020. responding rapidly to changes in net l ­oad. If Bilateral markets, conversely, rely on real- the price is increasingly set by renewables, time balancing services coordinated by the then there is no consistent energy price signal system operator, which penalize participants to incentivize the development of flexible for imbalances in their scheduled generation resources. For this reason, a key current con- ­ and ­ demand. India, for example, is develop- cern is to ensure that ancillary services (such ing markets for centralized scheduling of reg- as operating reserves, frequency regulation, ulation services at the regional level, although and so on) are sufficiently priced to incentiv- the outcomes of these reforms have yet to be ize fast-ramping, flexible ­ resources. 8 These seen (Mukhopadhyay and others 2016). essential reliability services would have to be In any case, both centralized and bilateral priced very high, however, to make up for markets can benefit from integration of non- the zero (or very low) energy price seen for generation resources to provide flexibility and much of the d ­ ay. There is growing concern other ancillary s ­ ervices. Demand response is by that, in power systems made up mostly of far the most underused flexibility resource renewable resources with zero marginal across developing ­ countries. Ancillary services prices, a market framework that incentivizes markets or mechanisms should allow investment decisions through price signals ­ market-based participation of large customers based on short-run marginal costs is unlikely and aggregators of smaller customers, because to ensure the correct mix of resources needed even moderate amounts of customer participa- to keep the lights on (see, for example, Sen tion can help to balance the grid, facilitating 2014; Tierney 2018). resources. the integration of variable generation ­ For this reason, grid systems are beginning Furthermore, distributed resources such as air to consider the introduction of new markets conditioners, refrigerators, electric vehicles, in ancillary ­services. Some are considering the rooftop solar photovoltaic, and household bat- introduction of specific ramping products or teries can and should participate with larger holding reserves that match the energy lim- generation resources in providing the ancillary itations of resources (for example, when the services that the system operator requires (MIT sun or wind is suddenly u ­ navailable.) In addi- 2016).9 Along with sound institutions, solid tion, in centralized power markets, reserves regulatory frameworks, and adequate eco- are integrated to some extent with the nomic incentives, demand participation could ­ s cheduling and dispatch ­ p rocesses. These flourish and help avoid politically costly optimization-based processes performed by ­ r ationing. Developing countries, however, ­ the system operator can directly accommo- have been slow to adopt the new technologies date complex bids (including start-up costs in that enable demand response and related adap- addition to variable costs) and technical tations by wholesale ­ markets. What Progress Has Been Made with Wholesale Power Markets? 231 The current wave of technological advances defined by a set of characteristics including has changed the relative costs of generation point of delivery, load curve, power quality, technologies and of energy ­ storage. Modern reliability, and ­ carbon-intensity. For example, forms of renewable energy, notably wind and the integration of renewable energy could be solar, have seen a dramatic decline in costs facilitated by contracts with providers of driven by economies of scale in equipment fast-ramping resources or other essential reli- manufacturing, technological breakthroughs, ability services, such as operating reserves, reg- and learning; they now compete with—and in ulation, voltage support, and black start some cases outcompete—conventional power, ­ capability. Contract designs will matter even even without subsidies and before environ- more than previously, because inflexible take- mental costs are factored i ­n. At the same time, or-pay contracts will make it more challenging a variety of storage options is becoming to integrate variable renewable e ­ nergy. In ­ available to compensate for the variable nature addition, contracts should be designed to pay of modern ­ renewables. Beyond the longstand- for ancillary ­ services. ing location-specific option of pumped storage, One of the challenges with planning for thermal storage with molten salts is becoming the least-cost mix of resources, given all of available in areas with adequate solar radiation these technological changes, is that traditional for concentrated solar power, to name just one cost-comparison mechanisms may no longer new storage ­ technology. Although electro- be ­relevant. For example, the concept of lev- chemical storage in batteries is still economic elized cost of energy (LCOE)—the expected only for small-scale or short-term storage, costs costs over the lifetime of a particular genera- are falling rapidly, and further technological tion asset per unit of energy produced— breakthroughs are a ­ nticipated. In the future, assumes that resources are dispatchable and the batteries of electric vehicles may also able to produce as needed during peak ­ hours. become storage sites that can supply grids Neither of these assumptions holds for solar when the vehicle is not n ­ eeded. Compressed photovoltaic or ­ wind. Power produced mostly air and flywheels could provide additional at night (for example, wind) is not as valuable forms of energy s ­ torage. Finally, there is the when the greatest demand occurs during the storage potential of “power-to-gas,” for exam- ­day.10 Solar resources are likely to produce ple, converting renewable electricity to be during the day but may not operate consis- stored as hydrogen or other ­ gases. tently, and both solar and wind require the Given such rapid technological change, availability of flexible ­ resources. In addition, markets can be useful to enable price discov- location matters for solar and wind more than ery of the most efficient resources to meet spe- for other t ­ echnologies. Some sites are better; cific system ­ needs. Even in contexts where if wind and solar are not sited at those loca- competitive wholesale generation markets tions, their overall output will be reduced or may not be applicable, market-like approaches c urtailed. Ignoring these key aspects of ­ could still bring considerable ­ value. In small renewables makes the LCOE comparison, on systems, this might take the form of a single its own, i ­nsufficient. Unless output restric- buyer conducting cost-based dispatch and tions are considered, renewable resources publishing information on the system’s mar- could be overvalued when compared to ginal c­ osts. Competitive auctions to procure a dispatchable resources (Joskow 2011). The ­ variety of resources could be just as h ­ elpful. levelized avoided cost of electricity (LACE) Auctions could be designed to procure not a attempts to consider the costs of integrating specific technology but a service requirement new technologies into an existing resource 232 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD mix, given their expected ­ utilization. It is a and wholesale power sales of at least US$1 much more ­ c omplex analysis, because it billion. Nevertheless, there are growing oppor- ­ requires an assessment of how a given project tunities for countries with smaller power sys- would operate within an existing grid ­ system. tems to harness some of the benefits of trade When combined with the LCOE concept, it by participating in regional power ­ markets. may offer a better indication of how benefits Creation of a wholesale power market is a exceed ­costs.11 If other resources are displaced complex process that may take some time and when new resources are added to the grid requires careful design ­ choices. Countries have system, the LACE concept can capture some adopted various transitional strategies, such as of the potential costs associated with this single-buyer markets, vesting contracts, and ­displacement.12 cost-based p ­ ools. These strategies may play Ultimately, the goal is to enable a power sys- useful roles, but they introduce the risk of bog- tem that integrates various disruptive technol- ging down the ­ transition. Once a market is ogies and ensures that an appropriate mix of operational, careful monitoring and regulatory resources will be available to grid ­ operators. oversight are needed to detect and act on signs Markets may be particularly helpful in the of abuse of market power, as well as to adapt transition to a low-carbon electricity s ­ ector. the market design to emerging c ­ onditions. Competition can also help protect against the System operators, which play a critical role in power of vested interests seeking to protect coordinating actions across the market, may be conventional fuel supply and generation tech- autonomous or housed within the transmis- nologies that may no longer be ­ cost-effective. sion utility—either way, they must ensure Certainly, markets can be used to incentivize broad-based stakeholder r ­epresentation. generation or load-management solutions Efficient security-constrained real-time dis- that meet system needs, but how those mar- patch is a critical requirement for correct kets will be created and enabled is an ongoing short-term ­ prices. Energy-only markets have ­discussion. struggled to attract adequate investment in future generation, and the investment deficit CONCLUSIONS needs to be separately handled through long- Developing country power markets have the term contracts complemented by some form of potential to improve the efficiency and reli- capacity mechanism or supply a ­ uctions. ability of electricity generation, once countries Enabling demand-side participation in the are ready to move to this stage; however, sig- power market significantly adds to its effi- nificant preconditions must be met before ciency and ­flexibility. countries are ready to move toward a whole- Experience shows that adhering to general sale ­ market. Those preconditions may include principles of market design tends to deliver structural reforms, financial health, a support- improvements in efficiency and ­ r eliability. ive regulatory environment, and resolution Conversely, deviations from sound design prin- of major generation and transmission ciples imperil future ­ performance. For exam- c onstraints. Full vertical and horizontal ­ ple, various degrees of centralized planning in unbundling of the sector is required, together India and Peru have proven successful in with significant investments in a market plat- attaining public policy objectives, but at the form, none of which is likely to make much expense of higher costs being passed through sense until the power system reaches a thresh- to ­end-users. Looking ahead, power systems old equivalent to 3 GW of peak demand, are bound to be transformed by disruptive 20 ­terawatt-​hours of annual energy demand, ­ technologies such as solar- and wind-fueled What Progress Has Been Made with Wholesale Power Markets? 233 generation, electric vehicles, and widespread on the market share (in terms of g ­ enerated distributed g ­ eneration. In this context, power energy) of each ­ participant. Although the markets can be an effective enabler for the average energy HHI inadequately reflects the extent of market power and its poten- transition of developing countries toward more tial impact in prices (which could be much sustainable, efficient, and reliable power higher during outages of generating facil- ­systems. ities), it does provide a useful s ­ignal for The wave of technological disruption the overall evolution of competition in the currently affecting the electricity sector only ­ power ­market. 5. Although in the United States the model of underscores the need for sophisticated market the independent system operator has been ­ d esign. The rapidly changing landscape of widely adopted, operators are also known as electricity services, with technological change regional transmission ­ operators. However, in driving down the costs of variable renewable practice they follow the same general princi- energy and battery storage, increases the ples as independent system operators. value of a wholesale power market that can 6. In addition, many renewable resources are often compensated outside of the provide continuous price ­ information. Market wholesale market model through various rules will also need to be adapted to remuner- ­ subsidies. This makes their already low ate the ancillary services that are so critical to bids even lower, because they would be the integration of variable renewable energy willing to produce up to the point where and to provide appropriate economic incen- they lost money, and is part of the reason why negative pricing has begun to appear tives for the use of ­ batteries. Moreover, the in competitive wholesale markets in the need to design wholesale power markets to United ­ States. A lack of transmission is also fully accommodate demand-side participation a reason for the negative pricing, because is becoming increasingly important as the evo- when transmission is insufficient, only lution of smart grids facilitates ever-more-­ generation within a particular area can sophisticated ways of aggregating responses to meet demand in the same area, effectively bottling ­renewables. ­demand. 7. Essentially, these actions mute the scarcity price signal often cited as the key mechanism NOTES to incentivize flexible ­resources. 1. This chapter draws on Rudnick and Velasquez 8. In countries that have them, there is also (2018, 2019a, 2019b, 2019c, and forthcom- a discussion about whether capacity mar- ing). Further original research was conducted kets should focus on procuring resources by a team led by Debabrata Chattopadhyay with specific characteristics, or whether the and comprising Hugh Rudnick, Constantin energy and ancillary services markets alone Velasquez, Martin Schroder, and Tatyana can produce price signals consistent with sys- Kramskaya. The work program was coordi- ­ tem ­needs. nated by Vivien Foster and Anshul ­ Rana. 9. This is a complicated regulatory challenge 2. Given the physical laws governing power in many of the advanced markets, because flows, transmission constraints can occur the transmission (wholesale) market system between subsystems owing to internal con- is separated from the distribution (retail) straints in the transmission network of a ­market ­system. subsystem, even with adequate transmission 10. This could change, for example, with the links between ­subsystems. advent of widespread night-time charging of 3. See, for example, One-Nation-One-Grid electric ­vehicles. announcement by Powergrid India ( ­https:// 11. However, neither LCOE or LACE consider www.powergridindia.com/one-nation-one​ environmental regulations or policy decisions -grid) and also Ryan (2017). that also influence project development. 4. The Herfhindahl-Hirschmann Index (HHI) 12. Or, benefits, especially if environmental or presented here measures concentration based policy goals are taken into ­account. 234 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD REFERENCES Boiteux, ­ M . 1960. “Peak-Load ­ P ricing.” The Abrenica, ­ M. ­J. ­V. 2009. “Detecting and Journal of Business 33 (2): 157–79. ­ http://www​ Measuring Market Power in the Philippine .jstor.org/stable/2351015. Wholesale Electricity M ­ arket.” The Philippine Bushnell, ­ B., ­ J. ­ E. ­T. Mansur, and ­ C. ­Saravia. Review of Economics XLVI (2): 5–46. 2008. “Vertical Arrangements, Market Andrews-Speed, ­ P. 2016. “Energy Security and Structure, and Competition: An Analysis of Energy Connectivity in the Context of ASEAN Restructured US Electricity ­ Markets.” American Energy Market ­ Integration.” ASEAN Studies Economic Review 98 (1): 237–66. ­ https://doi​ Center, Chulalongkorn University; Energy .org/10.1257­/aer.98.1.237­. Studies Institute, University of Singapore; Castalia. 2013. “International Experience with Norwegian Studies Institute, ­ F ebruary. Single Buyer Models for E ­ lectricity.” Report for ­h ttp://www.asean-aemi.org/wp-content​ Contact Energy, New ­ Zealand. ­ https://www. /­uploads/2016/03/AEMI-Forum-November​ yumpu.com/en/document/read/28356228​ -2015-Andrews-Speed-Feb2016­.pdf. /­castalia-review-international-experience-sin Arizu, ­ B ., ­ W. ­H . Dunn, and B ­. T­ enenbaum. gle-buyer-models. 2001. “Regulating Transmission: Why System Chattopadhyay, D ­ ., and P ­. N­.F­ ernando. 2011. Operators Must Be Truly ­ Independent.” Public “Cross-Border Power Trading in South Asia: Policy for the Private Sector ­ No. 226, World Bank, It’s Time to Raise the G ­ ame.” Electricity Journal Washington, ­DC. ­http://siteresources.worldbank​ 24 (9): 41–50. ­ h ttps://doi.org/10.1016­ / j​ .org​/­EXTFINANCIALSECTOR/Resources/282884​ .tej.2011.10.009. -1303327122200/226Ariz-1221­.pdf. Ching, ­ L. 2014. “Electricity Spot Market: The T. ­ Arizu, ­ D . Gencer, and L B ., ­ ­. ­ M aurer. 2006. Singapore E ­ xperience.” Presentation at the “Centralized Purchasing Arrangements: GCCIA 3rd Regional Power Trade Forum, International Practices and Lessons Learned on Abu Dhabi, September 29. http://www.gccia​ Variations to the Single Buyer ­ Model.” Energy .com.sa/Data/PressRelease​/­Press_39­.pdf. and Mining Sector Board Discussion Paper ­ No. Defeuilley, C ­ . 2009. “Retail Competition in 16, World Bank, Washington, ­ DC. Electricity ­Markets.” Energy Policy 37 (2): 377–86. Asian Development B ­ ank. 2015. “Knowledge and DOE (Department of Energy, ­ Philippines). 2014. Power: Lessons from ADB Energy ­ Projects.” “Department of Energy Power S ­ ituation.” Mandaluyong City, ­ Philippines. Report, Department of Energy, ­ M anila. Barker, ­ J., ­ B. Tenenbaum, and ­ F. ­ Woolf. 1997. ­h ttp://www.doe.gov.ph/sites/default/files​ “Governance and Regulation of Power Pools /­pdf/­electric_power/power_situationer/2014­​ and System Operators: An International _power_situationer.pdf. Comparison.” World Bank Technical Paper ­ ­ No. ESMAP (Energy Sector Management Assistance 382, World Bank, Washington, ­ DC. ­ Program). 2013. “International Experience Batlle, ­ C . 2013. “Electricity Generation and with Open Access to Power Grids: Synthesis Wholesale M ­ arkets.” In Regulation of the Report.” Knowledge Series 016/13, World ­ Power Sector, edited by ­ I. J­ . Pérez-Arriaga, Bank, Washington, ­ DC. ­ http://documents. 341–95. London: S ­ pringer-Verlag. https://doi​ worldbank.org/curated​/en/2013/11/18699545​ .org/10.1007/978-1-4471-5034-3. /­international​-experience-open-access-power​ Besant-Jones, John ­ E. 2006. “Reforming Power -grids​-synthesis-­report. Markets in Developing Countries: What Have Foster, Vivien, Samantha Helen Witte, Sudeshna We Learned?” Energy and Mining Sector Ghosh Banerjee, and Alejandro Vega Moreno. Board Discussion Paper n ­ o. 19, World Bank, 2017. “Charting the Diffusion of Power Sector Washington, ­DC. Reforms across the Developing World.” Policy Besant-Jones, ­ J ., and ­ B. ­Tenenbaum. 2001. Research Working Paper 8235, Rethinking “The California Power Crisis: Lessons for P o w e r S e c t o r R e f o r m , Wo r l d B a n k , Developing ­ Countries.” ESMAP Discussion Washington, DC. Paper N ­ o. 1, World Bank, Washington, D ­ C. Hogan, ­ W. ­ W. 1998­ a. “Competitive Electricity http://www-wds.worldbank.org/external​ Markets: A Wholesale ­ Primer.” Center for /­d efault/WDSContentServer/WDSP/IB/2 Business and Government, John ­ F. Kennedy 004/03/25/000090341_20040325143555​ School of Government, Harvard University, /­Rendered/PDF/280840California0Power0cri Cambridge, ­ M A. http://www.hks.harvard​ sis0EMS0­no.01­.pdf. .edu/fs/whogan/empr1298­.pdf. What Progress Has Been Made with Wholesale Power Markets? 235 Hogan, ­ W. ­ W. 1998­ b . “Nodes and Zones in Kim, S ­ . Kim, and J ­ ., Y ­.S ­ hin. 2013. “The Korean ­ .S Electricity Markets: Seeking Simplified Electricity Market: Stuck in ­ Transition.” In Congestion ­ Pricing.” In Designing Competitive Evolution of Global Electricity Markets, edited by Electricity Markets , edited by H ­ . Chao and ­ F. ­P. Sioshansi, 679–713. Academic ­ Press. H. ­ ­ G. Huntington, 33–62. New York: Springer Krishnaswamy, V ­ . 1999. “Non-payment in the Science+Business ­Media. Electricity Sector in Eastern Europe and Hogan, W ­ . W ­ . 2002. “Electricity Market the Former Soviet ­ Union.” Technical Paper Restructuring: Reforms of ­ Reforms.” Journal N o. 423, World Bank, Washington, ­ ­ D C. of Regulatory Economics 21 (1): 103–32. https:// http://documents.worldbank.org/curated​ doi.org/10.1023/A:1013682825693. /­en/1999/06/440475­/non-payment-electricity​ Holttinen, ­ H., ­ A. Tuohy, ­ M. Milligan, ­ E. Lannoye, -sector-eastern-europe-former-soviet-union. V. Silva, S ­ ­ . Muller, and L ­. S ­ oder. 2013. “The Krishnaswamy, ­ V., and G ­. ­S tuggins. 2003. Flexibility Workout: Managing Variable Private Participation in the Power sector in Europe Resources and Assessing the Need for Power and Central Asia: Lessons from the Last D ­ ecade. System ­M odification.” IEEE Power and Working Paper ­ No. 8. Washington, DC: World Energy Magazine 11 (6): 53–62. https://doi​ ­B ank. http://documents.worldbank.org​ .org/10.1109­/MPE.2013.2278000. /­curated/en/2003/07/2477725­/private​-­partici Hunt, ­ S . 2002. Making Competition Work in pation-power-sector-europe-central-asia-lessons​ ­Electricity. New York: John Wiley and Sons, I ­ nc. -last-decade. IEA (International Energy A ­ gency). 2016. Larsen, ­ E. ­ R., ­I. Dyner, ­ V. ­L. Bedoya, and ­ J. C. ­ Re-powering Markets: Market Design and Franco. 2004. “Lessons from Deregulation in ­ Regulation during the Transition to Low-Carbon Colombia: Successes, Failures and the Way Power ­Systems. Paris: ­ IEA. ­Ahead.” Energy Policy 32 (15): 1767–80. https:// Jamasb, T ­ ., ­R. Nepal, and G ­.­R. ­Timilsina. 2015. doi.org/10.1016/S0301-4215(03)00167-8. “A Quarter Century Effort Yet to Come of Age: Littlechild, ­ S . 2009. “Retail Competition in A Survey of Power Sector Reforms in Developing Electricity Markets: Expectations, Outcomes Countries.” Policy Research Working Paper ­ and ­Economics.” Energy Policy 37 (2): 759–63. 7330, World Bank, Washington, ­ D.C. Lovei, L ­ . 2000. “The Single-Buyer Model: Jamasb, T ­ ., ­ D. Newbery, and M ­ .­ Pollitt. 2005. “Core A Dangerous Path toward Competitive Indicators for Determinants and Performance Electricity ­ Markets.” Viewpoint Note 225, World of the Electricity Sector in Developing Bank, Washington, ­ DC. ­ Countries.” Policy Research Working Paper Maurer, ­ L ., and ­ L. ­B arroso. 2011. Electricity 3599, World Bank, Washington, ­ DC. Auctions: An Overview of Efficient ­ P ractices . Joskow, P ­. ­L . 2008­ a . “Capacity Payments in Washington, DC: World B ­ ank. https://doi​ Imperfect Electricity Markets: Need and .org/10.1596/978-0-8213-8822-8. ­Design.” Utilities Policy 16 (3): 159–70. https:// McRae, ­ S. ­D., and ­ A. ­ F. ­ Wolak. 2016. Diagnosing doi.org/10.1016­/j.jup.2007.10.003. the Causes of the Recent El Niño Event and ———. 2008­ b. “Lessons Learned from Electricity Recommendations for ­ Reform. Program on Energy Market ­Liberalization.” The Energy Journal, and Sustainable Development, Stanford Special Issue on the Future of Electricity, Papers in ­University. ­https://web.stanford.edu/group​ Honor of David Newbery, 9–42. http://­ economics​ /­fwolak/cgi-bin/sites/default/files/­diagnosing​ .mit.edu/files/2093. -el-nino_mcrae_wolak.pdf. ———. 2011. “Comparing the Costs of Intermittent MIT Energy ­ Initiative. 2016. Utility of the F ­ uture. and Dispatachable Electricty Generating Cambridge, MA: Massachusetts Institute ­Technologies.” American Economic Review of ­ Technology. http://energy.mit.edu/wp​ 101 (3): 238–41. -­c ontent/uploads/2016/12­/ Utility-of-the​ Joskow, P ­. ­ L ., and R ­. S­ chmalensee. 1983. -Future-Full-Report.pdf. Markets for Power: An Analysis of Electric Utility Mukhopadhyay, ­ S., ­V. ­K. Agrawal, ­ K. Soonee, S. ­ ­Deregulation. Cambridge, MA: MIT ­ Press. ­ P. ­K. Agarwal, ­ R. Anumasula, and ­ Kumar. C. ­ Kee, ­ E. ­ D. 2001. “Vesting Contracts: A Tool for 2016. “Detecting Low Frequency Oscillations Electricity Market T ­ ransition.” The Electricity through PMU-Based Measurements for Indian Journal 14 (6): 11–22. ­ https://doi.org/10.1016​ National G ­ rid.” In Power Systems Computation /S1040-6190(01)00211-1. Conference (PSCC), 1–7, ­ Genoa. 236 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Munoz, F ­. D­ ., S ­ . Wogrin, S ­. S­ . Oren, and B ­.F ­. Pollitt, M ­ . 2008. Electricity Reform in Argentina: Hobbs. 2017. “Economic Inefficiencies of Cost- ­ Lessons for Developing C ­ ountries.” Energy Based Electricity Market D ­ esigns.” USAEE Economics 30 (4): 1536–67. h ­ ttps://doi​ Working Paper N ­ o. 17-313, U ­ .S. Association .org/10.1016­/j.eneco.2007.12.012. for Energy Economics, Cleveland, ­ OH. ­ https:// Pollitt, M ­ .­ G. 2012. “Lessons from the History of doi.org/10.2139­/ssrn.2974353. Independent System Operators in the Energy Nepal, ­ R., and ­ T. ­ Jamasb. 2012. “Reforming the ­Sector.” Energy Policy 47 (August) 32–48. Power Sector in Transition: Do Institutions Raineri, ­ I. Dyner, ­ R., ­ J. Goñi, ­ N. Castro, ­Y. Olaya, Matter?” Energy Economics 34 (5): 1675–82. and ­ Franco. 2013. “Latin America Energy C. ­ ­h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6­/ j . e n e c o​. 2 0 1 2​ Integration: An Outstanding ­ D ilemma.” .02.002. Chapter 14 in Evolution of Global Electricity Neuhoff, ­ K., ­ J. Barquin, ­ J. ­ W. Bialek, ­ R. Boyd, ­M arkets . Academic P ­ ress. ­ h ttps://doi.org​ C. ­ ­ J. Dent, ­ F. Echavarren, ­ T. Grau, ­ C. von /10.1016/B978-0-12-397891​-2.00014-6. Hirschhausen, ­ B. ­ F. Hobbs, ­ F. Kunz, ­ C. Nabe, Rose, ­ A., ­R. Stoner, and ­ I. ­ Pérez-Arriaga. 2016. ­ G. Papaefthymiou, C ­ .Weber, and H ­.W ­ eigt. “Integrating Market and Bilateral Power 2013. “Renewable Electric Energy Integration: Trading in the South African Power ­ Pool.” Quantifying the Value of Design of Markets WIDER Working Paper ­ N o. 2016/132, for International Transmission C ­ apacity.” United Nations University World Institute Energy Economics 40 (November): 760–72. for Development Economic Research Newbery, ­ D . 2002. “Issues and Options in ­(UNU-WIDER). ­https://www.econstor.eu/­bits Restructuring Electricity Supply ­ Industries.” tream/10419/161515/1/873997999­.pdf. Cambridge Working Papers in Economics Roxas, ­ F., and ­ A. ­S antiago. 2010. “Broken No. 0210, Department of Applied Economics, ­ Dreams: Unmet Expectations of Investors in University of ­ C ambridge. ­ h ttps://pdfs​ the Philippine Electricity Restructuring and .­semanticscholar.org/2e6e/c9ddece0f664bbc72​ ­Privatization.” Energy Policy 38 (11): 7269–77. ec6a6c85d28ddee59­ba.pdf. ­https://doi.org/10.1016­/j.enpol​.2010.08.003. Newbery, ­ D. 2005. “Electricity Liberalisation in Rudnick, ­ H., ­R. Varela, and ­ W. ­Hogan. 1997. Britain: The Quest for a Satisfactory Wholesale “Evaluation of Alternatives for Power Market ­Design.” The Energy Journal 26 (Special System Coordination and Pooling in a Issue): 43–70. ­h ttps://www.­r esearchgate​ Competitive ­Environment.” IEEE Transactions .net/publication/227356675­_ Electricity​ on Power Systems 12 (2): 605–13. ­ https://doi​ _Liberalisation_in_Britain_The_Quest_for_a​ .org/10.1109/59.589615. _Satisfactory_Wholesale_Market_Design. Rudnick, ­ H., and C Velasquez. 2018. “Taking Stock ­ .­ Newbery, ­ D. 2015. “Missing Money and Missing of Wholesale Power Markets in Developing Markets: Reliability, Capacity Auctions Countries: A Literature ­ Review.” Policy Research and ­I nterconnectors.” Energy Policy 94 Working Paper 8519, World Bank, Washington, (July): 401–10. ­h ttps://doi.org/10.1016­/ j​ ­D.C. ­http://documents.worldbank.org/curated​ .enpol.2015.10.028. /­en/992171531321846513​ /­Taking-stock-of​ Ofgem (Office of Gas and Electricity M ­ arkets). -wholesale-power-markets-in-developing​ 1999. The New Electricity Trading ­ Arrangements. -countries-a-literature-review. Ofgem, ­L ondon. ­h ttps://archive.uea​.ac​ ———. 2019­ a . “Learning from Developing .uk/~e680/energy/energy_links/electricity​ Country Power Market Experiences: The Case /­neta1­.pdf. of the ­ Philippines.” Policy Research Working Oseni, M ­ . O ­ ., and M ­ . G ­. P ­ ollitt. 2016. “The Paper 8721, World Bank, Washington, D ­ C. Promotion of Regional Integration of ­h ttps://openknowledge.worldbank.org​ Electricity Markets: Lessons for Developing /­b itstream/handle/10986/31189/WPS8721​ ­Countries.” Energy Policy 88 (July): 628–38. .pdf?sequence=1­&isAllowed=y. ­https://doi.org/10.1016­/j.enpol.2015.09.007. ———. 2019­ b . “Learning from Developing Pollitt, ­ M. 2004. “Electricity Reform in Chile: Country Power Market Experiences: The Case Lessons for Developing ­ Countries.” Journal of ­Colombia.” Policy Research Working Paper of Network Industries 5 (3–4): 221–62. 8771, World Bank, Washington, ­ DC. ­ ­http://­heinonline.org/HOL/Page?handle=hein​ ———. 2019­ c . “Learning from Developing .journals/netwin5&id=221­&div=&collection=. Country Power Market Experiences: The Case What Progress Has Been Made with Wholesale Power Markets? 237 of ­Peru.” Policy Research Working Paper 8772, Carbon Power System in ­ California.” White World Bank, Washington, ­ DC. ­ Paper, Analysis ­ Group. ———. ­ Forthcoming. “Learning from Developing Timilsina, ­ R., ­ G. ­ M. Toman, ­ J. Karacsonyi, and Country Power Market Experiences: The Case L. De Tena ­ ­ Diego. 2015. “How Much Could of ­I ndia.” Policy Research Working Paper, South Asia Benefit from Regional Electricity World Bank, Washington, ­ D.C. Cooperation and Trade?” Policy Research Ryan, N ­ . 2014. “The Competitive Effects of Working Paper 7341, World Bank, Washington, Transmission Infrastructure in the Indian ­D C. ­h ttps://openknowledge.worldbank. Electricity ­M arket.” ­h ttp://campuspress. org​/­b itstream​/­h andle/10986/22224​/­H o yale.edu/nicholasryan/files/2014/10/ryan​ w0much0could0peration0and0trade00​ _­c ompetition_india_electricity_2014oct23​ .pdf?sequence=1­&isAllowed=y. -­1gdti64­.pdf. Vagliasindi, M ­ ., and ­ B esant-Jones. 2013. J. ­ ———. 2017. “The Competitive Effects of Power Market Structure: Revisiting Policy ­ Options. Transmission Infrastructure in the Indian Directions in ­ Development. Washington, DC: Electricity ­ M arket.” NBER Working Paper World ­Bank. ­https://doi.org/10.1596/978-0- 23106­ . National Bureau of Economic 8213-9556-1. Research, Cambridge, ­ MA. ­ https://www.nber​ Wijayatunga, P ­ . Chattopadhyay, and P ­ ., D ­. N­. .org/papers/w23106.pdf. Fernando. 2015. “Cross-Border Power Trading ­ Sen, ­ A. 2014. Divergent Paths to a Common Goal? in South Asia: A Techno Economic ­ Rationale.” An Overview of Challenges to Electricity Sector ADB South Asia Working Paper ­ No. 38, Asian Reform in Developing versus Developed ­ Countries. Development Bank, M ­ ttps://www​ ­ anila. h .adb​ OIES Paper EL 10. Oxford: Oxford Institute for .org/sites/default/files/publication/173198​ Energy ­Studies. /­south-asia-wp-038­.pdf. Sioshansi, ­ R ., ­ S . Oren, and ­ R. ­O ’Neill. 2008. Wilson, ­ R. 1998. “Design ­ Principles.” In Designing “The Cost of Anarchy in Self-Commitment- Competitive Electricity Markets, edited by ­ H. Chao Based Electricity M ­ arkets.” In Competitive and H­ .­G. Huntington, 212. Kluwer Academic Electricity Markets: Design, Implementation, ­Publishers. Performance, edited by ­ P. ­ F. ­ Sioshansi. Oxford, Wolak, ­ F. ­A . 2003­ a . “Designing Competitive ­U.K.: Elsevier ­Ltd. Wholesale Electricity Markets for Latin Stoft, S ­ . 2002. Power System Economics: Designing American C ­ ountries.” Paper prepared for Markets for ­ Electricity. Piscataway, NJ: Wiley​ the First Meeting of the Latin American -IEEE ­Press. Competition Forum, Paris, April 7–8. h ­ ttp:// Strbac, ­ G. 2008. “Demand Side Management: www.oecd.org/daf/competition/­prosecutiona Benefits and C ­ hallenges.” Energy Policy 36 ndlawenforcement/2490209­.pdf (12): 4419–26. ­h ttps://doi.org/10.1016­/ j​ ———. 2003­ b . “Diagnosing the California .enpol.2008.09.030. Electricity ­Crisis.” The Electricity Journal 16 Sweeney, ­ J. ­ L. 2002. The California Electricity ­Crisis. (7): 11–37. h ­ ttps://doi.org/10.1016/S1040​ Stanford University: Hoover Institution ­ Press. -6190(03)00099­-X. Thomas, M ­ . 2012. “Does Your Single Buyer ———. 2005. “Managing Unilateral Market Make the Very Best Bad Decision?” Lantau Power in ­ Electricity.” Policy Research Working Pique, ­July. ­http://www.lantaugroup.com/files​ Paper 3691, World Bank, Washington, D ­ C. /­pique_single_buyer.pdf. ­h ttp://documents.worldbank.org/curated​ Tierney, S ­. ­ F. 2018. “Resource Adequacy and /­en/2005/09/6246687­/managing-­unilateral​ Wholesale Market Structure for a Future Low -market-power-electricity. Gauging Impact PART III 239 Did Power Sector Reforms Improve Efficiency and Cost Recovery? 8 Guiding questions • Did countries make material progress toward improving cost recovery in the power sector? • To what extent was the evolution in cost recovery driven by increasing tariffs versus reducing costs? • How did cross-subsidies affect the achievement of cost recovery? • To what extent were any persistent shortfalls in cost recovery absorbed through fiscal transfers or simply allowed to undermine the financial viability of the utilities? Summary • Electricity tariffs are rarely high enough to cover the full costs of service delivery. This underrecovery of costs is common both for countries facing high costs of electricity and for those facing low costs of electricity. • The underlying reasons for lack of cost recovery can be found not only in relatively low tariffs but also in costs that are inflated because of inefficiency. • Costs in the power sector fluctuate greatly over time because of exposure to foreign exchange and oil price fluctuations and because of hydrological risk. As a result, even countries that achieve cost recov- ery struggle to sustain this achievement over time. • Power utilities often impose a substantial fiscal burden and contingent liabilities on government budgets. Nevertheless, financial losses due to the underrecovery of costs are seldom fully compen- sated by government subsidies. Rather, they are absorbed by the utility through the accumulation of arrears with suppliers and short-term loans from commercial banks. • Tariff levels are highly differentiated across customer groups often with a view to preserving afford- ability for residential consumers; however, such large cross-subsidies further undermine the achieve- ment of cost recovery. • Overall, the cost recovery ratio for power utilities has improved somewhat over the last 25 years. Progress has been very uneven, however, with cost recovery improving in about half the countries, and deteriorating in the rest. • In most cases, tariff erosion due to inflation outweighed the impact of tariff increases on cost recovery; however, system losses declined more consistently across the period and have been a major contribu- tor to improvements in cost recovery. 241 242 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION of electricity often leads to poor financial via- This chapter evaluates the extent to which bility of a utility, which, in turn, results in developing countries made progress over the explicit or implicit government subsidies. The period 1990–2015 on cost recovery and opera- terms are not identical, however. The following tional efficiency for power utilities.1 The theory paragraphs define each term and explains how of change underlying power sector reform they are interrelated. envisioned improved sector performance by Cost recovery is an attribute of electricity way of two intermediate outcomes. First, it was ­ tariffs—the average tariff corresponding to the thought that improved cost recovery and average cost of service. It is usually measured as financial viability would attract needed invest- the ratio between the average effective electric- ment in service quality and security of supply. ity tariff and the reference cost, often expressed Second, greater operational efficiency for the as a percentage.2 Average effective tariffs are utilities would ease cost recovery, also thought calculated by dividing total revenue from elec- to be critical to better service quality. Thus, the tricity sales by the volume of electricity sales in guiding questions for the chapter are as fol- kilowatt-hours (kWh). Various definitions of lows: Did countries increase cost recovery in costs exist, however, and all are useful in differ- the power sector? As cost recovery measures ent circumstances. Therefore, although cost evolved, were they driven by increasing recovery is sometimes reported in binary terms, ­ tariffs or by reducing costs? How did cross-­ the reality—particularly in the developing subsidies affect cost recovery? To what extent world—is that degrees of cost recovery exist on were persistent shortfalls absorbed through fis- a continuum, and that cost recovery can be cal transfers? Or were they allowed to under- analyzed from different perspectives. The World mine the financial viability of the utilities? Bank has developed a conceptual framework to Cost recovery, financial viability, and fiscal define different levels of cost recovery sustainability are closely related concepts (table 8.1) The definitions include three l­evels— sometimes used interchangeably. Underpricing operating cost recovery, limited capital cost TABLE 8.1 The cost recovery ladder from financial, fiscal, and economic perspectives A. Financial perspective B. Fiscal perspective C. Economic perspective Level 1: Level A1: Only those operating costs B1: Operating costs that are covered C1: Operating costs (excluding various Operating that are covered by the utility/sector on behalf of the utility/sector by reserves, such as depreciation, bad debt cost (excluding various reserves, such as the government through budgetary allowance, and revaluation of assets), recovery depreciation, bad debt allowance, and transfers and provision of subsidized irrespective of who bears them, required to revaluation of assets) goods and services adequately run the utility Level 2: A2: A1 plus any financing costs (to the B2: B1 plus the financing costs C2: C1 plus existing capital expenditure Operating utility) for existing capital expenditure, (assessed at the cost of existing (incorporated using the weighted average and limited such as debt service (interest and capital incurred by the government) cost of commercial capital assessed at the capital cost principal), required equity payments, for the capital expenditure covered opportunity cost of debt and equity) recovery and internally funded investments through sovereign funding/guarantee Level 3: A3: A2 plus financing costs (to the B3: B2 plus financing costs (to C3: C2 plus new capital investments required Full cost utility) and the associated O&M costs the government) for new capital to meet future demand (incorporated using recovery of for new capital investments (based on investments (based on an adequate weighted average cost of commercial capital current and an adequate investment prioritization investment prioritization framework) assessed at the opportunity cost of debt and future costs framework) required to meet future required to meet future demand equity capital)a demand Source: Kojima 2017. Note: O&M = operation and maintenance. a. The full definition of C3 also includes externalities, but these could not be assessed appropriately in this study due to a lack of data and are therefore omitted from this definition. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 243 recovery, and full cost recovery from financial, underrecovery of costs (Huenteler and others fiscal, and economic perspectives—to yield a 2017). In line with these findings, this chapter total of nine different cost recovery definitions starts off with a comprehensive economic defi- (A1–A3, B1–B3, and C1–C3). Over the past few nition of the cost of electricity service (levels C1 decades the literature on cost recovery has and C3). This analysis is then complemented by moved from a financial to an economic per- an analysis of cost recovery from a financial spective on cost of service. This shift reflects the perspective (A1, A2, and A3). See box 8.1 for literature’s increased focus on the macroeco- the details of the methodology for cost recovery nomic and environmental implications of the analysis. BOX 8.1 Methodology used in cost recovery analysis The analysis presented in this chapter distinguishes itself from the existing literature on cost recovery by comparing pre- and postreform performance on cost recovery levels over a 20- to 30-year period. This goes well beyond the exist- ing literature (annex 8A). The chapter analyzes power sector reform from the late 1980s and 1990s through the period 2010–17 of 17 jurisdictionsa (14 countries and 3 Indian states). These jurisdictions represent diverse geographies, income levels, and approaches to reform. Postreform cost recovery analysis is based on detailed cost and tariff data collected for the period 2010–17. The anal- ysis collected data on tariff levels from 2010 to 17 (or a subset of these years) for all 17 jurisdictions on the country and utility levels and compared those data with estimates of six cost recovery levels (A1–A3 and C1–C3),b using average cost metrics. The total sample of the utility-level analysis includes 18 majority publicly owned and 7 privately owned utilities. Where possible, the analysis included cash collected along with revenue billed. Compiled from the financial statements of all utilities in the sector, the data helped us obtain a full picture of cost and revenues in the jurisdictions, complemented with information on indirect government support (for example, in the form of subsidized fuels). Exceptions were made for independent power producers and small power producers, the cost of which were approximated by the electricity pur- chase cost of the off-taking utility. The only two cases where such a holistic picture of costs and revenues in the sector was not possible were the Dominican Republic, where the analysis focused only on the utility Edesur, and Morocco, where the analysis relied on a previous study by Camos and others (2018) for information on full cost recovery levels in the sector. All data presented in the chapter are expressed in 2017 U.S. dollars to adjust for inflation. An overview of data compiled for the case studies, the years covered, and the data sources is provided in annex 8b. Cost recovery analysis of the prereform period is based on historical studies of the 1980s–1990s that compare tariff levels with long-run marginal cost (LRMC). Data on tariff levels for this period were collected for all 17 jurisdictions and compared with LRMC estimates, which reflect the long-run marginal cost of supply that would need to be covered to expand the system, accounting for shadow prices for fuels, labor, and capital. Comparable LRMC estimates are available for most countries for the 1980s and 1990s, when the World Bank financed a number of LRMC studies. For 12 jurisdictions, the tariff and cost data are for 1987 and based on a World Bank study (1990) that uses a strictly compa- rable methodology. For two other jurisdictions, the data are based on the same LRMC methodology but for different years: 1991 (Tanzania) and 1993 (Vietnam) (World Bank 1993, 1995). For the remaining three cases, the assessment compares tariffs and current cost-of-service estimates (as opposed to LRMC) for 1991 (Senegal), 1994 (Ukraine), and 2003 (Tajikistan) (World Bank 1994, 1998, 2004). The exact years used for prereform analysis in each country are reported in annex 8B. The results of this second analytical step provide a baseline for assessments of reform impacts and allow assessment of whether countries were able to raise tariffs or reduce costs compared to the prereform period. The methodological differences between the historic analysis based on LRMC and the recent analysis based on average cost mean that the analysis cannot provide conclusions about whether or not actual costs have risen or fallen com- pared to prereform estimates. Because LRMC can legitimately be interpreted as the best historic estimate of future costs in fast-growing power systems,c however, the analysis can help us understand if countries’ actual costs from 2010 to 2015 were either higher or lower than long-term average cost in the 1980s. Cost recovery is assessed with respect to a utility’s actual costs without passing judgment on its efficiency. Over time, cost recovery may improve through higher tariffs or reduced costs—that is, improving operational efficiency. The analysis of utility efficiency focuses on (1) excessive transmission and distribution losses above a suitable norm (ranging between 7–13 percent, see annex 8D) taking into account the operating context of the utility; and (2) the undercollection of bills. Utility inefficiencies are measured as a percentage of revenue the utility foregoes because of these two factors. a. The cases are Colombia, Dominican Republic, Arab Republic of Egypt, India (Andhra Pradesh, Odisha, and Rajasthan), Kenya, Morocco, Pakistan, Peru, Philippines, Senegal, Tajikistan, Tanzania, Uganda, Ukraine, and Vietnam. b. Externalities were not assessed as part of C3 because data were unavailable. c. For Ukraine and Tajikistan, the prereform cost estimates are based on actual cost rather than LRMC to account for the fact that installed capacity did not grow by very much compared to the prereform period. 244 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Financial viability is an attribute of utility delivery; or (3) increase the revenues retained companies3 and is fulfilled when tariff reve- by the electricity producer or service provider nues and other sources of income cover the (Kojima 2017). This means that electricity can cost of service. Cost recovery of tariffs is obvi- be subsidized whether or not the utility incurs ously a key determinant, but financial viability a visible cash shortfall, and whether or not vis- also depends on accessible government trans- ible cash waterfall is covered by fiscal transfers fers, ready cash inflows (taking into account from the budget (as opposed to commer- collection losses and timely allocation of cial borrowing, deferred depreciation and government transfers), and proper spending ­ so forth.). priorities (for example, debts and payables are The quasi-fiscal deficit (hidden costs in the settled in a timely manner, or utilities pay out ­ case of private utilities, QFD) is a measure of large dividends or lend to other state-owned implicit fiscal costs of the power sector (or hid- enterprises [SOEs]). Therefore, although the den losses in the case of private utilities). The two mutually reinforce each other and the lit- QFD is measured as the difference between erature sometimes uses them interchangeably, the cash collected by the existing utility and a utility can be financially viable even when the revenues that would be collected without cost recovery is below 100 percent—for exam- bill collection losses by a utility applying cost ple, if tariffs are set below cost recovery level recovery tariffs (in this analysis, using cost but reliable fiscal transfers are made to com- benchmark level C3) and achieving commer- pensate for the shortfall. Similarly, a utility cial and operational efficiency. In general, may not be financially viable when tariffs are power utilities in most developing countries at cost recovery level, for example, when the are state-owned and can be considered utility uses its cash flows to finance new ­ quasi-fiscal entities. Typically, these utilities ­ investments while accumulating arrears to its display poor financial performance in part suppliers and financiers. Further, analyses of because they channel various transfers to con- financial viability usually do not differentiate sumers through underpricing, uncollected bills, between revenues from electricity sales and and unmetered consumption. The total cost of revenue not related to the sale of electricity such transfers, however, is not reflected in the (for example, government transfers); take public budget because it is implicit or involun- input cost at invoiced value (for example, fuels, tary (for example, theft). The resulting finan- capital, land, or labor at subsidized prices); and cial gap in the public utility has been called in count SOEs’ contribution to the government’s the literature QFD, typically expressed as per- revenues—for example, in the form of taxes, centage of gross domestic product (GDP), or duties, and, for SOEs and any mandatory allo- hidden cost, expressed in absolute terms.4 The cations from profits—as costs. When discussing QFD can usefully be disaggregated to clarify the financial viability in sectors with multiple how much is attributable to three main factors: utility companies, the term can apply to each (1) system losses, or the cost of electricity company separately or as an aggregate of all injected into the transmission system but not companies in the sector. metered/billed, minus the cost of electricity lost Electricity subsidies are understood as an for technical reasons within the normative attribute of the sector or the economy. level of 10 percent; (2) collection losses, or the Electricity subsidies can be defined as deliber- value of electricity billed but not collected from ate government policy actions targeting customers; and (3) underpricing, or the differ- ­ electricity services that (1) reduce the net cost ence between the amount billed to customers of electricity or fuels purchased; (2) reduce the and the cost of the corresponding amount of cost of electricity production or service electricity.5 Did Power Sector Reforms Improve Efficiency and Cost Recovery? 245 KEY FINDINGS figure 8.1). Egypt and the Indian states of (­ The most salient results from the analysis of Andhra Pradesh and Rajasthan are the lowest efficiency and cost recovery across the 25 util- performers. In addition, in several cases, bill ities are summarized in the following key collection losses introduce additional financial findings. These findings evaluate the current burden on utilities. Bill collection losses are performance of utilities against the intermedi- particularly high in Egypt and Tajikistan. These ate outcomes of efficiency and cost recovery findings are broadly in line with recent studies and go on to examine historical trends to see (see annex 8A), suggesting that the sample of to what extent performance has improved cases is representative of developing countries over time and why. more broadly. On the utility level, the analysis of postre- form cost recovery suggests that tariffs in only Finding #1: Electricity tariffs rarely 3 out of 24 case studies cover the full cost of cover the full costs of service delivery service delivery (C3). Tariffs in 10 out of 23 Cost recovery remains an elusive goal for many case studies do not even recover operating power sectors. Governments made noteworthy costs of electricity service. Ukraine (109 per- efforts on cost recovery, largely along the lines cent on average in 2010–17) and Colombia of the standard model. Consistent with the (104 percent) are the top performers in this methodological framework described above sample. Egypt and the Indian states of Andhra (table 8.1), cost recovery is analyzed from a full Pradesh (both utilities) and Rajasthan (all three economic cost perspective (C3), while provid- utilities) are the lowest performers. ing for comparison measures A1 and A2, From a financial perspective, which excludes which are the most immediate determinants of many implicit and “hidden” costs, the picture the financial viability of the electricity service. looks slightly better. Eleven out of 16 countries Cost recovery can be achieved through cost or states and 14 utilities out of 23 utilities are reduction or tariff adjustments or a combina- recovering their financial operating costs (A1; tion. Cost reduction takes time, however, see figure 8.2, panels a and c). Eight of 16 coun- because it requires major changes in genera- tries or states and 8 of 23 utilities are meeting tion sources and ­ e fficiency improvement, their operating and existing-capital expendi- whereas tariff adjustments may encounter tures from a financial perspective (see figure 8.2, political and social r­ esistance. Subsidized elec- panels b and d). tricity is so entrenched that it is seen as a social right in some countries (the Arab Republic of Finding #2: The underrecovery of costs Egypt, India, Pakistan, and Senegal). This view is common among countries with both is reflected in a continued pattern for under­ relatively high and relatively low costs recovered costs in most cases. of electricity On the country level, the analysis of post­ reform cost recovery levels suggests that tariffs It is notable that cost recovery appears unre- in only 2 out of 17 case studies cover the full lated to the absolute level of costs. The cost of cost of service delivery (C3). Tariffs in 7 out of electricity service differs by an order of magni- 17 case studies do not even recover operating tude between the case studies, but underre- costs of electricity service. Ukraine (109 per- covery of costs is an issue at both ends of the cent on average in 2010–17) and Colombia cost spectrum (see figure 8.3). When costs are (104 percent) are the top performers on full high, either because of the legacy of expen- economic cost recovery in this sample sive generation sources and high-cost take or 246 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.1 Although most achieved operating cost recovery, few attained full cost recovery a. Average full cost recovery level, at country level, 2010–17 (C3) b. Average full cost recovery level, at country level, 2010–17 (C1) Ukraine 2 out of 17 Colombia Colombia ≥ 100% Peru Peru Uganda Philippines Tajikistan Uganda 9 out of 17 Kenya 10 out of 17 ≥ 100% India – Odisha Ukraine ≥ 80% Kenya Philippines Vietnam Pakistan Morocco Vietnam Pakistan Egypt, Arab Rep. Senegal India – Odisha 13 out of 17 Tanzania ≥ 80% Tajikistan Senegal Dominican Republic Tanzania Egypt, Arab Rep. India – Andhra Pradesh India – Rajasthan India – Rajasthan India – Andhra Pradesh Dominican Republic 0 20 40 60 80 100 120 0 50 100 150 200 Cost recovery level (%) Cost recovery level (%) c. Average full cost recovery level, at utility level, 2010–17 (C3) d. Average operating cost recovery level, 2010–17 (C1) Ukraine Colombia 3 out of 24 Colombia Peru (Luz del Sur) ≥ 100% Peru (Luz del Sur) Peru (Hidrandina) Philippines (Meralco) Kenya India – Odisha (WESCO) Uganda Uganda Tajikistan Kenya 13 out of 23 Philippines (Meralco) Vietnam 13 out of 24 ≥ 100% ≥ 80% Ukraine Philippines (Beneco) Egypt, Arab Rep. India – Odisha (CESCO) Philippines (Beneco) Peru (Hidrandina) Vietnam Pakistan (LESCO) India – Odisha (WESCO) Morocco India – Odisha (CESCO) Tanzania India – Andhra Pradesh (APEPDCL) Senegal Tajikistan Pakistan (LESCO) Pakistan (KE) Senegal India – Andhra Pradesh (APSPDCL) 21 out of 23 Dominican Republic ≥ 80% India – Rajasthan (JVVN) Tanzania Egypt, Arab Rep. India – Rajasthan (JVVN) India – Rajasthan (JDVVN) India – Rajasthan (AVVN) India – Andhra Pradesh (APEPDCL) India – Rajasthan (JDVVN) India – Rajasthan (AVVN) Pakistan (KE) India – Andhra Pradesh (APSPDCL) Dominican Republic 0 50 100 150 0 100 200 Cost recovery level (%) Cost recovery level (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 247 FIGURE 8.2 About half of cases cover cash needs for limited capital costs a. Average financial operating cost recovery level, b. Average financial operating and limited capital cost at country level, 2010–17 (A1) recovery level, at country level, 2010–17 (A2) Colombia Uganda Egypt, Arab Rep. Pakistan Peru Colombia Pakistan Kenya 8 out of 16 Uganda 11 out of 16 Ukraine ≥ 100% Tajikistan ≥ 100% Peru Kenya Tajikistan Ukraine Philippines Philippines Vietnam Vietnam Senegal India – Odisha 13 out of 16 India – Odisha Egypt, Arab Rep. ≥ 80% 14 out of 16 Senegal Tanzania ≥ 80% India – Andhra Pradesh Tanzania India – Rajasthan India – Andhra Pradesh Dominican Republic India – Rajasthan 0 50 100 150 200 0 50 100 150 Cost recovery level (%) Cost recovery level (%) c. Average financial operating cost d. Average financial operating and limited capital recovery level, at utility level, 2010–17 (A1) cost recovery level, at utility level, 2010–17 (A2) Colombia Uganda Egypt, Arab Rep. Colombia Peru (Luz del Sur) Kenya Peru (Hidrandina) Peru (Luz del Sur) 8 out of 22 Uganda Ukraine ≥ 100% Tajikistan Tajikistan Kenya 14 out of 23 Philippines (Meralco) Ukraine ≥ 80% Philippines (Beneco) Philippines (Meralco) Vietnam India – Andhra Pradesh (APSPDCL) India – Odisha (WESCO) Philippines (Beneco) Peru (Hidrandina) Vietnam India – Odisha (CESCO) Senegal 17 out of 22 Pakistan (LESCO) India – Andhra Pradesh (APEPDCL) ≥ 100% Egypt, Arab Rep. India – Odisha (WESCO) Senegal India – Odisha (CESCO) Pakistan (LESCO) India – Andhra Pradesh (APEPDCL) 21 out of 23 Tanzania Tanzania ≥ 80% India – Rajasthan (JVVN) Pakistan (KE) India – Rajasthan (AVVN) India – Rajasthan (JVVN) India – Rajasthan (JDVVN) India – Andhra Pradesh (APSPDCL) Pakistan (KE) India – Rajasthan (JDVVN) Dominican Republic India – Rajasthan (AVVN) 0 100 200 0 100 200 Cost recovery level (%) Cost recovery level (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. 248 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.3 Below-cost tariffs remain an issue whether underlying costs are high or low 0.40 0.35 0.30 2017 US$/kWh 0.25 0.20 0.15 0.10 0.05 0 l ic h a s ia da n * ru an a a p. m e an ga co ne ny bi sh in es ha bl an Re na Pe st ist an ra ne m oc di ad pi pu Ke st nz ki et jik Ug lo Uk ab O ilip ja Se or Pa Pr Re Vi Ta Co Ta Ra – Ar M Ph ra a an – di t, dh ic yp a In An di in Eg m In – Do a di In Operating cost Capital cost Cash collected Billed tari revenue Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: No breakdown of operating and capital cost is available for Morocco. kWh = kilowatt-hour. pay contracts (the Philippines, Senegal, and Finding #3: Inefficiencies remain a Uganda) or because of high system losses big contributor to the underrecovery (Dominican Republic, the Indian states of of costs Odisha and Rajasthan, Kenya, and Tanzania), Several of the utilities under study could it is difficult to raise tariffs to cover the full reach or get much closer to cost recovery by cost of service. In Senegal, electricity tariffs improving efficiency without altering their have been frozen since 2009 amid fears of current level of tariffs. Good performance is profound social discontent and the eruption considered to be when a utility loses less than of protests. Nevertheless, even when cost 5 percent of revenues to inefficiency, because is low, such as in Egypt and Vietnam, the this share is equivalent to system (transmis- extreme public sensitivity toward electricity sion and distribution [T&D]) losses of about pricing prevents the government from abol- 10 percent and revenue collection close to ishing subsidies. In Vietnam, whereas end- 100 percent. 6 Both India (Odisha) and user tariffs follow cost, tariff increases above Pakistan would have achieved full cost recov- 7 percent require approval from the ministry ery in the most recent year of data had they of industry and trade or above 10 percent managed full bill collection, shrinking T&D from the prime minister. In Egypt, bold tariff losses to 5 percent of power fed into the trans- adjustments of up to 40 percent were enacted mission grid. Tajikistan and Uganda would in 2016 and 2017, but the low tariff base have both been 11 percentage points closer to means that even heftier increases are needed full cost recovery with these changes, though to cover costs. Political pressure to provide neither would reach full cost recovery with subsidized or free power to certain groups of these improvements alone. India (Rajasthan), consumers (for example, farmers) also under- Kenya, Senegal, Tanzania, and Vietnam cut the cost recovery effort, particularly in would also see improvements in cost recovery India. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 249 (all between 1 and 9 percentage points), (light-shaded bars) (Dominican Republic, Egypt, although full cost recovery could not be India (Odisha and Rajasthan), Pakistan, Senegal, achieved without tariff adjustments. Tajikistan, Tanzania, and Uganda). Some of the More than half of the cases exhibits ineffi- worst performers in the sample lose about ciencies exceeding 5 percent of revenue. As 20 percent of utility revenues to inefficiency, figure 8.4 shows, nine countries lose more than ­ which is equivalent to system losses of about 5 percent of revenue to excessive T&D losses 10 percent above efficient benchmarks and (dark-shaded bars) and noncollection of bills r evenue collection of about 90 percent. ­ FIGURE 8.4 Inefficiencies can account for as much as 10 to 20 percent of utility revenues a. Average ine ciencies as share of revenues, b. Average ine ciencies as share of revenues, at country level, 2010–15 at utility level, 2010–15 Colombia (EPM) Vietnam Vietnam (NPC) Colombia Vietnam (HCMCPC) Philippines (MERALCO) Peru Morocco (ONEE)a Peru (Hidrandina) Moroccoa Ukraine (Dniproblenergo) Philippines Peru (Luz Del Sur) Colombia (Codensa) India – Andhra Pradesh India – Andhra Pradesh (APSPDCL) Kenya Kenya (Kenya Power) Philippines (BENECO) Ukraine India – Andhra Pradesh (APEPDCL) Pakistan (LESCO) Uganda Ukraine (Khmelnytskoblenergo) Senegal Uganda (UMEME) Tanzania (TANESCO) Tanzania Senegal (SENELEC) Tajikistan (Barki Tojik) Pakistan Egypt, Arab Rep. (AEDC) Tajikistan India – Rajasthan (JDVVNL) India – Rajasthan (JVVNL) India – Rajasthan Dominican Republic (Edesur) India – Odisha (WESCO) Egypt, Arab Rep. India – Odisha (CESU) Dominican Republic Dominican Republic (Edenorte) Pakistan (KE) India – Odisha Egypt, Arab Rep. (NCEDC) 0 5 10 15 20 25 0 5 10 15 20 25 30 Percent Percent Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015 and World Bank World Development Indicators ­database 2018. Note: Dark-shaded parts of bars in figure show inefficiencies lost because of excessive transmission and distribution losses; light-shaded parts of bars show inefficiencies lost because of noncollection of bills. Countries/utilities with inefficiencies lower than 5 percent are considered good performers and are shaded green; countries/utilities with inefficiencies between 5 and 10 percent are considered moderate and are shaded yellow; countries/utilities with inefficiencies above 10 percent are poor performers and shaded red. a. ONEE (Morocco): collection rate data are not available. 250 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Even within specific countries, the variation in certain periods of time, cost volatility threat- efficiency performance can be huge, illustrated ens the sustainability of this outcome. This clearly by the case of India, with states such as finding is true for countries at both ends of the Odisha and Rajasthan losing about 20 percent of cost recovery spectrum (for example, for both revenues to inefficiency, and Andhra Pradesh Colombia and Tanzania), as well as countries losing less than 5 percent. By region, the cases that are completely reliant on fossil-fuels (for in East Asia and Pacific and Latin America and example, Dominican Republic) and countries the Caribbean perform best; the Middle East relying entirely on renewable energy (for and North Africa and South Asia perform worst. example, Tajikistan). See the full data in The key question then is who should bear the annex 8C. Year-on-year volatility of cost recov- cost of these inefficiencies—that is, to what ery is high in the Indian state of Rajasthan extent should these inefficiencies be included in (coefficient of variation of 16 ­ p ercent), tariffs as a cost to consumers or absorbed in sub- Pakistan (17 percent), Senegal (16 percent), sidies as a cost to society as a whole. In Pakistan, Tajikistan (19 percent), and Tanzania (23 per- the tariff-setting formula includes an allowance cent). The Indian states of Andhra Pradesh for normal T&D losses, but the excess losses are and Odisha are the two exceptions, with low in any case incorporated in circular debt when volatility in cost recovery levels (table 8.2). companies with scarce revenues fail to fully pay The sources of volatility most often observed their suppliers. The government is responsible are exchange rate, fuel cost, fuel mix, and debt for this debt in the long run through its owner- service costs. Colombia, for example, was above ship of state-owned companies. full capital cost recovery in 2011–16, but below Although more difficult to measure and that level in 2010 because of the impact of benchmark, inefficiencies in generation also fac- drought on hydropower costs. Kenya Power tor into the high cost. Generation accounts for and Lighting Company (KPLC) experienced most of the total cost of electricity supply. In worse financial performance in 2009–12 when addition, the least-cost planning and competi- drought reduced hydro capacity. The Tanzania tive procurement of generation encourages util- Electric Supply Company (TANESCO) has had ities to commit to suboptimal power purchase consistent losses in recent years but ran an oper- during periods of supply crisis. Both the ating profit in fiscal year 2013/14 (FY2013/14) Philippines and Senegal experienced a power and FY2014/15 because of favorable hydrologi- crisis (1990s and 2015, respectively). Their cal conditions that greatly reduced the cost of resulting power purchase agreements were sales and reduced the need for power purchases costly and have burdened the utilities for years. from third parties. Fuel prices have been a major Similarly, Tanzania’s reliance on noncompetitive factor in cost variations seen in the Philippines, procurement for generation projects has inflated which benefited from the decline of fuel prices costs and financially crippled the sector. since 2013. Senegal was also able to improve cost recovery despite tariffs having been frozen at 2009 levels, as fuel purchase costs declined Finding #4: The inherent volatility of 44 percent between 2012 and 2016. Because costs means few countries can sustain Tajikistan imports most materials and equip- cost recovery over time ment, Barki Tojik is susceptible to the devalua- None of the cases maintained full cost recov- tion of its local currency. ery throughout the observation period. Six of Despite consistent exposure to volatility, few the cases reached full cost recovery in at least utilities have any hedging strategy. Some utilities one of the observed years; however, even have opted for hydropower imports from neigh- when tariffs attain cost recovery levels for bor countries (Senegal and Vietnam) to reduce Did Power Sector Reforms Improve Efficiency and Cost Recovery? 251 TABLE 8.2 Full cost recovery is a moving target because of exogenous cost shocks Percent Country/state Average Min Max Coeff. of varb Main sources of volatility Colombia 104 94 117 7.0 Hydro availability, debt service costs, FEX rate Dominican Republic 66 62 73 7.0 Fuel prices Egypt, Arab Rep. 55 49 68 14.3 Gas availability, FEX rate, fuel prices India – Andhra Pradesh 38 36 44 8.9 Debt service costs India – Odisha 91 90 94 1.7 n.a. India – Rajasthan 45 38 57 16.0 Debt service costs Kenya 90 80 101 9.0 Hydro availability, capital costs Moroccoa 84 n.a. n.a. n.a. Fuel prices Pakistan 80 66 97 17.0 Tariff increases, fuel prices Peru 98 93 102 3.6 Fuel prices, hydro availability Philippines 98 95 100 1.7 Fuel prices, legacy PPAs Senegal 70 55 87 16.0 Fuel prices, expensive emergency power, debt service costs Tajikistan 67 52 83 18.7 FEX rate Tanzania 68 56 90 22.6 Hydro availability, FEX rate, fuel prices, expensive IPPs, emergency power, debt service costs Uganda 92 81 110 12.2 Capital costs, high rate of return for private utility, fuel prices, FEX rate Ukraine 106 93 115 7.1 FEX rate, fuel prices Vietnam 89 86 91 2.1 Hydro availability Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: Coefficients in bold highlight high year-on-year volatility. FEX = foreigh exchange; IPP = independent power producer; PPA = power purchase agreement; n.a. = not applicable. a. Data for Morocco are available for only one year (2013). b. Coefficient of variation (standard deviation divided by the geometric mean). reliance on costly oil-fired generation or during shortfall in cost recovery by utilities should major shortages. Going forward, many countries be covered through a compensating fiscal have committed to increasing the share of transfer. In practice, however, government ­ renewables in their energy mix, not only for support falls short of restoring utilities’ finan- their environmental benefits but also to enhance cial viability in almost all cases, which leaves the security of supply and diversify the fuel mix. utilities adopting other coping strategies to make ends meet financially (table 8.3). Most jurisdictions with below full cost recov- Finding #5: Financial losses due to the ery receive fiscal support in the form of opera- underrecovery of costs are seldom fully tional or capital transfers. Of the 11 countries compensated by government subsidies where fiscal support was registered, it was found but are instead absorbed through the to be more common in the jurisdictions with accumulation of arrears and short-term lower levels of cost recovery, and less common debts for private utilities. TANESCO, for example, Different mechanisms exist for absorbing the relies on government grants to finance underrecovery of costs, with different implica- ­ investments—totaling US$833 million in 2016 tions for the utilities’ ability to adequately (1.72 percent of GDP) and subsidized capital serve their customers. In principle, any from donor sources, reducing the average 252 TABLE 8.3 Utilities absorb financial shortfalls in various ways A1 Cost Large Large Sustained C3 Cost recovery government Subsidized Sustained debt Excessive Excessive negative operating Power utility Type recovery (%) (%) transfers fuels net losses service payables receivables cash flow Colombia (Codensa) Private 104 152 No No No No No No Yes Dominican Republic (Edesur) Public 66 67 No No Yes — — — Yes Egypt, Arab Rep. (EEHC) Public 55 135 No Yes No No Yes Yes No India – Andhra Pradesh (APEPDCL) Public 36 103 — No Yes No — — No India – Andhra Pradesh (APSPDCL) Public 42 100 — No Yes No — — No India – Odisha (CESU) Public 95 99 — No Yes No — — No India – Odisha (WESCO) Public 89 98 — No Yes No — — No India – Rajasthan (AVVN) Public 40 82 — No Yes Yes No No No India – Rajasthan (JDVVN) Public 48 81 — No Yes Yes No No No India – Rajasthan (JVVN) Public 56 87 — No Yes Yes No No No Kenya (KPLC) Public 90 117 No No No No No No Yes Morocco (ONEE) Public 84 — — — — — — — — Pakistan (KE) Public 86 92 — No No No Yes Yes No Pakistan (LESCO) Private 66 74 No No No No No No Yes Peru (Hidrandina) Private 102 135 — No No Yes No — No Peru (Luz del Sur) Public 87 127 — No No No No — No Philippines (Beneco) Private 99 111 — No No No No No Yes Philippines (Meralco) Private 89 101 — No No No No No No Senegal (Senelec) Public 70 100 No Yes No No Yes No No Tajikistan (Barki Tojik) Public 67 118 No No Yes Yes Yes No No Tanzania (TANESCO) Public 74 90 Yes No Yes Yes Yes No No Uganda (UMEME) Private 92 126 No No No No No No No Ukraine (Khmelnytskoblenergo) Mixed 109 113 — Yes — — — — — Vietnam (NPC) Public 89 101 No No No No — — No Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: Large government transfers = If total government transfers exceed 10 percent of C3 cost recovery. Subsidized fuels = If utility receives fuels below market prices (qualitative information). ­Sustained net losses = net losses exceed 5 percent of revenue (average of last three years). Large debt service = total debt service (interest and principal payments) exceed 20 percent of C3 cost recovery. Excessive payables = payables exceed 50 percent of annual revenues. Excessive receivables = receivables exceed 50 percent of annual revenues. Sustained negative operating cash negative net operating cash flow of more than 5 percent of revenue. Refer to table 8.1 for full definitions of A1 and C3 cost recovery. — = not available. flow = ­ Did Power Sector Reforms Improve Efficiency and Cost Recovery? 253 interest rates on TANESCO’s total borrowings Insufficient government transfers are aggra- from 5 percent in 2012 to 3 percent in 2016. In vated when public institutions don’t pay their addition, TANESCO does not always repay the electricity bills. Although utilities may benefit ministry of finance for its loans, which then from fiscal transfers from the finance ministry, become pseudogrants. Pakistan’s privatized they are vulnerable as well to nonpayment of Karachi Electric (KE) relies extensively on oper- regular electricity bills by a range of public ational transfers. Subsidies are provided to dis- institutions, including central government tribution companies in the form of a tariff differ- departments, state-owned enterprises, and ential subsidy, totaling US$418 million to KE in subnational jurisdictions. The government is a 2015, which compensates distribution compa- major contributor to receivables in Pakistan, nies for the difference between the regula- Senegal, and Tanzania. The issue of nonpay- tor-determined cost-based tariff (accounting for ment is especially important in Pakistan, where only efficient costs) and the uniform tariffs collection losses make up 47 percent of the (based on the costs of the most efficient distribu- QFD. Pakistan has the highest number of tion company). The government envisioned receivable days in the country sample (190 that, as distribution companies were privatized, days). As an illustration, Pakistan’s KE is con- the efficiencies of private management would tractually obligated to provide uninterrupted result in lower costs and lower subsidies. After service to Karachi Water & Sewerage Board almost 10 years of privatization, however, KE and City District Government Karachi, but still receives a subsidy. Total tariff differential their unpaid bills have been accumulating subsidies provided to the sector in 2016 (includ- since before 2010. Taken together, government ing KE and ex-Wapda [Water and Power and autonomous bodies make up 56 percent Development Authority] distribution compa- of KE’s trade receivables. In Tanzania, govern- nies, XWDISCOs) represented 0.4 percent of ment nonpayment used to be a major problem, GDP. Other governments also provide subsidies but the mainland government and TANESCO in the form of direct transfers to suppliers for recently took steps to slash accounts payable, fuel cost or for available capacity payment settling US$71 m ­ illion in unpaid invoices in (Senegal and Uganda). 2015. Collections losses make up 6 percent of Fiscal support meant to compensate for Tanzania’s QFD. shortfalls in tariff revenue often falls short of Utilities receiving no government support or required levels. Although fiscal transfers are facing large unpaid bills from public sector provided in most countries that fall short of institutions must use other coping strategies to cost recovery, rarely are these transfers large deal with underpricing. Often nonpayment enough to fully compensate utilities for the means utilities fall into arrears with suppliers, financial shortfall. In Senegal, for example, tar- creating contingent liabilities for the govern- iffs were frozen at their 2009 level, with the ment. Typically, energy providers and goods government agreeing to make quarterly pay- and services providers make up the bulk of a ments to compensate the national electric utility’s payables. Pakistan experiences this company, Senelec. These payments were not problem of “circular debt.” Distribution compa- timely, however, causing Senelec to take on nies often do not have the cash to pay the costly commercial debt. In 2011, Senelec’s con- National Transmission and Dispatch Company cession contract was updated so that, if the (NTDC) because of low collections, low cost government is unable to make payments, mak- recovery (even with the subsidies), or lack of ing Senelec borrow from commercial banks, timely payment of subsidies. NTDC then can- the government must then assume responsibil- not pay power producers, and power produc- ity for financial fees and principal repayment. ers cannot pay fuel suppliers. Tanzania has had 254 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD similar difficulty with its loan payments, with long-term loans but use some short-term payable days of 299. financing, which tends to be much more These issues of financial viability affect utili- expensive. For example, KPLC has one short- ties’ creditworthiness and ability to raise capi- term loan on record with an interest rate of 16 tal. As a result, many utilities in developing percent, well above its average commercial countries are overly reliant on high-cost, short- loan rate of 3.6 percent (otherwise comprising term debt, which they then can’t service. Debt medium- and long-term loans). repayment has been particularly problematic Overall, utilities benefit from substantial for utilities with high investment needs owing subsidies on investment financing. These subsi- to electrification (Kenya and Tanzania). dies lower the effective cost of capital relative TANESCO’s inability to pay its debts is apparent to the true cost of either public or private in its low debt–service coverage ratio (0.16 in finance. Weak financial performance persists FY2014/15 and –0.11 in FY2015/16). In 2015, despite the fact that utilities benefit conces- TANESCO’s financial reports also show that it sional sources of finance. The average cost of defaulted on government loans and World debt is 6 percent across the 13 jurisdictions Bank loans on-lent by the finance ministry, with available data. This cost of debt is less or about 19 percent of its 2016 capital than half the average commercial rate for these expenditure. For countries with detailed loan ­ jurisdictions (14 percent). In fact, every information (Kenya, Senegal, and Tajikistan), jurisdiction except Rajasthan in India receives ­ the average tenor of debt is between 5 and lower-cost loans than the average commercial 13 years. These countries rely primarily on borrowing rate in each country (see figure 8.5). FIGURE 8.5 Utilities often borrow at rates well below commercial benchmarks 30 26 25 24 Average borrowing rate (%) 20 19 16 17 15 15 14 13 13 10 10 10 10 8 7 7 7 6 6 6 5 6 5 4 4 3 3 4 0 a . s al m u da a an a sh a n p ne ni ny sh bi r ha g Re na Pe e ist an a ne m di pi ad Ke st nz et jik Ug lo b O ilip ja Se Pr Vi ra Ta Co Ta Ra – Ph ,A a a r – di dh t yp a In An di Eg In – a di In Average borrowing rate for the utilities Average commercial borrowing rate for the country Public distribution company Private distribution company Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015 and World Bank World Development Indicators database 2018. ­ Did Power Sector Reforms Improve Efficiency and Cost Recovery? 255 Government and international financial insti- jurisdictions at similar levels of cost recovery tution (IFI) loans have lower rates than the (Kenya, Senegal, Uganda, and Vietnam). commercial loans, as shown in figure 8.6 Tanzania maintains its high levels of invest- (except for Senegal’s IFI loans, most of which ment through government grants (totaling are from the West African Development Bank US$833 million, or 48 percent of investments and have a standard 8.5 percent rate). Uganda in 2016), in addition to loans, which it doesn’t is entirely dependent on IFI loans, while the always pay and which become pseudogrants. bulk of Senegal’s and Tajikistan’s loans are Odisha’s low levels of investment (14 percent from IFIs, and the lion’s share of Kenya’s and of revenues for CESU [Central Electricity Tanzania’s loans are from commercial sources. Supply Utility] and 1 percent for WESCO Vietnam Electricity (EVN) benefits from prefer- [Western Electricity Supply Company] in ential treatment, such as privileged access to 2015) are attributed to the utilities being credit, land, and contracts. annual loss-makers. Public investment subsidies are often the In line with the above, the qualitative evi- main driver of utilities’ investment levels, as dence suggests that countries that have mobi- opposed to financial performance. The cases lized large investment amounts in recent suggest no clear trend between investment lev- years—Kenya, Tanzania, and Uganda—have els (as a percentage of revenues) and cost done so through public investment that was recovery (see figure 8.7). The data presented in mobilized despite underrecovery of costs. In figure 8.7 show, for example, that Tanzania can line with this observation, full cost recovery invest at high levels despite low cost recovery, levels appear largely unrelated to progress in whereas the Indian state of Odisha is investing electrification, indicating a strong role for pub- at relatively low levels compared with other lic investment. FIGURE 8.6 International financial institutions are typically the cheapest source of borrowing 25 23.0 20 Average borrowing rate (%) 15 10 9.0 8.4 7.5 6.6 5 3.6 4.2 4.0 2.93 0.6 0 Tanzania Kenya Senegal Tajikistan Uganda Share of total loans (%) Government Commercial International financial institution Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015 and World Bank World Development Indicators ­database 2018. 256 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.7 Investment levels do not bear much relationship to cost recovery 80 70 R = 0.0055 Tanzania Investment as percentage of revenue (%) 60 50 40 Kenya 30 Egypt, Arab Rep. Uganda 20 India – Rajasthan Vietnam Senegal Colombia 10 Peru Dominican Republic Philippines India – Andhra Pradesh Tajikistan India – Odisha 0 30 40 50 60 70 80 90 100 110 120 Full cost recovery (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Finding #6: Power utilities often impose which suggests that tariff reforms or cost reduc- hefty fiscal burdens and contingent tions would be needed to reach cost recovery. liabilities on government budgets Underpricing is only one of several factors ­ contributing to the relatively poor performance The fiscal burden associated with the power on cost recovery of many cases. The two other sector ranges both above and below one per- factors are excessive T&D losses and the non- centage point of GDP. Estimates of the QFD of collection of bills expressed as a percentage of power utilities—or hidden losses in the case of total utility revenues. privately owned utilities—provide a strong reminder of the macroeconomic significance of the underrecovery of the cost of electricity ser- Finding #7: Tariff levels are highly vice. The QFD of the power sector stands at differentiated in many cases to make 0.93 percent of GDP on average (see ­figure 8.8, service affordable to certain consumer panel a). Again, these figures are broadly in groups; large cross-subsidies are often line with the literature (see annex 8A), sug- associated with low levels of cost gesting that the sample is representative of recovery developing countries. Separating the analysis by ownership of the Because of the extensive practice of cross-­ utility suggests that QFD/hidden costs are con- subsidies in tariff structures,7 industrial and centrated in state-owned utilities (see figure commercial users are much more likely than 8.8, panel b). The average QFD of sectors with residential and agricultural ones to pay at cost publicly owned distribution is 1.63 percent of recovery levels. Systematically, across jurisdic- GDP, compared to 0.16 percent for mixed pub- tions, industrial and commercial customers lic/private and 0.11 percent for fully p ­ rivate often pay a hefty tariff premium even though distribution utilities. the costs they impose on the network are no Below-cost-recovery tariffs are the leading greater (and are potentially lower) than contributor to the QFD in 9 out of 17 cases, those imposed by residential customers. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 257 FIGURE 8.8 Underpricing is the largest driver of quasi-fiscal deficits in the power sector 5 a. By source of deficit or hidden cost 4 3 Share of GDP (%) 2 1 0 –1 h* * an p. l ia ic co an a* a m da s ru a e ga an ne ny bi in bl an Re na sh Pe es oc ist st an ra ne m th pi pu Ke nz ki di ad et jik or Ug lo Uk ab ilip as Se Pa Re -O Vi Ta Co Pr M Ta aj Ar Ph -R an ra a t, di dh ic yp a In in di An Eg m In Do – a di In Collection losses T&D losses Underpricing b. By ownership type of utility 5 4 3 Share of GDP (%) 2 1 0 –1 * n* an p. al ia ic co an a* a m da s ru a e sh ne ny bi in bl eg an Re na sh Pe ha oc ist st an ra m de pi pu Ke nz ki n di et st jik or Ug lo Uk ab ilip Se a Pa Re -O Vi ja Ta Co Pr M Ta Ar Ra Ph an ra a t, di – dh ic yp In a in An di Eg m In Do – a di In Private Public Mixed public/private Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: T&D = transmission and distribution. Fewer countries have cross-subsidies from percent household) affordable to the bottom 40 ­ industrial to commercial customers, and these of the income spectrum. In eight cases—Egypt; cross-subsidies tend to be smaller. Several the Indian states of Andhra Pradesh, Odisha, countries use such cross-subsidies to make and Rajasthan; Pakistan; Senegal; Tajikistan; electricity affordable to politically favored and Ukraine—this ­ consumption costs less than groups, typically including but not limited to 1 percent of gross national income of the bot- the poor and vulnerable. As can be seen in fig- tom 40 percent. The Philippines and Uganda ure 8.9, several countries manage to make sub- stand out as being the only countries where sistence consumption (30 kWh per month per subsistence consumption can absorb as much 258 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.9 Many countries seem able to reconcile cost recovery and affordability 8 Philippines Uganda 7 Cost of 30 kWh/month as share of GNI for bottom 40 (%) 6 5 4 Kenya 3 Tanzania 2 Colombia Dominican Republic Vietnam Morocco Peru 1 India – Andhra Pradesh Tajikistan Pakistan Ukraine India – Rajasthan Egypt, Arab Rep. India – Odisha 0 Senegal 30 40 50 60 70 80 90 100 110 120 Full cost recovery (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: Bottom 40 = the bottom 40 percent of the income distribution; GNI = gross national income; kWh = kilowatt-hour. as 7–8 percent of the gross national income of This even distribution does not mean that cost the bottom 40 percent of the income distribu- recovery is being met or that there are no tion. Such cross-subsidies help governments cross-subsidies; it only means that all customer maintain affordable and politically legitimate classes are paying similar tariff rates for their power sector reforms. consumption. Some jurisdictions with flatter The critical question is at what point such distributions may not be recovering costs price discrimination across customer groups through tariffs on any customer class (as is the starts impeding cost recovery. One way to case with Rajasthan in India). In addition, it is quantify price discrimination is to use a Lorenz appropriate for customers who impose lower curve, which shows cumulative shares of con- costs on the system (such as industrial custom- sumption and revenue by customer class (see ers) to pay lower tariffs than other customer figure 8.10).8 A straight diagonal would mean groups, so a completely flat curve is not ideal. that all customers contribute shares of revenue A noticeable curvature, however, points to a equal to their shares of consumption, and misalignment of costs incurred to serve each major curvature would mean that customers’ customer group and the customers who pay shares of revenues are not well aligned with those costs. The three Andhra Pradesh (Indian) their shares of consumption. Many utilities distribution companies show the greatest show a relatively even distribution of shares of ­ disparity in the shares of consumption and revenue and consumption by customer class. ­ revenue contributed by each customer class. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 259 FIGURE 8.10 Cross-subsidies are pronounced in some countries and states 100 75 Share of revenue (%) 50 25 0 25 50 75 100 Share of consumption (%) Colombia India – Odisha (CESCO) Pakistan (KE) Tanzania Egypt, Arab Rep. India – Rajasthan (AVVNL) Peru (Luz del Sur) Uganda India – Andhra Pradesh (APSPDCL) India – Rajasthan (JDVVNL) Peru (Hidrandina) Ukraine (Dniprooblenergo) India – Andhra Pradesh (APEPDCL) India – Rajasthan (JVVNL) Philippines (BENECO) Ukraine India – Andhra Pradesh (APNPDCL) Kenya Philippines (MERALCO) (Khmelnytskoblenergo) India – Odisha (WESCO) Pakistan (LESCO) Senegal Vietnam Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: The level of cross-subsidization is quantified here as the area between the 45-degree line and the curve defined by the cumu- lative shares of consumption and revenue by customer class. The formula is the same as that of a Gini-coefficient of inequality. Data for latest available year. The distribution companies of Egypt, Tanzania, methodology defined earlier that measures the and Ukraine, and Pakistan’s KE also show area between the 45-degree line and the Lorenz curves that bow out more noticeably than the curve representing consumption and revenue rest of the sample. shares. Specifically, the analysis suggests that no There appears to be a limit of about country with a cross-subsidization indicator 15 ­­ percent on the degree of price differentiation above 15 percent comes even close to full cost among customer categories, which is compatible recovery (all five cases are below 80 percent; see with cost recovery. This limit is based on the figure 8.11). Egypt, India (Andhra Pradesh), 260 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.11 Higher levels of cross-subsidy are associated with lower levels of cost recovery 120 Colombia Ukraine 100 Philippines (MERALCO) Vietnam Peru India – Odisha (WESCO) Uganda India – Odisha (CESCO) Kenya 80 Pakistan (LESCO) Philippines Full cost recovery (%) Tanzania (BENECO) Senegal Pakistan (KE) 60 R2 = 0.32 India – Rajasthan (JVVN) Egypt, Arab Rep. India – Rajasthan (JDVVN) India – Andhra Pradesh (APEPDCL) 40 India – Rajasthan (AVVN) India – Andhra Pradesh (APSPDCL) 20 0 5 10 15 20 25 30 35 40 45 Level of cross-subsidization (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Pakistan, and Tanzania all feature large levels. KPLC has a uniform tariff for all domestic cross-subsidies and low cost recovery; the oppo- customers, despite rapid expansion of its net- site is true for Colombia, the Philippines, and work into lower-demand regions, with over a Vietnam. The Indian state of Rajasthan, with million new customers connected in both the low cross-subsidization and low cost recovery, is 2015–16 and 2016–17 financial years. a particular exception. This suggests that limit- Nevertheless, several jurisdictions that had ing cross-subsidization is a necessary but not reached full cost recovery preserved major life- sufficient condition for cost recovery. lines to protect low-income consumers via an Also challenging from a financial perspective increasing block tariff. Cross-subsidization are geographical cross-subsidies that preserve a aimed at protecting the poor often takes the uniform tariff across areas facing differential form of so-called lifeline tariffs up to a certain costs of supply. Another form of cross-subsidiza- number of kWhs per month. Notably, several tion is to maintain uniform tariffs across a coun- of the countries with the highest lifeline tariff try’s regions and distribution companies with block thresholds, such as Colombia (130–173 differences in cost of service. These forms of kWh), the Philippines (100 kWh), and Ukraine implicit cross-subsidization are common because (100 kWh or higher, based on household type) differentiated tariffs, particularly across urban are all at the high end of cost recovery. This and rural areas, often pose challenges to rural suggests that targeted cross-­ subsidies to protect electrification because the countryside has the poor can be part of a sustainable electricity higher connection costs and lower income pricing strategy. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 261 Electricity tariff structures avoid incentivizing recovery levels if real tariffs remained the same and grid defection in view of recent developments in actual costs materialized exactly as anticipated in distributed energy. Most low-income countries estimates of long-run marginal cost (LRMC). rely on energy charges to recover costs. Comparing these benchmarks to actual cost Consumers can self-generate to save almost recovery levels in 2010–17 allows us to draw their entire electricity bill, while benefiting from conclusions about (1) the change in electricity the grid’s backup services. If fixed costs are tariffs in real terms compared to the pre­ reform recovered through kWh energy charges, rather period; (2) if actual costs are now higher or than reflected in a separate fixed charge, each lower than expected in the LRMC estimates grid defection shifts these costs onto a smaller from the late 1980s and 90s; and (3) if the group of customers and further incentivizes grid combination of changes in real tariffs com- defection. Given that the costs of distributed pared to the prereform actuals and changes in generation may be even lower for nonresiden- real costs compared to prereform estimates of tial customers consuming at larger scales, LRMC led to a net increase or decrease in cost cross-subsidies further exacerbate incentives for recovery compared to the counterfactuals. grid defection. Odisha (India) illustrates the role There is evidence of some convergence of cross-subsidies in grid defection. After privat- in full cost recovery levels across coun- ization, support to low-income residential cus- tries (­figure 8.12). As shown in figure 8.13, tomers in the form of free connections was reduced in favor of cross-subsidies from indus- FIGURE 8.12 Only a handful of countries and states had trial customers, which led the industrial custom- achieved full cost recovery prior to reform ers to seek alternative sources to avoid paying Philippines 4 out of 17 higher tariffs. This sort of grid defection, com- ≥ 100% bined with residential tariffs that are kept artifi- Morocco cially low, can put utilities in increasing financial Kenya distress because they are unable to recover their Dominican Republic 7 out of 17 costs of service. ≥ 80% Pakistan Senegal Finding #8: Cost recovery levels have Tanzania risen on average, but progress has been Peru uneven, with more than half of case studies showing a drop compared with Vietnam the prereform period Colombia Although cost recovery remains challenging, India – Rajasthan a key question is whether it it has improved India – Odisha since the 1990s reforms. To understand the India – Andhra Pradesh impact of reforms, this section compares pre- Uganda reform cost recovery benchmarks from the late 1980s and early 1990s with cost recovery Tajikistan for 2010–17. As laid out in box 8.1, the pre­ Egypt, Arab Rep. reform benchmarks are based on a compari- Ukraine son of actual prereform tariffs and prereform 0 20 40 60 80 100 120 140 estimates of long-run marginal costs. The pre- Full cost recovery (%) reform cost recovery benchmarks can be Source: World Bank elaboration based on Rethinking Power Sector Reform utility understood as a counterfactual of today’s cost database 2015. 262 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 8.13 Prereform cost recovery was not necessarily sustained over time 140 R2 = 0.0128 120 Ukraine Colombia Philippines 100 Peru Postreform full cost recovery (%) Uganda Vietnam Kenya India – Odisha Morocco 80 Pakistan Senegal Tajikistan Dominican Republic Tanzania 60 Egypt, Arab Rep. India – Rajasthan 40 India – Andhra Pradesh 20 0 20 40 60 80 100 120 140 Prereform full cost recovery (%) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. the largest improvements were made by coun- Underlying drivers vary considerably across tries with low cost recovery in the prereform countries (­figure 8.14). For example, Kenya saw period, and vice versa (compare to figure 8.12). cost recovery decline despite hefty tariff The largest improvements in cost recovery increases because costs increased even more were observed in Europe and Central Asia rapidly. In Peru, by contrast, cost recovery (Ukraine +92 percentage points and Tajikistan improved despite a drop in real tariffs, because +43 percentage points) and Latin America and costs came down even faster. These findings are the Caribbean (Colombia +49; Peru +22), with broadly in line with the literature, which sug- South Asia and East Asia and the Pacific see- gests that, despite greater awareness about the ing, on average, relatively little change com- broad adverse impacts of electricity subsidies, pared to the prereform period. the aggregate level of cost recovery and financial Although the average level of cost recovery viability in developing countries has improved improved slightly since 1990, the record was only slightly between the late 1980s and the decidedly mixed, with about half of countries early 2010s (Huenteler and others 2017). seeing an improvement in cost recovery and the Notably, the average improvement in cost other half a deterioration. Average full cost recovery from 69 percent around 1990 to 79 recovery increased from 69 percent around percent in 2010–17 was largely driven by cost 1990 to 79 percent in 2010–17, but the reductions rather than tariff increases. In princi- increase was driven by a few strong perform- ple, improvements in cost recovery may result ers and over half saw a decline (9 out of 17). either from a reduction in costs or an increase in Did Power Sector Reforms Improve Efficiency and Cost Recovery? 263 tariffs, and these two effects may be disaggre- expand its network despite poor cash flow. Its gated (figure 8.14). Average real tariffs fell investments totaled US$481 million in 2016 slightly in all 17 cases from US$0.122/kWh to (about 45 percent of its revenue). US$0.119/kWh between the prereform (1990) Another cause of decline in cost recovery is a and postreform (2010–17) periods—or about reversal of tariff reforms or stalled tariff increases 2.5 percent on average. Over the same time due to sociopolitical pressure. The government period, costs fell slightly more on average in real of Senegal has been reluctant to authorize tariff terms from US$0.168/kWh to US$0.156/ increases and instead subsidizes the national util- kWh—a drop of about 7 percent. Thus, observed ity, Senelec, for the difference between tariffs improvements in cost recovery owe more to and the cost of service determined at quarterly cost reductions than to tariff increases. tariff revisions, equivalent to 18 percent of full In countries where cost recovery deterio- cost recovery for Senelec. These subsidies were rated, large investment programs promoting not provided in 2015 and 2016, and in 2017 tar- universal electrification were a contributing iffs were lowered by 10 percent. In the Indian f actor. Tanzania’s TANESCO, which saw a ­ state of Andhra Pradesh, where cost recovery 20 percent decline in cost recovery over this declined by 16 percentage points, the regulator period, has made increasingly large investments was unable to increase tariffs from 2004 to 2010, in recent years, totaling US$435.6 million in and in 2004 the government also announced a 2015–16 (69 percent of revenues) to fulfill its policy of free power to agriculture. requirements of funding distribution expansion The largest improvements in cost recovery and some of the cost of new connections. These compared to the prereform period were the large investments coupled with low tariffs result of both cost reductions and real tariff resulted in accounts payable exceeding reve- increases: Five out of eight countries with nues in 2015–16. Kenya’s KPLC, which faced a improved cost recovery, including three out of 16 percent decline in cost recovery, finds itself in the four best performers, witnessed both a hike a similar situation, making hefty investments to in tariffs and a fall in costs. Findings are in line FIGURE 8.14 Postreform cost recovery improved in about half of cases and deteriorated in half 100 80 8 out of 17 cases with net 60 improvement in 40 cost recovery Percent 20 0 9 out of 17 cases –20 with deterioration –40 in cost recovery –60 –80 Ph blic s co ia l ya Pa h Ra an n m ru . a Co an a da e ep ga ne sh bi in es ha an na Pe n oc ist ist an ra ne R m di pi ad u Ke st nz et p k jik or Ug lo Uk ab O ilip ja Se Pr Re Vi Ta M Ta – Ar ra a an – di t, dh ic yp a In An di in Eg m In – Do a di In Changes in tari levels Di erence between realized costs and LRMC Net change Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: LRMC = long-run marginal cost. 264 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD with the literature (see annex 8A). Notably, service delivery will reduce costs and, if tariffs cost recovery improved in four cases—Egypt, are not yet at cost recovery level, improved India, Peru, and Vietnam—without large quality of service eases the sociopolitical pres- increases in real tariffs. Kenya, conversely, saw sure that might otherwise suppress increases. cost recovery drop despite major tariff hikes. Similarly, the 2012 tariff increase in Tanzania Reforms on cost recovery were often was possible under emergency procedures prompted by crisis. Countries started to adopt because of drought. In Pakistan, the 2011 cir- drastic cost reductions and tariff increases after cular debt was crippling the energy sector as having experienced large power deficits that independent power producers threatened to require extensive investments that the govern- call in sovereign guarantees because of non- ment can no longer support alone. Uganda’s payment, pressuring the government to raise UMEME (+64-percentage-point increase in tariffs for the higher consumer blocks. cost recovery) has taken on aggressive invest- ments (totaling US$93 million in 2016), allow- Finding #9: Tariff erosion brought on by ing it to upgrade the distribution grid to keep inflation outweighed tariff increases in up with growing access and demand. Uganda is most cases also improving operating efficiency and reduc- ing losses and operating costs. Uganda’s tariffs Real tariffs declined on average as nominal have allowed UMEME to keep up with debt tariff increases did not keep pace with ­ inflation service payments, because tariffs are updated in most cases. As mentioned in the previous annually and subject to quarterly automatic section, average real tariffs dipped in all 17 adjustments for inflation, exchange rate, and cases from US$0.122/kWh to US$0.119/kWh oil price fluctuations. Investments that improve between the prereform (1990) and postreform service delivery are vital to sustain cost recov- (2010–17) periods, or about 2.5 percent on ery improvements, because efficiencies in average. Figure 8.15 breaks down the change FIGURE 8.15 Nominal tariff increases were largely eroded by inflationary pressures 0.15 0.10 0.05 US$/kWh 0 –0.05 –0.10 –0.15 –0.20 u s ic co a a an a a m ne an l n a p. h ga ne nd sh bi ny i r ha es l an Re na Pe ub oc ist ai ist m ne pi di ad Ke st a r nz et p k jik or Ug lo Uk b lip O Se ja Pa Pr Re Vi ra Ta Co M Ta Ra – i Ph ,A ra a an – di dh t ic yp a In di in An Eg m In – Do a di In Change in nominal terms Erosion due to inflation Net change Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: kWh = kilowatt-hour. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 265 in real tariffs into inflationary effects and tariff FIGURE 8.16 In most cases, system losses declined increases in nominal U.S. dollars. The graph substantially over time shows that real tariffs fell in 9 cases and India – Andhra Pradesh increased in 8—this despite a nominal tariff Vietnam increase, on average, in every one of the 17 India – Rajasthan observed cases. In Egypt and Tanzania, cur- Peru rency devaluation reduced the tariff value by 100 and 21 percent, respectively. Colombia Dominican Republic Finding #10: Declining transmission and Philippines distribution losses have led to major Uganda improvements in cost recovery Senegal Kenya Average system losses fell from 24 percent in the prereform era to 17 percent in the 2010– Morocco 15 period, contributing to the decline in aver- Tajikistan age costs across the case studies (figure 8.16). Pakistan In contrast to the story for cost recovery, where Tanzania roughly half the sample improved and half Ukraine deteriorated, there is a striking consistency Egypt, Arab Rep. with 14 of the 17 jurisdictions showing reduced India – Odisha system losses and only 1 jurisdiction (India – –40 –30 –20 –10 0 10 20 Odisha) showing any sizable deterioration. Change in average system losses compared to prereform (%) Even this instance can be attributed to artifi- Reduction in system losses cially low numbers expressed by the govern- Increase in system losses ment in the runup to reforms. The state Source: World Bank elaboration based on World Bank World Development government estimated losses at 29 percent in ­ Indicators database 2018. 1994, although some analysis done postreform estimated them to be closer to 40 percent. the sample cases. Had T&D losses remained at The largest improvements in system losses prereform levels for all countries, the average were observed in East Asia and Pacific and cost of service across the cases would be Latin America and the Caribbean, followed US$0.171/kWh, as compared to the average closely by South Asia, though absolute levels prereform cost of US$0.168 and average postre- are still high in South Asia. The cases from form cost of US$0.156. One of the policy tools Sub-Saharan Africa showed relatively little to improve performance has been the setting of improvement, in contrast. It is important to target losses under the utility’s calculation of note that system loss reduction often requires revenue requirement. Privatization, and major investments in the network; without performance-based regulation, was a contribu- ­ external financing or government subsidies, tor to the reduction of losses. In Colombia, loss-making utilities are then unable to main- losses fell from 25 to 19 percent in the first year tain or improve the network, further under- of privatization. Losses in Uganda came down mining their cost recovery, as is the case for from about 38 percent in the mid-2000s to Kenya, India (Odisha), and Senegal. 17 percent in 2017. Privatization also helped The reduction in transmission and distribu- keep losses in the Philippines below 10 percent. tion losses decreased the full economic cost of This policy has not always been successful, how- service by an average of US$0.015/kWh across ever: Pakistan is an example where unbundling 266 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD and privatization brought limited improvements Of the 17 jurisdictions, only 2 had achieved full in efficiency because the incentive for loss economic cost recovery (C3) for the power sec- reduction was weak. In India (Odisha), losses tor as of 2010–17, and as many as 7 jurisdic- were underestimated when distribution compa- tions had not yet achieved even operating cost nies were transferred to private licensees; they recovery (C1). Even when cost recovery is continue at a high level (39 percent in 2015). In attained, it remains permanently vulnerable to Senegal, confusion over ownership of T&D cost shocks arising from higher oil prices and assets also contributed to high losses. exchange rate fluctuations, as well as supply The effect of variability in the cost of supply shocks in the form of drought. Cross-subsidies is larger than the effect of T&D loss changes in among customer classes further undermine most cases. Figure 8.17 breaks down the net cost recovery, although the experience of some change in average full cost of service compared utilities suggests it is possible to attain cost to the prereform period into the effect of T&D recovery while offering limited discounts for loss reduction and changes in the cost of sup- so-called lifeline consumption by residential ply. On average, the decline due to the reduced users. Although most utilities experiencing a T&D losses was outweighed by the higher shortfall in cost recovery receive some degree average cost due to other factors in the of fiscal transfer from the state, in the form of observed cases (figure 8.17). operating or capital subsidies, these are seldom high enough to fully compensate for the short- CONCLUSION fall and are often at least partially offset by the Strong performance on efficiency and cost failure of many public institutions to pay their recovery remains the exception, not the norm, electricity bills. As a result, utilities often find among power utilities in developing countries. their financial viability compromised, so they FIGURE 8.17 Improvements in system losses were often outweighed by other cost movements 0.25 0.20 0.15 0.10 US$/kWh 0.05 0 –0.05 –0.10 –0.15 a ic a s sh an n e co da m a l ru a an p. ga ne bi ny ish ni in ha bl Re na Pe e oc st ist an a ra om ne pi ad pu Ke st d nz ki et jik or Ug Uk ab O lip ja Se Pa Pr Re l Vi Ta Co M Ta Ra – Ar i Ph a a an r – di t, dh ic yp a In An di in Eg m In – Do a di In Change in cost of supply Change in T&D losses Net change Sources: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015 and World Bank World Development Indicators database 2018. ­ Note: kWh = kilowatt-hour; T&D = transmission and distribution. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 267 resort to a range of coping strategies including recovery improve, whereas the other half saw costly short-term borrowing from commercial a deterioration. Indeed, average real tariffs fell banks or informal borrowing in the form of slightly over this time period by about payment arrears to various kinds of suppliers. 2.5 percentage points, even if average real costs Over time, greater progress has been made fell slightly faster at 7 percentage points. Thus, on improving efficiency than on recovering the limited progress on cost recovery over this costs; indeed, countries making greatest prog- period owed more to cost reductions than to ress on cost recovery have done so at least in tariff increases. More encouragingly, over the part by controlling costs. On aggregate, prog- same period, system losses fell from 24 to 17 ress toward cost recovery has been relatively percentage points with almost all jurisdictions ­ modest—from about 69 percent in the pre­ registering sizable improvements. Nevertheless, reform period (1990) to 79 percent in the post­ only about half the utilities in the sample had reform period (2010–17). Moreover, only achieved system losses commensurate with about half of the jurisdictions saw cost best-practice levels. ANNEX 8A. MAJOR STUDIES OF COST RECOVERY AND FINANCIAL VIABILITY IN THE POWER SECTOR IN DEVELOPING COUNTRIES Serial No. Study Coverage Time KPIs Main findings Observed trends 1 World Bank Argentina, 1955–70 Rate of return on All 10 analyzed utilities Significant (1972) Brazil, Colombia, assets (based on were profitable during the improvements in 1960s Ethiopia, Ghana, utility financial observation period, with Malaysia, Mexico, statements) return on assets mostly in the Singapore 8–9% range 2 Munasinghe, Recipient utilities 1966–84 Four financial Average rate of return for the Distinct deterioration Gilling, of 123 World Bank ratios (based on period 1966–85 was 7.9 in the trend of utilities’ and Mason power projects utility financial financial ratios for the (1989) worldwide statements) period 1973–85 3 World Bank 60 developing 1979–88, with Comparison of Tariffs on (weighted) average Real average tariffs (1990) countries LRMC for 1990s existing tariffs sufficient to recover 62% of constant in 1979–83, worldwide, to LRMC with LRMC; average tariff level then fell sharply until comparison to shadow prices 55% of the average level in 1988 OECD OECD countries 4 IEA (1999) China, India, 1998 Price gap Cost recovery ratio ranged n.a. Indonesia, Islamic between tariffs between 37% (Venezuela) Republic of Iran, and reference and >100% (Indonesia); Kazakhstan, price (LRMC average: 62.3% Russian based on current Federation, South fuel mix) Africa, Venezuela, 5 Foster 83 OECD and 1994–2002 Average tariff 15% of countries did not Slight real increase in and Yepes non-OECD compared cover O&M costs; 59% did tariffs in some regions (2006) countries to global not cover total cost; strong but no significant trend worldwide benchmark correlation with income per across sample values capita 6 Ebinger 20 countries in 2000–03 Disaggregated QFD between 0 (Belarus) Decline in the QFD in (2006) Eastern Europe QFD: T&D losses, and 16.53% (Tajikistan) in 17 out of 20 countries and Central Asia collection losses, 2003; mostly driven by between 2000 and underpricing underpricing (67%) 2003, by 48% (from US$30 billion to US$16 billion overall) ( Annex continued next page) 268 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Serial No. Study Coverage Time KPIs Main findings Observed trends 7 Saavalainen 8 countries in 2002 Disaggregated Cost recovery between 11.21% n.a. and ten Eastern Europe QFD: T&D losses, and 81.6%; QFD between 1.1% Berge and Central Asia collection losses, and 21.4% of GDP (2006) underpricing 8 Eberhard 21 Sub-Saharan 2001–05 Average tariff Despite comparatively high Real tariffs almost and others African countries revenues power prices only 57% doubled over the period (2008) compared to of Sub-Saharan African 2001–05, but cost average historical countries recovered OPEXs; recovery ratio declined cost, LRMC 36% recovered LRMC 9 Briceño- 20 Sub-Saharan 2006 Disaggregated 6 out of 20 countries n.a. Garmendia, African countries QFD: T&D losses, recovered average historic Smits, and collection losses, costs; hidden costs of power Foster underpricing mispricing amount to about (2009) 1% of GDP or 60% of total hidden costs 10 Briceño- 27 Sub-Saharan 2004–08 Average 80% countries recovered n.a. Garmendia African countries (latest effective tariff OPEX; 30% also recovered and available) and LRMC CAPEX; 38% recovered LRMC Shkaratan compared to (2011) OPEX (income statements) and CAPEX (LCOE benchmarks) 11 Vagliasindi 19 developing Late 1990s to Cost recovery Cost recovery index Tariffs increased over and Besant- countries late 2000s index (average correlated with indexes of the period, but cost Jones (2013) worldwide + 3 revenue divided competition and vertical recovery fluctuated Indian states by average unbundling supply cost) 12 Alleyne and Large sample 2005–09 Disaggregated Average tariffs were of 70% Average QFD constant Hussain of Sub-Saharan (latest year QFD: T&D losses, of cost; QFD was about 1.7% at 1.7% of GDP between (2013) African countries available) collection losses, of 2009, half of which from 2005–06 and 2009–10 (unspecified) underpricing underpricing 13 Mayer, Residential 2005, 2010 Average 87% of residential In real terms, the net Banerjee, electricity use in effective tariff consumption was subsidized cost of the average and Trimble 29 states in India (based on in 2010; average cost household subsidy (2015) household recovery was 68%; 2 out of in 2010 was 70 times surveys) 29 states had effective tariffs larger than in 2005 > average cost 14 Khurana and 29 states in India 2003–11 Comparison of Cost recovery averaged Cost recovery Banerjee average billed 82% in 2003–11; 7 states had fluctuated within a band (2013) tariff was higher tariffs below cost in 2003, 14 of 76–85%; with a low than AC in 2011 point in 2010 15 Di Bella 32 countries in 2011–13 Price-gap Electricity subsidies in Latin n.a. and others Latin America and (average) approach pretax American and the Caribbean (2015) the Caribbean subsidies (% of were almost as large as GDP) direct fuel subsidies, on average 0.8% of GDP in 2011–13 16 IEA (2015) 40 non-OECD 2012–14 Price gap All but four countries Decline in total countries approach (based subsidize electricity (excl. subsidies by 10.4% in worldwide on average cost renewable energy subsidies) 2012–14; 5 additional of production) countries reached cost recovery ( Annex continued next page) Did Power Sector Reforms Improve Efficiency and Cost Recovery? 269 Serial No. Study Coverage Time KPIs Main findings Observed trends 17 Trimble 39 countries in 2011–15 Disaggregated Average cash collected 57% Most of the countries and others Sub-Saharan QFD: Collection was of cost; 2 countries have with low QFDs (2016) Africa losses; a financially viable electricity improved over past T&D losses; sector; 19 countries cover decade, while most of overstaffing; OPEX; QFD average 1.5% of the countries with high underpricing GDP QFDs remained high 18 World Bank Utilities in 40 2003–13 Utilities’ 10 out of 40 utilities were Share of profitable (2016) developing profitability profitable; 2 out of 17 Sub- utilities increased from countries (based on Saharan African utilities were 10% to 35% in 2010, worldwide utility financial profitable in 2000, 4 in 2013 then fell to 25% in 2013 statements) 19 Coady 153 OECD and 2013, 2015 Price-gap 79 out of 119 developing Absolute decline of and others non-OECD approach countries had electricity subsidies by 36.5%; (2015) economies (reference subsidies in 2015, compared numbers of countries worldwide price including to 1 out of 34 “advanced with subsidies from 75% consumption economies” (Taiwan, China) to 66% taxes; excl. renewable energy subsidies) Source: Huenteler and others 2017. Note: AC = average cost; CAPEX = capital expenditure; KPI = key performance indicator; LCOE = levelized cost of energy; LRMC = long-run marginal cost; O&M = operation and maintenance; OECD = ­ Organisation for Economic Co-operation and Development; OPEX = operating expenditure; QFD = quasi-­ fiscal deficit; T&D = transmission and distribution; n.a. = not applicable. ANNEX 8B. COVERAGE OF QUANTITATIVE COST RECOVERY ANALYSIS UNDERTAKEN FOR THIS CHAPTER Year for prereform Years for postreform Scope of utility-level analysis Country/state analysis analysis (majority ownership) Colombia 1987 2010–16 Codensa (Private) Dominican Republic a 1987 2010–15 Edesur (Public) Egypt, Arab Rep. 1987 2011–16 EEHC (Public) India – Andhra Pradesh 1987 2011–15 APEPDCL (Public), APSPDCL (Public) India – Odisha 1987 2011–15 CESU (Public), WESCO (Private) India – Rajasthan 1987 2011–15 AVVN (Public), JDVVN (Public), JVVN (Public) Kenya 1987 2010–16 KPLC (Private) Morocco 1987 2013 ONEE (Public) b Pakistan 1987 2013–16 KE (Private), LESCO (Public) Peru 1987 2010–16 Hidrandina (Public), Luz del Sur (Private) Philippines 1987 2010–16 Beneco (Private), Meralco (Private) Senegal 1991 2010–16 Senelec (Public) Tajikistan 2003 2010–16 Barki Tojik (Public) Tanzania 1991 2012–16 TANESCO (Public) Uganda 1987 2012–16 UMEME (Private) Ukraine c 1994 2012–17 Dniprooblenergo (Private), Khmelnytskoblenergo (Public) Vietnam 1993 2010–16 NPC (Public) Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: Benchmark system loss level is 5 percent for the analysis in this table. a. Approximated by data for Edesur because full-country data were not available. b. Approximated by sector-wide data. c. C1–C3 cost recovery is approximated by A1–A3 because systematic information on government support to the utilities was not available. 270 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD ANNEX 8C. INDICATORS OF COST RECOVERY AND FINANCIAL VIABILITY OF POWER SECTORS AND UTILITIES IN 17 CASE STUDIES TABLE 8C.1 Full cost recovery for power sectors Percent Full cost recovery (C3 of approximate)a Country/state Type Region Prereform 2010 2011 2012 2013 2014 2015 2016 2017 Average 2010–17 Colombia Private LAC 55 94 109 117 103 100 105 102 — 104 Dominican Republicb Public LAC 106 71 65 62 65 62 73 — — 66 Egypt, Arab Rep. Public MENA 21 — 62 62 68 50 49 50 — 55 India – Andhra Pradesh Public SAR 54 — 36 36 38 37 44 — — 38 India – Odisha Public SAR 54 — 94 92 90 92 91 — — 91 India – Rajasthan Public SAR 54 — 38 41 47 43 57 — — 45 Kenya Public SSA 106 99 101 94 87 89 81 80 — 90 Moroccoc Public MENA 107 — — — 84 — — — — 84 Pakistan Public SAR 91 — — — 66 75 87 97 — 80 Peru Public LAC 76 102 102 101 98 95 97 93 — 98 Philippines Public EAP 132 98 99 100 96 98 99 95 — 98 Senegal Public SSA 88 77 68 55 66 72 86 87 — 70 Tajikistan Public ECA 24 — — — 52 73 83 70 — 67 Tanzania Private SSA 87 — — 60 56 — 90 71 — 68 Ugandac Private SSA 28 — — 110 96 91 85 81 — 92 Ukraine Public ECA 15 — — 107 115 110 105 103 93 106 Vietnam Private EAP 67 91 86 91 88 89 89 88 — 89 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SAR = South Asia; SSA = Sub-Saharan Africa; — = not available a. Excluding externalities. b. Approximated by Edesur. c. Approximated by full financial cost recovery. TABLE 8C.2 Full financial cost recovery for power sectors Percent Full financial cost recovery (A3 of approximate) a Average Country/state Type Region Prereform 2010 2011 2012 2013 2014 2015 2016 2017 2010–17 Colombia Private LAC — 94 109 117 103 100 105 102 — 104 Dominican Republicb Public LAC — 71 65 62 65 62 73 — — 66 Egypt, Arab Rep. Public MENA — — 62 62 68 70 65 66 — 65 India – Andhra Pradesh Public SAR — — 40 41 39 39 48 — — 41 India – Odisha Public SAR — — 94 92 90 92 91 — — 91 India – Rajasthan Public SAR — — 40 44 51 44 59 — — 47 (Table continued next page) Did Power Sector Reforms Improve Efficiency and Cost Recovery? 271 TABLE 8C.2 Full financial cost recovery for power sectors (Continued) Percent Full financial cost recovery (A3 of approximate) a Average Country/state Type Region Prereform 2010 2011 2012 2013 2014 2015 2016 2017 2010–17 Kenya Public SSA — 100 101 95 88 90 83 83 — 91 Morocco Public MENA — — — — 84 — — — — 84 Pakistan Public SAR — — — — 76 89 96 104 — 90 Peru Public LAC — 102 102 101 98 95 97 93 — 98 Philippines Public EAP — 98 99 100 96 98 99 95 — 98 Senegal Public SSA — 85 97 75 86 93 97 91 — 88 Tajikistan Public ECA — — — — 101 103 108 94 — 101 Tanzania Private SSA — — — 73 68 — 107 83 — 81 Uganda Private SSA — — — 114 105 100 94 88 — 100 Ukraine Public ECA — — — 107 115 110 105 103 93 106 Vietnam Private EAP — 92 87 93 89 91 90 89 — 90 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SAR = South Asia; SSA = Sub-Saharan Africa; — = not available a. Excluding externalities. b. Approximated by Edesur. TABLE 8C.3 Financial operating cost recovery for power sectors, based on cash collected Percent Financial operating cost recovery (A1 of approximate), adjusted for bill collection losses Average Country/state Type Region Pre-Reform 2010 2011 2012 2013 2014 2015 2016 2017 2010–17 Colombia Private LAC — 130 162 152 149 146 145 139 — 147 Dominican Republic a Public LAC — 67 62 58 61 59 71 — — 62 Egypt, Arab Rep. Public MENA — — 143 125 100 103 128 119 — 120 India – Andhra Pradesh Public SAR — — — — — — — — — — India – Odisha Public SAR — — 88 88 89 92 92 — — 90 India – Rajasthan Public SAR — — — — — — — — — — Kenya Public SSA — 112 114 115 113 115 123 124 — 116 Morocco Public MENA — — — — — — — — — — Pakistan Public SAR — — — — 90 113 124 144 — 116 Peru Public LAC — 137 135 132 129 130 130 131 — 132 Philippines Public EAP — 108 107 110 110 115 115 115 — 111 Senegal Public SSA — 89 105 93 97 105 104 102 — 99 Tajikistan Public ECA — — — — 68 84 108 98 — 88 Tanzania Private SSA — — — 81 87 — 120 88 — 93 Uganda Private SSA — — — 117 129 124 119 129 — 123 Ukraine Public ECA — — — — — — — — — — Vietnam Private EAP — 99 98 104 102 101 102 100 — 101 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: EAP = East Asia and Pacific; ECA = Europe and Central Asia; LAC = Latin America and the Caribbean; MENA = Middle East and North Africa; SAR= South Asia; SSA = Sub-Saharan Africa; — = not available a. Approximated by Edesur. 272 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 8C.4 Full cost recovery for power utilities Percent Full cost recovery (C3 of approximate)a Average Power utility Type 2010 2011 2012 2013 2014 2015 2016 2017 2010–17 Colombia (Codensa) Private 94 109 117 103 100 105 102 — 104 Dominican Republic (Edesur) Public 71 65 62 65 62 73 — — 66 Egypt, Arab Rep. (EEHC) Public — 62 62 68 50 49 50 — 55 India – Andhra Pradesh (APSPDCL) Public — 35 35 34 36 44 — — 36 India – Andhra Pradesh (APEPDCL) Public — 42 35 48 42 44 — — 42 India – Odisha (WESCO) Public — 94 97 94 94 94 — — 95 India – Odisha (CESU) Public — 94 87 86 90 87 — — 89 India – Rajasthan (AVVN) Public — 37 37 46 32 56 — — 40 India – Rajasthan (JDVVN) Public — 34 43 49 56 64 — — 48 India – Rajasthan (JVVN) Public — 57 53 57 53 61 — — 56 Kenya (KPLC) Public 99 101 94 87 89 81 80 — 90 Morocco (ONEE) b Public — — — 84 — — — — 84 Pakistan (KE) Public — — — 76 86 90 94 — 86 Pakistan (LESCO) Private — — — 53 61 71 89 — 66 Peru (Luz del Sur) Private 105 107 104 100 96 101 101 — 102 Peru (Hidrandina) Public 91 86 91 93 90 85 74 — 87 Philippines (Meralco) Private 99 100 101 98 98 100 96 — 99 Philippines (Beneco) Private 96 88 88 88 88 87 88 — 89 Senegal (Senelec) Public 77 68 55 66 72 86 87 — 70 Tajikistan (Barki Tojik) Public — — — 52 73 83 70 — 67 Tanzania (TANESCO) Public — — 60 56 — 108 93 — 74 Uganda (UMEME) Private — — 110 96 91 85 81 — 92 Ukraine (Khmelnytskoblenergo) Mixed — — 107 115 110 105 103 110 109 Vietnam (NPC) Public 91 86 91 88 89 89 88 — 89 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: — = not available a. Excluding externalities. b. Approximated by sector-wide cost recovery. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 273 ANNEX 8D. INDICATORS OF EFFICIENCY OF UTILITIES IN 17 CASE STUDIES TABLE 8D.1 Revenue lost due to undercollection for power utilities Percent of revenue lost Average Power utility Type 2010 2011 2012 2013 2014 2015 2010–15 Colombia (Codensa) Private — 0 0 0 0 0 0 Colombia (EPM) Public 0 0.44 0 0 0 0 0.1 Dominican Republic (Edesur) Public 6.66 6.67 3.55 3.71 2.80 3.21 4.4 Dominic Republic (Edenorte) Public 7.99 9.32 3.25 2.87 2.89 1.48 4.6 Egypt, Arab Rep. (Cairo North) Public 24.97 24.33 24.38 25.52 24.47 22.41 24.3 Egypt, Arab Rep. (Alexandia) Public 16.24 15.49 15.48 14.90 15.00 13.11 15.0 India – Andhra Pradesh (APEPDCL) Public 0.67 7.47 3.74 3.67 2.30 2.91 3.5 India – Andhra Pradesh (APSPDCL) Public 3.60 2.21 1.00 2.23 1.22 0.97 1.9 India – Odisha (CESU) Public 5.40 12.60 9.40 7.61 4.68 4.00 7.3 India – Odisha (WESCO)a Private 3.23 6.14 5.58 4.44 5.58 6.67 5.3 India – Rajasthan (JDVVNL) Public 6.06 1.35 4.92 0.85 3.51 2.98 3.3 India – Rajasthan (JVVNL) Public 3.50 2.78 4.58 1.83 3.50 1.77 3.0 Kenya (KPLC)b Public 0 0 0 0 0 1.60 0.3 Morocco (ONEE) Public 0 0 0 — 0 0 0 Pakistan (K-Electric Ltd) Private 8.68 10.14 8.24 11.12 10.19 8.05 9.4 Pakistan (LESCO) Public 6.24 1.84 3.67 2.08 2.07 3.95 3.3 Peru (Luz Del Sur) Private 0.55 0.60 0.69 1.43 0.13 1.18 0.8 Peru (Hidrandina) Public 0.11 0.53 0.92 0.40 0.55 1.59 0.7 Philippines (MERALCO) Private 0 0 0 0 0 0.04 0 Philippines (BENECO) Private 1.45 2.91 2.59 1.50 0 0 1.4 Senegal (Senelec) Public 5.80 2.61 3.50 6.26 0.37 6.88 4.2 Tajikistan (Barki Tojik) Public 13.04 0.98 26.74 23.41 14.28 12.84 15.2 Tanzania (TANESCO) Public 26.17 -16.79 11.14 -6.39 — 3.28 3.5 Uganda (UMEME) Private 4.35 0.92 5.20 0 0.85 1.70 2.2 Ukraine (Dniproblenergo) Private 1.62 1.73 0 1.69 1.82 0.25 1.2 Ukraine (Khmelnytskoblenergo) Public 0.31 1.04 0 0.79 0 1.21 0.6 Vietnam (NPC) Public 0 0.36 0.44 0 0.44 0 0.2 Vietnam (HCMCPC) Public 0.60 0.69 0.30 0.39 0.44 0.45 0.5 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: — = not available. a. WESCO in Odisha was under private ownership until 2015. b. KPLC is 51 percent government owned whereas 49 percent of its equity is floated in the Nairobi stock exchange. 274 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 8D.2 Revenues lost due to excessive system losses for power utilities Percent of revenue lost Threshold Average Power utility Type lossesc 2010 2011 2012 2013 2014 2015 2010–15 Colombia (Codensa) Private 7 — 1.6 2.0 1.7 1.8 1.8 1.8 Colombia (EPM) Public 10 0 0 0 0 0 0.2 0 Dominican Republic (Edesur) Public 10 18.2 16.9 17.1 15.5 16.6 14.8 16.5 Dominic Republic (Edenorte) Public 10 17.7 19.6 25.2 20.8 19.5 17.7 20.1 Egypt (Cairo North) Public 7 2.4 2.2 1.7 1.6 1.4 2.9 2.0 Egypt (Alexandia) Public 7 1.3 3.0 3.0 2.3 1.7 2.3 2.3 India- Andhra Pradesh (APEPDCL) Public 12 0 0 0 0 0 0 0 India- Andhra Pradesh (APSPDCL) Public 12 1.1 0.7 0 0 0 0 0.3 India- Odisha (CESU) Public 12 15.3 16.8 17.5 17.8 15.8 15.7 16.5 India- Odisha (WESCO)a Private 12 15.5 17.9 19.8 19.2 16.8 15.9 17.5 India- Rajasthan (JDVVNL) Public 12 21.8 16.1 14.5 16.1 13.1 13.4 15.8 India- Rajasthan (JVVNL) Public 12 16.4 16.6 13.6 16.8 18.3 18.2 16.7 Kenya (KPLC)b Public 13 2.1 2.1 3.1 3.8 3.4 3.0 2.9 Morocco (ONEE) Public 10 0.3 0.7 1.2 — 2.0 2.3 1.3 Pakistan (K-Electric Ltd) Private 12 21.9 19.6 18.8 15.4 11.1 8.3 15.9 Pakistan (LESCO) Public 12 1.8 1.1 1.6 1.2 1.2 1.6 1.4 Peru (Luz Del Sur) Private 7 0.4 0.4 0.5 0.3 1.1 0.8 0.6 Peru (Hidrandina) Public 10 0.3 0 1.0 0.6 0.6 0.3 0.5 Philippines (MERALCO) Private 7 0.8 0.3 0.1 0.4 0.8 0.8 0.5 Philippines (BENECO) Private 7 1.8 1.9 2.8 1.9 1.8 1.9 2.0 Senegal (Senelec) Public 13 10.1 10.5 9.1 7.9 6.2 6.1 8.3 Tajikistan (Barki Tojik) Public 10 0 1.1 1.0 1.1 1.7 1.6 1.3 Tanzania (TANESCO) Public 13 7.1 13.6 10.0 16.2 — 3.5 10.1 Uganda (UMEME) Private 13 8.6 6.7 8.1 7.6 5.1 3.8 6.7 Ukraine (Dniproblenergo) Private 7 0 0 0 0 0 0 0 Ukraine (Khmelnytskoblenergo) Public 10 5.1 5.1 5.5 5.2 4.8 4.9 5.1 Vietnam (NPC) Public 10 0 0 0 0 0 0 0 Vietnam (HCMCPC) Public 7 0 0 0 0 0 0 0 Source: World Bank elaboration based on Rethinking Power Sector Reform utility database 2015. Note: — = not available. a. WESCO in Odisha was under private ownership until 2015. b. KPLC is 51 percent government owned while 49 percent of its equity is floated in the Nairobi stock exchange. c. Threshold losses: Level of distribution losses of a comparable efficient utility in the region. Did Power Sector Reforms Improve Efficiency and Cost Recovery? 275 NOTES 4. According to the most common definition, 1. This chapter draws on Huenteler and others QFD is the difference between the actual rev- (2017) and further original research con- enue charged and collected at regulated elec- ducted by a team led by Ani Balabanyan tricity prices and the revenue required to fully and comprising Arthur Kochnakyan, Joern cover prudently incurred operating costs of Huenteler, Denzel Hankinson, Nicole service provision and capital depreciation: QFD Rosenthal, Arun Singh, and Tu Chi Nguyen. (as percent of GDP) = Cost of Underpricing of The chapter is based on a methodologically Electricity + Cost of Nonpayment of Bills + consistent set of financial models prepared Cost of Excessive Technical Losses (Alleyne for the 25 utilities across 14 countries and 3 and Hussain 2013). Indian states. The financial analysis was led 5. The literature includes relatively minor vari- by Arthur Kochnakyan, supported by a team ations of this generally accepted QFD for- of consultants including Martin Tarzyan, mula. For example, Briceño-Garmendia, Vazgen Sargsyan, Adrian Ratner, and Smits, and Foster (2009) and Kojima and Emiliano Lafalla. Vivien Foster and Anshul Trimble (2016) introduce overstaffing as an Rana coordinated the work program. additional “hidden cost” item. 2. Conceptually, cost recovery can be viewed 6. Because of lack of information, the chapter from the perspective of the power utility/ does not assess labor cost inefficiencies. sector, the fiscal perspective, or the over- 7. The term tariff structure is used here to all economic perspective. In each case, the describe the composition of end-consumer full costs would be defined differently, and prices (for example, one aggregate service which perspective is appropriate depends tariff compared with separate tariffs for gen- on the research question. Further, depend- eration, transmission, and distribution) as ing on the purpose, cost recovery may well as the differentiation of end-consumer include “full costs” with any inefficiencies tariffs by consumer groups (do tariffs differ (including excess losses) the power com- between groups and by how much?). pany/sector has or cost recovery assuming 8. Specifically, the level of cross-subsidization efficient operation of the company/sector. is quantified here as the area between the The latter approach is ideally that taken by 45-degree line and the curve defined by the the regulators so as not to pass inefficiencies cumulative shares of consumption and rev- to consumers. Importantly, full cost recov- enue by customer class. The formula is the ery of tariffs for the sector does not neces- same as that of a Gini-coefficient of inequality. sarily mean that all individual parts of the supply chain (generation, transmission, and distribution) recover their costs, depend- REFERENCES ing on how tariffs are set for the different Alleyne, T. S. C., and M. Hussain. 2013. Energy services. Furthermore, some studies in the Subsidy Reform in Sub-Saharan Africa: Experiences literature approximate tariffs with revenues and Lessons. Washington, DC: International and cost with actual cost incurred by the Monetary Fund. utilities, bringing the concept of “cost recov- Di Bella, G., L. Norton, J. Ntamatungiro, S. ery” closer to the common understanding of Ogawa, I. Samake, and M. Santoro. 2015. “financial viability.” “Energy Subsidies in Latin America and 3. Financial viability is also an attribute of the Caribbean: Stocktaking and Policy investment projects and in fact early World Challenges.” Working Paper 15, International Bank studies of financial viability in the Monetary Fund, Washington, DC. power sector were primarily interested in Briceño-Garmendia, C., and M. Shkaratan. 2011. the ability of individual investments to make “Power Tariffs: Caught Between Cost Recovery adequate returns. But this view has evolved and Affordability.” Policy Research Working (see Huenteler and others 2017) and now Paper 5904, World Bank, Washington, DC. the primary unit of analysis in the literature doi:10.1596/1813-9450-5904. is the utility. This is reflected in the term’s Briceño-Garmendia, C., K. Smits, and V. Foster. usage in this chapter given the focus on 2009. Financing Public Infrastructure in Sub- leveraging private solutions and improving Saharan Africa: Patterns and Emerging Issues. utility performance. Washington, DC: World Bank Group. 276 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD Camos, D., R. Bacon, A. Estache, and M. M. Mayer, K., S. Banerjee, and C. Trimble. 2015. Hamid. 2018. Shedding Light on Electricity Utilities Elite Capture: Subsidizing Electricity Use by Indian in the Middle East and North Africa: Insights from a Households. Washington, DC: World Bank. Performance Diagnostic. Washington, DC: World doi:10.1596/978–1-4648-0412-0. Bank. Munasinghe, M., J. Gilling, and M. Mason. 1989. Coady, D., I. Parry, L. Sears, and B. Shang. 2015. A Review of World Bank Lending for Electric Power. “How Large Are Global Energy Subsidies?” Washington, DC: World Bank. IMF Working Paper 105, International Saavalainen, T., and J. ten Berge. 2006. Quasi- Monetary Fund, Washington, DC. doi:10.5089​ Fiscal Deficits and Energy Conditionality in Selected /9781513532196.001. CIS Countries. Washington, DC: International Eberhard, A., V. Foster, C. Briceño-Garmendia, Monetary Fund. F. Ouedraogo, D. Camos, and M. Shkaratan. Trimble, C., M. Kojima, I. Perez Arroyo, and F. 2008. “Underpowered: The State of the Power Mohammadzadeh. 2016. “Financial Viability Sector in Sub-Saharan Africa.” Background of Electricity Sectors in Sub-Saharan Africa: Paper 6, Africa Infrastructure Country Quasi-Fiscal Deficits and Hidden Costs.” Policy Diagnostic, World Bank, Washington, DC. Research Working Paper 7788, World Bank, Ebinger, J. O. 2006. Measuring Financial Performance Washington, DC. in Infrastructure: An Application to Europe and Vagliasindi, M., and J. Besant-Jones. 2013. Central Asia. Washington, DC: World Bank. “Power Market Structure: Revisiting Policy Foster, V., and T. Yepes. 2006. “Is Cost Recovery Options.” World Bank, Washington, DC. a Feasible Objective for Water and Electricity? World Bank. 1972. “Operations Evaluation The Latin American Experience.” Policy Report: Electric Power.” Report 2-17, World Research Working Paper 3943, World Bank, Bank, Washington, DC. Washington, DC. — — —. 1990. “Review of Electricity Tariffs in Huenteler, J., I. Dobozi, A. Balabanyan, and Developing Countries during the 1980s.” S. G. Banerjee. 2017. “Cost Recovery and World Bank, Washington, DC. Financial Viability of the Power Sector in —— —. 1993. “Tanzania: Sixth Power Project.” Staff Developing Countries: A Literature Review.” Appraisal Report, World Bank, Washington, DC. Policy Research Working Paper 8287, World Bank, Washington, DC. — ——. 1994. “Ukraine: Electricity Market IEA (International Energy Agency). 1999. Development Project.” Staff Appraisal Report, “Looking at Energy Subsidies: Getting the World Bank, Washington, DC. Prices Right.” World Energy Outlook 1999 ——— . 1 9 9 5 . “ Vi e t n a m : P o w e r S e c t o r Insights, IEA, Paris. Rehabilitation Project.” Staff Appraisal Report, — — —. 2015. “Energy Subsidies: Fossil Fuel World Bank, Washington, DC. Subsidy Database.” IEA, Paris. —— —. 1998. “Senegal: Energy Sector Adjustment Khurana, M., and S. G. Banerjee. 2013. Beyond Operation.” Staff Appraisal Report, World Crisis: The Financial Performance of India’s Power Bank, Washington, DC. Sector. Washington, DC: World Bank. —— —. 2004. “Central Asia: Regional Electricity Kojima, M. 2017. “Identifying and Quantifying Energy Export Potential Study.” World Bank, Subsidies.” Energy Subsidy Reform Assessment Washington, DC. Framework (ESRAF) Good Practice Note 1, World —— —. 2016. “Financial Viability of the Electricity Bank, Washington, DC: World Bank Group. Sector in Developing Countries: Recent Kojima, M., and C. Trimble. 2016. “Making Trends and Effectiveness of World Bank Power Affordable for Africa and Viable for Its Interventions.” IEG Learning Product, World Utilities.” World Bank, Washington, DC. Bank, Washington, DC. Did Power Sector Reform Deliver Better Sector Outcomes? 9 Guiding questions • Did the 1990s model deliver on its stated objective of improving security of supply through improved utility performance and increased investment? • Did the reforms undertaken as part of the 1990s model help to support the wider policy goals of uni- versal access and decarbonization that emerged in the 21st century? • To what extent were countries that did not espouse the 1990s model able to deliver strong power sector outcomes by other means? Summary • While country-specific factors are the largest determinants of utility performance, there is evidence that regulation, restructuring, liberalization, and governance lead to significant improvements in some dimensions of efficiency for some groups of countries. When it comes to cost recovery, this is much more likely in countries that have introduced private sector participation. • Although almost all private utilities attain high levels of distribution efficiency, their performance is matched by a ­ substantial minority of public utilities. • While market reforms can be helpful in improving the overall efficiency and financial viability of the power sector, and creating a better climate for investment, they cannot—in and of themselves— deliver on the social and environmental aspirations of the twenty-first century. Complementary policy measures are needed to direct and incentivize the specific investments that are needed. ­ • Progress on electrification was made by countries with widely differing approaches to power sector reform and appears to be driven primarily by targeted public investment as countries reach higher income levels. At the same time, progress (or otherwise) on decarbonization of the electricity sector has historically been an unintended by-product of measures to improve security of supply. • Overall, although some of the deepest reformers delivered good sector outcomes, other countries with weaker starting conditions reformed without notable impacts on performance. Furthermore, a third group of countries showed that it was sometimes possible to achieve comparable performance by means of different institutional approaches. 277 278 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD INTRODUCTION investment? Did the reforms undertaken as This chapter evaluates the extent to which part of the 1990s model help to support the power sector reforms are associated with bet- wider policy goals of universal access and ter power sector outcomes. 1 Reform is a decarbonization that emerged in the 21st cen- means to an end, premised on the assumption tury? To what extent were countries that did that reform will lead to improved utility per- not espouse the 1990s model able to deliver formance and better outcomes for citizens. It strong power sector outcomes by other means? is therefore of fundamental importance to The 1990s power sector reform model was examine this hypothesis against the available based on an implicit theory of change that evidence. Despite some statistical limitations linked reforms to improved utility performance in definitively determining the direction of and better sector outcomes (figure 9.1). The causality, correlations between reform effort move toward commercially oriented utilities, and various measures of outcomes, combined ideally through private sector participation—or with more rigorous econometric analysis, at least through restructured corporatized shed some light on the underlying relation- utilities—was expected to create incentives for ­ ships. When this quantitative evidence is more efficient decision making, which would combined with the rich qualitative evidence be reinforced by competition where possible, underpinning the Power Sector Reform or incentive-based regulation where necessary. Observatory, some general inferences can be At the same time, competition or regulation or drawn. Thus, the guiding questions for this both would create the required environment chapter are as follows. for cost-recovery pricing, which would itself Did reforms lead to improved utility perfor- become easier to achieve thanks to greater effi- mance on the key dimensions of cost recovery ciency. The resulting improvement in financial and operational efficiency? Did the 1990s viability would make it possible for utilities to model deliver on its stated objective of improv- attract the larger volume of investment needed ing security of supply through increased to attain better sector outcomes. FIGURE 9.1 The theory of change underpinning the 1990s power sector reform model Reforms Intermediate outcomes Outcomes Liberalized, regulated markets for wholesale and retail electricity Outcomes before • 1st best: Competitive market 2000 • 2nd best: Regulated by independent agency Cost recovery • Reasonable return • Security of supply Attracts and quality of service on new investment Cost recovery prices • Cost recovery in • Fiscal a ordability ongoing operations Least-cost Outcomes after investment Enables 2000 Cost optimization / • Generation • Access and e cient operation Incentivizes • Transmission a ordability E ciency • Distribution • Environmental • E cient operations sustainability Economically e cient decision • Optimal resource Guides making allocation Commercially operating utilities • 1st best: Restructured and privatized utilities • 2nd best: Restructured, corporatized SOE(s) as Enables o -takers of private investment Source: World Bank elaboration. Note: SOE = state-owned enterprise. Did Power Sector Reform Deliver Better Sector Outcomes? 279 In the late 20th century, the main outcome in determining its continued relevance to of interest was security of supply; by the early ­policy makers. 21st century, policy objectives had broadened A mixture of quantitative and qualitative to encompass electrification and decarboniza- methods will be used to explore the effects of tion. Under the 1990s power sector reform the 1990s model on intermediate and final model, the main purpose of attracting invest- sector reform outcomes. In evaluating the ment was to expand generation capacity so impact of the 1990s model, a suite of comple- that supply could keep pace with rapidly mentary approaches will be used (see box 9.1 growing demand in the developing world, ­ for a methodological explanation). First, the and, to a lesser extent, also strengthen related most rigorous statistical analysis to be used is transmission and distribution infrastructure. based on cross-country panel data econo­ Although some countries (such as Morocco metrics covering a sample of 88 developing and Vietnam) had already chosen to pursue countries over the 20 years between 1995 and ambitious electrification programs in the late 2015. Because of data limitations, this analy- 20th century, this was never an explicit sis, however rigorous, is necessarily based on emphasis of the 1990s model, nor of the wider relatively simplistic measures of both reform contemporaneous development agenda as effort and sector outcomes. Second, narrow- articulated in the Millennium Development ing our focus to the 2015 cross-section of per- Goals. It was only with the adoption of the formance of the 15 observatory countries Sustainable Development Goals (SDGs) in allows us to use a much richer set of variables 2015, and the associated universal access that more effectively capture the quality of target for 2030 (SDG7.1.1), that electrification the institutional framework and the nature of became more prominent on the political the associated outcomes.2 This comes at the agenda and widely embraced across the devel- cost of employing more simplistic statistical oping world. Similarly, the decarbonization of methods, including simple cross-sectional the energy sector was widely adopted as a regressions and scatter plots, that cannot political goal only after the Paris Climate pretend to do more than identify patterns of ­ Agreement of 2015. It was also reflected in the correlation (as opposed to causation).3 Third, SDGs in the form of explicit targets for renew- both types of statistical analysis are comple- able energy penetration and acceleration of mented by qualitative information on the energy efficiency (SDG7.2 and SDG7.3). reform stories and associated outcomes from Care therefore needs to be taken when the 15 observatory countries. Going behind evaluating the performance of the 1990s sec- the quantitative analysis to look at the quali- tor reform model against electrification and tative stories that underlie the observed out- decarbonization goals. It is certainly legiti- comes enables us to gain some insights into mate to evaluate the 1990s sector reform patterns of causality. model against the achievement of its intended objectives of improving utility performance KEY FINDINGS and security of supply. It does not seem fair, The relationship between specific dimensions however, to judge the past performance of the of power sector reform and resulting interme- 1990s model in terms of whether or not it diate outcomes for utility performance in terms supported delivery of the later policy goals of efficiency and cost recovery are explored in of electrification and decarbonization. findings #1–5 and correspond to the left-hand Nevertheless, it is relevant to consider to what loop of figure 9.1. The latter part of the discus- extent the 1990s model was at least compati- sion, surrounding findings #6–9, examines the ble with such goals, because this factor is key impact of reforms on final outcomes in terms 280 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD BOX 9.1 The methodological challenges of inferring the impact of power sector reforms For purposes of the impact analysis presented in this chapter, two samples of countries are considered: a small one and a large one, the former being nested within the latter (see table 9.1 in the main text of the chapter for full details). The small sample comprises 17 economies from the Power Sector Reform Observatory. The information available for this sample includes a very detailed characterization of the depth of power sector reform measures developed through a rich variety of indexes cover- ing restructuring, privatization, utility governance, regulation, and competition that aggregate information from several hundred specific dimensions of the reform environment. These indexes were developed and presented in chapters 4–7. This detailed information is available only for 2015. The large sample comprises 88 developing countries from the Regulatory Indicators for Sustainable Energy (RISE) ­ project (Banerjee and others 2017). The information available for this sample includes a long-time series dating back to 1990, but for a much more basic set of variables than those available in the small sample. The descriptors of the reform process are significantly cruder and follow the approach described in chapter 2. For example, regulatory reform for this larger sample is represented simply as a dummy variable denoting the presence or absence of a regulatory agency, whereas for the smaller sample some 50 dimensions of the regulatory frame- work are quantified. The availability of data on sector outcomes is also considerably more limited. The analysis first examines the impact of reforms on intermediate outcomes, including efficiency and cost recovery. Efficiency is cap- tured both at the generation and distribution level and is available for both the small and the large samples. • Generation efficiency, defined as the average thermal efficiency of combustion in power plants, can be expected to improve over time either through greater efficiency in plant dispatch or through shifts in the generation portfolio toward more modern plants and more efficient technologies. • Distribution efficiency, defined as the percentage of the revenues of an efficient utility that are captured by the utility, expresses the shortfall attributable to both technical and commercial losses. (When this variable is not available, as in the case of the larger sample, system losses are used as a proxy for distribution efficiency.) Cost recovery is tracked at both the operating and the full (capital) cost recovery levels on the basis of the framework presented in chapter 8. This rich information is available only for the small sample of 15 countries during the period 2010–15. • Operating cost recovery (A1) captures the ratio of average billed revenue to the average operating costs borne directly by the utility (leaving aside any subsidized items). • Full cost recovery (C3) captures the ratio of average billed revenue to the average total cost, including all operating costs and associ- ated operating subsidies, as well as all capital costs and associated capital subsidies required to fund historic investments as well as the utility’s forward-looking five-year business plan. Various performance measures are used to track the extent to which final outcomes in the sector improved over time (table 9.2 in the main text of the chapter offers full details). Measurement of security of supply is notoriously difficult, both conceptually and empirically. The ultimate measure of interest to citizens is reliability, and this can be captured through the SAIFI (System Average Interruption Frequency Index), which is available only for 2015. Capacity-based measures suffer from fewer data limitations, and an often-used measure is the reserve margin (or ratio of peak demand to installed capacity). The use of this measure is problematic for cross-country comparisons, because the magnitude of the required reserve margin will vary across countries according to legitimate differences in their power sector structure, related particularly to renewable energy shares. Instead, several other capacity-based measures are calculated, in recognition of the fact that no single one of them is ideal. • The normalized capacity measure simply facilitates meaningful comparsons of how much capacity is available across countries and over time by normalizing against population. security of supply by looking at the index of c • Capacity diversification captures another aspect of ­ ­ oncentration across different types of generation technologies to examine to what extent countries are exposed to shocks affecting any particular energy source. • The capacity growth indicator looks at the ratio of peak demand growth to capacity growth to give a sense of whether supply and demand are keeping pace. As regards environmental sustainability, one popular measure used to track Sustainable Development Goal (SDG) 7.2 is the share of modern renewables in total final energy consumption. Nevertheless, this measure only partially captures the carbon footprint of the energy s­ ector, which is also affected by the different types of fossil fuel generation that may be in use, as well as the presence of nuclear energy. To capture this wider perspective on decarbonization, a second indicator is used, which is the carbon intensity of electricity generation. As regards social inclusion, electrification is comparatively easy to measure through the household access rate used to track progress on SDG7.1. For the year 2015, it can be complemented by a measure of affordability, measured as the percentage of the budget of those in the bottom 40 percent of the income distribution that is needed to purchase the average level of electricity consumption for the country at the prevailing tariff structure. For each indicator, performance is benchmarked against normative bands. On the basis of benchmarking across the sample and with reference to international norms, threshold values are designated for each indicator in terms of what describes a good (green), mediocre (yellow), or poor (red) performance (see table 9.2 in the main text). In a similar way, threshold values are defined for what constitute a good (green), mediocre (yellow), or poor (red) performance improvement over time. (Box continued next page) Did Power Sector Reform Deliver Better Sector Outcomes? 281 BOX 9.1 The methodological challenges of inferring the impact of power sector reforms (Continued ) Two types of statistical analysis are performed. For the large sample, it is possible to perform cross-country panel data econo- metrics. Owing to the relatively large sample (some 1,760 observations resulting from the panel of 88 countries over the 20 years between 1995 and 2015), each outcome measure can be regressed upon indicators of the adoption of the four major reform mea- sures (regulation, restructuring, privatization, and competition). Several different control variables are also explored, including a time trend, country fixed effects, and, as conditioning variables, power-system size and the country’s income per capita. The inclusion of country fixed effects goes some way toward addressing concerns about potential reverse causality from outcomes to reforms. Full statistical results of these models are reported in annex 9B, whereas the main highlights are incorporated into the text. In general, the explanatory power of the regressions increases dramatically (with R-squared increasing from under 0.10 to over 0.50) when country fixed effects are introduced into the specification, underscoring the importance of context in determin- ing reform outcomes. For the small sample, simple cross-sectional regressions are performed. Because of the limited number of observations, individ- ual regressions are performed for each outcome variable against each reform measure, always controlling for power system size and the country’s income per capita. Where relevant, these regressions are further clarified through visual cross-plots and qualitative analysis, with a particular focus on explaining outliers. Outlier analysis is also performed to check upon the robustness of the visual cross-plots. TABLE 9.1 Mapping of data availability across small and large country samples Small sample Large sample Source Power Sector Reform Observatory Regulatory Indicators for Sustainable Energy (RISE) Number of countries 15 88 Basic reform indexes Restructuring, regulation, competition, privatization, for period 1990–2015 Refined reform indexes Utility governance, restructuring, competition, None regulation, private sector participation, 2015 only Intermediate outcomes Generation efficiency Thermal efficiency of power generation Thermal efficiency of power (the percentage of energy content of fuel that generation (the percentage is turned into electricity) of energy content of fuel that is turned into electricity) Distribution efficiency Combined measure of system losses and Average national-level commercial efficiency at the utility level for 2010–17, system losses since 1990 plus prereform benchmark for system losses Cost recovery Wide range of operating and capital cost recovery None measures for 2010–17, plus prereform benchmark Final outcomes Normalized capacity 1990–2015 1995–2015 Capacity diversification 1990–2015 None Capacity growth 2010–17 None Reliability of supply (SAIFI) 2015 only 2015 only Fiscal sustainability (QFD) 2010–17 None Electrification rate 1990–2015 1995–2015 Carbon intensity of electricity 1990–2015 1995–2015 Modern renewable energy share 1990–2015 1995–2015 of TFEC Source: Rethinking Power Sector Reform Observatory. Note: QFD = quasi-fiscal deficit; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption. 282 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9.2 Framework for evaluating impact on final sector performance outcomes Environmental Security of supply Social inclusion sustainability Modern Reliability Normalized Capacity Capacity Carbon renewable of supply capacity diversification growth Access Affordability intensity energy share Current performance, 2015 Good SAIFI up Capacity greater Concentration Capacity Electrification Bottom 40% Carbon Over 10% of to 12 than 200 MW per index up to 0.33 growth rate exceeds spend <5% of intensity TFEC million population to peak 80% GNI per capita up to 250 growth >1 on average gCO2/kWh consumption Moderate SAIFI — Concentration Electrification Bottom 40% Carbon 5–10% of between index between rate of spend 5%–10% intensity of TFEC 13 and 52 0.34 and 0.66 60–80% of GNI per capita 250–500 on average gCO2/kWh consumption Poor SAIFI Capacity less Concentration Capacity Electrification Bottom 40% Carbon <5% of TFEC greater than 200 MW per index exceeds growth rate up to spend > 10% of intensity than 52 million population 0.66 to peak 60% GNI per capita exceeds growth <1 on average 500 gCO2/ consumption kWh Performance Improvement, 1990–2015 Good n.a. Capacity Concentration n.a. Electrification n.a. Carbon Grew by >5 growth exceeds index dropped rate rises by intensity fell percentage prereform by more than 0.1 over 25% by over 100 points capacity gCO2/kWh Moderate n.a. Capacity growth Concentration n.a. Electrification n.a. Carbon Grew by <=5 is 50–100% of index dropped rate rises by intensity fell percentage prereform by less than 0.1 12–25% by up to 100 points gCO2/kWh Poor n.a. Capacity Concentration n.a. Electrification n.a. Carbon Decreased growth is <50% index increased rate rises by intensity of prereform up to 12% increased capacity Note: gCO2/kWh = grams of carbon dioxide per kilowatt hour; GNI = gross national income; MW = megawatt; SAIFI = System Average Interruption F ­ requency Index; TFEC = total final energy consumption; n.a. = not applicable; — not available. of security of supply, social inclusion, and envi- above, the panel data regression provides a ronmental sustainability. These correspond to statistically rigorous overview of relationships the right-hand loop of figure 9.1. between high-level reform indicators and simple measures of sector efficiency. Five dif- ferent model specifications are considered, Finding #1: Country context is the largest and they vary according to whether control long-term driver of utility performance, variables are used, and whether time trends but regulation seems to be a consistent and country fixed effects are added to the contributor to improved outcomes, regression. The statistical analysis shows that particularly in low-income countries the results are quite sensitive to the model Econometric analysis underscores the major specification in many cases (see annex 9B). role of specific country conditions in explain- In general, the incorporation of country ing the level of utility performance. As noted fixed effects hugely improves the overall Did Power Sector Reform Deliver Better Sector Outcomes? 283 explanatory power of the regressions, indicat- model and display the correct negative sign. ing that local specificities appear to have a (Although only reported in the annex 9B, much larger impact on power sector outcomes table 9B.1, private sector participation appears than the extent of power sector reform. For to have a large impact on system losses under this reason, attention in table 9.3 will focus on some specifications; however, these losses the specification incorporating both time ­ disappear once country fixed effects are added, trend and country fixed effects. In addition to suggesting that the impact of private sector estimating the model on the full sample of 88 participation is subsumed within the country ­ countries, two separate nested models are context.) In terms of magnitude, the impact of estimated for the low-income and middle-­ regulation and restructuring on system losses is income countries of the sample, respectively. only moderate, because introducing a regula- Statistical tests reject the hypothesis of no tory agency or undertaking full vertical and hor- structural differences between these subcate- izontal unbundling of the sector is associated gories of countries, indicating that coefficients with just a two-percentage-point reduction in are significantly different between the two system losses. Disaggregation of the sample and hence that there is value in estimating between low- and middle-income countries separate models. indicates that the impact of regulation and When it comes to system losses, regulation restructuring on system losses is primarily a and restructuring reforms seem to have the low-income country phenomenon, with statisti- strongest impact. They are the only two reform cal significance disappearing in the middle-­ actions that remain statistically significant when income sample. Particularly striking is the larger country fixed effects are introduced into the magnitude of the coefficient for restructuring in TABLE 9.3 Impact of reform measures on intermediate outcomes for the large sample Regression coefficients Full sample Low-income countries Middle-income countries Dependent variable: system losses Regulation –2.167* –2.652* ~ Restructuring –2.024** –7.362*** ~ Competition ~ ~ ~ Private sector participation ~ ~ ~ Observations 1,358 635 720 R-squared 0.557 0.502 0.593 Dependent variable: generation efficiency Regulation 1.367*** 1.423** 1.890*** Restructuring ~ ~ ~ Competition ~ 4.201* ~ Private sector participation ~ ~ ~ Observations 1348 622 722 R-squared 0.678 0.613 0.764 Note: Ordinary least squares regressions on panel data set of 88 countries for the period 1995–2015. A separate regression is estimated for each of the intermediate outcomes on the full set of reform variables capturing regulation, restructuring, competition, and privatization. The analysis considers a number of different ways of capturing control variables in terms of time trend, country-specific fixed effects, and control variables capturing system size and GDP per capita. The preferred specification, incorporating time trend and country-specific fixed effects (but no control variables), is reported here. All the other specifications can be viewed in the annex (for complete econometric results refer to annex 9B, tables 9B.1 and 9B.2). Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, ~ = not significant. 284 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD the low-­ income country sample, suggesting that Finding #2: Certain elements of full vertical and horizontal unbundling is associ- utility governance—associated with ated with system losses that are seven percent- financial discipline and human resource age points lower than full vertical integration. management—are associated with When it comes to generation efficiency, regu- improved cost recovery and efficiency lation and private sector participation seem to The association between utility governance have the strongest impact. Although all reform and utility performance is particularly strong measures are statistically significant in the sim- for certain aspects of governance (table 9.4). pler model specifications, once again most of The panel data analysis reported in table 9.3 these effects disappear once country fixed effects was not able to capture the issue of utility cor- are introduced. Regulation is the one reform porate governance, because of a lack of time that seems to have a consistently positive impact series variables on this issue. Focusing atten- on generation efficiency, with the introduction tion on the small sample makes it possible to of a regulatory agency associated with an explore how utility governance interacts with increase of about two percentage points in the utility performance. The Utility Governance thermal efficiency of generation plants. (In some Index is built up from a series of subindexes specifications, reported only in annex 9B, that capture different components of gover- tables 9B.1 and 9B.2, private sector participation nance, including corporate governance of the is also significantly related to generation effi- board and quality of utility management in ciency, with a similar order of magnitude in terms of financial discipline, human resources, terms of the size of the effect.) The impact of and adoption of information technology (recall regulation on generation efficiency is consistent chapter 4). Statistical analysis at the subindex in significance and in size between the low-in- level helps to pinpoint which might be the come and middle-income subsamples. most critical aspects of utility corporate Additionally, the introduction of competition is ­ governance from a performance perspective. found to have a large impact on generation effi- The analysis finds statistically significant ciency for the specific case of low-income coun- ­ relationships between managerial practices tries, where the effect can be up to four percent- relating to financial discipline and human age points. resources, as well as various measures of TABLE 9.4 Impact of utility governance and intermediate outcomes for the 15-country sample Regression coefficients Corporate governance Managerial practices Financial Human Information and Reforms Autonomy Accountability discipline resources technology Operating cost recovery ~ ~ 0.995** 1.182** ~ Full capital cost recovery ~ ~ ~ 0.857* ~ Distribution efficiency ~ ~ 0.352** ~ ~ Generation efficiency ~ ~ ~ ~ ~ Note: Ordinary least squares multivariate regression analysis for the Rethinking Power Sector Reform country sample economies). The table reports the results of a series of cross-sectional regressions modeling the impact of each individual (17 ­ utility level governance reform measure (board autonomy, board accountability, financial discipline, human resource m ­ ­ anagement, and  information technology) on each individual intermediate outcome (cost recovery, distribution, and thermal efficiency). GDP per capita and system size have been used as control variables. For complete results refer to annex 9C, table 9C.3. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, ~ = not significant. Did Power Sector Reform Deliver Better Sector Outcomes? 285 efficiency and cost recovery (table 9.4), but not cost recovery reveals major differences in terms for measures relating to the autonomy and of whether they prepare accounts in compliance accountability of the utility’s board, or for the with international financial reporting standards adoption of information technology. (100 percent of countries with higher operating Stronger financial discipline is associated cost recovery versus 21 percent of the others), with improvements in operating cost recovery whether they have the liberty to issue equity and distribution efficiency. It makes sense (45 versus 14 percent), whether they are that tighter financial discipline would lead required to pay dividends to shareholders to improved operating cost recovery and distri- (67 versus 0 percent), and whether they pub- bution efficiency. The analysis shows relatively licly disclose their accounts (100 versus 57 per- high correlations between financial discipline cent). Similarly, comparing countries with good and operating cost recovery (R-squared of 0.40) and poor performance on distribution efficiency and between financial discipline and distribu- reveals that the former are much more likely tion efficiency (R-squared of 0.30) (­figure 9.2). to have clearly defined public service obliga- The effects of governance on cost recovery are tions (69 versus 17 percent), financial accounts relatively material in size, with a one-percent- ­ produced in compliance with international age-point increase in the respective governance financial reporting standards (81 versus 25 per- indexes associated with roughly one additional cent), and the requirement to pay ­ dividends to percentage point in operating cost recovery. shareholders (50 versus 0 percent). Drilling deeper and looking at the practices of Examples from Sub-Saharan Africa serve to financial discipline that underpin the index illustrate this point. Kenya and Uganda are two sheds further light on the drivers of improved countries whose utilities score relatively highly performance (­ figure 9.3). Comparing countries in terms of financial discipline and that are also with good and poor performance on operating doing well on distribution efficiency and FIGURE 9.2 Utilities with stronger financial discipline show better performance on operating cost recovery and distribution efficiency a. Operating cost recovery b. Distribution e ciency Ukraine Philippines Morocco Colombia Vietnam Peru 160 Colombia 100 India – Andhra Pradesh Kenya Senegal Pakistan Uganda Operational cost recovery, A1 (%) 140 Egypt, Arab Rep. Tajikistan Distribution e ciency as share Tajikistan Peru Dominican Tanzania Pakistan Uganda 80 India – Rajasthan 120 Republic Philippines Kenya India – Odisha Egypt, Arab Rep. Ukraine of revenue (%) 100 India – Odisha Senegal 60 Vietnam Tanzania 80 India – Rajasthan 60 40 Dominican India – Andhra Pradesh Republic 40 20 20 R = 0.404 R = 0.2966 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 Utility Financial Discipline Index (%) Utility Financial Discipline Index (%) Note: For panel a, sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of the Dominican Republic and the R 2 of 0.2116, and does not materially affect the analysis. For panel b, a sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of the Dominican Republic and the R 2 of 0.2069 and does not materially affect the analysis. A1 = operating cost recovery. 286 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 9.3 Certain financial discipline practices are much more widely practiced among utilities that perform well in terms of operating cost recovery and distribution efficiency a. Operating cost recovery b. Distribution e ciency Financial accounts meet 21 Financial accounts meet 25 international standards 100 international standards 81 Utility pays dividends 0 Public service obligations 17 to shareholders 67 are explicitly defined 69 Financial accounts are 57 Financial accounts are 50 publicly disclosed 100 publicly disclosed 100 Public service obligations 29 Utility pays dividends 0 are explicitly defined 67 to shareholders 50 Utility can issue new equity 14 Utility can issue new bonds 25 45 44 0 20 40 60 80 100 0 20 40 60 80 100 Share of utilities (%) Share of utilities (%) Bad performers Good performers operating cost recovery. Although these two underpinning the index puts the drivers of countries adopted different reform paths, each improved performance into greater relief of them incorporated comparable measures to figure 9.5). Comparing countries with good (­ boost the financial discipline of the utility. In the and poor performance on operating cost recov- case of Uganda, the national distribution com- ery, one sees major differences in terms of the pany was awarded as a concession to a private following human resource practices: good operator, introducing the stronger commercial employee performance awarded with bonuses orientation of a private sector board and man- (91 versus 50 percent); transparency in hiring agement. Additionally, the listing of the utility percent); and freedom to of staff (92 versus 57 ­ on the stock exchanges in Kampala and Nairobi fire employees for bad performance (100 ­ versus makes the financial reporting more transparent. 57 percent). Similarly, comparing countries In the case of Kenya, the utility remains under with good and poor performance on full capital majority public ownership, with a 49 percent cost recovery reveals major differences in free- stake floated on the Nairobi Stock Exchange. dom to fire (75 versus 14 percent) or hire This situation introduces a series of require- employees (50 versus 5 percent) and having ments in terms of the quality of financial report- no link to g ­ overnment pay scales (75 versus ing and the degree of transparency that serve to 43 percent) or public employment regulations enhance financial discipline. (50 versus 22 percent). Stronger human resource management is associated with improvements in operating Finding #3: The quality of the legislated cost recovery and capital cost recovery. The regulatory framework for tariffs has analysis shows relatively high correlations no real impact on cost recovery, but between human resource management and the quality of implementation of that operating cost recovery (R-squared of 0.50), regulatory framework does seem to and between human resource management have an effect and full capital cost recovery (R-squared of 0.42) (figure 9.4). Drilling deeper into the A closer look at the components of regulation practices of human resource management suggests that it is the way that tariff and quality Did Power Sector Reform Deliver Better Sector Outcomes? 287 FIGURE 9.4 Utilities with better human resource management show better performance on operating cost recovery and full capital cost recovery a. Operating cost recovery b. Full capital cost recovery 160 120 Colombia Uganda Colombia Ukraine 140 Egypt, Arab Rep. Peru Operational cost recovery, A1 (%) Full capital cost recovery, C3 (%) Peru 100 Uganda Pakistan India – Odisha 120 Tajikistan Vietnam Philippines Kenya Philippines Kenya Morocco Ukraine 80 100 Vietnam Senegal Senegal Pakistan India – Andhra Pradesh Tanzania Dominican Republic Tanzania India – Odisha Tajikistan 80 India – Rajasthan 60 Dominican Republic Egypt, Arab Rep. 60 India – Rajasthan 40 India – Andhra Pradesh 40 20 20 R = 0.5006 R = 0.4262 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 Utility Human Resource Index (%) Utility Human Resource Index (%) Note: For panel a, a sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of the Indian state of Rajasthan and the R 2 of 0.4236 and does not materially affect the analysis. For panel b, a sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of the Indian states of Andhra Pradesh and Rajasthan and the R 2 of 0.1372 and does not materially affect the analysis. A1 = operating cost recovery; C3 = full cost recovery. FIGURE 9.5 Certain financial human resource practices are much more widely practiced among utilities that perform well in terms of operating cost recovery and full capital cost recovery a. Operating cost recovery b. Full capital cost recovery Employees can be fired 57 Managers are free to 14 for poor performance 100 fire employees 75 Employees receive 50 Managers are free to 5 performance related bonuses 91 hire employees 50 Recruitment involves 57 Wages not based on 43 reference checks 92 government pay scales 75 71 Recruitment involves 70 Sta training policy exists 100 100 reference checks Managers are free to 0 Government employment 22 hire employees 27 regulation don't apply 50 0 20 40 60 80 100 0 20 40 60 80 100 Share of utilities (%) Share of utilities (%) Bad performers Good performers regulation are actually practiced that matters framework affects utility performance. This the most. The panel data regression analysis regulatory index is built up from a series of presented in table 9.3 was able to capture regu- sub­indexes that capture different components lation only in terms of a single dummy variable of regulation, including regulatory governance denoting the date when a regulatory agency and the substantive aspects of regulating tariffs, was created. Confining attention to the quality, and market entry (recall chapter 6). small sample, there is a much richer character- Statistical analysis at the subindex level helps ization of the regulatory environment, which to pinpoint which might be the most critical serves to explore how the quality of regulatory aspects of regulation from a performance 288 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD p ­ erspective. This exercise reveals that it is the Cost recovery is strongly associated with the substantive aspects of tariff and quality actual practice of tariff regulation. Because r egulation—rather than overall regulatory ­ achieving cost recovery is the purpose of tariff governance considerations—that are most regulation, it makes sense that the two would strongly associated with outcomes for effi- be associated. The visual cross-plots serve to ciency and cost recovery (table 9.5). It also contrast the impact of regulation on paper ver- shows little association between de jure regula- sus regulation in practice. The relationship tion and utility performance; rather it is per- between tariff regulation on paper and cost ceived regulatory practice that is associated recovery is very weak, with an R-squared of just with better outcomes. 0.03 (figure 9.6). The reason is that almost every TABLE 9.5 Impact of regulatory reforms on intermediate outcomes for small sample De jure Perceived Regulatory Quality Tariff Regulatory Quality Tariff Regression coefficients governance regulation regulation governance regulation regulation Operating cost recovery ~ ~ ~ ~ ~ 0.466** Full capital cost recovery ~ ~ ~ ~ 0.389* ~ Distribution efficiency ~ ~ ~ ~ 0.181* ~ Generation efficiency ~ ~ ~ ~ ~ ~ Note: Ordinary least squares multivariate regression analysis for the Rethinking Power Sector Reform country sample (17 economies). The table reports the results of a series of cross-sectional regressions modeling the impact of each regulatory reform measure (regulatory governance, tariff regulation, and quality regulation) on intermediate outcomes (various measures of efficiency and cost recovery). GDP per capita and system size have been used as control variables. For complete results refer to annex 9C, table 9C.4. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, ~ = not significant. FIGURE 9.6 Operating cost recovery bears little relation to the quality of tariff regulations on paper, but it is closely associated with the quality of tariff regulations in practice a. Regulation on paper (de jure) b. Regulation in practice (perceived) 160 160 Colombia Operating cost recovery ratio, A1 (%) Operating cost recovery ratio, A1 (%) Egypt, Arab Rep. Colombia 140 140 Egypt, Arab Rep. Uganda Peru Peru Pakistan Uganda Pakistan 120 Tajikistan Ukraine Kenya 120 Tajikistan Kenya Ukraine Senegal Philippines India – Odisha Philippines 100 India – Odisha Vietnam 100 Tanzania Vietnam Senegal India – Andhra Pradesh Tanzania 80 India – Rajasthan 80 India – Andhra Pradesh India – Rajasthan Dominican Republic 60 60 Dominican Republic 40 40 20 R = 0.0369 20 R = 0.3727 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Tari regulation index (%) Perceived tari regulation index (%) Note: For panel a, a sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of Tajikistan and the R 2 of 0.3554 and does not materially affect the analysis. For panel b, a sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of Tajikistan and the R 2 of 0.5529 and does not materially affect the analysis. A1 = operat- ing cost recovery. Did Power Sector Reform Deliver Better Sector Outcomes? 289 country has a well-written regulatory frame- to their renationalization. A similar story can be work for tariffs, as can be visualized in the told for the Indian state of Rajasthan, where a ­ cross-plot. This makes sense because sound reg- modern regulatory framework was introduced ulations can simply be drafted at the outset of a as required by federal regulation. Although the reform by lifting existing good practices from framework was respected for several years, a elsewhere, but their existence does not translate change of government eventually led to tariffs into any guarantee that such good principles being frozen for as long as 10 years in contradic- will be followed in practice. By contrast, the tion of the regulatory framework. This weaker relationship between tariff regulation as prac- perceived performance of regulation for the ticed and cost recovery is much stronger with an Dominican Republic and India (Rajasthan) is R-squared of 0.37 (figure 9.6). To put this result shown by the fact that both jurisdictions appear in perspective, a one-­ percentage-point increase much farther to the left in figure 9.6, panel b in the index for the quality of perceived tariff (perceived regulation) than in panel a (de jure regulation is associated with a half-percent- regulation). age-point increase in operating cost recovery. Distribution efficiency is much more closely Digging a little deeper, countries that achieve associated with the practice of quality regulation operational cost recovery have p ­ articular attri- than the practice of price regulation. In princi- butes of the regulatory framework that distin- ple, tariff regulation—particularly when it is guish them from the rest, including the adop- incentive-based—is expected to lead to greater tion of regulatory accounting standards operational efficiency among utilities. Despite (91 percent of cost-recovering countries versus some association between the practice of tariff 0 percent of those that do not achieve opera- regulation and distribution efficiency of the util- tional cost recovery), a clear d ­ efinition of cost ities (with an R-squared of 0.13) (­ figure 9.7, recovery in the regulatory framework (100 per- panel a), the relationship with the practice of cent versus 17 percent), and clearly articulated quality regulation is much stronger (with an principles for the setting of end-user tariffs R-squared of 0.33) (figure 9.7, panel b). Once (91 percent versus 17 percent). again, what might be at play is a reverse causal- These results are also consistent with what is ity with utilities performing at high levels of known about the actual practice of tariff regula- operational efficiency being better placed to tion in the countries concerned. For example, respond to quality of service regulation. Digging the Dominican Republic was a relatively late deeper, we find that the countries with more reformer and adopted a sophisticated regulatory efficient utilities tend to share certain aspects of framework for tariff setting that was able to quality regulation that separate them from the incorporate good practice lessons from earlier underperformers. These aspects include the fol- Latin American reformers. Shortly after the lowing: financial incentives exist for utilities to adoption of the tariff framework came an oil meet customer service standards or increase price shock that had a major impact on the costs customer satisfaction (75 percent of the more of the utilities that were heavily reliant on oil- efficient utilities versus 22 percent of the oth- fired power generation, costs that—according to ers); utilities are required to use an automated the regulatory framework—should have been information management system to measure passed on in consumer prices. Nevertheless, the quality or reliability of the power supply because of political concerns, tariffs remained (63 percent versus 11 ­ percent); and utilities are frozen for several years. The resulting financial fined for failing to meet quality of service stan- distress to the privatized utilities eventually led dards (50 percent versus 0 percent). 290 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 9.7 Distribution efficiency follows the practice of quality regulation more than the practice of tariff regulation a. Practice of tari regulation b. Practice of quality regulation India – Andhra Pradesh Vietnam Philippines Colombia Colombia Morocco Philippines 100 Morocco Peru 100 Kenya Vietnam Ukraine Uganda Distribution e ciency as share of revenue (%) Distribution e ciency as share of revenue (%) Kenya India – Andhra Pradesh Ukraine Peru Tajikistan Uganda Senegal India – Rajasthan Tanzania India – Rajasthan Pakistan Senegal Tajikistan Pakistan 80 Tanzania 80 India – Odisha Dominican Egypt, Arab Rep. Dominican India – Odisha Republic Republic Egypt, Arab Rep. 60 60 40 40 20 20 R = 0.1272 R = 0.3309 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Perceived tari regulation index (%) Perceived quality regulation index (%) Note: A sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean, but it found no outliers. Finding #4: Private sector recovery has not been achieved; however, the participation is much more strongly story may be different for generation and associated with full capital cost ­ distribution. When the private sector comes recovery than any other reform, but into generation, full cost recovery—though this finding may simply reflect that one ­ desirable—is not essential. Private generators is a precondition for the other do not typically sell to end-­consumers but to The relationship between private sector par- distribution companies and are willing to do ticipation and full cost recovery potentially so even if these off-takers have not achieved runs in both directions. The lack of availability cost recovery—as long as adequate contrac- of time series data for cost recovery did not tual protections and credit enhancements can make it possible for the panel data regression be put in place. In the distribution segment, analysis reported in table 9.3 to examine the by contrast, private operators will depend relationship between private sector participa- entirely on consumer revenues; therefore, tion and cost recovery. This relationship could privatization is unlikely to go ahead unless potentially run in both directions: from higher tariffs are relatively close to full capital cost private sector participation to greater cost recovery levels. recovery or from greater cost recovery to Almost all countries with significant pri- higher private sector participation. On the one vate sector participation in the distribution hand, countries that have private sector par- sector have reached close to full capital cost ticipation find it easier to commit politically to recovery (table 9.6). There is a strong positive tariffs that recover full capital costs, because of association between private sector participa- the presence of a nongovernment entity in tion and full capital cost recovery, with an the sector. On the other hand, the private R-squared coefficient of 0.49. Countries scor- s ector is unlikely to take much interest in ­ ing at least 40 percent on the private sector entering a power sector where full capital cost participation index, typically have the private Did Power Sector Reform Deliver Better Sector Outcomes? 291 TABLE 9.6 Impact of private sector participation on intermediate outcomes for small sample Private sector participation Regression coefficients Overall Generation Distribution Transmission Operating cost recovery ~ ~ ~ ~ Economic cost recovery 0.77*** ~ 0.63*** 0.38** Distribution efficiency 0.29** ~ 0.20** 0.17** Thermal efficiency ~ ~ ~ ~ Note: Ordinary least squares multivariate regression analysis for the Rethinking Power Sector Reform country sample (17 economies). The table reports the results of a series of cross-sectional regressions modeling the impact of private sector participation (overall, in generation, in distribution, and in transmis- sion) on intermediate outcomes (various measures of efficiency and cost recovery). GDP per capita and system size have been used as control variables. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, ~ = not significant. sector active not only in generation but also FIGURE 9.8 Private sector participation is strongly associated in distribution. These countries include with full capital cost recovery Colombia, Peru, Philippines, Uganda, and 120 Ukraine, all of which are close to full capital Ukraine Colombia cost recovery (­ f igure 9.8). This pattern 100 Uganda Peru Philippines Full cost recovery ratio, C3 (%) becomes even clearer in figure 9.9, which Vietnam Kenya India – Odisha Morocco presents results at the utility level as opposed 80 Senegal Tanzania Pakistan to the country level. Dominican Republic Tajikistan Unlike their publicly owned counterparts, 60 Egypt, Arab Rep. private utilities almost universally operate in India – Rajasthan jurisdictions with full capital cost recovery— 40 India – Andhra Pradesh except for those where privatization has been 20 reversed. The cost recovery ratio for the R = 0.491 ­ publicly owned utilities appears on the left- hand axis, whereas that for the private utili- 0 10 20 30 40 50 60 70 ties appears on the right-hand axis. Several Private sector participation index (%) countries have utilities that appear on both Note: A sensitivity analysis was carried out for exclusion of outliers two standard the public and private sides of the ledger (such deviations from the mean, which resulted in the exclusion of the Indian state of Andhra Pradesh and of Peru, and the R2 of 0.3381. It does not materially affect the analysis. as Pakistan and Peru). The public utilities C3 = full cost recovery. exhibit a wide range of performance on cost recovery, ranging from 30 percent to 90 ­percent. Some of the public utilities scoring highly on cost recovery hail from jurisdictions which although private remains distant from where public and private utilities coexist full capital cost recovery tariffs; however, under a common framework for tariff regula- under the Pakistani regulatory framework, tion (India [Odisha], Pakistan, and Peru) Karachi Electric receives a compensating although this is not the case for Morocco’s subsidy that (at least partially) compensates ­ Office National de l’Electricité et de l’Eau for this shortfall. Another striking result Potable (ONEE) and Vietnam Electricity is that the utilities involved in the failed pri- (EVN), which both do relatively well on cost vatizations of the Dominican Republic and recovery. By contrast, the private utilities are Senegal are only about 70 percent of the way all tightly clustered around the 90 to 110 per- toward full capital cost recovery—a factor cent range for capital cost recovery. A notable which contributed to the failure of these exception is Karachi Electric from Pakistan, privatizations. 292 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 9.9 Only a few publicly owned utilities are allowed to power sector, the utility efficiency improves by charge cost recovery tariffs 0.28 percentage points (table 9.6). The rela- 120 tionship between private sector participation Dniproblenergo EVN (Vietnam) (Ukraine) and distribution efficiency shows a relatively CESU (India – Odisha) Codensa strong correlation, with an R-squared coeffi- (Colombia) cient of 0.33 (figure 9.10). Nevertheless, it is 100 Hidrandina (Peru) Luz del Sur (Peru) clear from the scatter plot that the countries LESCO (Pakistan) Meralco (Philippines) with publicly owned distribution utilities fall ONEE (Morocco) into two distinct groups. Publicly owned utili- 80 WESCO (India – Odisha) ties in jurisdictions such as the Dominican Full cost recovery ratio, C3 (%) TANESCO (Tanzania) UMEME (Uganda) Republic, the Arab Republic of Egypt, India Senelec (Senegal) Barki Tojik (Tajikistan) Kenya Power (Odisha and Rajasthan), Pakistan, Senegal, and 60 Dominican Republic Tanzania display exceptionally low levels of Beneco (Philippines) efficiency, with utilities losing 15 to 25 percent Egypt, Arab Rep. Karachi Electric (Pakistan) of revenues. At the same time, publicly owned JVVNL (India – Rajasthan) utilities in jurisdictions such as India (Andhra 40 Pradesh), Morocco, and Vietnam display JDVVNL (India – Rajasthan) exceptionally high levels of efficiency, losing no more than 5 percent of revenues. Strikingly, 20 APEPDCL (India – Andhra Pradesh) the performance of this second group is as good as that of the handful of countries that APSPDCL (India – Andhra Pradesh) have largely privatized their power sectors: Colombia, Peru, and the Philippines. 0 A strong overlap exists, however, between Public Private Public vs. private utilities the best-performing public utilities and the Note: Red boxes indicate utilities that have seen privatization rollback. The cost worst-performing private utilities. The picture recovery ratio for the publicly owned utilities appears on the left-hand axis, whereas that for the private utilities appears on the right-hand axis. C3 = full cost becomes clearer by disaggregating results recovery. from the country level to the utility level (­ figure 9.11). As before, the efficiency perfor- mance of public utilities is plotted on the left Finding #5: Although private sector axis and that of private utilities is plotted on the participation is also strongly associated right axis. In contrast to the analysis of cost with distribution efficiency, a significant recovery (see figure 9.9) where public and pri- subset of publicly owned utilities is able vate utilities were clearly segmented, in the case to match this performance of efficiency there is substantial overlap in per- A moderately strong association exists between formance between the two groups, with the private sector participation and distribution overall range running from 5 percent to 25 per- efficiency. The panel data regression analysis cent of revenues lost to inefficiency in both reported in table 9.3 was able to capture only cases. As noted in figure 9.10, a significant clus- the relationship between private sector partici- ter of public utilities performs as ­ efficiently as pation and system losses. The small sample some of the private ones, including ONEE permits a deeper analysis of efficiency, which (Morocco), Hidrandina (Peru), EVN (Vietnam), incorporates both system losses and collection and the utilities from the Indian state of Andhra losses into a single aggregate measure of distri- Pradesh. It is also striking to see privatized utili- bution efficiency. For every percentage-point ties such as Karachi Electric (Pakistan) and increase in private sector participation in the WESCO (Indian state of Odisha) whose Did Power Sector Reform Deliver Better Sector Outcomes? 293 performance is comparable to the worst of the FIGURE 9.10 A substantial minority of publicly owned utilities publicly owned utilities. At the same time, some performs as efficiently as private ones of the worst-performing publicly owned utilities 100 Vietnam Kenya Morocco Peru include cases of failed privatization such as the Colombia Philippines India – Andhra Ukraine Distribution e ciency as share of revenue (%) Dominican Republic, CESU in the Indian state Pradesh of Odisha, and Senegal. 90 Tanzania Uganda These findings point to potential selectivity Senegal and survivor effects with the privatized utilities. Pakistan Tajikistan The comparison between public and private India – Rajasthan 80 utilities is not entirely even-handed for two Dominican Republic reasons, as the results for cost recovery Egypt, India – Odisha Arab Rep. figure 9.9) and efficiency (figure 9.11) appear (­ to show. First, there is a selectivity effect: utili- 70 ties that already perform relatively well (such as those close to cost recovery) tend to be those R = 0.3267 most likely to be selected for privatization. 60 Second, there is a survivor effect: privatized 0 10 20 30 40 50 60 70 utilities that experience poor performance on Private sector participation index (%) efficiency or cost recovery over time are likely Private Public Mixed to revert to state control. These two consider- ­ tandard Note: A sensitivity analysis was carried out for exclusion of outliers two s ations mean that the causality between private deviations from the mean. It resulted in the exclusion of Peru and the R 2 of 0.2803, and does not materially affect the analysis. KPLC in Kenya is a majority sector participation and utility performance is government-owned utility (51 percent of the shares owned by the government); ­ likely to run in both directions. however, it is listed on the Nairobi stock exchange, and its day-to-day functioning is more in line with a private utility. Thus, for the purpose of the Rethinking project we Well-performing public utilities show gover- classify KPLC as private. nance practices that are much closer to those of private utilities than their underperforming state- panel b). Human resource management is where owned counterparts. Comparing well-­ the largest differences across groups appear, with performing public utilities with their under­ greater ability to fire underperforming employ- performing state-owned counterparts serves to ees and increased tendency to conduct transpar- shed light on the governance practices that seem ent and objective hiring processes (figure 9.12, to be most critical to enhancing the performance performing public and pri- panel c). Finally, well-­ of public utilities. Interestingly, the g ­ overnance vate utilities are much more likely to make use of practices of this well-performing public sector modern information and technology systems group have more in common with well-per- (figure 9.12, panel d). forming private utilities than with their state- owned counterparts. In terms of board gover- Finding #6: Country context is again a nance, the largest differences across groups can strong determinant of sector outcomes; be found in their greater freedom to appoint a however, reform measures are also chief executive officer, higher exposure to audits, found to contribute significantly, and increased tendency to publish annual reports particularly in low-income countries (figure 9.12, panel a). In terms of financial disci- pline, the largest differences across groups can be ­ ector Econometric analysis shows that certain s found in their greater propensity to publish reform measures are significantly associated accounts that conform with international finan- with improved sector outcomes. As noted ear- cial reporting standards and their explicit costing lier in the chapter, the panel data regression of their public service obligations (­ figure 9.12, provides a statistically rigorous overview of 294 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 9.11 Among the least efficient utilities are those that that local specificities appear to have a much experienced privatization reversals larger impact on power sector outcomes than NPC (Vietnam) Meralco (Philippines) the extent of power sector reform. For this rea- 100 HCMCPC (Vietnam) EPM (Colombia) Luz del Sur (Peru) son, attention in table 9.7 focuses on the speci- Hidrandina (Peru) ONEE (Morocco) Codensa (Colombia) fication incorporating both time trend and APSPDCL (India – Andhra Pradesh) Dniproblenergo (Ukraine) APEPDCL (India – Andhra Pradesh) country fixed effects. In addition to estimating Beneco (Philippines) the model on the full sample of 88 countries, LESCO (Pakistan) Kenya Power two separate nested models are estimated Khmelnitskoblenergo (Ukraine) for the low- and m ­ iddle-income countries of the sample. Statistical tests reject the hypothesis UMEME (Uganda) Distribution e ciency as share of revenue (%) of no structural differences between these sub- 90 sets, indicating that ­ coefficients are significantly different between the two. Senelec (Senegal) When it comes to security of supply, private TANESCO (Tanzania) sector participation seems to be the only reform with a significantly positive effect, particularly Barki Tojik (Tajikistan) in low-income countries. Normalized capacity Alexandria (Egypt, Arab Rep.) per million population is the only measure of JDVVNL (India – Rajasthan) security of supply that is available as a long- 80 JVVNL (India – Rajasthan) term time series. None of the reform measures Edesur (Dominican Republic) except private sector participation shows any Edenorte (Dominican Republic) WESCO (India – Odisha) statistically significant relationship with nor- CESU (India – Odisha) malized capacity (table 9.7). This effect is both Karachi Electric (Pakistan) larger and more significant for the low-income Cairo North (Egypt, Arab Rep.) country group, where one megawatt per ­ million population is added for every ­ percentage-point increase in private sector participation; how- 70 ever, there is no statistical significance for the Public Private Public vs. private utilities middle-income country group. Source: World Bank elaboration based on Rethinking Power Sector Reform utility When it comes to electrification, private sec- database 2015. tor participation again emerges as significant, Note: Red boxes indicate utilities that have seen privatization rollback. The effi- ciency performance of the publicly owned utilities appears on the left-hand axis, particularly in low-income countries. For the whereas that for the private utilities appears on the right-hand axis. electrification regression, the sample is cur- tailed to those countries that had not yet relationships between high-level reform indica- reached 90 percent at the start of the study tors and different dimensions of sector out- period in 1995, because otherwise the depen- come. Five different model specifications are dent variable becomes truncated as countries considered, which vary according to whether reach universal access. There are some puz- control variables are used, and whether time zling results with both regulation and restruc- trends and country fixed effects are added to turing appearing with significantly negative the regression. The statistical analysis shows relationships. Private sector participation is the that the results are quite sensitive to the model only reform that emerges with a significant specification in many cases (annex 9B). In gen- positive effect on electrification. As before, this eral, however, the incorporation of country effect is larger for low-income countries, where fixed effects greatly improves the overall every percentage-point increase in private explanatory power of the regressions, indicating s ector participation is associated with a ­ Did Power Sector Reform Deliver Better Sector Outcomes? 295 FIGURE 9.12 Efficient public and private utilities score relatively higher on management and governance good practices a. Corporate governance b. Financial discipline Board is the final decision making 11 50 57 Financial accounts are 100 body in appointing CEO 100 publicly disclosed 100 50 Audit committee of the board 78 20 83 Financial accounts meet 67 international standards 100 Minority shareholders' rights 10 33 are protected 67 Public service obligations 20 67 are explicitly defined 83 Private or public shareholders 0 22 appoint board 83 Utility is subject to state 70 80 100 auditing procedures 50 Utility publishes an annual report 100 100 70 Utility pays dividends to 0 22 Function of company secretary 89 shareholders 83 exists 83 0 20 40 60 80 100 0 20 40 60 80 100 Share of utilities (%) Share of utilities (%) c. Human resources d. Information technology Employees can be fired for poor 50 IT system to support 60 100 100 performance distribution management 100 83 Recruitment involves short-listing 70 IT system to support energy 60 100 89 candidates management 100 50 Recruitment involves interviewing 60 IT system to support 60 89 89 candidates 100 incidence resolution 83 Annual sta performance 80 Advanced Metering 30 100 56 reviews exist Infrastructure (AMI) 80 100 70 20 Sta training policy exists 89 Commercial management 44 100 system (CMS) 60 0 20 40 60 80 100 0 20 40 60 80 100 Share of utilities (%) Share of utilities (%) Underperforming SOEs Well-performing SOEs Well-performing private utilities Note: CEO = chief executive officer; IT = information and technology; SOE = state-own enterprise. 0.3-­percentage-point increase in electricity is rather small, with an increase of just access, but disappears altogether in middle-­ 0.4 percentage points. In the case of middle-­ income countries. income countries, competition and private When it comes to renewable energy, there sector participation also have a significant is a consistent positive impact from regulation positive effect, although again the effect is not and private sector participation, whereas all that large. For low-income countries, the other effects differ by income grouping. The coefficient on competition, though significant, presence of a regulator has a significant posi- is surprisingly negative. tive effect on a country’s renewable energy These broad statistical results are comple- share across income levels, although the effect mented by a more detailed analysis of reform 296 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9.7 Impact of reform measures on final outcomes for each country and its relationship to sector large sample outcomes. Low-income Middle-income Regression coefficients Full sample countries countries Finding #7: Although many countries Dependent variable: normalized capacity made some progress toward security Regulation ~ ~ ~ of supply with the 1990s reform Restructuring ~ ~ ~ model, few were able to achieve that Competition ~ ~ ~ goal completely, despite harnessing Private sector participation 61.14** 134.0*** ~ substantial private investment Observations 1831 1008 819 Almost all developing countries face security R-squared 0.962 0.928 0.958 of supply challenges; addressing these chal- Dependent variable: access to electricitya lenges was one of the central concerns of the Regulation –1.775*** –2.369*** ~ 1990s power sector reform model. With elec- Restructuring ~ ~ –6.127*** tricity demand typically growing above five Competition ~ ~ ~ percentage points per annum, maintaining Private sector participation 19.25*** 29.66*** ~ supply–demand balance in the developing Observations 1314 924 378 world calls for substantial and timely invest- ments in new capacity. Furthermore, many R-squared 0.973 0.958 0.978 countries find their power supply overly Dependent variable: renewable energy share exposed to exogenous factors, in particular Regulation 0.461*** 0.417*** 0.437*** droughts and oil price shocks, calling for Restructuring ~ ~ ~ greater diversification of power generation Competition ~ –2.012*** 0.613* capacity. Even when these issues are over- Private sector participation 1.487*** ~ 1.733*** come, weaknesses in the transmission and Observations 1760 960 780 distribution grids may continue to undermine R-squared 0.974 0.975 0.973 the reliability of supply to consumers. According to the 1990s power sector reform Note: Ordinary least squares regressions on panel data set of 88 countries for the period 1995–2015. A separate regression is estimated for each of the intermediate model, security of supply could best be outcomes on the full set of reform variables capturing regulation, restructuring, addressed by opening the generation segment competition, and privatization. The analysis considers a number of different ways of capturing control variables in terms of time trend, country-specific fixed effects, to entry and investment by the private sector, and control variables capturing system size and GDP per capita. The preferred ideally in the context of a competitive whole- specification, incorporating time trend and country-specific fixed effects (but no control variables) is reported here. All the other specifications can be viewed in sale market, backstopped by a financially the annex (for complete results refer to annex 9B, tables 9B.3, 9B. 5, and 9B.6). robust distribution sector. ­ Coefficients are scaled in units of the intermediate outcome variable. a. For the specific case of the regression for access to electricity, observations Although such reforms were among the are dropped for any countries that had already reached electrification rates most widely adopted, achieving full security of above 90 percent in 1995. supply has proved quite challenging. Reflecting Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, ~ = not significant. the multidimensional nature of this concept, several different measures of security of supply, measures and associated sector outcomes which do not necessarily correlate closely with based on the smaller sample. This additional each other, are used here. They are normalized analysis makes it possible to examine a wider capacity, capacity diversification, supply reli- range of sector outcome indicators than those ability, and capacity growth. Overall, it is strik- that were available for the long-term panel ing that, although countries embracing exten- data analysis. It also allows for a deeper quali- sive power sector reforms tend to do quite well tative discussion of the actual reform path of overall on these various measures of security of Did Power Sector Reform Deliver Better Sector Outcomes? 297 supply, the converse is not necessarily the case Normalized capacity metrics reached levels (figure 9.13). Moreover, some of the worst per- typical of middle-income countries in most formers (notably, India and Pakistan) are parts of the world, except for Sub-Saharan among those that have pursued a significant Africa. The first measure relates to capacity amount of reform. normalized by population (figure 9.13, panel FIGURE 9.13 Performance on various measures of security of supply with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) a. Normalized capacity b. Capacity diversification Philippines Philippines Peru Peru Ukraine Ukraine Uganda Ugandaa Colombia Colombia Dominican Republic Dominican Republic India India Pakistan Pakistan Kenya Kenya Vietnam Vietnam Senegal Senegal Egypt, Arab Rep. Egypt, Arab Rep. Morocco Morocco Tanzania Tanzania Tajikistan Tajikistan –200 0 200 400 600 800 1,000 1,200 –0.75 –0.50 –0.25 0 0.25 0.50 0.75 1.00 Capacity (MW/million population) Fuel concentration index c. Capacity growth d. Supply reliability Philippines Philippines Peru Peru Ukraine Ukraine Uganda Uganda Colombia Colombia Dominican Republic Dominican Republic India India Pakistan Pakistan Kenya Kenya Vietnam Vietnam Senegal Senegal Egypt, Arab Rep. Egypt, Arab Rep. Morocco Morocco Tanzania Tanzania Tajikistan Tajikistana –0.5 0 0.5 1.0 1.5 2.0 2.5 0 50 100 150 200 250 300 Ratio of capacity growth to System Average Interruption Frequency growth in peak demand Index (SAIFI) 2015 Note: In panel a, dark-shaded bars represent capacity in the prereform era; the light-shaded bars represent the capacity added during 1990–2015. In panel b, dark-shaded bars represent average value in 2010–15; light-shaded bars represent the change in values from prereform era. For all panels the color of the bar follows the evaluation framework set up in table 9.2. MW = megawatt. a. Data not available. 298 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD a). For the Eastern European countries, avail- spread across a variety of different energy able capacity was already unusually high prior sources. Almost all countries succeeded in to the reform, reflecting the Soviet legacy. Most improving capacity diversification over the of the middle-income countries in the sample period 1990–2015, with Egypt being the main appear to operate adequately in the range of exception. Nevertheless, only a handful of 200–400 megawatts per million population. countries has actually achieved a well-balanced Many low-income countries started out in portfolio of energy sources, with Morocco, 1990 with capacity well below these levels; Pakistan, and the Philippines being the best however, there is a marked contrast between examples (figure 9.13, panel b). Looking across Sub-Saharan African countries that despite countries in the small sample, diversification some progress remain below this threshold, shows a significant relationship with reforms and countries elsewhere that succeeded in such as restructuring, regulation, and competi- growing their capacity into this range (such as tion, with competition ­having by far the largest India, Morocco, and the Philippines). The most effect (figure 9.15). A one-percentage-point dramatic example is Vietnam, whose normal- increase in competition leads to improvement ized capacity grew sixfold over the period of two percentage points on diversification.4 1990–2015. Overall, normalized capacity is Almost all countries are managing to grow most strongly associated with income per cap- generation capacity at least as rapidly as peak ita, which alone explains about 67 percent of demand. The third measure relates to the the cross-country variation in this indicator extent to which capacity expansions are keep- (figure 9.14). ing pace with the growth of peak demand in Few countries have achieved a fully diversi- recent years. On this measure, almost all coun- fied power mix. The second measure relates to tries do well, managing to grow capacity more capacity diversification and indicates the extent rapidly than demand. Notable exceptions, to which supply is concentrated rather than however, are Pakistan and Tanzania. FIGURE 9.14 Normalized capacity is strongly associated with income per capita 2,500 Normalized capacity, 2010–15 (MW/million population) 2,000 Russian Federation 1,500 Armenia Ukraine Kazakhstan Chile Romania 1,000 Malaysia Poland Belarus Tajikistan China South Africa Brazil Lao PDR Thailand Jordan Lebanon Mexico 500 Vietnam Colombia India Bolivia Pakistan Maldives R = 0.6726 Angola 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 GDP per capita, 2010–15 (US$) Note: A sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclu- sion of 11 countries and the R 2 of 0.6531, and does not materially affect the analysis. MW = megawatt. Did Power Sector Reform Deliver Better Sector Outcomes? 299 Performance on supply reliability varies FIGURE 9.15 Power sector diversification is across different geographical regions. The significantly related to various reform steps fourth measure relates to supply reliability 0 from the consumer perspective as measured –0.5 Coe cients by the standard System Average Interruption –1.0 Frequency Index (SAIFI). Generally, perfor- –1.5 mance on this measure is good across Latin –2.0 America, Eastern Europe, and North Africa –2.5 Restructuring** Regulation* Competition* and moderate across Asia and Sub-Saharan Note: Ordinary least squares multivariate regression ­ analysis Africa. Nonetheless, the Dominican Republic for the Rethinking Power Sector Reform country sample and especially India stand out as having (17 economies). The chart reports the results of a series of exceptionally high levels of service interrup- cross-sectional regressions modeling the impact of various reform steps on the concentration index for power generation, tions (figure 9.13, panel d). Perhaps not sur- with lower scores representing greater diversification. GDP per prisingly, SAIFI also seems to improve with capita and system size have been used as control variables. For complete results, refer to annex 9C, table 9C.5. more efficient utilities (figure 9.16) Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. Overall, the qualitative analysis shows that the implementation of the 1990s power sector FIGURE 9.16 Reliability of supply is associated with greater reform measures did not of themselves fully distribution efficiency guarantee the achievement of security of Ukraine ­supply (table 9.8). By looking at the underly- 120 Peru Morocco ing narratives for security of supply in the Kenya Utility e ciency as share of revenue (%) Vietnam Colombia 15 ­c ountries of the Power Sector Reform 100 India – Andhra Pradesh Uganda Observatory, it becomes possible to gain some Tanzania 80 Dominican insights into the channels of causality. Peru Philippines India – Odisha Republic and the Philippines stand out as the countries 60 Senegal Pakistan where the model was most successful. These Egypt, Arab Rep. countries went as far as introducing wholesale 40 power markets and, as a result, were able to 20 finance generation capacity expansions almost entirely through private investment, while R = 0.3086 0 100 200 300 400 500 maintaining capacity adequacy and diversifica- tion and providing quite reliable supply to System Average Interruption Frequency Index (SAIFI), 2015 c onsumers. Another group of countries—­ ­ Note: A sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of the Indian state of Odisha comprising Colombia, Morocco, and Vietnam— and Egypt, Arab Rep. and the R 2 of 0.4002 and does not materially affect the also did relatively well, although their private ­analysis. financing shares were substantially lower than the first group. A third group of countries— supply, but—following an early unsuccessful including the Dominican Republic, India, independent power producer program—did so Kenya, and Uganda—made some progress, but through a completely different approach based the countries continue to face challenges with on intergovernmental bilateral deals (primarily either reliability or diversification. A last group with Germany) to support new investments of countries—despite adopting significant in power generation. Since 2015, a similar reforms—cannot be considered to have approach has been adopted in Pakistan with the achieved security of supply. They are Pakistan, China–Pakistan Economic Corridor, which com- Senegal, and Tanzania. Finally, Egypt has (after gigawatts prises a support package including 17 ­ a period of supply crisis) achieved security of of new power generation capacity. 300 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9.8 Qualitative analysis of underlying causality for security of supply trends Hypothesis: Security of supply would be achieved by opening up generation to the private sector Performance Security PSP share in of supply generation Supports 2015 1990–2015 (%) Policy steps hypothesis Colombia Medium 70 Created successful wholesale market and attracted private Largely investment, but hydro exposure to drought risk still high. Dominican Medium 65 Created simple power market and attractive private sector Partially Republic investment, but power supply still unreliable (SAIFI). Egypt, Arab Rep. High 8 Initial IPP program stalled and capacity expansion achieved No through state-driven bilateral deals. India Medium 46 Created wholesale market for electricity attracting private Partially investment and growing reserve margin, but SAIFI an issue. Kenya Medium 40 IPP program attracted significant private investment; still scope to Partially improve SAIFI and capacity diversification. Morocco High 50 IPP program attracted significant private investment; adequate Largely capacity and reliability of service. Pakistan Low 57 IPP program attracted private investment, but not enough to meet No fast-growing demand because of sector financial crisis. Peru High 92 Created successful wholesale market largely reliant on private Yes investment; supply reliable, diversification ongoing. Philippines High 96 Created successful wholesale market largely reliant on private Yes investment; supply both reliable and diversified. Senegal Low 50 IPP program attracted private investment, but capacity remains No low, reliability poor, and diversification weak. Tanzania Low 32 IPP program had limited success, but capacity remains low, No reliability poor, and drought exposure substantial. Uganda Medium 55 IPP program eventually delivered investment, but capacity remains Partially low, reliability mediocre, and hydro dominant. Vietnam High 38 IPP program attracted significant private investment; capacity Largely grew rapidly and diversified; few reliability issues. ­ ajikistan. Note: Ukraine is excluded because it has excess capacity, demand is falling, and security of supply is therefore not an issue; the same can be said of T IPP = independent power producer; PSP = private sector participation; SAIFI = System Average Interruption Frequency Index. Finding #8: Progress on electrification some form of electricity service, whether on- was made by countries with widely or off-grid. Acknowledging that access with- differing approaches to power sector out affordability would be meaningless, the reform and appears to be driven 2030 goals also underscore the importance of primarily by targeted public investment affordability. Although the United Nations has as countries reach higher income levels no official indicator for affordability, in the lit- The social inclusion agenda that became erature it is often defined as the ability to avail ascendant after 2010 underscored the impor- an average volume of electricity consumption tance of universal access to affordable energy for no more than 5 percent of the budget of services. SDG7.1.1 sets the objective for uni- the bottom 40 percent of the income distribu- versal access to affordable, reliable, modern, tion. As noted earlier in the chapter, neither and sustainable electricity for all by 2030. access nor affordability was a central focus Universal access is usually measured in terms of the 1990s power sector reform model. of the percentage of households with access to Although some middle-income countries had Did Power Sector Reform Deliver Better Sector Outcomes? 301 already very much engaged in this agenda by this period and came close to achieving uni- the 1990s, that aspiration only became main- versal access. At the other extreme, Tanzania stream across the developing world following barely electrified 8 percent of its population the promulgation of SDG7. and had below 20 percent coverage even in Performance on electrification does not 2015. Strong performances on electrification seem to reflect the overall extent of power can be found both among countries that sector reform in a country but is closely asso- embraced power sector reform (Peru and the ciated with per capita income. During the pre- Philippines) and among ­ countries that took a reform period around 1990, countries already much more cautious approach to power sec- had a wide disparity of electrification rates, tor reform (Morocco and Vietnam). At the ranging from those in Latin America and same time, relatively weak performances on Eastern Europe that were already close to electrification can be found among countries universal access all the way to countries in that embraced reform (Uganda) as well as Sub-Saharan Africa with electrification rates those that eschewed it (Tanzania) (figure 9.17, no higher than 30 percent. The amount of panel a). By far the strongest and most consis- progress made on electrification over the 25 tent driver of electrification is gross domestic years between 1990 and 2015 also varies product per capita with a correlation of about greatly. At one extreme, Morocco electrified 0.35 in the cross-sectional analysis, dropping more than 50 percent of its population during to 0.17 for the difference in differences. FIGURE 9.17 Performance on social inclusion, countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) a. Access b. A ordability Philippines Philippines Peru Perua Ukraine Ukraine Uganda Ugandaa Colombia Colombia Dominican Republic Dominican Republic India India Pakistan Pakistan Kenya Kenya Vietnam Vietnam Senegal Senegal Egypt, Arab Rep. Egypt, Arab Rep. Morocco Morocco Tanzania Tanzania Tajikistan Tajikistan 0 20 40 60 80 100 0 5 10 15 20 25 Electrification rate, 2010–15 (%) Share of GNI of bottom 40 (%) Note: Dark-shaded bars represent electrification in the prereform era, and the light-shaded bars represent electrification during 1990–2015. For all panels the color of the bar follows the evaluation framework set up in table 9.2. Bottom 40 = population in the bottom 40 percent of the income distribution; GNI = gross national income. Affordability data for average consumption in Peru and Uganda not available. a. ­ 302 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD FIGURE 9.18 The strongest driver of electrification has been GDP per capita Vietnam Algeria Belarus 100 Bolivia Armenia Tajikistan Malaysia Argentina Chile Nepal India South Africa 80 Mongolia Pakistan Ghana Electrification rate, 2010–15 (%) Afghanistan 60 Cameroon Nigeria Cambodia Congo, Rep. 40 Benin Sudan Vanuatu Angola Kenya Zambia Ethiopia 20 R = 0.6457 Burundi 0 2,000 4,000 6,000 8,000 1,0000 1,2000 1,4000 GDP per capita, 2010–15 (US$) A graphical representation reveals a nonlinear Overall, the qualitative analysis confirms a relationship between gross domestic product consistent pattern of progress on electrification per capita and electrification with a particu- in countries with strong government-led tar- larly steep ascent in the US$1,000–3,000 per gets, institutions, and investment subsidies. For capita window, and the resulting R-squared the small sample, it is possible to go behind the rising to over 0.60 (­figure 9.18). quantitative analysis to examine the underly- Regarding affordability, this factor largely ing electrification narratives, which shed light affects the local economic context and little rela- on the process through which progress toward tion is found to reform measures. Most coun- universal access is achieved. These qualitative tries meet the basic affordability criterion of stories, summarized in table 9.9, show a keeping the cost of average consumption below remarkably consistent pattern across countries. 5 percent of the budget of the bottom 40 ­percent Specifically, in country after country, the drive of the income distribution (figure 9.17, panel b). for electrification entailed a strong political The few exceptions usually reflect at least one of commitment backed up by sustained public the following three factors: high price, low investment typically channeled through a rea- income, and high consumption. For instance, sonably competent utility. Moreover, for most the Dominican Republic, the Philippines, and countries, the main period of acceleration in Senegal all charge average residential tariffs in the electrification process is unrelated to the excess of US$0.20 per kilowatt-hour. Because period when the 1990s reform model was these factors are unrelated to reform, it is per- being implemented. In several of the mid- haps not surprising that little association is dle-income countries—such as India, Morocco, found between affordability and power sector and Vietnam—the electrification drive pre- reform measures, and even with respect to the dated the implementation of the 1990s reforms intermediate outcomes of efficiency and cost and simply continued in the background recovery. through its own established channels even as Did Power Sector Reform Deliver Better Sector Outcomes? 303 TABLE 9.9 Qualitative analysis of underlying causality for electrification trends Hypothesis: Progress on electrification had little to do with the 1990s reforms, but was driven by government-funded electrification programs led by public utilities and other state institutions Performance   Average Change Supports   2010–15 (%) 1990–2015 (%) Policy steps hypothesis Colombia 97.4 +5.5 Areas with low access were identified as special zones, and targeted Yes electrification programs were backed by resources from various special funds. Dominican 98.2 +15.0 Rural and suburban electrification unit in the CDE is responsible for access and Yes Republic made efforts to encourage development of isolated grids to improve access. Egypt, 99.8 +2.1 Rural electrification authority established in 1971 had completed its mandate Yes Arab Rep. by 2007 and was disbanded. India 78.8 +25.7 Multiple policies at national and state levels targeted resources toward utility- Yes led electrification programs. Kenya 31.3 +20.2 Since 2010, government sets electrification targets and directs special Yes funding to national utility for grid rollout; measures also taken to encourage off-grid solar. Morocco 94.6 +51.9 Rural electrification program targeted government resources to national Yes utility for electrification; off-grid solar systems also encouraged. Pakistan 71.0 +4.8 No national-level agency for rural electrification, and no special incentives for Yes utilities to expand access. Peru 91.3 +26.9 Rural electrification law in 2006–07, long after wider power sector reform Yes law in 1992; substantial government funds targeted to rural electrification. Philippines 86.8 +17.7 National Electrification Administration drives rural electrification program Yes implemented by cooperatives; utilities responsible for urban grid rollout. Senegal 57.7 +26.5 Attempt to use private rural concessions to expand access only partially Largely successful; current focus is on public funding for utility-based electrification. Tanzania 16.4 +8.0 Not until 2005–07 were the Rural Energy Fund and Rural Energy Agency Yes created to channel donor resources to on- and off-grid electrification programs. Uganda 16.4 +8.0 Rural private concessions were unsuccessful in expanding access; privatized Largely national utility made limited contribution to access, leaving government to drive electrification. Vietnam 99.3 +17.8 Access drive predates power sector reform, based on channeling public Yes. resources to electrification programs led by the national utility. Note: Tajikistan and Ukraine are excluded from this table since both countries had already achieved universal access to electricity at the time of independence. CDE = Corporación Dominicana de Electricidad. other reforms took place. In many low-income to the balance between financial viability and countries—such as Kenya, Tanzania, and tariff affordability and has largely been Uganda—progress in electrification came many dropped. years after the 1990s reform process and entailed the creation of additional institutions Finding #9: Progress on (Rural Electrification Agencies) and funds decarbonization of the electricity (Rural Electrification Funds). Only two coun- sector has historically been an tries (Senegal and Uganda) attempted to apply unintended by-product of measures to the principles of the 1990s reform model to the improve security of supply electrification process itself, by tendering rural concessions to the private sector. In both cases, The 2030 Agenda for Sustainable Develop- the approach encountered challenges related ment calls for rapid decarbonization of the 304 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD electricity sector, with a particular emphasis been just as likely to improve as to deteriorate on renewable energy. The 2015 Paris Climate over time historically. The only countries with Agreement holds the world accountable for low-carbon power sectors are Colombia and the implementation of rapid decarbonization Tajikistan, which in 1990 already enjoyed strong action across sectors aimed at limiting climate reliance on hydroelectric power, a reflection of change to 1.5 degrees Celsius. As the largest the availability and cost-effectiveness of this carbon-emitting sector, energy has featured resource rather than a particular focus on decar- among the Nationally Determined Contribu- bonization (figure 9.19). As of 2015, most coun- tions of many countries. Decarbonization of tries in the Power Sector Reform Observatory the sector can be achieved by improving still have highly carbon-intensive electricity sec- energy efficiency, substituting renewables for tors (more than 500 grams of carbon dioxide fossil fuel generation, and shifting toward fos- per kilowatt-hour). Moreover, a broadly equal sil fuels with lower ­carbon content (such as number of countries saw their carbon intensity natural gas). Further specificity is provided by rise and fall during the years 1990–2015, and SDG7.2, which calls for a doubling of the the rise and fall happened across the power sec- share of renewable energy in total final tor reform spectrum. Programs to explicitly pro- energy consumption globally by 2030. These mote the uptake of modern renewable energy new political agreements came right at the started to pick up only toward the end of this end of the historical period considered here, period, and the programs for most countries but some pioneering countries (such as have been small relative to the overall scale of Morocco) had already begun to espouse the electricity consumption. Nevertheless, Colom- decarbonization agenda. bia, Peru, the Philippines, and Vietnam all saw In many countries, however, electricity significant expansions in their shares of modern remains carbon intensive and this situation has renewable energy. FIGURE 9.19 Performance on environmental sustainability, with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) a. Carbon intensity of electricity b. Modern renewable energy share of TFEC Philippines Philippines Peru Peru Ukraine Ukraine Colombia Uganda Dominican Republic Colombia Dominican Republic India India Pakistan Pakistan Kenya Kenya Vietnam Vietnam Senegal Senegal Egypt, Arab Rep. Egypt, Arab Rep. Morocco Morocco Tanzania Tanzania 54 Tajikistan Tajikistan –400 –200 0 200 400 600 800 1,000 –10 –5 0 5 10 15 20 Carbon intensity (gCO2/kWh) Modern renewable share in TFEC (%) Note: Dark-shaded bars represent average value in 2010–15; light-shaded bars represent the change in values from prereform era. For all panels the color of the bar follows the evaluation framework set up in table 9.2. gCO2/kWh = grams of carbon dioxide per kilowatt-hour; TFEC = total final energy consumption. ­ Did Power Sector Reform Deliver Better Sector Outcomes? 305 Countries with relatively low carbon inten- diversified into fossil fuels (mainly natural gas) sity tended to do better in terms of operating as a way of reducing exposure to drought risk cost recovery (figure 9.20). This finding is (as in Colombia, Kenya, Peru, and Tanzania), at hardly surprising because fuel purchase the same time increasing the carbon intensity of expense tends to be one of the largest compo- their electricity sectors. In other cases, overreli- nents of operating costs for the power sector in ance on oil-fired power generation also created many countries. By contrast, renewable the need to diversify to reduce exposure to oil sources of energy have very low operating price shocks, which led to falling carbon inten- costs because of the fact that they do not run sity as a result of displacing oil with either hydro on fossil fuels. (as when Senegal began to import power from Overall, the qualitative analysis shows that Manantali) or less-carbon-intensive fossil fuels for most countries, changes in decarbonization (as with gas in the Dominican Republic, the have historically been an unintended by-­ Philippines, and Ukraine). The only clear excep- product of measures to promote security of tion to this pattern is Morocco, where a strategic ­supply (table 9.10). In order to understand what decision was taken to pursue an aggressive lies behind the numbers, it is helpful to zoom in renewable energy program with a view to posi- on the narrative for each of the 15 countries of tioning as a first mover, particularly with regard the Power Sector Reform Observatory. These to concentrated solar power. This decision led to stories provide strong support for the view that a sizable reduction in carbon intensity of elec- in most countries, power generation decisions tricity as renewables came in to displace fossil have been driven by security of supply consider- fuels. Nevertheless, as of 2015, the carbon ations, with unintended consequences—both intensity of electricity remained high in absolute positive and negative—for decarbonization. For terms: coal- and oil-fired plants still account for instance, several countries started out as overly more than half of generation capacity compared dependent on hydropower and consciously to 12 percent for wind and solar. FIGURE 9.20 Carbon intensity shows strong inverse correlation with operating cost recovery 900 India – Andhra Pradesh R = 0.4466 Carbon intensity of generation (gCO2/kWh) 800 India – Rajasthan India – Odisha 700 Senegal 600 Dominican Philippines Republic 500 Ukraine Egypt, Arab Rep. 400 Tanzania Pakistan Vietnam 300 Kenya Peru 200 Colombia 100 Tajikistan 0 20 40 60 80 100 120 140 160 Operating cost recovery (%) Private Public Mixed Note: A sensitivity analysis was carried out for exclusion of outliers two standard deviations from the mean. It resulted in the exclusion of Tajikistan and the R 2 of 0.5058, and does not materially affect the analysis. gCO2/kWh = grams of carbon dioxide per kilowatt-hour. The type of ownership of utilities is represented by the color of dots. 306 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9.10 Qualitative analysis of underlying causality for decarbonization trends Hypothesis: Decarbonization of the electricity sector has sometimes been an unintended consequence of policies to improve security of supply, rather than a deliberate policy effort Performance Average 2010–15 Change 1990–2015 Supports   (gCO2/kWh) (gCO2/kWh) Policy steps hypothesis Colombia 160.6 –27.9 Doubled natural gas generation to diversify from hydro Yes drought risk, but renewables also increased. Dominican 572.7 –294.7 Massive financial crisis due to oil price shocks prompted Yes Republic diversification toward natural gas and coal. Egypt, Arab Rep. 445.9 +99.4 Until recent renewable program, capacity expansions Yes have been entirely done using natural gas. India 794.2 –37.7 Generation coal dominated because of security of Yes supply concerns; recent shift toward scaling renewables. Kenya 228.6 +146.0 Diversification away from hydropower toward rapid Yes expansion of oil and geothermal capacity. Morocco 700.7 –136.5 Explicit decarbonization policy objective and creation of No MASEN as specialized entity to promote solar energy. Pakistan 416.9 +5.1 Most new generation added is from coal and natural gas Yes driven by security of supply concerns. Peru 271.0 +116.1 Shift toward natural gas generation to diversify from Yes hydro drought risk for security of supply. Philippines 549.0 +65.4 Almost all new generation added since EPIRA in 2001 is Yes natural gas and coal, reducing share of oil-fired plants. Senegal 627.1 –313.6 Almost all reduction is because of hydro capacity Yes coming online; oil remains the mainstay. Tanzania 401.5 +366.1 Expanded oil and gas generation since 2004 to diversify Yes from hydro drought risk. Ukraine 446.4 –129.6 Because of declining demand, costly oil-fired generation Yes could be removed to result in lower carbon footprint. Vietnam 398.6 +66.5 Generation expansion for security of supply based on oil Yes and coal; explicit renewables program more recent. Note: Tajikistan is excluded from the table because it was already almost entirely hydropower based in 1990 and has continued to be so. Uganda is excluded because of data limitations. EPIRA = Electric Power Industry Reform Act; gCO2/kWh = grams of carbon dioxide per kilowatt-hour; MASEN = Moroccan Agency for Sustainable Energy. Finding #10: Overall, although some are shaded in green, ­yellow, and red according of the deepest reformers delivered to whether their performance on any particu- good sector outcomes, other countries lar matrix is good, mediocre, or poor, as laid showed that it was possible to achieve out in the rubric in table 9.2. Table 9.11 evalu- comparable performance through ates countries on the basis of their absolute different approaches performance in 2015, without taking starting The multidimensionality of power sector per- conditions into account. Table 9.12 looks at the formance makes it difficult to reach very crisp changes in performance from 1990 to 2015— conclusions regarding the relationship between at least for those variables for which a signifi- reforms and outcomes. Tables 9.11 and 9.12 cantly long time series is available. This exercise bring together the eight different dimensions of serves to reveal that few countries are able to power sector reform outcomes that have been perform consistently well on all eight measures discussed throughout this chapter. Countries of power sector outcomes. It also illustrates Did Power Sector Reform Deliver Better Sector Outcomes? 307 TABLE 9.11 Overview of cross-sectional performance on final outcomes as of 2015 with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) Environmental Security of supply Social inclusion sustainability Capacity Capacity Modern Normalized diversification growth RE capacity (Fuel (Capacity Carbon share Reliability (MW/million Concentration growth/peak Electrification Affordability intensity in TFEC 2015 (SAIFI) population) index) dd growth) (% population) (% of GNI) (gCO2/kWh) (%) Philippines 4 209 0.26 1.05 87 7.35 549 6.8 Peru 10 391 0.44 1.14 91 — 271 12.24 Ukraine 10 1,260 0.38 — 100 3.71 446 1.15 Uganda 42 23 — 1.22 16 — — 1.09 Colombia 11 345 0.57 2.34 97 7.32 161 13.21 Dominican 105 354 0.36 1.04 98 10.01 573 3.46 Republic India 248 0.52 1.89 79 3.09 794 2.28 Pakistan 37 121 0.31 0.35 71 4.18 417 3.05 Kenya 13 49 0.35 1.47 31 4.32 229 2.89 Vietnam 14 433 0.35 1.43 99 10.84 399 7.14 Senegal 31 64 0.75 1.19 58 13.45 627 0.79 Egypt, 0 415 0.62 1.83 100 0.49 446 2.37 Arab Rep. Morocco 3 231 0.32 0.9 95 4.59 701 2.15 Tanzania 47 22 0.42 0.45 16 21.32 401 0.78 Tajikistan — 637 0.98 — 100 0.67 3 54.79 Note: Country results are shaded in green, yellow, and red according to whether their performance on any particular matrix is good, mediocre, or poor. dd = demand; gCO2/kWh = grams of carbon dioxide per kilowatt-hour; GNI = gross national income; MW = megawatt; RE = renewable energy; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption; — = not available. that, overall, there are some dimensions of per- If such reforms are systematically associated formance where many countries are doing rel- with better performance, we would expect to atively well—notably electrification and see the countries toward the top of the table capacity expansion—compared to others— shaded mainly in green, and those toward such as supply reliability, capacity diversifica- the bottom of the table shaded mainly in tion, and carbon intensity—where many red. Reality proves to be more complex. It is countries are struggling. c ertainly true that among the deepest ­ Although top reformers show good perfor- reformers —­ ­ Colombia, Peru, the Philippines, mance, not all reformers perform well and and Ukraine—good performance is in evi- several cautious reformers achieve strong out- dence. For a number of countries engaging in comes. Tables 9.11 and 9.12 list countries in substantial reforms, however, performance is order of their extent of adoption of the 1990s more checkered—the Dominican Republic, power sector reform model, from the most India, Pakistan, and Uganda. Furthermore, reformed country (the Philippines) to the several countries that did not reform so least (Tajikistan). This treatment provides a ­ e xtensively—such as Egypt, Morocco, and straightforward visual check for the relation- Vietnam—exhibit positive performance. ship between reform and sector outcomes. And Kenya, the best performer in 308 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9.12 Overview of performance on differences in final outcomes from 1990 to 2015 with countries ranked by extent of power sector reforms from most reformed (Philippines) to least reformed (Tajikistan) Environmental Change in security of supply Social inclusion sustainability Capacity Capacity Modern Normalized diversification growth Carbon RE capacity (fuel (capacity intensity share Reliability (% change in concentration growth/peak Electrification Affordability (gCO2/ in TFEC 1990–2015 (SAIFI) MW/million) index) dd growth) (% population) (% of GNI) kWh) (%) Philippines — 68 –0.07 — 18 — 65 3.10 Peru — 114 –0.16 — 27 — 53 2.82 Ukraine — 23 a 0.08 — 1a — –130 0.45 Uganda — 95 — — 8 — — 0.49 Colombia — 3 –0.05 — 5a — –28 a 2.51 Dominican — 70 –0.41 — 15 a — –295 2.32 Republic India — 149 –0.01 — 26 — –38 0.47 Pakistan — 37 –0.03 — 5 — 5 –0.41 Kenya — 79 –0.28 — 20 — 146 –0.19 Vietnam — 611 –0.19 — 18 — 67 3.61 Senegal — 137 –0.05 — 26 — –314 0.79 Egypt, Arab — 59 0.22 — 2a — 99 –1.36 Rep. Morocco — 127 –0.25 — 52 — –136 1.47 Tanzania — 20 –0.52 — 8 — 366 –0.36 Tajikistan — –1 a 0.03 — 1a — –4 a –5.79 Note: Country results are shaded in green, yellow, and red according to whether their performance on any particular matrix is good, mediocre, or poor. dd = demand; gCO2/kWh = grams of carbon dioxide per kilowatt-hour; GNI = gross national income; MW = megawatt; RE = renewable energy; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption; — = not available. a. Instances where countries were already performing well and therefore the change during study period is marginal. Sub-Saharan Africa, adopted a middle road competition is significantly associated with with regard to reforms. This pattern becomes higher thermal efficiency of generation plants. further marked when one considers perfor- Further insights are gleaned from the richer mance improvements (table 9.12) as opposed set of variables available for the small sample. to performance alone (table 9.11). Certain aspects of utility governance—notably financial discipline and human resource CONCLUSIONS ­ management—are found to have a significant Certain aspects of power sector reform are association with cost recovery and distribu- strongly associated with utility performance tion efficiency. Cost recovery is also found to in distribution efficiency and cost recovery. be significantly associated with the real func- The panel data econometrics for the large tioning of the regulatory system—as opposed sample confirm that the efficiency of the gen- to the way the system is described on paper. eration and distribution segments is signifi- On private sector participation, the effects cantly associated with regulatory reform, that are particularly difficult to disentangle. Whereas restructuring has a positive influence on sys- the panel data analysis does not find any signif- tem losses in distribution, and that icant impact of private sector participation on Did Power Sector Reform Deliver Better Sector Outcomes? 309 utility performance once country fixed effects broad-based security of supply based on almost are taken into account, the cross-sectional anal- ­ exclusive private investment in generation ysis finds a strong link between private sector capacity, as envisaged in the 1990s model. In participation and cost recovery and, to some other countries—despite significant reforms—­ extent, distribution efficiency. This finding may challenges remain on one or more of the reflect a certain circularity, with cost recovery dimensions of security of supply, and private being both a precondition for private sector par- sector contributions to investment have been ticipation and a means of committing govern- closer to 50 percent in most cases. In some ment to ongoing cost recovery. When the cases, serious security-of-supply issues remain detailed performance of individual public and despite significant reform, with Pakistan one of private utilities in the sample is compared, it is the most striking examples. The panel data striking that all of the private sector arrange- econometrics suggest that private sector partici- ments that have been sustained over time are pation may have been the only reform to in environments where full capital cost recov- contribute significantly to the development of ­ ery has been attained—something that has not normalized capacity, particularly in low-income been fully attained anywhere with state-owned countries, whereas the cross-sectional analysis utilities. At the same time, the relatively good suggests that regulation, restructuring, and levels of distribution efficiency reached by most competition may contribute to improved (but not all) privatized utilities are matched by diversification. a substantial minority of the state-owned utili- With regard to electrification, countries ties in the sample. progressed in ways that were unrelated to the Overall, the link between power sector 1990s reforms. Efforts proceeded in parallel to reforms and final sector outcomes is much the 1990s reforms in almost every country, weaker, despite some evidence that private occasionally predating them but more often sector participation has made a positive con- coming later, as access objectives surfaced on tribution. Some of the countries that carried the political agenda in the 21st century. Just out the deepest reforms (Colombia, Peru, and about every country that made progress on the Philippines) have also done well in terms electrification did so through state-led efforts of final outcomes. Several other countries to plan and finance the rollout of grid tech- implementing major reforms (the Dominican nologies (and increasingly to catalyze the Republic, Pakistan, and Uganda) did not uptake of off-grid technologies) through com- achieve anywhere near the same amount of petent national utilities sometimes supported progress. At the same time, a third group of by specialized rural electrification entities. countries that took a much more cautious Isolated attempts to apply the 1990s approach approach to reform (Morocco, Vietnam) made to rural electrification were largely unsuccess- remarkable gains. Nevertheless, econometric ful. Nevertheless, the panel data econometrics analysis supports the finding that, among the suggest that private sector participation has elements of power sector reform, private sec- had a positive influence on electrification, tor participation is the one that seems to be particularly in low-income countries. most strongly associated with improved sector With regard to decarbonization, progress has outcomes in terms of security of supply, elec- historically been an unintended by-­ product of trification, and renewable energy share. measures to improve security of supply, rather With regard to security of supply, the picture than a conscious effort. Generally, ­ countries is complex. Two countries (Peru and the needing to diversify away from hydropower Philippines) succeeded in achieving a increased their carbon footprint, whereas 310 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD those needing to diversify away from oil-fired For the large sample (N=88), Int Outcome data power generation decreased their footprint. include generation efficiency and system losses Nevertheless, competition reforms do seem to between 1995 and 2015. Reform is the main have been associated with greater progress by variable of interest and includes regulations, encouraging entry of the private sector, which restructuring, competition, and privatization is more inclined to invest in cheaper forms for each country starting from 1995. Xs are of energy, such as natural gas or solar power. country specific control variables that may The panel data econometrics found that reg- have an impact on the Int Outcome such as gross ulatory reforms provided a ­ helpful impetus domestic product per capita and system for the uptake of renewable energy, whereas size. The variables m c, t t, and ∈c,t are country competition and private sector participation and year fixed effects and the error term, seem to have played a hand in middle-income respectively. countries. We assess the impact of reforms on final outcomes over time: ANNEX 9A. ECONOMETRIC ANALYSIS OF POWER SECTOR a0 + a1Reformc,t + a2Xc,t + mc + Final Outcomec,t =  REFORM IMPACTS BASED ON tt + ∈c,t LARGE SAMPLE (88 COUNTRIES) where the dependent variable is the Final Specification Outcome in country c, year t. Final Outcome We assess the impact of reforms on intermedi- includes access, affordability, modern renew- ate output over time: able energy share, carbon intensity, reliability of supply, and normalized capacity. Because of a0 + a1Reformc,t + a 2Xc,t + m c + Int Outcomec,t =  data limitation, however, the estimation results tt + ∈c,t are presented for access, modern renewable where the dependent variable is the Intermedi- energy share, carbon intensity, and normalized ate Outcome (Int Outcome) in country c, year t. capacity. ANNEX 9B. RESULTS OF ECONOMETRIC ANALYSIS OF POWER SECTOR REFORM IMPACTS BASED ON LARGE SAMPLE (88 COUNTRIES) TABLE 9B.1 Impact of reforms on system losses (intermediate outcome), 1995–2015 Complete sample (87) Low-income countries (48) Middle-income countries (39) Dep. var.: system losses (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation 0.438 0.571 –2.167* –2.472** –2.528** –2.336* –1.463 –2.652* –2.218 –3.205** 0.995 1.130 –1.931 –2.223 –1.795 (0.739) (0.723) (1.130) (1.187) (1.145) (1.288) (1.358) (1.595) (1.601) (1.617) (0.897) (0.895) (1.729) (1.735) (1.721) Restructuring –0.172 0.785 –2.024** –1.781** –1.815** –4.319*** –3.662*** –7.362*** –6.930*** –6.614*** 1.997*** 2.509*** –0.177 –0.140 –0.173 (0.643) (0.642) (0.944) (0.903) (0.924) (1.313) (1.305) (2.674) (2.382) (2.415) (0.745) (0.748) (0.919) (0.872) (0.924) Competition –2.117** –0.991 1.853 2.541** 2.279* 4.134* 5.421** 5.977 9.049** 7.970* –0.791 –0.851 0.981 1.094 0.756 (0.866) (0.824) (1.325) (1.268) (1.350) (2.179) (2.125) (4.311) (3.867) (4.220) (0.960) (0.935) (0.915) (0.892) (0.912) Privatization –6.106*** –4.850*** 1.699 0.911 0.925 –2.764 –0.735 10.69 13.29*** 7.471 –7.283*** –6.929*** –2.885 –4.197* –2.600 (0.976) (0.935) (2.881) (2.400) (2.925) (1.903) (1.895) (6.861) (4.975) (6.552) (1.119) (1.055) (2.386) (2.292) (2.423) Observations 1,358 1,350 1,358 1,350 1,350 635 627 635 627 627 720 720 720 720 720 R-squared 0.030 0.094 0.557 0.551 0.555 0.018 0.046 0.502 0.504 0.516 0.043 0.110 0.593 0.581 0.593 Year fixed No No Yes No Yes No No Yes No Yes No No Yes No Yes effect Country fixed No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes effect Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. 311 312 TABLE 9B.2 Impact of reforms on generation efficiency (intermediate outcome), 1995–2015 Complete sample (87) Low-income countries (48) Middle-income countries (39) Dep. var.: generation efficiency (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation 1.440*** 1.462*** 1.367*** 2.259*** 1.559*** 1.007 0.164 1.423** 1.846*** 1.736** 2.330*** 2.042*** 1.890*** 2.966*** 1.857*** (0.381) (0.378) (0.455) (0.426) (0.473) (0.700) (0.696) (0.711) (0.580) (0.732) (0.389) (0.390) (0.678) (0.706) (0.676) Restructuring –1.035** –1.820*** 0.336 0.343 0.249 –2.162** –2.396** –0.133 –0.308 –0.231 –1.583*** –2.247*** 0.163 0.520 0.271 (0.405) (0.400) (0.629) (0.624) (0.630) (1.024) (1.062) (1.218) (1.197) (1.216) (0.407) (0.400) (0.854) (0.858) (0.863) Competition 1.930*** 1.318** 1.208 1.572 1.180 3.870** 1.977 4.201* 4.443** 4.109* 0.147 0.487 –0.355 0.173 –0.357 (0.576) (0.530) (0.986) (1.000) (1.013) (1.605) (1.742) (2.273) (2.217) (2.330) (0.542) (0.497) (1.055) (1.087) (1.030) Privatization 2.001*** 1.424** 0.0372 2.592** 0.639 –0.0222 –0.545 –2.142 –0.767 –0.787 2.393*** 2.369*** 1.375 5.231*** 1.244 (0.640) (0.590) (1.471) (1.269) (1.502) (1.094) (1.097) (2.856) (2.129) (2.920) (0.683) (0.628) (1.724) (1.647) (1.716) Observations 1,348 1,339 1,348 1,339 1,339 622 613 622 613 613 722 722 722 722 722 R–squared 0.038 0.128 0.678 0.672 0.679 0.021 0.057 0.613 0.608 0.619 0.059 0.199 0.764 0.741 0.765 Year fixed effect No No Yes No Yes No No Yes No Yes No No Yes No Yes Country fixed effect No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. TABLE 9B.3 Impact of reforms on modern renewable share in TFEC (final outcome), 1995–2015 Dep. var.: Complete sample (87) Low-income countries (48) Middle-income countries (39) renewables in TFEC (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation –0.854*** –0.976*** 0.461*** 0.570*** 0.424*** –1.106** –1.520*** 0.417*** 0.618*** 0.345** –0.731* –0.599 0.437*** 0.459*** 0.457*** (0.330) (0.341) (0.109) (0.101) (0.117) (0.543) (0.577) (0.151) (0.150) (0.154) (0.422) (0.426) (0.160) (0.143) (0.171) Restructuring 2.798*** 2.853*** –0.289 –0.303 –0.290 –1.597** –1.631** 0.473 0.353 0.475 5.051*** 5.143*** –0.556 –0.545 –0.575 (0.501) (0.508) (0.264) (0.266) (0.265) (0.786) (0.773) (0.371) (0.380) (0.364) (0.643) (0.662) (0.369) (0.382) (0.377) Competition 1.281** 1.181** –0.194 –0.185 –0.222 1.518 0.811 –2.012*** –1.863*** –2.140*** 1.903*** 1.800*** 0.613* 0.539* 0.582* (0.508) (0.506) (0.312) (0.302) (0.316) (1.135) (1.078) (0.640) (0.664) (0.665) (0.545) (0.535) (0.336) (0.305) (0.321) Privatization 0.0605 0.313 1.487*** 1.972*** 1.390 1.851** 2.678*** 0.799 1.650 0.518 –1.287* –1.306* 1.733*** 1.899*** 1.770*** (0.484) (0.506) (0.557) (0.599) (0.553) (0.748) (0.855) (1.147) (1.163) (1.140) (0.668) (0.669) (0.421) (0.350) (0.433) Observations 1760 1339 1760 1705 1705 960 922 960 922 922 780 780 780 780 780 R–squared 0.035 0.033 0.974 0.974 0.974 0.006 0.010 0.975 0.975 0.975 0.195 0.199 0.973 0.973 0.973 Year fixed effect No No Yes No Yes No No Yes No Yes No No Yes No Yes Country fixed No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes effect Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: TFEC = total final energy consumption. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. 313 314 TABLE 9B.4 Impact of reforms on carbon intensity (final outcome), 1995–2015 Complete sample (87) Low-income countries (48) Middle-income countries (39) Dep. var.: carbon intensity (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation 29.21* 34.90** –2.639 –17.88* –7.985 61.18* 55.09* 12.96 –25.95* 0.497 24.34 18.78 –10.76 –12.40 –11.17 (17.32) (17.46) (11.12) (10.14) (11.29) (32.00) (32.49) (19.89) (15.71) (19.69) (18.22) (19.09) (11.16) (11.13) (11.23) Restructuring –126.7*** –133.0*** 2.849 3.164 2.023 –145.9*** –139.8*** 3.545 17.78 1.628 –147.8*** –150.6*** –5.099 –4.772 –2.675 (18.23) (18.45) (10.45) (10.37) (10.35) (38.64) (40.31) (25.35) (24.24) (23.77) (22.18) (22.36) (10.88) (10.46) (10.82) Competition 57.58** 40.71* –25.15 –25.33 –22.34 126.6* –3.614 69.11 36.28 69.56 –6.777 –2.586 –59.15*** –54.66*** –58.99*** (23.84) (23.48) (18.74) (18.64) (19.07) (65.94) (67.76) (51.42) (49.99) (51.28) (23.96) (24.14) (15.79) (16.58) (16.77) Privatization –145.2*** –151.1*** 44.61 –16.15 28.29 –277.9*** –285.3*** 120.3* –54.62 90.46 –89.30*** –87.72** 1.479 –9.267 –2.276 (27.36) (27.72) (35.06) (32.09) (35.08) (53.50) (54.93) (63.92) (54.80) (61.08) (34.31) (34.24) (42.97) (38.95) (42.20) Observations 1364 1355 1364 1355 1355 625 616 625 616 616 735 735 735 735 735 R–squared 0.054 0.066 0.897 0.897 0.899 0.047 0.090 0.885 0.884 0.891 0.135 0.139 0.922 0.921 0.923 Year fixed effect No No Yes No Yes No No Yes No Yes No No Yes No Yes Country fixed effect No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. TABLE 9B.5 Impact of reforms on normalized capacity (final outcome), 1995–2015 Dep. var.: Complete sample (87) Low-income countries (48) Middle-income countries (39) normalized capacity (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation –66.39*** –76.24*** –10.30 7.061** –1.038 17.26 –29.66*** 4.167 –9.298*** –8.028** –93.34*** –76.39*** 3.421 47.30*** 41.43*** (17.57) (12.87) (6.853) (3.567) (4.839) (10.57) (10.30) (6.321) (2.994) (3.831) (32.06) (22.40) (12.14) (7.742) (8.877) Restructuring 262.1*** 171.0*** 10.14 –4.942 –7.099 112.9** 112.1** 7.757 18.11 18.55* 287.9*** 221.7*** –20.82 –27.34** –37.06*** (25.91) (21.09) (11.43) (8.770) (8.965) (50.28) (43.97) (20.92) (11.13) (11.02) (30.63) (21.42) (15.73) (11.46) (11.51) Competition 180.0*** 53.98** –6.192 –26.78* –38.58*** 103.9*** –7.542 –18.01 –57.58*** –59.37*** –2.383 –42.34 –8.594 –61.58*** –72.03*** (32.83) (23.88) (19.55) (13.68) (13.87) (40.17) (35.48) (45.80) (21.32) (21.40) (42.96) (26.97) (20.74) (17.73) (17.40) Privatization 43.53 –15.50 61.14** 104.5*** 73.89*** 128.4*** 80.11*** 134.0*** 83.59** 87.59** –92.78** –100.3*** –8.357 86.70*** 50.50 (30.24) (22.44) (30.43) (22.97) (24.43) (25.13) (22.57) (46.49) (32.47) (34.90) (43.45) (26.96) (42.47) (28.91) (31.49) Observations 1831 1792 1831 1792 1792 1008 969 1008 969 969 819 819 819 819 819 R–squared 0.176 0.548 0.962 0.976 0.977 0.145 0.344 0.928 0.970 0.971 0.090 0.581 0.958 0.975 0.977 Year fixed No No Yes No Yes No No Yes No Yes No No Yes No Yes effect Country fixed No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes effect Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. 315 316 TABLE 9B.6 Impact of reforms on access (final outcome), 1995–2015 Complete sample (87) Low-income countries (48) Middle-income countries (39) Dep. var.: access (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) Regulation 1.491 –3.871*** –1.775*** 5.268*** –1.313** 1.016 –5.077*** –2.369*** 4.384*** –1.569*** –2.099 –13.67*** 1.349 6.624*** 1.286 (1.430) (1.153) (0.529) (0.469) (0.518) (1.393) (1.183) (0.642) (0.521) (0.608) (3.231) (2.227) (0.897) (1.191) (0.851) Restructuring –12.32*** –11.69*** –1.698 –5.413*** –2.198** –22.09*** –19.30*** –1.891 –5.817** –2.614 7.376* 11.40*** –6.127*** –8.169*** –5.486*** (2.696) (2.167) (1.063) (1.413) (1.081) (3.097) (2.875) (1.714) (2.253) (1.772) (3.842) (2.868) (1.452) (1.420) (1.481) Competition 68.46*** 48.58*** –2.650 6.289*** –1.924 72.90*** 54.98*** –2.984 4.126 –1.846 37.51*** 32.26*** 1.110 11.76*** –0.187 (2.744) (2.308) (1.692) (1.909) (1.663) (3.464) (3.557) (2.509) (3.015) (2.499) (4.361) (3.270) (2.043) (1.950) (2.146) Privatization –5.724** –2.868 19.25*** 41.94*** 19.74*** 6.107** 1.626 29.66*** 50.58*** 31.05*** –27.48*** –12.87*** 0.611 20.79*** –2.384 (2.658) (2.718) (3.263) (4.383) (3.306) (2.668) (2.641) (5.152) (6.694) (5.377) (3.968) (4.262) (3.138) (3.291) (3.421) Observations 1314 1267 1314 1267 1267 924 885 924 885 885 378 378 378 378 378 R-squared 0.361 0.599 0.973 0.961 0.975 0.358 0.558 0.958 0.948 0.963 0.277 0.549 0.978 0.955 0.980 Year fixed No No Yes No Yes No No Yes No Yes No No Yes No Yes effect Country No No Yes Yes Yes No No Yes Yes Yes No No Yes Yes Yes fixed effect Controls No Yes No Yes Yes No Yes No Yes Yes No Yes No Yes Yes Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent, Dep. var. = dependent variable. TABLE 9B.7 Correlation between reform steps and final outcomes Final outcomes Environmental Intermediate outcomes Security of supply Social inclusion sustainability Operating Full capital Modern cost cost System Generation Normalized Capacity Capacity Carbon RE in Reforms recovery recovery losses efficiency Reliability capacity diversification growth Electrification Affordability intensity TFEC Total reform n.a. n.a. –0.19 0.16 –0.33 0.31 n.a. n.a. 0.41 –0.13 –0.15 0.13 Regulation n.a. n.a. 0 0.31 –0.19 –0.05 n.a. n.a. –0.05 0 0 –0.06 Restructuring n.a. n.a. –0.20 –0.05 –0.25 0.41 n.a. n.a. 0.46 –0.14 –0.21 0.18 Competition n.a. n.a. –0.20 0.05 –0.32 0.38 n.a. n.a. 0.58 –0.16 –0.09 0.18 Privatization n.a. n.a. –0.18 0.13 –0.25 0.20 n.a. n.a. 0.25 –0.10 –0.16 0.11 Note: Correlations with R squared higher than 0.1 are highlighted. RE = renewable energy; TFEC = total final energy consumption; n.a. = not applicable. 317 318 ANNEX 9C. CROSS-SECTIONAL REGRESSION ANALYSIS ON THE IMPACT OF POWER SECTOR REFORM BASED ON SMALL SAMPLE (17 ECONOMIES) TABLE 9C.1 Correlation between reform steps and outcomes Final outcomes Environmental Intermediate outcomes Security of supply Social inclusion sustainability Full Operating capital Modern cost cost Distribution Generation Normalized Capacity Capacity Carbon RE in Reforms recovery recovery efficiency efficiency Reliability capacity diversification growth Electrification Affordability intensity TFEC Restructuring index −0.10 0.18 −0.04 0.21 0.26 0.11 −0.53 0.04 0.29 −0.31 0.24 −0.27 Regulation index −0.12 −0.18 0.23 0.57 0.16 −0.25 −0.49 −0.11 −0.14 0.21 0.42 −0.64 Perceived regulation index 0.42 0.35 −0.29 0.32 −0.27 −0.06 −0.16 −0.33 −0.29 0.47 0.01 −0.35 Competition index −0.17 0.14 −0.20 0.43 0.27 0.20 −0.39 0.18 0.50 −0.20 0.51 −0.36 Private sector participation index 0.47 0.70 −0.57 0.05 −0.17 0.04 −0.40 −0.03 0.12 −0.01 −0.17 −0.04 Utility governance index 0.40 0.45 −0.33 −0.23 −0.39 −0.11 −0.27 −0.21 −0.45 0.23 −0.30 −0.17 Regulatory framework De jure governance −0.11 −0.24 0.30 0.51 0.23 −0.20 −0.37 −0.07 −0.13 0.15 0.45 −0.60  De jure accountability −0.19 −0.15 0.07 −0.04 0.22 −0.18 −0.40 0.00 −0.26 0.19 0.37 −0.40  De jure autonomy −0.04 −0.18 0.30 0.64 0.13 −0.18 −0.30 −0.11 −0.08 0.13 0.39 −0.57 De jure substance −0.12 0.12 −0.02 0.39 0.01 −0.22 −0.47 0.06 −0.20 0.24 0.07 −0.47  De jure tariff regulation 0.19 0.32 −0.21 0.40 −0.09 −0.19 −0.44 0.01 −0.33 0.29 −0.04 −0.51  De jure quality regulation 0.17 0.11 0.01 0.39 0.10 0.22 −0.34 0.09 0.34 −0.18 0.05 −0.22  De jure market entry regulation −0.51 −0.19 0.16 −0.02 −0.02 −0.52 −0.16 0.00 −0.46 0.44 0.12 −0.23 De facto governance 0.10 −0.04 0.06 0.35 0.17 −0.21 −0.16 −0.09 −0.34 0.27 0.48 −0.55  De facto accountability 0.07 0.01 −0.03 0.30 0.13 −0.28 −0.32 −0.14 −0.47 0.34 0.48 −0.64  De facto autonomy 0.05 −0.09 0.21 0.35 0.15 −0.13 −0.04 −0.04 −0.16 0.27 0.39 −0.43 De facto substance 0.59 0.65 −0.55 0.16 −0.53 0.10 −0.16 −0.29 −0.12 0.32 −0.43 −0.06  De facto tariff regulation 0.60 0.47 −0.37 0.25 −0.33 −0.04 −0.16 −0.16 −0.26 0.26 −0.26 −0.24  De facto quality regulation 0.56 0.66 −0.58 0.14 −0.51 0.38 −0.25 −0.32 0.29 0.05 −0.54 0.14  De facto market entry regulation 0.04 0.18 −0.19 −0.12 −0.28 −0.31 0.20 −0.10 −0.51 0.54 −0.03 −0.10 Utility governance reforms Utility corporate governance 0.29 0.34 −0.13 −0.20 −0.19 −0.13 −0.16 −0.10 −0.43 0.29 −0.17 −0.20  Accountability 0.29 0.36 −0.19 −0.02 −0.11 −0.30 −0.37 −0.03 −0.53 0.27 −0.09 −0.37  Autonomy 0.24 0.26 −0.04 −0.34 −0.23 0.06 0.06 −0.15 −0.27 0.28 −0.21 0.00 Utility management 0.49 0.50 −0.66 −0.19 −0.70 0.00 −0.42 −0.36 −0.27 −0.04 −0.47 −0.01  Financial discipline 0.64 0.37 −0.54 −0.32 −0.55 −0.06 0.02 −0.10 −0.38 −0.01 −0.50 0.14  Human resources 0.71 0.65 −0.40 0.07 −0.43 0.01 −0.26 −0.32 −0.09 0.24 −0.40 −0.02  Information and technology −0.23 0.05 −0.39 −0.14 −0.48 0.05 −0.51 −0.29 −0.10 −0.25 −0.08 −0.12 Note: Correlations with R squared higher than 0.1 are highlighted. RE = renewable energy; TFEC = total final energy consumption. Did Power Sector Reform Deliver Better Sector Outcomes? 319 TABLE 9C.2 Regression analysis: reforms and intermediate outcomes Operating cost Economic cost Distribution Thermal recovery recovery efficiency efficiency A Restructuring –0.055 0.282 0.036 1.879 GDP 0 0 0 0.002 System size 0 0 0 0.004 R squared 0.232 0.369 0.119 0.182 B Private sector participation 0.450 0.77*** 0.282** –17.198 GDP 0 0 0 0.003* System size 0 0 0 0.004 R squared 0.332 0.608 0.370 0.234 C Perceived regulation 0.413 0.290 0.110 15.985 GDP 0 0 0 0.002 System size 0 0 0 0.005 R squared 0.323 0.324 0.144 0.236 D Utility governance 0.417 0.468 0.154 –22.242 GDP 0 0 0 0.002 System size 0 0 0 0.002 R squared 0.293 0.368 0.170 0.253 E Competition –0.911 1.703 0.750 56.395 GDP 0 0 0 0.001 System size 0 –0.0004** –0.0001* –0.001 R squared 0.261 0.395 0.248 0.230 Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. TABLE 9C.3 Regression analysis: utility governance reforms and intermediate outcomes Operating Cost Economic cost Distribution Thermal recovery recovery efficiency efficiency A Corporate governance Autonomy 0.098 0.113 –0.016 –16.252 GDP 0 0 0 0.002* System size 0 0 0 0.002 R squared 0.241 0.293 0.113 0.319 B Corporate governance Accountability 0.207 0.256 0.056 –4.155 GDP 0 0 0 0.002 System size 0 –0.0002* 0 0.004 R squared 0.281 0.375 0.138 0.189 C Managerial practices Financial discipline 0.995** 0.473 0.352** –23.366 GDP 0 0 0 0.002 System size 0 0 0 0.002 R squared 0.566 0.358 0.381 0.252 (Table continued next page) 320 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD TABLE 9C.3 Regression analysis: utility governance reforms and intermediate outcomes (Continued) Operating Cost Economic cost Distribution Thermal recovery recovery efficiency efficiency D Managerial practices Human Resources 1.182** 0.857* 0.217 2.891 GDP 0 0 0 0.002 System size 0 0 0 0.005 R squared 0.509 0.448 0.176 0.181 E Managerial practices Information and Technology –0.398 –0.012 0.206 –4.291 GDP 0 0 0 0.002 System size –0.0002* 0 0 0.004 R squared 0.313 0.276 0.254 0.184 Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. TABLE 9C.4 Regression analysis: regulatory reforms and intermediate outcomes Operating cost Economic cost Distribution Thermal recovery recovery efficiency efficiency A De jure regulation Governance –0.045 –0.261 –0.172 29.338 GDP 0 0 0 0.002 System size 0 0 0 0 R squared 0.230 0.308 0.192 0.347 B De jure regulation Quality regulation 0.384 0.224 0 11.832 GDP 0 0 0 0.002 System size –0.0002* –0.0002* 0 0.001 R squared 0.288 0.309 0.111 0.216 C De jure regulation Tariff regulation 0.109 0.286 0.074 16.687 GDP 0 0 0 0.002 System size 0 –0.0002* 0 0.003 R squared 0.234 0.347 0.139 0.265 D Perceived regulation Governance 0.265 0.095 –0.001 17.205 GDP 0 0 0 0.002 System size –0.0002* 0 0 0.001 R squared 0.288 0.282 0.098 0.276 E Perceived regulation Quality regulation 0.334 0.389* 0.181* –1.003 GDP 0 0 0 0.002 System size 0 0 0 0.004 R squared 0.333 0.458 0.343 0.181 F Perceived regulation Tariff regulation 0.466** 0.296 0.102 9.815 GDP 0 0 0 0.002 System size 0 0 0 0.005 R squared 0.449 0.403 0.200 0.232 Note: Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. TABLE 9C.5 Regression analysis: reform steps and final outcomes Security of supply Social inclusion Environmental sustainability Normalized Capacity Capacity Carbon Modern RE share SAIFI capacity diversification growth Access Affordability intensity in TFEC A Utility restructuring 16.00 29.10 –0.4** 0.02 3.97 –0.05 37.47 –13.14 GDP per capita 0 0.04 0 0 0.01** 0 –0.01 0 System size (GWh) 0.22*** 0.01 0 0 0.02 0 0.36** 0 B Private sector participation –7.04 –238.93 –0.48 0 –32.73 –0.02 –5.34 –6.70 GDP per capita 0 0.05 0 0 0.01** 0 0 0 System size (GWh) 0.22*** 0.01 0 0 0.02 0 0.37** –0.01 C Regulation 78.83 –792.00 –0.68* –0.01 –85.63* 0.15 499.73 –62.61** GDP per capita 0 0.06 0 0 0.01*** 0 –0.02 0 System size (GWh) 0.21*** 0.09 0 0 0.028* 0 0.34*** 0 D Perceived regulation –58.29 –211.24 –0.14 –0.04 –66.34* 0.19 228.73 –32.52 GDP per capita 0 0.04 0 0 0.01*** 0 –0.01 0 System size (GWh) 0.21*** 0 0 0 0.01 0 0.41*** –0.01 E Utility governance –54.671 –275.838 –0.358 –0.074 –93.31** 0.061 –159.861 –23.908 GDP per capita 0 0.040 0 0 0.009*** 0 –0.004 0 System size (GWh) 0.21*** –0.013 0 0 0.007 0 0.35** –0.009 F Competition –83.68 439.86 –2.24* 0.05 82.05 –0.05 2108.25* –125.51 GDP per capita 0 0.03 0 0 0.01 0 –0.05 0 System size (GWh) 0.23** –0.02 0 0 0.01 0 0.16 0.01 Note: GWh = gigawatt-hour; RE = renewable energy; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. 321 322 TABLE 9C.6 Regression analysis: regulatory reforms and final outcomes Security of supply Social inclusion Environmental sustainability Normalized Capacity Capacity Carbon Modern RE SAIFI capacity diversification growth Access Affordability intensity share in TFEC De jure regulation Regulatory governance 56.22 –527.73 –0.40 –0.02 –72.11* 0.12 366.08 –48.8** GDP 0 0.05 0 0 0.01*** 0 –0.01 0 System size 0.21*** 0.09 0 0 0.03* 0 0.34** 0 De jure regulation Quality regulation –52.352 178.239 –0.388 –0.023 –5.547 –0.019 –221.426 –13.392 GDP 0.003 0.029 0 0 0.009** 0 0.009 0.001 System size 0.23*** –0.024 0 0 0.019 0 0.42** –0.003 De jure regulation Tariff regulation 1.033 –385.430 –0.410 0.049 –69.89** 0.093 –38.471 –34.87** GDP 0 0.049 0 0 0.01*** 0 –0.004 0.001 System size 0.22*** 0.023 0 0 0.018 0 0.37** –0.006 Perceived regulation Regulatory governance –5.82 –316.64 –0.13 –0.04 –65.64** 0.16* 245.24 –32.90** GDP 0 0.04 0 0 0.008*** 0 –0.01 0 System size 0.22*** 0.07 0 0 0.03** –0.00* 0.35** 0 Perceived regulation Quality regulation –100.913 413.576 –0.165 –0.016 15.531 –0.032 –222.007 4.255 GDP 0.007 0.012 0 0 0.008* 0 0.010 –0.001 System size 0.19*** 0.148 0 0 0.023 0 0.30** –0.005 Perceived regulation Tariff regulation –60.233 –76.345 –0.099 0.004 –30.628 0.048 –105.131 –15.044 GDP –0.001 0.041 0 0 0.009** 0 –0.004 0 System size 0.21*** 0.008 0 0 0.013 0 0.36** –0.008 Note: RE = renewable energy; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. TABLE 9C.7 Regression analysis: utility governance reforms and final outcomes Security of supply Social inclusion Environmental sustainability Normalized Capacity Capacity Carbon Modern RE SAIFI capacity diversification growth Access Affordability intensity share in TFEC Corporate governance Autonomy 5.656 46.427 0.070 –0.083** –31.263 0.046 –39.424 –2.797 GDP 0.000 0.040 0.000 0.000 0.009** 0.000 –0.005 0.000 System size 0.22*** 0.026 0 0 0.013 0 0.36** –0.007 Corporate governance Accountability 19.533 –364.391 –0.266 –0.025 –63.67*** 0.062 –70.035 –19.347 GDP 0 0.043 0 0 0.009*** 0 –0.003 0 System size 0.22*** 0.008 0 0 0.016 0 0.37** –0.007 Managerial practices Financial discipline –231.601 –49.851 0.034 –0.011 –57.496 –0.049 –634.023 8.201 GDP –0.003 0.040 0 0 0.008** 0 –0.010 0 System size 0.18** 0.013 0 0 0.011 0 0.30** –0.005 Managerial practices Human Resources –23.392 –224.647 –0.457 –0.065 –41.355 0.033 126.068 –23.719 GDP 0 0.043 0 0 0.009** 0 –0.007 0 System size 0.22** –0.026 0 0 0.009 0 0.39** –0.011 Managerial practices Information and Technology –168.770 175.934 –0.63** 0.095 7.292 –0.111 67.803 –14.652 GDP –0.003 0.043 0 0 0.009** 0 –0.004 0 System size 0.19*** 0.036 0 0.00* 0.018 0 0.38** –0.008 Note: RE = renewable energy; SAIFI = System Average Interruption Frequency Index; TFEC = total final energy consumption. Significance level: *** = 1 percent, ** = 5 percent, * = 10 percent. 323 324 RETHINKING POWER SECTOR REFORM IN THE DEVELOPING WORLD NOTES results could be regarded as more powerful 1. This chapter is based on original research by than positive ones. Vivien Foster, Joern Huenteler, Anshul Rana, 4. Diversification index is calculated such that and Tu Chi Nguyen. results closer to 1 show increasing depen- 2. The methodological approach draws from dence on a single fuel whereas results closer recent statistical literature underpinning to 0 show more diversification. machine learning (Dean 2014; Friedman, Hastie, and Tibshirani 2000; Witten and others REFERENCES 2017). It proceeds by examining all possible Banerjee, S. G., F. A. Moreno, J. E. Sinton, T. bilateral correlations between variables that Primiani, and J. Seong. 2017. “Regulatory arise from the theoretical framework described Indicators for Sustainable Energy: A Global in figure 9.1, reporting those for which the Scorecard for Policy Makers.” World Bank, coefficient of determination (R-squared) is Washington, DC. above 0.06 (equivalent to a correlation coef- Dean, J. 2014. Big Data, Data Mining, and Machine ficient of over 0.25) with the expected sign. Learning: Value Creation for Business Leaders and 3. In particular, where no association of any Practitioners. Hoboken, NJ: John Wiley & Sons. kind is found between reform and outcome Friedman, J., T. Hastie, and R. Tibshirani. 2000. variables, it is unlikely that this association “Additive Logistic Regression: A Statistical would increase as a result of the adoption View of Boosting.” The Annals of Statistics of more sophisticated statistical methods. 28 (2): 337–407. Conversely, where a strong association is Witten, I., E. Frank, M. Hall, and C. Pal. 2017. found, it is possible that the size of this effect Data Mining: Practical Machine Learning Tools and could weaken if more sophisticated statistical Techniques (Fourth Edition). Cambridge, MA: tools were employed. In that sense, negative Morgan Kaufmann. ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental foot- print. In support of this commitment, we leverage electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be low- ered and shipping distances decreased, resulting in reduced paper con- sumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), pro- cessed chlorine–free (PCF), or enhanced elemental chlorine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. Rethinking Power Sector Reform in the Developing World During the 1990s, a new paradigm for power sector reform was put forward emphasizing the restructuring of utilities, the creation of regulators, the participation of the private sector, and the establishment of competitive power markets. Twenty-five years later, only a handful of developing countries have fully implemented these Washington Consensus policies. Across the developing world, reforms were adopted rather selectively, resulting in a hybrid model, in which elements of market orientation coexist with continued state dominance of the sector. This book aims to revisit and refresh thinking on power sector reform approaches for developing countries. The approach relies heavily on evidence from the past, drawing both on broad global trends and deep case material from 15 developing countries. It is also forward looking, considering the implications of new social and environmental policy goals, as well as the emerging technological disruptions. A nuanced picture emerges. Although regulation has been widely adopted, practice often falls well short of theory, and cost recovery remains an elusive goal. The private sector has financed a substantial expansion of generation capacity; yet, its contribution to power distribution has been much more limited, with efficiency levels that can sometimes be matched by well-governed public utilities. Restructuring and liberalization have been beneficial in a handful of larger middle-income nations but have proved too complex for most countries to implement. Based on these findings, the report points to three major policy implications. n First, reform efforts need to be shaped by the political and economic context of the country. The 1990s reform model was most successful in countries that had reached certain minimum conditions of power sector development and offered a supportive political environment. n Second, countries found alternative institutional pathways to achieving good power sector outcomes, making a case for greater pluralism. Among the top performers, some pursued the full set of market- oriented reforms, while others retained a more important role for the state. n Third, reform efforts should be driven and tailored to desired policy outcomes and less preoccupied with following a predetermined process, particularly since the twenty-first-century century agenda has added decarbonization and universal access to power sector outcomes. The Washington Consensus reforms, while supportive of the twenty-first-century century agenda, will not be able to deliver on them alone and will require complementary policy measures. The full report, together with a substantial collection of case studies and background papers, can be accessed at http://www.esmap.org/rethinking_power_sector_reform. ISBN 978-1-4648-1442-6 90000 9 781464 814426 SKU 211442