Time to ACT Realizing Indonesia’s Urban Potential Mark Roberts, Frederico Gil Sander, and Sailesh Tiwari, Editors Time to ACT Time to ACT Realizing Indonesia’s Urban Potential Mark Roberts, Frederico Gil Sander, and Sailesh Tiwari, Editors © 2019 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 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: Roberts, Mark, Frederico Gil Sander, and Sailesh Tiwari, editors. 2019. Time to ACT: Realizing Indonesia’s Urban Potential. Washington, DC: World Bank. doi:10.1596/978-1-4648-1389-4. 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: 978-1-4648-1389-4 ISBN (electronic): 978-1-4648-1400-6 DOI: 10.1596/978-1-4648-1389-4 Cover photo: © Mulya Amri / World Bank. Further permission required for reuse. Cover design: Bill Pragluski, Critical Stages LLC. Library of Congress Cataloging-in-Publication Data has been requested. Contents Forewords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xvii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 By 2045, 220 million Indonesians will live in urban areas . . . . . . . . . . . . . . . . . . . . . . 2 Delivering on the promise of urbanization requires managing congestion forces . . . . . . 4 Policies to realize Indonesia’s urban potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Annex OA Tailored policy options by type of place . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35 Defining prosperity, inclusiveness, and livability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 A framework for assessing urbanization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 A reader’s guide to this report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Part 1 Indonesia’s Urban Trends and Performance 1 Patterns of Urbanization and Structural Transformation . . . . . . . . . . . . . . . . . . . .45 Measuring urbanization in Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Recent urbanization trends in Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Indonesia’s evolving “portfolio of places” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Looking forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Annex 1A BPS’s composite scoring system for identifying urban settlements . . . . . . . . 68 Annex 1B Urban population growth decomposition methodology . . . . . . . . . . . . . . . . 69 v Annex 1C Multidistrict and single-district metro areas in Indonesia, by island-region, sorted by metro area total population. . . . . . . . . . . . . . . . . . . . . . . . . 71 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 2 Is Urbanization Delivering?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 What should urbanization deliver? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Is urbanization in Indonesia delivering prosperity for all? . . . . . . . . . . . . . . . . . . . . . . . 76 Is urbanization in Indonesia delivering livability for all? . . . . . . . . . . . . . . . . . . . . . . . . 89 Urbanization’s potential to deliver more benefits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 3 Drivers of Productivity and Prosperity across the Portfolio of Places. . . . . . . . 103 The cross-district relationship between prosperity and average productivity . . . . . . . . 105 What explains urban productivity? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 The importance of differences in underlying productivity . . . . . . . . . . . . . . . . . . . . . . 107 Underlying productivity highest in multidistrict metro core and single-district metro area districts, followed by urban periphery districts . . . . . . . . . . . . . . . . . . . . . 109 Understanding agglomeration forces and how tailored policies can improve productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Varying effects of the business environment on productivity . . . . . . . . . . . . . . . . . . . . 114 Conclusion and policy implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Annex 3A Construction of the domestic market access variable . . . . . . . . . . . . . . . . . 120 Annex 3B Determinants of underlying productivity. . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Annex 3C Effects of the local business environment on firm productivity . . . . . . . . . . 123 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 4 Drivers of Urban and Spatial Inclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Determinants of between-place inequality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Determinants of within-place inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 Annex 4A Methodology to analyze spatial disparities . . . . . . . . . . . . . . . . . . . . . . . . 153 Annex 4B Returns to aggregate human capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Annex 4C Returns to aggregate human capital by skill level . . . . . . . . . . . . . . . . . . . . 156 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Spotlight 1  Strengthening the Disaster Resilience of Indonesian Cities . . . . . . . 161 How exposed are Indonesian cities to disasters? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 A look ahead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 What are the drivers of urban disaster risk? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 What needs to be done: A holistic approach to urban disaster resilience . . . . . . . . . . . 166 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 Spotlight 2  Urbanization for Human Capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Indonesia’s low score on the human capital index. . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Human capital across places. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 vi  TIME TO ACT Urbanization policy sensitive to human capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Part 2  How Can Urbanization in Indonesia Deliver More? 5 Urban Governance, Institutions, and Finance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Urbanization in the context of Indonesia’s decentralization. . . . . . . . . . . . . . . . . . . . . 180 Why don’t Indonesian cities deliver more and better urban services and infrastructure? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 Opportunities and policy options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 Annex 5A Evolution of decentralization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 6 Infrastructure and Policies to Connect the Portfolio of Places . . . . . . . . . . . . . . 219 Lower aggregate infrastructure investment since the 1997 Asian crisis . . . . . . . . . . . . 221 Problems arising from the spatial targeting of infrastructure investment under decentralization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 Implications for productivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 Barriers to migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 Infrastructure, the decentralization conundrum, and the need for coordination. . . . . . 239 Annex 6A Aggregate infrastructure investment and stock data . . . . . . . . . . . . . . . . . . 241 Annex 6B Econometric methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 Annex 6C Results of the econometric model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 7 Connected and Integrated Cities: A Focus on Housing and Transport . . . . . . . 247 Planning for connected growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 Housing and transport: Key sectors for urban connectivity and integration . . . . . . . . . 251 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271 8 Targeting Places and People Left Behind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 Characterizing lagging places in Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 Place-based policies in Indonesia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 Rethinking place-based policies in Indonesia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Targeting people left behind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 Spotlight 3  The Invisible Crisis of Wastewater Management in Indonesia . . . . . 297 The consequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 The causes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298 What can be done? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 C o n te n ts   vii Spotlight 4  The Potential of Smart Cities. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 Livability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 Inclusiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 Prosperity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 Implications for Indonesia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303 Notes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 Boxes O.1 Indonesia’s portfolio of places. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 O.2 The basic policy principles: Augment, Connect, and Target. . . . . . . . . . . . . . . . . . . . 14 O.3 The three coordination challenges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 I.1 The basic policy principles: Augment, Connect, and Target. . . . . . . . . . . . . . . . . . . . 36 I.2 Consistency of this report with the Sustainable Development Goals and the New Urban Agenda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 I.3 Sources and types of positive agglomeration forces. . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.1 Three of a kind—Indonesia’s three definitions of urban areas . . . . . . . . . . . . . . . . . . 48 1.2 What is the correct measure of the pace of urbanization? . . . . . . . . . . . . . . . . . . . . . 50 1.3 Explaining Indonesia’s urbanization slowdown from rapid to near normal. . . . . . . . 52 1.4 Comparing the contribution of migration to urban population growth across Indonesia, China, and India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 1.5 Methodology for defining Indonesia’s portfolio of places . . . . . . . . . . . . . . . . . . . . . 55 1.6 Is Jakarta too big?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 2.1 Economic profile of Indonesia’s most prosperous metro areas. . . . . . . . . . . . . . . . . . 80 2.2 Mixed health and education outcomes for slum dwellers compared to non-slum dwellers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 3.1 Data and methodology for estimating underlying productivity . . . . . . . . . . . . . . . . 108 3.2 Sources and types of agglomeration forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.1 Chronic poverty: A profile of those left behind . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.2 The various types of migrants in Indonesia: Are they all captured in the data? . . . . 139 4.3 Technology, the changing nature of work, and urbanization . . . . . . . . . . . . . . . . . . 146 4.4 Measuring residential segregation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 S1.1 Floods of many types: The example of Mataram City. . . . . . . . . . . . . . . . . . . . . . . 163 S1.2 An inundated capital: The 2007 floods in Jakarta. . . . . . . . . . . . . . . . . . . . . . . . . . 165 S1.3 Risk-informed spatial planning in Indonesia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 S1.4 Disaster information crowdsourcing: PetaBencana.id . . . . . . . . . . . . . . . . . . . . . . . 169 5.1 Cross-jurisdictional coordination in the Jakarta metropolitan area. . . . . . . . . . . . . 186 5.2 Fragmentation among Indonesia’s planning systems . . . . . . . . . . . . . . . . . . . . . . . . 188 5.3 Regulatory complexity in conducting public-private partnerships. . . . . . . . . . . . . . 198 5.4 Some models of metropolitan management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 5.5 Improving reporting and accountability through the Local Government and Decentralization project. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 6.1 Road quality and structural transformation in Indonesia . . . . . . . . . . . . . . . . . . . . 222 viii  TIME TO ACT 6.2 Attracting private infrastructure investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 6.3 Literature analyzing the impact of decentralization. . . . . . . . . . . . . . . . . . . . . . . . . 232 7.1 Two key planning systems for Indonesian cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 7.2 Government of Indonesia housing policy interventions. . . . . . . . . . . . . . . . . . . . . . 253 7.3 Affordable housing land suitability perspective in Semarang. . . . . . . . . . . . . . . . . . 256 7.4 Land tenure: Streamlining property rights with blockchain technology. . . . . . . . . . 258 7.5 Key principles for mixed-use, mixed-income developments. . . . . . . . . . . . . . . . . . . 259 7.6 Rental housing voucher programs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 7.7 Public transport deficiencies and TransJakarta BRT . . . . . . . . . . . . . . . . . . . . . . . . 263 7.8 Mobility apps and urban transport. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 7.9 High-occupancy vehicle policy in Jakarta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 7.10 Tanjung Barat: Leveraging partnerships for transit-oriented development in Jakarta. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 8.1 Batam and the Singapore–Johor–Riau growth triangle . . . . . . . . . . . . . . . . . . . . . . 280 8.2 Assessment of the impact of the Integrated Economic Development Zone program. . . 282 8.3 Special economic zones throughout the world. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 8.4 Indonesia’s transmigration program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 8.5 The potential of ecotourism in Indonesia’s lagging districts. . . . . . . . . . . . . . . . . . . 289 8.6 The mobility challenge for people with disabilities . . . . . . . . . . . . . . . . . . . . . . . . . 292 8.7 Inclusion along Samoa’s Apia waterfront . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 S4.1 Indonesia 100 Smart City Movement and the Association of Southeast Asian Nations Smart Cities Network . . . . . . . . . . . . . . . . . . . . . . . . . . 304 Figures O.1 Indonesia’s post-2000 pace of urbanization is typical by international standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 BO.1.1 Stylized depiction of the different types of place within the “portfolio of places” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 O.2 Urbanization outcomes are jointly determined by the interplay of agglomeration and congestion forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 O.3 Workers in more urban areas earn higher wages than comparable rural workers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 O.4 Better economic opportunities in urban areas have led to less poverty and vulnerability to poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 O.5 Urbanization has delivered less prosperity in Indonesia than in the rest of the region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 O.6 Inequality has increased everywhere and remains highest in multidistrict metro cores. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 O.7 Learning gaps between the top and the bottom socioeconomic groups are largest for cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 O.8 One-fifth of Indonesia’s urban population lives in slums, and overcrowded housing is commonplace in metro cores and single-district metro areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 C o n te n ts   ix O.9 Indonesia’s cities are among the most congested in the East Asia and Pacific region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 O.10 Unsafe to breathe: Most Indonesian metro areas have unacceptable air pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 O.11 Despite convergence, significant urban–rural gaps remain in access to basic services. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.12 Indonesia’s road and railway networks are not as extensive as in many other countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 O.13 Indonesia has lower property tax revenue than most other G20 countries . . . . . 17 BO.3.1 Districts account for about half the public spending in education, health, and infrastructure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 I.1 A framework for assessing urbanization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 1.1 Indonesia is at an intermediate level of urbanization, 2017 . . . . . . . . . . . . . . . . . 49 1.2 Almost 70 percent of the national urban population lives on Jawa-Bali, which is the only island-region with more than half of its population in urban areas, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 1.3 Indonesia’s pace of urbanization accelerated through the turn of the century and then slowed, 1950–2017. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 1.4 Rising and then falling—benchmarking Indonesia’s pace of urbanization against global experience, 1950–2015. . . . . . . . . . . . . . . . . . . . . . . 51 1.5 The pace of urbanization slowed across all Indonesian island-regions between 2004–10 and 2010–16. . . . . . . . . . . . . . . . . . . . . . . . . . . 52 1.6 In Indonesia, and especially in Jawa-Bali, natural population growth and reclassification from rural to urban contributed more to urban population growth than did rural–urban migration, overall and by island-region, 2000–10. . . . 53 B1.4.1 Decomposition of urban population growth for Indonesia, India, and China. . . . 53 1.7 Growth in Indonesia’s urban population peaked in 1990–2000 but remains high. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 B1.5.1 Stylized depiction of the different types of place within the “portfolio of places”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 B1.6.1 Urban primacy levels for 156 countries globally and urban primacy levels for the world’s 10 most populous countries, 2012. . . . . . . . . . . . . 59 1.8 Collectively, nonmetro rural areas account for a large share of the population, including the urban population, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 1.9 Sectoral structure of employment by type of place, 2017 . . . . . . . . . . . . . . . . . . . 62 1.10 Changes in employment composition by sector and type of place, 2007–17. . . . . 62 1.11 Average annual growth rates of urban population shares and of urban population, by type of place, 2004–16. . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 1.12 Relative contributions of natural population growth, migration, and reclassification to urban population growth, by type of place, 2000–10 . . . . 64 1.13 Indonesia’s projected path of urbanization, up to 2045 . . . . . . . . . . . . . . . . . . . . 67 1.14 Projected absolute increase in Indonesia’s urban population, 2016–45 . . . . . 67 2.1 More urbanized countries tend to have higher levels of income and lower levels of poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 x  TIME TO ACT 2.2 More urbanized districts are richer, especially when mining-intensive districts are removed, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 2.3 Urban areas contribute nearly 60 percent of Indonesia’s GDP . . . . . . . . . . . . . . . 78 2.4 Economic prosperity is concentrated mostly in metro cores, especially in Jakarta. . . 79 2.5 The higher the level of urbanization, the larger are industry and services sectors in the economy, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 2.6 High-end services account for a larger share of output in metro cores and single-district metros than in other places, 2016. . . . . . . . . . . . . 79 B2.1.1 Share of nominal GDP, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 2.7 Urban areas employ a higher share of workers in industry and services, but jobs tend to be in lower-value-added services, 2015. . . . . . . . . . . . . 81 2.8 Urban areas tend to offer more opportunities for formal employment and better-paying jobs than rural areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 2.9 Faster urbanization was accompanied by more prosperity in Indonesia, but not as much as in China and other developing countries in the East Asia and Pacific region, 1996–2016. . . . . . . . . . . . . . . . . . . 82 2.10 Districts that urbanized faster have not grown faster, 1996–2016 . . . . . . . . . . . . 83 2.11 Per capita incomes have grown fastest in the metro cores and single-district metro areas, whereas other types of urban districts grew more slowly than rural districts, 1996–2016 . . . . . . . . . . . . . . . . . . . . . . . . 84 2.12 Indonesian districts that were poorer in 1996 grew faster over 1996–2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 2.13 Urbanization is associated with increases in the share of workers employed in industry and services across districts, but not with changes in the share of output from industry and services, 2007–15 . . . . . . . . . . 85 2.14 Poverty and vulnerability are lowest in metro cores and single-district metros, even when the number of people who are poor or vulnerable are counted in absolute terms, 2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 2.15 The urbanization process is associated with poverty reduction in Indonesia, but not as much as in China and Vietnam. . . . . . . . . . . . . . . . . . . . . . 86 2.16 Poverty and vulnerability decreased in all types of place, with sharper reductions in metro peripheries and nonmetro rural areas . . . . . . . . . . . . . . . . . . 87 2.17 Middle-class population shares are the largest in metro areas, 2017 . . . . . . . . . . . 87 2.18 Urban inequality is high in Indonesia compared to regional peers . . . . . . . . . . . . 88 2.19 Inequality has risen in all places, especially in rural peripheries, 2001–17 . . . . . . 88 2.20 Overall inequality is driven by within-place inequality. . . . . . . . . . . . . . . . . . . . . 89 2.21 Urban households have greater access to health facilities and to safe drinking water and sanitation, 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.22 Rural households are particularly deprived of access to secondary education, 2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 2.23 Gaps in services remain between urban and rural places . . . . . . . . . . . . . . . . . 91 2.24 Disparity in access to safe sanitation and to safe drinking water remains high between urban and rural areas, 2002 and 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . 91 C o n te n ts   xi 2.25 Indonesia trails several emerging market economies in providing basic sanitation services and basic drinking water services, 2015 . . . . . . . . . . . . . . . . . 92 2.26 Urban households are less deprived than rural households on a wide range of health and education outcomes, 2016 . . . . . . . . . . . . . . . . . . . . . . 93 2.27 Ratios of house price to income are higher in Jakarta than in New York, contributing to substantial overcrowding in metro cores and single-district metro areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 2.28 One-fifth of Indonesia’s urban population lives in slums, mostly in urban peripheries and nonmetro urban areas, 2015 . . . . . . . . . . . . . . . 95 B2.2.1 Slum dwellers have poorer access to doctors and preschools, 2014 . . . . . . . . . . . 96 2.29 Large shares of commuters in Jakarta metro and single-district metro areas spend more than an hour to get to work, partly because of long distances between home and work, 2015 . . . . . . . . . . . . . 97 2.30 Indonesia’s cities are among the most congested in the East Asia and Pacific region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 2.31 It takes far more time to travel the same distance in Indonesia than in other East Asia and Pacific countries . . . . . . . . . . . . . . . . . . . . 98 2.32 Only eight metro areas in Indonesia have acceptable levels of air pollution, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 2.33 Indonesia’s cities are among the most polluted in the world, 2015. . . . . . . . . . . 100 2.34 Metro cores and single-district metro areas, especially their middle-class households, are more likely to experience crime than other places in Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 3.1 ­ GDP per capita and average productivity are strongly correlated across districts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3.2 ­ Differences in average productivity across districts can be explained by differences in economic composition (“sorting”) or differences in underlying productivity (“place-based advantages”) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 ­ 3.3 Average estimated levels of underlying district productivity, relative to nonmetro rural areas, by type of district . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 ­ .4 3 Dimensions of the local business environment . . . . . . . . . . . . . . . . . . . . . . . . . . 114 ­ 3.5 Variations in the local business environment across Indonesia’s metro areas: Power outages, use of web to conduct business, skilled labor obstacle, and land access obstacle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 ­ 3.6 Some aspects of the business environment improved and some worsened across locations in Indonesia between 2009 and 2015. . . . . . . . . . . . . . . . . . . . 117 ­ 3.7 Estimated effects of the local business environment on firm productivity . . . . . 117 B4.1.1 Where do Indonesia’s chronically poor live? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.1 Even after accounting for cost-of-living differences, welfare differences across space vary in Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 4.2 The returns profile has a stronger correlation with, and explains a larger variation of, the actual welfare distribution than the endowments profile does . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 xii  TIME TO ACT 4.3 Metropolitan areas have lower rates of stunting and better learning outcomes on average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 4.4 Multidistrict metro cores provide a quicker pathway out of poverty and into the middle class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 4.5 Large gaps exist between average growth of per capita consumption and growth for the bottom 40 ­ percent. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 4.6 Highly populated places generally have high inequality . . . . . . . . . . . . . . . . . . . 143 4.7 Average years of schooling and population size are positively correlated across Indonesia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 4.8 High-skilled workers see the largest return to aggregate human capital, followed by medium-skilled workers and low-skilled workers. . . . . . . . . . . . . . 145 4.9 Both within the portfolio of places and within Indonesia’s major metropolitan areas, differences in educational attainment account for as much as one-third of overall within-place inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 4.10 Children from households in the top 20 percent of the socioeconomic distribution outperform those in the bottom 20 percent most in cities on the Programme for International Student Assessment’s science test. . . . 146 S1.1 Indonesia’s metro areas face several different disaster types, 2003–17. . . . . 162 S2.1 Average human capital index scores vary widely across Indonesia, 2017 . . . . . . 175 5.1 Indonesia’s decentralization has gone through three phases . . . . . . . . . . . . . . . . 181 5.2 Indonesia’s public governance has improved since the start of decentralization but remains far from perfect. . . . . . . . . . . . . . . . . . . . . . . . . . . 181 5.3 Subnational government financial report audit scores have improved since 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 5.4 Subnational government expenditures have risen as a share of total government expenditure after decentralization. . . . . . . . . . . . . . . . . . . . . . . . . . 183 5.5 Across the portfolio of places, personnel account for more than half of subnational government spending, 2010–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 5.6 Districts account for about half of public spending in education, health, and infrastructure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 5.7 Steady growth in the number of civil servants across places and island-regions, 2005–11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 5.8 Urban districts and those in Jawa-Bali generated relatively more own-source revenue than other types of place, 2014–16 . . . . . . . . . . . . . . . . . . . 193 5.9 Own-source taxes make up a higher proportion of the total revenue in urban areas and in Jawa-Bali, 2014–16 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 5.10 General purpose funds are important throughout all districts but particularly in rural districts and in Maluku-Papua, Nusa Tenggara, and Sulawesi, 2014–16. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 5.11 More densely populated districts received less transfer revenue per capita than less densely populated districts, 2016 . . . . . . . . . . . . . . . . . . . . . . . 196 5.12 Rural districts had higher proportions of poor people, but urban peripheries had greater numbers of poor people, 2016. . . . . . . . . . . . . . . . . . . . 196 C o n te n ts   xiii 5.13 Urban periphery districts and those in Jawa-Bali received less transfer revenue per capita than other districts, 2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 5.14 There is a “missing middle” of available instruments for subnational governments to finance infrastructure investment. . . . . . . . . . . . . . . . . . . . . . . . 199 5A.1 Subnational governments proliferated after decentralization . . . . . . . . . . . . . . . 212 6.1 Infrastructure investment has fallen as a share of GDP, by source and sector, 1995–2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 6.2 Transport infrastructure investment has rebounded as a percentage of GDP, with a greater share coming from subnational governments, 1995–2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 6.3 Indonesia’s Liner Shipping Connectivity Index performance lags neighbors. . . . 225 6.4 Telecommunications infrastructure investment has fallen as a percentage of GDP, with almost all coming from the private sector and state-owned enterprises, 1995–2014. . . . . . . . . . . . . . . . . . . . . . . . . 228 6.5 Mobile cellular subscription rates are higher in Indonesia than in most comparator countries, 2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 6.6 Internet use remains low in Indonesia despite the high number of secure Internet servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 6.7 Subnational government revenues have grown greatly, 1994–2016 . . . . . . . . . . 230 6.8 Transfers per capita skew toward less populated districts, by current population quartiles, 1994–2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 6.9 Spending per capita skews toward less populated districts, by current population quartiles, 1994–2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 6.10 Infrastructure spending per capita skews toward less populated districts, by current population quartiles, 2001–16. . . . . . . . . . . . . . . . . . . . . . . 231 6.11 Electricity access increased more in high-resource districts, but water and toilet access did not, 1996–2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 6.12 Districts in favored provinces show a differential increase in highways and total stock of roads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 6.13 Higher resources do not differentially increase market access, 1997–2014. . . . . 235 6.14 Electricity coverage, but not water or sanitation coverage, increases in districts in favored provinces, 1996–2016 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 6.15 Additional resources in low-population districts did not improve employment, value added, or productivity, 1990–2015 . . . . . . . . . . . . . . . . . . . 236 6.16 Firms that used more vehicles in high-resource districts did not improve employment, value added, or productivity, 1990–2015 . . . . . . . . . 237 7.1 In 2017, few housing loan liquidity facility subsidies went to urban areas. . . . . 255 7.2 Congestion in Indonesian metropolitan areas peaks in the evening. . . . . . . . . . . 264 7.3 Population density is correlated with peak hour congestion intensity. . . . . . . . . 265 7.4 The annual cost of congestion in Indonesian metropolitan areas exceeds US$100 million in six cities, and exceeds US$2.6 billion in DKI Jakarta . . . . . . 266 B8.5.1 Foreign travelers to Indonesia and Bali, and Bali’s share of total, 1969–2016. . . . 289 S3.1 Access to safely managed sanitation in Indonesia is desperately lagging. . . . . 297 xiv  TIME TO ACT Maps O.1 Neighborhoods in Jakarta are highly segregated by skill level, with more-skilled residents living closer to basic services. . . . . . . . . . . . . . . . . . . 12 O.2 Indonesian cities face large infrastructure investment financing gaps, 2014 . . . . . 17 1.1 Jakarta is by far the largest of Indonesia’s metro areas, 2016. . . . . . . . . . . . . . . . 58 1.2 Slow versus fast: Built-up areas expanded more slowly in the already built-up Jakarta metro area than in the Mojokerto metro area, between 2000 and 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.1 Poverty and vulnerability rates in Indonesia, by district . . . . . . . . . . . . . . . . . . . 133 4.2 In nine major multidistrict metro areas, pockets of poverty exist alongside pockets of prosperity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 4.3 Access to health and education services for Jakarta core . . . . . . . . . . . . . . . . . . 149 4.4 Human capital and skills are concentrated in metro cores . . . . . . . . . . . . . . . . . 149 S1.1 There will be a substantial increase in the number of people exposed to fluvial flooding, between 2016 and 2055. . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 S1.2 The number of people exposed to coastal flooding will increase significantly between 2016 and 2055. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 5.1 For 14 subnational governments, investment needs for infrastructure projects far exceed borrowing capacity and revenue, 2014. . . . . . . . . . . . . . . . . 192 6.1 Consolidated air passenger flows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 B7.3.1 FLPP housing developments according to the suitability index in Semarang. . . . 256 B8.1.1 The Singapore–Johor–Riau growth triangle. . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 8.1 The Integrated Economic Development Zone program included 35 districts . . . 281 8.2 Twelve special economic zones have been established or are planned. . . . . . . . . 284 Tables O.1 Importance of the ACT principles to different types of place. . . . . . . . . . . . . . . . . 15 OA.1 Matrix A: Reforms to governance and subnational finance. . . . . . . . . . . . . . . . . . 27 OA.2 Matrix B: Policies for more connected places. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 1.1 The different types of place in Indonesia’s portfolio. . . . . . . . . . . . . . . . . . . . . . . . 57 1.2 Summary characteristics for different types of place in Indonesia, circa 2016 . . . . 61 1.3 Built-up area growth has been rapid in rural places but slower elsewhere, 2000–14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 1.4 Urban population growth and its components in major multidistrict metro areas, 2000–10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 1A.1 Scoring system for urban/rural classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 1C.1 Multidistrict and single-district metro areas in Indonesia. . . . . . . . . . . . . . . . . . . . 71 3.1 ­ Disparities in average and underlying productivity across districts, 2008–15. . . . 108 3.2 ­ Estimated elasticity of a district’s underlying productivity with respect to different agglomeration forces, by type of district . . . . . . . . . . . . 113 ­ B.1 3 Regression results for key drivers of a district’s underlying productivity. . . . . 121 ­ 3C.1 Regression results for effects of the local business environment on firm productivity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 C o n te n t s   xv 4.1 Proximate district-level correlates of the welfare return to secondary education. . . . 137 4.2 Densely populated places also generally have high inequality . . . . . . . . . . . . . . . 143 4.3 More segregated multidistrict metro areas generate weaker human capital externalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 4A.1 Proximate district-level characteristics correlated with welfare returns to secondary and tertiary education. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 4B.1 Regression results for the returns to aggregate human capital. . . . . . . . . . . . . . . 155 4C.1 Regression results for the returns to aggregate human capital by skill level. . . . . 156 5.1 Types of metropolitan governance mechanisms in Indonesia. . . . . . . . . . . . . . . . 187 5.2 Several plans related to urban development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 5.3 Indonesia’s intergovernmental transfers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 5.4 Policy options to address coordination challenges, build capacity, and expand opportunities for financing. . . . . . . . . . . . . . . . . . . . . . . . . 200 6.1 Toll road projects, complete and in the pipeline, 2017. . . . . . . . . . . . . . . . . . . . . 224 6.2 District groups in the analysis of the impact of transfers on road construction. . . . . 234 6C.1 Differential effect of 1995 population on roads. . . . . . . . . . . . . . . . . . . . . . . . . . 243 6C.2 Differential effect of current population on roads . . . . . . . . . . . . . . . . . . . . . . . . 243 6C.3 Differential effect of current population on roads—triple difference results, urban versus rural. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 7.1 Annual land price increases within the Jakarta multidistrict metro area, 2010–14. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254 xvi  TIME TO ACT Foreword Indonesia, the world’s fourth most populous classifying the different types of urban and country, is becoming increasingly urban. rural places across the country. The face of Today over half of the population lives in cit- urbanization ranges from large, thriving ies and towns; by 2045, the centenary of metropolises spanning multiple districts such Indonesia’s independence, nearly three-­ as Jakarta and Bandung (“multi-district quarters will. Urbanization promises a better metro areas”) to smaller metropolitan areas, life for Indonesians: around the world, major such as Lampung (“single-district metro global cities are centers of economic prosper- areas”), and even smaller towns such as ity and desirable places to work and live. No Manado and Ambon (“non-metro” urban large country has ever reached high-income areas). Distinctions are also made between the status without also becoming urbanized. experiences of Indonesians living and work- Although urbanization has been, on bal- ing in city centers (“cores”) and those who ance, a positive force in Indonesia, the coun- live on the outer edges of metropolitan areas try can do more to fully reap its benefits. For (“peripheries”). every one percent increase in the level of Drawing on a wide range of data sources, urbanization, Indonesia’s income per capita Part 1 of the report takes stock of the extent has risen less than in other developing coun- to which urbanization in Indonesia has deliv- tries in East Asia and the Pacific. And even ered on three key outcomes—prosperity, with over half of the population residing in inclusiveness, and livability. Indeed, the bene- urban areas, Indonesia remains a lower-­ fits of urbanization are tangible and substan- middle-income country. What can policy tial: urban areas are more productive and makers do to ensure that the promise of provide better access to services and infra- urbanization is realized in the most effective, structure, on average, compared to rural inclusive, and sustainable manner? areas. However, not all places and all people This report explores the challenges and have benefitted to the same degree. The gap opportunities associated with Indonesia’s between rich and poor has risen within all urbanization. Honoring the diversity of the types of places, and disparities in well-being archipelago, the report offers a novel way of between metro and non-metro areas remain xvii large by international standards. Even within expanding options for subnational financing metro areas, those living on the fringes have to meet basic infrastructure and service needs; more difficulties accessing amenities than building local capacity to better plan, imple- those living in the core. Furthermore, many ment, and finance urban development; and urban areas in Indonesia are strained by con- improving institutional coordination across gestion forces: choked roads, polluted air and all levels of government and across jurisdic- slums are common phenomena, even in some tions. In addition to these reforms, specific smaller cities. actions in the policy areas of housing and While cities everywhere face challenges of transportation can help to spread the benefits congestion and inequality, good policies can of urbanization within and across places. help shape the future of Indonesia’s urban The report recognizes that no one-size-fits- areas for the better. Part II of the report pro- all approach can tackle the challenges faced poses three basic policy principles for by Indonesia’s diverse “portfolio of places” Indonesia to leverage the promise of urban- and hence offers tailored advice to policy ization: Augment, Connect, and Target makers. Regardless of the approach, now is (“ACT”). Augment refers to expanding and the time to ACT to ensure that Indonesia ben- equalizing access to high-quality basic ser- efits fully from urbanization. As more and vices across all places, both urban and rural. more Indonesians settle in urban areas, it will Connect refers to enhancing the connections become increasingly difficult and costly to between places and between people and jobs, alter the trajectory of urbanization. This opportunities, and services. Target refers to report hopes to help policy makers chart a addressing persistent inequalities across roadmap of integrated, coordinated actions regions and groups of people that may endure that will foster prosperous and livable cities even if the first two policy principles were that can be enjoyed by all Indonesians. fully enacted. Key to implementing the ACT principles Rodrigo A. Chaves are institutional reforms to subnational Country Director, Indonesia and Timor-Leste g overnance and finance. These include ­ The World Bank xviii  TIME TO ACT Foreword It is with the highest pleasure that I welcome 2.7  percent increase in per capita income the publication of the Indonesia Urbanization for every percentage point increase in their Flagship Report. This report is the result of urban population shares, but Indonesia’s strong collaboration between the World Bank c orresponding “growth return” from ­ and the Ministry of National Development u rbanization has only been 1.4 percent. ­ Planning/National Development Planning Indonesia needs to be forward-looking in Agency, the Ministry of Finance, and the line developing its urban areas, so that potential ministries in Indonesia. Already, the report’s problems of congestion and sprawl can be preparation has served to provide evidence- anticipated and addressed through careful based policy recommendations for the next planning and design before they manifest Technical Draft of the National Medium- themselves too badly. Term Development Plan, 2020–24. The Urbanization must become a driver not report can also be used as a basis for only of economic growth but also of social Indonesia’s policy makers to start thinking change. In fact, Indonesia’s urbanization is about innovations in urban planning and facing critical issues of equity and accessibil- development, for both the medium- and long ity. Limited access to basic services, such as terms. Indonesia’s experience suggests that health and education, as well as to basic infra- urbanization policy reform is urgently structure, such as piped water, sanitation, and required to reestablish the country’s course housing, contribute significantly to urban toward an upper-middle-income status. To issues. Facing those issues, we recognize the borrow from the report’s title, now is the importance of making urbanization more time to act. inclusive by ensuring that all citizens have The report reveals that Indonesia has good access to basic services and infrastruc- become more prosperous as a result of urban- ture. To achieve this, Indonesia needs to take ization. Despite this progress, the full promise immediate policy actions. Our traditional of urbanization has yet to be realized. approach to urban policy needs to be replaced Between 1996 and 2016, other developing by a more comprehensive urban policy, where East Asian and Pacific countries enjoyed a all stakeholders—from national to local xix governments, from private to nonprofit orga- implementation, and development control. nizations—collaborate to determine the vision Along with that, President Joko Widodo has and interventions needed to achieve better taken the very important decision to relocate urban development. Successful urban policy the national capital city to East Kalimantan. requires a multisector and multistakeholder This decision is in line with the concept of approach; urban development is the responsi- integrating development in Java and bility not only of the government but of Kalimantan and spreading it to other big everyone. islands in the country. Through this approach, In 2016, the New Urban Agenda (NUA) was we hope that all regions, including rural areas, adopted by the United Nations. In Indonesia, will benefit from urbanization, thereby reduc- the NUA’s principles are integrated into ing spatial disparities. For the development of Presidential Regulation Number 59/2017 on metropolitan areas, new cities, and the new the Sustainable Development Goals (SDGs). capital city, this report provides insights on Currently, the Ministry of National urban management systems, urban financial Development Planning/National Development capacity, institutional coordination, and inte- Planning Agency is developing a long-term gration of national programs, as well as local, vision to achieve sustainable urban develop- provincial, and national capacity in planning, ment in the 2045 National Urban Policy financing, and managing urban areas. (Kebijakan Perkotaan Nasional). To monitor I appreciate the financial contributions of and evaluate the implementation of this policy, the governments of Switzerland and Australia especially at the local level, we are developing that have supported the production of this the Sustainable Cities Index (Indeks Kota high-quality flagship report. The collabora- Berkelanjutan). This includes important indica- tion with the World Bank, other ministries, tors—many in line with the SDG indicators—to and the governments of Switzerland and enable local governments to see their progress Australia to produce this report has been a and identify areas in which they need to very fruitful experience, one that we hope to improve to achieve sustainable urbanization. further strengthen and carry forward. For the next National Medium Term Development Plan, the Ministry of National Bambang P. S. Brodjonegoro Development Planning/National Minister of National Development Planning/ Development Planning Agency is promoting Head of National Development equitable development through the develop- Planning Agency ment of metropolitan areas in outer Java, Jakarta exploiting big data analytics in planning, October 2019 xx  TIME TO ACT Foreword Urbanization has brought many benefits to the government recognizes that districts need Indonesia. People living in urban areas are to improve their capacity to raise locally gen- more productive, have higher incomes, and erated revenues, including from local taxes. enjoy better access to services and infrastruc- This is because in our decentralized system of ture. However, we have not benefited from governance, districts are the main actors urbanization as much as some other East responsible for delivering good quality educa- Asian countries. The quality of life can tion, health, and local infrastructure to the improve in many of our cities, where people public. The government is revising the neces- face long, congested commutes; live in sary regulations to strengthen the capacity of cramped quarters; and are vulnerable to natu- the districts to raise their own revenues in ral disasters. order to fund the necessary infrastructure. The answer is not to stop the forces of On the spending side, the 2020 National urbanization altogether but to design the right Budget is very much aligned with the policies to minimize these costs. For these rea- “Augment, Connect, and Target” (ACT) sons, I am very pleased that the World Bank framework discussed in this World Bank has published this comprehensive report on report. Recognizing the need for better access urbanization in Indonesia. This report will to clean water, sanitation, waste management, help guide Indonesian policy makers to ensure and mass transport in urban areas, the gov- that everyone can benefit from urbaniza- ernment continues to support public and pri- tion—regardless of whether they are living in vate investments in these areas. We will also a large metropolitan city in Java or in a rural continue to maintain a high allocation for area in West Nusa Tenggara. The lessons of infrastructure, investing especially in arterial this book are particularly timely as we discuss roads to ensure that small towns are con- the planned relocation of Indonesia’s capital nected to larger cities. Finally, we recognize city. that all Indonesians, no matter where they Fiscal policy can help to make urbaniza- live, have the right to good quality health care tion more inclusive and can contribute more and education. This is why we will continue to growth. On the revenue side, for example, to prioritize programs that help develop our xxi human capital—especially to alleviate stunt- this opportunity, urbanization can help ing, which affects many children. Indonesia become a high-income country. Fiscal policy matters, but it will not drive Thus, we must be bold in our resolve to make better urbanization outcomes by itself. The urbanization work for all Indonesians. Ministry of Finance looks forward to collabo- rating with other ministries, agencies, and Sri Mulyani Indrawati local governments, as well as development Minister of Finance, Republic of Indonesia partners, to ensure that Indonesia reaps the Jakarta benefits of urbanization. If we capitalize on October 2019 xxii  TIME TO ACT Acknowledgments This report was prepared by a team led by The Spotlights in the report were prepared Mark Roberts, Frederico Gil Sander, and by Zuzana Stanton-Geddes and Yong Jian Sailesh Tiwari. The core team of chapter Vun (Spotlight 1), Sailesh Tiwari (Spotlight 2), authors also consisted of Mulya Amri, Judy Christophe Prevost (Spotlight 3), and Natsuko Baker, Souleymane Coulibaly, Nancy Lozano Kikutake (Spotlight 4). The report was also Gracia, Jane Park, Giuliana De Mendiola informed by a series of background papers. Ramirez, Stephane Straub, and Pui Shen Authors and contributors to these background Yoong. Other important contributors included papers who have not already been named Marcus Lee, Matthew Wai-Poi, David Ingham, include Hamidah Alatas, Maarten Bosker, Gayatri Singh, Christopher Crow, and Massimiliano Cali, Keerthana Chandrashekar, Jonathan Hasoloan. The work was conducted Sheng Fang, Taufik Hidayat, Claire Hollweg, under Victoria Kwakwa (Vice President, East Vitalijs Jascisens, Jonathan Lain, Ririn Salwa Asia and Pacific [EAP] region) with the g ­ eneral Purnamasari, Mayla Safuro Lestari Putri, guidance of Rodrigo Chaves (Country Husnul Rizal, Alexander Rothenberg, Audrey Director, Indonesia), Abhas Jha (EAP I Sacks, Imam Setiawan, Akhmad Rizel Shidiq, Practice Manager for Urban and Disaster Risk Della Temanggung, and Lixin C. Xu. Empirical Management), Ndiame Diop (EAP Practice work for the report was underpinned by an Manager for Macroeconomics, Trade and extensive subnational database for Indonesia Investment), and Salman Zaidi (EAP Practice that was developed by Ratih Dwi Rahmadanti, Manager for Poverty and Equity). Further Lourentius Dimas Setyonugroho, Abigail guidance was provided by Taimur Samad, Ho, and Pui Shen Yoong, with help from Kevin Tomlinson, and Stephan Garnier. The Katie McWilliams, Benjamin Stewart, Jane report was requested by the Government of Park, Brian Blankespoor, Massimiliano Cali, the Republic of Indonesia and prepared in Taufik Hidayat, Muhammad Hazmi, and close collaboration with both Indonesia’s Shiyan Zhang. Ministry of National Development Planning / The team was fortunate to receive excel- National Development Planning Agency lent advice and guidance from the following (Bappenas) and its Ministry of Finance. peer reviewers at various points in the report xxiii preparation process: Peter Ellis, Samuel Bruce Ross-Larson was the principal edi- Freije-Rodriguez, Bert Hofman, Somik Lall, tor, working with his Communications Sandeep Mahajan, Barjor Mehta, and Martin Development Incorporated colleagues Meta Rama. Although we are very grateful for the de Coquereaumont and Joseph Brinley, and guidance received, these reviewers are not Joe Caponio was the production editor, responsible for any remaining errors or omis- working with Mike Crumplar. Patricia sions. Additional insights from Sudhir Shetty, K a t a y a m a , f r o m t h e Wo r l d B a n k ’s Vivi Alatas, Camilla Holmemo, Adri Asmoro Development Economics Strategy and Laksono Poesoro, Nicholas Menzies, Operations unit, and Mary Fisk and Yaneisy Wicaksono Sarosa, Daniel Van Tuijll, Martinez, from the Bank’s formal publishing Kathleen Whimp, Thalyta Nandya Yuwono, unit, were responsible for the design, typeset- Alanna Simpson, Abigail Baca, Brenden ting, printing, and dissemination of both the Jongman, and Andrew Mason are also grate- hard- and soft-copy versions of the report. fully acknowledged. Last, but not least, we thank Inneke Herawati In preparing the report, the team benefited Ross, Marleyne (Alin) Danuwidjojo, and from feedback received during several work- Rebekka Hutabarat for unfailing administra- shops that were organized in Indonesia in col- tive support. laboration with Bappenas and the Ministry of This work received financial support from Finance. These workshops involved partici- the Swiss State Secretariat for Economic pants from a wide range of central govern- Affairs (SECO) through the Indonesia ment ministries and agencies, as well as from Sustainable Urbanization Multi-Donor Trust city governments, academic institutions, and Fund (IDSUN MDTF) and from the research think tanks. Australian government through the Local The team is also grateful for the support Solutions to Poverty (LSP) and Partnership provided by senior management of the World for Knowledge Based Poverty Reduction Bank’s Social, Urban, Rural and Resilience; (PKPR) trust funds. Macroeconomics, Trade and Investment; and Poverty and Equity Global Practices. May 2019 xxiv  TIME TO ACT Abbreviations ACT augment, connect, target AP1 Angkasa Pura 1 AP2 Angkasa Pura 2 ATR/BPN Kementerian Agraria dan Tata Ruang/Badan Pertanahan Nasional (Ministry of Agrarian Affairs and Spatial Planning / National Land Agency) Bappenas Badan Perencanaan Pembangunan Nasional (Ministry of National Development Planning/National Development Planning Agency) BNBP Badan Nasional Penanggulangan Bencana (National Disaster Management Authority) BPS Badan Pusat Statistik (Statistics Indonesia) BRT bus rapid transit BSPS Bantuan Stimulan Perumahan Swadaya (Self-Help Housing Stimulus) CAGR compound annual growth rate DKI Jakarta Daerah Khusus Ibukota Jakarta EMS energy management system FLPP Fasilitas Likuiditas Pembiayaan Perumahan (Housing Loan Liquidity Facility) GDP gross domestic product GHSL Global Human Settlement Layer HCI human capital index HRD-FP high-resource districts in favored provinces HRD-UP high-resource districts in unfavored provinces IFLS Indonesia Family Life Survey INPRES instruksi presiden (presidential instruction program) xxv KAPET Kawasan Pengembangan Ekonomi Terpadu (integrated economic development zone) KK Kartu Keluarga (identification card) KTP Kartu Tanda Penduduk (identification card) LRD-FP low-resource districts in favored provinces LRD-UP low-resource districts in unfavored provinces LSCI Liner Shipping Connectivity Index MOHA Ministry of Home Affairs PISA Programme for International Student Assessment PODES Potensi Desa (Survey of Village Potential) PPP public–private partnership Rp Indonesian rupiah RPJMD Rencana Pembangunan Jangka Menengah Daerah (district-level medium-term development plan) RPJMN Rencana Pembangunan Jangka Menengah Nasional (national medium-term development plan) RT Rukun Tetangga (neighborhood association) RTBL Rencana Tata Bangunan Lingkungan (local‑level urban design guidelines) RTRW Rencana Tata Ruang Wilayah (district-level spatial plan) SAKERNAS Survei Angkatan Kerja Nasional (National Labor Force Survey) SDG Sustainable Development Goal SEZ special economic zone SI Survei Industri (Industrial Survey) SIM subscriber identity module SNG subnational government SSB Subsidi Selisih Bunga (Interest Rate Buy-Down Subsidy) SSL sector-specific law SUSENAS Survei Sosial Ekonomi Nasional (National Socio-Economic Survey) TFP total factor productivity UNCTAD United Nations Conference on Trade and Development xxvi  TIME TO ACT Overview Indonesia stands as a country transformed by The unprecedented growth of urban areas has urbanization. When its independence was given rise to negative congestion forces, associ- p roclaimed in 1945, only one in eight ­ ated with the pressure of urban populations on Indonesians lived in towns and cities, and the infrastructure, basic services, land, housing, and country’s entire urban population stood at the environment, that are undermining the liva- about 8.6 million, roughly equal to that of bility of cities and dampening the prosperity London today. By contrast, today about gains from urbanization. 151 million, or 56 percent, of Indonesians live Put differently, urbanization has not in urban areas, roughly 18 times the popula- f ulfilled its potential to drive sustainable ­ tion of London.1 improvements in prosperity, inclusiveness, As Indonesia has urbanized, so has it and livability in Indonesia. This, in turn, can climbed the ladder of development and be traced to a failure to adequately ACT: ­ prosperity. Since 1950, average gross domes- tic product (GDP) per capita has increased •  Augment the coverage and quality of basic almost ninefold in real terms, and the average services and urban infrastructure to better Indonesian today enjoys a standard of living manage congestion forces and address far surpassing that of previous generations.2 large disparities in human capital out- In part, a more prosperous Indonesia today is comes both across and within places. due to the productivity benefits that arise •  Connect urban areas of different sizes with from urban agglomeration and the associated each other, with surrounding rural areas, transformation from an agrarian society to and with international markets—and to one more based on industry and services. connect people with jobs and basic services That climb, however, has been slower and within urban areas—to e ­ nhance inclusive- more arduous than the rapid pace of ness both within and among areas. urbanization. Hence, Indonesia remains a lower- ­ •• Target places and people that may be left middle-income country, and although almost behind by the urbanization process to ensure everyone has benefitted in absolute terms, the that they share in the prosperity benefits of relative gains from urbanization have been urbanization and that urban areas are uneven within cities and across the country. livable for everyone. ­ 1 To overcome these shortcomings and people—or more than 70 percent of its ensure that Indonesia gets the most out of population—will live in towns and cities. ­ urbanization, policy makers need to under- Because the urban environment is difficult take bold institutional reforms and imple- and costly to change once built, delays in ment decisive policies to ACT. This action will risk locking Indonesia further improvement involves reforming the ways into a suboptimal trajectory of urban devel- urban areas are governed and financed, opment. In the meantime, policy makers can with a focus on expanding options for do plenty to ensure that urbanization deliv- financing infrastructure and basic services, ers a prosperous and inclusive Indonesia of as well as improving coordination between livable cities. different levels and sectors of government and between districts that belong to a com- mon metropolitan area. It also involves By 2045, 220 million building stronger capacities to plan, imple- Indonesians will live in ment, and finance urban development. Across the board, it will be necessary to tai- urban areas lor many of the policy actions required to Although Indonesia urbanized rapidly in ACT according to the type of place—for the past, its current pace of urbanization example, according to whether an urban can be described as near “normal” or “typi- area is a large metropolis such as Jakarta or cal.” In the 1980s and 1990s, Indonesia’s Surabaya or a smaller, less connected urban urbanization growth rate averaged more area such as Bima. than 3 percent a year—faster than in other Even though such measures provide the developing East Asian countries at the time, foundations to ACT, they are unlikely to be including China.3 Since the turn of the cen- sufficient to connect people with jobs and ser- tury, however, the pace of urbanization has vices in urban areas, for which additional begun to slow, returning close to that seen policies and investments are needed. These in the 1950s and 1960s. Between 1990– include policies to facilitate the supply of 2000 and 2010–17, Indonesia’s pace of well-located affordable housing, better urban urbanization more than halved, lagging public transport, and the better management other countries in the region but in line of traffic within a framework of more effec- with its level of urbanization.4 When bench- tive urban and spatial planning. Adequately marked against the historical experiences of connecting urban areas with each other, with countries globally, the recent moderation surrounding rural areas, and with interna- represents a return to what can be regarded tional markets will also require addressing as a “typical” pace of urbanization for the key regulatory issues in transportation country (figure O.1). markets. In Indonesia, the urbanization process has Finally, making sure that no island and primarily been driven by the densification of no place are left behind will require settlements and their acquisition of infrastruc- Indonesia to rethink its approach to place- ture and amenities, leading to their reclassifi- based policies, putting a stronger emphasis cation from rural to urban,5 followed by the on human capital in the design of those pol- natural growth of population in urban areas. icies. A paradigm shift in urban planning These factors accounted for more than and design is also needed to ensure that all 80 percent of Indonesia’s urban population groups of society—especially women and growth between 2000 and 2010. By contrast, girls, the elderly, and people with disabili- net rural–urban migration contributed less ties—fully benefit from all that urban areas than 20 percent of overall urban population have to offer. growth.6 The role of migration in explaining To succeed, Indonesia needs to ACT now. urban population growth is relatively small By 2045, the centenary of Indonesia’s inde- in  Indonesia compared to India and espe- pendence, approximately 220 million cially China, where migration contributed 2  TIME TO ACT 56  percent of urban population growth FIGURE O.1  Indonesia’s post-2000 pace of between 2000 and 2010 (World Bank and urbanization is typical by international standards DRC 2014). The urban transformation has given rise to 10 y = –1.31 ln(x) + 6.05 a diverse and vibrant “portfolio of places” R2 = 0.36 Growth rate of urban share of population (%) (box O.1). Today, about 57 percent of 8 Indonesia’s urban population lives in metro- politan areas that span multiple districts 6 (“multidistrict metro areas”) or that comprise only a single district (“single-district metro 4 1980s 1990s areas”). The remaining 43 percent of the 1970s urban population lives outside metropolitan 2 1950s 2000s 1960s 2010s areas. 0 The experience of urbanization can differ not just between metropolitan and nonmetro- –2 politan areas, but even within each type of area. In multidistrict metro areas, for exam- –4 ple, many Indonesians live in the “periphery” 0 10 20 30 40 50 60 70 80 90 100 districts, commuting to the “core” to work Initial urban share of population (%) and to access services. These periphery areas All countries Indonesia Fitted line (all countries) can be predominantly urban or rural. Outside metropolitan areas, most Indonesians live in Source: Calculations based on data from the United Nations World Urbanization Prospects: 2018 “nonmetro rural” areas, but some also live in Revision database (https://esa.un.org/unpd/wup/). “nonmetro urban” areas—districts in which Note: Each data point shows the average growth rate of a country’s urban population share (that is, the share of its population that lives in urban areas) over a given period for its urban population share at the most of the population lives in small cities beginning of that period. The periods considered are 1950–60, 1960–70, 1970–80, 1980–90, 1990–2000, and towns that provide, for example, market 2000–10, and 2010–15; therefore, the figure contains seven observations for each of 231 countries. BOX O.1  Indonesia’s portfolio of places Four broad types of urban and rural places can predominantly rural (“rural periphery”), d be distinguished in Indonesia (figure BO.1.1).a where a predominantly urban district is one Multidistrict metro areas are large met- with at least 50 percent of the population liv- ropolitan areas, such as Jakarta, Surabaya, ing in urban settlements. Medan, and Makassar, with labor markets Single-district metro areas are kota districtse that cut across multiple districts, as defined with a population of at least 500,000 and aver- using commuting flow data. A multidistrict age population densities that resemble those of metro, in turn, consists of the following types multidistrict metro areas, but whose labor mar- of subareas: kets are largely confined within the boundaries •  Metro core corresponds to the district within of a single district. Examples include Palembang, the metro area that exhibits the highest average Pekanbaru, and Samarinda. population density,b except for Jakarta, where Nonmetro urban areas are districts that do the core is taken to be Daerah Khusus Ibukota not meet the criteria to be classified as either a (DKI, or special capital region) Jakarta.c single-district metro area or part of a multidis- •  Metro periphery corresponds to districts trict metro area, but within which most of the linked to the core through strong commuting population lives in urban settlements. Such dis- flows. Metro periphery districts can be either tricts may be either kota or ­ kabupaten. Thirty- predominantly urban (“urban periphery”) or two of 57 nonmetro urban areas are  kota. Box continued on next page O v er v ie w   3 BOX O.1 Continued FIGURE BO.1.1  Stylized depiction of the different types of place within the “portfolio of places” Nonmetro Metro Single-district urban core metro Urban periphery Rural periphery Nonmetro rural Examples include Cirebon, Manado, and of urbanization. On the basis of 1996 administrative boundaries, metro Medan contains three kota districts, two of which have 100 percent of their population Mataram. living in urban areas. Nonmetro rural areas are nonmetro dis- c. DKI Jakarta comprises six districts—Jakarta Pusat, Jakarta Barat, Jakarta tricts in which most of the population lives in Selatan, Jakarta Timur, Jakarta Utara, and Kepulauan Seribu. d. For brevity, we refer to “predominantly urban” and “predominantly rural” rural settlements. The majority (354 of 377) are periphery areas as simply “urban periphery” and “rural periphery” in the ­ kabupaten. Examples include Kabupaten Ciamis, remainder of this overview. However, it is important to remember that a Kabupaten Kampar, and Kabupaten Kupang.f predominantly urban area may still have a large rural population, and vice versa. e. There are two types of districts in Indonesia—kota and kabupaten. Kota translates as “city,” while kabupaten designates what has traditionally been a. For a more detailed description of the methodology used to derive these considered a rural district. typologies see chapter 1 and Park and Roberts (2018). f. Collectively, these areas still possess a large urban population, which lives b. Except for metro Medan, the identification of metro cores is robust to other mostly in smaller cities and towns that may act more as local market centers. criteria for their selection, such as a district’s status as a kota and its level In 2016, nonmetro rural areas had a collective urban population of 37 million. centers for agricultural output produced in surrounding rural settlements. Delivering on the promise Whereas urbanization is at an intermediate of urbanization requires stage for Indonesia as a whole, as well as for managing congestion the Jawa-Bali region, it is still at an incipient stage in the rest of the country. 7 By 2045, forces when Indonesia will celebrate the centenary Urbanization can boost economic prosperity. of its independence, approximately 220 mil- This is because urbanization fosters positive lion people—or more than 70 percent of its agglomeration forces, creating an environment population—will live in towns and cities. The that is conducive to innovation and enhanced promise of urbanization is that this process productivity. As people and firms cluster in can lead to a more prosperous and inclusive settlements, it becomes easier to match talent Indonesia of livable cities. to jobs, exchange ideas and knowledge, share 4  TIME TO ACT FIGURE O.2  Urbanization outcomes are jointly determined by the interplay of agglomeration and congestion forces Agglomeration forces Prosperity Outcomes Urban Inclusiveness growth Congestion Livability forces inputs, and access markets. With economies of connectivity between places, prosperity may scale, larger cities can provide more and better- remain largely “locked up” in the cores of quality services and infrastructure, because the metropolitan areas rather than being shared ­ fixed costs of doing so are spread over more more broadly. Likewise, disconnected growth beneficiaries. With greater mobility and con- within urban areas is associated with residen- nectivity among places, labor and capital can tial segregation between the high and the low be allocated more efficiently, creating more skilled. Such segregation exacerbates inequal- opportunities for people to prosper. ity, hampers the ascension of the poor to the Realizing the promise of urbanization, how- middle class, and reduces the strength of ever, requires managing negative “congestion knowledge spillovers. forces” that intensify as areas urbanize and that This report therefore assesses whether threaten to limit the benefits of agglomeration. urbanization in Indonesia has delivered pros- These congestion forces arise from the pressure perity, inclusiveness, and livability. Figure O.2 of urban populations on basic services, infra- describes the report’s analytical framework. structure, land, housing, and the environment. The failure to adequately manage these forces gives rise to, among other things, grid-locked Urbanization has led to prosperity in streets, slums and overcrowded housing, and Indonesia, but has not reached its full potential inequitable access to good schools and hospi- tals. Such congestion forces directly undermine To some extent, urbanization and economic the ­livability of urban areas, reducing their prosperity have gone together in Indonesia. attractiveness as places to live and work. They Districts with higher shares of their popula- also undermine human capital accumulation tions living in urban settlements have higher and encourage urban areas to sprawl outward, per capita incomes. This is partly due to better with negative implications for knowledge spill- labor market opportunities in urban areas, overs and other prosperity-enhancing agglom- which offer more jobs in industry and services eration forces. than rural areas. These sectors tend to gener- Not everyone may benefit from the prosper- ate more formal, better-paid, and stable jobs ity and livability generated by urbanization. than those in the agricultural sector. In many countries around the world, larger cit- Consistent with the existence of positive ies are more unequal, while some cities lead agglomeration forces, workers in more urban and others lag. The benefits of urbanization areas are also more productive, and thus better may also fail to spill over to those who remain paid, than otherwise identical workers in less in the countryside, creating widening gaps urban areas. A person who works in a multi- between urban and rural areas, which can district metro core or a single-district metro threaten social cohesion. Along with invest- area earns 25 percent more than a comparable ments aimed at the universal provision of basic worker—of the same age, gender, marital sta- services, the inclusiveness of urbanization rests tus, and education level—who is employed in on connectivity both across and within places, the same industry in a nonmetro rural area. and therefore on integration. Without a ­ dequate Workers in urban periphery areas also enjoy a O v er v ie w   5 FIGURE O.3  Workers in more urban areas earn similar, albeit slightly smaller, “wage premium” higher wages than comparable rural workers due to their greater productivity (figure O.3). Better economic opportunities in 30 more  urban areas have also helped many 25.4 25.4 Indonesians escape poverty. Rates of poverty 25 and vulnerability to poverty in multidistrict Wage premium over nonmetro rural (%) 20.7 20 metro cores, urban peripheries, and single-­ 15 district metro areas are substantially lower than 10.7 those in nonmetro rural areas (figure O.4). 10 Nonetheless, significant pockets of poverty 5 2.4 persist in urban areas, as reflected by the ­ 0 38 ­percent of the population—equivalent to Metro Single-district Urban Nonmetro Rural 21.5 million Indonesians—in nonmetro urban core metro periphery urban periphery areas who are poor or vulnerable to poverty. Urban areas have also provided a robust Source: The wage premium was estimated using data from the August 2008–August 2015 rounds of pathway to the middle class. Nonmetro rural Indonesia’s National Labor Force Survey (SAKERNAS) following the methodology described in chapter 3. households that moved to multidistrict metro Note: The values reported in the figure are the average wage premium that workers in districts that belong to a given type of place command over observationally equivalent workers in areas have better prospects of becoming mid- nonmetro rural districts, controlling for the island-region that a district belongs to. A district’s dle class than those that moved to other rural wage premium is estimated using an augmented Mincerian wage regression (Mincer 1974) that areas. The odds of a migrant to a multidistrict controls for both the observable characteristics of workers (gender, marital status, age and its square, and educational attainment) and the jobs (industry of employment, average number of metro core entering the middle class have hours worked per week, and the length of the worker’s tenure with the current employer and its weakened since 2010, but the chances have square) they occupy. remained robustly high for those moving to FIGURE O.4  Better economic opportunities in urban areas have led to less poverty and vulnerability to poverty Metro core 3.1 13.9 Urban periphery 6.3 18.2 Rural periphery 11.2 23.4 Single-district metro 5.0 18.1 Nonmetro urban 11.4 26.1 Nonmetro rural 14.6 27.9 0 5 10 15 20 25 30 35 40 45 Share of population (%) In poverty Vulnerable to poverty Source: Calculations using data from Indonesia’s 2017 National Socio-Economic Survey (SUSENAS). Note: Poverty rates are based on official poverty lines of the government of Indonesia. Vulnerability is defined as having per capita household consumption above the poverty line, but below 1.5 times the poverty line. 6  TIME TO ACT urban peripheries. This finding suggests that FIGURE O.5  Urbanization has delivered less urban peripheries have retained the advan- prosperity in Indonesia than in the rest of the region tages of being close to the prosperity of the metro cores while avoiding the worst of the 3.5 costs of congestion. 3.0 3.0 Growth return, 1996–2016 (%) Notwithstanding these gains in prosperity, 2.7 Indonesia has not benefitted as much from 2.5 urbanization as some other East Asia and 2.0 Pacific (EAP) countries (figure O.5). Between 1.4 1996 and 2016, every percentage point increase 1.5 in the share of Indonesia’s population living in 1.0 urban areas was associated with a 1.4 percent increase in GDP per capita, but for developing 0.5 EAP countries, the corresponding increase in 0 GDP per capita was 2.7 ­ percent. For China, the China EAP-developing Indonesia “growth return” to urbanization was even higher, with every percentage point increase in Sources: Calculations based on data from World Bank World Development Indicators (http://datatopics​ the level of urbanization being associated with .worldbank.org/world-development-indicators/) and the United Nations World Urbanization percent increase in GDP per capita. a 3.0 ­ Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). Note: Figure shows the percentage increase in GDP per capita (constant 2011 international dollars) associated with a one-percentage-point increase in the share of population living in urban areas. EAP-developing refers to East Asia and Pacific countries excluding high-income countries, China, and Benefits of urbanization have spread, but Indonesia. inequality remains high, especially within places FIGURE O.6  Inequality has increased everywhere To what extent have all households in and remains highest in multidistrict metro cores Indonesia, regardless of where they live, been able to benefit from the prosperity that urban- 0.45 ization generates? Looking at real consumption per capita as a measure of welfare, households Gini coefficient, 2001–17 0.40 in more urban areas remain ­ significantly better- off than those in rural areas.8 Although the gap has narrowed over time, the level of welfare in 0.35 nonmetro rural areas remained 43 percent lower than that of Jakarta’s metro core (Daerah 0.30 Khusus Ibukota [DKI] Jakarta) in 2015. Rural periphery areas and nonmetro urban areas also welfare penalties of 35 percent and suffer large ­ 0.25 27 percent respectively, relative to DKI Jakarta. 03 04 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 20 Urban periphery areas, by contrast, suffer a 20 Metro core Urban periphery Rural periphery much smaller welfare penalty—only 7 percent Single-district metro Nonmetro urban Nonmetro rural relative to DKI Jakarta—partly because of their proximity to the prosperity of the multidistrict Source: Calculations based on consumption data from Indonesia’s National Socio-Economic Survey (SUSENAS). metro cores. Note: Three-year moving average of Gini coefficient. Despite these large welfare disparities between urban and rural areas, the bulk of consumption inequality is accounted for by within all types of place and it is highest inequality within places.9 Indeed, within-place in exactly the most prosperous areas—that is, inequality accounted for nearly 86 percent of in multidistrict metro cores and their urban total inequality in 2017, up from 82 percent peripheries (figure O.6).10 in 2001. Moreover, since 2001, the gap That more prosperous places are more between the rich and the poor has increased unequal is not unusual. Globally, within O v er v ie w   7 countries, larger and more prosperous cities Uneven learning outcomes among children tend to be more unequal than smaller, less are a further indication that the benefits of prosperous, cities. This is true, for example, urbanization are not equally spread within cit- within both the United States and within Latin ies (figure O.7). Although children in urban American and Caribbean countries (Ferreyra areas generally have better access to schools 2018). In both cases, higher inequality in larger and average educational attainment is higher cities is driven by the fact that they are home to than in other types of area, children from lower- more-skilled populations. This phenomenon is income households appear to be falling behind. driven, in part, by the greater tendency of Looking at science test scores on the Programme skilled workers to migrate, which leads to their for International Student Assessment (PISA), being disproportionately concentrated in larger the gap between children in the bottom 20 per- cities, creating a larger spread of skills and thus cent of the socioeconomic distribution and wages in those cities. Furthermore, even though those in the top 20 percent is largest in cities.12 both low- and high-skilled workers might ben- In 2015, children in the top 20 percent scored efit from higher overall levels of human capital about 86 points higher than children in the bot- in larger cities, they may not do so equally, thus tom 20 percent in large cities, whereas in other further contributing to higher inequality. cities, children in the top 20 percent scored In Indonesia, as in the United States and 103 points higher. As this generation enters the other countries, more densely populated places labor market, such learning gaps may com- also have more highly educated workforces pound within-place inequalities, further limit- and hence are more productive. High-skilled ing the potential of urbanization to bring more workers also benefit more from being in such a prosperity to all Indonesians. place. Whereas a one-year increase in an area’s average years of schooling yields a 10 percent return in earnings for high-skilled workers, With better management of congestion the corresponding returns for medium- and forces, Indonesian cities can become more low-skilled workers are about 6 and 3 percent, prosperous and livable respectively.11 Larger cities therefore tend to be Around the world, people are drawn to the more unequal in Indonesia. bright lights of large cities, not only because of the better economic opportunities they offer FIGURE O.7  Learning gaps between the top and the but also because of the perceived excitement bottom socioeconomic groups are largest for cities that goes along with their bustle. However, with bustle also comes congestion of infra- 120 structure and markets that, without the right policies and investments, can undermine both top 20 percent and bottom 20 percent 102.6 Di erence in science scores between 100 livability and prosperity. In Indonesia, many 85.8 urban areas are showing signs of strain from 80 their inability to manage congestion forces. 67.0 Negative congestion forces are especially prev- 60 alent in multidistrict metro cores and single- 40.4 district metro areas, which suffer from a lack of 40 29.6 affordable housing, severe traffic congestion, and unacceptable air pollution levels. These 20 conditions are helping to push people and firms 0 to the peripheries of cities as they search for Village Small town Town City Large city cheaper land and housing, contributing to dis- connected and low-density urban growth that Source: Calculations based on 2015 science test score data from the Programme for International limits the productivity gains of agglomeration. Student Assessment (PISA), (http://www.oecd.org/pisa/). Although reliable and comparable data Note: Villages have fewer than 3,000 people, a small town has 3,000–15,000 people, a town has 15,000–100,000 people, a city has 100,000–1,000,000 people, and large cities have more than are difficult to come by, evidence suggests 1,000,000 people. These definitions cannot be mapped into the portfolio of places. that house price–to–income ratios in 8  TIME TO ACT FIGURE O.8  One-fifth of Indonesia’s urban population lives in slums, and overcrowded housing is commonplace in metro cores and single-district metro areas a. Share of urban population living in slums, 2015 b. Share of households in overcrowded housing 60 40 35 50 30 40 25 Percent Percent 30 20 15 20 10 10 5 0 0 2002 2016 lo . In bia sia il Th ia d Vi a La m ilip DR s a Co ep ne az n ny an d na i ne oP Ch R In m Br pi Ke ail et b do ra t, A Ph Jakarta core Jakarta periphery yp Eg Other metro core Other urban periphery EAP (excluding high-income) average Rural periphery Single-district metro Lower-middle-income average Nonmetro urban Nonmetro rural Sources: Panel a calculations based on data from World Bank World Development Indicators (http://datatopics.worldbank.org/world-development-indicators/); panel b calculations based on data from Indonesia’s 2002 and 2016 National Socio-Economic Surveys (SUSENAS). Note: EAP = East Asia and Pacific. Bandung, Denpasar, and Jakarta are higher cities in the world. On the TomTom Traffic than in New York.13 Along with a lack of Congestion Index, Jakarta was the third-most adequate access to mortgage financing, the congested of 18 megacities worldwide,16 with high cost of housing contributed to one-fifth an estimated 58 percent extra travel time of Indonesia’s urban dwellers living in slums needed for any trip, anywhere in the city, at any in 2015.14 Although this share is substan- time compared with a free-flow situation.17 tially lower than the share of the urban pop- Indonesian cities are also among the most con- ulation living in slums in countries such as gested on the Inrix Global Traffic Scorecard Kenya, the Lao People’s Democratic (figure O.9). Even in smaller cities such as Republic, and the Philippines, it still equates Padang and Yogyakarta, drivers spend about a to 29 million Indonesians—roughly equiva- quarter of their driving time stuck in conges- lent to three times the entire population of tion. According to new estimates prepared for Sweden—living in slums (figure O.8, panel this report, the total cost of traffic congestion a). There is also significant overcrowding of for Indonesia’s 28 multi- and single-district housing in multidistrict metro cores and sin- metro areas is at least US$4 billion (equivalent gle-district metro areas, especially in to 0.5 percent of national GDP) a year in Jakarta’s core, where 35 percent of house- excess travel time and fuel consumption, with holds lived in overcrowded housing in 2016, US$2.6 billion for Jakarta’s core alone.18  up from 28 percent in 2002 (figure O.8, Such traffic congestion, together with the panel b).15 use of polluting fuels in motorized vehicles and Traffic congestion is another congestion industrial coal plants that power urban areas, force that undermines the livability and pro- means that pollution is yet another congestion ductivity of Indonesia’s metro areas. Jakarta is force that afflicts Indonesia’s metro areas. consistently rated one of the 10 most congested Twenty of 28 multi- and single-district metro O v er v ie w   9 FIGURE O.9  Indonesia’s cities are among the most availability of affordable housing to help cre- congested in the East Asia and Pacific region ate metro areas in Indonesia that are highly residentially segregated. Multidistrict metro Padang areas are strongly sorted into high- and low- Yogyakarta skill neighborhoods (map O.1, panel a), with Malang high-skill neighborhoods centrally located Bangkok close to both the best jobs and services Pontianak (map O.1, panel b). High-skill workers there- Bandung fore face relatively little inconvenience from Bengkulu congestion in their daily commutes from home Tarogong to the office and in travelling to access basic Medan services. By contrast, low-skill households and Jakarta workers, priced out of good-quality housing in Sungai Pinang Semarang the cores, are concentrated in less central Tasikmalaya neighborhoods disconnected from both the Surabaya best jobs, and health and education services. Denpasar Not only is this strong residential sorting a Bogor source of heightened inequality and hence a Sakaka lack of inclusiveness within metro areas, but it Lat Krabang may also be undermining the aggregate Riyadh strength of human capital spillovers between Fujairah high- and low-skilled workers, with detrimen- Chiang Mai tal effects for productivity and prosperity. Kuwait City Empirical evidence undertaken for this report Jeddah shows that multidistrict metro areas with Dubai more residential segregation between high- Singapore and low-skilled workers generate weaker 0 5 10 15 20 25 aggregate human capital externalities.19 Share of driving time spent in congestion (percent) Source: Based on data from Inrix Global Traffic Scorecard 2017 (http://inrix.com/scorecard/). Note: The scorecard covers 1,360 cities, 15 of which are in Indonesia. Congestion refers to road speeds Closing service delivery gaps would extend that are less than 65 percent of free-flow speeds. Red bars indicate Indonesian cities. the livability and prosperity benefits of urbanization to all areas had unsafe ambient outdoor air pollu- Although metro and larger urban areas in tion in 2015 (figure O.10, panel a). Jakarta’s Indonesia struggle with issues of slums and air is fouler to breathe than that of Ho Chi overcrowding, traffic congestion, and pollu- Minh City, Kampala, Mexico City, and São tion, urban residents tend to do better, on Paulo; Pekanbaru’s air is also more polluted average, than their rural counterparts in terms than that of Mumbai and Shanghai of access to hospitals and schools, clean water, ­ (figure O.10, panel b). These high pollution and safe sanitation, all of which are impor- levels are associated with a host of diseases and tant determinants of both an area’s livability other health-related effects that undermine the and its level of human capital and, therefore, livability of metro areas. In Jakarta, an esti- prosperity. Virtually all Indonesians in multi- mated 60 percent of the population suffers district metro cores and single-district metro from air pollution–related diseases, and, in a areas have easy access to primary care facili- perceptions survey conducted for this report, ties (puskesmas), delivery facilities, and hospi- 70 percent of surveyed city dwellers identified tals. By contrast, more than 20 percent of “less pollution” as the most important urban those residing in nonmetro rural areas do not environment issue (World Bank 2018c). have easy access to hospitals, and over In addition to pollution, severe traffic con- 80 ­percent lack easy access to a private prac- gestion has combined with the limited tice doctor.20 10  TIME TO ACT FIGURE O.10  Unsafe to breathe: Most Indonesian metro areas have unacceptable air pollution a. Annual average PM2.5 concentration, b. Annual average PM2.5 concentration, Indonesian cities, 2015 Indonesian and international cities, 2015 Pekanbaru Dhaka Jambi Beijing Palembang Pekanbaru Bukittinggi Shanghai Padang Johannesburg Pontianak Mumbai Medan Seoul Jakarta Bangkok Semarang Medan Bandar Lampung Los Angeles Bandung Jakarta Salatiga London Surabaya Mexico City Surakarta Kampala Magelang São Paulo Mojokerto Surabaya Sukabumi Banjarmasin Moscow Yogyakarta Bandung Pasuruan Tehran Blitar Ho Chi Minh City Malang Istanbul Probolinggo Tokyo Balikpapan Paris Denpasar Bogota Samarinda Buenos Aires Banda Aceh Nairobi Makassar Makassar 0 10 20 30 40 50 60 70 0 30 60 90 120 150 Micrograms per cubic meter Micrograms per cubic meter Source: Calculations based on satellite-derived data from Dalhousie University where, for global consistency, cities are defined as “high-density clusters” using the algorithm of Dijkstra and Poelman (2014) as applied to Landscan-2012 gridded population data. Note: PM2.5 is particulate matter of 2.5 microns or less in diameter. In panel a, dark blue–colored bars indicate PM2.5 levels that exceed the World Health Organization’s standard of 10 micrograms per cubic meter for safe air. In panel b, red-colored bars indicate Indonesian cities. With better access to basic services, chil- portfolio of places. This convergence, in dren in urban areas are healthier and better turn,  has reduced the pressure for “push” educated. Even after controlling for differ- ­ migration—stimulated by the push of distress ences in access to services, as well as house- rather than the pull of opportunity—to the hold characteristics, children in multidistrict metro areas. Nevertheless, significant gaps metro cores are taller and heavier for their remain between urban and rural areas in age than children in nonmetro rural areas.21 access to education, health, water, sanitation, Individuals living in households in multidis- and hygiene services (figure O.11). trict metro cores and single-district metro Access to basic services continues to be areas are also more likely to be literate and to more challenging for both households living have finished primary school than those in on the rural fringes of large metro areas and rural areas. households in nonmetro rural areas. These Over time, some convergence has occurred areas continue to significantly lag urban areas in access to services across Indonesia’s on health and educational outcomes, O v er v ie w   11 MAP O.1  Neighborhoods in Jakarta are highly Better connectivity would further help in segregated by skill level, with more-skilled residents spreading prosperity from more urban to living closer to basic services less urban places Although service delivery gaps provide part of the explanation for disparities in prosperity and well-being across Indonesia’s portfolio of places, another important part of the story is a lack of connectivity. There is inadequate infra- structure to connect the core and periphery areas of multidistrict metro areas, and metro and nonmetro areas. This is partly due to underinvestment in transport infrastructure. Total transport investment fell from about 2 percent of GDP in 1995 to less than 1 ­ percent in 2000; since then, it has slowly recovered to pre-Asian financial crisis levels, reaching 2.2 percent in 2016. By contrast, during this period, other major developing countries were investing heavily in improving their transport connectivity between cities. In recent decades, China has constructed a 96,000 kilometer net- work of highways connecting the country’s largest cities and is building the world’s longest high-speed rail network to connect its main population centers (Bosker, Deichmann, and Roberts 2018). Meanwhile, between 2001 and 2012, India built its Golden Quadrilateral highway network, which, at nearly 6,000 kilo- meters in length, links the country’s four major metro areas—namely, Delhi, Kolkata, Mumbai, and Chennai (Ghani, Goswami, and Kerr 2016, 2017). Although transport investment in Indonesia has recovered since the Asian finan- cial crisis, the current level of investment is unlikely to be sufficient to meet the demand for transport, and, relative to population, Indonesia’s stocks of both main roads and rail- ways are low compared to many other coun- tries (figure O.12).23 Of course, as the world’s largest archipe- lagic country, Indonesia and its portfolio of Sources: For panel a, calculations are based on Indonesia’s 2010 Population Census. For panel b, travel time calculations are based on data from Google Maps and Trafi, while poverty data are from Smeru places could never be fully connected by Institute’s 2015 poverty map (http://www.smeru.or.id/en/content/poverty-and-livelihood-map- roads alone, especially outside Jawa. Both indonesia-2015). maritime and air transport connectivity are Note: The PPP $3.10 poverty rate refers to the share of the population living on less than $3.10 per day at 2011 constant international prices. PPP = purchasing power parity. also important, but here, too, Indonesia suf- fers from issues that hamper integration. In undermining the formation of human capital the maritime transport sector, which is espe- in these areas and hence their productivity. cially important for goods market integra- This is particularly significant as urban popu- tion, insufficient investment and an lation growth was fastest between 2004 and inadequate legal and regulatory framework 2016 in precisely these areas.22 contribute to poor performance of the ports 12  TIME TO ACT FIGURE O.11  Despite convergence, significant urban–rural gaps remain in access to basic services a. Share of households b. Share of households with a child c. Share of households lacking access to a preschool, under age 5 not delivered by a skilled without access to safe drinking 2003 and 2014 health worker, 2002 and 2014 water, 2002 and 2016 50 50 70 60 40 40 50 30 30 40 Percent Percent Percent 30 20 20 20 10 10 10 0 0 0 2003 2014 2002 2014 2002 2016 Jakarta core Jakarta periphery Other metro core Other urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Sources: Adapted from Lain 2018, using data from Indonesia’s 2002, 2014, and 2016 National Socio-Economic Surveys (SUSENAS) and 2003 and 2014 Surveys of Village Potential (PODES). FIGURE O.12  Indonesia’s road and railway networks are not as extensive as in many other countries a. Main roads per 1,000 population, circa 2015 b. Railways per 1,000 population, 1994 and 2014 1.4 0.8 1.17 Length of main roads (km) 1.2 0.7 Length of railways (km) 1.0 0.90 0.6 0.78 0.5 0.8 0.66 0.4 0.6 0.50 0.44 0.37 0.3 0.4 0.2 0.2 0.11 0.05 0.1 0 0 es Ge ico Th ny d a de il In tion sia a ed es n ca ico Th zil d a do a sia Fe raz n di di n So atio an an at n F Stat a fri a i i ne ne ex ex Ch Ch In In Br rm ail ail ra St B hA er do M M d d ut In ite ite Un Un ian ia ss ss Ru Ru 1994 2014 Sources: For panel a, calculations based on roads data from the Global Roads Inventory Project v 4 (GRIP4) dataset (https://www.globio.info/download- grip-dataset) (Meijer, Huijbegts, and Schipper 2018); for panel b, calculations based on railway data from World Bank World Development Indicators (http:// datatopics.worldbank.org/world-development-indicators/). Population data for both panels is from World Bank World Development Indicators. Note: In panel a, main roads include both highways and primary roads; roads data are for various years between 2010 and 2015 with population data chosen to match the year of the roads data. O v e r v i e w   13 and operations that link Indonesia’s islands. This poor performance, in turn, translates Policies to realize into wide differences in shipping costs: Indonesia’s urban potential sending a 6-meter container within Basic policy principles for getting more out of Indonesia from Tanjung Priok to Jayapura, urbanization: Augment, Connect, and Target Banjarmasin, and Padang costs approxi- Within the next quarter of a century, Indonesia mately US$1,000, US$650, and US$600, will have reached an advanced stage of urban- respectively, compared with shipping costs ization, with more than 70 percent of its popu- from Tanjung Priok to Guangzhou (China) lation living in towns and cities. As time goes of about US$400. by, the costs of shifting the country’s urban Similarly, domestic air travel remains development trajectory increase because, once extremely concentrated along a few routes built, the urban environment is very difficult to between Jakarta and other major metro change. To sustainably shift Indonesia to a tra- areas such as Surabaya. Although Indonesia’s jectory that brings more prosperity, inclusive- market for domestic air travel is both large ness, and livability, policy makers can follow and growing rapidly, a lack of competition, the basic ACT principles of Augment, Connect, compounded by relatively weak regulatory and Target (box O.2): enforcement, adversely affects safety and ser- vice standards, and is likely to become an •  Augment the coverage and quality of basic increasingly important constraint as the mar- services and infrastructure for all people in ket undergoes ­further growth. all places. BOX O.2  The basic policy principles: Augment, Connect, and Target Three basic policy principles— A ugment , •  Connect refers to enhancing the connections Connect, and Target—can guide the central and between places through investments in trans- subnational governments in Indonesia as they port infrastructure and reforms that increase work to ensure that urbanization brings more the spatial integration of markets for goods, prosperity, inclusiveness, and livability to the services, labor, and capital. It also refers to country. better connecting people to jobs, opportuni- ties, and services within places. By contribut- •  Augment refers to expanding and equalizing ing to the spread of prosperity and facilitating access to high-quality basic services across access to basic services whose provision it all places. Ensuring that all Indonesians have might make sense to centralize in certain loca- access to good-quality health and education tions, connecting helps make the urbanization services, safe water, proper sanitation, and process more inclusive, both within urban other basic services would mean that people places and across urban and rural places. move to urban areas because of the •  Target refers to using customized policies to opportunities these areas offer rather than address stubborn and lingering inequalities, because of the lack of basic services in rural which may persist even if the first two pol- areas. Expanding access to basic services and icy principles of augment and connect were infrastructure in line with the population fully enacted. Taking account of the needs would also reduce the speed at which of lagging regions and groups of people (for negative congestion forces mount as cities example, women and girls, the elderly, and grow in population. As such, augmenting people living with disabilities) whose needs basic service access and local infrastructure require extra consideration in urban plan- provides the foundation for economic ning and design would help ensure that the prosperity and for improving the livability of urbanization process benefits all people in cities. all places. 14  TIME TO ACT •  C onnect to integrate both across and cores and single-district metro areas, ensuring within places. universal access to good-quality basic services •• T arget any places and any people left would also help ease the strong congestion behind. pressures already so evident in these areas. The connect principle matters for rural and These three principles matter in different nonmetro areas because it provides the main ways for different types of place (table O.1).24 channel for such places to benefit from the For all types of place, actions to augment the higher productivity of the multidistrict metro coverage and quality of basic services cores and the single-district metros, contribut- are important to close the gaps between those ing to a more inclusive urbanization process. who do and do not currently have access to The connect principle is also important in such services. Doing so would also raise the facilitating the access of residents in rural livability of all types of place, ensuring that, and  nonmetro areas to more advanced when people move to the metro areas, they do ­ s ervices—for example, hospitals with so not out of distress but out of opportunity. advanced facilities for the treatment of com- Moreover, to the extent that improving the plicated ­illnesses and universities that provide coverage and quality of basic services lifts a world-class tertiary education—that it human capital, such actions would improve might not make sense to provide locally. productivity and therefore prosperity, thus Finally, improving connectivity within multi- contributing to the overall inclusiveness of the district metro areas is especially important for urbanization process. For multidistrict metro their prosperity and inclusiveness, because the TABLE O.1  Importance of the ACT principles to different types of place Principle Type of place Why it matters All types ·  Close gaps in access to basic services between individuals. ·  Elevate human capital as people get better access to services that directly (for example, health and education services) or indirectly (for example, better wastewater management) boost skills and health. Augment ·  Expand infrastructure and basic services in line with population growth to counter growing congestion forces. Multidistrict metro cores; ·  Help to ease strong congestion forces as already reflected in a single-district metro areas lack of affordable housing, severe traffic congestion, and high pollution. Nonmetro urban areas; ·  Integrate these areas with the multi- and single-district metro nonmetro rural areas areas, allowing for the spillover of prosperity and, therefore, a more inclusive urbanization process. ·  Facilitate access to more advanced services in multi- and single- district metros. Connect Multidistrict metro cores; ·  Connect people with jobs and services, creating more integrated single-district metros labor markets and boosting positive agglomeration forces. Urban peripheries; ·  Integrate these areas with multidistrict metro cores, countering rural peripheries the negative effects of residential segregation. Nonmetro rural areas ·  Provide extra assistance to places where the policy principles of augment and connect are not by themselves enough to generate prosperity. Target Multidistrict metro cores; ·  Ensure benefits of urbanization are fully shared with women urban peripheries; and girls, the elderly, and people with disabilities through urban single-district metros; planning and design that take account of the needs of all. nonmetro urban areas O v er v ie w   15 strong residential segregation between high- Expanding subnational financing options and low-skilled households and workers in would increase the envelope for much- metro areas is a source of inequality, as well needed investments in basic infrastructure as of diminished productivity and prosperity. services For the peripheries of multidistrict metro The needs for basic infrastructure and ser- areas, it is also important because empirical vices in Indonesia’s urban areas are enormous. evidence presented in this report shows that As an indication of the costs, a 2015 market enhanced access to domestic markets, which assessment of 14 large Indonesian cities esti- can be achieved through strategic investments mated an overall subnational infrastructure in transport infrastructure, is a significant investment financing gap of US$11.1 billion driver of their underlying productivity.25 (map O.2). Meanwhile, during 2011–13, total Finally, the target principle, to reach lag- infrastructure investment in Indonesia stood ging places through better-designed place- at only about 3–4 percent of GDP, having based policies, is of most importance to fallen from approximately 7 percent for nonmetro rural areas, especially those outside 1995–97 (World Bank 2017b). This level of Jawa-Bali. These areas are most likely to fall investment is unlikely to bridge the urban behind even if the policy principles of aug- infrastructure gap, especially considering that ment and connect are fully implemented. neighboring countries such as China, When it comes to improving urban planning Thailand, and Vietnam were each investing and design to address the needs of women, between 7 and 10 percent of their GDP in girls, the elderly, and people living with dis- infrastructure annually during 2011–13. abilities, however, the target principle applies The potential sources of additional financ- to all types of urban area. It takes on an extra ing to bridge the infrastructure and service dimension in the multi- and single-district delivery gaps depend on the type of place. For metro areas, at least for transport, because multidistrict metro cores, urban peripheries, travel patterns for women in these areas tend and single-district metro areas, considerable to be more complex (Greed and Reeves 2005), scope exists for improving own-source reve- and the risk of social isolation of the elderly nues, especially through property taxes. and people living with disabilities is greatest. Indeed, at 11–12 percent, own-source reve- nues already account for at least double the The key to ACT-ion: Institutional reforms to share of total revenue for these areas than for subnational governance and finance other types of area.26 Even so, political con- Successfully implementing the ACT principles siderations have led Indonesia to focus fiscal requires fundamental reforms to Indonesia’s decentralization on the expenditure side. The system of subnational governance and “money follows function” principle stipulates finance, particularly in three key areas: that subnational governments are mainly responsible for spending, whereas the central •  Expanding options for subnational financ- government generates most of the revenue. ing to meet basic infrastructure and ser- Even in the most urbanized areas, therefore, vice needs the revenue raised through property tax pales •  Building capacity for planning, implement- in comparison to advanced countries such as ing, and financing urban development the United States, where 30 percent of local •• Improving institutional coordination ver- government revenues came from property tax tically and horizontally at all levels of in 2014. Nationwide, property tax revenue is government. equivalent to 0.57 percent of GDP in Again, the exact ways these policy action Indonesia, which is among the lowest for areas apply partly depend on the type of G20 countries (figure O.13). Issues that will place, but their implementation would ulti- need to be addressed by districts to expand mately enable Indonesia’s urbanization pro- property tax revenue include low coverage of cess to deliver more prosperity, inclusiveness, the cadaster, low tax rates for urban areas, and livability. and low collection rates. 16  TIME TO ACT MAP O.2  Indonesian cities face large infrastructure investment financing gaps, 2014 Investment need gap Borrowing capacity Surabaya Revenue (excl. salary, earmarked 2,954 and contingency fund) Batam 825 Pontianak 361 Banjarmasin Balikpapan Bangka 651 339 Makassar 449 860 Semarang 1,262 Gresik Sidoarjo Bogor 875 521 642 Lombok Surakarta Denpasar Barat 279 606 339 Source: Joshi et al. 2015. Note: Investment needs gaps are expressed in millions of U.S. dollars (2014 prices). Beyond property tax, metro areas can FIGURE O.13  Indonesia has lower also draw on alternative sources of own- property tax revenue than most other G20 source revenue through transparent and sus- countries tainable urban development practices. Land value capture and transfers of development France 4.3 rights are instruments that subnational gov- United Kingdom 4.21 ernments could explore for raising infra- Canada 3.54 structure development funds. Using these United States 3.14 types of instruments will, in turn, require Argentina 2.9 building the familiarity of district govern- Japan 2.69 Korea, Rep. 2.62 ment staff with the tools of real estate Australia 2.47 development. China 1.73 Financing investments in urban infrastruc- South Africa 1.39 ture, especially public transport, may also Brazil 1.25 require going beyond resources available from Russian Federation 1.2 the annual budget, particularly in multidistrict Italy 0.81 Germany 0.8 metro areas where the investment requirements Indonesia 0.57 are higher. For example, in a large metro like India 0.48 Jakarta, a mass rail system may be required, but Turkey 0.27 such a system costs 12 times more per kilometer Mexico 0.25 than a bus rapid transit system, which might 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 suffice to meet the transportation needs of a Property tax revenue (percent of GDP) smaller city.27 Currently, subnational govern- ment debt in Indonesia is very low at only Source: Prakash 2013. 0.04 percent of GDP.28 Developing a framework Note: Figure omits the European Union and Saudi Arabia due to missing data. O v er v ie w   17 to gradually empower financially sound subna- services in line with their populations, con- tional governments to access capital markets tributing to the mounting of congestion and official financing would help them to meet forces. their infrastructure needs. The fact that increased resources to less A case can also be made for revamping populous districts have not translated into Indonesia’s fiscal transfer system, which was results suggests that resources may not be the put in place by the “big bang” decentraliza- binding constraint for nonmetro rural areas tion in 2001. The reforms not only shifted when it comes to the provision of infrastruc- resources for service delivery and infrastruc- ture and basic services. Indeed, there may be ture provision from the central to subnational capacity constraints for implementation, governments29 but also skewed the allocation implying that the resources may be more effi- of those resources more toward less populous, ciently directed elsewhere. Gradually moving nonmetro, and more remote areas. This skew- toward a transfer formula that places more ing is a consequence of the transfer formula weight on population could improve the abil- for the Dana Alokasi Umum (DAU, or ity of growing urban areas—most notably, the General Allocation Grant), in which the size urban peripheries of multidistrict metro of the transfer that a district receives is largely areas—to tackle mounting congestion forces independent of its population. Although such without adversely affecting infrastructure and a design would seem to promote improve- service provision in rural areas. ments in basic service delivery and infrastruc- ture in the most lagging areas, regression Building capacity for planning, implementing, evidence presented in this report suggests that and financing infrastructure and urban it has failed to translate into results—less development populous districts favored by the decentral- ization reforms and transfer formula have The failure of fiscal transfers and infra- failed to improve basic services and infra- structure spending in less populous districts structure at a faster pace than less favored, to translate into results suggests that resources more populous districts.30 Districts favored may not be everything. A more important by the post–big bang transfer system have constraint for nonmetro areas in implement- likewise failed to improve their stocks of ing the ACT principles may be related to local roads and access to markets through trans- capacity issues in planning and executing portation networks at a faster pace than less infrastructure projects. More generally, capac- favored districts, which suggests that the more ity needs to be improved across the portfolio intense spatial targeting of nonmetro areas of places in three interrelated areas: has not helped to improve their relative con- 1. The capacity of subnational govern- nectivity and integration with metro areas. ment staff to address issues of urban and They also failed to experience significant regional development improvements in manufacturing firm out- 2. The capacity to use information sys- comes in response to the “windfall” in trans- tems and data for both transparent and fer revenues. Indeed, the extra revenues failed informed decision making to halt an already evident downward trend in 3. The capacity to regulate and monitor firm performance for less populous districts subnational government performance. in relation to more populous districts. The transfer formula also penalizes places Building staff capacity to manage urban and where population growth, driven by urban- regional development issues. With Indonesia’s ization, is occurring. This is because, all else “big bang” decentralization in 2001, adminis- equal, population growth translates into trative and fiscal responsibilities were trans- declining per capita transfers. Unless these ferred quickly from central to subnational places can plug the gaps through increased governments without transitional staff train- own-source revenue, it will be harder for ing. Between 1999 and 2001, the share of them to expand infrastructure and basic civil servants on subnational government 18  TIME TO ACT payrolls increased from just over 12 percent at the university level will also prepare future to almost 67 percent (World Bank 2003). In government staff in these core areas. Financial addition, the total civil service expanded by support for such training could come from 25 percent from 2006–17, from about capacity-building grants to subnational gov- 3.6  million to more than 4.5 million ernments. Moreover, reducing the frequency of (World Bank 2018b). rotation of government staff among depart- Although the quality of the Indonesian ments would create more incentives to invest civil service has also improved because of in skills specific to a department and to allow more meritocratic hiring and promotion for retaining specific capacities. practices (World Bank 2018b), most district- With respect to information systems and level public works departments still have too data, ATR/BPN, together with the National few professionally qualified staff, and the Geospatial Information Agency, is well- quality of available training is questionable placed to take on the responsibility of pro- (World Bank 2012). Many staff are also viding an integrated data and mapping unfamiliar with calculating the fiscal needs platform for spatial planning and analysis. of complex projects. Often, when subna- Meanwhile, districts should integrate cur- tional staff members improve their skills, rently disparate statutory plans into a com- they are rotated to other departments for mon spatial development framework. On political or administrative reasons, leaving tax administration, relevant government capacity gaps behind. agencies could use satellite imagery, Building capacity to use information unmanned drones, and automated methods ­ systems and data. The absence of consistent, to prepare effective cadasters. The central reliable base maps and data impedes produc- government should also support subnational tion of local spatial plans in Indonesia. governments in building and enhancing their Similarly, the lack of access to reliable land current tax information and management records and spatial data inhibits infrastruc- system (Sistem Informasi dan Manajemen ture investments and service delivery. 31 Objek Pajak or SISMIOP). Moreover, there is no common framework to Building capacity to regulate and monitor integrate different plans from different min- subnational government performance . istries and agencies, creating confusion that To  improve regulatory and monitoring undermines urban development and capacity at both the central and subnational ­ investment. levels, the central government could consider A systemic lack of capacity also hampers the following: local taxation, particularly the management of property taxes, which, as argued earlier, are •  Setting up an accountability system requir- an underleveraged source of own-source rev- ing subnational governments to report to enue for metropolitan areas. Subnational gov- the central government on their perfor- ernments lack the expertise, capacity, and mance. This could be in the form of a web- cadastral information from the Ministry of based system to standardize reporting by Agrarian Affairs and Spatial Planning/ subnational governments. National Land Agency (ATR/BPN) to manage •  Having an independent body such as property taxes. the government’s auditor or the relevant Policy options to improve capacity include line ministry verify such reports, following establishing well-designed, on-the-job training the lead of the Local Government and programs to improve data quality and man- Decentralization project.32 agement in urban and spatial planning (World •• Aligning the incentives of local government Bank 2017a), improving geospatial education leaders to deliver better services and provide (through, for example, professional courses, a basis for central government intervention organized with leading Indonesian universi- by conditioning fiscal transfers to subna- ties), and delivering training in tax collection tional governments on the verification of and management. Upstream skill development such performance reporting. O v er v ie w   19 Improving institutional coordination vertically multiplied the number of decision points, and horizontally at all levels of government increasing the complexity of coordinating across different sectors, levels of government, The decentralization of decision making for and, crucially for multidistrict metro areas, providing basic services and infrastructure jurisdictional boundaries. Finding effective has had many benefits. It was, however, solutions to these coordination challenges is almost inevitably accompanied by inertia in central to the successful implementation of the functioning of central public institutions, the ACT policy principles. so that the new decentralized decision-­making More specifically, policy needs to better process coexisted with the pre–big bang cen- address three key coordination challenges tralized decision culture. Even today, the old (box O.3). Overcoming these challenges will central planning process is still in place, with require addressing their underlying origins— annual, five-year, and long-term plans; and a namely, the excessive complexity of decision large fraction of the administrative staff making in Indonesia that has resulted from across the country is centrally managed by the the successive addition of new institutional Ministry of State Apparatus Empowerment layers on top of existing ones. The first step to and Bureaucratic Reform (PANRB). doing this will be to identify redundancies and Although decentralization has served the simplify the country’s governance structure— invaluable function of bringing the manage- an enormous task requiring strong commit- ment and planning of places, especially urban ment, but essential for improving efficiency. places, closer to the people, it has also BOX O.3  The three coordination challenges Challenge 1: Vertical coordination designs, and implements water infrastructure In Indonesia, 32 types of activity are shared development projects in subnational government among the central, provincial, and district gov- jurisdictions. It later hands over these assets to ernments, with each level playing a different role subnational water utilities, but follow-up invest- depending on the sector. Districts play the larg- ment in distribution ­ networks and connections est role in education, health, and infrastructure, may not occur, leading to increased idle capacity whereas the central government plays the largest of the utilities (World Bank 2015). role in general administration, social protection, and housing and public facilities (figure BO.3.1). Challenge 2: Horizontal coordination Sectoral departments at the subnational level, across sectors however, do not map hierarchically to the Urban development is a cross-cutting issue that corresponding national ministries. ­ requires the involvement of many ministries and The consequence is that “everybody’s business agencies. In Indonesia, at least eight ministries is nobody’s business.” Often, neither the central and agencies, each in charge of different sectors, nor the provincial nor the district government need to coordinate on this issue. An intermin- takes responsibility for vertical coordination and isterial steering committee on urban develop- ensuring that activities in each sector deliver the ment under the stewardship of the Ministry of intended results. For example, in urban water, National Development Planning/National the lack of coordination between central, pro- Development Planning Agency, Bappenas, vincial, and district governments and the poor includes representatives from relevant minis- prioritization of subnational government capital tries and agencies. The committee is chaired, expenditure have led to disappointing increases however, by an Echelon 1–level official (a dep- in the number of homes connected to pipes. The uty minister) who reports to the Minister of Ministry of Public Works and Housing plans, National Development Planning, whereas other Box continued on next page 20  TIME TO ACT BOX O.3 Continued committee members report to their respective networks to connect and integrate metro areas, ministers, hampering intersectoral coordination. sewage treatment, storm-water systems, and solid waste treatment facilities. One example Challenge 3: Horizontal coordination of the problem is the increased flooding risk within multidistrict metro areas in Jakarta’s core, which is associated with the Whereas the first two coordination challenges failure to prevent haphazard and often illegal affect all types of place, the third—horizontal development in the upstream water catchment coordination of neighboring jurisdictions— areas of Kabupaten Bogor in the south of metro applies especially to multidistrict metro areas. Jakarta. This contributed to major floods in the This challenge affects the efficient delivery of capital in February 2007, causing more than 70 large-scale infrastructure at the metro level, deaths and displacing 340,000 residents (World including the delivery of urban transport Bank 2008a). FIGURE BO.3.1  Districts account for about half the public spending in education, health, and infrastructure 100 27 25 33 Share of total government 80 54 48 spending (%), 2014 57 8 14 60 16 40 19 8 18 67 59 51 20 35 33 28 0 Education Health Infrastructure General Social Housing and administration protection public facilities Central Province District Source: World Bank 2016. To improve coordination between different a coordinating minister and established by a levels of hierarchy, more clarity on the role for presidential decree—could be an effective way each level of government (central, provincial, to gain traction on reform. This platform district) in delivering infrastructure and ser- would replace the current interministerial vices would help. The Ministry of Home coordinating committee and be similar in Affairs could also empower provincial gov- spirit to, for example, the National Team for ernments to ensure coordination between the Acceleration of Poverty Reduction and the central and district governments. National Coordination Team for the To improve horizontal coordination across Achievement of the Sustainable Development sectors, a national platform on urban trans- Goals. Because it would be a national plat- formation—led by either the vice president or form reporting directly to the president, O v er v ie w   21 it would carry more authority than the exist- development, and to combat severe traffic ing coordinating committee and, therefore, be ­ congestion are needed. Such measures are more effective in bringing about coordination central to making places more inclusive and ­ and elevating the urban agenda more prosperous, countering the residential segrega- broadly.33 tion of high- and low-skilled workers that is Finally, to improve horizontal coordina- undermining the beneficial spillover of ideas tion of neighboring districts—most notably, between workers. within multidistrict metro areas—the central For housing, key actions include scaling up government has several policy options. These current efforts to improve the fluidity of land include, for example, elevating metropolitan and housing markets, as well as to enhance management functions to provincial local technical capacity for coordinated and government(s).34 The options also include spatially appropriate land and housing plan- encouraging greater contracting between sub- ning. This could involve measures to improve national governments to deliver complemen- land administration, the complexity of which tary services and promoting the establishment currently constrains the availability of land in of multigovernment enterprises for specific well-located areas for infrastructure and purposes, such as transport or solid waste housing development. It could also include a disposal.35 review of existing regulations by subnational governments to remove artificial constraints Additional policy actions are required on well-located land, to simplify procedures to ACT and deliver on the promise of for land acquisition, and to accelerate (and urbanization reduce costs) for building registration and permits, and to implement occupancy certifi- Although institutional reforms to subnational cate processes. An integrated housing plat- governance and finance provide the backbone form—like Mexico’s National Housing for implementing the ACT policy principles Registry—could also be considered.36 Such a and, hence, the better leveraging of urbaniza- platform would combine housing and land tion for prosperity, inclusiveness, and livabil- market information to improve accountabil- ity, they are not, by themselves, sufficient for ity and efficiency across all levels of govern- the full implementation of the Connect and ment and support transparent sharing of Target principles. Improved connectivity and market information about housing demand, integration of urban areas—especially metro supply, location, prices, and finance. Finally, areas—also requires additional policy actions measures could be implemented to strengthen at the interface of housing and urban trans- subnational government capacity to develop portation, and connecting places requires and enforce regulatory systems relating to, for additional actions to address regulatory issues example, zoning and construction quality that in transportation markets. Finally, targeting support integrated and resilient urban growth. any places and any people left behind requires Policy makers could also explore scaling a rethinking of Indonesia’s approach to place- based policies and a paradigm shift in the up examples of innovative mixed-use, mixed- country’s approach to urban planning and income development like Vida Bekasi, 37 design. located in Bekasi City in West Jawa, to coun- ter strong residential sorting of workers by income and skill level. Vida Bekasi is a new Enhancing the connectivity and integration sustainable urban development of 15,000 res- of urban areas—especially metro areas—also idents, designed with a master plan that inte- requires specific policy interventions at the grates mixed-use design with place-specific interface of housing and urban transport considerations. It also integrates a public In addition to reforms in governance and transportation system to connect with finance, specific policy actions to ensure a Bekasi’s train station and a bus terminal—and ­ sufficient supply of well-located affordable has a school, shopping center, and market- housing, to promote more mixed-income place. To promote integrated growth more 22  TIME TO ACT generally, subnational governments can guidelines for subnational governments. encourage the development of varied housing These guidelines will help ensure that trans- typologies, such as two-story townhomes, port system designs are based on robust and incremental strategies, and midrises that cater reliable demand forecasts for the medium to to different income (and skill) groups in the long term—and that investments are pro- same development. The supply of affordable posed under a well-defined funding and rental housing could also be increased. financing framework and targeted to the most Finally, reforming housing finance would effective modes of transport for different help the government to target unserved places. A comprehensive transport policy groups and crowd in the private sector. This would outline actions that improve supply requires continuing efforts to improve and manage demand, while discouraging pri- Indonesia’s current credit-linked subsidy pro- vate vehicle use. grams—the Housing Loan Liquidity Facility To manage traffic demand, subnational (Fasilitas Likuiditas Pembiayaan Perumahan, governments in metro areas may want to FLPP) and the Interest Rate Buy-Down revisit Jakarta’s 3-in-1 high-occupancy vehicle Subsidy (Subsidi Selisih Bunga, SSB)—which (HOV) policy. Although it was revoked currently focus on middle-income households because of its unpopularity, the HOV policy and generate high fiscal costs. To the extent was successful in reducing congestion while it that schemes such as FLPP have helped pro- was in place (Hanna, Kreindler, and Olken mote affordable housing in metro areas, it has 2017). A better-designed variant of the policy, often been in their peripheries. One option is if introduced, might prove more long lasting. to use the BP2BT ( Bantuan Pembiayaan The reduction of explicit and implicit subsi- Perumahan Berbasis Tabungan ) savings- dies to private car use can also be considered. linked down-payment subsidy program, For example, in Vienna, reorganizing parking which provides unserved informal and self- spaces in the city, eliminating free parking, built sectors access to subsidies and intends to removing parking from historic places, and crowd in private sector funding and participa- instituting street and parking permissions for tion at market-based pricing. residents were key actions for reducing car Tackling severe traffic congestion, an issue use (Buehler, Pucher, and Altshuler 2017). that particularly afflicts multi- and single-­ In addition, congestion charges, such as those district metro areas, will similarly require a in cities like London, may also form part of coordinated and spatially driven approach to the policy mix for better demand manage- planning and developing transport infrastruc- ment, especially in multidistrict metro cores ture. Such an approach requires strong coor- and single-district metro areas. dination in increasing the supply of To enhance local capacity to plan, operate, infrastructure and services, and the use of and maintain urban transport systems, key demand management instruments to reduce actions along three lines are important: congestion and ensure that the use of private vehicles is priced to correctly reflect their true •  Strengthen transit regulatory and manage- social costs. Key actions include strengthening ment authorities so that they can better the central government’s role in guiding urban coordinate public transport services across transport policy; enhancing local capacity to administrative boundaries. plan, operate, and maintain urban transport •  Introduce measures to improve operations, systems; and promoting transit-oriented including optimized routing and schedul- development. ing, together with actions to prioritize allo- To strengthen the central government’s cation of space to public transport. role, a national urban transport policy—a •• Implement technological improvements subject discussed in the Ministry of that can considerably improve the quality Transportation for several years—needs to be of service, such as smart ticketing or fare brought to a successful conclusion. Such a collection systems, fleet management sys- policy is key to developing transport-related tems, and user information systems. O v er v ie w   23 F i n a l l y, p r o m o t i n g t r a n s i t - o r i e n t e d Connecting places requires additional actions development may involve redeveloping resi- ­ to address regulatory issues in the maritime dential structures or encouraging new build- and air transport markets ings with more vertical development by Ensuring connectivity and integration across permitting higher floor–area ratios, thereby the portfolio of places will also require addi- loosening height restrictions, or allowing tional actions to address regulatory issues in greater density in target zones. Target zones the maritime and air transport sectors. In the can be selected to promote local objectives, maritime sector, domestic connectivity across such as reduced dependence on private island-regions is hampered by, among other vehicles or development of mixed-use, things, the small size of vessels and, conse- pedestrian-friendly areas. Indonesian metro quently, the fact that the fleet is larger than areas could follow the examples of Hong necessary; the low fleet use and low service Kong and Seoul, which have already inten- frequency; and a domestic network that con- sified land use around transit stops sists of single-port-to-single-port routes rather (MGI 2016). than multiport itineraries. Although some policies, such as the Encouraging more private participation in reform of the housing finance programs and maritime transport is crucial to easing these the national urban transport policy, are constraints. Since 1992, all Indonesian com- national policies, many of the detailed pol- mercial ports have been operated by four icy actions for affordable housing and state-owned enterprises, known as Pelindo urban transport will, again, need to be tai- I to IV, each covering a designated region. The lored across the portfolio of places. Multi- 2008 shipping law was intended to change district metro areas, as evidenced by their this by introducing a new public body as the higher inequality and the strong residential Port Authority to regulate the sector and by sorting of workers by skill level, face the the easing of entry of private operators most complicated challenges to successfully through the obligation to concession port connect people with jobs, opportunities, operations. The transition to private opera- and basic services. For these areas, policy tion has been slow, however, with the main makers need to focus on ensuring that hous- issue stalling progress being distortions in the ing is available in both core and peripheral tariff structure. Current tariffs for domestic areas—ideally, in mixed income develop- users appear too low to justify necessary port ments—and that traffic congestion is com- investments. One obstacle to adjusting tariffs batted by using the full range of instruments is the need to consult with all port stakehold- on both the supply and the demand sides. ers, including associations for domestic ship- That includes expanding the supply of high- ping lines, freight forwarders, and importers quality mass transit, implementing parking and exporters. These stakeholders tend to charges and restrictions, and (re-)introduc- resist tariff increases, so policy action is ing HOV and congestion charging policies. needed to redefine the perimeter and attribu- For single-district metro areas, where hous- tion of the regulator—the Port Authority. ing supply constraints may be more binding For air transport, related issues apply. The than demand pressures, a focus on making sector is facing growing demand that capacity serviced land available for development will is unlikely to be able to keep up with, absent be important. In those areas, it may not be significant changes in the sector’s governance necessary to use congestion charging on the aimed at increasing competition and private demand side of traffic management. sector participation. Indonesian airports are Nonmetro areas, where urban population operated by two state-owned companies, growth is strong, will need to ensure that Angkasa Pura 1 (AP1) and Angkasa Pura 2 new development remains connected with (AP2), which have more than 90 percent of the city and that infill development for new seats and act as de facto monopolies in their housing is possible for, and attractive to, the respective territories. On the basis of successful private sector. 24  TIME TO ACT experiences in other countries, the possibility The places most likely to be left behind of allowing private participation in some of the are remotely located nonmetro rural areas main airports could be considered (World outside Jawa-Bali, especially those lacking Bank 2018a). the advantage of natural resources. The evi- dence in this report suggests that, although these places fail to experience other types of Targeting any places and any people left positive agglomeration forces, they do bene- behind requires a rethinking of place-based fit from positive human capital externali- policies and a paradigm shift in urban ties.38 Spatially targeted policies to help these planning and design places should be designed with these results Even if all the institutional reforms and poli- in mind. Rather than using targeted incen- cies discussed above are successfully imple- tives to attract firms to remote regions, mented to provide good, universal coverage which has been shown to have little impact of basic services across a well-connected port- outside Jawa-Bali, the government could folio of places where those places themselves consider focusing more on developing are highly integrated, some places and people human capital. Such a strategy may be com- may still be left behind. Even the most devel- plemented by incentives for firms (through oped countries have places considered “lag- dedicated infrastructure development or ging” and for which governments often deem reduced red tape, for example), but specifi- additional policy measures necessary. cally based on their contribution to the Similarly, even if a place has a well-developed human capital agenda by investing in skills, and extensive public transport system that, in especially skills that align with an area’s theory, provides good connectivity of people comparative and competitive advantages, with jobs, opportunities, and basic services, and creating jobs with significant learning subgroups of the population may still face content. Rigorous but transparent criteria barriers to using the system. This will be the for selecting targeted industries, as well as case if the system is not planned and designed monitoring and evaluating the effectiveness in a way that considers the needs of people of these policies, are important, since their with disabilities or if the system is considered cost can be high. unsafe to use by women and girls—say, Finally, for urbanization to be truly inclu- because of fear of harassment. Challenges sive for women, girls, the elderly, and people with sidewalks and access to hospitals, clinics, living with disabilities, urban planning needs and so forth may likewise constrain the abil- to apply design principles and construction ity of people with disabilities and the elderly standards for public spaces, sidewalks, transit to access services, even if those services are in facilities, and buildings that recognize the principle available. needs of all, not only the few. Within the next quarter of a century, Indonesia’s transition to an urban society will be almost complete. To realize urbanization’s full potential to deliver sustainable gains in pros- perity, inclusiveness, and livability, Indonesia’s policy makers need to ACT now to better aug- ment urban infrastructure and basic services, to connect the country’s portfolio of places and integrate its cities, and to target lagging places and the needs of women, children, the elderly, and those living with disabilities. This transition will require, first and foremost, reforms to Indonesia’s system of subnational governance and finance. It will further require regulatory reforms to key transportation markets, policy actions to improve the supply of well-located affordable housing within urban areas and to tackle traffic congestion, a rethinking of place- based policies with a focus on human capital, and a paradigm shift in urban planning and design to better recognize the needs of all segments of the population. By acting now, Indonesia’s policy makers can help ensure that, when the country celebrates the centenary of its independence in 2045, it does so as a prosperous, inclusive, and livable urban society. O v er v ie w   25 Annex OA Tailored policy options by type of place KEY TO TABLES Time horizon S = Short term (next 2 years) M = Medium term (next 2–5 years) L = Long term (> 5 years) Symbols “✓” in a cell indicates that a recommendation applies to the type of place “↔” in a row indicates that a recommendation applies to all types of place Leading institution(s)/Champion(s) ATR/BPN Ministry of Agrarian Affairs and Spatial Planning/National Land Agency Bappeda Local Development Planning Agency Bappenas Ministry of National Development Planning/National Development Planning Agency BI Bank of Indonesia BIG Geospatial Information Agency BKPR Spatial Planning and Coordination Agency BPK National Audit Agency BPN National Land Agency CMEA* Coordinating Ministry for Economic Affairs CMMA** Coordinating Ministry of Maritime Affairs Core Multidistrict metro core MCIT Ministry of Communication and Information Technology MFI Microfinance Institutions MOF Ministry of Finance MOHA Ministry of Home Affairs MOT Ministry of Transportation MPWH Ministry of Public Works and Housing MSOE Ministry of State-Owned Enterprises OJK Financial Services Authority Perum Perumnas National Housing Development Corporation SDM Single-district metro area SMF Sarana Multigriva Finansial – State-owned secondary mortgage financing company SNGs Subnational governments * Finance; Industry; Trade; Agriculture; Manpower; Cooperation and Small & Medium Enterprises; State-Owned Enterprises; Public Works and Housing; Land and Spatial Planning; Environment and Forestry ** Transportation; Maritime Affairs and Fisheries; Tourism; Energy and Mineral Resources 26  TIME TO ACT TABLE OA.1  Matrix A: Reforms to governance and subnational finance Multidistrict metro Nonmetro Single- Periphery district Leading institution(s) Time BROAD RECOMMENDATION Core Urban Rural metro Urban Rural /champion(s) horizon 1. Expand opportunities for urban financing Own-source revenue: Support vertical development, ¸ ¸ ¸ ¸ ATR/BPN; S–M more collection of property taxes, and use of real estate MOF; MSOE instruments Transfers: Keep supporting through transfers with increasing ¸ ¸ ¸ ¸ MOHA; MOF; M portion of results-based transfers Sectoral ministries Own-source revenue: Allow to keep a larger portion of ¸ ¸ ¸ ¸ MOHA; MOF M–L income tax Borrowing: Increase debt financing from both public and ¸ ¸   ¸ ¸   MOF M–L private sources Private financing: Continue regulatory reforms to encourage ¸ ¸ ¸ ¸ ¸ MOF; Sectoral M–L and support PPPs ministries Transfers: Change transfer formula to place more weight on MOHA; MOF L population 2. Build capacity for urban planning, management, and finance Personnel: Build capacity for planning, geospatial analysis, MOHA; Sectoral S–M project management, and public finance ministries; Universities; Professional associations Monitoring: Establish a system that enables subnational MOHA; MCIT; M governments to report on their outputs and benchmark their Sectoral ministries performance Information system: Strengthen common data and ¸ ¸ ¸ ¸ ATR/BPN; M–L mapping platforms for capital investment and property tax BIG; MOF management Information system: Strengthen common data and mapping ¸ ¸ ATR/BPN; M–L platforms for development control (monitoring of spatial BIG; MPWH plan) Monitoring: Enable local internal audit officers to play a BPK M–L larger monitoring role 3. Enhance institutional coordination Interjurisdictional coordination: Engage actively in ¸ ¸ ¸       Core; Provincial S interdistrict cooperation to deliver complementary services governments; MOHA Interjurisdictional coordination: Take anticipatory       ¸ ¸ ¸ SDM; Provincial S coordination measures with surrounding or nearby districts governments; MOHA Vertical coordination: Work with provincial government to Provincial S–M improve vertical alignment across levels of government governments; Sectoral ministries Intersectoral coordination: Integrate local sectoral plans into ATR/BPN; Bappenas S–M a common data and mapping platform Intersectoral coordination: Establish a national-level Bappenas; CMEA M platform for urban transformation led by the president O v er v ie w   27 TABLE OA.2  Matrix B: Policies for more connected places Multidistrict metro Nonmetro Single- Leading Periphery district institution(s) / Time BROAD RECOMMENDATION Core Urban Rural metro Urban Rural champion(s) horizon 1. Enhance integration within urban areas through operationalization of spatial planning including land management To support alignment of spatial and sectoral plans, Bappeda; SNGs; S consider aligning their preparation process timelines Bappenas; MPWH Scale up current efforts to improve technical capacity at ¸ ¸   ¸ ¸   Bappeda; BPN; M the local level, strengthening data collection and analysis Bappenas; BKPR; for evidence-based decision making in contexts of rapid Perum Perumnas urbanization Continue efforts to clarify property rights and improve ATR/BPN; M efficiency of land-management practices (including Bappeda; reforms to reduce transaction costs) Bappenas; MOHA; SNGs Ensure operationalization of plans by explicitly linking ATR/BPN; BKPR; M socioeconomic development outcomes to spatial planning BSN; MPWH Apply spatially enabled prioritization of city investments ¸ ¸   ¸ ¸   Bappeda; BKPR; M to match funding Bappenas; BPN; MPWH; Perum Perumnas Strengthen local government capacity to develop and Bappenas; L enforce regulatory systems that support integrated growth MPWH; Bappeda Expand use of capital investment planning and asset ¸ ¸   ¸ ¸   Bappeda; MPWH; L management tools to strengthen links between planning Bappenas; SNGs and financing decisions 2. Promote a holistic housing agenda to ensure access to adequate, affordable, and resilient housing with good access to basic urban services Strengthen local capacity to prioritize coordinated and ATR/BPN; S spatially driven actions in the housing sector, including Bappeda; effective land management Bappenas; MOHA; SNGs Strengthen local capacity to accelerate permitting processes MPWH; Bappeda; S and to enforce construction quality and resilience Bappenas; MOHA; standards SNGs Carry out an assessment of rental housing demand, supply, ¸ ¸   ¸ ¸   MPWH; Bappenas S and challenges to explore opportunities for expanding rental markets Expand support to mixed-used, mixed-income solutions to ¸ ¸   ¸ ¸   MPWH; Bappeda; S enhance integration and prevent segregation BKPR; BPN; Perum Perumnas Improve availability of well-located housing through MPWH; Bappeda; M information system and tools for spatial planning ATR/BPN; including location guidelines Perum Perumnas Table continued on next page 28  TIME TO ACT TABLE OA.2 Continued Multidistrict metro Non metro Single- Leading Periphery district institution(s)/ Time BROAD RECOMMENDATION Core Urban Rural metro Urban Rural champion(s) horizon Collaborate across ministries to design incentives for development of land for housing in suitable areas, closest MPWH; MOT; to availability of existing infrastructure and services ¸ ¸   ¸ ¸   Bappenas; M and where exposure to risks are minimized (such as Bappeda tax undeveloped land or constrain subsidies to central locations) MPWH; Bappeda; Support expansion of rental solutions to increase the ¸ ¸   ¸ ¸   BKPR; BPN; M supply of affordable housing for low-income households Perum Perumnas Continue efforts to improve the efficiency of the credit-link MPWH; Bappenas; M housing subsidy MOF; SMF; Banks Jump-start the housing microfinance sector to support the BI; MPWH; OJK; L housing needs of the informal sector SMF; MFI’s/Banks 3. Strengthen involvement of national government guidance of urban transport policy and systematically support a sustainable public transport investment program Develop a National Urban Transport Program to provide MOT; CMMA; clear policy, institutional, investment, and operational Bappenas; S guidelines for sustainable urban transport programs at the Bappeda; BPN; local level MPWH As part of the National Urban Transport Program, MOT; CMMA; systematically strengthen financial support from national MOF; Bappenas; S government to sustainable urban transport projects Bappeda Design incentives (or legal mechanisms) to promote MOT; CMMA; coordination across jurisdictions in metropolitan areas, for ¸ ¸ ¸       Bappenas; S example, through a single management authority Bappeda; MPWH Implement actions to manage demand and reduce explicit MOT; CMMA; and implicit subsidies to private car use (such as congestion ¸ ¸   ¸ ¸   S Bappeda; MPWH pricing) MPWH; MOT; Increase investments in nonmotorized transport and CMMA; Bappenas; S improve links to public transport systems Bappeda MOT; CMMA; Reform public transport to promote high-quality service Bappenas; that fosters city connectivity and integration and reduces M Bappeda; MPWH; private vehicle use intensity BPN Develop financing mechanisms to incentivize local Bappenas; MOT; governments toward coordinated land use and transport ¸ ¸   ¸     CMMA; Bappeda; M planning, coordination of supply and demand-side MPWH policies, and cross-jurisdiction coordination Bappenas; Promote transit-oriented development that encourages Bappeda; BPN; ¸ ¸   ¸     M densification in transport corridors of high demand MOT; CMMA; MPWH; BKPR Note: Leading institutions are shown in boldface. PPP = public–private partnership. O v er v ie w   29 Notes 11. Workers are considered low skilled if they have education below the elementary level, 1. All urban population figures cited in this medium skilled if they have education paragraph are based on data from the United between junior secondary and secondary lev- Nations World Urbanization Prospects: 2018 els, and high skilled if they have education revision database (https://population.un.org​ above the upper secondary level. /­wup/). 12. Similar results hold for PISA math test scores. 2. Based on historical GDP data from the 2018 13. Based on calculations using BPS data, version of the Maddison Project Database house price–to–(median) income ratios for (Bolt et al. 2018) (https://www.rug.nl­ /­ ggdc­ Bandung, Denpasar, and Jakarta are 12.1, /­h istoricaldevelopment/maddison­/­r eleases­­ 11.9, and 10.3, respectively. In contrast, data /­maddison-project-database-2018). from Demographia and Nomura point to 3. A country’s urbanization growth rate is given house price–to–income ratios of 7.7, 5.7, and by the growth rate of the share of its popula- 4.8 for Bangkok, New York, and Singapore, tion that lives in urban areas. For a discussion respectively. of why this rate rather than, for example, the 14. Based on data from the World Bank’s World simple growth rate of urban population, rep- Development Indicators (http://datatopics. resents the most appropriate metric of the worldbank.org/world-development-indicators). pace of urbanization, see box 1.2, chapter 1. 15. Following Health Ministerial Decree 4. Whereas during 1990–2000, Indonesia’s (Kepmenkes) No. 829/1999, a house is urbanization pace exceeded that of China, ­ considered overcrowded if the floor area per Thailand, and Vietnam, by 2010–17, its pace person is less than 8 square meters. lagged all three. 16. A megacity is defined as a city with 10 ­ million 5. Indonesia’s national statistics office, Badan or more inhabitants. Pusat Statistik (BPS, also known as Statistics 17. For more information, see https://www. Indonesia), classifies settlements as urban or tomtom.com/en_gb/trafficindex/. Rankings rural using a composite scoring system that based on 2016 data. considers population density, the structure of 18. In the analysis, congestion was defined as the local economy, and the extent to which a occurring when a journey’s travel time exceeds settlement has certain types of infrastructure the “free flow” travel time by 25  ­ percent, and amenities (see chapter 1). where the “free flow” travel time is the fastest 6. The decompositions of Indonesia’s urban reported travel time, according to the Google ­ population growth are by Wai-Poi et al. (2018). Maps API, at which the journey can be under- 7. The World Bank’s 2009 World Development taken during a typical day (typically, between Report (World Bank 2008b) defines areas of the hours of 2 a.m. and 3 a.m.). Redefining con- incipient urbanization as those in which about gestion so that it occurs whenever a journey’s 25 percent of the population lives in urban set- travel time exceeds the “free flow” time would tlements. Areas with urban population shares result in an approximate fourfold increase in of about 50 percent are at an “intermediate” the estimated costs of congestion. These figures stage of urbanization, and those that are at represent lower-bound estimates of the true an “advanced” urbanization stage have urban costs of congestion because they exclude esti- shares of about 75 percent or higher. mated costs associated with travel time uncer- 8. As described in more detail in chapter 4, real tainty (additional time that must be allocated consumption per capita is calculated by using to a trip to ensure that the vehicle reaches its the ratio of the poverty line of a given prov- destination at a specific time), vehicle operat- ince to the poverty line of Jakarta’s core (that ing costs due to additional starts and stops, is, DKI Jakarta) as a spatial price deflator. carbon monoxide and other emissions, public 9. This finding also holds true for island-regions health impacts from emissions, traffic collision and districts. Indonesia’s island-regions, as damage, and losses due to excess freight trans- defined in this report, are Jawa-Bali, Kalimantan, port time. Crucially, they also exclude indirect Maluku-Papua, Nusa Tenggara, Sulawesi and costs arising from the impacts of traffic conges- Sumatera. tion on both urban form and the overall level 10. As is evident from figure O.6, the bulk of the of urbanization that may constrain the realiza- increase in within-place inequality took place tion of positive agglomeration forces. prior to 2012; since that time, inequality lev- 19. See chapter 4 for a detailed discussion of this els have stabilized. evidence. 30  TIME TO ACT 20. For health facilities, accessibility is rated by worth US$770 million that ran from 2011 to respondents to Indonesia’s Village Potential 2017. Before the project, no reporting or inde- Statistics (PODES) survey as very easy, easy, pendent verification process was mandated on difficult, or very difficult. Households lacking subnational government use of specific pur- “very easy” or “easy” access to a facility are pose funds or compliance of completed work classified as “deprived.” with technical guidelines. Governments only 21. Controlling for access to services and house- submitted financial reports to the Ministry of hold characteristics, children in multidis- Finance stating that the funds had been spent, trict metro cores are taller and heavier for and no technical reporting was required. their age than children in nonmetro rural Because financial reports were submitted areas by 0.21 and 0.15 standard deviations, manually, sharing them among relevant minis- respectively. tries was delayed. The project was designed to 22. See chapter 1 of the report for a detailed dis- improve reporting, transparency, and account- cussion of urban population growth trends ability in the use of specific purpose funds. for the different types of place. 33. This platform would need to include all 23. Similar results hold when comparing relevant ministries such as the Ministry of Indonesia’s densities (per square kilometer of Finance, Ministry of Home Affairs, Ministry of land area) of main roads and railways with Agrarian Affairs and Spatial Planning, Ministry those of other countries. of Public Works and Housing, Ministry of 24. Annex OA sets out the tailoring of detailed Transportation, Ministry of Communications policy actions by type of place. and Information, and the National Geospatial 25. In chapter 3 of the report, it is estimated that Information Agency. Meanwhile, Bappenas a doubling of a measure of access to domestic could be assigned the critical role of technical markets for urban periphery districts is associ- secretariat and hub of the platform. ated with a statistically significant 2.9 percent 34. This move would require a change in the increase in worker productivity in regression overall decentralization hierarchy and Law analysis that controls for a wide range of observ- 23. It would also require specific transfers to able characteristics of both workers and the jobs the provincial governments that take on such they occupy. For rural periphery districts, the functions. Finance could be achieved through effect of access to domestic markets on produc- the voluntary agreement of districts to allo- tivity is significant at the 12 percent level. cate some part of their funding to provinces, in 26. For nonmetro urban areas in 2014–16, own- accordance with the functions being taken on. source revenues accounted for about 5 percent 35. These multigovernment enterprises would of total revenue. The corresponding figures have the constituent district governments on for rural periphery and nonmetro rural areas their boards. Financing could come through were 3.4 percent and 1.2 percent, respectively. a mix of user fees, contributions from mem- 27. Using U.S. case studies, mass rail system costs ber subnational governments, precept pow- are US$104.5/km versus US$8.4/km for bus ers, central transfers, and borrowing. rapid transit (Cervero 2013). 36. For a description of Mexico’s National 28. This figure is based on December 2018 Housing Registry, see Kim and Zangerling Ministry of Finance data. (2016). 29. This is particularly true for transport invest- 37. For more information, see http://vidabekasi​ ment. The share of subnational governments .com/. in total transport investment increased from 38. See chapter 3 of the full report. about 25 percent in the mid-1990s to more than 50 percent in the second half of the 2000s, whereas the central government share underwent an inverse evolution. References 30. See chapter 6 for a full discussion of this Bolt, J., R. Inklaar, H. de Jong, and J. L. van analysis. Zanden. 2018. “Rebasing ‘Maddison’: New 31. For example, partly because of the lack of Income Comparisons and the Shape of Long- base maps, only 90 of 1,400 detailed spatial run Economic Development.” Working Paper plans for targeted priority areas nationwide 10, Maddison Project, Groningen Growth and have been developed and issued. Development Centre, University of Groningen, 32. The Local Government and Decentralization Groningen, Netherlands. https://www­.­rug­.­nl/ project was a World Bank–funded loan project ggdc­/­historicaldevelopment­/­maddison­­/­research. O v er v ie w   31 Bosker, M., U. Deichmann, and M. Roberts. 2018. Meijer, J. R., M. A. J. Huijbegts, and A. M. Schipper. “Hukou and Highways: The Impact of China’s 2018. “Global Patterns of Current and Future Spatial Development Policies on Urbanization Road Infrastructure.” Environmental Research and Regional Inequality.” Regional Science and Letters 13 (6): 064006. Urban Economics 71 (2018): 91–109. MGI (McKinsey Global Institute). 2016. “Housing Buehler, R., J. Pucher, and A. Altshuler. 2017. Affordability: A Supply-Side Tool Kit for “Vienna’s Path to Sustainable Transport.” Cities.” Briefing note, MGI, New York. I n t e r n a t i o n a l Jo u r n a l o f S u s t a i n a b l e Mincer, J. 1974. Schooling, Experience, and Transportation 11 (4): 257–71. Earnings . New York: National Bureau of Cervero, R. 2013. “Bus Rapid Transport Economic Research. (BRT): An Efficient and Competitive Mode Park, J., and M. Roberts. 2018. “A New Typology of Public Transport.” Berkeley Institute of of Districts for Indonesia.” Background paper Urban Regional Development, University of for this report, World Bank, Washington, DC. California. Prakash, P. 2013. “Property Taxes across G20 Dijkstra, L., and H. Poelman. 2014. “A Harmonised Countries: Can India Get It Right?” Oxfam Definition of Cities and Rural Areas: The New India Working Paper Series, Oxfam India, New Degree of Urbanization.” Regional Working Delhi. Paper, Directorate-General for Regional and Wai-Poi, M., H. Alatas, K. Chandrashekar, and Urban Policy, European Commission, Brussels. J. Lain. 2018. “The Different Faces of Urban Ferreyra, M. M. 2018. “Human Capital in Cities.” Indonesia: Recent Urban Trends in Indonesia.” In Raising the Bar for Productive Cities in Latin Background paper for this report, World Bank, America and the Caribbean, edited by M. M. Washington, DC. Ferreyra and M. Roberts. Washington, DC: World Bank. 2003. Decentralizing Indonesia: World Bank. A Regional Public Expenditure Review Report. Ghani, E., A. G. Goswami, and W. R. Kerr. 2016. World Bank Regional Public Expenditure “Highway to Success: The Impact of the Review Overview Report 26191-IND, World Golden Quadrilateral Project for the Location Bank, Washington, DC. and Performance of Indian Manufacturing.” ———. 2008a. Spending for Development: Making The Economic Journal 126: 317–57. the Most of Indonesia’s New Opportunities. ———. 2017. “Highways and Spatial Location Washington, DC: World Bank. within Cities: Evidence from India.” The World ———. 2008b. World Development Report 2009: Bank Economic Review 30 (Supplement 1): Reshaping Economic Geography. Washington, S97–S108. DC: World Bank. Greed, C., and D. Reeves. 2005. “Mainstreaming ———. 2012. “Investing in Indonesia’s Roads: Equality into Strategic Spatial Planning Policy: Improving Efficiency and Closing the Financing Are Town Planners Losing Sight of Gender?” Gap.” Road Sector Public Expenditure Review Construction Management and Economics 2012, World Bank, Washington, DC. 23 (10): 1059–70. ———. 2015. “More and Better Spending: Hanna, R., G. Kreindler, and B. Olken. 2017. Connecting People to Improved Water Supply “Citywide Effects of High-Occupancy Vehicle and Sanitation in Indonesia.” Water Supply and Restrictions: Evidence from ‘Three-in-One’ in Sanitation Public Expenditure Review (WSS- Jakarta.” Science 357 (6346): 89–93. PER), World Bank, Washington, DC. Joshi, R., R. Gayathri, A. Permono, B. Ladda, ———. 2016. “Size of the Public Sector: A. Parasher, R. Hutagaol, and R. Prastiwan. Government Wage Bill and Employment.” 2015. “Consultant Services for Market World Bank, Washington, DC. http://www. Assessment and Operating Framework for the worldbank.org/en/topic/governance/brief/size- Indonesia Regional Infrastructure Development of-the-public-sector-government-wage-bill-and​ Fund: Final Report.” Consultant Report for the -employment. World Bank, CRISIL, Mumbai. ——— . 2 0 1 7 a . “ I n t e r n a t i o n a l B a n k f o r Kim, Y., and B. Zangerling. 2016. Mexico Reconstruction and Development Project Urbanization Review: Managing Spatial Appraisal Document on a Proposed Additional Growth for Productive and Livable Cities in Loan in the Amount of US$50 Million to the Mexico. Washington, DC: World Bank. Republic of Indonesia for a National Urban Lain, J. 2018. “Multi-Dimensional Urban Poverty Development Project (NUDP) (P163896).” in Indonesia.” Background paper for this report, Project Appraisal Document PAD2646, World World Bank, Washington, DC. Bank, Washington, DC. 32  TIME TO ACT ———. 2017b. “Indonesia: Regional Infrastructure ———. 2018c. “People’s Perspectives of Urban D e v e l o p m e n t F u n d P r o j e c t .” P r o j e c t Poverty and Rural-to-Urban Migration.” Appraisal Document PAD1579, World Bank, Background paper for this report, World Bank, Washington, DC. Washington, DC. —— —. 2018a. Indonesia Infrastructure Sector World Bank and DRC (Development Research Assessment Program. Indonesia: World Bank. Center of the State Council of the People’s ———. 2018b. Mapping of Indonesia’s Civil Republic of China). 2014. Urban China: Service. ID: P163712. Analytics of Indonesia’s Toward Efficient, Inclusive, and Sustainable Civil Service. Jakarta: World Bank. Urbanization. Washington, DC: World Bank. O v er v ie w   33 Introduction Indonesia has urbanized rapidly since its country (chapters 1–4). Part 2 turns to an independence in 1945, profoundly reshaping ­ analysis of the main institutional issues and its economic geography and giving rise to a policy areas that, if addressed, would help diverse array of urban places ranging from the urbanization deliver sustainable improve- bustling metropolis of Jakarta to rapidly ments in prosperity, inclusiveness, and liv- emerging urban centers in previously rural ability. It provides policy recommendations to parts of the country. Along the way, Indonesia help Indonesia harness the power of urbaniza- has ascended the international ladder of devel- tion to improve these outcomes by following opment and prosperity. This ascent, however, the basic policy principles to ACT (box I.1):4 has been relatively slow; today, even with •  Augment the coverage and quality of local more than half its population living in towns infrastructure and basic services for all and cities,1 Indonesia remains a lower-middle- people in all places. income country.2 It continues to be a country •  C onnect to improve integration both of deep inequalities, both among people across and within places. within cities and among people across differ- •• Target any places and any people left ent places and islands. Many of its cities are behind. grappling with the livability challenges of high traffic congestion and pollution, inadequate Chapter 1 focuses on the description of pat- supplies of affordable housing, and gaps in terns of urbanization and structural transfor- access to water, sanitation, and quality educa- mation. It introduces the “portfolio of places,” tion and health s­ ervices, while also facing seri- a concept fundamental to this report. This ous and increasing risks of natural disasters.3 concept captures the wide diversity of urban Drawing on a wide range of data sources, areas in Indonesia, ranging from large metro- part 1 of this report describes the progress politan areas that cover multiple districts of urbanization and structural transfor- (“multidistrict metro areas”) to rapidly mation within Indonesia and documents emerging urban settlements in otherwise whether urbanization has delivered pros- largely rural areas. Chapter 2 assesses whether perity, inclusiveness, and livability for the urbanization has been delivering prosperity, 35 BOX I.1  The basic policy principles: Augment, Connect, and Target This report emphasizes three basic policy even if the previous two policy principles were ­principles—A ugment, Connect, and Target— enacted, would still leave some places lagging that can guide both the national government in their inclusion in the prosperity and livabil- and subnational governments in Indonesia as ity benefits of urbanization. It further includes they work to achieve sustainable enhancements customizing urban planning and design to in the prosperity, inclusiveness, and livability take account of the needs of women and girls, outcomes that arise from urbanization. the elderly, and people living with disabilities. •  Augment refers to augmenting and equalizing Together, the ACT principles form a logical hier- the coverage and quality of basic services and archy that suggests a clear sequence—Augment, local infrastructure across all places. a Key then Connect, and then, if prosperity and livabil- to this principle are institutional and policy ity are still lagging, Target—even if policy actions reforms in urban and subnational governance to implement the principles proceed in parallel. and financing. Augmenting fosters positive Successfully implementing the principles to ACT agglomeration forces and reduces the speed at will also require policy actions and institutional which negative congestion forces, particularly reforms that are tailored by place. This report those associated with the pressure of urban therefore distinguishes among four types of populations on basic services and infrastructure, place—multidistrict metro areas, s ­ ingle-district mount as cities grow in size. Augmenting metro areas, nonmetro urban areas, and non- basic service access and local infrastructure metro rural areas. For multidistrict metro areas, provides the foundation for urban and national it further distinguishes their cores from both prosperity and for improving the livability their urban and their rural peripheries. These of cities. different types of place face different challenges. •  Connect refers to enhancing the connections For example, a large multidistrict metro area between places (including between urban and such as Surabaya needs to grapple with issues rural places) through investments in transport of horizontal coordination across jurisdictional infrastructure and regulatory and other reforms boundaries in the provision of basic services. that increase the spatial integration of markets Those issues are largely absent for a  single-­ for goods, services, labor, and capital. It also district metro area such as Palembang or a refers to better connecting people to jobs, nonmetro urban area such as Kota Palangkaraya. ­ opportunities, and services within places. By Conversely, Kota Palangkaraya may not have the helping spread prosperity more widely, same own-source revenue-raising potential as connecting makes the urbanization process more Palembang or Surabaya. inclusive, both within places and across places. •  Target refers to using customized policies to a. In this report, we define local infrastrucure, as associated with the Augment policy principle, to exclude transport and information and communications address stubborn and lingering inequalities, technology infrastructure. We instead associate both those types of infrastructure including persistent spatial inequalities that, with the Connect policy principle. inclusiveness, and livability, including by deliver sustainable improvements in prosper- benchmarking Indonesia’s urban performance ity, inclusiveness, and livability, where many against that of other countries. Chapters 3 of these policies are tailored according to the and 4 analyze the underlying drivers of urban type of place. Chapter 5 diagnoses key issues and subnational differences in prosperity and in Indonesia’s system of urban and sub­ the drivers of inclusiveness both within cities national governance and financing, and rec- and across places. ommends policy actions that can provide the Part 2 discusses policies, based on the guid- foundations for augmenting and equalizing ing ACT principles, to help urbanization the coverage and quality of basic services and 36  TIME TO ACT local infrastructure across urban and subna- Agglomeration forces include the beneficial tional areas. These same policy actions also effects for productivity from clustering people provide necessary preconditions for successful and firms within cities and from the connec- policies that connect and target. Chapters 6 tivity of a city to the surrounding markets of and 7 then provide tailored policy recommen- other cities—and to international markets. dations for achieving better connectivity and Congestion forces arise from the pressure of a integration both across urban and subnational city’s population on urban infrastructure, areas and among people, firms, and basic ser- basic services, land and housing markets, and vices within cities. Chapter 6 focuses on trans- the environment, as well as from the increased port infrastructure policies and investments exposure to natural disasters that often that connect, and thus integrate, the portfolio accompanies urban growth. ACT principles of places, with some discussion also of poli- create the conditions to “optimize” the inter- cies that relate to barriers to geographic labor play of agglomeration and congestion forces mobility. Chapter 7 shifts to constraints on to help urbanization deliver sustainable the provision of affordable housing, policies improvements in prosperity, inclusiveness, for urban transport, and the need for more and livability (see the section below on effective urban and spatial planning in cities. “A framework for assessing urbanization”). All are fundamental to achieving more con- nected and integrated growth within cities. Even if all the policy recommendations in Defining prosperity, chapters 5–7 are implemented, additional tar- inclusiveness, and livability geted policy measures may still be required to As discussed, this report focuses on three ensure the full inclusion of all places and all key outcomes—prosperity, inclusiveness, people in the benefits of urbanization. For and livability—in considering Indonesia’s places, this may include remote areas and urbanization performance and policies for islands; for people, it may include women improving it: and girls, the elderly, those with disabilities, and other groups that are disadvantaged or •  Prosperity denotes urbanization’s contribu- discriminated against. Chapter 8 thus dis- tion to material well-being and the absence cusses targeted, place-based policies to boost of national and local material poverty, the fortunes of lagging places and additional where material well-being is driven, to a urban planning and design considerations large extent, by productivity. Throughout important in ensuring that basic services and the report, we associate prosperity with the opportunities of urbanization are accessi- variables such as gross domestic product ble to all. per capita, per capita consumption, and In addition to its eight chapters, the report whether a household is in poverty, vulner- includes four spotlights on strengthening the able to poverty, or part of the middle class. disaster resilience of Indonesian cities, the We measure productivity using such vari- nexus between urbanization and human capi- ables as total factor productivity for firms tal, the “invisible” crisis of wastewater man- and nominal wages for workers. agement, and the potential for smart cities in •  Inclusiveness refers to how well prosper- Indonesia. Although these topics—especially ity is shared among different types of peo- disaster resilience and the role of human ple living in different places, within and ­ capital—are woven into the report’s eight among places. We consider urbanization chapters, their importance also merits sepa- to be inclusive when all people in all rate discussion. places share in the material well-being The policy areas considered in the report benefits that it generates. To be inclusive, influence the prosperity, inclusiveness, and liv- the benefits of urbanization must spill ability of Indonesia’s urbanization process over from cities to rural areas, spread through their impacts on the agglomeration equally across men and women of all and congestion forces that cities give rise to. ages, and be accessible to people living I n troductio n   37 with disabilities and those who might nonmonetary aspects of welfare. The focus on otherwise be discriminated against. the three outcomes also aligns the report with •• Livability indicates the quality of life of a the Sustainable Development Goals and the city beyond the level of material well- New Urban Agenda (box I.2). being that it generates, and it is thus related to the nonmonetary dimensions of welfare. In a livable city, residents and A framework for assessing workers enjoy access to high-quality basic urbanization services—including access to clean water and sanitation and to high-quality health Consistent with basic theory in urban and and education services. A livable city also spatial economics,5 this report views urban- has affordable housing and a relative lack ization’s prosperity, inclusiveness, and livabil- of traffic congestion, pollution, and ity outcomes as jointly determined by the crime. interaction of agglomeration and congestion forces (figure I.1, which also provides the By focusing on the three outcomes of structure for this report). In turn, augment, ­ ­ prosperity, inclusiveness, and livability, the ­connect, and target provide the basic high- report takes a broad view of urbanization’s level principles to guide policy in managing performance, covering both monetary and these forces to deliver better outcomes. BOX I.2  Consistency of this report with the Sustainable Development Goals and the New Urban Agenda Adopted by the United Nations General Assembly at the Habitat III conference in October 2016 in 2015 as part of the wider 2030 Agenda and endorsed by the United Nations General for Sustainable Development, the Sustainable Assembly in December 2016. By “redressing the Development Goals (SDGs) are a set of 17 way cities and human settlements are planned, interlinked goals designed to meet the urgent designed, financed, developed, governed and environmental, political, and economic chal- managed,” the New Urban Agenda aims to help lenges facing the world. Goal 11 of the SDGs— countries achieve a shared vision of “cities for Sustainable Cities and Communities—includes all,” defined as 10 targets for countries to meet by 2030 or the equal use and enjoyment of cities and before. These t ­argets include ensuring access for ­ h uman settlements, seeking to promote all to adequate, safe, and affordable housing and inclusivity and ensure that all inhabitants, of basic ­services; ­providing access for all to safe, present and future generations, without dis- affordable, a­ ccessible, and sustainable transport crimination of any kind, are able to inhabit systems; enhancing inclusive and sustainable and produce just, safe, healthy, accessible, urbanization and capacity for participatory, inte- affordable, resilient, and sustainable cit- grated, and sustainable human settlement plan- ies and human settlements to foster prosper- ning and management; significantly reducing ity and quality of life for all (United Nations both the human and ­ economic losses arising from Resolution 71/256). natural disasters; reducing the per capita environ- mental impacts of cities, with special attention With its focus on policy actions to help to air quality and waste management; provid- urbanization deliver sustainable improve- ing universal access to safe, inclusive, accessible, ments in prosperity, inclusiveness, and livabil- green, and public spaces; and strengthening links ity, this  report provides a road map to help between urban and nonurban areas. Indonesia achieve many of the SDGs, including Following fast on the heels of the adoption of SDG 11, and realize the New Urban Agenda’s the SDGs, the New Urban Agenda was adopted vision. 38  TIME TO ACT FIGURE I.1  A framework for assessing urbanization Agglomeration Prosperity Augment OUTCOMES forces URBAN Inclusiveness Connect GROWTH Congestion forces Livability Target PART 1: Chapters 1–4 PART 2: Chapters 5–8 Source: World Bank. As an urban area grows in population— forces benefit the urban area they favor, but at whether a multidistrict metro area such as the expense of the national economy. Jakarta, a single-district metro area such as If agglomeration forces operated without Samarinda, or a newly emerging urban settle- hindrance, they would encourage all people ment in an otherwise rural area—the growth and economic activity to locate in a single gives rise to both agglomeration and conges- city. Because of congestion forces, however, tion forces. Agglomeration forces can arise this clustering does not occur.6 These conges- from a variety of sources and include “tradi- tion forces arise from the pressure of an urban tional” agglomeration economies associated area’s population on its local infrastructure, with urban size and density, human capital its supply of basic urban services, its land and externalities associated with an urban area’s housing markets, and the environment. stock of human capital (how skilled the area’s Therefore, if infrastructure and basic urban workforce is), and market access effects asso- services are not augmented as a city grows, ciated with an urban area’s connectivity to their quality and per capita availability tend both domestic and international markets to deteriorate. If housing fails to expand to (box I.3). Agglomeration forces can also arise accommodate a city’s growth, it tends to because urban population growth spreads the become increasingly unaffordable, giving rise fixed costs of new infrastructure—including to overcrowding and slums, where housing is infrastructure for basic urban services and less able to withstand climate events because amenities (transportation, utilities, solid waste of poor-quality construction and is often management, health and education facilities, located in areas more prone to flooding and and others)—over a larger number of people, natural disaster.7 And, if pressure on the envi- lowering the average cost of provision ronment is not appropriately managed, air (Armstrong and Taylor 2000). In general, quality can deteriorate, contributing to respi- agglomeration forces increase the productiv- ratory illnesses and global climate change.8 ity of both an urban settlement and a national Agglomeration forces tend to boost the economy. They can also arise for more nega- prosperity of an urban area (and, where such tive reasons—because of policy distortions forces are positive, that of the national econ- that bias the spatial allocation of resources omy), but congestion forces tend to dampen either toward certain cities or, more generally, prosperity, in large part by constraining urban toward urban areas at the expense of rural density and, therefore, the realization of posi- ones. For example, some cities may be favored tive agglomeration forces. Congestion forces on noneconomic grounds in the provision of also impair, obviously and very directly, an infrastructure or basic services—or firms may urban area’s livability. benefit from the lobbying power that comes Although the effects of agglomeration and from locating in the same city as the national congestion forces on prosperity and livability government. Such negative agglomeration are clear, their effects on inclusiveness, as I n troductio n   39 BOX I.3  Sources and types of positive agglomeration forces Urban economists have identified three main, ideas among people. In doing so, it also closely interrelated, and largely overlapping hypothesizes that ideas are more likely to spill theories of positive agglomeration forces that over from more highly skilled workers than contribute to local and national prosperity: from less-skilled workers. That leads to the agglomeration economies, human capital exter- prediction that a worker’s individual nalities, and market access. productivity will be an increasing function of the average human capital of the city where •  Agglomeration economies. According to this she or he works (Moretti 2004; Rauch 1993). theory, metro and urban areas generate higher •  Market access. The third theory is that metro underlying productivity, and therefore and urban areas can generate higher produc- prosperity, than rural areas because of the tivity and prosperity because they also tend positive externalities created by their large to benefit from greater access to large con- population sizes and/or densities (Duranton sumer markets and to supplier markets for and Puga 2004; Jacobs 1969; Marshall 1890). intermediate inputs. This superior access Agglomeration economies can arise through stems from a city’s own “internal” market several mechanisms. The “thick” labor and from its connectivity to other surround- markets that characterize cities can generate ing areas and cities, including, potentially, better matches between workers and firms, so cities overseas through international gate- that each person is more likely to find his or ways such as ports. Greater consumer and her “perfect” job. Cities can also provide the supplier market access makes it easier for conditions for the growth of a large and firms to cover the fixed costs of setting up diversified array of specialized suppliers of shop, thus stimulating increases in profits goods and services, which provide the and productivity (Fujita, Krugman, and intermediate inputs that help fuel the growth Venables 1999; Krugman 1991a, 1991b; of the local economy. And the geographic Krugman and Venables 1995). Again, this proximity of people and firms within cities theory is closely related to the theory of can give rise to the (often unintended) agglomeration economies in that it focuses spillover of ideas, as workers learn from each on a specific (sub)set of mechanisms that other through observation and interaction. allow positive agglomeration forces to arise. •  Human capital externalities. The second It therefore shares with agglomeration econ- theory is that metro and urban areas can omies the hypothesis that a larger “­internal” generate higher productivity and prosperity market aids city productivity by stimulating not so much because of their size or density the growth of a large and diversified array of but because they tend to have more human specialized suppliers of intermediate goods. capital. That is, more educated and skilled It then goes beyond this idea by also empha- workforces generate positive human capital sizing connectivity to the markets of other externalities. In many ways, this theory can be surrounding areas and cities. considered a special case of the theory of agglomeration economies, which emphasizes Chapter 3 provides empirical evidence on the several channels for agglomeration to increase strength of these different types of positive a city’s productivity. The theory of human agglomeration forces in Indonesia. capital externalities, by contrast, focuses on just one of these channels—the spillover of Source: Based on Roberts 2018. defined in this report, are less obvious. widen gaps in prosperity across places. By encouraging economic activity to agglom- Similarly, because the more skilled tend to be erate in a single city, agglomeration forces, more likely to migrate, agglomeration forces where they exceed congestion forces, can also tend to concentrate more highly skilled 40  TIME TO ACT workers in larger cities, widening the inequal- The report is also expected to be a resource ities within cities.9 The key to sharing the and reference for academic and nonacademic gains of agglomeration, and thus of urbaniza- researchers with an interest in Indonesia’s tion, is for policies to ensure the mobility of urbanization and for the public at large. people, goods, and firms across the portfolio Although the report has been written as an of places—as well as the upward economic integrated whole, different parts of the report mobility of people, especially the poor, within are likely to be of more interest to some audi- urban areas. ences than to others. All these considerations give rise to the For high-level policy makers, the report’s ACT policy principles. Augmenting implies overview contains its main policy messages. converging to a higher level of effective access For Echelon 1 officials, in addition to the to basic services, where effective access covers overview, the chapters of part 2, along with both the availability of a service and its qual- the spotlights, are likely to be of deepest ity, across all places and without any spatial interest. For technical staff below Echelon 1, ­ biases in resource allocation. Ensuring effec- chapters 1 and 2 in part 1 provide essential tive access avoids negative ­ agglomeration background on patterns of urbanization and forces and encourages urban growth in the structural transformation in Indonesia and “right” locations. It also requires creating the current performance of urbanization on the framework and conditions to ensure that the key outcomes of prosperity, inclusiveness, urban areas growing in population have the and livability. For diehard technical readers, financing, capacity, and incentives to expand chapters 3 and 4 leverage econometric tech- their infrastructure and supply of basic niques to provide empirical evidence on the services in line with their growth, thus mitigat- ­ underlying drivers of urban and spatial pros- ing the “choking” effects of mounting conges- perity and the inclusiveness of urbanization. tion forces. Meanwhile, connecting among Each chapter begins with a summary of its and within cities and places allows the benefits key messages. of urbanization to be widely shared, promot- ing inclusiveness. And customizing urban planning and design to take account of the Notes needs of women and girls, the elderly, and 1. In 2018, 55.3 percent of Indonesians lived in those living with disability, and targeting officially classified urban settlements, accord- places still left behind—even if the policy prin- ing to data from the United Nations World ciples of augment and connect are effectively Urbanization Prospects: 2018 revision data- implemented—further enhances inclusion. base (https://esa.un.org/unpd/wup/). 2. For more information, see World Bank data at https://datahelpdesk.worldbank.org​ A reader’s guide to this /​knowledgebase/articles/906519-world-bank​ report -country-and-lending-groups. As chapter  1 documents, in 1950, Indonesia’s average gross This report has been written in collabora- domestic product per capita was 7.7  percent tion with, and primarily for, the govern- of that of the United States. By 2016, its ment of Indonesia. Its primary audiences are average GDP per capita had increased to high-level policy makers in the national and 19.8 percent—an improvement, to be sure, but local governments in Indonesia and tech- still far short of the world’s largest economy. nical staff across government ministries, 3. See chapters 2–4 of this report and spotlight 1 on strengthening the disaster resilience of including, among others, the Ministry of Indonesian cities. National Development Planning / National 4. The ACT policy principles are partly inspired Development Planning Agency (more com- by the “3 I” (institutions, investments, and monly known as Bappenas), the Ministry of interventions) policy framework of the World Finance, the Ministry of Public Works and Bank’s World Development Report 2009: Housing, the Ministry of Home Affairs, and Reshaping Economic Geography. The con- the National Board for Disaster Management. cept of the “portfolio of places,” which this I n troductio n   41 report draws on, was also first introduced in Fujita, M., P. Krugman, and A. J. Venables. 1999. the World Development Report 2009. The Spatial Economy: Cities, Regions, and 5. See, for example, Fujita, Krugman, and International Trade . Cambridge, MA: MIT Venables (1999); Henderson (1974, 1982); Press. and Krugman (1991a, 1991b), among many Henderson, J. V. 1974. “The Sizes and Types of others. Cities.” The American Economic Review 64 (4): 6. Locational preferences also play a role in pre- 640–56. venting the clustering of all people and eco- ———. 1982. “Systems of Cities in Closed and nomic activity in a single city—for example, Open Economies.” Regional Science and Urban people may prefer to live in a certain location Economics 12 (3): 325–50. because it is close to the beach, even if wages Jacobs, J. 1969. The Economy of Cities. New York: in that location are lower than elsewhere. Random House. 7. For example, the 2007 floods in Jakarta Krugman, P. 1991a. Geography and Trade . disproportionately affected those living in Cambridge, MA: MIT Press. slums—see spotlight 1 on strengthening the ———. 1991b. “Increasing Returns and Economic disaster resilience of Indonesian cities. Geography.” Journal of Political Economy 8. Congestion forces include both negative 99 (3): 483–99. externalities such as traffic congestion and Krugman, P., and A. J. Venables. 1995. pollution and more general “crowding “Globalization and the Inequality of Nations.” effects” in land, housing, and other markets Quarterly Journal of Economics 110 (4): associated with an insufficiently elastic sup- 857–80. ply of provision. Marshall, A. 1890. Principles of Economics . 9. See, for example, Ferreyra and Roberts London: Macmillan. (2018) on this point. Moretti, E. 2004. “Human Capital Externalities in Cities.” In Handbook of Regional and Urban Economics, Volume 4: Cities and Geography, References edited by J. V. Henderson and J.-F. Thisse, 2243–91. Amsterdam: Elsevier. Armstrong, H., and J. Taylor. 2000. Regional Rauch, J. E. 1993. “Productivity Gains from Economics and Policy, 3rd Edition. Oxford, Geographic Concentration of Human Capital: U.K.: Wiley-Blackwell. Evidence from the Cities.” Journal of Urban Duranton, G., and D. Puga. 2004. “Micro- Economics 34 (3): 380–400. Foundations of Urban Agglomeration Roberts, M. 2018. “The Empirical Determinants Economies.” In Handbook of Regional of City Productivity.” In Raising the Bar and Urban Economics, Volume 4: Cities and for Productive Cities in Latin America and Geography , edited by J. V. Henderson and the Caribbean , edited by M. M. Ferreyra J.-F. Thisse, 2063–117. Amsterdam: Elsevier. and M.  Roberts, 89–115. Washington, DC: Ferreyra, M. M., and M. Roberts, eds. 2018. World Bank. Raising the Bar for Productive Cities in Latin World Bank. 2008. World Development Report America and the Caribbean. Washington, DC: 2009: Reshaping Economic Geography. World Bank. Washington, DC: World Bank. 42  TIME TO ACT PA R T 1 Indonesia’s Urban Trends and Performance S ince gaining independence in 1945, Chapter 2 then assesses whether urbaniza- Indonesia has urbanized rapidly overall, tion has been delivering prosperity, inclu- resulting in a profound reshaping of its siveness, and livability, including by economic geography and the associated emer- benchmarking Indonesia’s urbanization gence of a rich “portfolio of places.” These performance against that of other countries. places range from the bustling metropolis of Chapter 3 analyzes the underlying drivers Jakarta to rapidly emerging urban centers in of urban and subnational differences in hitherto largely rural parts of the country. prosperity, and chapter 4 examines the driv- Chapter 1 describes patterns of urban- ers of inclusiveness both within and across ization and structural transformation. places. Patterns of Urbanization and Structural Transformation 1 KEY MESSAGES For much of the time since its independence, century, this growth has been largely driven by Indonesia has been a country of rapid urbaniza- two very different urbanization dynamics. The tion, resulting in a rich “portfolio of places” that first is rapid population growth in the already range from large metropolitan areas spanning urban peripheries of the major metro areas, multiple districts to new urban areas in otherwise with migration making a notable contribution. rural districts. Jakarta, with a population almost The second is a process of largely in-place four times that of the next-largest metropolitan urbanization in predominantly rural areas, area in the country, stands out. In recent decades, leading to the reclassification of settlements however, Indonesia’s urbanization has slowed to from rural to urban. a pace more consistent with the country’s overall •  Indonesia possesses a rich portfolio of level of urbanization. Urban population growth places—with urban and rural areas of has been driven more by natural population different sizes and economic characteristics: increase and the reclassification of previously ŊŊ Large metropolitan areas that span multiple rural settlements than by rural–urban migration. districts and that consist of both a core and a periphery (multidistrict metros), by far •  Between 1960 and 2000, Indonesia was one the most populous of which is Jakarta of the fastest-urbanizing countries in East ŊŊ Metropolitan areas that are contained Asia and the world, outstripping even China. within the administrative boundaries of a Since the turn of the century, however, single district (single-district metros) Indonesia’s urbanization has slowed to a rate ŊŊ Nonmetro areas, consisting of both more typical of a country at its level of predominantly urban and predominantly urbanization. rural nonmetro districts •  Although Indonesia’s overall urban population •  The cores of multidistrict metros are more has continued to grow since the turn of the dense, built up, service oriented, and 45 productive than their peripheries. Although leading to the reclassification of rural this is the case for all multidistrict metros, it is settlements as urban. particularly so for Jakarta. Such relative riches •  By 2045, when Indonesia will celebrate the have not, however, been sufficient to stimulate centenary of its independence, more than positive net migration into the cores: migrants 70 percent of its people will live in towns have preferred to settle in the urban and cities. How Indonesia manages urban- peripheries of multidistrict metro areas. Also, ization will go a long way to determining the pace of transformation has been fastest in whether it reaches the upper rungs of the more rural areas, which are experiencing a global ladder of prosperity, inclusiveness, strong process of urbanization in place, and livability. When Indonesia gained its independence in improvement to be sure, but still far short of 1945, only one in eight people lived in towns the world’s largest economy. Even with more and cities, and the country’s entire urban pop- than half of its population living in towns and ulation stood at about 8.6 million, roughly cities, Indonesia remains a lower-middle- equivalent to that of London today.1 In the income country. Moreover, although virtually more than 70 years since then, Indonesia has everyone has benefitted in absolute terms, the undergone a remarkable urban transforma- relative gains from urbanization have been tion. Today, nearly 151 million people uneven across the country and within places (more than 55 percent of the population) live (see chapters 2–4). In addition to the gains in towns and cities, almost 18 times the popu- from agglomeration forces, the unprecedented lation of London.2 This transformation has growth of cities over the last seven decades given rise to a diverse and vibrant collection has given rise to negative congestion forces of urban places, ranging from Jakarta—one that policy has failed to adequately address— of the world’s largest megacities, with a popu- or, in some cases, has even exacerbated—and lation exceeding that of many countries3—to that are undermining the prosperity, inclusive- rapidly emerging urban settlements in hith- ness, and livability of places. erto largely rural areas. Part 1 of this report examines patterns of As Indonesia has urbanized and trans- urbanization and structural transformation formed its economy, so too has it climbed the within Indonesia. This chapter describes how international ladder of development and rapid urbanization until the turn of the century prosperity. Since 1950, average gross domes- reshaped the country’s economic geography. It tic product (GDP) per capita has increased also shows how, in recent decades, Indonesia’s almost ninefold in real terms, and the average overall pace of urbanization has slowed as Indonesian today enjoys a standard of living urban population growth has been driven far surpassing that of previous generations.4 more by natural population growth and rural The strong productivity-enhancing agglomer- transformation, leading to the reclassification ation forces that cities generate have helped of previously rural settlements, than by migra- propel Indonesia’s climb up the ladder (see tion from the countryside. Underlying this chapter 3 for estimates of the strength of these overall picture are two very different urbaniza- agglomeration forces).5 tion dynamics. The first is rapid population All the same, however, the climb has been growth in the already urban parts of the slow and arduous compared with the rapid peripheries of Indonesia’s major metro areas pace of urbanization. Whereas in 1950, (especially Jakarta), to which migration is mak- Indonesia’s average GDP per capita stood at ing a notable contribution. The second is a pro- 7.7 percent of that of the United States, by cess of in-place urbanization in predominantly 2016, it had increased to 19.8 percent—an rural areas that is being driven by natural 46  TIME TO ACT population growth and the transformation and This definition is highly sophisticated. reclassification of rural settlements as urban. Whereas other countries may use one or more Chapter 2 considers whether these patterns of of the criteria Indonesia employs—population urbanization are delivering the desired out- density, structure of the local economy, exis- comes of prosperity, inclusiveness, and livabil- tence of certain types of urban infrastructure, ity. Chapter 3 analyzes the underlying drivers and the presence of certain urban amenities— of urban and subnational differences in pros- Indonesia combines them using a unique perity, and chapter 4 examines the drivers of composite scoring system. inclusiveness both within and across places. Recent urbanization trends Measuring urbanization in in Indonesia Indonesia This section looks at the level and pace of Before analyzing patterns of urbanization and urbanization for Indonesia overall and for its related trends of structural transformation island-regions. It also looks at the contribu- within Indonesia, it is important to under- tions to urban population growth made by stand how urbanization is defined and mea- natural growth of the population within sured. Official definitions of urban areas vary urban areas, net rural–urban migration, and markedly across countries (Ellis and Roberts reclassification of rural settlements as urban. 2016; Roberts et al. 2017; World Bank 2008). Note that reclassification will be triggered Whereas some countries, such as Argentina, when a settlement records a higher score on have definitions that rely on a single criterion, BPS’s composite index by virtue of it becom- such as settlements that exceed a certain ing more urban in character (because of some population threshold, others employ more ­ combination of it becoming more densely complex definitions.6 And, whereas some defi- populated, having experienced transforma- nitions are precise, others are vague. Many tion away from agriculture, and having countries have no explicitly stated criteria to acquired more infrastructure and “urban” identify urban areas but either list their cities facilities). Reclassification thus reflects a real, by name or designate administrative units on-the-ground process and not just the that constitute cities (Roberts et al. 2017). bureaucratic relabeling of a settlement. Some countries even have multiple definitions of urban, such as a statistical definition for enumerating the urban population in the Indonesia is at an intermediate level of national census and alternative definitions for urbanization overall, but with considerable determining whether a settlement is governed variation across island-regions as an urban or rural unit and for making With an urban share of the population in planning decisions. 2017 of almost 55 percent, Indonesia has Indonesian law recognizes three definitions reached an intermediate level of urbaniza- of urban areas: a statistical definition estab- tion.9 Thus, although more highly urbanized lished by Badan Pusat Statistik (BPS, also than other developing East Asia and Pacific known as Statistics Indonesia), and two plan- countries, such as the Philippines, Thailand, ning/legal definitions set by the Ministry of and Vietnam, Indonesia still lags the level of Home Affairs and the Ministry of Agrarian urbanization in the developed countries of the Affairs and Spatial Planning / National Land region and in other, more developed, compar- Agency (see box 1.1 for a discussion of these ator countries such as Brazil (figure 1.1). definitions).7 This report uses Indonesia’s offi- In 2017, about 144 million Indonesians lived cial statistical definition of urban areas in urban areas out of a total population of because it underpins most data on urbaniza- approximately 264 million. This urban popu- tion trends and because comprehensive infor- lation is more than the entire population of mation on the boundaries consistent with the the Russian Federation, the world’s ninth planning/legal definitions is hard to come by.8 most populous country.10 P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   47 BOX 1.1  Three of a kind—Indonesia’s three definitions of urban areas •  Statistical definition (BPS Regulation The identification of areas as urban does not 37/2010). This definition used by Badan have any consequence for the governance of an Pusat Statistik (BPS) is based on a composite area. The MOHA and ATR/BPN definitions pro- scoring system that assesses and scores areas vide the basis by which district governments may as urban or rural according to four “urban initiate the legal (as opposed to statistical) reclas- characteristics”: population density, structure sification of an area from rural to urban. This of the local economy (specifically, the share of influences both the way an area is governed and households that are nonagricultural), presence its treatment within Indonesia’s national urban of certain types of infrastructure (electricity system. MOHA and ATR/BPN definitions allow and telephone networks), and the presence of for the delineation of urban areas of any shape or certain urban amenities (schools, hospitals, size, crossing any number of provincial or district markets, shops, and recreational facilities, administrative boundaries. such as hotels, cinemas, salons, and massage Although all three definitions emphasize the parlors). An area is classified as urban for economic structure of an area in considering statistical purposes if its score exceeds a set whether to designate it as urban, the ATR/BPN threshold (see annex 1A for further details). and statistical definitions give extra weight to •  Ministry of Home Affairs (MOHA) definition: the presence of government services and addi- Law 23/2014 on Provincial and Local tional types of economic activity. The statistical Governance states, “A territory may be definition also further considers population den- defined as urban (perkotaan) if it has specified sity. Finally, whereas the statistical definition is borders, within which the primary economic broad but precise in the factors it considers, the activities of the inhabitants are in the fields of ATR/BPN and MOHA definitions are narrower industry and services.” but vaguer. •  Ministry of Agrarian Affairs and Spatial It might seem confusing that a country has Planning (ATR/BPN) definition: Law 26/2007 multiple definitions of urban areas, but Indonesia on Spatial Planning states, “An urban area is not unique. Austria, Bangladesh, China, India, (kawasan perkotaan) is a territory the primary and the United Kingdom all have both statistical functions of which are non-agricultural, but and planning/legal definitions of urban areas. rather comprise urban settlements and the cen- Part 1 of this report uses the statistical defi- tralization and distribution of government ser- nition for analyzing urbanization and related vices, social services, and economic activities.”a trends because that is the definition BPS uses for its data. Every three years, BPS, in its Survey of Village Potential (Potensi Desa, PODES), uses the statisti- a. At the time of the passing of Law 26/2007, the relevant ministry was the Ministry cal definition to classify every desa and kelurahan of Spatial Planning. b. Both desa and kelurahan correspond to fourth-level administrative units in in Indonesia as urban or rural.b The definition is Indonesia. A desa is a “village” and is governed in accordance with local traditions. also used in the enumeration of Indonesia’s urban A kelurahan is an “urban community.” A kelurahan has less autonomy from higher population in the national population census. levels of government than a desa. Most of Indonesia’s urban dwellers (almost Kalimantan is the next most urbanized island- 70 percent in 2016) live on Jawa-Bali region (43.5  percent). Nusa Tenggara and figure 1.2, panel a).11 Jawa-Bali is the most (­ Maluku-Papua are the least urbanized island- populous of Indonesia’s six island-regions12 regions, with fewer than one in three people and by far the most urbanized (figure 1.2, living in towns and cities. Consequently, panel b). Jawa-Bali is the only island-region whereas urbanization overall in Indonesia and where more than half of the popula- on Jawa-Bali is at an intermediate stage, it is at tion  (60.8  percent) lives in urban areas. a much earlier stage in the rest of the country. 48  TIME TO ACT After accelerated urbanization in the decades over only a five-year period given its level of following independence, the pace slowed at urbanization in 2010).13 The fitted line in the the turn of the century to a more “normal” figure shows the predicted pace of urbaniza- rate for Indonesia’s stage of urbanization tion at any given level of urbanization based The growth rate of the share of the popula- FIGURE 1.1  Indonesia is at an intermediate level of tion that lives in urban areas—Indonesia’s urbanization, 2017 overall pace of urbanization—is the relevant metric for gauging how fast Indonesia is tran- sitioning from a rural to an urban society EAP–developed 88.9 (box 1.2). In the two decades following its Brazil 86.3 independence in 1945, this growth rate aver- Malaysia 75.4 aged just over 1.5 percent a year (figure 1.3). Beginning around 1970, the pace of urbaniza- China 58.0 tion began to accelerate, exceeding 3 percent World 54.8 a year on average in the 1980s and 1990s, EAP–developing 54.7 which was faster than in the other developing countries of East Asia and Pacific overall and Indonesia 54.7 in China during the same period. Thailand 49.2 The pace of urbanization was also faster Philippines 46.7 than the pace that would be predicted on the basis of Indonesia’s level of urbanization at Vietnam 35.2 the time. Figure 1.4 plots the average growth 0 10 20 30 40 50 60 70 80 90 100 rate of the urban population share over Share of population in urban areas (%) 10-year intervals from 1950 onward against the initial level of urbanization at the begin- Source: Calculations based on data from the United Nations World Urbanization Prospects: 2018 ning of each 10-year period for 231 coun- Revision database (https://esa.un.org/unpd/wup/). Note: EAP denotes East Asia and Pacific. Following World Bank definitions, EAP–developed includes tries (the final interval of 2010–15 shows the high-income economies, whereas EAP–developing includes non-high-income economies. World and growth rate of the urban population share EAP–developing exclude Indonesia. FIGURE 1.2  Almost 70 percent of the national urban population lives on Jawa-Bali, which is the only island-region with more than half of its population in urban areas, 2016 a. Share of national urban population (percent) b. Share of population living in urban areas (percent) 70 68.9 70 60.8 Share of national urban population (%) Share of population in urban areas (%) 60 60 50 50 43.5 40.2 40 40 35.0 31.6 31.3 30 30 20 16.9 20 10 10 5.1 5.0 2.4 1.6 0 0 a n li i a a li n a i a ua es es er Ba Ba ar pu er ar ta ta p law law at g g at an an a- a- Pa Pa ng ng m m w w lim m Su Su - - Su ku ku Te Te Su Ja Ja li Ka Ka alu alu sa sa Nu Nu M M Source: Calculations based on data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   49 BOX 1.2  What is the correct measure of the pace of urbanization? There tends to be a lack of precision and clarity 2. Growth rate of the urban population, which is in how researchers and commentators describe a the growth rate of the absolute urban country’s pace of urbanization. They frequently population. It measures the speed at which the talk about a “rapid” pace of urbanization, often number of people in urban areas is increasing. referring to a large increase in the absolute popu- This growth rate may be high even after a lation of a city or urban areas in general or a per- country’s urban transition is complete (i.e., even ceived high rate of urban population growth. It is once it has reached a high level of urbanization) possible, however, for a country’s urban popula- if the rate of natural population increase is tion to increase by a large number without much high. It is a particularly relevant metric when change in the share of its population that lives in thinking about how quickly congestion forces urban areas (its level of urbanization). This will may be mounting in urban areas. be the case if the populations of urban and rural 3. Absolute increase in urban population, which areas are increasing at an equal rate and if there is is the absolute increase in the number of neither any reclassification of rural areas as urban people who live in urban areas. A country can nor any (net) migration between rural and urban experience a large absolute increase in its areas. It could also occur if the country is starting urban population without having a high from a high base of urban population. growth rate of urban population if its urban This report uses the following three measures population base is large. For example, an of urban population growth. Each conveys dif- absolute increase in urban population of ferent information and offers different insights. 10 million people a year translates into an annual growth rate of 100 percent for a 1. Growth rate of the urban share of the population, country that starts off with an urban which is the growth rate of the share of a population of 10 million but an annual country’s population that lives in urban areas. It growth rate of just 10 percent for a country measures how fast a country is transitioning that has an initial urban population of from rural to urban. When the share of a 100  million. Nevertheless, the absolute country’s population that lives in urban areas increase in urban population is a relevant reaches 100 percent, this growth rate becomes metric for thinking about the number of zero, no matter how fast the populations of its additional people for which governments and cities is increasing. This report uses “pace of the private sector need to provide housing and urbanization” to refer to this rate only. urban infrastructure and services. on the experiences of these 231 countries and 1960s. Between 2000 and 2017, Indonesia’s over the last seven decades. As the figure pace of urbanization more than halved. shows, Indonesia’s pace of urbanization was Consequently, Indonesia no longer leads the higher than predicted in the 1980s and region in pace of urbanization but now lags 1990s, indicating rapid urbanization by both China and other developing countries in global standards. Consistent with this find- East Asia and Pacific (see figure 1.3). ing, researchers and commentators have used This slowing pace of urbanization has both phrases such as “hyper-urbanization”14 and a structural component and a mean-reverting “mega-urbanization” to describe the pace of component. The structural component reflects urbanization in Indonesia, also characteriz- the fact that, as a country approaches higher ing it as “unstoppable” (Firman 2017). levels of urbanization, its pace of urbanization Since the turn of the century, however, naturally slows.15 The mean-reverting compo- Indonesia’s pace of urbanization has slowed, nent is shown in figure 1.4 as Indonesia’s pace returning closer to the pace in the 1950s of urbanization moves from above the global 50  TIME TO ACT trendline in the 1980s and 1990s to being FIGURE 1.3  Indonesia’s pace of urbanization much closer to the trendline since the turn of accelerated through the turn of the century and then the century. Indonesia’s pace of urbanization slowed, 1950–2017 can thus be said to have slowed to a near “nor- 3.5 Growth rate of urban share of population (%) mal” or “average” level, contradicting any pos- sible claims that it is currently occurring too 3.0 fast and that it needs to be “controlled” by, for 2.5 example, measures to deter rural–urban migra- tion. The pace has slowed across all six of 2.0 Indonesia’s main island-regions, particularly 1.5 since 2010 (­figure 1.5).16 Some possible expla- nations for Indonesia’s slowing pace of urban- 1.0 ization are discussed in box 1.3. 0.5 0 Recent urban population growth has been 1950–60 1960–70 1970–80 1980–90 1990–2000 2000–10 2010–17 driven by natural population growth and Indonesia China EAP–developing reclassification from rural to urban rather than by migration Source: Calculations based on data from the United Nations World Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). Consistent with the slowing pace of urbaniza- Note: EAP–developing includes all non-high-income economies in East Asia and Pacific except tion, net rural–urban migration made only a Indonesia. Growth rates are calculated as compound annual growth rates of the urban share of the population over 10-year intervals, with the exception of 2010–17. The black dashed line divides pre- modest contribution to overall urban popula- 2000 and post-2000 growth. tion growth in Indonesia between 2000 and 2010 (figure 1.6).17 Less than 20 percent of urban population growth was attributable to internal migration, whereas more than FIGURE 1.4  Rising and then falling—benchmarking 80 ­ percent came from the reclassification of Indonesia’s pace of urbanization against global experience, 1950–2015 rural settlements as urban and natural popula- tion growth in urban areas. As highlighted ear- 10 lier, the reclassification of settlements reflects y = –1.31 ln(x) + 6.05 real, on-the-ground transformation rather than R2 = 0.36 Growth rate of urban share of population (%) 8 just a bureaucratic change of designation.18 By island-region, the contribution of net migra- 6 tion to urban population growth was smallest for Jawa-Bali (14 percent), which is consistent 4 with its higher overall level of urbanization and 1980s 1990s 1970s with its population density contributing to 2 1950s 1960s 2000s stronger congestion forces within its urban 2010–15 areas.19 The contribution of net rural–urban 0 migration to Indonesia’s urban population growth was smaller than in China and India, –2 although differences in methodology mean that such comparisons need to be treated with care –4 (box 1.4). 0 10 20 30 40 50 60 70 80 90 100 Initial urban share of population (%) In absolute terms, however, the recent All countries Indonesia Fitted line (all countries) increase in Indonesia’s urban population has been huge Source: Calculations based on data from the United Nations World Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). Despite the slowing pace of urbanization, the Note: Each data point shows the average growth rate of a country’s urban population share over a given period for its urban population share at the beginning of that period. The periods considered absolute increase in the number of people are 1950–60, 1960–70, 1970–80, 1980–90, 1990–2000, 2000–10, and 2010–15; therefore, the figure living in Indonesia’s towns and cities since contains seven observations for each of 231 countries. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   51 FIGURE 1.5  The pace of urbanization slowed across the beginning of the century has been huge. all Indonesian island-regions between 2004–10 and In the world’s fourth most populous country, 2010–16 even a slower pace of urbanization still translates into a large absolute increase in 3.0 2.76 urban population. And, although Indonesia’s 2.64 2.5 2.39 2.27 natural rate of population increase declined 2.09 from 21.4 per 1,000 population in 1950–55 2.0 1.87 to 13.2 per 1,000 population in 2010–15 as Percent 1.5 the country transitioned demographically, it 1.0 0.98 remains high compared with that of devel- 0.71 0.67 0.63 0.60 0.48 oped countries such as the United States, 0.5 where the natural population growth rate 0 was 4.3 per 1,000 population in 2010–15. Sulawesi Jawa-Bali Maluku- Kalimantan Nusa Sumatera As a result, Indonesia’s population, including Papua Tenggara its urban population, remains relatively 2004–10 2010–16 young: although the median age of Indonesia’s population increased from 18.5 Source: Calculations based on data from Indonesia’s 2004, 2010, and 2016 National Socio-Economic years in 1975 to 28 years in 2015, it remains Surveys (SUSENAS). Note: The pace of urbanization is calculated as the compound annual growth rate of the urban share of low compared with the United States (37.6 the population by island-region over the periods 2004–10 and 2010–16. years) and, especially, Japan (46.3 years). BOX 1.3  Explaining Indonesia’s urbanization slowdown from rapid to near normal The slowing of Indonesia’s overall pace of urban- positively value—for example, clean water, safe ization can be thought of as having both struc- sanitation, good schools and hospitals, restau- tural and mean-reverting components. From this rants, open spaces, and recreational ­ facilities— perspective, it is the rapid urbanization of the and the relative absence of disamenities, such as 1980s and 1990s that looks unusual rather than traffic congestion, crime, and pollution, which in the slower, near “normal,” pace of urbanization turn are associated with the strength of conges- in evidence since 2000. tion forces in a particular area.a A country’s pace of urbanization depends in Thus, the unusually rapid urbanization of the part on the decisions of potential migrants to 1980s and 1990s could have its origins in lower move from rural to urban areas. These decisions costs of migrating during that period or, more reveal information on the relative attractive- likely, in the stronger relative attractiveness of ness of different areas. Economic theory states urban areas—in terms of expected income gains that a rural resident’s decision of whether to and quality of life—during that period. Stronger migrate to an urban area will be based on the relative attractiveness of urban areas in the extra “utility” (happiness) expected from mov- 1980s and 1990s could, in turn, have been due to ing relative to the costs of moving. In turn, the greater prosperity, inclusiveness, and livability in extra happiness from moving will depend on cities than prevail today or worse conditions in the expected income gain from moving and rural areas that have since been alleviated, with- the relative quality of urban and rural life. The out being eliminated, by rural catch-up (possibly expected income gain from moving consists of spurred by the decentralization reforms of the a static component—a one-time jump in income last two decades). Evidence in the report sug- from moving to a city—and a dynamic compo- gests that both sets of forces may have been at nent, which depends on the migrant’s ability to play. progress faster up the income ladder in an urban area than in a rural area. Meanwhile, the relative a. The migration-choice problem described here corresponds well to that modeled in, for example, Bosker, Deichmann, and Roberts (2018). For a good, although quality of urban  and rural life depends on the slightly dated, review of theoretical and empirical models of rural–urban availability of amenities that potential migrants migration, see Lall, Selod, and Shalizi (2006). 52  TIME TO ACT FIGURE 1.6  In Indonesia, and especially in Jawa-Bali, natural population growth and reclassification from rural to urban contributed more to urban population growth than did rural–urban migration, overall and by island-region, 2000–10 100 18.8 14.0 17.6 20.4 23.5 80 32.2 40.7 37.4 60 37.6 41.6 33.8 Percent 42.4 37.9 26.8 40 48.6 45.8 20 43.5 40.8 34.1 29.9 32.5 0 Indonesia Jawa-Bali Nusa Sulawesi Sumatera Kalimantan Maluku Tenggara Reclassi cation Natural population growth Migration Sources: Wai-Poi et al. 2018, using data from the 2000 and 2011 Survey of Village Potential (PODES), and the 2000 and 2010 population censuses. Note: Papua is excluded for data reasons. For a summary of the urban population growth decomposition methodology see annex 1B. BOX 1.4  Comparing the contribution of migration to urban population growth across Indonesia, China, and India How does the roughly 19 percent contribution of permanent household registration (Hukou) sys- net rural–urban migration to urban population tem imposed on ­ internal migration throughout the growth in Indonesia over 2000–10 compare with period (see,  for example, World Bank and DRC the contribution in other large Asian develop- 2014). The contribution of net migration to urban ing countries? Answering this question precisely population growth for Indonesia is also less than is difficult because of methodological differ- for India, but the difference, although not negli- ences in decomposition exercises across studies. gible, is smaller. The greater difference between Nonetheless, broad patterns can be discerned. Indonesia and India is in the relative contributions The contribution of net migration to urban popu- of reclassification and natural population growth. lation growth between 2000 and 2010 in Indonesia Whereas reclassification has been the largest single (19 percent) was only about one-third of its contri- driver of urban population growth in Indonesia, bution in China (56 percent) (figure B1.4.1). This natural growth within preexisting urban areas has result is despite the severe restrictions that China’s been the largest single driver in India. FIGURE B1.4.1  Decomposition of urban population growth for Indonesia, India, and China Indonesia, 2000–10 India, 2001–11 China, 2000–10 19% 27% 29% 35% 43% 56% 38% 44% 9% Reclassi cation Natural population growth Migration Sources: Derived from figures in Wai-Poi et al. 2018 for Indonesia, World Bank and DRC 2014 for China, and Pradhan 2013 for India. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   53 FIGURE 1.7  Growth in Indonesia’s urban the national political or administrative capital population peaked in 1990–2000 but remains high or the dominant economic and cultural center of a country. Because of its size, the primate 40 city will often be home to a diverse set of industries and serve as the primary interna- Urban population increase (millions) 35 33.4 32.2 tional gateway of the country. In tiers below 30 26.6 the primate city are other large cities with 25 22.9 many of the same characteristics as the primate city though without the same scale. Then come 20 medium-size cities (sometimes also referred to 15 13.0 as secondary or intermediate cities), which are 10 often regional centers of economic activity. At 6.8 the bottom are small cities and towns, which 5 4.2 act as the interface between rural and urban 0 areas, providing, for example, market centers 1950–60 1960–70 1970–80 1980–90 1990–2000 2000–10 2010–18 for the agricultural output produced in sur- rounding rural areas. In developed countries, Source: Calculations based on data from the United Nations World Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). medium-size and smaller cities are more manu- facturing oriented than larger cities, whose economies tend to be driven more by human It also remains low compared with China capital–intensive tradable service activities (37  years), where the One Child Policy, (World Bank 2008). although recently relaxed, has contributed to an older society.20 Between 2010 and 2018, Indonesia’s urban This report considers four types of place population grew by almost 27 million To analyze the rich diversity of urbanization figure  1.7)—equivalent to more than the (­ dynamics within Indonesia, it is useful to dif- entire population of Australia—which trans- ferentiate between different types of place lates into an annual rate of urban population using districts (which correspond to second- growth of about 2.5 percent. This growth is level administrative units) as the building important because it is the absolute increase blocks (see box 1.5).21 Crucially, such differ- in population in an urban area that deter- entiation also provides a key basis for provid- mines the potential for agglomeration forces ing tailored policy recommendations in part 2 and the strength of congestion pressures on of this report. We distinguish the four follow- basic urban services, infrastructure, land and ing types of place: housing markets, and the environment (see, for example, Jedwab and Vollrath, forthcom- •  Multidistrict metro areas are large met- ing). If policies do not respond adequately to ropolitan areas with functional labor ease these pressures, or, even worse, if the pol- markets (defined using commuting flow icies exacerbate these pressures, they can data) that cut across multiple districts. dampen the prosperity, inclusiveness, and liv- A  multidistrict metro consists of two ability benefits of urbanization. subareas: ◦ Metro core, which is the district within the metro area with the highest average Indonesia’s evolving population density, 22 except in the case of Jakarta, where the core is taken to “portfolio of places” be DK I Jakar ta (Daerah K husus Countries have a variety of urban areas that Ibukota Jakarta). can vary by size, economic characteristics, and ◦ Metro periphery, which corresponds to the functions they serve. Countries possess a the noncore districts within the metro hierarchy of these urban areas—or a “portfolio area. These districts are linked to the of places.” At the apex is the primate city, often core through strong commuting flows. 54  TIME TO ACT BOX 1.5  Methodology for defining Indonesia’s portfolio of places Using districts as the building blocks, a three- of its population that lives in urban settlements. step methodology distinguished four basic types Where the share of the population living in urban of place: multidistrict metro areas; single-district settlements is at least 50 percent, the district is metro areas; nonmetro, predominantly urban classified as urban; where the share is less than areas; and nonmetro, predominantly rural areas. 50 percent, it is classified as rural. Multidistrict metro areas are distinguished from single-district metro areas by having labor markets Step 2: Identify multidistrict metro areas that are spread across multiple districts rather than confined within a single district.a Within multidis- Multidistrict metro areas have labor markets that trict metro areas, a district or districts that make span multiple districts. These areas are identified up the “core” of the metro area are distinguished using data on daily commuting flows between dis- from a district or districts that form the metro tricts, applying an algorithm devised by Duranton periphery, which may be either predominantly (2015). This algorithm identifies metro areas urban or predominantly rural ­(figure B1.5.1). through an iterative process of aggregating districts. The three-step methodology used to distin- As an example, in the first round of this process, a guish types of place is described briefly here district (say, district A) is aggregated to a second (for a more complete description, see Park and district (district B) to form a multidistrict metro Roberts 2018). area if the share of workers who live in district A but commute daily to work in district B exceeds a predefined threshold level, T. In the second round, Step 1: Classify districts as predominantly a further district (district C) is then aggregated to urban or predominantly rural the multidistrict metro area formed by districts A Data from the 2014 round of the National and B if the share of workers who live in district C Socio-Economic Survey (SUSENAS) was used but commute daily to districts A and B (combined) to classify each district as predominantly urban exceeds the same threshold level T. This iterative or predominantly rural according to the share process of aggregating districts continues until FIGURE B1.5.1  Stylized depiction of the different types of place within the “portfolio of places” Nonmetro Metro Single-district urban core metro Urban periphery Rural periphery Nonmetro rural Source: Based on Park and Roberts 2018. Box continued on next page P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   55 BOX 1.5 Continued there is no district left to be aggregated that satis- threshold of 500,000 is the same as that applied by fies the daily commuting flow threshold. In the case the Organisation for Economic Co-operation and where a district satisfies the threshold with two or Development in classifying urban areas as metro more other districts, the algorithm aggregates it areas (OECD 2013), and the population density to the district to which it sends most commuters. threshold is the same as that used by, for exam- When commuting flows between two districts are ple, both the European Commission (Dijkstra and above the threshold in both directions, the algo- Poelman 2014) and Henderson, Nigmatulina, and rithm also ensures that the smaller district is aggre- Kriticos (2018) to identify cities. gated to the larger one. A total of 21 multidistrict Applying the three criteria led to the iden- metro areas were identified. tification of four single-district metro areas in A commuting flow threshold of T = 7.5 ­ percent Sumatera (Bandar Lampung, Jambi, Palembang, was used after experimenting with a range of and Pekanbaru). Following consultations with thresholds. This threshold was chosen on the basis country experts, three more kota districts— of an assessment of the plausibility of the resultant Padang in Sumatera, and Balikpapan and metro areas, which was informed by consultations Samarinda in Kalimantan—with populations of with government of Indonesia officials and World more than 500,000 but that narrowly fell short Bank country experts.b Data on daily commut- of meeting the population density criterion were ing flows were derived from the August 2013–15 also classified as single-district metro areas.f rounds of Indonesia’s National Labor Force Survey a. In applying the methodology, districts were defined by their 1996 administrative (SAKERNAS).c For the Jakarta multidistrict metro boundaries rather than their current boundaries to facilitate the analysis of area, the core was set as the districts that form urbanization and related trends of structural transformation. However, all Daerah Khusus Ibukota (DKI) Jakarta.d For each single-district metro areas identified in step 2 remain single-district metro areas even when overlaid with 2016 boundaries—despite the proliferation of districts of the remaining multidistrict metro areas, the core through splitting that has occurred in Indonesia since 1996 (see chapter 5). district is defined as the district with the highest b. These consultations took place in February 2018 at a workshop convened by average population density in 2014.e the Ministry of National Development Planning / National Development Planning Agency (Bappenas). c. SAKERNAS defines workers as all employed wage workers (including casual Step 3: Identify single-district metro areas workers), self-employed workers, and unpaid family workers. Anyone who worked for at least one hour consecutively in the previous week is considered employed. Single-district metro areas are individual districts The 2013–15 August rounds of SAKERNAS are representative at the district level. resembling multidistrict metro areas in key respects d. DKI Jakarta is composed of six districts—Jakarta Pusat, Jakarta Barat, Jakarta Selatan, Jakarta Timur, Jakarta Utara, and Kepulauan Seribu. Technically, Duranton’s that also distinguish them from nonmetro districts, algorithm fails to aggregate Kepulauan Seribu with the other districts of DKI Jakarta but which fail, because of insufficiently strong and the rest of the Jakarta multidistrict metro area. It is nevertheless included as part commuting flows, to aggregate with other districts of the definition of the metro core given its status as part of DKI Jakarta. e. One agglomeration of districts—consisting of Kota Sibolga and Kabupaten Tapanuli in step 2. To qualify as a single-district metro area, Tengah—that was identified as a multidistrict metro area in step 2 was dropped because a district must satisfy three criteria: it must be a of its failure to meet the criteria set in step 3. A second agglomeration—consisting kota, its population must be at least 500,000, and of Kota Solok and Kabupaten Solok—identified by the algorithm was dropped as a multidistrict metro area in step 2 because of its low overall level of urbanization. its average population density must be at least f. A fourth criterion—whether 75 percent or more of a district’s workers are 1,500 people per square kilometer. The population employed outside of the primary sector—was considered but proved redundant. Metro periphery districts can be pre- make them resemble multidistrict metro dominantly urban or predominantly areas but whose functional labor markets rural, where a predominantly urban dis- are confined within the administrative trict is one with at least half of the popu- boundaries of a single district. lation in urban settlements.23 •  Nonmetro urban areas are districts that do •  Single-district metro areas are kota dis- not meet the criteria to be classified as either tricts with populations of at least 500,000 a single-district metro area or part of a mul- and average population densities that tidistrict metro area but within which most 56  TIME TO ACT of the population is urban. Such districts (box 1.6). Because of its size and its role as may be either kota or kabupaten.24 Indonesia’s primate city, Jakarta is discussed •  Nonmetro rural areas are districts away separately from the other multidistrict metro from the metro areas in which most of the areas in much of this report’s analysis.29 population is rural. Again, such districts The methodology identified seven single- may be either kota or kabupaten.25 district metro areas. Collectively, they were home to almost 5 percent of Indonesia’s A total of 21 multidistrict metro areas, which urban population in 2016. Five of them together are composed of 67 districts plus (Bandar Lampung, Jambi, Palembang, DKI Jakarta, were identified (table 1.1). Padang, and Pekanbaru) are on Sumatera, Together, they accounted for 52 percent of the and the remaining two (Samarinda and national urban population in 2016. Of these Balikpapan) are on Kalimantan. This leaves multidistrict metro areas, 15 are on Jawa-Bali, 434 nonmetro districts, 57 of which are non- 3 on Sumatera, 2 on Kalimantan, and 1 on metro urban areas with the remainder being Sulawesi. By far the biggest multidistrict nonmetro rural areas. Because of their large metro area is Jakarta (map 1.1, panel a). In number, even though no one nonmetro rural 2016, Jakarta had a population just shy of area has a majority urban population, collec- 31 million, which is 3.8 times the population tively they were home to 37 million people, or of the second-largest multidistrict metro area, just over one-quarter of Indonesia’s urban Bandung, which had a population of 8.1 mil- population in 2016, which is more than the lion (map 1.1, panel b). Jakarta’s share of the collective population (26.7 million) of the national urban population is 22.6 percent, of metro cores. Again, this large collective urban which roughly one-third live in the core (DKI population underscores the importance of not Jakarta).26 The Jakarta metro area spreads neglecting the nonmetro rural areas when across 14 districts,27 by far the most of any thinking about Indonesia’s urbanization pro- multidistrict metro.28 The question naturally cess. Finally, figure 1.8 shows the absolute arises whether the Jakarta metro area is too numbers of people—both overall and urban— big , but the evidence suggests it is not that lived in each type of area in 2016. TABLE 1.1  The different types of place in Indonesia’s portfolio Number of Percent of national Type of place Description districts urban population Metropolitan areas Multidistrict metro area Core District with highest population density (except for DKI Jakarta and 20.1 Jakarta, where DKI Jakarta is the core) 20 others Urban Predominantly urban, noncore districts 27 27.0 periphery Rural Predominantly rural, noncore districts 20 4.9 periphery Single-district metro Predominantly urban kota districts with ≥ 500,000 7 4.9 population and ≥ 1,500 people per square kilometer on average Nonmetropolitan areas Urban Predominantly urban nonmetro districts 57 15.3 Rural Predominantly rural nonmetro districts 377 27.9 Source: Calculations based on data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS) for percent of national urban population. Note: Typology is derived following the methodology described in box 1.5. Number of districts is based on 2016 district administrative boundaries. DKI Jakarta = Daerah Khusus Ibukota Jakarta. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   57 MAP 1.1  Jakarta is by far the largest of Indonesia’s built up than their peripheries, single-district metro areas, 2016 metro areas, and nonmetro areas (table 1.2). Along with single-district metro areas, they a. Multi-district metro areas in Jawa-Bali are also, on average, significantly more urban- ized, more productive (in terms of GDP per capita), and more service oriented than other types of place and their own peripheries. Although nonmetro urban areas are not as Jakarta Salatiga Mojokerto productive, on average, as multidistrict metro Semarang Surabaya cores and single-district metro areas, they still have higher average GDP per capita than Sukabumi Pasuruan nonmetro rural areas, which have low aver- Bandung Magelang age population density and low average levels Yogyakarta Surakarta Blitar of built-up area. Malang Probolinggo Denpasar Within multidistrict metro areas, there is also a striking difference between the core 0 100 200 and its peripheries. The average picture for Kilometers these metro areas is of a relatively productive core and less productive peripheries. The b. Size distribution of Indonesia’s metro areas mean GDP per capita is two-thirds higher for Jakarta 30 the metro cores than for their urban peripher- ies and more than two times higher than for their rural peripheries. Indeed, the average Population (millions, natural log scale) GDP per capita in the rural peripheries is 9 almost identical to that in nonmetro rural 8 Bandung 7 6 Surabaya areas despite rural peripheries’ higher average 5 Medan Surakarta population density. 4 Semarang Malang However, differences in GDP per capita 3 Makassar Sukabumi between the cores of multidistrict metro Denpasar Yogyakarta 2 areas and their peripheries may well over- Pasuruan Pontianak Palembang Magelang Probolinggo state differences in prosperity and welfare. Blitar Mojokerto Salatiga Pekanbaru Banjarmasin This is because GDP captures value added, 1 Bandar Lampung Padang Samarinda much of which goes back to workers in the Banda Aceh Balikpapan form of wages. Hence, some of the GDP Bukittinggi Jambi generated in the cores will be shared, 10 20 30 through wages, with residents in the Rank of metro areas by population (natural log scale) ­ p eripheries—in particular, with the resi- Sumatera (multi) Kalimantan (multi) Jawa Bali dents in the peripheries who commute to Sumatera (single) Kalimantan (single) Sulawesi the cores for work. This “sharing” of GDP, Sources: Map based on nighttime lights data from the 2015 VIIRS (Visible Infrared Imaging Radiometer which increases with the strength of com- Suite) annual composite product (https://ngdc.noaa.gov/eog/viirs/download_dnb_composites.html). muting flows, does not take place in the Population data in panel b are derived from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). nonmetro areas 30 and helps explain why Note: Map shows patterns of brightness of nighttime lights for metro areas in Jawa-Bali derived using the methodology set out in box 1.5. Different colors are used to distinguish between contiguous metro real consumption per capita is, on average, areas. Darker shades within a given metro area indicate higher levels of brightness of nighttime lights. higher in rural peripheries than in nonmetro The intensity of lights is not comparable across metro areas because of different scales. The white lines rural areas. Another notable difference correspond to Indonesia’s 2016 district administrative boundaries. between multidistrict metro cores and their peripheries is that the peripheries, at least in Economic characteristics vary considerably their more urban parts, are more industry by place, including for the cores and oriented. This difference is as expected peripheries of multidistrict metro areas given the land-intensive nature of industry On average, the cores of multidistrict metro compared to services, in which the cores areas are more densely populated and more specialize (figure 1.9). 58  TIME TO ACT BOX 1.6  Is Jakarta too big? With more than 30 million people calling the the primate city. This overconcentration, in turn, Jakarta metro area home, its population exceeds leads to excessive congestion forces, which could that of many countries, including Australia, be avoided if resources were more evenly spread Nepal, the Netherlands, and Sri Lanka. Concerns across cities. These excessive congestion forces are frequently expressed that Jakarta is too then lower national economic growth. large—or “excessively” primate, meaning that the Is Jakarta excessively primate? The evidence share of Indonesia’s national urban population suggests no. Under an internationally compa- living in Jakarta may be larger than is optimal for rable definition of cities, the share of Indonesia’s national economic growth. urban population that lives in Jakarta is a little The idea that excessive primacy creates a drag more than 22 percent,a which places Indonesia on national economic growth stems from a paper in the second decile of urban primacy interna- published in 2000 by urban economist Vernon tionally on the basis of a sample of 156 countries Henderson, who several times referenced Jakarta. figure B1.6.1, panel a).b It may be argued that (­ Henderson associated excessive primacy with this level is misleading, because Indonesia is much an overconcentration of resources in a country’s more populous than most countries in the world, largest city, thereby depriving other cities of ade- and, as the empirical results in Henderson (2000) quate resources. Such overconcentration can arise show, a country’s “optimal” degree of primacy for a variety of reasons, including favoritism by declines with the overall size of its  ­ population. the national government in the allocation of pub- However, Indonesia’s level of urban primacy lic services and infrastructure, a lack of auton- does not look out of place when set against the omy of subnational governments, insufficient levels observed in the other nine most populous investment in national transport and telecom- countries in the world (­ figure  B1.6.1, panel  b). munications networks, and restrictions in capital Furthermore, Henderson also finds that although markets, export/import markets, and licensing of Indonesia’s level of urban primacy is higher than production rights that favor firms that locate in optimal, it is not excessively so. FIGURE B1.6.1  Urban primacy levels for 156 countries globally and urban primacy levels for the world’s 10 most populous countries, 2012 b. Primacy levels for world’s 10 most a. Distribution of urban primacy populous countries 18 China 16 India 14 Includes United States Number of countries Jakarta 12 Indonesia 22.4 10 Brazil 8 Pakistan 6 Nigeria Bangladesh 4 Russia 2 Mexico 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 0 10 20 30 40 Level of primacy (%) Level of primacy (%) Source: Calculations apply the Dijkstra and Poelman (2014) algorithm to Landscan-2012 gridded population data. Note: Primacy is defined using a globally consistent definition of cities as the share of a country’s overall high-density cluster population that lives in its most populous high-density cluster. Panel a excludes countries with a primacy rate of 100 percent (mainly small countries that possess only a single high- density cluster). Cities in panel b are sorted in descending order of the populations of the countries within which they are located. Box continued on next page P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   59 BOX 1.6 Continued Since the Henderson paper in 2000, Indonesia urbanization process to sustainably increase pros- has completed a “big bang” fiscal decentralization, perity, inclusiveness, and livability. which dramatically increased the local autonomy a. The internationally comparable definition of cities used here is the European of subnational governments, and this reduced Commission’s high-density cluster definition, which identifies a city as a spatially some of the potential reasons for an overconcen- contiguous area of land that is densely populated throughout its entire area and tration of monetary resources in the primate city has a reasonably large overall population (Dijkstra and Poelman 2014). “Densely populated” is defined as at least 1,500 people per square kilometer, whereas (see chapters 5 and 6). This does not mean, how- “reasonably large” is defined as a population of at least 50,000. ever, that further reforms to Indonesia’s system of b. Although data are available for 182 countries, 26 countries with only a single urban governance and finance are not needed to city are excluded from the sample because they have primacy rates of 100 percent. These are mainly small countries or city-states such as Singapore. Including them ensure the augmented coverage of quality basic would only strengthen the conclusion that Jakarta is not excessively primate. Similar services across Indonesia’s full portfolio of places, results are obtained using data on levels of urban primacy based on official national and to lay the foundations for better integration definitions of cities taken from the World Bank’s World Development Indicators database. According to those data, Indonesia’s level of urban primacy in 2016 was both within and across the places. Such reforms only 7.4 percent, which places it in the first decile globally in primacy. The lower level are crucial to fulfilling the potential of Indonesia’s of urban primacy stems from a narrower definition of the Jakarta metropolitan area. FIGURE 1.8  Collectively, nonmetro rural areas Structural transformation is rapid in rural account for a large share of the population, including areas, with multidistrict metro cores the urban population, 2016 moving toward services and industry shifting to their peripheries a. Overall population The structure of employment has been chang- 140 ing across Indonesia’s portfolio of places. 120 Although agriculture’s share of employment 100 has been declining everywhere, the areas that Millions 80 have experienced the biggest transformation 135.0 60 away from agriculture since 2007 are the 40 rural peripheries of multidistrict metro areas 21.4 20 16.4 29.4 and the nonmetro rural areas, followed by the 22.0 16.9 6.6 0 10.3 nonmetro urban areas and the urban periph- Metro Urban Rural Single- Nonmetro Nonmetro eries of multidistrict metro areas. Much of the core periphery periphery district urban rural movement away from agriculture has been metro toward services rather than industry, although b. Urban population the change has been more balanced in the 40 rural peripheries of the multidistrict metro 35 areas outside of Jakarta (figure 1.10). 30 As employment has been shifting from 16.0 25 industry toward services within the cores of Millions 20 37.0 multidistrict metro areas, a shift that is par- 16.4 15 ticularly pronounced in Jakarta, industry has 10 19.8 20.3 been moving to the peripheries. To some 5 10.3 6.5 6.4 extent, this suburbanization of industry may 0 be considered a “natural” part of urbaniza- Metro Urban Rural Single- Nonmetro Nonmetro core periphery periphery district urban rural tion and development. Metro cores are metro already more built up (see table 1.2), so it is Jakarta metro Others natural for congestion forces to be stronger in cores than in peripheries. The effect is to push Source: Calculations based on data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). development outward, as people and busi- Note: Places are defined using the methodology described in box 1.5. nesses trade off the advantages of proximity 60  TIME TO ACT TABLE 1.2  Summary characteristics for different types of place in Indonesia, circa 2016 Multidistrict metro area Nonmetro area Urban Rural Single-district Predominantly Predominantly Characteristic Core periphery periphery metro area urban rural Population (thousands) Mean 1,271 5,429 1,125 940 653 746 Median 514 2,995 1,146 909 335 649 Population density (per sq. km) Mean 7,212 1,693 641 2,696 2,546 278 Median 6,612 1,654 651 1,736 1,197 126 Urban population (percent of total population) Mean 99.4 76.0 36.3 97.6 80.0 25.3 Median 100.0 77.8 35.3 97.5 92.1 24.5 Built-up area (percent of total land area) Mean 55.1 18.7 6.9 18.7 19.2 2.2 Median 57.6 17.5 5.1 18.0 10.1 0.6 Share of national GDP (percent) Mean 1.4 1.8 0.3 0.5 0.3 0.2 Median 0.2 0.9 0.2 0.4 0.1 0.1 GDP per capita (million rupiah) Mean 70.1 42.1 33.6 70.6 61.1 34.1 Median 58.0 36.7 27.5 63.4 48.3 27.2 Real consumption per capita (thousand rupiah) Mean 1,347.2 1,034.0 801.2 1,380.0 1,085.4 763.4 Median 1,277.0 1,033.4 792.0 1,387.4 1,069.6 749.7 Manufacturing employment (percent of total employment) Mean 15.7 21.0 16.7 8.9 14.3 8.4 Median 15.3 20.9 13.8 8.5 11.4 7.1 Services employment (percent of total employment) Mean 75.5 54.6 41.5 77.3 63.4 37.2 Median 76.7 54.1 40.7 77.6 69.2 37.7 Sources: Population, population density, share of population that is urban, and real consumption per capita based on data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). Built-up area is calculated using European Commission Global Human Settlement Layer data for 2014 (https://ghsl.jrc.ec.europa.eu/). Percent national GDP and GDP per capita are based on nominal GDP data for 2016 provided by Badan Pusat Statistik (Statistics Indonesia). Manufacturing and services employment are based on data from the August round of Indonesia’s 2017 National Labor Force Survey (SAKERNAS). Note: Places are defined using the methodology set out in box 1.5. Mean is the unweighted mean across all districts within each type of place. For core, urban periphery, and rural periphery areas of a multidistrict metro area, all districts belonging to the same metro area were considered a single unit within each type. GDP = gross domestic product. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   61 FIGURE 1.9  Sectoral structure of employment by type of place, 2017 100 80 38 41 52 54 65 60 82 76 77 Percent 18 28 40 31 30 20 45 31 22 32 19 18 17 16 0 Jakarta core Jakarta Other Other urban Rural Single-district Nonmetro Nonmetro periphery metro core periphery periphery metro urban rural Agriculture Industry Services Source: Calculations based on data from the August 2017 round of Indonesia’s National Labor Force Survey (SAKERNAS). Note: Places are defined using the methodology described in box 1.5. Industry includes manufacturing, mining and quarrying, electricity/gas/water supply, and construction. Employment is based on the areas in which workers live rather than where they work. FIGURE 1.10  Changes in employment composition by sector and type of place, 2007–17 10.0 8.4 8.5 7.4 6.7 6.3 5.6 5.0 4.4 3.4 Percentage point change 2.7 1.6 1.6 0.6 0.3 0.9 0 –0.2 –0.9 –0.7 –0.9 –5.0 –5.4 –5.0 –10.0 –9.3 –9.4 –12.9 –13.7 –15.0 Jakarta core Other Single-district Jakarta Other urban Nonmetro Nonmetro Rural metro core metro periphery periphery urban rural periphery Agriculture Industry Services Source: Calculations based on data from the August 2007 and 2017 rounds of Indonesia’s National Labor Force Survey (SAKERNAS). Note: Places are defined using the methodology described in box 1.5. Industry includes manufacturing, mining and quarrying, electricity/gas/water supply, and construction. Employment is based on the areas where workers live rather than where they work. to the core with the availability of cheaper By freeing up space for development in land and housing, particularly as households urban cores, such suburbanization may be become richer and demand more living space considered beneficial for city economies. and as intracity transport networks improve.31 Nonetheless, the suburbanization of industry Similarly, with improved transport networks may still be undesirable if industry is pushed and the emergence of more human capital– out prematurely by poor urban management intensive industries, more land-intensive eco- and policy failures that lead to “excessive” nomic activities, such as manufacturing, congestion forces in metro cores—as naturally tend to suburbanize as they seek ­ chapter 2 suggests may be the case given evi- cheaper land. dence, for example, of relatively high levels of 62  TIME TO ACT traffic congestion, overcrowded housing, and The average annual growth rate of the urban crime—rather than being pulled out by population share was 1.0 percent in nonmetro improved connectivity, infrastructure, and urban areas and essentially zero for multidis- basic services in the peripheries. trict metro cores and single-district metro areas, because they were already fully or nearly The pace of urbanization is fastest in more fully urbanized by 2004. rural areas Consistent with these patterns of structural Migration is contributing to urban transformation, the pace of urbanization population growth in urban peripheries, between 2004 and 2016 was appreciably faster but not as much elsewhere in the rural peripheries of the multidistrict The pace of urbanization was fastest over metro areas and in nonmetro rural areas than 2004–16 in both the rural peripheries of in other types of place (figure 1.11, panel a). multidistrict metro areas and nonmetro For these areas, the growth rate of the urban rural areas (see figure 1.11, panel a), but population share was 2.6–2.7 percent a year, at the fastest urban population growth was in least double that in the urban peripheries of Jakarta’s periphery (figure 1.11, panel b). multidistrict metro areas, including Jakarta. A slow pace of urbanization combined FIGURE 1.11  Average annual growth rates of urban population shares and of urban population, by type of place, 2004–16 a. Urban population share 3.0 2.7 2.6 2.5 2.0 Percent 1.5 1.3 1.2 1.0 1.0 0.5 0.3 0.1 0.0 0 Rural Nonmetro Other urban Jakarta Nonmetro Single-district Other metro Jakarta periphery rural periphery periphery urban metro core core b. Urban population 6 5 4.8 4.1 4 3.7 Percent 2.9 2.7 3 2.7 2 1.4 1.3 1 0 Jakarta Nonmetro Rural Other urban Nonmetro Single-district Jakarta Other metro periphery rural periphery periphery urban metro core core Source: Calculations based on data from the 2004 and 2016 rounds of Indonesia’s National Socio-Economic Survey (SUSENAS). Note: Urban population share is the share of an area’s population that lives in urban settlements; growth rates are calculated as compound annual growth rates. Places are defined using the methodology described in box 1.5. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   63 FIGURE 1.12  Relative contributions of general pattern of a strong positive contri- natural population growth, migration, and bution of migration to urban population reclassification to urban population growth, growth in the urban peripheries of multi- by type of place, 2000–10 district metro areas. Between 2000 and 2010, migration contributed more than 100 6 16 13 10 one-third of the urban population growth 25 35 80 in urban peripheries of multidistrict metro 8 46 61 areas (figure 1.12). 60 61 95 The large contribution of migration to Percent 37 40 67 urban population growth in the urban periph- 20 38 eries of multidistrict metro areas stands in 28 26 29 stark contrast with the overall modest contri- 0 -1 bution of net rural–urban migration to urban –20 population growth for both other types of Metro core Single- Urban Nonmetro Rural Nonmetro place (see figure 1.12) and at the national district periphery urban periphery rural level (see figure 1.6). The multidistrict metro metro cores even experienced a slight negative Natural population growth Reclassi cation Migration (1 percent) contribution of migration, mean- ing that, overall, the number of people mov- Source: Wai-Poi et al. 2018, using data from the 2000 and 2011 Survey of Village Potential (PODES), and the 2000 and 2010 population censuses. ing out of urban settlements in these areas Note: Papua is excluded for data reasons. For a summary of the urban population growth decomposition slightly outnumbered the number of people methodology see annex 1B. moving in. Consistent with all of this, and except for the urban peripheries of multidis- TABLE 1.3  Built-up area growth has been rapid in trict metro areas, overall urban population rural places but slower elsewhere, 2000–14 growth has been driven overwhelmingly by a combination of natural population growth Built-up area, Absolute increase, Percentage and reclassification of rural areas as urban. Type of place 2000 (sq. km) 2000–14 (sq. km) increase, 2000–14 Natural growth has made the larger contribu- Multidistrict metro tion in multidistrict metro cores and single- district metro areas, and reclassification in   Jakarta core 591.3 20.9 3.5 rural peripheries and nonmetro urban and   Jakarta periphery 2,305.5 210.4 9.1 rural areas.   Other metro cores 1,072.8 165.6 15.4 Alongside the reclassification of settle- ments in the rural peripheries and non-  Other urban 1,621.3 572.5 35.3 metro rural areas as urban, there have been peripheries large increases in the amount of built-up   Rural peripheries 1,060.4 667.1 62.9 area between 2000 and 2014 (table 1.3). Single-district metro 434.1 67.7 15.6 Total built-up area increased almost 63 per- cent in the rural peripheries and nearly Nonmetro 45  percent in nonmetro rural areas. The  Urban 2,157.6 871.6 40.4 expansion of built-up area has been more  Rural 6,562.0 2,921.8 44.5 limited in the urban peripheries of multi- district metro areas, which were already Source: Calculations based on European Commission Global Human Settlement Layer (GHSL) data downloaded from http://ghsl.jrc.ec.europa.eu/datasets.php. more built up in 2000 and where urban Note: Places are defined using the methodology in box 1.5. population growth has been driven more by migration into already urban areas. with rapid urban population growth in Finally, the expansion of built-up area has Jakarta’s periphery can  be explained by been most restricted in the already highly strong net inward migration into the built-up multidistrict metro cores. The lim- (already highly urbanized) periphery from ited expansion of built-up area within the other places. This is consistent with a more Jakarta metro area (map  1.2, panel a) 64  TIME TO ACT stands in vivid contrast to the expansion MAP 1.2  Slow versus fast: Built-up areas expanded within the Mojokerto metro area, which more slowly in the already built-up Jakarta metro has a much more rural periphery (map 1.2, area than in the Mojokerto metro area, between panel b). 2000 and 2014 All these patterns offer insights into the wider story of Indonesia’s slowing pace of a. Jakarta metro area urbanization since the beginning of the cen- tury and the associated low overall contribu- tion of migration to urban population growth. That low contribution could be Jakarta Utara Jakarta Barat Jakarta Pusat attributable either to mounting congestion Jakarta Timur forces within cities, which exert a negative Jakarta Selatan influence on prosperity, inclusiveness, and livability, or to more rapidly improving con- ditions in rural areas. The migration data suggest that, in aggregate, the metro cores District classi cation are straining under the weight of congestion Metro core Metro periphery forces and, as a consequence, are failing to Single-district metro Nonmetro district attract population. At the same time, GHSL Built up 2000–2014 although conditions may be improving and Built up before 2000 0 7.5 15 22.5 km structural transformation may be occurring b. Mojokerto metro area in rural areas, this is not happening quickly enough to make urban settlements in other- wise predominantly rural areas attractive to  potential migrants. Meanwhile, single-­ district metro areas and nonmetro urban areas, where congestion pressures are likely Kota Mojokerto less severe than in multidistrict core areas, retain some modest pull for migrants. Drivers of urban population growth vary across multidistrict metro areas District classi cation Metro core Metro periphery Underlying the aggregate picture of slight net Single-district metro Nonmetro district out-migration from multidistrict metro cores GHSL Built up 2000–2014 0 7.5 15 22.5 km (figure 1.12), which is suggestive of strong and Built up before 2000 mounting congestion forces, is a diverse pat- tern of forces across multidistrict metro areas. Source: Based on European Commission Global Human Settlement Layer (GHSL) data downloaded from http://ghsl.jrc.ec.europa.eu/datasets.php. For the 11 most populous multidistrict met- ros, migration made a negative contribution to the urban population growth of six of their cores (Yogyakarta, Surakarta, Surabaya, almost completely offset natural population Medan, Malang, and Bandung; table 1.4). In growth of 46,446 people.32 In the other five general, this negative contribution was more major metro cores (Jakarta, Semarang, than offset by natural population growth— Sukabumi, Makassar, and Denpasar), migra- meaning that population in the cores still grew tion made a positive contribution to urban between 2000 and 2010. An exception was population growth. However, only in Yogyakarta, whose core population fell by Denpasar did this contribution (66 percent) 8,084 to 388,627. Surakarta’s core narrowly exceed 20 percent. For Jakarta’s metro core, escaped overall population decline because of the contribution was just 4  ­p ercent. The net out-migration of 37,323 people that attractiveness of Denpasar—very much the P a t t e r n s o f U r b a n i z a t i o n a n d S t r u c t u r a l T r a n s f o r m a t i o n   65 TABLE 1.4  Urban population growth and its components in major multidistrict metro areas, 2000–10 CORE PERIPHERY % share of growth % share of growth Multidistrict Growth Natural population Net Growth Natural population Net metro area rate growth migration Reclassification rate growth migration Reclassification Yogyakarta –0.2 356 –456 0 3.2 29 29 41 Surakarta 0.2 509 –409 0 2.7 69 –16 47 Surabaya 0.7 191 –91 0 3.3 35 34 30 Medan 1.1 152 –52 0 3.6 42 16 42 Malang 1.1 97 –23 26 3.8 27 8 65 Bandung 1.2 119 –19 0 4.8 30 21 49 Jakarta 1.4 96 4 0 7.6 25 43 32 Semarang 1.9 62 11 27 5.6 39 –6 67 Sukabumi 2.2 68 10 22 7.0 17 7 77 Makassar 2.4 68 20 12 6.4 21 31 48 Denpasar 4.4 34 66 0 2.7 18 42 40 Sources: Calculations for the growth rate of urban population based on data from the 2000 February and 2010 July rounds of Indonesia’s National Socio-Economic Survey (SUSENAS). Figures for natural population growth, net migration, and reclassification are based on Wai-Poi et al. 2018, using data from the 2000 and 2011 Survey of Village Potential (PODES) and the 2000 and 2010 population censuses. Note: Table is sorted in ascending order of urban population growth rate in the cores. Growth rates are compound annual growth rates of urban population during 2000–10. Components of urban population growth are based on direct estimates of the contributions of natural population growth and reclassification and indirect estimates of the contribution of net migration. For a given area, the components of urban population growth sum to + 100 percent (once rounding is accounted for) where such growth is positive, and – 100 percent where such growth is negative. For a summary of the growth decomposition methodology, see annex 1B. outlier among metro cores in its ability to multidistrict metro peripheries, however, the attract net inward migration—can probably contribution of migration to urban popula- be explained, at least in part, by the strength tion growth was modest (Medan, Malang, of its tourism economy (Wai-Poi et al. 2018).33 and Sukabumi) or even negative (Semarang Although, overall, there is strong net inter- and Surakarta). This indicates quite dramatic nal in-migration into the peripheries of multi- variation in the attractiveness of metro district metro areas, beneath that picture is a peripheries to migrants. diverse pattern across peripheries. Migration contributed more than 30 percent to urban population growth only in the peripheries of Looking forward Jakarta, Denpasar, Surabaya, and Makassar. If Indonesia continues to urbanize in line Because Jakarta’s periphery accounted for with global historical standards, more than 55 percent of the aggregate population of all 70 percent of its population will be living urban peripheries in 2016, it dominates the in towns and cities by the time the ­country overall picture (see table 1.4). Migration also celebrates the centenary of its indepen- contributed large shares to urban population dence in 2045 (­ fi gure 1.13, panel a). This growth in Yogyakarta (29  percent) and trend means that Indonesia’s urbaniza- Bandung (21 percent). In the other five tion will be almost complete within the 66  TIME TO ACT FIGURE 1.13  Indonesia’s projected path of urbanization, up to 2045 a. Projected pace of urbanization b. Projected levels of urbanization by island-region 10 80 Annual growth rate of urban share of population (%) y = –1.31 ln(x) + 6.05 R2 = 0.36 8 5.2 5.4 6 60 6.7 6.9 Urban share (%) 4 6.6 7.1 6.7 7.2 7.2 2015–20 40 6.8 2 6.8 6.8 2030s 60.8 2040–45 2020s 0 43.5 40.2 20 35.0 31.6 31.3 –2 –4 0 0 10 20 30 40 50 60 70 80 90 100 Jawa-Bali Kali- Sumatera Sulawesi Nusa Maluku- Initial urban share of population (%) mantan Tenggara Papua All countries Indonesia, projected 2016 Increase, 2016–30 Increase, 2030–45 Indonesia, historical Fitted line (All countries) Source: Calculations based on projections from the United Nations World Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). Note: In panel a, each light blue or (for Indonesia) red marker shows the average growth rate of a country’s urban population share over a given period for its urban population share at the beginning of that period. The periods considered are 1950–60, 1960–70, 1970–80, 1980–90, 1990–2000, 2000–10, and 2010–15, so the figure contains seven observations for each of 231 countries. The dark blue markers also indicate data points for Indonesia but are based on the country’s projected path of urbanization between 2015 and 2045. Projections in panel b of the figure have been calculated assuming that the path of urbanization for each island-region obeys the dynamics implied by the fitted line in panel a of the figure. This implies a faster pace of urbanization for the less urbanized island-regions than for the more urbanized island-regions. FIGURE 1.14  Projected absolute next three decades. By 2045, the level of increase in Indonesia’s urban urbanization will also have reached almost population, 2016–45 50  ­p ercent even for the island-regions of Nusa Tenggara and Maluku-Papua 46 44.8 f igure  1.13, panel b). The number of (­ Urban population increase 44 people living in the country’s urban areas 42 will increase by a projected 83 m ­ illion between 2016 and 2045 (­ figure 1.14). How (millions) 40 Indonesia manages this continued expan- 38.4 38 sion of its urban population—and the mounting congestion forces that expan- 36 sion brings—will do much to determine 34 whether the country reaches the upper 2016–30 2030–45 rungs of the global ladder of prosperity, Source: Calculations based on projections from the United Nations World inclusiveness, and livability. Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   67 Annex 1A BPS’s composite scoring system for identifying urban settlements BPS has evaluated whether a territory is clas- telephone and electricity), primary and sec- sified as urban at the Admin-4 level (the desa/ ondary schools, markets, shops, cinemas, hos- kelurahan level) on the basis of a composite pitals, hotels, and personal and entertainment scoring system that considers population den- services including billiard salons, hair salons, sity, percentage of households employed in and massage parlors (table 1A.1). Desa/kelu- agriculture, and the presence of a specified list rahan receiving total scores of 10 or higher of urban facilities—infrastructure (such as are classified as urban. TABLE 1A.1  Scoring system for urban/rural classification Population density Percentage of households (per sq. km.) Score employed in agriculture Score Urban facilities Criteria Score < 500 1 > 70 1 Primary school Within boundary 1 or <= 2.5 km 500–1,249 2 50–69.99 2 Lower secondary school > 2.5 km 0 1,250–2,499 3 30–49.99 3 Upper secondary school 2,500–3,999 4 20–29.99 4 Market Within boundary 1 or <= 2 km 4,000–5,999 5 15–19.99 5 Shops > 2 km 0 6,000–7,499 6 10–14.99 6 Cinema Within boundary 1 or <= 5 km 7,500–8,499 7 5–9.99 7 Hospital > 5 km 0 > 8,500 8 <5 8 Hotel/Billiards/Discotheque/ Within boundary 1 Massage parlor/Salon Not available 0 Percentage of households >= 8 1 with telephone <8 0 Percentage of households >= 90 1 with electricity < 90 0 Source: Based on Badan Pusat Statistik (Statistics Indonesia), Peraturan Kepala BPS 37/2010. 68  TIME TO ACT Annex 1B Urban population growth decomposition methodology The absolute change in urban population moved from another district and as out- over the intercensal period 2000–10 can be migrants if they moved away. This results in decomposed into three components: two problems. First, people who moved into the district between 2000 and 2005 will not ∆(Urban population) = (Natural population be counted because the question captures growth) + (Net migration) + (Reclassification) only movement in the last five years. To  address this gap in the data, total net Where Natural population growth is the migration for 2000–10 can be estimated to natural growth of the preexisting urban pop- be double that for 2005–10; however, if ulation, Net migration is net immigration into migration is trending upward or downward, urban settlements from rural settlements, and this method will, respectively, overestimate Reclassification is population change associ- or underestimate net migration over the ated with the reclassification of rural settle- decade. Second, the data indicate whether ments as urban settlements by BPS as they people have moved into an urban or a rural pass a score of 10 based on its composite part of a district, but they do not indicate scoring system for classifying settlements (see whether people have departed from an urban annex 1A). Decomposing urban population or rural area. growth into these three components requires To address this issue, Wai-Poi et al. (2018) direct estimates of at least two of the compo- developed a direct measure of reclassifica- nents. With those estimates plus data on the tion from rural to urban that produces more absolute change in urban population, the confident estimates. The Survey of Village remaining component can be indirectly esti- Potential (PODES) classifies villages as rural mated as the residual. For example, Net (desa) or urban (kelurahan). The 2001 and migration can be calculated as ∆ ( Urban 2010 PODES were merged into the 2000 ­p o p u l a t i o n ) – ( N a t u ra l p o p u l a t i o n and 2010 population censuses, respectively, growth) – (Reclassification). at the village level. These resultant datasets Direct estimates are available for natu- were then matched with each other at the ral population growth and net migration, village level to see which villages had been but net migration is imperfectly estimated. reclassified from rural to urban during the That means that indirect estimates of decade. This required aggregating the 80,000 reclassification are also imperfect. The villages in 2010 back to the 70,000 villages total change in urban population can be under the 2000 definitions, because many directly estimated as the difference in villages and districts have been split since urban population between two censuses. Indonesia’s “big bang” decentralization The census also captures births and deaths, reforms (see chapters 5 and 6). Finally, the so well-established demographic methods population living in each district in an area can be used to calculate fertility and mor- that had been reclassified from rural to tality rates, which are then used to directly urban was calculated to provide a direct estimate natural urban population growth measure of reclassification. in each district. Having a direct measure of reclassification The direct estimate of net migration is not only improves confidence in this estimate imperfect because the census asks only where but also allows systematic indirect estimates a person was born and in which district he of the other two drivers to more accurately or she lived five years ago, which means that calculate net migration. As discussed people can be classified as recent migrants above,  only two components require (that is, moved in the last five years) if they direct estimates: the total change in urban P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   69 population is available, and the third compo- component is indirectly estimated each time. nent is indirectly estimated as the residual The degree of similarity between the direct (the difference between total urban popula- and indirect estimates for each component tion change and the two directly estimated provides a basis for confidence (or lack components must be the third component). thereof) in the final estimates. This helps Indirect estimates of all three components address concerns about the direct migration can be derived by substituting which estimate. 70  TIME TO ACT Annex 1C Multidistrict and single-district metro areas in Indonesia, by island-region, sorted by metro area total population TABLE 1C.1  Multidistrict and single-district metro areas in Indonesia Population Number of Percent Core districts Periphery Core Total Urban urban Jawa-Bali DKI Jakarta 14 Kota and Kab. Bekasi; Kota and Kab. 10,076,332 30,962,372 28,627,190 92.5 Tangerang; Kota Tangerang Selatan; Kota and Kab. Bogor; Kota Depok Bandung 4 Kota Cimahi; Bandung; Bandung Barat 2,470,336 8,121,474 6,962,478 85.9 Surabaya 3 Sidoarjo; Gresik 2,833,881 6,167,832 5,476,349 88.9 Surakarta 5 Sukoharjo; Klaten; Karanganyar; Boyolali 509,915 4,325,759 2,690,286 62.2 Semarang 3 Demak; Kendal 1,674,470 3,713,708 2,496,601 67.2 Malang 3 Kota Batu; Malang 845,575 3,569,742 2,227,844 62.5 Sukabumi 2 Sukabumi 314,872 2,735,825 1,328,911 48.9 Denpasar 5 Badung; Gianyar; Tabanan; Klungkung 862,179 2,581,421 1,945,334 75.6 Yogyakarta 3 Bantul; Sleman 407,247 2,552,038 2,261,131 88.9 Pasuruan 2 Pasuruan 193,227 1,761,658 879,423 50.0 Magelang 2 Magelang 120,383 1,353,586 466,733 34.5 Probolinggo 2 Probolinggo 205,738 1,337,679 587,991 43.8 Blitar 2 Blitar 136,845 1,277,350 568,448 44.6 Mojokerto 2 Mojokerto 124,624 1,194,071 641,327 53.8 Salatiga 2 Semarang 181,057 1,167,503 574,426 49.2 Sumatera Medan 5 Deli Serdang; Serdang Bedagai; 2,189,955 5,192,973 4,316,753 82.2 Kota Binjai; Kota Tebing Tinggi Palembang 1 1,556,692 1,556,692 1,540,981 99.0 Pekanbaru 1 1,010,371 1,010,371 991,496 98.1 Bandar Lampung 1 959,224 959,224 949,563 99.0 Padang 1 890,937 890,937 853,321 95.8 Banda Aceh 2 Aceh Besar 248,663 633,357 358,009 56.4 Bukittinggi 2 Agam 120,344 592,953 276,956 46.8 Jambi 1 567,450 567,450 550,425 97.0 Sulawesi Makassar 4 Gowa; Takalar; Maros 1,428,228 2,757,084 1,837,228 66.7 Kalimantan Pontianak 4 Pontianak; Mempawah; Landak 597,449 1,741,065 849,784 48.8 Banjarmasin 2 Barito Kuala 665,470 959,287 714,406 74.3 Samarinda 1 805,284 805,284 757,523 94.1 Balikpapan 1 610,741 610,741 576,559 94.4 Source: Population data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). Note: There are two types of district in Indonesia: kota and kabupaten. Kota translates as “city”; kapubaten designates what has traditionally been considered a rural district. All the metro cores are kota districts. Unless specified as kota, all the peripheral districts are kabupaten. District names and number of districts are based on 2016 administrative boundaries. DKI Jakarta = Daerah Khusus Ibukota Jakarta. Kab = kabupaten. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   71 Notes Prospects: 2018 Revision database (https:// esa.un.org/unpd/wup/), Indonesia is home to 1. Urban population figures in this paragraph approximately 3.5 percent of the population are based on data from the United Nations globally that lives in officially defined urban World Urbanization Prospects: 2018 revision areas. database (https://esa.un.org/unpd/wup/). 11. Indonesia consists of more than 17,000 2. Indonesia’s total population in 2018 was esti- islands. This report’s grouping of these into mated as just shy of 267 million. island-regions follows World Bank (2012). 3. A megacity is defined as a city whose population 12. According to census data, 59.6 percent of is at least 10 million. On the basis of the meth- Indonesians lived on Jawa-Bali in 2010. odology outlined in this chapter, the Jakarta 13. Because we calculate average annual growth metro area had an estimated population of rates, the growth rates for the 2010–15 inter- 30.96 million in 2016 (see also annex 1C). vals can be safely compared with those for 4. Based on historical GDP data from the 2018 the earlier 10-year intervals. version of the Maddison Project Database 14. See, for example, https://bluenotes.anz.com​ (Bolt et al. 2018) (https://www.rug.nl/ggdc​ /metropolis-now/part3.html. /historicaldevelopment/maddison/releases​ 15. This slowdown happens because the share /maddison-project-database-2018). of a country’s population that lives in urban 5. As box 3 in the introductory chapter and areas is bounded above at 100 percent. box 3.2 in chapter 3 document, positive 16. Indonesia’s slowing pace of urbanization is agglomeration forces include “traditional” also evident from calculations of the abso- agglomeration economies, human capital lute percentage point change in the level of externalities, and market access effects. urbanization. Indonesia’s urban share of 6. Roberts et al. (2017) identify four criteria the population jumped by 8.48 percentage that countries commonly use to define settle- points between 1980 and 1990 and by 11.42 ments as urban: population size, population percentage points between 1990 and 2000. density, the existence of “urban” services or The jump then declined to 7.91 percentage characteristics, and the composition of local points between 2000 and 2010, and to 3.40 economic activity. Countries vary in the percentage points between 2010 and 2015 number of these criteria that they use and the (which is equivalent to a 10-year jump of thresholds that they apply to population size, 6.80 percentage points). population density, and the composition of 17. The decompositions of urban population economic activity. The most commonly used growth presented here and in later parts criterion is population size of a settlement. of this chapter are taken from the back- 7. For brevity, in the remainder of the report, ground paper for this report by Wai-Poi we simply refer to the Ministry of Agrarian et  al. (2018). Wai-Poi et al. present decom- Affairs and Spatial Planning / National Land positions of urban population growth based Agency as the Ministry of Agrarian Affairs on both direct and indirect estimates of the and Spatial Planning. contribution of net rural–urban migration. 8. A detailed comparison of urban areas as This chapter focuses on their preferred set delineated by the three definitions would be of decompositions based on the indirect a worthy subject of future research. One may estimates of the contribution of migration. hypothesize that the definitions coincide bet- A summary of Wai-Poi et al.’s decomposition ter for larger and more established urban cen- methodology is in annex 1B. ters, but this requires empirical verification. 18. Consistent with this finding, Firman (2017) 9. The World Development Report 2009 defines notes that the number of urban localities in areas of intermediate urbanization as those Jawa increased by about 30 percent, from in which roughly 50 percent of the popula- 7,510 to 9,239, over the period 2000–10. tion lives in urban settlements. These are to This increase, he observes, reflects “the in situ be contrasted with areas of incipient urban- urbanization process in Jawa, resulting from ization and areas of advanced urbanization gradual changes from rural to urban locali- where roughly 25 percent and 75 percent or ties.” (Firman 2017, 54). greater of the population, respectively, lives 19. In 2016, the average population density for in urban settlements (World Bank 2008). Jawa-Bali was 1,093 people per square kilo- 10. According to the latest estimates from meter. Nusa Tenggara, Sumatera, Sulawesi, and the United Nations World Urbanization Kalimantan followed with average population 72  TIME TO ACT densities of 153, 118, 101, and 29 people per between multidistrict cores and other types square kilometer, respectively. Finally, Maluku- of place could also be overstated if there Papua was the island-region with the lowest are firms that have their headquarters in the population density of 14 people per square cores but also operate plants elsewhere. This kilometer. will be the case if the GDP generated by the 20. Figures on the natural rate of population plants is attributed to the headquarters. increase and on the median age of the pop- 31. Consistent with this, Roberts (2018) reports ulation are from the United Nations World a negative partial correlation between a Population Prospects 2018 database (https:// country’s level of real GDP per capita and a esa.un.org/unpd/wpp/). Chapter 8 discusses weighted measure of average urban popula- policies and issues relating to urban planning tion density, meaning that, globally, richer and design to help ensure the inclusiveness countries tend to have less dense cities. of the elderly in the prosperity and livability 32. Between 2000 and 2010, Surakarta’s core benefits of urbanization. population grew from 490,214 to 499,337— 21. There are two types of district in Indonesia: an overall increase of 1.9 percent (or 0.18 kota and kabupaten. Kota translates as percent a year). “city”; kabupaten designates what has tradi- 33. In 2015, tourism-related industries accounted tionally been considered a rural district. for nearly 50 percent of employment in the 22. The identification of the metro cores is robust core of the Denpasar metro area, compared to other selection criteria, such as a district’s with about 40 percent in 2007. status as a kota or the share of a district’s population that is urban. The exception is the multidistrict metro area of Medan, which, using 1996 administrative boundaries, con- References sists of four districts, three of which are Bolt, J., R. Inklaar, H. de Jong, and J. L. van kota and two of which have entirely urban Zanden. 2018. “Rebasing ‘Maddison’: New populations. Income Comparisons and the Shape of Long- 23. For brevity, we may refer to predominantly run Economic Development.” Working Paper urban and predominantly rural areas as 10, Maddison Project, Groningen Growth and simply urban and rural areas. However, it is Development Centre, University of Groningen, important to remember that a predominantly Groningen, Netherlands. https://www.rug.nl​ urban (rural) area may still have a significant /ggdc/historicaldevelopment/maddison/research. rural (urban) population. Bosker, M., U. Deichmann, and M. Roberts. 2018. 24. Thirty-two of 57 nonmetro urban areas are “Hukou and Highways: The Impact of China’s kota. Spatial Development Policies on Urbanization 25. Of 377 nonmetro rural areas, 354 are and Regional Inequality.” Regional Science and kabupaten. Urban Economics 71: 1–162. 26. The Jakarta metro area as defined in this Dijkstra, L., and H. Poelman. 2014. “A Harmonised report, using the methodology described in Definition of Cities and Rural Areas: The New box 1.5, corresponds perfectly to the official Degree of Urbanization.” Regional Working Jabodetabek metropolitan area. Paper, Directorate-General for Regional and 27. Based on 2016 district administrative Urban Policy, European Commission, Brussels. boundaries. Duranton, G. 2015. “A Proposal to Delineate 28. A full list of multi- and single-district metro Metropolitan Areas in Colombia.” Desarrollo y areas and their associated characteristics is in Sociedad 6 (75): 223–64. annex 1C. Ellis, P., and M. Roberts. 2016. Leveraging 29. Where we do not explicitly separate Jakarta Urbanization in South Asia: Managing Spatial from other multidistrict metro areas, this is Transformation for Prosperity and Livability. typically because of data limitations. Washington, DC: World Bank. 30. More accurately, it will not exist to the same Firman, T. 2017. “The Urbanization of Jawa, degree given that there may still be some 2000–2010: Toward ‘the Island of Mega-urban commuting flows between nonmetro areas Regions.’” Asian Population Studies 13 (1): 50–66. and metro cores, though these will not meet Henderson, J. V. 2000. “The Effects of Urban the 7.5 percent commuting flow threshold Concentration on Economic Growth.” NBER required for them to be part of a multidistrict Working Paper 7503, National Bureau of metro area. Differences in GDP per capita Economic Research, Cambridge, MA. P atter n s of U rba n i z atio n a n d S tructural T ra n sformatio n   73 Henderson, J. V., D. Nigmatulina, and S. Kriticos. Bar for Productive Cities in Latin America and 2018. “Measuring Urban Economic Density.” the Caribbean, edited by M. M. Ferreyra and M. Discussion Paper No. 1569, Centre for Roberts, 89–115. Washington, DC: World Bank. Economic Performance, London School of Roberts, M., B. Blankespoor, C. Deuskar, Economics and Political Science, London. and B. Stewart. 2017. “Urbanization and Jedwab, R., and D. Vollrath. Forthcoming. “The Development. Is Latin America and the Urban Mortality Transition and Poor Country Caribbean Different from the Rest of the Urbanization.” American Economic Journal: World?” Policy Research Working Paper 8019, Macroeconomics. World Bank, Washington, DC. Lall, S. V., H. Selod, and Z. Shalizi. 2006. “Rural– Wai-Poi, M., H. Alatas, K. Chandrashekar, and Urban Migration in Developing Countries: A J. Lain. 2018. “The Different Faces of Urban Survey of Theoretical Predictions and Empirical Indonesia: Recent Urban Trends in Indonesia.” Findings.” Policy Research Working Paper Background paper for this report, World Bank, 3915, World Bank, Washington, DC. Washington, DC. OECD (Organisation for Economic Co-operation World Bank. 2008. World Development Report and Development). 2013. “Definition of 2009: Reshaping Economic Geography. Functional Urban Areas (FUA) for the OECD Washington, DC: World Bank. Metropolitan Database.” ———. 2012. Indonesia: The Rise of Metropolitan Park, J., and M. Roberts. 2018. “A New Typology Regions—Towards Inclusive and Sustainable of Districts for Indonesia.” Background paper Regional Development . Washington, DC: for this report, World Bank, Washington, DC. World Bank. Pradhan, K. C. 2013. “Unacknowledged Wo r l d B a n k a n d D RC ( D e v e l o p m e n t Urbanization: The New Census Towns of Research Center of the State Council of the India.” Center for Policy Research, New Delhi. People’s Republic of China). 2014. Urban Roberts, M. 2018. “The Many Dimensions of China:  Toward Efficient, Inclusive, and Urbanization and the Productivity of Cities in Sustainable Urbanization. Washington, DC: Latin America and the Caribbean.” In Raising the World Bank. 74  TIME TO ACT Is Urbanization Delivering? 2 KEY MESSAGES As in most countries, urban areas in Indonesia strong congestion forces. About one-fifth of are associated with higher incomes and lower the urban population lives in slums, mostly in poverty. This is especially true in multi- and urban peripheries and nonmetro urban areas. single-district metro areas. Strong and mount- •  Economic inequality is higher in urban areas, ing congestion forces, however, limit the full especially more populous ones. Disparities in realization of urbanization’s potential to deliver income and access to services between urban prosperity, inclusiveness, and livability through and rural places remain significant, but they positive agglomeration forces. have diminished. As a result, most economic inequality today is driven by inequality within •  Multidistrict metro cores and single-district places. Tackling economic inequalities metro areas offer higher-paid formal requires policies to accelerate connectivity employment opportunities and better access and integration. to services. They are associated with higher •  In the past 20 years, the process of urbaniza- incomes, fewer poor and vulnerable people, tion—measured by increases in the share of and a larger middle class than nonmetro the population living in urban areas—has not rural areas. been associated with faster economic growth •  People living in multidistrict metro cores or faster poverty reduction. This reflects partly enjoy better access to basic amenities and are the weight of mounting congestion forces and therefore healthier and better educated than partly the faster progress in  raising living those residing in rural areas, but they face standards in nonurban areas. 75 What should urbanization prosperity has been inclusive across people and places. Next, it focuses on whether urban deliver? areas provide a good quality of life to all resi- Cities, and more generally urban areas, are dents, relative both to rural areas in Indonesia the economic, political, and social pulse of a and to urban areas elsewhere in the world. country. As people and firms cluster, they start Where the data permit, the analysis bench- to share amenities and suppliers, find work marks Indonesia’s performance against that and workers more easily, and exchange ideas of peer countries to assess whether and knowledge more frequently. Along with Indonesia’s urbanization experience con- the better access to markets that tends to go forms with other international experiences. together with a city’s growth, these mecha- The chapter finds that urban areas in nisms of “sharing, matching and learning” Indonesia are indeed more prosperous and (Duranton and Puga 2004) give rise to provide better access to services than rural agglomeration forces that lead to higher pro- areas. Especially in metro cores, per capita ductivity, economic growth, and, ultimately, incomes are higher, poverty rates are lower, prosperity.1 Because of economies of scale in and various basic services, from sanitation to building infrastructure, cities can also provide health and education, are more readily acces- more and better services than rural areas, sible. Urbanization has also not left rural including in health and education, and more areas behind: gaps between districts, both in leisure options. This is the promise of income and service access, have declined. urbanization. However, disparities remain large, both However, not all urban areas manage to between places and between people, as provide a good quality of life to most resi- inequality within places is large and increas- dents. As cities become more crowded, ing. Moreover, some evidence since the late many cannot fully support basic needs for 1990s indicates that growing congestion drinking water and sanitation. Land and forces have limited the potential of urbaniza- housing markets can become strained, tion to propel growth and that urban areas leading to the emergence of slums. Urban in Indonesia also tend to underperform areas are also more likely to suffer high regional comparators. Improving the man- levels of air and water pollution from the agement of these negative congestion forces concentration of industrial activity and could help Indonesia to grow faster. If urban- vehicle use and are potentially more vul- ization is well managed, in 2045, when an nerable to the harmful effects of climate estimated 70 percent of Indonesians will live change and natural disasters (see spotlight in urban areas, Indonesia’s gross domestic 1 on strengthening the disaster resilience product (GDP) per capita could reach that of of Indonesian cities). Commuting can be Hungary (about $27,000 in constant 2011 lengthy, often in crowded or dangerous international dollars), which currently has a forms of transportation. These negative similar urban share. congestion forces reduce livability and also affect businesses and overall productivity. Urbanization may also contribute to higher Is urbanization in Indonesia inequality, 2 because skilled workers con- delivering prosperity for all? centrate in cities and returns to scarce skills and capital rise faster in more pro- Indonesia’s per capita income and ductive areas (see chapter 4). poverty rate are in line with its level of urbanization This chapter assesses whether urbaniza- tion in Indonesia has delivered prosperity Globally, living in urban areas goes hand in that is inclusive of all, and to what extent hand with economic prosperity: countries congestion forces have been mitigated to with a higher share of the population living in make Indonesian cities livable. First, it con- urban areas tend to have higher incomes, as siders the extent to which urban areas in measured by real GDP per capita (figure 2.1, Indonesia have prospered and whether this panel a), and lower shares of people living in 76  TIME TO ACT FIGURE 2.1  More urbanized countries tend to have higher levels of income and lower levels of poverty a. GDP per capita, 2012 b. Share of population living on $3.20 or less per day, 2015–16 12 100 Log, constant 2011 international dollars y = 0.043x + 6.51 y = –0.86x + 84.24 R² = 0.54 R² = 0.36 11 80 10 China 60 Share (%) 9 40 8 Indonesia Indonesia Vietnam 20 7 Vietnam China 6 0 20 40 60 80 100 0 20 40 60 80 100 Urban population share, 2012 (%) Urban population share, 2012 (%) Sources: The globally consistent calculation of the urban population share is based on application of the Uchida and Nelson 2010 agglomeration index to LandScan-2012 gridded population data. GDP per capita is from World Bank World Development Indicators (http://datatopics.worldbank.org/world​ -development-indicators/). Poverty rates are from World Bank PovcalNet (http://iresearch​.worldbank.org/PovcalNet/povOnDemand.aspx). Note: The bands colored in green in panel a indicate the 95 percent confidence interval. The data for panel b exclude high-income countries that have nearly zero poverty. GDP = gross domestic product. poverty (figure 2.1, panel b).3 On average, Within Indonesia, urban areas are more a 1 percent increase in the level of urbaniza- prosperous than rural areas, but not all tion is associated with a 4 percent increase in urban areas are alike GDP per capita and an almost 1 percent The relationship between levels of urbanization reduction in poverty. and income per capita in Indonesia is c ­ onsistent However, for a given level of urbaniza- with the cross-country pattern. Districts that tion, there is wide variation. Some countries have a larger share of the ­ population l­iving in are richer than expected given their level of urban areas tend to be richer, as m ­ easured by urbanization, whereas others are poorer. per capita income (­ f igure  2.2, panel a). 4 When placed in a global context, Indonesia’s However, the association is not as strong across per capita income is slightly lower than Indonesian districts as it is across countries. On expected given the share of population living average, a 1 percent increase in the share of the in urban areas. In the East Asia and Pacific population l­iving in urban areas is associated context, Indonesia fares less well. Although percent increase in income per capita in with a 1 ­ China and Indonesia have similar urban Indonesia, ­compared to a 4 ­ percent increase population shares, China’s per capita income across countries. is 20 percent higher than that of Indonesia. This positive relationship between urban- Wi t h r e s p e c t t o p o v e r t y r e d u c t i o n , ization and income per capita holds for differ- Indonesia’s poverty rate is also about as ent measures of economic prosperity, such as expected given its urban share in a global household consumption.5 The association context, but several countries such as China strengthens when districts that are reliant on and Vietnam outperform Indonesia on this mining and quarrying activities are excluded score (figure 2.1, panel b). About one-third (figure 2.2, panel b).6 No district with less of Indonesians live on $3.20 (at constant than 80 percent of its population living in 2011 international dollars) or less a day, urban areas enjoyed real per capita income whereas less than one-tenth of the popula- above 80 million rupiah (Rp) (or about tions of China and Vietnam do. I s U rba n i z atio n D eli v eri n g ?   77 FIGURE 2.2  More urbanized districts are richer, especially when mining-intensive districts are removed, 2016 a. All districts b. Excluding mining-intensive districts 7 7 y = 0.0096x + 2.87 y = 0.011x + 2.67 6 R = 0.23 6 R = 0.36 Real GDP per capita (log) Real GDP per capita (log) 5 5 4 4 3 3 2 2 1 1 0 20 40 60 80 100 0 20 40 60 80 100 Urban population share (%) Urban population share (%) Non-mining intensive Mining intensive Sources: Calculations of real GDP per capita based on data from Badan Pusat Statistik (Statistics Indonesia). Urban population share is from Indonesia’s National Socio-Economic Survey (SUSENAS). Note: In panel a, the fitted line captures the relationship between income per capita and urbanization for all districts; in panel b it captures the corresponding relationship for non-mining-intensive districts only. In both panels, definitions of districts are based on their 1996 administrative boundaries. GDP = gross domestic product. FIGURE 2.3  Urban areas contribute Nonetheless, not all urban areas are equally nearly 60 percent of Indonesia’s GDP prosperous, nor are all people within urban Average share of nominal GDP, 2010–16 (%) areas equally well-off. Economic prosperity is concentrated mostly in metro areas, especially Jakarta, which accounts for almost a quarter Rural periphery, 4.4 of the country’s GDP (and 12 percent of the population). Except for the cores of multidis- Jakarta trict metro areas (metro cores) and for single- core, 16.4 district metro areas, which account for a small Jakarta periphery, 7.2 Nonmetro share of the population, all other types of place rural, have per capita incomes substantially below 38.8 Other metro core, 10.8 the national average (­ figure 2.4). There are also striking disparities between the cores and Other urban periphery, 6.7 the peripheries of multidistrict metro areas: the Single-district metro, per capita income of Jakarta’s metro core is Nonmetro about five times that of its periphery and twice 3.6 urban, 12.2 that of other metro cores. This gap, however, may be slightly overstated given that many Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). people in Jakarta’s periphery commute to the Note: Places are defined using the methodology described in box 1.5, chapter 1. core for work (see also chapter 1). The positive relationship between levels of urbanization and income is partly due to dif- US$5,680) in 2016.7 Moreover, despite being ferences in the economic structure of urban home to only half the country’s population, and rural areas. The economies of more urban- urbanized districts8 contributed 57 percent of ized districts rely more on industry and services Indonesia’s GDP over 2010–16, up from compared to more rural districts (­ figure 2.5). 54 percent over 1996–2003 (figure 2.3). About 70 percent of total output in metro 78  TIME TO ACT FIGURE 2.4  Economic prosperity is FIGURE 2.5  The higher the level of urbanization, concentrated mostly in metro cores, the larger are industry and services sectors in the especially in Jakarta economy, 2016 Real GDP per capita, 2016 (Rp millions) 100 Share of output from industry and services (%) Jakarta core 90 Jakarta periphery 80 Other metro core 70 Other urban periphery Rural periphery 60 Single-district metro 50 y = 0.39x + 59.55 Nonmetro urban R = 0.69 National real 40 GDP per capita: 0 20 40 60 80 100 Nonmetro rural Rp 56 million Urban population share (%) 0 40 80 120 160 Sources: Calculations of the share of output in industry and services based on data from Badan Pusat Sources: Calculations of real GDP per capita for each type of place based on Statistik (Statistics Indonesia). Urban population share is based on data from Indonesia’s National data from Badan Pusat Statistik (Statistics Indonesia). National real GDP per Socio-Economic Survey (SUSENAS). capita is from World Bank, World Development Indicators (http://datatopics. Note: Industry includes manufacturing, mining and quarrying, electricity/gas/water supply, and construction. worldbank.org/world​-development-indicators/). Note: Places are defined using the methodology described in box 1.5, chapter 1. GDP = gross domestic product; Rp = Indonesian rupiah. FIGURE 2.6  High-end services account for a larger share of output in metro cores and single-district metros than in other places, 2016 cores comes from the services sector, more than double the share in nonmetro rural areas. 100 0 3 1 10 2 8 13 14 19 Manufacturing activities are more prevalent in 12 12 20 28 19 periphery areas of multidistrict metro areas 80 20 13 Share of output (%) and nonmetro urban areas. 38 42 27 30 60 23 25 There is wide variation in economic activity 23 across urban areas of different income levels. 14 40 48 Areas with higher per capita income, metro 19 36 26 37 22 32 cores and single-district metro areas, have a 20 35 16 higher share of output from high-end services 24 22 17 15 than nonmetro urban areas (­ figure 2.6). The 12 9 12 0 four multidistrict metro areas with the highest re ry re y ro n l y ra er r per capita incomes in 2016—Balikpapan, ba he he co co et ru h ur tm rip ip ip ta ro ro er er ro Surabaya, Jakarta, and Banda Aceh (in r et pe et ric ka np lp et rm nm st Ja rta nm ra ba di descending order of per capita income)—have he No Ru ka e- No ur Ot Ja gl r Sin he a higher share of high-end services compared Ot to the average metro area (box 2.1). High-end services Manufacturing Low-end services Nevertheless, across all urban areas, low- Other industry Agriculture value-added activities tend to prevail. Across all areas except for urban peripheries, agricul- Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). ture, other industry, and low-end services con- Note: Places are defined using the methodology described in box 1.5, chapter 1. High-end services comprise finance, real estate, business, transport, communication, and storage services. Low-end tribute a larger share of total output than services are wholesale retail trade and other services. Other industry refers to construction, mining and manufacturing and high-end services. quarrying, and utilities. I s U rba n i z atio n D eli v eri n g ?   79 BOX 2.1  Economic profile of Indonesia’s most prosperous metro areas Of Indonesia’s 21 multidistrict metro areas, (see figure B2.1.1). Compared to the average those with the highest per capita incomes in multidistrict metro area, Balikpapan, Surabaya, Indonesia in 2016 were Balikpapan, Surabaya, and Jakarta tend to engage more in higher-value- Jakarta, and Banda Aceh (in descending order of added activities such as manufacturing and per capita income). These four metro areas were high-end services (transport, storage, communi- among the only ones with per capita incomes in cations, and financial, real estate, and business 2016 above the national average of 55.6 million services). Although Banda Aceh’s manufacturing rupiah. sector is substantially smaller than that of the Economic structure is one reason why some average multidistrict metro, its high-end services metro areas are more prosperous than others sector is larger. FIGURE B2.1.1  Share of nominal GDP, 2016 100 1 2 1 10 8 15 11 13 14 15 80 15 33 39 60 33 Percent 45 48 40 24 29 25 2 20 29 28 21 19 19 0 Balikpapan Surabaya Jakarta Banda Aceh Average multidistrict metro High-end services Manufacturing Low-end services Other industry Agriculture Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). Note: Indonesia’s 21 multidistrict metro areas are identified using the methodology described in box 1.5, chapter 1. Average multidistrict metro excludes Balikpapan, Surabaya, Jakarta, and Banda Aceh. High-end services comprise finance, real estate, business, transport, communication, and storage services. Low-end services are wholesale retail trade and other services. Other industry refers to construction, mining and quarrying, and utilities. GDP = gross domestic product. Job opportunities in urban areas are have higher shares of workers in industry mostly in low-value-added services, but and services (figure 2.7, panel a), which tend urban jobs are more stable and pay better to provide better-paying and more stable jobs than jobs in other places than those in agriculture. However, the domi- Consistent with the prediction that people nance of low-value-added services in urban move away from agriculture into higher- areas is even more apparent in employment value-added activities as rural areas become than it is in output. Whereas high-end ser- urbanized, or as those people migrate from vices account for about 20 percent of output rural to urban areas, Indonesian districts in urban areas on average, they account for with higher urban population shares tend to only 12 percent of jobs. Even in the Jakarta 80  TIME TO ACT FIGURE 2.7  Urban areas employ a higher share of workers in industry and services, but jobs tend to be in lower-value-added services, 2015 a. Share of workers in industry and services b. Share of total workers 100 100 0 2 4 5 3 7 6 17 16 10 32 80 80 45 9 10 9 49 61 60 60 62 Percent Percent 64 8 44 46 40 40 35 33 24 13 20 16 20 y = 0.66x + 33.17 22 20 9 18 R² = 0.84 10 21 16 14 14 5 6 8 8 0 0 0 20 40 60 80 100 l ry ry n ry re re ro ra ba he he he co co et ru Urban population share (%) ur m ip rip rip rta ro ro er ro ict et pe pe et ka lp et rm str nm Ja an rta nm ra di he No rb Ru ka e- No Ot ru Ja gl Sin he Ot High-end services Manufacturing Low-end services Other industry Agriculture Sources: Calculations of the share of workers based on data from Indonesia’s National Labor Force Survey (SAKERNAS; panels a and b). Urban population share is from Indonesia’s National Socio-Economic Survey (SUSENAS; panel a). Note: In panel a, industry includes manufacturing, mining and quarrying, electricity/gas/water supply, and construction. In panel b, high-end services comprise finance, real estate, business, transport, communication, and storage services; low-end services are wholesale retail trade and other services; and other industry refers to construction, mining and quarrying, and utilities. In panel b, places are defined using the methodology described in box 1.5, chapter 1. metro core, where high-end services make up in urban areas associated with positive 35 percent of the economy, only 21 percent agglomeration forces. of workers are employed in these activities (figure 2.7, panel b).9 Jobs in low-value-added services are, how- The process of urbanization in recent years has not been associated with faster growth ever, still better paying and more stable than agricultural and mining jobs available in rural Although urban places tend to be more pros- areas. Urban areas have a higher proportion perous than rural places, the process of urban- of formal jobs (figure 2.8, panel a). ization in the past 20 years does not appear to Furthermore, even within individual sectors, have contributed significantly to greater pros- urban areas, especially metro cores and sin- perity. One piece of evidence that the process gle-district metro areas, provide higher-paying of urbanization over 1996–2016 did not jobs (figure 2.8, panel b). Even after control- deliver as much prosperity as it could have ling for differences in sector, educational level, comes from the fact that urbanization in workforce experience, and other characteris- Indonesia was not associated with the same tics, workers in multi- and single-district level of per capita GDP growth as in China metro areas clearly earn a premium over and other developing East Asia and Pacific those in nonmetro rural areas. As discussed in countries (figure 2.9). Every 1 percent increase chapter 3, this premium can be interpreted in the share of population living in urban as evidence of higher underlying productivity areas was associated with a 1.4 percent I s U rba n i z atio n D eli v eri n g ?   81 FIGURE 2.8  Urban areas tend to offer more opportunities for formal employment and better-paying jobs than rural areas a. Share of total employment, 2017 b. Average wage premium, net of worker and job characteristics, over nonmetro rural areas, 2008–15 100 30 25.4 25 80 33 34 38 25 49 51 20.7 62 20 69 60 Percent Percent 15 40 10.7 75 67 66 62 10 51 49 20 38 5 31 2.4 0 0 e e ro ry ry n ry l et - ry n ry ra m gle r r ba ba he he he he he co co et ru ro ur ur ict in m rip rip ip ip ip rta ro ro str /S er er er ro ro ict et pe pe et ka di core lp np lp et et rm str nm Ja rta an nm nm ra ra ba di he No ro rb Ru Ru ka e- No No Ur et Ot ru Ja gl M Sin he Ot Formal Informal Sources: Calculations based on data from Indonesia’s National Labor Force Survey (SAKERNAS), 2017 August round for panel a and August 2008–August 2015 rounds for panel b. Note: For both panels, places are defined using the methodology described in box 1.5 of chapter 1 and workers are enumerated at their place of work. In panel a, formal employees are defined as salaried permanent workers or employers with at least one worker. In panel b, the reported values are mean levels of estimated wages net of worker and job characteristics for districts that belong to a given type of place, expressed as a percentage difference from the corresponding mean estimated net wage for nonmetro rural districts. These mean levels are estimated by regressing a district’s net wage (in natural log) on a series of dummy variables for the different types of place while controlling for the island- region that a district belongs to. For the methodology to estimate a district’s net wage, see box 3.1 in chapter 3. FIGURE 2.9  Faster urbanization was accompanied by more prosperity in Indonesia, but not as much as in China and other developing countries in the East Asia and Pacific region, 1996–2016 a. China b. Other EAP developing economies c. Indonesia 60 16 60 16 60 16 Thousands of constant 2011 Thousands of constant 2011 14 14 14 Thousands of constant 2011 50 50 50 international dollars international dollars 12 12 12 international dollars 40 40 40 10 10 10 Percent Percent Percent 30 8 30 8 30 8 6 6 6 20 20 20 4 4 4 10 10 10 2 2 2 0 0 0 0 0 0 96 02 08 14 96 02 08 14 96 02 08 14 19 20 20 20 19 20 20 20 19 20 20 20 Urban share (left scale) GDP per capita (right scale) Sources: Calculations based on data from World Bank World Development Indicators (http://datatopics.worldbank.org/world-development-indicators/) and the United Nations World Urbanization Prospects: 2018 Revision database (https://esa.un.org/unpd/wup/). Note: “Other EAP developing economies” refers to East Asia and Pacific countries excluding high-income countries, China, and Indonesia. For all panels, left-hand axis denotes the share of population living in urban areas measured in percent and right-hand axis indicates gross domestic product (GDP) per capita measured in thousands of constant 2011 international dollars. 82  TIME TO ACT annual increase in per capita income in annually on average. Congestion forces, Indonesia, but a 3.0 percent increase in China especially in the Jakarta multidistrict metro and a 2.7 percent increase in other developing area, may be dampening the impact of East Asia and Pacific countries. urbanization on prosperity in these areas. Within Indonesia, districts that urbanized •• Second, rural areas performed relatively faster do not appear to have grown faster well during this period. Between 1996 and figure 2.10). This finding is similar to the (­ 2016, per capita incomes of nonmetro rural finding at the global level that more urbaniza- areas and rural peripheries grew by 3.4 per- tion is associated with higher productivity, cent and 3.3 percent annually on average, but not necessarily with faster economic respectively (­ f igure  2.11). One possible growth (Henderson, Roberts, and Storeygard explanation for this relatively good growth 2013). In Indonesia, two factors may have performance of rural areas is that they ben- contributed to the lack of a relationship. efitted disproportionately from the com- modity boom of 2003–14, helping them to •  First, the economic performance of urban catch up with more urban areas. Another areas has been uneven. Between 1996 and possibility is that rural areas benefitted 2016, per capita incomes of metro cores from spillovers from urbanization through (excluding Jakarta) and single-district economic links such as remittances and metro areas grew by 4.1 percent and greater demand for agricultural goods. 4.7 percent annually on average, respec- figure 2.11), higher than the national tively (­ In more recent years (that is, between 2010 average during this period, 3.1 percent. and 2016 in figure 2.11), rural peripheries Jakarta’s core, in contrast, grew at about have continued to outperform nonmetro rural the national average, as did the urban and even nonmetro urban areas. Better con- peripheries (excluding Jakarta) and non- nectivity may have helped rural peripheries to metro urban areas. Jakarta’s periphery, increasingly access economic opportunities in meanwhile, grew at only about 1 percent metro cores, boosting growth. Because rural districts tend to be poorer FIGURE 2.10  Districts that urbanized and performed relatively well over 1996– faster have not grown faster, 2016, there has been convergence in per cap- 1996–2016 ita incomes across Indonesian districts over time. Districts that were poorer (and largely 9 rural) in 1996 grew faster than richer districts Annual growth rate of real GDP per capita (%) 8 y = 0.011x + 3.47 R = 0.0003 over 1996–2016 (figure 2.12). In other words, 7 incomes across Indonesian districts have con- 6 verged over time, regardless of the districts’ 5 urban share. Districts that are more reliant on mining 4 and quarrying activity show a faster speed of 3 income convergence, as reflected by the 2 steeper fitted line. This result of income con- 1 vergence holds for different sample periods 0 and is robust to alternative measures of –1 income, such as nighttime lights or median –5 0 5 10 15 per capita expenditures. Annual growth rate of urban share (%) Although the growth of rural areas may have dampened the relationship between Sources: Calculations based on data from Badan Pusat Statistik (Statistics urbanization and economic growth over 1996– Indonesia) for the growth rate of real GDP per capita, and Indonesia’s National 2016, more urbanization has also not resulted Socio-Economic Survey (SUSENAS) for the growth rate of the urban share. Note: The data exclude 13 districts with negative gross domestic product in faster structural transformation. From 2007 (GDP) per capita annual growth rates, mostly in Aceh and Papua. Annual to 2015,10 increases in urban shares are associ- growth rates are calculated as compound annual growth rates. ated with increases in the shares of I s U rba n i z atio n D eli v eri n g ?   83 FIGURE 2.11  Per capita incomes have grown fastest in the metro cores and single-district metro areas, whereas other types of urban districts grew more slowly than rural districts, 1996–2016 7 Annual growth rate of per capita income (%) 6 5 4 3 2 1 0 –1 –2 1996–2003 2003–10 2010–16 1996–2016 Jakarta core Jakarta periphery Other metro core Other urban periphery National Rural periphery Single-district metro Nonmetro urban Nonmetro rural averages Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). Note: Annual growth rates are calculated as compound annual growth rates. Places are defined using the methodology described in box 1.5, chapter 1. FIGURE 2.12  Indonesian districts that economy, as evidenced by sectoral output were poorer in 1996 grew faster over figure 2.13, panel b). Even in districts shares (­ 1996–2016 that have not urbanized rapidly, industry and services increased as a share of output. ­ 9 y = –1.46x + 7.74 That workers have moved from agriculture 8 R = 0.26 to industry and services while the share of Annual growth rate of real GDP per 7 these sectors in total output has not risen sug- 6 gests that productivity must be falling in indus- capita, 1996–2016 (%) 5 try and services. Indeed, cross-district analysis 4 suggests that urbanization is not strongly asso- 3 ciated with labor productivity growth as mea- 2 sured by output per worker. Much of the 1 y = –0.65x + 5.1 R = 0.09 growth in services in metro cores and single- 0 district metro areas has been in low-value- –1 added activities, which have absorbed many of –2 0 1 2 3 4 5 6 the workers coming from agriculture. Real GDP per capita, 1996 (log) Non–mining intensive Mining intensive Metro areas are associated with lower poverty, though many people remain vulnerable Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). With better-paying jobs and a larger formal Note: Orange fitted line shows the relationship for non-mining intensive districts; red fitted line shows the relationship for mining-intensive districts. Annual sector, more urbanized districts have higher growth rates are calculated as compound annual growth rates. GDP = gross purchasing power.12 Household consump- domestic product. tion (as measured in household surveys) is highest in metro cores. Median monthly per employment in industry and services across capita expenditure is highest in metro cores districts (figure 2.13, panel a) as expected,11 (Rp 1.2 million, or US$86) and in single-­ but not with changes in the structure of the district metros (Rp 1.1 million, or US$76). 84  TIME TO ACT FIGURE 2.13  Urbanization is associated with increases in the share of workers employed in industry and services across districts, but not with changes in the share of output from industry and services, 2007–15 a. Share of workers in industry and b. Share of output from industry and services services 30 30 y = 0.39x + 5.26 y = 0.023x + 2.56 25 25 Growth of industry and services as Growth of industry and services as R = 0.10 R = 0.00 a share of employment (%) a share of total output (%) 20 20 15 15 10 10 5 5 0 0 –5 –5 –10 –10 –10 –5 0 5 10 15 20 25 30 –10 –5 0 5 10 15 20 25 30 Change in urban share of population (%) Change in urban share of population (%) Sources: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia) for the growth of the shares of workers and of output, and Indonesia’s National Socio-Economic Survey (SUSENAS) for the change in the urban share of population. Note: In both panels, the changes in the shares are measured as percentage points between 2007 and 2015. In panel b, output is measured as nominal gross domestic product (GDP). In both panels, industry includes manufacturing, mining and quarrying, electricity/gas/water supply, and construction. This expenditure is almost double that of ­ ulnerable, double the share in metro cores v nonmetro rural areas (Rp 597,973, or (figure 2.14, panel a). In Jakarta, the situation US$42). is the reverse, with a greater share of poor or Similarly, poverty and vulnerability are vulnerable people in the core. All in all, lowest in metro cores and in single-district 2.7  million poor people and 7.7 million metro areas. Just 2.7 percent of the popula- ­ vulnerable people reside in urban peripheries, tion in metro cores (excluding Jakarta) and including Jakarta’s periphery (­ figure 2.14, 5.0 percent in single-district metro areas live panel b). The situation is even more alarming below the national poverty line, far lower in nonmetro urban areas, where 6.5 million than 14.6 percent in nonmetro rural areas people are poor and 15 ­ million are vulnerable (figure 2.14, panel a). Similarly, the share of to poverty. the population vulnerable to falling back into Significant differences exist in poverty and poverty is also lower in urban areas, though vulnerability across metro areas. In some by a narrower margin. About 11 percent of metro areas, poverty has been virtually eradi- residents in metro cores (excluding Jakarta) cated. In Surabaya, for example, just 1.3 per- and 18 percent in single-district metro areas cent of the population lives below the national are considered vulnerable, compared with poverty line. In contrast, in Surakarta, 28 percent in nonmetro rural areas. Even in Malang, and Yogyakarta, the poverty rate absolute terms, the number of people who remains around 10 percent. are poor or vulnerable in metro cores and single-district metros is low relative to non- metro rural areas (figure 2.14, panel b). More urbanization facilitates faster poverty Nevertheless, significant pockets of p ­ overty reduction, but not as much as in other countries in the region persist in some types of urban places. Excluding Jakarta, about one-third of resi- Urbanization and poverty reduction have dents in the urban peripheries are poor or gone hand in hand in Indonesia, but not I s U rba n i z atio n D eli v eri n g ?   85 FIGURE 2.14  Poverty and vulnerability are lowest in metro cores and single-district metros, even when the number of people who are poor or vulnerable are counted in absolute terms, 2016 a. Share of population that is poor or vulnerable b. Number of people who are poor or vulnerable 29.8 30 27.9 30 26.1 23.4 21.5 20 18.1 18.3 20 15.6 Millions Percent 14.6 14.5 14.9 11.4 11.2 11.2 10 9.1 10 6.5 5.0 4.2 4.8 3.7 3.1 2.7 2.0 2.0 2.9 1.8 1.9 0.3 1.2 0.4 0.6 0.4 0 0 l n ry ry ro re ry re ry l n ro ry re ry re ra ra ba ba he he he he he he co co co co et et ru ru ur ur m m ip rip rip ip rip rip rta ro ro rta ro ro er er ro ro ict ict et et pe pe pe pe et et ka ka lp lp et et rm rm str str nm nm Ja Ja an rta an rta nm nm ra ra di di he he No No rb rb Ru Ru ka ka e- e- No No Ot Ot ru ru Ja Ja gl gl Sin Sin he he Ot Ot Poor Vulnerable Source: Based on Lain 2018, using Indonesia’s National Socio-Economic Survey (SUSENAS). Note: “Poor” refers to people in households that consume below the national poverty line (Rp 350,000 a day); “vulnerable” refers to people who are not poor but live in households that consume below 1.5 times the national poverty line (Rp 530,000 a day). Places are defined using the methodology described in box 1.5, chapter 1. Rp = Indonesian rupiah. FIGURE 2.15  The urbanization process is associated as  strongly as in China and Vietnam with poverty reduction in Indonesia, but not as much figure 2.15). In Indonesia, the 23-percentage- (­ as in China and Vietnam point increase in the urban share of the popu- lation over 1990–2015 was accompanied by 100 a 54-percentage-point reduction in poverty Share of population living on $3.20 or less a day (%) 1990 1990 (at the global $3.20 or less a day poverty line). 13 In China, a 29-percentage-point 80 1992 2002 increase in the urban share was associated 2002 2002 with an 83-percentage-point reduction in 60 poverty. Vietnam achieved a 70-percentage- point reduction in poverty despite increasing its urban share by only 14 percentage points. 40 These reductions in poverty are broadly simi- lar for the global $1.90 a day poverty line. 2015 The weaker relationship between urbaniza- 20 tion and poverty reduction in Indonesia is partly due to the fact that rural poverty has not 2015 2015 declined as quickly as it did in China. Although 0 20 25 30 35 40 45 50 55 60 about 93 percent of the rural population in Urban population share (%) both China and Indonesia lived on under $3.20 a day in 1990, only 13 percent of China Indonesia Vietnam China’s rural population remained poor by this measure in 2015, compared with Sources: Calculations based on data from World Bank World Development Indicators database (http://datatopics.worldbank​.org/world-development-indicators/) and PovCalNet (http://iresearch. 37 ­percent of Indonesia’s rural population. worldbank.org/PovcalNet/povOnDemand.aspx). One possibility is that growth in urban areas 86  TIME TO ACT has not benefitted rural areas in Indonesia as FIGURE 2.16  Poverty and vulnerability decreased much as it has in China, where rural–urban in all types of place, with sharper reductions in migration has made a much stronger contribu- metro peripheries and nonmetro rural areas tion to overall urban population growth. Nonetheless, poverty and vulnerability in 70 62 64 Indonesia declined more in rural places than 60 Share of population that is poor it did in urban places (­ figure  2.16).14 The 51 share of the population that is poor or vulner- 50 44 or vulnerable (%) able declined by 28 percentage points (from 40 39 34 34 62 to 34 percent) between 2001 and 2017 in 29 rural peripheries of multidistrict metro areas, 30 23 21 19 and by 25 percentage points in nonmetro 20 16 rural areas (from 64 to 39 percent). In con- 10 trast, poverty and vulnerability declined by only 7 and 13 percentage points in metro 0 cores and single-­ district metro areas, respec- re ro ry n ry l ra ba he he co et ru tively, although these areas already had much ur m ip ip ro ro er er ro ict et et np lp lower rates to begin with. et M str nm nm ra ba di No Ru e- No The decline in poverty and vulnerability in Ur gl Sin rural areas, together with the areas’ robust 2001 2017 growth performance, suggests that push fac- tors in rural–urban migration have become less Source: Calculations based on data from Indonesia’s National Socio-Economic Survey (SUSENAS). key to the urbanization process. This indica- Note: “Poor” refers to people in households that consume below the national poverty line (Rp 350,000 tion is consistent with the finding that urban a day); “vulnerable” refers to people who are not poor but live in households that consume below population growth in Indonesia is driven 1.5 times the national poverty line (Rp 530,000 a day). Places are defined using the methodology described in box 1.5, chapter 1. Rp = Indonesian rupiah. mostly by natural population growth and reclassification of rural areas as urban, rather than by rural–urban migration (see chapter 1). FIGURE 2.17  Middle-class population shares are the largest in metro areas, 2017 Metro cores are home to a growing middle class, but inequality is high 50 46.5 44.9 Indonesia’s middle class has grown from 5 percent of the total population in 2001 to 40 Share of population that is 36.8 22 percent in 2017.15 Although that share has middle class (%) grown everywhere, it grew the most in metro 30 27.9 26.7 areas: by 25 percentage points in peripheries 21.8 of multidistrict metro areas, 23 percentage 20 points in metro cores, and 22 percent- 13.6 9.8 age points in single-district metro areas. These 10 areas also had the largest shares of the ­ middle-class population in 2017 (figure 2.17). 0 About 43 percent of Indonesians in metro re ry re ro y n y l ra r r ba cores are middle class, 16 compared with he he he co co et ru ur m rip rip ip ro rta ro er ro ict et pe pe 10 percent in nonmetro rural areas. Even after et ka lp et rm str nm Ja rta an nm ra di he No rb Ru controlling for household characteristics such ka e- No Ot ru Ja gl Sin he as size, education of the head, assets, and Ot employment, households in metro cores and single-district metro areas are about 50 per- Source: Calculations using data from Indonesia’s National Socio-Economic Survey (SUSENAS). Note: Middle class refers to the households whose consumption is more than 3.5 times the national cent more likely to be middle class than those poverty line (Rp 1,240,000 a day). Places are defined using the methodology described in box 1.5, in nonmetro rural areas. chapter 1. Rp = Indonesian rupiah. I s U rba n i z atio n D eli v eri n g ?   87 More urbanized areas, especially metro individuals, who command a higher wage areas, also tend to be more unequal. premium. Whereas in most countries inequal- discussed in chapter 4, this may be because As ­ ity is higher in urban areas, in Indonesia, the metro areas have a higher share of consumption Gini coefficient, a measure of ­ employment in high-value-added activities inequality, is 0.41 points in urban areas, and are more likely than other places to have higher than urban areas in  regional peers labor markets that demand highly educated figure 2.18). The urban Gini coefficient also (­ exceeds the rural Gini coefficient by 8 per- FIGURE 2.18  Urban inequality is high in Indonesia centage points, which is more than in China, compared to regional peers Malaysia, and Thailand.17 The metro cores of Bandung, Yogyakarta, Surakarta, and Malang 0.45 0.41 are among the 10 most unequal districts in 0.39 Indonesia (Bank  Indonesia, Ministry of Gini coefficient on income or consumption 0.40 0.38 0.36 0.36 0.36 0.34 Finance Indonesia, and World Bank 2018). 0.35 0.33 0.33 0.29 0.30 Inequality widened in all types of places 0.25 All types of places experienced increases in 0.20 the Gini coefficient over the period 2001–17, 0.15 with the fastest increase in the Gini coeffi- cient seen in multidistrict metro areas, espe- 0.10 cially their rural peripheries (figure 2.19).18 0.05 Because income disparities between districts have declined, but overall inequality in 0 Indonesia has increased in this period, Indonesia Malaysia India Thailand China inequality in Indonesia has become increas- Urban Rural ingly driven by inequality within, rather than between, places. Sources: India—National Sample Survey; Indonesia—National Socio-Economic Survey (SUSENAS); Malaysia— Today, within-place inequality accounts Household Income and Expenditure Survey; Thailand—National Statistics Office; China—Knight 2013. Note: Data refer to 2016 except for China (2007) and India (2012). Gini coefficient is measured for for much more of the overall inequality in consumption in Indonesia and India; Gini coefficient on income elsewhere. Indonesia than between-place inequality does. This finding is true across island- FIGURE 2.19  Inequality has risen in all places, regions, the portfolio of places, and districts. especially in rural peripheries, 2001–17 For the portfolio of places, within-place inequality accounts for 86 percent of overall 0.45 inequality, and between-place inequality accounts for 14 percent, according to 2017 Gini coefficient on consumption data (figure 2.20). Likewise, for districts, 0.40 within-place inequality accounts for 78 per- cent of overall inequality, and between-place 0.35 inequality accounts for 22 percent. The increase in inequality within places suggests 0.30 that the supply of human capital has not increased at the same pace as the demand, which has led to an increase in the education 0.25 premium. This is not to say, however, that 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 inequality between places does not remain 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Metro core Urban periphery Rural periphery a major issue for Indonesian policy makers. Single-district metro Nonmetro urban Nonmetro rural By international standards, spatial income disparities in Indonesia are very high (Ezcurra Source: Calculations based on data from Indonesia’s National Socio-Economic Survey (SUSENAS). and Rodriguez-Pose 2013). Also, although Note: Three-year moving average of Gini coefficient. some degree of inequality between people 88  TIME TO ACT FIGURE 2.20  Overall inequality is rural areas—and, if so, to what extent this driven by within-place inequality is true for all inhabitants. Jakarta tends to do poorly on global indexes 100 of livability, ranking 142nd of 231 cities on the 14.5 18.3 2018 Mercer Quality of Life Index and 118th 80 of 140 cities on the Economist Intelligence Unit’s Global Livability Index. 19 On the Total inequality (%) 60 Mercer Index, Jakarta fares worse than every major Southeast Asian city except Hanoi 85.5 40 81.7 (155th), Vientiane (170th), Phnom Penh (197th), and Yangon (203rd). However, no 20 other Indonesian city appears in these rankings. 0 To assess the extent to which Indonesia 2001 2017 capitalizes on the opportunities of living in Within place Between places cities while minimizing the negative conges- tion forces that may arise from urbanization, Source: Tiwari and Shidiq 2018. this section relies on some common elements Note: The inequality shares are obtained from the generalized entropy class of inequality measures that can be additively decomposed into inequality of these livability indexes: access to public within and between groups. The measure used is Theil’s L. services (utilities, education, health, and sani- tation), affordability and quality of housing, and levels of traffic congestion, pollution and within places is inevitable, it is less clear that crime. differences in average incomes between places should always exist. Within Indonesia, on average, urban areas provide better access to services than rural areas do Is urbanization in Indonesia delivering livability for all? Indonesia has made steady progress toward universal access to basic services, including The previous section showed that, although health, sanitation, and education facilities. urban areas in Indonesia are associated However, urban–rural differences remain sig- with higher incomes and lower poverty, the nificant. With respect to health, virtually all process of urbanization since 1996 has not urban dwellers in metro cores and single-­ been associated with faster growth or pov- district metro areas have access to primary erty reduction. In part, this outcome was care facilities (puskesmas), delivery facilities, due to the relatively robust growth of rural and hospitals located in or near their desa/ areas. In addition, the poor quality of struc- kelurahan.20 By contrast, more than 80 percent tural transformation and low productivity of households in rural areas lack “very easy” growth in urbanizing areas suggest that or “easy” access to a doctor (figure 2.21, congestion forces may be dampening the ­panel a).21 However, not all urban areas are impact of urbanization on prosperity in alike: over one-third of households in non- Indonesia. Congestion forces arise from the metro urban areas and urban periphery areas pressure of urban populations on basic ser- (excluding Jakarta) find it difficult to see a vices and infrastructure, land and housing doctor. markets, and the environment. These pres- Urban areas are also generally ahead of sures have a direct negative impact on liva- rural areas in access to water and sanitation. bility, while also constraining the positive Fewer than 1 in 10 households in metro cores agglomeration forces that raise productivity and single-district metro areas lack “easy” or and prosperity. This section assesses “very easy” access to safe drinking water and whether Indonesia’s cities and urban areas sanitation, compared with about 4 in 10 house- are “livable”—or at least, more livable than holds in rural areas (figure 2.21, panel b). I s U rba n i z atio n D eli v eri n g ?   89 FIGURE 2.21  Urban households have greater access to health facilities and to safe drinking water and sanitation, 2014 a. Share of households without access to health care b. Share of households without access to safe water and sanitation 90 45 80 40 70 35 60 30 Percent Percent 50 25 40 20 30 15 20 10 10 5 0 0 Hospital Doctor Drinking water Sanitation Jakarta core Jakarta periphery Other metro core Other urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Source: Adapted from Lain 2018, using data from Indonesia’s Survey of Village Potential (PODES). Note: Places are defined using the methodology described in box 1.5, chapter 1. In panel a, “doctor” refers to a doctor’s private practice. FIGURE 2.22  Rural households are particularly peripheries lack access to safe drinking water deprived of access to secondary education, 2014 and sanitation. There is also a substantial dif- ference between Jakarta core, where 7 ­ percent 45 of households do not have access to good sani- tation, and other metro cores, where 19 percent Share of households without access (%) 40 35 of households face this issue. While access to primary school is nearly 30 universal in Indonesia, urban areas gener- 25 ally do better than rural areas in providing 20 o pportunities to attend secondary school ­ 15 (­f igure 2.22). About 13 percent of metro core (excluding Jakarta) inhabitants live in 10 a desa/kelurahan without easy access to a 5 secondary school, whereas in Jakarta’s 0 metro core virtually all have easy access. Preschool Secondary school For urban peripheries (excluding Jakarta) Jakarta core Jakarta periphery and nonmetro urban areas, easy access is Other metro core Other urban periphery somewhat poorer, with 25 percent of Rural periphery Single-district metro households in each type of place living in a Nonmetro urban Nonmetro rural desa / kelurahan without such access. However, these places still do better than Source: Adapted from Lain 2018, using data from Indonesia’s Survey of Village Potential (PODES). both rural periphery and especially non- Note: Places are defined using the methodology described in box 1.5, chapter 1. metro rural areas, where 34.5 percent and 41.6 percent of households, respectively, However, living in an urban area does not lack easy access. guarantee such access: one in five households Over 2002–16, there has been conver- in urban peripheries and nonmetro urban areas gence across districts in some indicators of does not have access to safe drinking water. local infrastructure and basic service deliv- The poor in these areas are even more deprived: ery such  as access to education and the nearly 4 in 10 poor people living in urban share of births delivered by a skilled health 90  TIME TO ACT worker (­ f igure 2.23). On some other ser- from 25.8 p ­ ercent to 7.9 percent in single- vices, there has been substantial progress. district metros (­figure 2.24, panel a). For example, between 2002 and 2016, the Despite these convergence patterns, proportion of households with inadequate ­ significant gaps remain between urban and sanitation dropped from 80.4 percent to rural places in access to many services, 44.3 percent in nonmetro rural areas and including access to early childhood services FIGURE 2.23  Gaps in services remain between urban and rural places a. Share of households with b. Share of households with a child c. Share of households with a child under age 5 no access to a preschool (age 5–16) not enrolled in school not delivered by a skilled health worker 50 30 50 40 25 40 30 20 30 Percent Percent Percent 20 15 20 10 10 10 0 5 0 2003 2014 2002 2016 2002 2014 Jakarta core Jakarta periphery Other metro core Other urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Source: Adapted from Lain 2018, using data from Indonesia’s 2003 and 2014 Survey of Village Potential (PODES) and 2002, 2014, and 2016 National Socio-Economic Survey (SUSENAS). Note: Places are defined using the methodology described in box 1.5, chapter 1. FIGURE 2.24  Disparity in access to safe sanitation and to safe drinking water remains high between urban and rural areas, 2002 and 2016 a. Share of households without access b. Share of households without access to sanitation to safe drinking water 90 70 80 60 70 50 60 40 Percent Percent 50 40 30 30 20 20 10 10 0 0 2002 2016 2002 2016 Jakarta core Jakarta periphery Other metro core Other urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Source: Adapted from Lain 2018, using data from Indonesia’s National Socio-Economic Survey (SUSENAS). Note: Places are defined using the methodology described in box 1.5, chapter 1. I s U rba n i z atio n D eli v eri n g ?   91 such as ­preschools and skilled health work- managed sanitation in urban areas are dis- figure 2.23) and in access to safe water ers (­ cussed in detail in spotlight 3, “The Invisible and sanitation (figure 2.24). These spatial Crisis of Wastewater Management in inequalities persist within urban areas and Indonesia.” This spotlight also diagnoses across Indonesia’s portfolio of places. the underlying causes of poor access and policy options for improving access. Indonesia trails its emerging market economy peers on providing many basic Along with better access, urban areas have services better health, nutrition, and education outcomes Despite the progress in closing service deliv- ery gaps between districts, Indonesia’s abil- With better access to health and education ity to provide basic services to all services, urban areas also have better health inhabitants is lower than expected for its and education outcomes than rural areas. level of development and given its level of People residing in metro cores and single- urbanization. At best, about 3 in 4 people district metro areas are more likely than have access to basic sanitation services in ­ people residing in nonmetro rural areas to Indonesia, compared with more than 8 in have infants or young children who were 10 in Colombia, and nearly everyone delivered with the help of a medical profes- in Malaysia (figure 2.25, panel a). Indonesia sional, to be literate, and to have finished also lags country peers that have a similar primary school (­ figure 2.26). However, some level of urbanization in providing basic deprivations affect urban and rural dwellers drinking water services (­ f igure  2.25, evenly. Regardless of place, 40–50 percent of panel  b). The health and environmental households include a young child lacking full consequences of poor access to safely immunizations. FIGURE 2.25  Indonesia trails several emerging market economies in providing basic sanitation services and basic drinking water services, 2015 a. Share of population with access to sanitation b. Share of population with access to basic drinking water services 100 97 98 98 100 95 100 96 96 96 89 90 91 84 86 80 75 75 76 80 68 60 60 Percent Percent 40 40 20 20 0 0 a zil sia s a ) a il ico d ia ia s a ia d ico AS ne ne in az in bi bi an an ys es ys a ne ex ex Ch Ch m m Br Br EN pi pi ala ala ail ail on lo lo do M M lip ilip US Th Th d M M Co Co In In i Ph Ph (S ia es on d In Sources: Calculations based on data from World Health Organization/United Nations Children’s Fund (WHO/UNICEF; panels a and b) and Indonesia’s National Socio-Economic Survey (SUSENAS; panel a). Note: SUSENAS data are for sanitation facilities that are not shared with other households. WHO/UNICEF definition of basic sanitation further specifies that these facilities must flush to piped sewer systems, septic tanks, or pit latrines. Basic drinking water services refers to piped water, boreholes or tubewells, protected dug wells, protected springs, and packaged or delivered water, with collection time of no more than 30 minutes. 92  TIME TO ACT Children in urban places also have better FIGURE 2.26  Urban households are less deprived nutritional outcomes. Regression analyses than rural households on a wide range of health and that control for differences in household education outcomes, 2016 characteristics and access to services find Share of households without access (%) that children in metro cores are taller (by 0.21 standard deviation) and heavier (by Skilled births 0.15 standard deviation) for their age than 60 children in nonmetro rural areas.22 Children 50 in urban peripheries are likewise taller (by 40 0.33 standard deviation) and heavier (by 30 0.13 standard deviation) for their age than Primary 20 Immunization children in nonmetro rural areas and are school 10 6.8 percentage points less likely to be stunted attainment than children in nonmetro rural areas. 0 Intraurban differences are large for some outcomes. Births without a skilled medical attendant are far more prevalent in urban periphery areas, affecting about 11 percent of poor households, than in metro cores and Literacy Exclusive single-district metro areas. Thus, there are breastfeeding strong core–periphery disparities in health Jakarta core Jakarta periphery Other metro core and education outcomes. Other urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Better access to services in urban areas Source: Adapted from Lain 2018, using data from Indonesia’s National Socio-Economic Survey comes at the price of higher housing costs (SUSENAS). and cramped quarters Note: A household is considered to be deprived of access to exclusive breastfeeding if it contains an infant ages 6–23 months who was not exclusively breastfed for at least six months. Better access to economic opportunities and services in urban areas has driven up demand Crucially, paying more is not necessarily for housing in Indonesian cities, leading to associated with more living space. In fact, over- congestion in land and rental markets crowding (defined as less than 8 square meters (see ­ chapter 7). Although not as high as in cit- of floor area per person23), which is perhaps an ies such as Shenzhen, Hong Kong, and even more telling indicator of congestion pres- Manila, ratios of house price to income are sures in housing markets than the house price- nevertheless higher in Indonesia’s cities than to-income ratio, is more prevalent among urban in rural areas and higher than in cities such as inhabitants, especially those living in metro New York, Singapore, and Tokyo (figure 2.27, cores, than it is among rural inhabitants. This is panel a). These higher ratios occur in a con- true across all economic classes. More than half text where Indonesian households also have of poor and vulnerable households residing in much poorer access to mortgage finance than metro cores are overcrowded. Overcrowding is do, for example, households in the United especially severe in Jakarta’s core. Hence, States. As a result, households at all income whereas the share of households experiencing levels in the metro cores spend more on hous- overcrowding in other metro cores and types of ing. Overall, people residing in metro cores place fell between 2002 and 2016, the share devote 33 percent of consumption to housing, increased from 28 percent to 35 percent in compared with 23 percent for people living in Jakarta’s core (figure 2.27, panel b). nonmetro rural areas. The differences are less Although housing in metro cores is more stark among the poor, but poor people in expensive and more crowded, it does have the metro cores still devote a higher proportion of benefit of being built of better materials. consumption to housing than poor people in Unsuitable wall materials are used for more rural areas. houses in nonmetro rural areas (42.6 percent) I s U rba n i z atio n D eli v eri n g ?   93 FIGURE 2.27  Ratios of house price to income are higher in Jakarta than in New York, contributing to substantial overcrowding in metro cores and single-district metro areas a. Ratio of house price to income b. Share of overcrowded households, 2002 and 2016 Shenzhen 19.8 40 Hong Kong 19.4 35 Taipei 15.1 Beijing 14.5 30 Manila 14.0 25 Percent Shanghai 14.0 Bandung 12.1 20 Mumbai 11.9 15 Denpasar 11.9 10 Seoul 10.8 Jakarta 10.3 5 London 8.5 0 Bangkok 7.7 2002 2016 New York 5.7 Singapore 4.8 Jakarta core Jakarta periphery Tokyo-Yokohama 4.8 Other metro core Other urban periphery Kuala Lumpur 4.0 Rural periphery Single-district metro Osaka-Kobe-Kyoto 3.5 Nonmetro urban Nonmetro rural 0 5 10 15 20 Percent Sources: Panel a calculations based on data from Demographia, Nomura, and Euro-money Institutional Investor Company (CEIC). Panel b adapted from Lain 2018, using data from Indonesia’s National Socio-Economic Survey (SUSENAS). Note: In panel a, data are for latest year available between 2015 and 2017; red bars indicate Indonesian cities. In panel b, overcrowding is defined as less than 8 square meters of floor area per person. Places in panel b are defined using the methodology described in box 1.5, chapter 1. than in metro cores (6.5 percent) and urban Nonmetro urban areas and urban periphery areas (6.9 percent), although the peripheries, in particular, face a large share is also large among people living in housing deficit and large slum populations ­ single-district metro areas (22.0 percent). The relatively high cost of housing, coupled Similarly, less than 1 percent of households in with a lack of access to mortgage finance metro cores and single-district metro areas (chapter 7), has led to a substantial housing and just 2.1 percent of households in urban deficit in Indonesia. The Ministry of Public periphery areas have dirt floors compared Works and Housing estimated a deficit of with 9.6 percent of households in nonmetro 13.5 million units in 2014. Although this rural areas. estimate may be overstated (it is based on the There are large differences in deprivation number of people who do not own a home by economic class within all types of place and does not take into account the prefer- for inadequate roofing, walls, and flooring. ences of people who may want to rent rather One  of the starkest examples: in single-­ than own a house), it is clear that the annual district metro areas, just 4.5 percent of supply of 400,000 new housing units is not ­ m iddle-class households have inadequate enough to address the formation of up to walls, whereas 51.2 percent of poor house- 920,000 new households annually. In addi- holds do. Living in urban areas therefore tion, some 3.4 million units are deemed to be provides only limited assurance that house- substandard on the basis of one or more holds will avoid deprivation in housing indicators, such as substandard housing quality. 94  TIME TO ACT materials or lack of access to water or Long commutes, congestion, and sanitation. pollution reduce the quality of life in New affordable housing supply is also metro cores often poorly located, pushing the urban poor In part because of high housing costs in the to the city peripheries. Lack of finance and metro core, and high rates of private vehicle the high cost of land in metro cores mean use that have clogged roads and contributed that low-income households often resort to to urban sprawl, many urban households purchasing housing in periphery areas (see contend with long commutes to reach better chapter 7). As a result, transportation needs economic opportunities in the core. Thus, and costs rise for these households, which 35 percent of commuters in Jakarta’s core contributes to congestion and patterns of spend more than an hour on the road to get sprawl that are neither economically efficient to work, and 47 percent of commuters in nor sustainable. Jakarta’s periphery do. Similarly, 43 percent High housing costs in the metro cores and of commuters in single-district metros travel poor-quality subsidized housing have facili- more than an hour to work. This compares to tated the emergence of urban slums.24 About a ­figure of 24 percent for nonmetro rural 22 percent of Indonesia’s urban population areas (figure 2.29, panel a). lives in slums. This is lower than the average However, there is significant variation of 26 percent for developing East Asia and across metro areas and between Jakarta and Pacific countries, not to mention the average other metro cores. In other metro cores, f o r l o w e r- m i d d l e - i n c o m e c o u n t r i e s only 17 percent of commuters spend more figure 2.28, panel a), but still means that (­ than an hour on the road. This difference is some 29 million Indonesians live in slums. in part due to shorter travel distances: Most slum dwellers live in urban peripheries whereas about 30 percent of commuters in and nonmetro urban areas (figure 2.28, Jakarta’s core and 43 percent of commuters panel b) and, except for greater difficulty in in its periphery travel more than 30 kilome- accessing some health and education ser- ters to their job, only about 12 percent do vices, are not necessarily worse off than in other metro cores and urban peripheries those people in the same areas who do not (figure 2.29, panel b). live in slums (box 2.2). FIGURE 2.28  One-fifth of Indonesia’s urban population lives in slums, mostly in urban peripheries and nonmetro urban areas, 2015 a. Share of urban population living in slums b. Share of population 60 100 50 22.3 80 43.0 40 Percent 23.8 Percent 60 30 40 33.1 28.0 20 10 20 4.9 19.1 25.9 0 0 Slum Non-slum lo . In bia sia Th ia Vi ina La m a Co Rep Ke s ilip R il d ne ny d Ph o PD az na an ne Ch In m pi Br et b ail Metro core Single-district metro do ra ,A Urban periphery Nonmetro urban t yp Average, EAP (except high income) Eg Average, lower-middle income Sources: Panel a calculations based on data from World Bank World Development Indicators (http://datatopics.worldbank.org/world-development-indicators/). Panel b calculations based on data from Badan Pusat Statistik (Statistics Indonesia). Note: EAP=East Asia and Pacific. Places in panel b are defined using the methodology described in box 1.5, chapter 1. I s U rba n i z atio n D eli v eri n g ?   95 BOX 2.2  Mixed health and education outcomes for slum dwellers compared to non-slum dwellers Indonesians who live in slums appear to be slums do not have worse health and education more deprived of certain health and education outcomes simply because they are monetarily services compared with those who do not live poorer. Even among poor people in multidistrict in slums (figure B2.2.1). About 40 percent of metro areas, living in a slum comes with a two to slum dwellers lack “easy” or “very easy” access three times higher chance of giving birth without to a doctor, double the share of non-slum a skilled attendant. dwellers who face the same difficulty. Similarly, However, for other health and education 6 percent of slum dwellers lack access to pre- outcomes, slum dwellers appear to be no worse schools, compared to about 2 percent of non- off. For example, outcomes for primary school slum dwellers. enrollment and exclusive breastfeeding are sim- Living in a slum comes with a greater chance ilar, if not slightly worse, for urban residents of nonmonetary deprivation in terms of health living outside of slums. In addition, some appar- and education outcomes, regardless of where ent differences between slum dwellers and oth- the household is located. For example, in multi- ers disappear when economic class is controlled district metro areas, women in slum-dwelling for. For example, taking all economic classes households are more than four times as likely together, slum households in m ­ultidistrict to have experienced childbirth without a skilled metro areas are slightly more likely to be illiter- assistant as women living outside slums. Because ate. But when economic classes are considered these differences persist even after socioeco- separately, this difference persists only for poor nomic class is controlled for, households in households. FIGURE B2.2.1  Slum dwellers have poorer access to doctors and preschools, 2014 a. Share of households without access to health care (%) b. Share of households without access to education (%) Primary care Preschool 45 20 35 15 25 15 10 Midwife Hospital 5 5 –5 0 Junior Primary Doctor Delivery high school school Slum Non-slum Sources: Lain 2018, based on data from Indonesia’s National Socio-Economic Survey (SUSENAS) and Survey of Village Potential (PODES). Note: Sample of all households in metro cores, single-district metro areas, urban peripheries, and nonmetro urban areas. In the left panel, “doctor” refers to a doctor’s private practice; “delivery” refers to a delivery facility, which could consist of a maternity puskesmas (community health center) or a polindes (village maternity post); “midwife” refers to a midwife’s practice. Source: Lain 2018. 96  TIME TO ACT FIGURE 2.29  Large shares of commuters in Jakarta metro and single-district metro areas spend more than an hour to get to work, partly because of long distances between home and work, 2015 a. Share of commuters b. Distance traveled by daily commuters who travel more than 60 minutes a day to work 50 47 100 43 18 17 15 13 12 30 40 80 43 38 35 41 51 48 30 60 43 52 Percent Percent 23 24 20 47 20 17 40 47 15 43 10 20 39 41 33 36 40 23 14 15 0 0 e ry e ry ry ro n l ry ro e n l ry ry e ra ra r r r r ba ba he he he he he he co co co co et et ru ru ur ur m m rip rip ip rip ip rip rta ro rta ro ro ro er er ro ro ict ict et et pe pe pe pe et et ka ka lp lp et et rm rm str str nm nm Ja Ja rta an rta an nm nm ra ra di di he he No No rb rb Ru Ru ka ka e- e- No No Ot Ot ru ru Ja Ja gl gl Sin Sin he he Ot Ot <10 km 10–30 km >30 km Source: Calculations based on 2015 Indonesia’s National Labor Force Survey (SAKERNAS). Note: In both panels, sample includes only workers who live in one district and commute to another for work. Long commute times are a result not only of inhabitants and degrading the environ- of distance but also of traffic congestion (see ment. According to World Bank estimates also chapter 7). The lack of adequate public constructed for this report, traffic congestion transport contributes to the use by most costs Jakarta US$2.6 billion a year, or about Indonesians of private vehicles (cars or motor- 1.7 percent of its GDP. This is nearly half the cycles) to get to work. As a consequence, total congestion cost for all metro areas in Jakarta is consistently rated one of the 10 cit- Indonesia, which is estimated at US$4 billion ies in the world with the worst traffic. On the a year, or about 0.5 percent of national GDP. TomTom Traffic Congestion Index, Jakarta These are lower-bound estimates that do not was the third most congested of 18 megacities take into account the health impacts from worldwide, with an estimated 58 percent emissions, costs of road collisions, and extra travel time needed for any trip, any- greenhouse gas emissions (see chapter 7). where in the city, at any time compared with a Partly because of traffic congestion, but free-flow situation. Indonesian cities are also also because of the use of polluting fuels and among the most congested on the Inrix Global industrial coal plants, pollution is a major Traffic Scorecard. Even in smaller cities such detriment to living in Indonesian cities. Only as Padang and Yogyakarta, drivers spend eight metro areas in Indonesia have safe lev- about a quarter of their driving time in con- els of air pollution, as measured by the World gestion (figure 2.30). On average, it takes Health Organization standard of 10 micro- more than twice as long to travel the same dis- grams of fine particulate matter of 2.5 tance with the same type of transport in microns or less in diameter per cubic meter Indonesia as it does in Malaysia (figure 2.31). (figure 2.32). Jakarta is more polluted than Congestion slows economic growth and Tokyo, London, São  Paulo, and Mexico productivity in Indonesia’s urban areas, City; and Pekanbaru is more polluted than while diminishing the health and well-being Mumbai and Shanghai (figure 2.33). I s U rba n i z atio n D eli v eri n g ?   97 FIGURE 2.30  Indonesia’s cities are among the most FIGURE 2.31  It takes far more time to congested in the East Asia and Pacific region travel the same distance in Indonesia than Share of driving time spent in congestion, 2017 in other East Asia and Pacific countries 3.0 Padang 2.7 Yogyakarta 2.5 Hours per 100 km Malang 2.0 2.0 Bangkok 1.5 1.3 1.3 Pontianak 1.1 Bandung 1.0 Bengkulu 0.5 Tarogong 0 Medan ia na m sia d an ys na i ne Jakarta Ch ala ail et do Th Vi M In Sungai Pinang Semarang Source: IndII 2012. Tasikmalaya Surabaya Denpasar Bogor Crime is more prevalent in metro cores and Sakaka single-district metro areas than in other places Lat Krabang Riyadh Globally, urban areas tend to be associated Fujairah with more crime than rural areas, and Chiang Mai Indonesia is no different. Household surveys Kuwait City indicate that crime tends to be more preva- Jeddah lent in the core and in single-district metro Dubai areas than elsewhere. The proportion of Singapore households that reports having experienced crime is higher in the core and single-district 0 5 10 15 20 25 metro areas (­ figure 2.34, panel a), as is the Percent number of violent incidents reported in Indonesian cities Cities not in Indonesia figure 2.34, newspapers at the district level (­ panel b). Crime mainly seems to affect higher Source: Inrix Global Traffic Scorecard 2017 (http://inrix.com/scorecard/). Note: The scorecard covers 1,360 cities, 15 of which are in Indonesia. Congestion refers to road speeds economic classes—especially the middle that are less than 65 percent of free-flow speeds. Red bars indicate Indonesian cities. class—potentially because these households have greater material wealth and therefore High pollution is associated with a host of more goods  and  assets at risk of theft diseases and other health-related effects that (figure 2.34, panel c). can worsen the quality of life and reduce pro- ductivity. In Jakarta, an estimated 60 percent of the population suffers from air p­ ollution– Urbanization’s potential to related diseases. Greenpeace (2017) estimates that emissions from eight coal-fired power deliver more benefits plants in the Jakarta metro area already To a large degree, urbanization in Indonesia result in 5,260 premature deaths and 1,690 has delivered on its promise. Urban areas, underweight births per year. Seventy percent especially multi- and single-district metro of city dwellers surveyed identified “less pol- areas have higher incomes per capita than lution” as the most important urban environ- nonmetro rural areas. They offer more stable, ment issue (World Bank 2018). higher-paying jobs in industry and services, 98  TIME TO ACT which contribute to higher median incomes, a FIGURE 2.32  Only eight metro areas in Indonesia larger middle class, and lower poverty rates. have acceptable levels of air pollution, 2015 Finally, urban areas—especially metro cores and single-district metro areas—facilitate Pekanbaru ­ better access to services such as health, educa- Jambi tion, safe drinking water, and sanitation. Palembang Better access is associated with better Bukittinggi outcomes in those areas—and a better quality ­ Padang of life. Pontianak Importantly, urbanization in Indonesia Medan does not appear to have left rural areas completely behind. Rural peripheries have Jakarta grown as fast as metro cores and single-­ Semarang district metro areas in recent years, and Bandar Lampung gaps in access to basic services between Bandung urban and rural areas have declined. Salatiga However, several pieces of evidence sug- Surabaya gest that the process of urbanization of the Surakarta past 20 years has fallen short of its potential. Magelang Districts that urbanized faster over the Mojokerto 1996–2016 period have not enjoyed faster Sukabumi economic growth. In particular, nonmetro Banjarmasin urban areas and urban peripheries have seen Yogyakarta slower than average growth in per capita Pasuruan incomes over 1996–2016, and Indonesia’s Blitar neighbors have experienced faster growth Malang for a given increase in the urban share of the population. Probolinggo Urbanization has been associated with Balikpapan shifts in employment away from agricul- Denpasar ture, but not with higher shares of services Samarinda and industry in total output, suggesting that Banda Aceh jobs have been created in relatively low-­ Makassar productivity services and industry sectors. 0 10 20 30 40 50 60 70 The concentration of scarce human capital Annual average PM2.5 concentration (micrograms per cubic meter of air) in metro cores limits the potential of struc- Cities that meet the WHO standard tural transformation and economic growth Cities that do not meet the WHO standard of the peripheries and nonmetro urban areas, while leading to higher inequality in Source: Calculations based on satellite-derived data from Dalhousie University. urban cores. In  fact, although inequality Note: PM2.5 is particulate matter of 2.5 microns or less in diameter. Dark blue bars indicate Indonesian cities that satisfy the World Health Organization’s (WHO) standard of 10 micrograms per cubic meter between places, while remaining very high for safe air. by international standards, has declined, inequality has increased in all types of face long and lengthy commutes to work, places, driving an overall increase in high pollution levels, and higher levels of inequality in Indonesia during the past crime. 20 years. These congestion forces reduce the poten- Growing congestion may also limit the tial of agglomeration forces in urban areas to potential of urban areas to provide greater generate greater productivity and prosperity. prosperity for all Indonesians. Households The subsequent chapters elaborate on the in metro cores and single-district metro drivers of these agglomeration and congestion areas tend to suffer overcrowding and to forces. I s U rba n i z atio n D eli v eri n g ?   99 FIGURE 2.33  Indonesia’s cities are FIGURE 2.34  Metro cores and single-district metro among the most polluted in the world, areas, especially their middle-class households, are more 2015 likely to experience crime than other places in Indonesia Dhaka a. Households that experienced any type of crime in the previous year, circa 2016 Beijing 10 Pekanbaru 8 6 Percent Shanghai 4 Johannesburg 2 Mumbai 0 Seoul IFLS SUSENAS Bangkok Metro core Urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Medan b. Number of crimes reported in newspapers, 2014 Los Angeles 400 Jakarta 300 London 200 Mexico City 100 Kampala São Paulo 0 Metro All Single- All National Surabaya core peripheral district nonmetro areas metro areas Moscow c. Households that experienced any type of crime Bandung the previous year, by economic class, 2016 Ho Chi Minh City 6 5 Istanbul 4 Percent Tokyo 3 Paris 2 1 Bogotá 0 Buenos Aires Metro Urban Rural Single- Nonmetro Nonmetro core periphery periphery district urban rural 0 30 60 90 120 150 metro Annual average PM2.5 concentration (micrograms per cubic meter of air) Poor Vulnerable Aspiring middle class Middle class Cities not in Indonesia Indonesian cities Source: Calculations based on satellite-derived data from Dalhousie Sources: Panel a calculations based on data from Indonesia’s 2017 Indonesia Family Life Survey (IFLS) University. and 2016 National Socio-Economic Survey (SUSENAS). Panel b calculations based on data from Badan Note: PM2.5 is particulate matter of 2.5 microns or less in diameter. Red bars Pusat Statistik (Statistics Indonesia) and National Violence Monitoring System Indonesia. Panel c indicate Indonesian cities. calculations based on data from Indonesia’s 2016 National Socio-Economic Survey (SUSENAS). Note: Regarding data in panel a, IFLS uses all available observations, without population weights; SUSENAS uses a sample of all households, weighted by sample weights and household size. Panels b and c show a sample of all households, weighted using sample weights (number of households in the district) and household size. 100  TIME TO ACT Notes economy—specifically, the share of employ- ment that is nonagricultural (see chapter 1). 1. Chapter 3 estimates the strength of positive 12. Much of the analysis in this section and the agglomeration forces for Indonesia. discussion on access to services later on in the 2. In this chapter, inequality refers to consump- chapter draws on Lain (2018). tion inequality calculated using data from 13. Global poverty lines are expressed in 2011 Indonesia’s National Socio-Economic Survey constant international dollars. (SUSENAS), unless otherwise noted. 14. Huppi–Ravallion decompositions suggest 3. Figure 2.1 uses a globally consistent definition that almost 90 percent of the poverty decline of urbanization that is calculated on the basis was associated with within-place declines of an application of the Uchida and Nelson as opposed to compositional changes in the (2010) agglomeration index to LandScan grid- underlying population. In addition, declines ded population data from 2012. However, the in poverty in rural areas (using official BPS same relationships hold using data on urban definitions) accounted for 85 percent of the population shares from the United Nations total within-place declines. World Urbanization Prospects: 2018 Revision 15. The middle class are those households that have database. less than a 10 percent chance of being poor or 4. Per capita income is obtained by dividing vulnerable. In other words, their household real GDP (from Badan Pusat Statistik [BPS], consumption is more than Rp 1, 240,000, or or Statistics Indonesia) by the total popula- 3.5 times the national poverty line. See World tion (from Intercensal Population Surveys, Bank (forthcoming) for the methodology used the Census, and linear interpolations for to define the middle class and for the character- other years). This calculation applies to all istics of Indonesia’s emerging middle class references of district GDP per capita in this 16. This is the population weighted average of the chapter unless otherwise noted. For all inter- shares of the population that are middle class temporal analysis of Indonesian districts in in Jakarta’s core and the cores of other multi- this chapter, districts are defined on the basis district metro areas shown in figure 2.17. of their 1996 administrative boundaries. 17. The Indonesian household survey, SUSENAS, 5. Measured as average or median monthly per measures consumption, and therefore the capita expenditure from SUSENAS house- Gini coefficient is calculated for consumption. hold survey data. Income Gini coefficients are typically higher 6. We identify districts that are reliant on mining and quarrying activities as those for which the than consumption Gini coefficients because share of mining and quarrying in local GDP rich households save more of their income exceeds 8 percent, the national average. than poorer ones, meaning their incomes are 7. Using the nominal exchange rate of US$1 = more unequal than their consumption. Rp 14,100. This exchange rate is used 18. Mookherjee–Shorrocks decomposition of throughout the chapter. overall inequality changes during this period 8. Districts are classified as urban areas when shows that most (more than 80 percent) of 50 percent or more of the population lives in the inequality increase can be attributed to urban settlements. They include all districts inequality increases within urban and rural except those classified as nonmetro rural or areas within the country as opposed to the rural periphery areas (see chapter 1). compositional changes in the underlying 9. The fact that high-end services account for population over time. This result, together a greater share of output than they do of with the fact that the urbanization process employment is consistent with the relatively itself is driven to a larger extent by admin- high productivity of such services. istrative reclassification and conversion of 10. The choice of sample period is due to the erstwhile rural areas into urban areas as fact that the National Labor Force Survey shown in chapter 1, suggests that the changes (SAKERNAS) is representative at the district in inequality have a stronger association with level only for August rounds during the years the dynamics within urban and rural areas as 2007–15 and again in August 2017. opposed to the process of urbanization itself. 11. This relationship between increases in urban This theme is further picked up in chapter 4. shares and increases in the share of work- 19. These indexes can be accessed at https:// ers employed in industry and services could www.mercer.com/newsroom/2018-quality​ be partly mechanical, because part of the -of-living-survey.html and https://pages.eiu​ information that BPS uses to define dis- .com/rs/753-RIQ-438/images/The_Global​ tricts as urban is the structure of the local _Liveability_Index_2018.pdf. I s U rba n i z atio n D eli v eri n g ?   101 20. Desa and kelurahan are fourth-level admin- Indonesia: Beyond Survey Data.” Draft version istrative units that lie below districts. A desa February 2018; full publication forthcoming. is a “village” and is governed in accordance Cameron, L., C. Chase, G. Joseph, and R. Pinto. with local traditions. A kelurahan is an “urban 2017. “Child Stunting and Cognitive Impacts community.” A kelurahan has less autonomy of Water, Sanitation and Hygiene in Indonesia.” from higher levels of government than a desa. Policy Research Working Paper, World Bank, 21. For health facilities, accessibility is rated by Washington, DC. respondents to Indonesia’s Survey of Village Duranton, G., and D. Puga. 2004. “Micro- Potential (PODES) as very easy, easy, dif- Foundations of Urban Agglomeration ficult, or very difficult. Households lacking Economies.” In Handbook of Regional and “very easy” or “easy” access to a facility are Urban Economics, Volume 4: Cities and classified as deprived. For education facilities, Geography , edited by J. V. Henderson and households must live either in the same desa/ J.-F. Thisse, 2063–117. Amsterdam: Elsevier. kelurahan or within 1 kilometer of a facility; Ezcurra, R., and Rodriguez-Pose, A. 2013. “Does otherwise, they are classified as deprived. Economic Globalization Affect Regional 22. Regression analysis updating Cameron et  al. Inequality? A Cross-Country Analysis.” World (2017) using pooled Indonesian Family Living Development 52 (December): 92–103. Survey (IFLS) waves 1–4 for children in utero Greenpeace. 2017. “Jakarta’s Silent Killer.” or under the age of 2. Dependent variables are Available at http://www.greenpeace.org/seasia​ height-for-age and weight-for-age z-scores, /Press-Centre/publications/Jakartas-Silent-Killer/. moderate or severe stunting (< –2 height-for- Henderson, J. V., M. Roberts, and A. Storeygard. age z-scores), and moderate or severe under- 2013. “Is Urbanization in Sub-Saharan Africa weight (< –2 weight-for-age z-scores). Controls Different?” Policy Research Working Paper include access to improved sanitation, access 6481, World Bank, Washington, DC. to improved water source, ­ average community Knight, J. 2013. “Inequality in China: An open-defecation-free rate, gender, household Overview.” Policy Research Working Paper size, floor condition, age at time of survey, 6482, World Bank, Washington, DC. mother’s age at birth, parents’ education, log Lain, J. 2018. “Multi-Dimensional Urban Poverty per capita expenditure, urban ­ status, exclusive in Indonesia.” Background paper for this report, breastfeeding, prenatal checks, migration sta- World Bank, Washington, DC. tus of parents, slum status, year of birth fixed IndII (Indonesia Infrastructure Initiative). 2012. effects, and subdistrict fixed effects. Modernizing the national road network: a plan- 23. This definition follows Health Ministerial ning framework to improve connectivity and Decree (Kepmenkes) No. 829/1999. development. Jakarta. 24. This chapter adapts the United Nations Tiwari, S., and A. R. Shidiq. 2018. “Sources of Human Settlements Program’s approach to Welfare Disparities in Indonesia.” Background defining a slum household as any household paper for this report, World Bank, Washington, that meets at least one of the following four DC. criteria: house size less than 7.2 square meters Triyana, M. 2014. Slum Externalities Study per person; poor housing materials, defined as Report: Indonesia Slum Alleviation Policy and having at least two of earth floor, palm/fiber Action Plan, Jakarta: World Bank. roof, or bamboo walls; unsuitable drinking Uchida, H., and A. Nelson. 2010. “Agglomeration water; or unsuitable sanitation. Households Index: Towards a New Measure of Urban are first classified as slum and nonslum, then Concentration.” WIDER Working Paper aggregated up to the desa/kelurahan level Series 029, World Institute for Development (Triyana 2014; UN–Habitat 2006). A  slum Economic Research. desa/kelurahan is then defined as having UN–Habitat (United Nations Human Settlements 75 percent or more slum households using Program). 2006. United Nations Human pooled data from Indonesia’s 2014 National Settlements Programme 2006 Annual Report. Socioeconomic Survey (SUSENAS). Nairobi. World Bank. 2018. “Urban Perception Survey.” Background document for this report, World References Bank, Washington, DC. ———. Forthcoming. “Indonesia’s Aspiring Middle: Bank Indonesia, Ministry of Finance Indonesia, and Expanding the Middle Class.” World  Bank, World Bank. 2018. “Estimating Top Incomes in Washington, DC.  102  TIME TO ACT Drivers of Productivity and Prosperity across the 3 Portfolio of Places KEY MESSAGES Productivity, the ultimate driver of national pros- •  More than 75 percent of the variation in perity, differs widely across districts in Indonesia’s underlying productivity can be explained by portfolio of p ­ laces. These differences are mostly the type of place that a district either is or driven by differences in the characteristics of the belongs to (multidistrict metro core, single- places themselves, with the composition of local district metro area, urban periphery, rural economies and their workforces playing an periphery, nonmetro urban area, or non- important secondary ­ role. The characteristics of metro rural ­ a rea). On average, multidistrict places that matter for productivity depend on the metro core districts and single-district metro type of place, implying the need for tailored poli- areas have the highest levels of underlying cies to enhance productivity and, therefore, productivity, followed by urban periphery ­prosperity. ­ d istricts. Rural periphery districts have underlying productivity levels on a par with •  Differences in average levels of worker produc- nonmetro rural ­ districts. tivity (as measured by average nominal wages) •  The determinants of underlying productivity across districts are explained, to a large extent, vary according to the type of place a district by differences in the underlying productivity either is or belongs ­to. For multidistrict metro associated with the characteristics of the places core, single-district metro areas, and urban themselves (“place-based ­ a dvantages”). periphery districts, traditional agglomeration Meanwhile, sorting effects associated with dif- economies associated with the overall size ferences in the composition of local economies and density of the metro area are a strong and their workforces play a secondary, and significant determinant of underlying although still important, role in accounting for ­ p roductivity. Market access matters for average productivity ­ differences. urban and, to a lesser extent, rural districts 103 in the peripheries of the multidistrict metro investing in human capital appear to be high- areas, but not for other types of districts, est in urbanized metropolitan ­ districts. whereas the presence of a major port has a •• Differences in the business environment also large positive impact on nonmetro urban shape productivity differences across cities ­ districts. Positive agglomeration forces asso- and ­ d istric ts. Modern infrastruc t u re, ciated with human capital are an important availability of finance, and protection from ­ driver of underlying productivity for most crime are among the variables associated types of district, but the social returns to fi rms. with higher productivity of ­ Chapters 1 and 2 described how widely harbor or the existence of natural resources) prosperity varies across Indonesia’s portfolio ­ or advantages that a place or district acquires of ­places. If labor is geographically immo- over time, including advantages that may bile, such differences in prosperity will be come from being part of a larger metropoli- driven by differences in productivity across tan ­ a rea. These acquired advantages can ­places.1 This chapter analyzes the extent to include both the benefits of a good business which variations in prosperity across dis- environment and positive agglomeration tricts and places are reflected in variations in forces, which are beneficial externalities ­ productivity. It then uses detailed microlevel and spillovers that enhance the productivity data on workers and firms to analyze the of workers and firms beyond what determinants of average productivity across would be expected from their individual Indonesian places and districts and to disen- ­characteristics.2 tangle their ­roles. Knowing which of these reasons explains Urban economic theory identifies two most of the differences in productivity and, basic reasons why one place or district is thus in prosperity, across places and districts more productive than another (Combes, is important for ­policy. If differences in pro- Duranton, and Gobillon 2008; Quintero ductivity are explained by differences in the and Roberts 2018; Roberts ­ 2018). First, the composition of the workforce and the local place or district’s workers and firms may economy, this explanation implies that there have characteristics that make them more is nothing special in itself about the agglom- ­ p roductive. For example, it may have an eration of people and firms that occurs in unusually talented workforce whose mem- metro and urban ­ areas. The same is true if bers would be equally productive no matter differences in productivity are explained by where they ­ worked. Such differences in the natural advantages, because policy makers composition of a location’s economy and have little control over ­ these. However, if workforce are associated with the process of productivity differences are explained by “sorting,” as, for example, people with positive agglomeration forces and the busi- higher skills and ability migrate from non- ness environment in a place, that leaves more metro to metro districts or from rural to room for policy to affect ­ productivity. Thus, urban districts at a greater rate than people policy can improve a place or district’s pro- with lower ­ skills. Second, the more produc- ductivity and, therefore, its prosperity, to the tive place or district may have attributes extent that it can stimulate positive agglom- associated with its location that give it a pro- eration forces and enhance the business ductivity advantage over other places or environment. Examples include investing in ­ ­d istricts. These “place-based advantages” human capital through, for example, aug- may be natural advantages (such as a favor- menting the provision of high-quality educa- able coastal location with a natural deep tion and health services and addressing the 104  TIME TO ACT underlying causes of stunting, to take advan- tage of human capital externalities; investing The cross-district in transport infrastructure that ­ better con- relationship between nects, and, therefore, integrates, places; prosperity and average improving relevant dimensions of the busi- ness environment; and addressing the productivity ­ congestion constraints that arise from the Chapters 1 and 2 documented large differ- pressure of urban population on land, hous- ences in prosperity across Indonesia’s portfo- ing, local infrastructure, the supply of basic lio of places, as measured by gross domestic services, and the environment, which may product (GDP) per capita (see table ­ 1.2 in otherwise constrain ­ density. chapter 1 and figure ­ 2.4 in chapter ­2). As The analysis in this chapter uses data on ­ 3.1 shows, differences in GDP per cap- figure ­ more than 1 million workers drawn from ita across districts that belong to different Indonesia’s National Labor Force Survey types of place are, in turn, strongly and posi- (SAKERNAS) and business data from the tively correlated with differences in average World Bank’s Enterprise Surveys for worker productivity, where we measure a ­Indonesia.3 The analysis of the labor force district’s productivity by the average nomi- data reveals that differences in average nal wage that is paid to workers, who are ­ p roductivity across districts are mostly not in school, age 15 and a ­ bove. The use of accounted for by differences in the charac- nominal rather than real wages as a measure teristics of the places themselves (by place- of productivity follows an extensive urban based differences) with the composition of economics literature on the empirical deter- local economies and their workforces (sort- minants of variations in productivity across ing) playing an important secondary r ­ ole. areas.4 cities and other spatial ­ Moreover, and importantly for the tailoring The fact that variations in prosperity are of policy, the exact characteristics that mat- reflected so clearly in variations in worker ter vary according to the type of place a productivity is consistent with a relative ­ district either is or belongs ­ to. Whereas tradi- lack of geographic labor mobility, and, more tional economies associated with population generally, factor market integration, within size and density are important for districts Indonesia (see chapter ­ 6). With greater labor that are part of multidistrict metro cores, mobility, one would expect spatial varia- their urban peripheries, and single-district tions in prosperity to be relatively smooth metro areas, they are not important for other compared with spatial variations in types of d ­ istrict. Positive agglomeration ­productivity. forces associated with access to domestic markets through transportation networks matter only for urban and rural districts in What explains urban the peripheries of multidistrict metro a ­ reas. Finally, positive agglomeration forces associ- productivity? ated with human capital matter for most Two theories from urban economics can help types of district but are strongest in urban- explain the large variations in productivity, ized metropolitan d ­ istricts. Meanwhile, the and therefore prosperity, across Indonesian analysis of the enterprise survey data shows places and ­ districts. One is related to compo- that metro and urban areas with more mod- sitional differences in the structure of the local ern infrastructure, greater availability of economy associated with the sorting of work- finance, less crime, and a greater concentra- ers and firms, and the other is related to dif- tion of “high capacity” firms benefit from ferences in the underlying productivity of the more productive f ­ irms. Other elements of the districts associated with place-based advan- business environment, by contrast, appear tages (figure ­3.2; see also Combes, Duranton, insignificant for ­ productivity. and Gobillon 2008; Quintero and Roberts D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   105 FIGURE ­3.1  GDP per capita and average productivity are strongly correlated across districts 9.0 (log of 2008–15 Indonesian rupiah) Mean nominal hourly wage 8.5 8.0 y = 0.23x + 5.85 R2 = 0.30 7.5 9 10 11 12 13 14 GDP per capita (log of 2016 Indonesian rupiah, thousands) Metro core Urban periphery Rural periphery Single-district metro Nonmetro urban Nonmetro rural Sources: Calculations based on GDP per capita data from Badan Pusat Statistik (Statistics Indonesia)—from Euro-money Institutional Investor Company (CEIC) and the Indonesia Database for Policy and Economic Research ­(https://datacatalog.worldbank.org/dataset/indonesia-database-policy-and-economic- research)—and on wage data from the August 2008–15 rounds of Indonesia’s National Labor Force Survey ­(SAKERNAS). Note: Average nominal wages are based on nominal wages for workers ages 15–80 in their primary occupation and are calculated using pooled data for 2008–15, where the average wage has been detrended using survey-year fixed ­effects. The detrending of nominal wages in this figure removes the general effects of (national) ­inflation. The sample excludes zero wage ­workers. Data are for districts as defined by their administrative boundaries in ­1996. The definition of places that districts belong to is based on the methodology described in box ­1.5, chapter ­1. GDP = gross domestic product. FIGURE ­3.2  Differences in average productivity across districts can be explained by differences in economic composition (“sorting”) or differences in underlying productivity (“place-based advantages”) Di erences in average productivity People and rms Places Di erences in Underlying variation economic composition associated with places Sorting on Sorting on Natural Acquired observables unobservables di erences di erences Agglomeration Business e ects environment 106  TIME TO ACT 2018; Roberts ­ 2018). Compositional differ- characteristics of a nationally average worker ences can occur either because of sorting on employed in a nationally average job characteristics that can be observed in the (Combes, Duranton, and Gobillon ­ 2008). data (“sorting on observables”), such as a Unsurprisingly, in the table, both the mean worker’s level of schooling and workforce and standard deviation for underlying produc- experience, or because of sorting on charac- tivity are less than the corresponding values for teristics that cannot be observed in the data average ­productivity. These lower values are to (“sorting on unobservables”), such as a work- be expected because underlying productivity is “grit.” Differences in the er’s level of ability or ­ the component of average productivity that is underlying productivity of districts can be left over after controlling for s­ orting. However, either natural (due to favorable physical geog- the more important result from our perspective raphy, climate factors, or natural resources) is that, once we employ a relative measure of or ­acquired. Acquired differences may be sep- dispersion, underlying productivity shows arated into agglomeration forces, which may more variation across districts than the average derive, in part, from belonging to a larger productivity of which it is a ­ component. This metropolitan area, and differences associated result is true when we take the coefficient of with the business ­ environment. variation, which normalizes the standard devi- This chapter disentangles variation in the ation of a variable by its mean, as the relative underlying productivity of districts associated measure of dispersion and also when we use with place-based advantages from the effects other relative measures of dispersion such as of sorting and then focuses on how different (Max–Min)/Min, where Max is the maximum types of acquired differences, which are poten- observed value of a variable (average produc- tially open to policy influence, explain the tivity or underlying productivity) and Min is observed productivity variation across the minimum observed ­ value. ­districts.5 The analysis controls for natural dif- Put differently, variations in underlying ferences (natural resource advantages, favor- productivity account for much of the varia- able physical geography, and climate) across tion in average productivity across Indonesian districts, which policy cannot easily ­ influence. districts. One measure of exactly how much ­ underlying productivity “accounts” for is the ratio of the standard deviation of underlying The importance of productivity to the standard deviation of aver- differences in underlying age productivity, which is equal to 0 ­ .56 (­634.74 / ­1,136.05). Hence, on this measure, productivity differences in underlying productivity account To assess the relative importance of variations for 56 percent of the differences in average in underlying productivity associated with productivity across ­ districts. Interestingly, this place-based differences versus sorting in result is the opposite of what Combes, explaining differences in average worker pro- Duranton, and Gobillon (2008) find in their ductivity across districts, table 3­ .1 follows comparable exercise for ­ France. They find Combes, Duranton, and Gobillon (2008) by that sorting accounts for most of the variation systematically comparing disparities across in average productivity across French employ- districts in average productivity (as measured ment areas over the period 1976–96, although by average nominal wages) and underlying the role of differences in underlying produc- productivity. In doing so, we use the data ­ tivity remains ­ important. One potential expla- sources and follow the methodology outlined nation for the difference in results could be in box ­3.1 to derive our estimates of underly- that workers are less geographically mobile in productivity. We also note that a district’s ing ­ Indonesia than in France, which leaves less estimated underlying productivity level may scope for ­ sorting. This explanation is consis- equivalently be thought of as a measure of its tent with a failure of policy to adequately con- “net wage”—that is, the wage that is paid in a nect, and therefore integrate, Indonesia’s given district to a worker who has the portfolio of places (chapter ­ 6). D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   107 TABLE ­3.1  Disparities in average and underlying productivity across districts, 2008–15 Average productivity Underlying productivity (Mean hourly wage) (Net hourly wage) (Max-Min)/Min ­2.62 ­3.33 (P90-P10)/P10 ­1.01 ­1.29 (P75-P25)/P25 ­0.55 ­0.60 Coefficient of variation ­0.27 ­0.31 Mean ­4,200.52 ­2,044.81 Standard deviation ­1,136.05 ­634.74 Median ­4,046.63 ­1,983.90 Minimum ­2,190.00 ­1,000.14 Maximum ­7,919.84 ­4,332.23 Source: Underlying productivity was estimated using data from the August 2008–August 2015 rounds of Indonesia’s National Labor Force Survey ­(SAKERNAS). Average productivity was calculated using the same ­source. Note: Average productivity refers to the detrended mean hourly nominal wage for the period 2008–15 for workers ages 15–80 and is based on the wage received in the main occupation; workers who received a zero wage were ­excluded. Levels of underlying productivity, which may also be interpreted as the “net” mean hourly nominal wage paid within a district to a nationally “average” worker in a nationally “average” job, are estimated using the methodology described in box ­3.1. Max, Min, P10, P90, P25, and P75 are the maximum, the minimum, the first decile, the last decile, the first quartile, and the third quartile, ­respectively. Both average and underlying productivity values are reported in Indonesian ­rupiah. BOX ­3.1  Data and methodology for estimating underlying productivity Data for the analysis were pooled from the of district dummy variables that are set equal to 2008–15 August rounds of Indonesia’s National zero for any given worker except for the vari- Labor Force Survey ( ­SAKERNAS).a The sample able associated with the district in which the of nearly ­1 .2 million workers consists of all worker is employed, which is set to o ­ ne.d The workers reported as receiving a nonzero wage estimated coefficients on the district dummy in cash,b excluding all workers who were still in variables are interpreted as estimates of the school or younger than ­ 15.c The sample includes underlying productivity of districts after work- male and female workers in both the private and force and job characteristics are controlled ­ for. public sectors and in both the formal and the (This method follows that of Ahrend, Gamper, informal ­sectors. Unlike most empirical studies and Schumannet 2014; Combes, Duranton, and of agglomeration forces, the analysis is based Gobillon 2008; De la Roca and Puga 2017; on the districts in which workers are employed Quintero and Roberts 2018; and Roberts 2018, as opposed to the districts in which they live, among ­others.) which is important given the significant com- The worker characteristics controlled for muting flows between districts in multidistrict are gender, marital status, age and its square,e metro ­areas. and educational ­ attainment. Age is a proxy for To isolate differences in underlying produc- ­ experience. Educational attainment is captured tivity across districts from compositional differ- by a series of dummy variables indicating the ences in the workforce associated with sorting, highest level of schooling or qualification that a a worker’s nominal wage in his or her main worker has acquired, ranging from incomplete occupation is regressed on a series of variables primary school to S2/S3, which is equivalent to that capture characteristics of the worker and possessing a graduate ­ degree.f The job charac- the job on which sorting may occur and a set teristics controlled for are the industry in which Box continued on next page 108  TIME TO ACT BOX 3.1 Continued the worker is employed, the average number of As a robustness check, levels of underlying hours that the worker worked in the primary productivity were also estimated for a much occupation in the week before the survey, and narrower sample restricted to prime-age men the length of the worker’s tenure with the cur- (ages 20–54) who work in the private ­ sector. The rent employer and its s ­ quare.g Length of tenure results were similar to those for the full sample controls for the experience that a worker has and so are not reported ­here. ­ acquired with the current employer, allowing a. Each of these rounds is representative at the district ­level. this to have a differential effect on a worker’s b. Within the survey data, some workers are reported as receiving payment nominal wage and productivity than that asso- in goods as well as in ­cash. The results reported in this chapter are robust to ciated with more general workforce ­ experience. whether the reported monetary value of goods received is included or not in the measurement of a worker’s nominal ­wage. One problem with the regression is that it con- c. Although the legal age for receiving a pension in Indonesia is 55, many trols only for a broadly defined set of 17 ­industries. workers who are older than this reported in SAKERNAS as receiving a nonzero The regression is therefore imperfectly able to ­wage. Indeed, the oldest individual reporting a nonzero wage in our data is ­98. To make sure that the results are not driven by exceptionally old workers, the distinguish among, for example, different types oldest 1 percent of workers was trimmed out of the sample, which is equivalent of extractive industries ­ associated with natural to excluding workers older than 80 ­years. Reported results are robust to resource endowments, which may be important further restricting the sample to workers ages 15–64 following the working- age population defined by the Organisation for Economic Co-operation and in driving productivity ­ differences.h To mitigate Development (see ­https://data.world/oecd/working-age-population). this problem, the effects of a worker’s industry d. This regression includes year fixed effects, thereby controlling for the general effects of employment on the nominal wage (and there- of (national) ­inflation. These fixed effects imply that it is the cross-sectional variation in nominal wages that identifies the estimates of underlying ­productivity. fore measured level of productivity) are allowed e. The square of age is included to allow for the possibility of a nonlinear impact to differ by island-region to capture the fact that of age on earnings and ­productivity. More generally, the estimated regression is a different island-regions have different natural Mincerian wage regression (Mincer 1974) augmented with job characteristics and location effects (the district dummy ­variables). resource endowment p ­ rofiles. A second problem f. More generally, the dummy variables capture 11 levels of educational is that, although the regression controls for sort- attainment: no education; incomplete primary school; primary school / Package ing on observable characteristics of workers and A; general junior high school / Package B; vocational junior high school; general senior high school / Package C; vocational senior high school; Diploma I/II; jobs, it does not explicitly control for sorting on Diploma III; Div/S1; and ­S2/S3. such unobservable characteristics as a worker’s g. The square of the length of job tenure is included to allow for the possibility of a ability or ­ ­ “grit.” The resultant potential for bias nonlinear impact of tenure length on earnings and ­productivity. h. The 17 industries controlled for are agriculture, hunting, and forestry; fishing; in estimates of underlying district productivity mining and quarrying; manufacturing; electricity, gas, and water supply; construction; is mitigated, however, to the extent that these wholesale and retail trade; hotels and restaurants; transport, storage, and unobservable characteristics are correlated with communication; financial intermediation; real estate, renting, and business activities; public administration, defense, and compulsory social security; education; health the worker’s observable characteristics, which are and social work; other community, social, and personal service activities; private controlled ­for. households with employed persons; and extra-territorial organizations and ­bodies. One major caveat to this conclusion, how- ever, is that, unlike Combes, Duranton, and Underlying productivity Gobillon, who had the advantage of being highest in multidistrict able to draw on a detailed panel of data on metro core and single- French workers that tracked their full employ- ment histories, our analysis is unable to con- district metro area trol for the sorting of workers on the basis of districts, followed by urban their unobservable characteristics (such as periphery districts ability and ­“grit”).6 This means that our esti- ­ nderlying Consistent with the strong role for u mate that differences in underlying productiv- productivity differences (and, therefore, place- ity account for 56 percent of the differences in based advantages) in explaining average average productivity across districts should be worker productivity differences, the estimated regarded as an upper-bound ­ estimate.7 D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   109 level of underlying productivity is, on average, highest for districts that are either part of a Understanding multidistrict metro core or that form their agglomeration forces and own single-district metro areas (figure ­ 3.3). how tailored policies can Predominantly urban districts in the peripher- ies of the multidistrict metro areas closely improve productivity ­ f ollow in terms of estimated underlying Different types of agglomeration force ­ p roductivity. These ­ d istricts are, in turn, imply different variables matter for lagged by nonmetro urban, and, especially, productivity predominantly rural districts, be these rural Various types of positive agglomeration districts in the peripheries of the multidistrict forces can play a part in driving differences metro areas (rural periphery districts) or non- in underlying productivity across districts metro ­ districts. Importantly, once we control of different t­ ypes. As a reminder, positive for the island-region that a district belongs to, agglomeration forces are the beneficial 77 percent of the variation in underlying pro- externalities and spillovers that arise from ductivity across districts can be explained by the concentration of people and economic the type of place a district either is or belongs activity within places and d ­ istricts. These to: whether a district is part of, for example, externalities and spillovers can enhance the the core of a multidistrict metro area or is a productivity of workers and firms beyond nonmetro rural area makes a big difference to what can be expected on the basis of their its underlying productivity and, therefore, by individual characteristics ­ a lone. The exis- extension, to its average level of productivity tence of such forces implies an important and its overall average level of ­prosperity. potential role for policy in influencing underlying productivity and therefore FIGURE ­3.3  Average estimated levels of underlying prosperity. Urban economists have formu- ­ district productivity, relative to nonmetro rural lated three closely related theories of posi- areas, by type of district tive agglomeration forces: agglomeration economies, human capital externalities, and 30 market access (box 3 ­ .2; see also box I.3 in 25.4 the introduction to this ­ report). 25 Each of these theories can be associated 20.7 with one or more variables whose effects on a 20 district’s underlying productivity can be Percent di erence examined using regression techniques: 15 10.7 •  Agglomeration economies : The popula- 10 tion size of the multidistrict metro area to which a district belongs or, if the district 5 is not part of a multidistrict metro area, 2.4 the population size of the district i ­tself.8 0 In either case, population size is measured Metro core/ Urban Nonmetro Rural as the average over 2008–15 for consis- Single-district periphery urban periphery tency with the wage data used to estimate metro levels of underlying ­productivity. Source: Underlying productivity was estimated using data from the August 2008–August 2015 rounds of •  Human capital externalities : Average Indonesia’s National Labor Force Survey (SAKERNAS) following the methodology described in box ­3.1. years of schooling for the population ages Note: The values reported in the figure are the percentage differences in mean levels of estimated 15 and older for the multidistrict metro underlying productivity for districts that belong to a given type of place from the mean level of estimated underlying productivity for nonmetro rural districts. These mean levels are estimated by area to which a district belongs or, if the regressing a district’s underlying natural log level of productivity on a series of dummy variables for district is not part of a multidistrict metro the different types of place while also controlling for the island-region that a district belongs to. For area, the average years of schooling for consistency with subsequent analysis in this chapter, we group single-district metro districts with multidistrict metro core ­districts. As described in chapter 1, single-district metro districts possess the district itself, again measured as an similar characteristics to multidistrict metro core ­districts. average over ­ 2008–15. 110  TIME TO ACT BOX ­3.2  Sources and types of agglomeration forces Urban economists have identified three closely •  Market ­access. Metro and urban areas can related and somewhat overlapping theories of generate higher productivity because they positive agglomeration forces that may help also tend to benefit from access to large con- explain the variation in underlying productiv- sumer markets and to supplier markets of ity levels across districts and places: agglomera- intermediate i ­nputs. This access stems both tion economies, human capital externalities, and from an area’s own internal market and from market ­access. its connectivity to other surrounding areas, potentially including overseas cities through •  Agglomeration ­economies. Metro and urban international gateways, such as p ­ orts. areas generate higher underlying productivity Greater access to consumer and supplier than rural areas because of the positive markets makes it easier for firms to cover the externalities created by their large population fixed costs of setting up a new plant, which size and/or densities (Duranton and Puga stimulates higher profits and productivity 2004;  Jacobs 1969; and Marshall ­ 1 890). (Fujita, Krugman, and Venables 1999; Agglomeration economies can arise through Krugman 1991a, 1991b; Krugman and mechanisms. The “thick” labor markets several ­ Venables 1 ­ 995). Again, this theory is closely in metro and urban areas can generate better related to the theory of agglomeration econ- matches between workers and firms, so that omies insofar as it focuses on a specific sub- each worker is more likely to find the most set of mechanisms through which positive suitable ­job. Metro and urban areas can also agglomeration forces may ­ a rise. It shares provide the conditions for the growth of a large with agglomeration economies theory the and diversified array of specialized suppliers of hypothesis that a larger internal market aids goods and services that provide the intermediate productivity by stimulating the growth of a inputs that help fuel the growth of the local large and diversified array of specialized sup- ­ economy. Finally, the geographic proximity of ­ oods. It also goes far- pliers of intermediate g people and firms within such areas can give rise ther by emphasizing connectivity to the to the spillover of ideas as workers learn from markets of other surrounding metro, urban, each other through observation and ­ interaction. and other ­ areas.b •  Human capital ­ e xternalities. This theory posits that metro and urban areas generate In estimating the strength of different types of higher productivity not so much because of agglomeration forces, economists have focused their size or density but because they have almost exclusively on developed countries more highly educated and skilled workforces, (Duranton 2016; Henderson 2010; Overman which generate positive ­ e xternalities. This and Venables 2 ­ 005). Recently, however, the theory can be considered a special case of the methods developed for empirically estimating theory of agglomeration economies by agglomeration forces in developed ­countries are focusing on just one of the several channels being applied to developing countries, including through which agglomeration economies can Colombia (Duranton 2016); Brazil, China, and positively affect a metro or urban area’s al. 2017); and other countries India (Chauvin et ­ productivity—the spillover of ideas between in Latin America and the Caribbean (Quintero people. The theory also hypothesizes that the ­ and Roberts 2018; Roberts ­ 2018). spillover of ideas is more likely to occur from higher-skilled workers than from lower-skilled Source: Based on Roberts ­2018. a. If unskilled and skilled workers are complementary inputs in the production workers, leading to the prediction that a processes within a district’s firms, then this can also generate observationally worker’s individual productivity will be an similar effects to human capital externalities (see, for example, Ferreyra 2018; increasing function of the average human Moretti ­2004). This idea is also discussed, to some extent, in chapter ­4. b. The analysis below excludes a district’s own market, as captured by capital of the location in which she or he its population, from the calculation of its market ­access. This helps avoid works (Moretti 2004; Rauch ­ 1993).a multicollinearity with the measure of agglomeration economies ­used. D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   111 •  Market access : Two variables capture core districts (see chapter 1), we combine access to domestic and international multidistrict metro core districts and single- markets: district metro areas into a single c ­ ategory. The °° Domestic market access is measured regression further includes fixed effects for using a variable for how well connected the type of district and the island-region to a district is to areas of population con- which the district belongs, as well as controls centration through Indonesia’s national for a district’s geography (average terrain rug- road and domestic ferry ­ networks. The gedness) and climate (annual mean tempera- variable is derived from digitized road ture and annual total ­ rainfall).10 and ferry network maps for the year The results, summarized in table 3 ­ .2, show 2011 (see annex ­ 3A). Even for districts the estimated elasticity of a district’s underly- that are part of a multidistrict metro ing level of productivity with respect to each area, we calculate this variable at the of the four agglomeration force variables and level. This allows the measure district ­ allow for the possibility of different elastici- to capture the effect of proximity and ties and, therefore, effects by type of district connectivity to the core for both urban (see annex 3B for complete ­ r esults). The and rural periphery ­ districts. results show that different variables—and, °° International market access is measured therefore, different types of agglomeration using a dummy variable that takes the force—matter for different types of district, value one if a district contains a “major” thereby implying the need for policies that are port within its administrative boundar- tailored by ­ place. Hence, for multidistrict ies and the value zero ­ otherwise.9 metro core districts and single-district metro districts, the population size of the metro area to which the district belongs is a key, statisti- Different types of agglomeration forces cally significant, determinant of underlying matter for districts belonging to different ­ productivity. For these districts, a doubling of places metro area population is associated with a To assess the importance of the different types large 1­ 0.9 percent increase in underlying of agglomeration forces for a district’s under- ­ productivity. A similar result also holds for lying productivity and analyze how these urban periphery districts, for which a dou- forces vary according to the type of place, a bling of metro area population is associated regression of the determinants of underlying with an even larger ­ 13.1 percent increase in productivity was e ­ stimated. This regression underlying ­ productivity. These results provide includes as independent variables population, evidence of strong traditional agglomeration average years of schooling, the measure of economies for urban districts—be they core domestic market access (all entered as natural or periphery—that belong to metro ­ areas. By logs), and the dummy variable for whether a contrast, such effects are almost entirely district has a major ­ port. To allow for the pos- absent for other types of district, be they rural sible existence of different effects by type of periphery districts or nonmetro districts of place, the regression further includes interac- any ­kind.11 tion terms between each of the aforemen- Domestic market access, meanwhile, tioned variables and the type of place that a ­ m atters—in terms of being statistically district either is or belongs to (whether the ­ significant—only for urban periphery d ­ istricts. district is part of a multidistrict metro core, an For these districts, a doubling of domestic urban periphery district, a rural periphery market access is associated with a ­ 2.9 percent d istrict, a single-district metro district, a ­ increase in underlying ­ productivity. The only ­ n onmetro urban district, or a nonmetro other type of district for which market access rural ­district). Because there are only seven comes close to having a statistically significant single-district metros in Indonesia and these positive impact is rural periphery districts, for areas, by construction, are similar in both which a doubling of market access is associ- population and density to multidistrict metro ated with a ­ 1.5 percent increase in underlying 112  TIME TO ACT TABLE ­3.2  Estimated elasticity of a district’s underlying productivity with respect to different agglomeration forces, by type of district Domestic market District type Population Schooling access Port Metro Core/single-district metro ­0.109 ­0.666 ­– 0.007 ­– 0.041 Urban periphery ­0.131 ­0.997 ­0.029 – Rural periphery ­– 0.002 ­0.069 ­0.015 – Nonmetro Nonmetro urban ­0.030 ­0.308 ­– 0.008 ­0.148 Nonmetro rural ­0.029 ­0.274 ­– 0.008 ­0.004 Note: Each cell in the table shows the estimated elasticity of a district’s underlying productivity with respect to a variable from a regression of the natural log of underlying productivity on all variables, where population, schooling, and domestic market access enter the regression in natural ­logs. The regression allows estimated elasticities to vary by type of place using interaction ­terms. Island fixed effects and geography and climate controls (terrain ruggedness, annual mean temperature, and annual total rainfall) are also included in the ­regression. Where a cell is highlighted in orange, this indicates a positive significant effect on underlying productivity at the 10 percent level or ­higher. For the full set of regression results, see table ­3B.1 in annex ­3B. ­ roductivity.12 For all other types of district, p metro core and single-district metro districts, the estimated elasticity of underlying produc- this result could be because the positive effect tivity with respect to market access is slightly on underlying productivity provided by better negative, but far from statistically significant access to international markets is negated by, cases. The fact that domestic market in all ­ for example, traffic congestion in and around access matters for a district, especially an port ­areas. Having a major port located at the urban district, in the periphery of a multidis- heart of a metro area also tends to represent a trict metro area is probably driven by its relatively inefficient use of l ­and. The signifi- proximity and strength of commuting flows cance of the port variable for nonmetro urban to the metro c ­ ore. This result is consistent districts should also not be interpreted as with this report’s ACT (Augment, Connect, implying that such districts should prioritize Target) policy ­framework. In particular, it sug- the building of ports, which clearly makes gests that the integration of urban and rural sense only for certain ­ locations. Rather, it periphery districts to the cores, through should be taken as signifying the more general improvements in metro transportation and importance of access to international markets, tackling traffic congestion problems, should which could, for example, be enhanced be a key policy priority for such districts through better land transport connectivity to (chapter ­7). existing ports located in other ­ districts. In contrast to domestic market access, the Finally, human capital externalities are presence of a major port within a district, the type of positive agglomeration force that which proxies for better access to interna- has the most general importance in the sense tional markets, matters only for nonmetro that average years of schooling have a urban ­districts. For such districts, the presence ­ significant positive influence on underlying of a major port has a large positive and statis- productivity for three of the five types of tically significant impact on underlying ­ district. Hence, the effect of average years of productivity. In particular, the presence of a ­ schooling is significantly positive for multi- port is associated with a 15 percent increase in district metro core and single-district metro underlying ­ productivity. By contrast, for other districts, urban periphery districts, and non- types of district for which the effect of a port metro rural ­ districts. The effect of schooling could be estimated,13 there is no significant also comes close to being significant for non- ­ impact. For districts belonging to multidistrict metro urban d ­ istricts.14 It is important to D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   113 note that the positive effect of average schooling on underlying productivity is in Varying effects of the addition to the direct effect of an individual’s business environment on own level of human capital—as driven by an productivity increase in either education level or work- Along with agglomeration forces, the local force experience—on his or her own produc- business environment can also drive spatial tivity and nominal ­ wage. In other words, the productivity.15 To examine its differences in ­ existence of a positive and significant effect role, the analysis switches from a focus on of average schooling is consistent with the workers to a focus on firms using data on presence of positive social returns to human productivity, as measured by firm labor pro- capital over and above the private r ­ eturns. ductivity and total factor productivity (TFP), These social returns are estimated to be high- and on different elements of the local business est in multidistrict metro core, single-district environment for a sample of up to 1,700 pri- metro, and urban periphery ­ districts. The vate firms drawn from the 2009 and 2015 bottom line is that—although policies to rounds of the World Bank Enterprise Survey improve human capital through, for exam- for ­Indonesia.16 These firms cover 38 places ple, augmenting the coverage and quality of (11 multidistrict metro areas, 1 single-district education services and tackling the underly- metro area, 11 nonmetro urban areas, and ing causes of stunting, can be expected to 15 nonmetro rural a ­ reas). A location’s improve productivity and nominal wages ­ business environment consists of three main everywhere because of positive private elements: the basic business environment, returns to human capital—the social returns the refined business environment, and the to such policies are highest in the urbanized agglomeration environment (figure ­ 3.4). parts of metro ­ areas. Basic business environment refers to those As shown in annex 3B (see columns [2a] aspects of the business environment that are and [2b] of table ­ 3B.1), the above results are fundamental for ­ development. It includes the qualitatively robust to replacing the simple basic functions of government protection, measure of population that we use to capture including containing corruption and safe- traditional agglomeration economies with an guarding against ­crime. It also encompasses a alternative weighted population variable that good supply of human capital and infrastruc- has recently been proposed by De la Roca and ture and access to finance (Beck, Demirgüç- Puga ­ (2017). This alternative variable—which Kunt, Maksimovic 2005; Bloom et ­ al. 2010; is more akin to a measure of density than population per se—measures the average number of people located within 10 kilome- FIGURE ­3.4  Dimensions of the local ters of a person within a metro or nonmetro business environment area, depending on which type of area a dis- to. With this alternative measure, trict belongs ­ • Basic protection we find evidence of even stronger traditional Basic business • Infrastructure agglomeration economies in multidistrict environment • Human capital metro core and single-district metro districts, • Access to nance and urban periphery d ­ istricts. This evidence is • Barriers to entry/exit consistent with the argument that policies, Refined business • Labor regulations such as those discussed in chapter 7, that environment • Tax environment work to enhance density in such areas through, for example, promoting transit-­ • In a big city oriented development and addressing conges- Agglomeration • Capacity agglomeration environment tion forces that might otherwise constrain the • Informal competition attractiveness of such areas to households and workers, are central to boosting the prosper- Source: Fang, Roberts, and Xu ­2018. Business environment is defined offer. ity that they ­ following Reyes, Roberts, and Xu ­2017. 114  TIME TO ACT Demirgüç-Kunt and Maksimovic 1998; the variation for four business environment La Porta et ­al. 1998; Levine ­ 1997). variables across metro areas (both multi- and Refined business environment includes the single-district) for the sample of firms entry and exit barriers for firms in a location, (figure ­ ­ 3.5). The share of firms that experi- as well as the location’s labor regulations and ence power outages ranges from a low of tax ­environment. 20 percent in Medan, already a high figure, to Agglomeration environment includes a high of 73 percent in Bandung (figure 3 ­ .5, whether a firm is based in a large city and panel ­ a). Levels of modern infrastructure therefore has the potential to benefit from (using the web to conduct business) are low positive agglomeration forces; 17 capacity throughout metro ­ areas. Even in Jakarta, agglomeration, or the concentration of firms only 42 percent of firms use the web to con- in a location with high capacity in technology, duct business, whereas in Mojokerto, just management, or adaptation to a changing 7 percent do ­ 3.5, panel ­ (figure ­ b). competitive environment; and exposure to Mojokerto performs well, however, for the informal competition in a location, measured share of firms that report a lack of availability by the share of firms that self-report as com- of skilled labor (see figure ­ 3.5, panel c) and peting with informal ­ firms. access to land (figure ­ 3.5, panel d) as moder- Capacity agglomeration is proxied by the ate or severe obstacles to doing b ­ usiness. The share of firms in a location that employ more situation is worse in other metro a ­ reas. In four than 50 ­workers. The use of this proxy is con- metro areas (Bandar Lampung, Semarang, sistent with evidence showing that large firms Makassar, and Salatiga), more than 35 percent are more productive and export more than of firms identify obstacles to hiring skilled smaller firms (Bernard et a ­ l. 2007; Melitz and labor as a moderate or severe constraint, Ottaviano 2008); are more innovative (Cohen and in an additional three (Surakarta, Medan, and Levin 1989); and conduct research and and Bandung), 20 percent or more of firms development more efficiently (Cohen and identify such an obstacle as a moderate or Klepper ­ 1996). Perhaps because the research severe ­constraint. For Jakarta, Surabaya, and and development activities of large firms pro- Denpasar, hiring skilled labor poses less of a vide key spillovers for small firms (Acs, ­ constraint. Access to land is a moderate or Audretsch, and Feldman 1994), large firms severe obstacle for more than 35  percent are associated with higher industrial agglom- of firms in four multidistrict metro  areas eration (Barrios, Bertinelli, and Strobl 2006; (Salatiga, Medan, Semarang, and Makassar) Holmes and Stevens ­ 2002). For the United and for 20 percent or more of firms in a fur- States, the relocation of large firms has been ther three multidistrict metro areas (Bandung, found to positively affect incumbent firms’ Bandar Lampung, and ­ Surakarta). Perhaps TFP (Greenstone, Hornbeck, and Moretti surprisingly, access to land, as with access to 2010), and firms are more likely to become skilled labor, appears to be less of a constraint large when they are colocated with other for firms in Jakarta, Surabaya, and ­ Denpasar. large firms (Li, Lu, and Wu ­ 2012). The same A possible explanation for this is that firms in proxy for capacity agglomeration helps these metro areas tend to be more service-­ explain China’s productivity advantage over oriented and therefore less land intensive in India in a quantitatively important way production (see chapters 1 and ­ 2). (Li, Long, and Xu 2017), and it has predictive power for firm-level job growth (Clarke, Qiang, and Xu ­ 2015). Indonesia has experienced improvements in infrastructure and capacity agglomeration, but deterioration in trade Indonesia’s business environment varies credit and perception of several obstacles widely across places Between 2009 and 2015, some aspects of the The business environment exhibits consider- business environment improved across loca- able variation across Indonesia, as shown by 3.6). tions, whereas others deteriorated (figure ­ D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   115 FIGURE ­3.5  Variations in the local business environment across Indonesia’s metro areas: Power outages, use of web to conduct business, skilled labor obstacle, and land access obstacle a. Power outages b. Use of web to conduct business Bandung Jakarta Bandar Lampung Salatiga Jakarta Denpasar Surakarta Semarang Mojokerto Makassar Makassar Medan Denpasar Bandung Semarang Bandar Lampung Salatiga Surabaya Surabaya Surakarta Medan Mojokerto 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 Percent of rms Percent of rms c. Moderate to severe lack of skilled labor d. Moderate to severe lack of access to land Bandar Lampung Salatiga Semarang Medan Makassar Semarang Salatiga Makassar Surakarta Surakarta Medan Bandar Lampung Bandung Bandung Jakarta Jakarta Surabaya Surabaya Denpasar Denpasar Mojokerto Mojokerto 0 10 20 30 40 50 0 10 20 30 40 50 60 Percent of rms Percent of rms Source: Calculations based on data from World Bank Enterprise Surveys for 2009 and 2015 ­(http://www.enterprisesurveys.org/). Note: Figures show the share of sample firms, which was calculated using a pooled sample of 2009 and 2015 World Bank Enterprise ­Surveys (http://www​ .enterprisesurveys.org/). Metro areas with fewer than 20 observations are not ­included. There have been statistically significant (at the constraints to their businesses rose between 10 percent level or higher) improvements in 2009 and 2 ­ 015. There has also been a signifi- the share of firms not reporting power out- cant deterioration in the share of firms with ages and the share using the web to conduct credit. access to trade ­ business, indicating improvements in both traditional and modern infrastructure across ­ locations. Likewise, there has been a signifi- Several aspects of the business environment significantly affect firm productivity cant improvement in capacity a ­ gglomeration. Against this, however, are the worsening con- To determine which aspects of the business ditions for access to skilled labor, for labor environment matter most for productivity, a regulation, and for tax obstacles, as the share firm’s level of productivity (labor productiv- of firms reporting these as moderate or severe ity or TFP) was regressed on aspects of the 116  TIME TO ACT local business environment of the place in FIGURE ­3.6  Some aspects of the business which the firm is ­located. To control for the environment improved and some worsened across sorting of firms across locations, which locations in Indonesia between 2009 and 2015 results in differences in the composition of the local economy, the regressions also No power outage 0.35 include controls for observable firm charac- Capacity agglomeration 0.16 teristics including size, age, industry, and Web use 0.11 ownership ­structure.18 The regressions reveal that only a few Access to skilled labor –0.10 aspects of a location’s basic business environ- Labor regulation obstacle –0.13 ment have a statistically significant associa- Tax rate obstacle –0.13 tion with firm productivity (figure ­3.7). A lack Access to trade credit –0.31 of basic protection against crime—as proxied by high average spending by firms on security –0.4 –0.2 0 0.2 0.4 (high value of “payments for s ­ ecurity”)—has Mean di erence in the shares of rms (2015 minus 2009) a significant negative impact on a firm’s labor productivity and ­ TFP. The presence Source: Calculations based on data from World Bank Enterprise Surveys for 2009 and 2015 ­(http://www​ of modern infrastructure (“web use”) and .enterprisesurveys.org/). access to finance (“access to trade credit”) in Note: The figure reports business environment aspects with a statistically significant change (at the 10 percent level or higher) between 2009 and ­2015. Positive (negative) values correspond to an a location has a significant positive associa- improvement (deterioration) in a business environment variable. It includes only multi- and single- tion with firm  ­ p roductivity. The refined district metro areas and nonmetro urban ­districts. FIGURE ­3.7  Estimated effects of the local business environment on firm productivity Payment for security * * Corruption Power outages ** Basic Web use ** Access to skilled labor Access to overdraft facility Access to trade credit * ** Access to land Re ned Labor regulation obstacle Tax rate obstacle Agglomeration In a big city Capacity agglomeration *** *** Informal competition –1.0 –0.5 0 0.5 1.0 1.5 2.0 2.5 3.0 Estimated e ect on rm productivity Labor productivity Total factor productivity Source: Calculations based on data from World Bank Enterprise Surveys for 2009 and ­2015 (http://www.enterprisesurveys.org/). Note: Figure reports estimated elasticities (based on the regression results reported in annex 3C) for a firm’s level of productivity (labor productivity or total factor productivity) with respect to different dimensions of the business environment of the area in which it is ­located. The regressions control for observable characteristics of the firm (firm size, age, industry, and ownership structure) on which sorting may occur, and they also include year fixed ­effects. * Significant at the 10 percent level; ** significant at the 5 percent level; *** significant at the 1 percent ­level. D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   117 business environment appears not to matter congestion constraints discourage density ­ roductivity. As for the agglomera- for firm p because people are deterred by the unpleas- tion environment, a location’s “capacity antness of living in these ­ areas. As discussed agglomeration” has a significant positive in chapter 7, density can also be cultivated by impact on firm productivity, whereas whether more integrated and effective land use and a firm is “in a big city” has no significant transportation planning within large cities productivity. impact on its ­ that emphasize, for example, compact and transit-oriented ­development. For urban and rural districts in the periph- Conclusion and policy eries of multidistrict metro areas, the signifi- cance of domestic market access, which implications largely stems from the proximity of these dis- Consistent with a lack of geographic labor tricts to multidistrict metro cores, is consistent mobility, which may be associated with a rela- with a policy priority to better connect tive lack of national factor market integration peripheries with their c ­ ores. This can be and thus a failure of policy to adequately con- achieved through measures that, for example, nect places, the wide variations in prosperity improve transport connectivity between across the portfolio of places that were periphery and core districts and tackle prob- described in chapters 1 and 2 are mirrored by lems of traffic congestion (chapter ­ 7). For large variations in average productivity across nonmetro urban districts, market access also districts that belong to different types of matters, but here it is more access to interna- ­ place. Although differences in the composi- tional markets—which can be improved by, tions of local economies and their workforces for example, better connectivity to major attributable to sorting play an important role international ports—that appears important in explaining their variations in productivity, (chapter ­6). the bigger role is played by differences in Meanwhile, policies to improve human underlying productivity associated with capital—among which policies to augment “place-based ­ advantages.” These place-based the coverage and quality of basic services are advantages arise from a variety of agglomera- key—are important for all districts regardless tion forces, as well as from variations in the of the type of place they belong ­ to. This is local business environment, thereby opening because investments in human capital directly an important role for policy to influence boost the individual productivity, and there- p roductivity and, through productivity, ­ fore nominal wages, of w ­ orkers. However, in ­prosperity. addition to the private returns to human capi- Because, however, different types of tal, there are also significant positive social agglomeration forces matter for districts that returns associated with the presence of human belong to different types of place, policies capital ­externalities. These social returns are need to be ­tailored. The importance of tradi- highest in the cores of multidistrict metro tional agglomeration economies for multidis- areas, single-district metro districts, and trict metro core districts, single-district metro urban peripheries: from a productivity per- areas, and urban periphery districts, especially spective, policies to improve human capital in when estimated using a weighted measure of these areas bring an extra bang for the policy population, points to the importance of better maker’s ­buck. managing and cultivating density in these Finally, the fact that there is such a ­ areas. This improvement can be achieved strong connection between variations in through policies, including policies to aug- average productivity and prosperity across ment, to address key congestion constraints portfolio-of-place districts is consistent that arise in these areas from the pressure of with the need for more general policies and urban population on local infrastructure, reforms to improve connectivity and inte- basic services, land, housing, and the gration across the portfolio of places ­ environment. By undermining livability, these ­ ( chapter 6­ ). The reason is that policies to 118  TIME TO ACT connect and integrate nationally across areas—which arise from both traditional places can help to decouple spatial varia- agglomeration ­ e conomies and stronger tion in prosperity from spatial variation in human capital ­ e xternalities—to be more ­ productivity. Doing so will allow the higher widely shared with other types of place, underlying productivity of multidistrict thus promoting urbanization that is more metro cores and single-district metro places. inclusive across the portfolio of ­ D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   119 Annex 3A Construction of the domestic market access variable Following Blankespoor et a ­ l. (2017) and road network and ferry routes between Jedwab and Storeygard (2018), among oth- islands that was published by Nelles Verlag in ers, we calculated domestic market access for 2011. Information on road categories from ­ a given district as a function of the weighted the map were used to produce travel time esti- sum of the populations of all other districts, mates by assuming different average travel with a weight that decreases with transport speeds for the different c­ ategories. The fol- ­time: lowing average speeds were assumed: express- way–105 kilometers/hour (km/h); principal MAi ,t = ∑P −θ j ,t τ ij ,t ,­(3A.1) highway–100 km/h; highway–90 km/h; sec- j≠i ondary road–75 km/h; track–50 km/h; road where MA i,t denotes the level of market of unknown category–6 km/h; and 5 km/h in access of district i in year t, Pj,t the popula- the absence of a r­ oad. Travel times are esti- tion of district j in year t, and t ij,t the esti- mated between district capitals, as identified mated travel time between districts i and j in by Humanitarian Data Exchange ­ (2013). For year ­t . q is a trade elasticity parameter districts with more than one city within their which, following Donaldson (2018), is set 1996 boundaries, the grid cell with the high- equal to ­3.8. est population from Landscan-1998 is u ­ sed. Market access for districts as defined by Where no city location can be identified, the their 1996 administrative boundaries for location was added manually considering the 2011 is calculated using travel times derived population density and snapped to the road from a digitized paper map of Indonesia’s network ­data. 120  TIME TO ACT Annex 3B Determinants of underlying productivity TABLE ­3B.1  Regression results for key drivers of a district’s underlying productivity Population Weighted population Dependent variable: Underlying productivity (ln) [1a] [1b] [2a] [2b] ln (population) ­0.051*** ­0.029 ­– 0.009 ­– 0.045**   ­[2.64] ­[1.41] ­[–0.46] ­[–2.00] (Core / SDM) × ln (population) – ­0.080*** – ­0.161***   ­ 3.37] [ ­[3.01] (Urban periphery) × ln (population) – ­0.102*** – ­0.205***   ­ 3.34] [ ­[3.90] (Rural periphery) × ln (population) – ­– 0.032 – ­– 0.058   ­[–0.58] ­[–1.08] (Nonmetro urban) × ln (population) – ­0.001 – ­0.013   ­ 0.03] [   ­[0.49] ln (schooling) ­0.475*** ­0.274* ­0.567*** ­0.351**   ­[5.40] ­[1.76] ­[4.80] ­[2.20] (Core / SDM) × ln (schooling) –  ­0.392 – ­0.191     ­[1.23] ­[0.48] (Urban periphery) × ln (schooling) –  ­0.723 – ­0.123     ­[1.55] ­[0.26] (Rural periphery) × ln (schooling) –  ­– 0.205 – ­0.042     ­[–0.58] ­[0.14] (Nonmetro urban) × ln (schooling) –  ­0.033 – ­– 0.093     ­[0.13]   ­[–0.43] ln (market access) ­– 0.007** ­– 0.008** ­– 0.006* ­– 0.003   ­[–2.52] ­[–2.18] ­[–1.82] ­[–0.88] (Core / SDM) × ln (market access) –  ­0.000 – ­– 0.004     ­ 0.01] [ ­[–0.36] (Urban periphery) × ln (market access) –  ­0.036** – ­0.041**     ­[1.97] ­[2.49] (Rural periphery) × ln (market access) –  ­0.023** – ­0.019***     ­[2.27] ­[2.60] (Nonmetro urban) × ln (market access) –  ­– 0.001 – ­– 0.005     ­[–0.21]   ­[–1.16] Table continued on next page D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   121 TABLE ­3B.1  Continued Population Weighted population Dependent variable: Underlying productivity (ln) [1a] [1b] [2a] [2b] Port ­0.033 ­0.004 ­0.042 ­0.025   ­[1.10] ­[0.13] ­[1.21] ­[0.97] (Core / SDM) × Port –  ­– 0.044 –  ­– 0.098     ­[–0.84]   ­[–1.58] (Nonmetro urban) × Port –  ­0.145 –  ­0.130     ­[1.78]   ­[1.52] Constant ­6.251*** ­8.311*** ­7.174*** ­9.220***   ­ 3.73] [ ­[5.12] ­[4.33] ­[5.54] Observations 285 285 285 285 R-squared ­0.808 ­0.846 ­0.784 ­0.839 Adjusted R-squared ­0.799 ­0.827 ­0.774 ­0.819 Source: Calculations based on data on workers from the August 2008–August 2015 rounds of Indonesia’s National Labor Force Survey ­(SAKERNAS). Results in columns [1a] and [1b] are based on population and schooling measured at the metropolitan area level, where, for nonmetro areas, population and schooling are given by district-level population size and average years of ­schooling. Results in columns [2a] and [2b] are based on the same specifications as those in columns [1a] and [1b], respectively, except that, following De la Roca and Puga (2017), the measure of population is replaced by a weighted population measure that measures the average number of people located within 10 kilometers of a person within a metropolitan or nonmetro ­area. Population and average years of schooling were compiled using data from Indonesia’s National Socio-Economic Surveys (SUSENAS) for 2008–15 (July rounds for 2008–10; pooled multirounds [March, June, September, and December] for 2011–14; and March round for ­2015). Weighted population was calculated using Landscan-2012 gridded population ­data. Note: Schooling refers to average years of schooling for the population ages 15 and ­older. “Core” refers to multidistrict metro core district and SDM to single- district ­metro. Numbers in brackets are robust ­t-statistics. Standard errors are clustered by place (that is, by metro or, for nonmetro areas, by ­district). Regressions also include measures of terrain ruggedness (ln), annual mean temperature (ln), and annual total rainfall (ln) as geographic ­controls. Area-type and island fixed effects were controlled as ­well. * Significant at the 10 percent level; ** significant at the 5 percent level; *** significant at the 1 percent ­level. 122  TIME TO ACT Annex 3C Effects of the local business environment on firm productivity TABLE ­3C.1  Regression results for effects of the local business environment on firm productivity Dependent variable (ln) Total factor productivity Labor productivity Basic business environment Payments for security ­– 0.639* ­– 0.497* ­[–1.81] ­[–1.75] Corruption obstacle ­– 0.137 ­– 0.523 ­[–0.27] ­[–1.59] Power outage ­– 0.070 ­0.081 ­[–0.23] ­[0.29] Web use ­1.045** ­0.814** ­[2.31] ­[2.29] Skilled labor obstacle ­– 0.179 ­0.454 ­[–0.39] ­[1.15] Overdraft facility ­0.742 ­0.653 ­[1.34] ­[1.39] Trade credit ­1.064** ­0.697* ­[2.04] ­[1.70] Refined business environment Land access obstacle ­– 0.218 ­– 0.585 ­[–0.40] ­[–1.44] Labor regulation obstacle ­– 0.304 ­– 0.203 ­[–0.49] ­[–0.38] Tax rate obstacle ­– 0.541 ­– 0.024 ­[–0.93] ­[–0.06] Agglomeration environment In a big city ­0.135 ­0.003 ­[0.48] ­[0.02] Capacity agglomeration ­1.548*** ­2.781*** ­[2.65] ­[6.03] Informal competition ­0.081 ­0.342 ­[0.23] ­[1.22] Observations 1,121 1,653 Adjusted R-squared ­0.102 ­0.300 Source: Calculations based on data from World Bank Enterprise Survey for 2009 and 2015 for ­Indonesia (http://www.enterprisesurveys.org/). Note: Numbers in brackets are robust ­t-statistics. Standard errors are heteroskedasticity corrected and clustered at the location-industry ­level. Regressions includes measures of a firm’s size, age, industry, and ownership structure to control for sorting of firms across ­locations. Year fixed effects were also ­controlled. * Significant at the 10 percent level; ** significant at the 5 percent level; *** significant at the 1 percent ­level. D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   123 Notes measure of productivity, see Combes and Gobillon ­(2015). 1. When labor is geographically very mobile, 5. As will become evident, we do this by adopt- differences in welfare tend to equalize, at the ing a two-stage regression design that mir- margin, for workers of a given type across rors the logical structure set out in figure ­ 3.2. places even in the presence of substantial Such a regression design has been widely variations in the productivity of the places used in the urban economics literature that ­ themselves. This mobility allows for spread- tests for agglomeration forces, including ing the welfare gains of urbanization and Ahrend, Gamper, and Schumannet (2014); agglomeration across ­ places. Combes, Duranton, and Gobillon (2008); 2. The two basic reasons identified in this para- De la Roca and Puga (2017); Quintero and graph—sorting and place-based ­ advantages— Roberts (2018); and Roberts ­ (2018). One for why one location may be more productive big advantage of this design is that it permits than another are not mutually e ­ xclusive. For the explicit estimation of underlying levels of example, one important source of positive district productivity and, in so doing, facili- agglomeration forces is human capital exter- tates the decomposition of the role of sorting nalities, which arise from differences in the versus underlying productivity in accounting average level of human capital across p ­ laces. for variations in average district ­ productivity. However, such differences are themselves in A potential concern with the approach, how- part a reflection of ­sorting. Similarly, place- ever, is that, because the levels of underly- based advantages can act as a stimulus for ing productivity that act as the dependent migration and net firm entry, thereby lead- variable in the second-stage regression are ing to sorting, which generates compositional themselves estimated (in the first stage), the differences in the local economies and work- standard errors in the second stage may be forces of ­ places. unreliable. Given the number of observa- ­ 3. This work is similar in spirit to that recently tions in our first-stage regression (more than published for Latin America and the 1 million), however, all estimates of under- Caribbean in the World Bank report, Raising lying productivity in the first stage are very the Bar for Productive Cities in Latin America precise, which helps to minimize concerns and the Caribbean (Ferreyra and Roberts about standard errors in the second-stage 2018, chapter ­ 3). The work also contributes ­ regression. Discussions by the team with to an emerging literature that tests for agglom- leading urban economists have also pro- eration forces in developing countries, about vided reassurance about the legitimacy of which there has been little rigorous ­ research. ­ the two-stage ­ approach. In an accompanying 4. Although, as noted earlier, this literature background paper to this report by Bosker, has been confined largely to developed Park, and Roberts (2018), we also estimate ­ countries. Notable exceptions are Chauvin the strength of agglomeration forces using a al. (2017), Duranton (2016), Quintero and et ­ single-stage regression ­ design. Roberts (2018), and Roberts ­ (2018). Whereas 6. The availability of a panel of workers enables Chauvin et ­al. consider the empirical determi- Combes, Duranton, and Gobillon (2008) to nants of city productivity for Brazil, China, ­ egressions. include worker fixed effects in their r and India (and the United States), Duranton This allows them to control for time-invariant considers them for C ­ olombia. Quintero and unobservable worker ­ characteristics. Roberts (2018) and Roberts (2018), mean- 7. In addition to the exercise reported in table while, analyze the empirical determinants of 3.1, we also undertook a complete variance ­ spatial productivity variations for 16 Latin decomposition of nominal wages at the indi- American and Caribbean ­ countries. Nominal vidual worker level similar to that reported wages reflect productivity because the only by Combes, Duranton, and Gobillon (2008) way a firm in a given place or district can, in for French employment a ­ reas. Unsurprisingly, spatial equilibrium, afford to pay a worker a at the level of the individual worker, differ- higher nominal wage is if that worker is more ences in the characteristics of the worker productive. Spatial equilibrium, meanwhile, is ­ and the job he or she occupies explain more a situation in which no worker or firm has of the  variation than does the district in an incentive to move ­ location. For a more which the worker is e ­ mployed. Nevertheless, detailed explanation of why nominal rather even in this exercise, underlying productiv- than real wages provide the appropriate ity shows both a reasonably high standard 124  TIME TO ACT deviation across workers (it is about one- agglomeration is somewhat cruder than the third of the standard deviation for individual continuous measure of population used in nominal wages) and is moderately correlated the earlier worker-based ­ analysis. ­ with individual wages, implying that location 18. The regression specifies productivity (labor ­matters. productivity or TFP) in natural l ­ogs. Because 8. For single-district metro areas, the popula- the regressions are based on the pooling of data tion size of the metro area is simply equal to for 2009 and 2015, they also include year fixed the population size of the district ­ itself. ­ effects. Standard errors are h­ eteroskedasticity-​ 9. The 2015 “main” port hierarchy cat- corrected standard errors and clustered at egory was retrieved from Australian AID the location-industry l ­evel. Three ownership (2012) to identify major ports and com- control variables are included—the gender of bined with location data from World Port the owner of the firm as captured by a binary Index ­(https://msi.nga.mil/NGAPortal/MSI​ dummy variable, the share of foreign owner- .portal?_nfpb=true&_pageLabel=msi_portal​ ship in the firm, and the ownership share of _page_62&pubCode=0015) and Google owner. the largest ­ Maps to identify the districts in which major ports are ­ located. 10. In the regression, we take nonmetro rural areas as the omitted ­ category. References 11. Even though they are statistically insignifi- Acs, Z ­.J ­ ., ­ D. ­B. Audretsch, and M ­ .P Feldman. ­. ­ cant, the estimated elasticities of underlying 1994. “R&D Spillovers and Recipient Firm ­ productivity with respect to population are, ­Size.” Review of Economics and Statistics 76 at 3 percent and ­ 2.9 percent, respectively, (2): ­336–40. still quite large for nonmetro urban and rural Ahrend, ­ R ., ­C . Gamper, and ­ A. ­S chumannet. areas. These estimated elasticities are in the ­ ­ 2014. “The OECD Metropolitan Governance range that have been found in the literature ­S urvey.”  Regional Development Working for developed countries when testing for tra- Paper 2014/04, Organisation for Economic ditional agglomeration economies (Rosenthal Co-operation and Development, P ­ aris. http:// and Strange ­ 2004). doi:10.1787/5jz43zldh08p-en. 12. The p-value for domestic market access for Australian ­ Aid. ­2012. “Academic Paper to Support rural periphery districts is ­ 0.117, indicating National Port Master Plan Decree: Creating significance at the 12 percent ­ level. an Efficient, Competitive and Responsive Port 13. No effect could be estimated for urban and System for I ­ndonesia.” Indonesia Infrastructure rural periphery districts due to the absence of Initiative Working Paper, Australian Aid, major ports in these ­ districts. ­Canberra. 14. For this type of district, p = ­ 0.121. Barrios, ­ S ., ­ L . Bertinelli, and ­E. ­S trobl. ­2 006. 15. The analysis and results presented here are “Geographic Concentration and Establishment based on the background paper by Fang, Scale: An Extension Using Panel ­ Data.” Journal Roberts, and Xu (2018) prepared for this of Regional Science 46 (4): ­ 733–46. ­report. Beck, ­T., ­A. Demirgüç-Kunt, and ­V. ­Maksimovic. 16. We estimate TFP as the residual from an ­ 2 005. “Financial and Legal Constraints to industry-specific production function with Growth: Does Firm Size Matter?” Journal of log value added as the dependent variable Finance 60 (1): ­ 137–77. and log capital and log labor as the inde- Bernard, ­ A. ­B., J B. Jensen, S ­. ­ J. Redding, and ­. ­ pendent ­ variables. Capital is measured as ­P.  ­K.  ­Schott. ­2007. “Firms in International the replacement cost of land and machin- ­Trade.” Journal of Economic Perspectives ery while labor is measured as the number 21 (3): ­105–30. of permanent employees plus 0 ­ .5 times the Blankespoor, ­ B., ­T. Bougna, ­R. ­G. Rivera, and number of temporary ­ employees. World ­ H. ­ Selod. ­ 2017. “Roads and the Geography Bank Enterprise Survey data can be accessed of Economic Activities in M ­ exico.” Policy at http://www.enterprisesurveys.org/. Research Working Paper 8226, World Bank, 17. In this case, a large city is defined as either Washington, ­DC. Indonesia’s capital (Daerah Khusus Ibukota Bloom, ­ N ., ­ A . Mahajan, ­ D. McKenzie, and Jakarta) or a location with more than ­ Roberts. ­ J. ­ 2010. “Why Do Firms in Developing 1  million ­ residents. Because it is a simple Countries Have Low Productivity?” American binary dummy variable, this measure of Economic Review 100 (2): ­ 619–23. D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   125 Bosker, ­ M ., ­ J. Park, and ­ M. ­ R oberts. ­ 2 018. Latin America and the Caribbean, edited by “Definition Matters: Metropolitan Areas ­ M. ­ M . Ferreyra and ­ M . Roberts, ­ 1 41–66. and Agglomeration Economies in a Large Washington, DC: World ­ Bank. Developing ­ Country.” Policy Research Working Ferreyra, ­ M. ­ M ., and ­ M . Roberts, ­ 2 018. e ds. ­ Paper 8641, World Bank, Washington, ­ DC. Raising the Bar for Productive Cities in Latin Chauvin, ­ J. ­ P., ­E. Glaeser, ­ Y. Ma, and ­ K. ­ Tobio. America and the ­ Caribbean. Washington, DC: 2017. “What Is Different About Urbanization ­ World ­Bank. in Rich and Poor Countries? Cities in Brazil, Fujita, ­M., ­P. Krugman, and ­A. ­J. ­Venables. ­1999. China, India and the United ­ States.” Journal of The Spatial Economy: Cities, Regions, and Urban Economics 98 (C): ­ 17–49. International ­Trade. Cambridge, MA: MIT P ­ ress. Clarke, ­ G., ­C. ­ Z. Qiang, and L ­.­ C. ­ Xu. ­2015. “The Greenstone, ­ M., ­ R. Hornbeck, and ­ E. ­Moretti. Internet as a General-Purpose ­ Technology.” ­2010. “Identifying Agglomeration Spillovers: Economics Letters 135 (C): ­ 24–27. Evidence from Winners and Losers of Large Cohen, ­ W. ­ M., and ­ S. ­ Klepper. ­ 1996. “Firm Size and Plant ­Openings.” Journal of Political Economy the Nature of Innovation within Industries: The 118 (3): ­ 536–98. Case of Process and Product ­ R&D.” Review of Henderson, ­J. ­V. ­2010. “Cities and ­Development.” Economics and Statistics 78 (2): ­ 232–43. Journal of Regional Science 50 (1): ­ 515–40. Cohen, ­W. ­M., and ­R. ­C. ­Levin. ­1989. “Empirical Holmes, ­T. ­J., and ­J. ­J. ­Stevens. ­2002. “Geographic Studies of Innovation and Market S ­ tructure.” Concentration and Establishment ­ Scale.” Review In Handbook of Industrial Organization, of Economics and Statistics 84 (4): ­ 682–90. Volume 2 , edited by ­ R . Schmalansee and Humanitarian Data ­ Exchange. ­ 2013. “Indonesian ­R. ­D. Willing, ­1059–107. Amsterdam: ­Elsevier. Capital C ­ ities.” https://data.humdata.org​ Combes, ­ P., ­G. Duranton, and ­ L. ­ Gobillon. ­ 2008. /dataset/indonesian-capital-cities. “Spatial Wage Disparities: Sorting Matters!” Jacobs, ­J. ­1969. The Economy of C ­ ities. New York, Journal of Urban Economics 63 (2): ­ 723–42. NY: Random ­ House. Combes, ­ P., and L ­.­Gobillon. ­ 2015. “The Empirics Jedwab, ­R ., and ­A . ­S toreygard. ­2 018. “The of Agglomeration E ­ conomies.” In Handbook Average and Heterogeneous Effects of of Regional and Urban Economics, Volume 5, Transportation Investments: Evidence from edited by G ­ . Duranton, J V. Henderson, and ­. ­ Sub-Saharan Africa ­ 1960–2010.” Discussion ­W. Strange, ­247–348. Amsterdam: ­Elsevier. Paper, Department of Economics, Tufts De la Roca, ­ J., and ­ D. ­ Puga. ­ 2017. “Learning by University, Medford, ­ MA. Working in Big ­ Cities.” Review of Economic Krugman, ­P. ­1 991a. Geography and ­Trade . Studies 84 (1): ­ 106–42. Cambridge, MA: MIT ­ Press. Demirgüç-Kunt, ­A., and ­V. ­Maksimovic. ­1998. ———. ­ 1991b. “Increasing Returns and Economic “Law, Finance, and Firm ­ Growth.” Journal of ­Geography.” Journal of Political Economy 99 Finance 53 (6): ­ 2107–37. (3): ­483–99. Donaldson, ­ D. ­ 2 018. “Railroads of the Raj: Krugman, ­P., and ­A. ­J. ­Venables. ­1995. “Globalization Estimating the Impact of Transportation and the Inequality of N ­ ations.” Quarterly Journal ­Infrastructure.” American Economic Review of Economics 110 (4): ­ 857–80. 108 (4–5): ­ 899–934. La Porta, R ­ . Lopez-de-Silanes, ­ ­ ., F A. Shleifer, and Duranton, ­ G. ­ 2016. “Agglomeration Effects in ­R. ­Vishny. ­1998. “Law and ­Finance.” Journal of ­Colombia.” Journal of Regional Science 56 (2): Political Economy 106 (6): ­ 1113–55. ­210–38. Levine, R ­.­ 1997. “Financial Development and Duranton, ­ G ., and ­ D. ­P uga. ­ 2 004. “Micro- Economic Growth: Views and ­ Agenda.” Journal Foundations of Urban Agglomeration of Economic Literature 35 (2): ­ 688–726. ­Economies.” In Handbook of Regional and Li, ­D., ­Y. Lu, and ­M . ­Wu. ­2 012. “Industrial Urban Economics , Volume 4: Cities and Agglomeration and Firm Size: Evidence from Geography, edited by ­ J. ­ V. Henderson and ­ J.-F. ­China.” Regional Science and Urban Economics Thisse, ­2063–117. Amsterdam: ­Elsevier. 42 (1–2): ­ 135–43. Fang, ­ M. Roberts, and ­ S., ­ L.C. X ­ u. ­2018. “Female Li, ­ W., ­ X. ­ C . Long, and ­ L. ­ C. ­ X u. ­ 2 017. Corporate Leaders, Business Environment, and “Regulation, Agglomeration, and the Reversal Short- and Long-term Productivity Growth: of Fortune between China and ­ India.” Working Evidence from Indonesian ­ Firms.” Background Paper, World Bank, Washington, ­ DC. paper for this ­ report. Melitz, ­ M. ­J., and ­ Ottaviano. ­ G. ­ 2008. “Market Ferreyra, ­ M. ­ M. ­ 2018. “Human Capital in ­ Cities.” Size, Trade and ­ P roductivity.” Review of In Raising the Bar for Productive Cities in Economic Studies 75 (1): ­ 295–316. 126  TIME TO ACT Mincer, ­J. ­1 974. Schooling, Experience, and Evidence from the ­ Cities.” Journal of Urban ­Earnings. New York, NY: National Bureau of Economics 34 (3): ­ 380–400. Economic ­Research. Reyes, J M. Roberts, and L ­ ., ­ ­ u. 2 ­ .C. X ­ 017. “The Moretti, ­ E. ­ 2004. “Human Capital Externalities in Heterogeneous Growth Effects of the Business ­Cities.” In Handbook of Regional and Urban Environment: Firm-Level Evidence for a Global Economics, Volume 4: Cities and Geography, Sample of ­ Cities.” Policy Research Working edited by J ­. ­ V. Henderson and J ­.-F. Thisse, Paper 8114, World Bank, Washington, ­ DC. ­2243–91. Amsterdam: ­Elsevier. Roberts, M ­ .2 ­ 018. “The Empirical Determinants Overman, ­H. ­G., and ­A. ­J. ­Venables. ­2005. “Cities of City P ­ roductivity.” In Raising the Bar for in the Developing ­ World.” Discussion Paper Productive Cities in Latin America and the Carib- ­ 695, Center for Economic Performance, bean, edited by ­ M. ­M. Ferreyra and ­ M. Roberts, London School of Economics and Political ­89–115. Washington, DC: World ­Bank. Science, London. Rosenthal, ­S. ­S., and ­W. ­C. ­Strange. ­2004. “Evidence Quintero, L ­ ., and M­ .R ­ oberts. 2 ­ 018. “Explaining on the Nature and Sources of Agglomeration Spatial Variations in Productivity: Evidence ­E conomies.” In Handbook of Urban and from 16 Latin American and Caribbean Regional Economics, Volume 4, edited by Countries.” Policy Research Working Paper ­ ­J. ­V. Henderson and ­J.-F. Thisse, ­2 119–71. 8560, World Bank, Washington, ­ DC. New York, NY: North ­ Holland. Rauch, J ­. ­E. 1 ­ 993. “Productivity Gains from Verlag, ­N. ­2011. Indonesia ­1:4,500,000. Munich, Geographic Concentration of Human Capital: Germany: Nelles Verlag ­ GmbH. D ri v ers of P roducti v ity a n d P rosperity across the P ortfolio of P laces   127 Drivers of Urban and Spatial Inclusion 4 KEY MESSAGES As a large and far-flung archipelago, Indonesia undermines the potential benefits of agglomera- faces unique challenges in achieving inclusive tion through human capital externalities. urbanization. Welfare disparities across space are sizable and have remained persistent, sug- •  Spatial patterns of welfare in Indonesia gesting that the gains from urbanization are not m irror the wide productivity differences ­ being broadly shared across the country. Although across places documented in chapter 3. place-specific characteristics related particularly Welfare differences are large and persistent to road density and access to markets explain a across island-regions and the portfolio of larger share of the disparities, lingering differ- places. Although some places have caught up, ences in the levels of human capital of the popula- others continue to lag. tion matter, too. Migration has yielded economic •  Consistent with the results for productivity, rewards for those who migrated, but it has the the characteristics of places explain a larger potential to do much more to help spread the share of the spatial variation in welfare than benefits of urbanization to the rest of the coun- do the characteristics of their residents. This try. Metro areas provide better economic oppor- finding highlights the need for greater inte- tunities, but the benefits of agglomeration appear gration across places, through policies that to accrue much more to the high skilled than to better connect the portfolio of places, to share the low skilled, suggesting relatively weak human the prosperity of leading areas. The charac- capital spillovers across skill categories. A high teristics of residents also matter, because degree of residential segregation has led to con- access to some basic services, especially when centrations of human capital in specific neighbor- adjusted for quality, remains uneven. hoods within metros. Together with the lack of •  Within metro areas, the benefits of agglom- integration within these metros, this segregation eration have not accrued evenly to workers of 129 all skill levels: high-skilled workers benefit •• High concentrations of human capital in spe- from living and working in urban areas, but cific neighborhoods within metro areas and spillovers to low-skilled workers are less pro- the lack of integration due to policy failures nounced. As a result, differences in educa- at the housing–transportation nexus are tional attainment explain a large share of undermining the potential benefits of agglom- within-place inequality. eration through human capital externalities. As described in chapter 2, Indonesia has consumption increased by more than made considerable progress in reducing pov- 8 ­ percentage points in all areas of Indonesia’s erty and vulnerability. The national poverty portfolio of places except single-district metro and vulnerability rate declined from areas. The largest increase was in the rural 54 ­ percent in 2001 to 31 ­ percent in 2017, peripheries of multidistrict metro areas a  23-percentage-point drop. 1 Jawa-Bali, (10 percentage points), which are most likely Maluku, Nusa Tenggara, and Papua had the to have labor markets with blended charac- largest declines, at close to 25  ­ percentage teristics of urban and rural economies: prox- points, compared with 18 ­ percentage points imity to metro cores provides ample in Sumatera, which had the smallest. Except opportunities outside the primary sector, but for nonmetro rural areas, where poverty was sizable employment remains in agriculture already highest, poverty and vulnerability (see figure 1.9 in chapter 1). Inequality declined most in urban peripheries of multi- increased 8 ­ percentage points in multidistrict district metro areas (24 ­ percentage points) metro cores and 9 ­ percentage points in urban and rural peripheries of multidistrict metro peripheries of multidistrict metro areas (see areas (28 ­ percentage points). The smallest chapter 2). During the same period, inequality declines were in multidistrict metro cores increased in all major multidistrict metro percentage points) and single-district metro (7 ­ areas in Indonesia. Sukabumi saw the largest areas (13 ­percentage points), where poverty increase (10.9 ­ percentage points); Denpasar and vulnerability rates were already low. (3.3  ­p ercentage points) and Yogyakarta The share of the middle-class population p ercentage points) saw the smallest. (3.0  ­ has grown commensurately, from 5 ­ percent in Jakarta, Bandung, Surabaya, and Medan— 2001 to 22 ­ percent in 2017.2 It grew most in the largest and most populous metro areas— urban peripheries of multidistrict metro areas all saw inequality increase more than (25 ­ percentage points), multidistrict metro 5 percentage points over this period. cores (23  ­p ercentage points), and single-­ Against this backdrop and given the evi- district metro areas (22 ­ percentage points), dence on the spatial differences in productiv- whose populations were most middle class in ity documented in chapter 3, this chapter 2017 (33–40 ­ percent of the populations of probes deeper into the underlying drivers of these areas were middle class in 2017). By welfare disparities between and within places contrast, just over 10 ­percent of Indonesians in Indonesia. A key question that the chapter in nonmetro rural areas were in the middle addresses is the extent to which the differ- class in 2017. ences in productivity translate to differences At the same time, although overall inequal- in welfare and the extent to which these dif- ity trends at the national level are beginning ferences persist over time. In theory, if house- to moderate after a long spell of steady holds and workers could freely move increase, there is wide variation in the evolu- (or “sort”) across space or benefit from the tion of inequality by places and specific metro productivity in leading areas through other areas. Between 2001 and 2017, inequality mechanisms, welfare differences would be measured by the Gini coefficient of moderate and disappear over time, even 130  TIME TO ACT though productivity differences might persist. such as access to education, health, housing, In reality, several factors may constrain this sanitation, safety, and voice. However, mobility, including the country’s unique geog- because livability of urban areas is an explicit raphy as a vast, far-flung archipelago. The outcome that this report analyzes, gaps in the chapter thus also assesses the extent to which access to many of these nonmonetary mea- Indonesian workers and households move sures are presented in chapter 2 and discussed between places to capitalize on opportunities further in chapter 7. Some of these gaps will outside their home regions and districts. be discussed here but only in the context of Finally, the chapter analyzes the implications how much they contribute to differences in of mobility within cities and other forces of monetary well-being. urbanization on the sharing of economic ben- efits within urban places. This chapter is closely linked to chapter 3 Determinants of between- in two respects. First, the chapter’s main anal- ysis of the drivers of between-place welfare place inequality differences uses a conceptually similar frame- This section looks at the spatial variation in work to the one used in chapter 3. Both chap- well-being and inequalities across places in ters are concerned with unpacking the roles Indonesia. Urbanization is spatially inclusive played by characteristics of places versus the when the benefits of urbanization spill over to characteristics of the people who reside in help lift the living standards and well-being of those places, so the underlying methods used people living in other parts of the country. have similar flavors. One important differ- This process can happen through several ence is that, because chapter 3 is concerned channels. Rural households can migrate to with productivity differences, it operates with urban areas, where they can access more pro- nominal wages at the level of individual ductive jobs and higher earnings. At the same workers. This chapter is concerned with wel- time, such migration can also boost wages fare differences and as such will work with and incomes in rural areas by reducing labor real per capita expenditures at the level of the supply in these areas, leading to tighter local household.3 The method used here can essen- labor markets. Households can send out tially be regarded as the welfare analogue of migrant workers to more productive urban the productive-places (that is, “place-based centers and benefit not only from remittances advantages”) versus productive-workers (that sent home by these workers but also from is, “sorting”) analysis in chapter 3. Second, ideas and experiences. Agricultural house- the analysis of within-place inequality pre- holds in rural areas can become more inte- sented here reinforces the evidence on human grated into urban value chains and seize capital externalities in chapter 3 and builds opportunities that may arise in high-value on it to explore how externalities accrue to agriculture (such as horticulture, dairy, and workers at different skill levels, thereby con- livestock) and begin to transition out of sub- tributing to within-place inequality. sistence or low-value agriculture (such as Another clarification that is necessary rice). Finally, as rural areas themselves begin before proceeding is the definition of welfare to urbanize and take on urban characteristics, as it is used in this chapter and a brief discus- the labor markets in these areas are likely to sion of notions it does and does not include. transform, generating more opportunities What do we mean by welfare here? For the outside the farm, which can create greater most part, the analysis of inequalities in this opportunities for farm households. chapter will be done using real household per However, the extent to which these spill- capita expenditures as the main measure of overs can happen depends crucially on how well-being. This monetary measure is admit- well the more productive urban places are tedly limited: one may argue that a broader connected and integrated with the surround- definition of welfare and well-being would ing metro and nonmetro rural areas. also include other nonmonetary dimensions Geographic, institutional, and cultural D ri v ers of U rba n a n d S patial I n clusio n   131 barriers to migration may hinder mobility of What is interesting, though, is that, just as dis- workers and households, and the lack of con- tricts outside Jawa-Bali have relatively low nective infrastructure may impede possible poverty incidence and have made decent prog- backward links. Collectively, these mobility ress in reducing poverty and vulnerability, dis- barriers may contribute to the prosperity gen- tricts in Jawa-Bali are afflicted with chronic erated in urban areas being locked in place— poverty (box 4.1). Indeed, this diversity of out- benefitting only those who live in these areas comes within places, irrespective of whether but not doing much for others who are unable places are defined as island-regions, provinces, to connect to, participate in, and contribute to or even districts is one of the key features of this growth. welfare disparities in Indonesia.4 Spatial patterns of poverty and Welfare differences are large and persistent vulnerability in Indonesia mirror the wide across island-regions and the portfolio of productivity differences across the country places Despite the success in reducing overall poverty In addition to poverty and vulnerability, and vulnerability at the national level, another more general way of looking at differ- Indonesia today continues to be characterized ences in welfare is to consider the overall by sizable spatial variation in poverty living standards or welfare levels as measured ­ and ­vulnerability. At the district level, poverty by per capita household consumption. 5 and vulnerability rates vary from more than Although appealing and intuitive, comparing 90 percent in places like Mamberamo Tengah welfare across space can be tricky. Prices and and Nduga of Papua to less than 25 percent in their implications for cost-of-living differences some of the less poor places in Jawa and Bali. are important considerations. Chapter 3 fol- The broad geographic pattern mirrors the lows the urban economics literature in using results on spatial differences in productivity nominal wages as the relevant metric for pro- across space in Indonesia, but it is also note- ductivity; however, inferences for welfare and worthy that some of the lowest poverty rate living standards drawn on unadjusted nomi- districts in the country are actually outside nal consumption or income can be misleading. Jawa. The fact that some natural resource– For example, spending 10 million rupiah (Rp) dependent districts (especially in Papua) sit as does not yield the same welfare in Jakarta as islands of prosperity in a neighborhood of dis- in rural areas of Papua. Unlike for productiv- tricts with notably higher poverty and vulner- ity, it is important to take this cost-of-living ability rates suggests that the benefits of growth difference into account when comparing wel- in these areas is accruing to a few (map 4.1). fare differences between the two. Comparing the poverty and vulnerability Indonesia does not have an official spatial map of 2016 with the corresponding map for price index that captures differences in the 2010 reveals interesting patterns. First, the spa- cost of living across regions. As in many tial variation in the incidence of poverty and countries, the official measure of inequality is vulnerability rates is broadly similar in both based on the Gini coefficient, calculated using years. Second, the country-level decline in pov- nominal consumption without adjusting for erty and vulnerability rates is somewhat mir- cost-of-living differences across places. rored in the decline at the district level for most However, the analysis here uses the ratio of districts. Third, however, some districts, most the poverty line of a given province to the notably in Papua but to some extent also in poverty line of Jakarta core (that is, DKI Jawa and Nusa Tenggara, show relatively little [Daerah Khusus Ibukota] Jakarta) as a spatial progress over time. These concentrated pockets price deflator.6 The deflator is used to spa- of chronic poverty, which might ultimately tially adjust all the nominal consumption require special targeted place-based policies to series to generate welfare ratios—ratios of address (see ­chapter 8), are important frontiers household expenditure to the poverty line in for Indonesia’s ongoing efforts to fight poverty. the region, expressed in Jakarta core rupiah. 132  TIME TO ACT MAP 4.1  Poverty and vulnerability rates in Indonesia, by district Source: Smeru Institute and World Bank, using data from the 2010 and 2016 rounds of Indonesia’s National Socio-Economic Survey (SUSENAS). Note: Households with per capita household consumption below Indonesia’s official poverty line are considered poor. Households that are not poor but consume below 1.5 times the official poverty line are regarded as vulnerable. The maps show rates of poverty and vulnerability by districts. For residents of Jakarta core, the welfare Even after accounting for cost-of-living dif- ratios will equal household expenditures; for ferences, welfare varies quite substantially residents of a region where the poverty line is across space in Indonesia, according to half that of Jakarta core’s, the welfare ratio 2015 data. All the island-regions, including the will be double their household expenditure.7 rest of Jawa-Bali, have lower average welfare D ri v ers of U rba n a n d S patial I n clusio n   133 BOX 4.1  Chronic poverty: A profile of those left behind A large literature offers several ways of concep- in agriculture, have household heads that tualizing and defining chronic poverty. One defi- are older and less educated than the rest of nition that lends itself easily to the analysis of the population, and have higher dependency economic mobility is “poverty that persists for ratios. In terms of places, 60  ­ percent of the many years or a life course that may be trans- chronically poor in Indonesia live in nonmetro mitted across generations” (see, for example, rural areas, whereas 31  ­ percent live in metro Hulme and Shepherd 2003). By this definition, percent) and their urban peripheries cores (14  ­ chronic poverty can be understood as poverty percent). Although the higher concentra- (17  ­ recurring over the lifetime of an individual or a tion of chronic poverty in nonmetro rural areas household and often across generations. highlights the continuing gaps in economic The Indonesia Family Life Survey (IFLS), prospects across Indonesia’s portfolio of places, a long-running panel survey of households the high incidence of chronicity even within from 1993 to 2014, reveals that households in metro areas possibly underscores a less appreci- Indonesia have been able to achieve substantial ated fact about the extent of existing disparities upward economic mobility. More than half of within places (figure B4.1.1). Indonesian households that were poor in 1993 The IFLS was designed to be representative had escaped poverty and vulnerability by 2014. of 13 of Indonesia’s 26 provinces in 1993. It has Entry into the middle class was, however, more also been noted that the provinces not covered in restricted—of all those who were below the the survey are more likely to be the poorer prov- middle class in 1993, only 17.5  ­ percent had inces (see Strauss, Witoelar, and Sikoki 2016). succeeded in entering the middle class by 2014 For that reason, some caution is warranted in (Setiawan, Tiwari, and Rizal 2018). interpreting these results to extrapolate to the The same dataset, however, also reveals that country as a whole. However, within the IFLS percent of households in the original sample 4 ­ sample, it is striking that 10 ­percent of chroni- were chronically poor; they started out poor cally poor households live in metro Jakarta, and remained poor throughout the two decades whereas more than 50 ­ percent of the chronically that they were tracked. These left-behind house- poor are in the rest of Jawa. Sumatera accounts holds have certain peculiar characteristics: they for about 15 ­ percent, and the other regions for are significantly more likely to be concentrated the remaining 20 ­ percent. FIGURE B4.1.1  Where do Indonesia’s chronically poor live? a. Portfolio of places b. Island-regions Urban periphery Sumatera B&NT Metro core Nonmetro rural Nonmetro urban Rest of Jawa Jakarta KMP Source: Setiawan, Tiwari, and Rizal 2018, using data from various rounds of Indonesia Family Life Survey (IFLS). Note: B&NT refers to Bali and Nusa Tenggara; KMP refers to Kalimantan, Maluku, and Papua. The sizes of the boxes are proportional to the share of chronically poor living in each type of place. In panel a, places are defined using the methodology described in box 1.5, chapter 1. 134  TIME TO ACT than Jakarta core (figure 4.1, panel a). Papua FIGURE 4.1  Even after accounting for cost-of-living and Maluku and Nusa Tenggara are the differences, welfare differences across space vary in farthest behind with welfare levels 50 and Indonesia 54 ­percent, respectively, below that of Jakarta core on average, while Kalimantan and a. Island-regions, relative to Jakarta’s multidistrict metro core, 1993 and 2015 Sulawesi are the closest with welfare levels 0 23 and 25 ­ percent, respectively, below that of –0.1 Jakarta core. Across the portfolio of places, multidistrict metro cores (excluding Jakarta) –0.2 Welfare ratio have higher average welfare than Jakarta core, –0.25 –0.23 –0.26 –0.25 possibly reflecting stronger congestion forces in –0.3 –0.33 Jakarta core (figure 4.1, panel b). Urban periph- –0.34 –0.4 –0.37 eries of multidistrict metro areas trail Jakarta –0.43 –0.41 core marginally, but nonmetro rural areas, rural –0.5 –0.47 –0.50 peripheries of multidistrict metro areas, and –0.54 other nonmetro urban areas are well behind. –0.6 Jawa-Bali Kalimantan Maluku Papua Sulawesi Sumatera Comparing these results with those based on and Nusa 1993 data yields several interesting observations. Tenggara Across island-regions, the rest of Jawa-Bali and Kalimantan closed much of the welfare gap b. Portfolio of places, relative to Jakarta’s multidistrict metro core, 1993 and 2015 with Jakarta core, whereas Sulawesi’s gap 0.2 0.15 barely changed and Maluku, Nusa Tenggara, 0.1 Sumatera, and Papua fell further behind (figure 0 4.1, panel a). Across the portfolio of places, all Welfare ratio predominantly urban places—multidistrict –0.1 –0.07 –0.07 –0.12 metro cores, urban p ­ eripheries of multidistrict –0.2 metro areas, single-district metro areas, and –0.21 –0.20 nonmetro urban areas—made gains, whereas –0.3 –0.27 predominantly rural areas—rural peripheries of –0.4 –0.35 –0.39 –0.38 multidistrict metro areas and nonmetro rural –0.40 –0.43 areas—have continued to lag far behind Jakarta –0.5 Other Urban Rural Single- Nonmetro Nonmetro core. metro core periphery periphery district urban rural metro The characteristics of places explain more 1993 2015 of the spatial variation in welfare than do the characteristics of their residents Source: Tiwari and Shidiq 2018, using data from the 1993 and 2015 rounds of Indonesia’s National Socio- Economic Survey (SUSENAS). What explains these gaps in welfare, and why Note: The welfare ratio is the ratio of household expenditure to the contemporaneous poverty line in have these gaps persisted? This section takes the province of residence expressed in Jakarta core rupiah. The bars represent the percentage by which the framework used in chapter 3 (figure 3.2) average household welfare in each of the places varies from the average welfare in Daerah Khusus Ibukota Jakarta. In panel b, places are defined using the methodology described in box 1.5, chapter 1. to distinguish between characteristics of places and the characteristics of people as drivers of productivity and applies it to welfare.8 The with identical attributes would have the same attempt is to distinguish the relative impor- economic prospects everywhere. Any differ- tance of two distinct, but potentially self-­ ences in the return to those endowments reinforcing, views. One view is that an area is across space would be equalized because peo- poor because it has a higher concentration of ple would freely move to capitalize on initial households or people with low human capital differences in earnings premiums, increasing or portable endowments. Elevating those supply in places with high returns and decreas- endowments in lagging regions would there- ing supply in places with low returns.9 fore help welfare catch up. An underlying The other view is that an area is poor not assumption of this view is that households because of the compositional deficiencies in D ri v ers of U rba n a n d S patial I n clusio n   135 endowments but because of the low underly- structure, level of education of the household ing productivity of the place itself, which head and spouse, and the like) is identical yields a low return to endowments. Areas across all districts. The second is the returns with better local public goods—such as profile, which assumes that the endowment is ­ i nfrastructure and other basic services identical across districts but that the return to (­electricity, water, and sanitation)—tend to be the endowment varies. The endowment pro- more productive and thus provide higher file lets differences in the endowment drive all returns. According to this view, households the differences in welfare, whereas the returns with the same endowment can have different profile lets the differences in the return do so. welfare profiles, depending on which part of Analyzing how each of these synthetic pro- the country they live in. files relates to the actual welfare profile can The policy emphasis of these two views help make inferences about the relative also differs. The “poor places” view suggests strengths of the underlying mechanisms. enhancing poorer regions’ connectivity by Three key results emerge: investing in physical infrastructure and other basic services; the “poor people” view sug- •  First, the returns profile has a stronger gests investing in stronger education and correlation with actual welfare distribu- health services to enhance individual charac- t ion t h a n t he endow ment s prof i le teristics, including through direct assistance to fi gure 4.2, panel a). (­ families in the form of conditional cash trans- •  Second, the returns profile also explains a fers (such as Indonesia’s Program Keluarga larger share of the overall variance of the Harapan). Both views equally stress the need actual welfare profile (54 ­ p ercent) than for strong institutions and local governance. the endowments profile does (19 ­ percent) To determine whether places or people (figure 4.2, panel b). play a larger role in Indonesia, two synthetic •• Third, the preceding two statements are as welfare profiles were created. The first is the true in 2015 as they were in 1993—that endowments profile, which assumes that the is, the stronger relationship between the welfare return to all household-specific demo- returns profile and the actual welfare pro- graphic characteristics (age, dependency file appears to be consistent across time. FIGURE 4.2  The returns profile has a stronger correlation with, and explains a larger variation of, the actual welfare distribution than the endowments profile does a. Correlation between the actual b. Share of variance of the actual welfare distribution and the two synthetic profiles explained by the variance of the two synthetic profiles 1.0 0.90 0.7 0.63 0.83 0.6 0.54 Correlation coefficient 0.8 0.72 Share of variance 0.61 0.5 0.6 0.4 0.4 0.3 0.22 0.19 0.2 0.2 0.1 0 0 1993 2015 1993 2015 Actual and endowments profile Endowments profile Actual and returns profile Returns profile Source: Tiwari and Shidiq 2018, using data from the 1993 and 2015 rounds of Indonesia’s National Socio-Economic Survey (SUSENAS). Note: For the endowments profile, district-level endowments are paired with the national average return to each endowment; for the returns profile, district- level returns to household characteristics are paired with the national average of these characteristics. The methodology is a variant of the Oaxaca–Blinder decomposition and is also used in Ravallion and Wodon 1999 and Skoufias and Olivieri 2013. See annex 4A for further details. 136  TIME TO ACT What this implies is that differences TABLE 4.1  Proximate district-level in the return to characteristics—as opposed correlates of the welfare return to to the overall levels of the characteristics secondary education t hemselves—are, and continue to remain, ­ the stronger driver of welfare differences in Characteristic Sign and significance Indonesia. In other words, geography mat- Share of villages with road + (**) ters: poor areas are poor not just because access (percent) they have a higher concentration of house- holds with characteristics that are condu- Share of villages with a primary + (*) and secondary school (percent) cive to higher poverty, but perhaps more so because the same characteristics fetch a Market access + (***) lower return relative to other places in the Source: Tiwari and Shidiq 2018, using data from the 2015 round of country. Indonesia’s National Socio-Economic Survey (SUSENAS). The following question arises next: What Note: See annex 4A for a description of methodology, table 4A.1 for the full regression results, and annex 3A in chapter 3 for details on how the domestic are the characteristics of places that dictate market access variable was constructed. Levels of significance are the highest whether they have high and low returns to reported levels in table 4A.1. endowments? Again, following the concep- * Significant at the 5 ­percent level; ** significant at the 1 ­percent level; *** significant at the 0.1 ­percent level. tual framing in chapter 3, but continuing with the focus on welfare rather than pro- ductivity and with households rather than workers as the unit of observation, we can Characteristics of people also matter for variation in welfare exploit the richness of Indonesia’s house- hold survey data to answer this question. Although the returns profile explains a much Household-level welfare regressions are run higher proportion of the overall variance of the at the district level limiting attention to just actual welfare profile, it is noteworthy from fig- the household characteristics such as gender, ure 4.2, panel a, that the correlation between age, educational attainment of the house- the synthetic endowments profile and the hold head, and the overall demographic actual welfare profile is also strong. There are composition of the household. In the second several possible explanations for this correla- stage, the estimated “returns” to each of tion. First, as noted in chapter 2, although these characteristics are then regressed on poorer Indonesian districts have caught up district-level variables such as share of vil- with the rest of the country in providing basic lages with road access, share of households services, sizable gaps remain. Across the portfo- with access to safe drinking water, share of lio of places, urban areas generally have better households with electricity, share of house- access to education, health care, safe drinking holds with improved sanitation, share of water, and sanitation facilities. Because all these households with nonearth floor, share of vil- services are key inputs into the human capital lages with a health facility (puskesma), share of the population, the high correlation between of villages with a primary and secondary the endowment profile and the actual welfare school, population size, population density, profile reflects the gaps. For example, stunting and market access (calculated using digi- rates in multidistrict metro cores were tized maps of road infrastructure; see 29 ­percent relative to 42 ­ percent in nonmetro annex 3A in chapter 3). rural areas (­figure 4.3, panel a). The main district-level characteristics asso- Another possible explanation is that the ciated with the welfare return to secondary quality of services varies widely across education are domestic market access, road Indonesia’s portfolio of places. The amount of penetration, and school infrastructure actual learning attained in a reasonably good (table 4.1; see table 4A.1 in annex 4A for the public school in Jakarta likely differs greatly full results). The evidence on returns to other from the amount attained in a school in rural household characteristics is inconclusive and Papua. Data on science scores in the presented in Tiwari and Shidiq (2018). Programme for International Student D ri v ers of U rba n a n d S patial I n clusio n   137 FIGURE 4.3  Metropolitan areas have lower rates of stunting and better learning outcomes on average a. Stunting rates b. Average PISA science scores, 2015 45 450 42.0 440 40 38.2 430 35 33.8 410 Percent 401 Score 31.4 29.3 30.1 389 30 390 382 25 370 361 20 350 Metro Urban Rural Single- Nonmetro Nonmetro Village Small Town City Large core periphery periphery district urban rural town city metro Sources: Panel a calculations based on survey data from Riskesdas 2013; panel b calculations based on 2015 data from the Programme for International Student Assessment (PISA) (http:// www.oecd.org/pisa/). Note: In panel a, places are defined using the methodology described in box 1.5, chapter 1. In panel b, PISA instead classifies places in the following manner: villages or hamlets have fewer than 3,000 people; small towns have 3,000–15,000 people; towns have 15,000–100,000 people; cities have 100,000–1,000,000 people; and large cities have more than 1,000,000 people. This taxonomy cannot be mapped to the taxonomy of places used in this report. Assessment (PISA), for example, show that stu- on migration complicate estimating the long- dents in more populous areas in Indonesia have term time trend, and it is unclear whether all markedly higher overall learning levels than types of migrants are captured in the surveys students in similar age and grade groups in (box 4.2). Nevertheless, available evidence other parts of the country. The difference in test suggests that migration is an increasingly scores between what PISA considers a large city common phenomenon in Indonesia.10 The (places larger than population of 1 million) and proportion of Indonesians who are migrants a village (a place with fewer than 3,000 people) has been slowly increasing over time, reaching is a gaping 79 points. With every 30 points in 21 ­percent in 2015. The rate of provincial the PISA test scale corresponding roughly to migration, considering all categories of one year of education, this difference implies migrants (recent, frequent, long term, and almost a three-year gap in years of education returned) increased from 9.8 ­ percent of the between students in the same grade in cities population in 1990 to 11.9 ­ percent in 2010. and villages (see spotlight 2, “Urbanization for percent District migration increased from 18.2 ­ Human Capital,” on the spatial disparities in in 2000 to 21.0 ­ percent by 2015. Long-term the human capital index in Indonesia). migrants make up the majority, representing 70–80  ­ p ercent of all migrants. However, about 15 ­ percent of migrants have migrated Migration bears returns, but not in the last five years for the first time, many Indonesians appear to seize 5–7 ­percent are frequent migrants and have this opportunity—suggesting hidden moved again in the last five years, and the constraints to moving smallest number (2–6 ­ percent) are migrants Given large earnings premiums in metropoli- who have returned to their district of birth. tan areas (documented in figure 2.8, panel b, Most migrants stay within the same island- and figure 3.3), and the welfare gaps across region. Of those who leave their region, most Indonesia’s portfolio of places, a key question are from Jawa. Nationally, 75  ­ p ercent of that arises is the extent to which Indonesians migrants stay within their island-region, with move across the places to capitalize on oppor- that share reaching 87  ­ p ercent for Jawa. tunities outside their region of birth. However, all non-Jawa islands receive signifi- Inconsistent questions in household surveys cant migration from Jawa, particularly 138  TIME TO ACT BOX 4.2  The various types of migrants in Indonesia: Are they all captured in the data? Qualitative work was undertaken for this report Moreover, the cell phone estimates may be in Jakarta, Makassar, Medan, and Tangerang to low for three reasons. First, cell phone data were deepen the understanding of the nature of migra- available only for a four-month period (August tion in Indonesia (see Wai-Poi and Sacks 2018; to November 2017) and so have been annualized, World Bank 2018). On the basis of a combination but how these four months reflect general sea- of the nature of their work, motivations to migrate, sonal patterns is unknown. The time frame cov- and length of residence in the urban areas among ers the shorter dry season rice crop’s harvest and those who migrate, seven distinct categories of planting and the wet season rice crop’s planting, migrants can be identified.a The following four which may suggest that seasonal effects are pres- types of migrants work largely within the infor- ent in limiting the supply of labor. This would mal sector and include casual workers s ­ egmented indicate that the cell phone–based estimates are commuter (stay s as (1) long stay (2) ­ ­ everal weeks), an underestimate when annualized. Second, a and (3) “­follow the money” migrants, who move common phenomenon in Indonesia is SIM (sub- frequently for work. Also in the informal sector are scriber identity module) card churn. Subscribers (4) self-employed migrants, who work for them- abandon SIM cards (and their cell phone num- selves, typically as entrepreneurs. Within the for- bers) for new ones because of promotional offers, mal sector are (5) migrants who are wage workers so these migrants would not be captured in the with employment contracts and (6) migrants pur- data. Third, further potential coverage errors suing higher education with the aim of getting sal- and biases are introduced by the possibility that aried work. Finally, there are migrants who move some migrants do not own cell phones, so they to the urban area for reasons other than work, would never appear in the mobile network data particularly to (7) accompany family members. in any form. The result of these coverage errors These migrants are typically either not working or and biases is that the type of migration captured earning very little money. by the mobile network data analysis, however Some of these migrants are short term, com- defined, is underestimated. muting, or seasonal and thus may not stay in urban locations long enough to be covered by a. The qualitative work used a “Reality Check Approach” methodology. This household surveys like Indonesia’s National approach entails immersive research by trained researchers who live with people in their own homes and share in their everyday lives. It is based on Socio-Economic Survey (SUSENAS). It is there- principles of ethnography but is narrower in focus, and the short time for fore possible that not all types of migrants are immersions distinguishes it from ethnography. The qualitative analysis for this covered in national surveys. Data from Pulse Lab work was undertaken in 10 locations—5 urban and 5 rural. The urban locations were identified within the cities of Jakarta, Makassar, Medan, and Tangerang, Jakarta’s cell phone network to identify the and included both slum and nonslum areas, core and periphery areas, and number and origins of the migrants on the basis areas with many long-term and short-term migrants. The rural areas were of anonymized mobile network activity records identified—after the immersion in the urban areas—as the origination areas of the migrants researchers met in the 5 urban areas. Further details on the confirm that the number of migrants recorded in methodology, study design, and findings are available in the background report surveys may indeed be underestimated. World Bank (2018). Kalimantan (45 ­ percent of all migrants) and outward migration is twice as high at Sumatera (31 ­percent). Kalimantan, Maluku, 4.6 ­ million, making Jawa by far the largest and Papua also receive significant outside net sending region. The main receivers of this migration from places other than Jawa. Jawa outward migration are Sumatera and sends and receives many migrants but is a net Kalimantan (World Bank 2018). sender, whereas Kalimantan and Sumatera are Most recent migrants come from rural the largest net recipients. Although inward areas, and most migrants go to urban areas— migration to Jawa is nearly as high as it is to but urban–urban migration is also very preva- Sumatera and Kalimantan, with 2.3 million lent. Fifty-five percent of recent migrants coming in over the five years to 2015, come from a rural area, most of them from D ri v ers of U rba n a n d S patial I n clusio n   139 nonmetro rural areas. Seventy-one percent of middle class in the multidistrict metro cores migrants go to an urban area, spread over the seems to have come before the 2000s; the esca- metro core (20 ­ percent), the urban periphery lator does not seem to have worked very well (23  ­ percent), and nonmetro urban for those who made the move after the 2000s. (24 ­percent) areas, with a small number going In contrast, prospects of middle class entry for to single-district metro areas (4  ­ p ercent). newer entrants into the urban peripheries have Urban–urban migration is the most common remained robustly high over time, suggesting type of migration (2.9 million), followed by that the urban peripheries have retained the rural–urban migration (2.3 million). However, advantages of proximity to the prosperity of rural–rural (1.2 million) and urban–rural the metro cores while avoiding their costs (785,000) migration have also been quite (Setiawan, Tiwari, and  Rizal 2018).11 This common in the last five years. result is consistent with the sources of urban Moving to urban areas, particularly multi- population growth in c ­ hapter 1: at the country district metro cores, has been a robust pathway level, migration has made a somewhat limited for upward mobility for those who made the contribution, but for specific regions, and espe- move. Analysis suggests that, since the mid- cially in urban peripheries of multidistrict met- 1990s, the odds of escaping poverty were twice ros, where the escalator out of poverty and as large for households that were able to move into the middle class has remained strong, to a core than for those that moved to other migration made large contributions. nonmetro rural areas. Similarly, the chances of This apparent erosion of the ability of the entering the middle class were also better for metro cores to lift Indonesians into the middle those who moved to metro cores and their class indicates a variety of congestion forces urban peripheries than for those who made at play. One such force is related to housing rural–rural moves (figure 4.4). Interestingly, affordability. Data from an urban perceptions most of the success in moving people to the survey commissioned for this report show that urban residents, including migrants and FIGURE 4.4  Multidistrict metro cores provide nonmigrants, tend to prioritize proximity to a quicker pathway out of poverty and into the family members when they choose where to middle class live (Wai-Poi and Sacks 2018). After that con- cern, the two other considerations that come 2.5 up most frequently are distance to place of work and housing affordability.12 Roughly 2.0 ** one in five core residents chose location to be *** close to work. Given the high cost of housing Odds ratio 1.5 * in metro cores (see chapters 2 and 7), how- 1.0 ever, most migrants get priced out of the mar- ket and are forced to take up residence in the 0.5 surrounding urban periphery (see chapter 1). On the whole, the persistent welfare dis- 0 Metro Urban Rural Single- Nonmetro parities across space in Indonesia, together core periphery periphery district urban with the fact that migrants who move gener- metro ally are able to climb up the economic ladder Exiting poverty Entering middle class to escape poverty even if not sufficiently enough to enter the middle class, raise ques- Source: Setiawan, Tiwari, and Rizal 2018. tions about why more Indonesians do not Note: The bars represent odds ratios (relative to nonmetro rural) estimated using conditional logit take this opportunity. Part of the reason could regressions of moving out of poverty and entering the middle class for places as defined using the methodology described in box 1.5, chapter 1. Controls include changes in dependency ratio and be the many observed and unobserved con- household size; head of household education; number of migrant workers in the household; number straints to migration. Potential movers may of household workers in different sectors, by employment type; income by source; public and private be uncertain about migration’s benefits or transfers; and asset value by type. Bars with solid colors represent estimated odds ratios that are significant at the level shown by the asterisks; shaded bars represent estimated odds ratios that are wary of risking unemployment or toiling in a statistically insignificant. low-paying job. Qualitative analyses for this * Significant at the 10 ­percent level; ** significant at the 5 ­percent level; *** significant at the 1 ­percent level. 140  TIME TO ACT report suggest that migrants to cities often not have an official address or, at that time, underestimate the extent to which the cost of a Rukun Tetangga (RT, neighborhood associ- living in cities erodes their expected earnings. ation) to help them obtain the KTP (Wai-Poi (World Bank 2018) There may be consider- and Sacks 2018; World Bank 2018).15 ations related to transportation costs or strong cultural preferences to maintain social ties in their origins. Migrants tend to prefer Determinants of within- destinations closer to their origins, often place inequality choosing the nearest city or provincial capital Next, we turn to the examination of drivers of over larger agglomerations farther away.13 within-place welfare inequality. For urbaniza- Another key constraint migrants in tion to be inclusive, the gains from urbaniza- Indonesia may face, at least notionally, is the tion must be shared not only across the difficulty associated with obtaining identity and country’s diverse portfolio of places but also other administrative documents. Having an within these places. In other words, the evolv- identity card such as a Kartu Tanda Penduduk ing process of urbanization must enable all city (KTP) or a Kartu Keluarga is important for dwellers to benefit from the opportunities cre- access to various public services, such as obtain- ated in the urban and urbanizing areas. This ing a driving license and receiving cash trans- section looks at some of the forces relevant to fers. Evidence on this is conflicting, however. In the within-place inclusivity of urbanization in the four cities in the Rural Urban Migration in Indonesia. As before, the gaps in basic services China and Indonesia survey (Makassar, Medan, and amenities (access to clean water, sanita- Samarinda, and Tangerang), all migrants who tion, and pollution-free air) that determine liv- lived outside slums obtained a local KTP within ability are discussed in other parts of the report three months of arrival, suggesting that obtain- (chapter 2). Here, the focus is on the extent to ing this card may not be that much of an issue which urbanization is inclusive in the sense of (Wai-Poi and Sacks 2018). sharing the economic benefits of urbanization. More recent evidence from qualitative sur- The relevant metrics are per capita household veys conducted for this report suggests that expenditures and, building on chapter 3, wages many slum dwellers in fact do not have KTPs, for workers of different skills. either because they do not see the value or because they have no official address. Long- stay migrants in the Jakarta urban slum Benefits of urbanization have not been reported they had updated their cards when shared evenly within places, and that they first came to the city as proof of resi- inequity is the key driver of the increase in within-place inequality dence but then had not bothered to update them when they moved within the city. Some Part of the reason for the increase in within- others had not bothered to update their vil- place inequality is that not everyone has ben- lage KTP because they saw no value in having efitted equally from opportunities in urban a city KTP and found the administrative pro- areas (see chapter 2). In all areas of the portfo- cesses “a hassle.” Many people reported using lio of places, the bottom 40 ­ percent of the dis- nembak (speed money/bribes) to circumvent tribution of per capita household consumption administrative requirements for KTP and saw much slower growth in their welfare lev- documents like the Surah Pindah.14 Although els between 2001 and 2017 than the overall birth certificates are important to migrants, population did. The cores and urban and rural many rural Indonesians have not obtained peripheries of multidistrict metro areas saw birth certificates because of the administrative the largest increase in overall living standards, hurdles involved. When staying in the Jakarta averaging 3 ­percent a year or more (figure 4.5, urban slum in 2015, the qualitative survey panel a); however, these were also the areas team found that only about 10 ­ percent of the with the widest gap between the growth rate residents they spoke with had a city KTP. for the overall population and that for the bot- Most did not obtain KTPs because they did tom 40  ­ percent. In other words, although D ri v ers of U rba n a n d S patial I n clusio n   141 FIGURE 4.5  Large gaps exist between average growth of per capita consumption and growth percent for the bottom 40 ­ a. Portfolio of places b. Major multidistrict metro areas Jakarta 3.4 3.1 1.6 Metro core 1.1 Bandung 3.2 1.3 3.9 Surabaya 4.4 Urban periphery 3.0 1.9 1.2 Medan 0.1 3.1 Surakarta 2.8 Rural periphery 1.0 1.2 Semarang 3.1 1.7 2.4 3.6 Single-district metro Malang 1.1 2.4 Makassar 4.4 2.0 2.9 Nonmetro urban 1.3 Sukabumi 3.5 1.7 Denpasar 3.6 2.6 2.2 Nonmetro rural 3.0 1.1 Yogyakarta 1.4 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 Annualized growth of per capita Annualized growth of per capita consumption, 2001–17 (%) consumption, 2001–17 (%) Overall population Bottom 40 percent Source: Calculations based on data from the 2001 and 2017 rounds of Indonesia’s National Socio-Economic Survey (SUSENAS). Note: In panel a, places are defined using the methodology described in box 1.5, chapter 1. overall living standards (at least from a mone- multidistrict metro areas in Indonesia such as tary perspective) improved most in multidis- Bandung, Jakarta, Makassar, Surabaya, and trict metro areas, the gap between the living Yogyakarta are among the most unequal standards of the wealthiest and the poorest places in the country. This phenomenon is not groups also increased the most.16 unique to Indonesia: the relationship is simi- The pattern is similar for larger multidis- lar for the United States and for countries in trict metro areas. In all 11 multidistrict metro Latin America and the Caribbean (see, for areas analyzed, the growth rate of consump- example Ferreyra 2018). Further, places that tion for the bottom 40 ­ percent consistently have become more populous—and thus more lags the growth rate for the overall ­population densely populated—have also seen larger (figure 4.5, panel b). In Bandung, Jakarta, changes in inequality (table 4.2). In other Makassar, Medan, Sukabumi, Surakarta, and words, within places, inequality changes are Yogyakarta, the difference is more than a fac- positively correlated with population changes. tor of two. Malang and Surabaya stand out as What drives this pattern? One possible cities where the bottom 40 percent have done answer is differences in aggregate human cap- relatively well, both in absolute terms and in ital within areas and the opportunities they relation to the mean. create for workers in different skill groups. As  cities grow and become denser, they acquire higher stocks of aggregate human Although everyone benefits from living capital, either through increases in the attain- and working in high human capital areas, ment of younger cohorts of residents or the benefits of agglomeration accrue much through the skill-selective migration of indi- more to high-skilled workers than to low- viduals from other places. Higher-density skilled workers places in Indonesia are also, on average, Highly populated places generally have high places with a higher stock of aggregate human inequality (figure 4.6). Accordingly, the large capital (­figure 4.7). In other words, districts 142  TIME TO ACT FIGURE 4.6  Highly populated places generally have high inequality 0.45 Yogyakarta Bandung Makassar 0.40 Malang Probolinggo Jakarta Jambi Blitar Magelang Surabaya Palembang Pontianak Semarang Salatiga Pasuruan Surakarta Banjarmasin Denpasar Bandar Lampung Pekanbaru Gini coefficient 0.35 Banda Aceh Sukabumi Mojokerto Medan Balikpapan Padang Samarinda 0.30 Bukittinggi 0.25 y = 0.0063x + 0.25 R2 = 0.0193 0.20 10 11 12 13 14 15 16 17 18 Population, 2017 (log) Metro Nonmetro Source: Calculations based on data from the 2017 round of Indonesia’s National Socio-Economic Survey (SUSENAS). Note: The unit of analysis is districts, except for multidistrict metro areas, for which both the Gini coefficient and the population are calculated over all districts in the metro area. TABLE 4.2  Densely populated places in cities. It occurs not only through the direct also generally have high inequality effect of skills on productivity but also through indirect spillover effects between Dependent variable: workers as they learn from each other Gini coefficient (district) through, for example, observation and mim- icking (see chapter 3). How the benefits of Pooled cross-section Fixed effects Variable (levels) (changes) higher aggregate human capital accrue to workers of different skill levels, however, Population (log) 0.006*** 0.182*** depends on the strength of the overall human (0.001) (0.006) capital externalities and complementarities Constant 0.219*** –2.111*** across skill categories. If complementarities (0.013) (0.078) are strong—that is, high- and low-skilled Observations 4,277 4,277 workers are imperfect substitutes for each other—low-skilled workers benefit at the R-squared 0.010 0.190 expense of more highly skilled workers. 17 Source: Calculations based on data from the 2001–15 rounds of Indonesia’s Conversely, if strong enough externalities National Socio-Economic Survey (SUSENAS). exist, all workers will benefit: high-skilled Note: Numbers in parentheses are robust standard errors. *** Significant at the 1 ­percent level. workers through greater cross-fertilization of ideas and learning across firms and workers, which spurs innovation, and low-skilled that become more populous (which can hap- workers through spillovers of those pen only through an increase in population, externalities. because area is fixed) also accumulate more Across most types of place, aggregate human capital (see table 4.2). human capital is correlated with productivity, A higher concentration of human capital in indicating the existence of positive social urban areas is a major source of productivity returns, over and above the private returns to D ri v ers of U rba n a n d S patial I n clusio n   143 FIGURE 4.7  Average years of schooling and population size are positively correlated across Indonesia 12.5 Average years of schooling for population age 15 and older Padang 11.5 Pekanbaru Banda Aceh Balikpapan Yogyakarta Palembang Medan Samarinda Jambi Bandar Lampung 10.5 Denpasar Surabaya Jakarta Makassar Banjarmasin Semarang 9.5 Bukittinggi Bandung Surakarta Pontianak Mojokerto Malang Salatiga 8.5 Blitar Magelang Pasuruan Sukabumi 7.5 Probolinggo y = 0.13x + 8.056 R2 = 0.0401 6.5 0 1 2 3 4 5 6 7 8 9 10 Population density, 2016 (log) Metro Nonmetro Source: Tiwari and Shidiq 2018. Note: The unit of analysis is districts, except for multidistrict metro areas, for which both average years of schooling and population density are calculated over all districts in the metro area. human capital (see chapter 3; Bosker, Park, the key finding relevant to the discussion here and Roberts 2018). Here we examine how is that agglomeration benefits accrue much these aggregate benefits accrue to workers of more to high-skilled workers than to low- different skill levels. 18 On the basis of a skilled ones, which, in turn, helps drive inequal- nationwide sample of workers in the IFLS, the ity within places. medium-skilled group sees a 6.3  ­ p ercent return to aggregate human capital (­ figure 4.8). In other words, a one-year increase in the Differences in educational attainment and average years of schooling in an area yields a lack of integration are key drivers of within- place inequality 6.3 ­percent return in earnings to medium- skilled workers in that area. Low-skilled This narrative is broadly supported by data workers also benefit: the return to aggregate on the drivers of within-place inequality. human capital for them is about 2.8 ­ percent. Inequality decomposition analyses show that, They benefit much less, however, than high- both within the portfolio of places and within skilled workers, whose return is 10.0 ­ percent. Indonesia’s major multidistrict metro areas, It also appears that most of the effects are differences in educational attainment account driven by Jawa-Bali, where the pattern of for as much as one-third of overall within- returns across skill groups is roughly the place inequality (figure 4.9). This is more than same.19 the share accounted for by such dimensions The findings that, on average, all types of as gender, occupation, and sector of workers benefit from living and working in employment. areas with high human capital conform with The inequality generated in the labor the evidence of agglomeration forces through ­ m arket is compounded by within-place human capital externalities presented ­ differences in education quality. Although chapter 3 using a different dataset. However, in ­ average educational attainment is higher in 144  TIME TO ACT FIGURE 4.8  High-skilled workers see cities—and there are signs that education qual- the largest return to aggregate human ity has improved over time—the learning gaps capital, followed by medium-skilled between children in different socioeconomic workers and low-skilled workers classes are wider in cities than in smaller towns 10.0 and rural areas. Science test scores on the PISA, Estimated earning premium 12 ** 10.3 a standardized international assessment, show 7.9 *** that, in cities, children from households in the 8 6.3 *** top 20  ­ p ercent of the socioeconomic 4.2 2.8 ** ­ distribution outperform those in the bottom 4 ** 20 ­percent by more than 80 points—­ equivalent to more than two and one-half full years of 0 schooling (figure 4.10).20 This finding suggests Full sample Jawa-Bali that, if productivity spillovers across skill cate- Low skilled Medium skilled High skilled gories remain weak, within-place inequality may be compounded as the current generation Source: Setiawan, Tiwari, and Rizal 2018. enters the labor market. This effect is likely to Note: Data are for a pooled sample of working-age individuals in all rounds of the Indonesia Family Life Survey (IFLS), 1993, 1997, 2000, 2007, and 2014. become stronger as the increasing use of vari- The bars represent coefficients on the average number of years of schooling ous forms of technology begins to fundamen- for each district, where this is measured using data from Indonesia’s National tally transform the nature of work. Newer Socio-Economic Survey (SUSENAS), interacted with the skill level of each individual (in three categories: high, low, and medium skilled) in augmented technologies, however, could also help make Mincer regressions for the entire country as well as separately for Jawa-Bali and urbanization more inclusive by helping close Jakarta-Bali. All specifications include individual characteristics such as age, distances and connect places across the age-squared, gender, marital status, hours of work, and dummy variables for portfolio of place areas, occupation, and sector of work. Unobserved individual Indonesian archipelago (box 4.3). heterogeneities are captured by individual-level fixed effects. The bars with solid In addition to disparities in levels, poor colors are statistically significant at the level indicated by the number of asterisks integration associated with poor connectivity (see note on significance, below). Medium skilled is used as the base category, so the significance on the returns to the high and low skilled refers to their difference within Indonesia’s metro areas could also from the medium skilled. See annexes 4B and 4C for full specification and results. partly explain why low-skilled workers bene- * Significant at the 10 ­percent level; ** significant at the 5 ­percent level; *** fit less from human capital externalities significant at the 1 ­percent level. in  ­c ities than high-skilled workers do. FIGURE 4.9  Both within the portfolio of places and within Indonesia’s major metropolitan areas, differences in educational attainment account for as much as one-third of overall within-place inequality a. Portfolio of places b. Major multidistrict metros 45 45 Share of within-place inequality (%) Share of within-place inequality (%) 40 40 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 2001 2003 2005 2007 2009 2011 2013 2015 2017 2001 2003 2005 2007 2009 2011 2013 2015 2017 Gender Education Occupation Sector Source: Tiwari and Shidiq 2018. Note: The shares are obtained by decomposing Theil’s L separately for each demographic characteristic within six spatial categories: multidistrict metro core, urban periphery, rural periphery, single-district metro, nonmetro urban, and nonmetro rural. Shares for each year do not need to sum to 100. D ri v ers of U rba n a n d S patial I n clusio n   145 FIGURE 4.10  Children from households in the For example, because of limited affordable top 20 percent of the socioeconomic distribution housing, especially in good locations, and outperform those in the bottom 20 percent most in inadequate transport systems that connect cities on the Programme for International Student people to each other and to work opportuni- Assessment’s science test ties, poorer people are forced to reside in informal settlements or in the outer peripher- 120 ies of cities and to take lower-quality jobs in Difference between top 20 percent 102.6 100 the vicinity of their neighborhoods as opposed and bottom 20 percent on PISA science test, 2015 85.8 to potentially better ones if labor markets 80 were better integrated. This can lead to iso- 67.0 60 lated enclaves that impede the wider sharing 40.4 of the benefits of urbanization (see chapter 7). 40 Poverty maps of a selection of Indonesia’s 29.6 20 multidistrict metro areas show that many of these areas are characterized by segmented 0 neighborhoods where pockets of poverty Village Small town Town City Large city exist alongside pockets of relative prosperity Source: Calculations based on 2015 data from the Programme for International Student Assessment (map 4.2). The central parts of Bandung, (PISA) (http://www​.oecd.org/pisa/). Jakarta, and Yogyakarta and the coastal areas Note: PISA classifies places in the following manner: villages or hamlets have fewer than 3,000 people; of Makassar and Surabaya clearly have the small towns have 3,000–15,000 people; towns have 15,000–100,000 people; cities have 100,000– 1,000,000 people; and large cities have more than 1,000,000 people. This taxonomy cannot be mapped lowest poverty rates, whereas the southern to the taxonomy of places used in this report. parts of Bandung, the southeastern and BOX 4.3  Technology, the changing nature of work, and urbanization Increasing use of technology in the produc- spatial transformation. First, automation and tion process, including through greater use greater use of technology can lower demand of automation-enabled technologies such as for manufacturing workers. Coupled with robotics and artificial intelligence, carries tre- the greater rate of mechanization in agricul- mendous promise and potential in the produc- ture, which may in turn accelerate the transi- tivity growth it can spark. Digital technologies tion out of agriculture, the use of automation are already challenging traditional production and technology may amplify the pressure to patterns, blurring firm boundaries, spurring create more jobs, especially in urban areas. new business models built on digital platforms, Three factors potentially mitigate this pres- and leading to a vibrant ecosystem of start- sure. First, Indonesia created 1.8 million jobs ups and global behemoths alike. Despite lower per year between 2010 and 2017, and projec- overall levels of Internet connectivity relative tions suggest that, in the short run at least, that to other neighboring countries, Indonesia also momentum can be sustained. Second, automa- has witnessed the growth of start-ups led by tion notwithstanding, aging workforces glob- e -commerce (Alphacart.com), transportation ­ ally, together with an increasingly prosperous (Go-Jek), and financial services sectors (Kartuku global middle class, may create enough scope and HaloMoney). for industrial jobs to grow in Indonesia. Third, In spite of its advantages, technology even though Indonesia is still a relatively young can also be a disruptive force that can pro- country, it will get over the demographic hump foundly change the very nature of work. For within the next 30 years, which may mean that a lower-middle-income country like Indonesia, it will require that fewer jobs be created. two aspects of this trend are relevant given The second aspect of the trend is prob- where the country stands in its structural and ably the one that poses a greater challenge Box continued on next page 146  TIME TO ACT BOX 4.3 Continued for  Indonesia. Rather than displacing jobs By contrast, newer forms of digital tech- altogether, technology is profoundly reshap- nology could also help bridge spatial divides ing the skills content of work. The demand for through a variety of channels. First, digital tech- advanced cognitive and sociobehavioral skills nology can reduce costs of accessing informa- together with more adaptable skill combina- tion and markets, which can enable workers, tions is rising—whereas the demand for less including potential migrants, to connect better advanced skills that are often applied to repeti- to opportunities in cities. Access to online mar- tive tasks is declining. In addition, an ability ketplaces and e-commerce platforms may help to work with some form of technology (at all producers in more remote locations connect to skill levels) is becoming a stronger discriminant buyers in other parts of the country, helping in the labor market for newer cohorts every- strengthen backward links and spillovers from where, including in Indonesia. If not adequately urbanization. Second, specific digital platforms addressed with urgency, the inequalities in and applications may be useful not only to close foundational cognitive and noncognitive abili- gaps but also to improve quality of basic services ties in the younger generation today are likely and better deliver social assistance, which would to amplify inequalities in income in the future also enable lagging places to catch up. (see spotlight 2 for more on human capital dif- ferences across Indonesia). Note: This box draws on findings from World Bank 2019. southwestern parts of Jakarta, the eastern for some neighborhoods. 21 Accessing the parts of Makassar, and the northern parts of three top-ranked senior high schools in each Surakarta have much higher poverty rates. In kota (city) takes 35  minutes and costs general, across most metro areas, pockets of Rp 3,500 for the median neighborhood but poverty appear to be concentrated largely on takes 107 minutes and costs Rp 16,000 for the outer periphery of the areas. some Jakarta residents (see map 4.3 panel b) When neighborhoods are segmented along (Lain 2018). income lines in this manner, equitable access When neighborhood-level data are aggre- to services and opportunities depends criti- gated to the kelurahan level, travel time to the cally on the efficiency of the within-city trans- top-ranked senior high schools is positively portation infrastructure. Analysis of time and correlated with kelurahan-level poverty rates costs associated with reaching health and (using the $3.10 a day—2011 constant inter- education facilities based on big data reveals national dollars—global poverty line). This that distance to certain facilities may be an finding is consistent with the idea that richer unsuitable indicator of access in settings with households are located in areas that offer bet- high congestion. The analysis also finds that ter access to good schools. At the same time, regular survey data may not capture the full there is little evidence that kelurahan-level pov- extent of spatial inequality and constraints to erty rates are correlated with travel time to a access in Indonesia’s cities. For example, data regional public hospital, training center, or from Google Maps and Trafi show that it government office (Lain 2018). However, this takes residents of an average Jakarta neigh- analysis covers only the core of Jakarta. If borhood (rukun warga) about 40 minutes to greater Jakarta is considered as a single metro- reach a regional public hospital using public politan unit, as is the case in most of this transportation but that, for some neighbor- report, these inequalities are likely to be even hoods, the travel time is almost 2 hours more pronounced because the results in (map 4.3 panel a). Similarly, the median cost c hapter 2 confirm not only that service ­ for Jakartans to reach a regional public hospi- ­ provision in Jakarta’s periphery is lower but tal is Rp 4,000, but the cost rises to Rp 20,000 also that the congestion challenges make the D ri v ers of U rba n a n d S patial I n clusio n   147 MAP 4.2  In nine major multidistrict metro areas, pockets of poverty exist alongside pockets of prosperity Source: Calculations based on data from Smeru Research Institute’s Poverty and Livelihood Map of Indonesia 2015 (http://www.smeru.or.id/en/content/poverty-and-livelihood-map- indonesia-2015). journey to access these services in the cores level shows a heavy concentration of human more costly. capital around the core of the city (map 4.4). The ­ surrounding areas in the outer periph- ery, largely around the city’s outer quadrants, More segregated cities generate weaker have markedly lower human capital. This human capital externalities difference is a manifestation of high housing Mapping the share of college-educated costs in the city and the difficulties newer adults in Greater Jakarta and a few other migrants face in finding a toehold, especially multidistrict metro areas at the kelurahan in the core. When human capital is so 148  TIME TO ACT MAP 4.3  Access to health and education services for Jakarta core a. Travel times to Jakarta core’s nine regional public hospitals b. Travel times to the three top-ranked senior high schools in each of Jakarta core’s kota Travel time Travel time (Minutes) (Minutes) 1–20 1–20 21–40 21–40 41–60 41–60 61–80 61–80 81–100 81–100 101+ 101+ PPP $3.10 Poverty rate (%) PPP $3.10 Poverty rate (%) 2 .57 2 .5 7 0.1 0.1 13 13 Sources: Calculations of travel times based on data from Google Maps and Trafi and of poverty rates based on data from Smeru Research Institute’s Poverty and Livelihood Map of Indonesia 2015 (http://www.smeru.or.id/en/content/poverty-and-livelihood-map-indonesia-2015). Note: PPP = purchasing power parity. MAP 4.4  Human capital and skills are concentrated in metro cores Source: Based on data from Indonesia’s 2010 Census. D ri v ers of U rba n a n d S patial I n clusio n   149 ­ oncentrated, there is very little scope for c experiences of intergenerational mobility. ­ spillovers from high-skilled to lower-skilled Neighborhoods and local environments play workers. Spillovers can happen only with a strong role in the long-term outcomes of integration and intermingling. All the chan- children, including their income in adulthood. nels through which spillovers occur—peer According to this research, neighborhood effects, role models, job contacts, norms, characteristics such as income segregation, behaviors, and social capital—are activated concentrated poverty, inequality, racial segre- when communities and neighborhoods are gation, quality of schools, and crime rates are diverse and when social networks overlap. important determinants of the extent to Residential segregation—a measure of the which children can break out of the intergen- spatial separation or sorting of the residential erational poverty trap (Chetty et al. 2014). living patterns of various social and economic How much do residential segregation and groups within a given geographic unit—has the concentration of human capital within cer- been studied extensively by sociologists and tain parts of metro areas matter for Indonesia? urban economists. The literature is especially We investigate this question by constructing rich for the United States, where there is much measures of human capital segregation for the interest in understanding the relationship 21 multidistrict metro areas and examining between segregation and modern racial differ- whether more highly segregated metro areas ences in socioeconomic outcomes. Segregation do worse in terms of the strength of the human in the United States has been linked to poorer capital externalities they can generate. schooling and labor market outcomes and Specifically, we construct two measures of seg- poorer health outcomes for blacks (see Logan regation that are widely used in the literature: and Parman 2017 for a review). Chetty et al. the index of dissimilarity and the index of iso- (2014) have also documented evidence link- lation (box 4.4). On the one hand, the index ing residential segregation to differential of dissimilarity is a measure of evenness of BOX 4.4  Measuring residential segregation The literature offers a variety of ways to measure in the geographic area. For our application, residential segregation. Each of these measures we compute this index for kelurahans within the captures different dimensions of segregation 21 multidistrict metro areas using data from the such as centralization, concentration, exposure, Population Census in 2010. evenness, and clustering. The measures we adapt The dissimilarity index captures how evenly to our study of residential segregation are the high- and low-skilled individuals are distrib- index of dissimilarity and the index of isolation. uted across kelurahans within each multidis- These indexes can be represented in the follow- trict metro area. If high-skilled individuals are ing manner: distributed identically to low-skilled individu- N als (for example, a kelurahan with 10  ­percent 1 Li Hi Dissimilarity =   2 ∑L total −  Htotal high-skilled individuals also has 10 ­ percent low- skilled individuals), then the measure of dissimi- i =1 larity will be zero. Conversely, as high-skilled N Li Li individuals become less evenly distributed across Isolation = ∑L total × Li + H i , these kelurahans, the index of dissimilarity will i =1 take on higher values. The isolation index, by where Hi and Li represent, respectively, the num- contrast, captures possible interactions between ber of high- and low-skilled individuals living the two groups. In a metro area with the least in a geographic subunit i within a geographic segregation, this measure will approach the unit that is constituted by N such subunits. share of the low-skilled population in the area, Htotal and Ltotal represent, respectively, the total whereas, in a metro area with the highest segre- number of high- and low-skilled individuals gation, the measure will approach 1. 150  TIME TO ACT groups across geographic subunits. Lower human capital externalities may work, at least evenness (more dissimilarity) implies that the in Indonesia, is not so much the result of skills lower-skilled groups are more likely to have concentration in particular areas within the differential access to schooling, health, and metro but the limited possibility of interaction labor market opportunities because of their between workers of high and low skills concentration within specific geographic units. through more connected and integrated neigh- The index of isolation, on the other hand, is a borhoods. This finding again highlights the measure of exposure and proxies for the importance of better integration within cities, potential contact—and by extension, for the and hence the Connect policy principle within possibility of cross-group social networks and this report’s ACT (Augment, Connect, Target) peer effects—between the high- and low- framework, through better housing policies skilled groups. and improved infrastructure for intracity Combining these measures of segregation connectivity. with the analysis of the strength of agglomera- tion forces presented in chapter 3, we find that human capital segregation within metro areas does indeed dampen the externalities that a Conclusion metro area’s stock of human capital can gener- This chapter analyzed some of the drivers of ate (table 4.3). The effect is statistically inde- the inequality between and within Indonesia’s terminate for segregation measured by the portfolio of places. The main findings are that dissimilarity index but is unambiguously neg- sizable and persistent welfare disparities exist ative for the isolation index, implying that the across space in Indonesia, suggesting that the channel through which the dampening of gains from urbanization have not been shared TABLE 4.3  More segregated multidistrict metro areas generate weaker human capital externalities Dependent variable: Hourly wage (ln) Dissimilarity index Isolation index Model 1 Model 2 Model 1 Model 2 Average years of schooling x Dissimilarity index 0.175 (0.567) Tertiary share of population x Dissimilarity index 0.049 (0.165) Average years of schooling x Isolation index –0.646** (0.236) Tertiary share of population x Isolation index –0.184*** (0.062) Observations 303,085 303,085 303,085 303,085 R-squared 0.499 0.499 0.500 0.500 Sources: Estimations are based on pooled data from the 2008–15 rounds of Indonesia’s National Labor Force Survey (SAKERNAS). Dissimilarity and isolation indexes are computed from data in Census 2010. Note: Each regression includes controls for individual characteristics (gender, age, age squared, length of tenure in current job, tenure squared, marital status, educational attainment), place-specific characteristics (population size, market access, access to ports), an interaction of the industry of employment with island-regions, and a dummy for year to pick out year effects. Standard errors are clustered at the metro level. Coefficients on the controls are not reported for presentation purposes. Full results are available upon request. Clustered standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. D ri v ers of U rba n a n d S patial I n clusio n   151 broadly across the country. Although place- targeting of transport investments and specific characteristics related particularly to policies to address barriers to migration ­ road density and access to markets explain a (chapter 6). larger share of the disparities, lingering differ- •  Second, Indonesia needs its decentraliza- ences in the levels of human capital of the tion to deliver better. In the poorer and population also matter. Migration has yielded more remote districts where access to ser- economic rewards for those who migrated, vices and human capital levels are low, the but it has the potential to do much more to capacity as well as the accountability of help spread the benefits of urbanization to the subnational governments can be strength- rest of the country. Metro areas provide better ened to augment the quality of service economic opportunities, but the benefits of delivery. In addition, in the more urbanized agglomeration appear to accrue much more and fast-urbanizing places, where resources to the high skilled than the low skilled, sug- to cater to the growing population may be gesting relatively weak spillovers across skill an issue, fixing the imbalances inherent in categories. The high degree of residential seg- the intergovernmental transfer formula regation has led to concentration of human may be required (chapters 5 and 6). capital in certain neighborhoods within metro •  Third, a whole host of policies at the areas, and the lack of integration within these housing–transportation nexus will need metro areas—which is associated with a lack to be addressed to better tackle issues of of intrametro connectivity—undermines the residential segregation and facilitate potential benefits of agglomeration through greater mobility within cities and to make human capital externalities. Indonesian cities more dynamic and These findings have several implications inclusive (chapter 7). for policy on how Indonesia could do better •  Finally, specific targeted place- and in spreading the benefits of urbanization more people-based policies may also be required ­ broadly, both across the country and within to address the needs of parts of the coun- urban areas. try that remain chronically excluded from the growth in the leading regions, and of •  First, to facilitate greater sharing of the segments of the population (elderly, peo- growth and prosperity in Indonesia’s metro ple living with disabilities) who, despite areas with other nonmetro areas, there is a living in cities, may be locked out of clear need to better connect, and hence inte- opportunities provided in urban areas grate, places through improved spatial (chapter 8). 152  TIME TO ACT Annex 4A Methodology to analyze spatial disparities The methodology to analyze the drivers of The methodology broadly proceeds in the spatial disparities in welfare follows Ravallion following steps. and Wodon (1999) and its extension in 1. Regress log of the welfare ratio on the Skoufias and Olivieri (2013). The method household characteristics separately for entails modeling household welfare as a func- each year for which the analysis is done. tion of household characteristics where the “returns” to these characteristics are in turn 2. Generate the “actual” welfare profile as: recognized as functions of the characteris- lnW p ( actual ) = α p + β p X p. tics  of the places in which the household 3. Generate simulated “returns” profile: resides. Places better endowed in general lnW p ( returns ) = α p + β p X N d i s t r i c t infrastructure—related to adequate provision ­ of education, health, drinking water, and returns, but national average of the Xs. sanitation—as well as connective infrastruc- ­ 4. Generate simulated “endowments” pro- ture are, theoretically at least, places where the f i l e : lnW p ( endowments ) = α N + β N X p returns on household characteristics such as national, population-weighted average demographic composition of households and returns, but district-level Xs. educational attainment of the household head 5. Examine whether the simulated returns and the spouse can be expected to be higher. or the simulated endowments profile Formally, the model is written as follows: correlates better with the actual welfare ( lnW p ( i ) = α p + β1 Z1 p p ,  Z2 )  ; Q X1 ( i ) profile and whether there is a time varia- tion in these correlations. Also examine ( p + β2 Z1 p ,  Z2 )  ; Q X2 ( i ) , whether the simulated returns or the sim- ulated endowments profile accounts for a where, for exposition, larger share of the variation in the actual p =f welfare profile. Z1  raction of households with access to 6. In the second stage, regress district- paved roads in district p, level returns to specific household Z2p = fraction of households with access to charac teristics X s on distric t-level “immobile” endowments— Z s such as health facilities in district p, roads, schools, health posts, water, sani- Q= institutional factors common across tation facilities, and so on. all districts, years of schooling of the household X1(i) =  The details of the application of this method- head, and ology are in Tiwari and Shidiq (2018), and number of children in the X2(i) =  table 4A.1 shows a regression result from one household, of the second-stage regressions. A key point to note is that the provincial poverty lines in and where the welfare profile is calculated by Indonesia take into account not only the dif- adjusting the per capita expenditure of house- ferences in the cost of procuring the same hold i in district p (PCEp(i)) for the spatial bundle of commodities that are in the poverty variation in the cost of living across line but also the provincial differences in pref- Indonesia’s provinces by using the ratio of the erences and consumption patterns. For that provincial poverty line (PL(r)) and the pov- reason, using the ratio of the poverty lines to erty line relative to a numeraire region, which account for spatial variation in prices, as we in this application is chosen to be DKI Jakarta. do, is likely to understate actual spatial dis- PCEp ( i ) parities because it is not the same bundle W p (i) =  PL ( r ) / PL(DKI )   being valued at different prices in the  Laspeyres sense. D ri v ers of U rba n a n d S patial I n clusio n   153 TABLE 4A.1  Proximate district-level characteristics correlated with welfare returns to secondary and tertiary education Dependent variable: District-level mean of welfare Dependent variable: District-level mean of returns to HH head w/secondary education welfare returns to HH head w/ tertiary education (percent of welfare) (percent of welfare) Model 1 Model 2 Model 3 Model 4 Model 1 Model 2 Model 3 Model 4 Road 0.0862** 0.0728* 0.0711* 0.0613 0.260** 0.239** 0.215* 0.204*   (0.0299) (0.0319) (0.0308) (0.0329) (0.0896) (0.0831) (0.0966) (0.0890) Health 0.00972 0.00454 –0.00783 –0.00763 –0.0953 –0.0976 –0.135 –0.123 (0.0292) (0.0281) (0.0298) (0.0287) (0.0937) (0.0932) (0.0964) (0.0960) School 0.0975* 0.100* 0.115* 0.118* 0.323* 0.340* 0.362* 0.377*   (0.0495) (0.0484) (0.0494) (0.0488) (0.156) (0.156) (0.160) (0.161) Water 0.0443 0.0620 0.0286 0.0532 0.0903 0.156 0.0525 0.137 (0.0598) (0.0564) (0.0617) (0.0586) (0.188) (0.184) (0.194) (0.190) Electricity –0.0283 –0.0978 0.0195 –0.0386 0.120 –0.0857 0.278 0.0686 (0.0679) (0.0698) (0.0921) (0.0868) (0.250) (0.247) (0.292) (0.271) Sanitation –0.0315 0.00525 –0.0272 0.00349 –0.237 –0.153 –0.248 –0.175 (0.0521) (0.0504) (0.0536) (0.0519) (0.145) (0.141) (0.147) (0.144) Floor –0.155* –0.105 –0.155* –0.102 –0.281 –0.138 –0.319 –0.168 (0.0667) (0.0686) (0.0695) (0.0713) (0.200) (0.203) (0.205) (0.209) Urban –0.0145 0.0279 –0.0484 –0.00416 0.0805 0.188 0.0170 0.135 (0.0384) (0.0385) (0.0385) (0.0395) (0.133) (0.128) (0.140) (0.135) Proportion –0.349 –0.183 –0.216 0.191 of pop. 15+ (0.183) (0.181) (0.563) (0.580) with high education Average –3.307*** –2.806*** –6.472** –5.421* years of (0.762) (0.778) (2.301) (2.393) schooling, ages 15+ Market     0.0110*** 0.00972***     0.0193* 0.0153 access     (0.00219) (0.00236)     (0.00782) (0.00793) Constant 29.56*** 54.52*** 26.82*** 46.63*** 72.52*** 123.0*** 66.98** 108.5*** (4.953) (7.357) (7.868) (8.850) (15.81) (22.55) (22.39) (26.35) Prob >F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 overall Prob >F. Joint 0.0017 0.0077 0.0000 0.0001 0.0031 0.0024 0.0029 0.0050 key variables Observations 298 298 286 286 298 298 286 286 R–squared 0.106 0.151 0.146 0.184 0.133 0.155 0.140 0.156 Note: Each column represents a separate regression of estimated district-level welfare return on the education level of the household (HH) head against a variety of district-level characteristics. The characteristics in the regression are road (percentage of villages in district with asphalt road), health (percentage of villages in district with health facility, puskesmas), school (percentage of districts with any primary and secondary school facility), water (percentage of households in district with protected pump/well/spring with the distance to the septic tank 10m), floor (percentage of household in district with nonearth floor), and urban (percentage of households in districts living in urban areas). The market access measure is described in annex 3A. Standard errors in parentheses. * p<0.05, ** p<0.01, ***p<0.001. 154  TIME TO ACT Annex 4B Returns to aggregate human capital TABLE 4B.1  Regression results for the returns to aggregate human capital Dependent variable: log of monthly earnings Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Years of education 0.096*** 0.084*** 0.053*** 0.053*** 0.053*** 0.053*** (0.001) (0.001) (0.006) (0.006) (0.006) (0.006) Average years of education (district) 0.114*** 0.078*** 0.061*** 0.059*** 0.092*** 0.057***   (0.003) (0.006) (0.017) (0.017) (0.021) (0.018) Average years of education         –0.056***   (district level) x Jawa-Bali (= 1) (0.018) Jawa-Bali (= 1)         0.228             (0.176)   Average years of education           0.032 (district level) x Jakarta-Bali (= 1) (0.029) Jakarta-Bali (= 1)           –0.147             (0.298) Age 0.058*** 0.057*** 0.096*** 0.096*** 0.096*** 0.096*** (0.003) (0.003) (0.004) (0.004) (0.004) (0.004) Age squared –0.001*** –0.001*** –0.001*** –0.001*** –0.001*** –0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Gender 0.393*** 0.404*** (0.009) (0.009) Marital status 0.137*** 0.154*** 0.028 0.031 0.031 0.031 (0.011) (0.012) (0.020) (0.020) (0.020) (0.020) Number of hours worked in primary 0.002*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** occupation (per month) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Year dummy variables Yes Yes Yes Yes Yes Yes Type of work, sector of work, and No Yes Yes Yes Yes Yes portfolio of places dummy variables Individual fixed effects No No Yes Yes Yes Yes District-level total factor productivity No No No Yes Yes Yes in manufacturing Constant 10.093*** 10.110*** 10.141*** 10.167*** 10.087*** 10.106*** (0.051) (0.076) (0.161) (0.162) (0.204) (0.165) Observations 47,329 43,115 43,115 42,788 42,788 42,788 R-squared 0.288 0.304 0.162 0.162 0.163 0.163 Number of individuals     25,415 25,293 25,293 25,293 Source: Calculations based on data from the pooled sample of the Indonesia Family Life Survey (IFLS) from 1993, 1997, 2000, 2007, and 2014. Note: Numbers in parentheses are robust standard errors. *** Significant at the 1 ­percent level. D ri v ers of U rba n a n d S patial I n clusio n   155 Annex 4C Returns to aggregate human capital by skill level TABLE 4C.1  Regression results for the returns to aggregate human capital by skill level Dependent variable: log monthly earnings Variable Full sample Jawa-Bali Jakarta-Bali Years of education 0.039*** 0.053*** 0.004 (0.009) (0.011) (0.017) Average years of education (district) 0.063*** 0.079*** –0.043 (Reference category: medium skilled) (0.019) (0.022) (0.045) Average years of education (district level) x low skilled –0.035** –0.037** –0.031   (0.016) (0.019) (0.031) Average years of education (district level) x high skilled 0.037* 0.024 0.079**   (0.019) (0.023) (0.037) Low skilled 0.343** 0.395** 0.202 (0.146) (0.172) (0.286) High skilled –0.175 –0.123 –0.455 (0.173) (0.206) (0.331) Individual characteristics (age, age squared, gender, Yes Yes Yes marital status, hours of work) Year dummy variables Yes Yes Yes Type of work and sector of work dummy variables Yes Yes Yes Individual x portfolio of places dummy variables Yes Yes Yes District-level total factor productivity in manufacturing Yes Yes Yes Constant 10.203*** 10.104*** 10.957*** (0.188) (0.220) (0.424) Observations 42,788 27,968 14,820 R-squared 0.163 0.159 0.178 Number of individuals 25,293 16,166 9,246 Source: Calculations based on data from the pooled sample of the Indonesia Family Life Survey (IFLS) from 1993, 1997, 2000, 2007, and 2014. Note: Numbers in parentheses are standard errors. **Significant at the 5 ­percent level; ***significant at the 1 ­percent level. 156  TIME TO ACT Notes 8. A specific description of how this has been adapted can be found in annex 4A. 1. Poverty is defined as the proportion of peo- 9. “Returns” here implies welfare returns—that ple with per capita household consumption is, per capita expenditure of the household below the national poverty line, and vulner- rather than nominal wages of a worker. ability is defined as the proportion of people 10. Indonesia’s National Socio-Economic Survey with consumption above the poverty line but (SUSENAS) began asking about migra- less than 1.5 times the poverty line. tion only in 2011. The Census asked about 2. Individuals are considered to belong to the migration in 1990, 2000, and 2010; but in middle class if their per capita household 1990, it asked only which province a person consumption is more than 3.5 times the pov- was living in five years ago, rather than the erty line. See World Bank (forthcoming) for district. The Indonesia Family Life Survey the methodology used to define the middle (IFLS) asked about migration consistently class and for the characteristics of Indonesia’s between 1993 and 2014, but this survey is emerging middle class. not fully nationally representative, is panel 3. Another distinction relates to the fact that, in nature, and so does not have the same when one focuses on workers, the place that time consistency as repeated cross-sections. is perhaps more relevant is not where work- Although the IFLS has admirably low attri- ers reside but where they work. Therefore, tion, migration is one outcome that will be chapter 3 works with places where workers disproportionately affected by attrition. actually work (using information on com- 11. Because the data use the IFLS, a much smaller muting, for example) as opposed to where sample than the SUSENAS, it was not pos- they live, whereas here we focus on places sible to unpack this finding to look at the where households live. extent to which Jakarta drives the result. 4. As shown in chapter 2, most of the overall 12. The perceptions survey was implemented inequality in Indonesia today is inequal- between March and April of 2018 by Survey ity within places, as opposed to inequality Meter. The sample consisted of 3,529 respon- between places. dents from a combination of multidistrict 5. In addition to, or instead of, per capita metro cores and peripheries, and it spanned household consumption, one could also con- both slum and nonslum areas. The districts sider using household expenditures adjusted included Bogor, Deli Serdang, DKI Jakarta for adult equivalence to take account of the (including West, Central, South, East, and differences in economies of scale within the North Jakarta), Makassar, Medan, Sidoarjo, household based on differences in family Surabaya, and Tangerang. The survey asked composition. We recognize that this would be migrants about their migration histories, an important refinement. However, because employment, assets, housing quality, percep- Indonesia’s official poverty measurement has tions of mobility, social integration, access to historically operated with per capita house- services, experiences with crime and disas- hold consumption, we stick to that measure ters, and policy preferences. Detailed results in this analysis. from the survey are in Wai-Poi and Sacks 6. See annex 4A regarding details of the meth- (2018). odology, including the exact nature of the 13. This preference is confirmed not only in spatial price adjustment. The method closely the survey data but also in data using cell follows Ravallion and Wodon (1999) and phone records. Medan draws migrants Skoufias and Olivieri (2013). mainly from around Suamatera, whereas 7. Provincial poverty lines in Indonesia con- the edges of Kalimantan are more closely sider not only the differences in the cost of linked with Makassar. In contrast, Jakarta procuring the same bundle of commodities, and Tangerang, most notably Jakarta, attract but also the provincial differences in prefer- migrants from all over the country. ences and consumption patterns. For that 14. These are official letters required to effect a reason, using the ratio of the poverty lines to change of address in an individual’s national account for spatial variation in prices, as we identification document. do here, is likely to understate actual spatial 15. For information about administrative docu- welfare disparities because it is not the same ments and services, migrants tend to turn bundle being valued at different prices in the to their RT and existing local organizations Laspeyres sense. such as paguyuban (informal organizations D ri v ers of U rba n a n d S patial I n clusio n   157 with shared origins but also professions) 20. According to PISA, a 30-point difference in and religious organizations, all of which test scores translates to roughly a year of are known and trusted. In what people see schooling. as safer neighborhoods, people talked about 21. The monetary values reported are in 2017 having a strong and effective head of RT. In rupiah. Tangerang, the survey team was told that landlords work together with the RT and the Rukun Warga (the administrative unit one level below the village or kelurahan level) to References protect rented homes, and newcomers usu- Bosker, M., J. Park, and M. Roberts. 2018. ally are accepted only with a recommenda- “Definition Matters: Metropolitan Areas tion from existing residents. The RT requests and Agglomeration Economies in a Large visitors to report to the RT so it can see who Developing Country.” Policy Research Working goes in and out of the community. In the Paper 8641, World Bank, Washington, DC. Jakarta urban slum, the survey team was told Chetty, R., N. Hendren, P. Kline, and E. Saez that the RT helps renew KTPs for residents 2014. “Where is the Land of Opportunity? for Rp 150,000 because people were “afraid The Geography of Intergenerational Mobility to go to the (kelurahan) office as you have to in the United States.” The Quarterly Journal of be clean, they ask lots of questions, too com- Economics 129 (4): 1553–623. plicated” (man, Jakarta Urban Slum). Ferreyra, M. M. 2018. “Human Capital in Cities.” 16. The difference in the growth rate in aver- In Raising the Bar for Productive Cities in Latin age per capita consumption of the bot- America and the Caribbean, edited by M. M. tom 40  ­ percent relative to the growth rate Ferreyra and M. Roberts. Washington, DC: in average per capita consumption for the World Bank. overall population can also be regarded as Hulme, D., and A. Shepherd. 2003. “Chronic the “shared prosperity premium,” which Poverty and Development Policy.” World is a measure of how overall growth has Development 31 (3): 55–75. been shared in a given geographic area Lain, J. 2018. “Multi-dimensional Urban Poverty (World  Bank 2016). Over the period 2001– in Indonesia.” Background paper for this report, 17, the shared prosperity premium has been World Bank, Washington, DC. negative in all of Indonesia’s places, and the Logan, T., and J. Parman. 2017. “The National largest (in absolute values) in the multidis- Rise in Residential Segregation.” The Journal of trict metro cores and urban peripheries. Economic History 77 (1): 127–70. 17. This follows from the standard neoclassical Moretti, E. 2004. “Estimating the Social model. A growth in the relative supply of Return to Higher Education: Evidence from skilled workers drives down the wages of the Longitudinal and Repeated Cross-sectional high skilled and increases the wages of the Data.” Journal of Econometrics 121 (1–2): low skilled. This could happen also through 175–212. the demand channel: an increase in the rela- Ravallion, M., and Q. Wodon. 1999. “Poor Areas, tive supply of high-skilled workers could or Only Poor People?” Journal of Regional increase the marginal product of labor of the Science 39 (4): 689–711. low-skilled group. Setiawan, I., S. Tiwari, and H. Rizal. 2018. 18. Workers are considered low skilled if they “Economic and Social Mobility in Urbanizing have education below the elementary level, Indonesia.” Background paper for this report, medium skilled if they have education Washington, DC, World Bank. between junior secondary and secondary lev- Skoufias, E., and S. Olivieri. 2013. “Sources els, and high skilled if they have education of Spatial Welfare Disparities in Indonesia: above the upper-secondary level. Household Endowments or Returns?” Journal 19. Unlike the analysis in chapter 3, which was of Asian Economics 29 (C): 62–79. based on a pooled cross-section of Indonesia’s Strauss, J., F. Witoelar, and B. Sikoki. 2016. “User’s Labor Force Survey (SAKERNAS), this anal- Guide for the Indonesia Family Life Survey, ysis uses an unbalanced panel of all workers Wave 5.” Jakarta, Indonesia: Rand. over age 18 from the Indonesia Family Life Tiwari, S., and A. R. Shidiq. 2018. “Sources of Survey (IFLS), which allows us to account for Welfare Disparities in Indonesia.” Background unobserved heterogeneities. The methodol- paper for this report, Washington, DC, ogy closely follows Moretti (2004). World Bank. 158  TIME TO ACT Wai-Poi, M., and A. Sacks. 2018. “Better Living Background paper for this report, World Bank, in the City: Trends, Patterns, and Outcomes of Washington, DC. Migration in Indonesia.” Background paper for ———. 2019. World Development Report 2019: this report, World Bank, Washington, DC The Changing Nature of Work. Washington, DC: World Bank. 2016. Poverty and Shared Prosperity World Bank. 2016: Taking on Inequality. Washington, DC: ———. Forthcoming. “Indonesia’s Aspiring Middle: World Bank. Expanding the Middle Class.” World  Bank, ———. 2018. “People’s Perspectives of Urban Washington, DC. Poverty and Rural-to-Urban Migration.” D ri v ers of U rba n a n d S patial I n clusio n   159 SPOTLIGHT 1 Strengthening the Disaster Resilience of Indonesian Cities Through the high-level policy principles of disasters across Indonesia’s 28 multi- and Augment, Connect, and Target, Indonesia’s ­ ingle-district metropolitan areas (as defined in s urbanization opens the door for enhanced chapter 1), or about 25 percent of all disaster prosperity, inclusiveness, and livability. At the events in Indonesia, were recorded in the same time, urbanization processes and urban Indonesia Disaster Data Information (DiBi) development increase the exposure of people database, maintained by the National Disaster and assets to disaster risk. Some 110 million Management Authority ( Badan Nasional people, or 42 percent of the population, Penanggulangan Bencana, or BNPB). Bandung, across some 60 Indonesian cities are exposed Jakarta, and Surakarta alone account for to natural hazards (Gunawan et al. 2015). 39 percent of the disasters hitting metro areas, The number is expected to increase because with all three far exceeding the average of of urban population growth and associated 173 events per metro area between 2003 and transformation of the built and natural envi- 2017. According to BNPB data, hydrological ronment, projected effects of climate change, disasters are the most frequent, affecting the and more widespread land subsidence. highest number of people, and these disasters Despite the factors that increase exposure to impair the prosperity and livability of metro natural disaster risk, Indonesia can reduce the areas, often disrupting public and economic vulnerability of its cities and their inhabitants life. Between 2003 and 2017, floods adversely by improving access to quality disaster and cli- affected more than 4.7 ­ million Indonesians and mate risk information for planning, by upgrad- caused 720 fatalities in Indonesia’s metro areas. ing the structural integrity of urban The concentration of people and assets in infrastructure, by enhancing early warning and cities is directly linked with higher damages emergency management systems, and by per disaster event. Excluding the impacts of strengthening the capacity of communities and the 2004 Indian Ocean earthquake and tsu- subnational governments to manage disaster nami, Indonesian metro areas accounted for risks. Policy reforms, investments in disaster about one-half of all recorded disaster-related risk management, and better institutional coor- fatalities between 2003 and 2017, despite dination are critical to minimize loss of life, being impacted by only one-quarter of all reduce damage to assets and the economy, and events in the country. This phenomenon protect and further enhance the prosperity, is especially stark in the case of geophysical inclusiveness, and livability of Indonesia’s cities. events, during which an overwhelming major- ity of the fatalities occur in metro areas. Because of the country’s location in a seis- How exposed are Indonesian mically active region, all metro areas (except for those on Kalimantan) are exposed to geo- cities to disasters? physical hazards. These less frequent but more Indonesian cities face a wide range of natural severe geophysical events have the most devas- hazards, particularly floods, earthquakes and tating impacts on the population (­ figure S1.1), tsunamis, volcanic activities, landslides, and recording by far the highest number of fatali- storms. 1 Over the past 15 years, 4,856 ties since 2003. With approximately five FIGURE S1.1  Indonesia’s metro areas face several different disaster types, 2003–17 a. Disaster type b. People a ected per event by disaster type c. People a ected by disaster type Tidal wave/ Strong wind, Flood, Landslide, Tidal wave/ Earthquake, Volcanic abrasion, Tsunami, 78,815 2,407 (0%) 70,119 (1%) abrasion, 22 (0%) eruption, 707 (0%) 317,072 (1%) 42 (1%) 9 (0%) Strong wind, (4%) Volcanic Tidal wave/ 54 (0%) Volcanic Tsunami, eruption, abrasion, 0.4 (0%) 64,888 (7%) Landslide, eruption, 29,518 (1%) 52 (0%) 559,792 Earthquake, (7%) 92,217 (10%) Landslide, Flood, 1,361 (28%) 1,976 (41%) Earthquake, 2,069,275 Tsunami, (26%) Flood, 4,756,929 747,384 (83%) (60%) Strong wind, 1,446 (30%) Source: Calculations based on data from the National Disaster Management Authority’s Disaster Data Information (DiBi) database. Note: The 28 multi-and single-district metropolitan areas are defined using the methodology described in box 1.5, chapter 1. “People affected” includes fatalities, missing, injured, and displaced. events each, Bandung, Sukabumi, and Yogyakarta were the metro areas most prone A look ahead to geophysical disasters, though Padang also percent It is expected that, by 2055, about 64 ­ faces significant risk. In the urban context, in of Indonesia’s population will be living addition to their impact on population, such in seismic hazard zones, up from 53 percent events can cause significant destruction to the in 2016, with the largest increase in expo- stocks of both infrastructure and housing and sure across Jawa island (Deltares 2018). can inflict severe economic damage when net- The  exact timing and severity of tectonic works are disrupted and services cease to plate movements is difficult to predict, mean- function. For example, the 2006 earthquake in ing that assumptions about the future Yogyakarta province caused approximately impacts of earthquakes are difficult to calcu- 5,800 fatalities, injured and displaced more late accurately. According to probabilistic than 2.2 million people, and created losses loss models, there is a 2 percent chance each equivalent to an estimated 41 percent of pro- year of a significant earthquake event that vincial gross domestic product at the time causes damages of about US$1.3 billion (World Bank 2011).2 In 2018, Lombok island (World Bank 2011). 4 Secondary hazards suffered a series of moderate to strong earth- including tsunamis and liquefaction also quakes affecting the entire island’s population pose significant threats to cities located in of about 3.5 million and thousands of tourists. districts with high seismic risk, as witnessed Damage estimates point to at least 515 fatali- in Palu in the aftermath of the Central ties and the displacement of more than Sulawesi earthquake in 2018. The approxi- 400,000 residents, with preliminary eco- mately 130 active volcanoes in Indonesia can nomic  damages and losses estimated at also affect urban areas through lahar flows US$528  ­ m illion (AHA Centre 2018). 3 (destructive mud or debris flows from volca- In 2018, Central Sulawesi also witnessed a nic eruptions) or by indirectly causing eco- powerful earthquake and subsequent tsunami nomic damage triggered by air traffic and liquefaction, which caused loss of lives, disruptions, particularly for major hubs such loss of infrastructure, and economic damages as Bali (Llewellyn 2018). with the banking and retail sector coming to a Today, many Indonesian metro areas halt for a period. already experience pluvial flooding linked to 162  TIME TO ACT heavy rainfall events, storm water runoff, events and high tides—or a combination of and drainage system overflow; fluvial various types (box S1.1). It is expected the ­ ­ f looding linked to upstream rainfall, that fluvial flooding alone will put an esti- flash  floods, and overflowing rivers; and mated 75 ­ percent more Indonesians at risk coastal flooding linked to extreme weather by 2055 (map  S1.1), mainly because of BOX S1.1  Floods of many types: The example of Mataram City With a population of about 400,000, Mataram coastlines to high tides, and fluvial flooding occurs City, which is a nonmetro urban area in the when the rivers overflow during intense rainfall province of Nusa Tenggara Barat, has developed events. Riparian settlements have encroached on around four major rivers—Sungai Ancar, Sungai river banks and coastal areas, and poorly designed Jangkok, Sungai Midang, and Sungai Unus— water infrastructure impedes drainage, exacerbat- which pass through the city from the slopes of ing floods. In March 2018, some 1,000 houses in Mount Rinjani. Pluvial flooding severely inundates Mataram were inundated by flash floods. streets and ground‑floor building spaces, coastal flooding frequently exposes low‑lying land along Source: Nugraha 2018. MAP S1.1  There will be a substantial increase in the number of people exposed to fluvial flooding, between 2016 and 2055 Source: Calculations based on National Disaster Management Authority’s hazard maps overlaid with the aqueduct global models, prepared by Deltares for this report. Note: The maps show the combined effect of urban growth and climate change using the Badan Pusat Statistik (BPS or Statistics Indonesia) 2016 and 2055 (projected) population figures in combination with the aqueduct flood hazard maps. The flood maps apply very coarse rainfall-runoff grid cells of about 55 by 55 kilometers, which limits their usefulness for smaller islands and small rivers. The results of the future flood hazard maps are based on significant uncertainty regarding the impact of climate change on peak precipitation levels in Indonesia. “People exposed” includes all residents living in urban flood zones. S P O T L I G H T   163 MAP S1.2  The number of people exposed to coastal flooding will increase significantly between 2016 and 2055 2016 2055 Number of people exposed 0–10,000 10,001–20,000 20,001–50,000 50,001–100,000 100,001–200,000 200,001–500,000 >500,000 Source: Calculations based on National Disaster Management Authority’s hazard maps overlaid with the aqueduct global models, prepared by Deltares for this report. Note: The maps show the combined effect of urban growth and climate change using the Badan Pusat Statistik (BPS or Statistics Indonesia) 2016 and 2055 (projected) population figures in combination with the aqueduct flood hazard maps. The dynamic interactive vulnerability assessment projects sea level rises on a regional level only (island groups), which limits the usefulness for smaller islands and watersheds. The models used for coastal flooding do not consider expected future land subsidence and land use change. “People exposed” includes all residents living in urban flood zones. population growth (47 ­ percent), with cli- mate change having a lower impact (19 per- What are the drivers of cent). Along Indonesia’s coastline, with some urban disaster risk? 8,000 inhabited islands, cities will be even Unplanned or poorly planned urbanization. more exposed to coastal flooding because of Cities in Indonesia have grown rapidly over projected sea level rises. There will be an the past several decades, though the overall estimated 73 percent increase in the number pace of urbanization has started to slow (see of Indonesians living in coastal flood hazard chapter 1). Urbanization, linked mainly to zones by 2055, with 36 percent stemming natural population growth and rural–urban from population growth and 26  percent transformation, has increased the exposure from the effects of climate change (map S1.2) of cities to natural hazards (Deltares 2018).5 (Deltares 2018). Within Asia, Jawa is seen as Large-scale urban development—often one of the regions most affected by increas- poorly planned and inadequately ­ regulated— ing surface temperature and the changing also increases the vulnerability of cities to rainfall patterns that are expected to follow natural hazards. New urban migrants and (Willner et al. 2018). assets are often pushed toward vulnerable 164  TIME TO ACT areas, such as flood-prone land and steep Land subsidence and environmental hillsides.6 In addition, poor‑quality infra- ­degradation. Land subsidence is increasing structure is constructed in hazard-prone the risk of flooding in major metro and areas with inadequate consideration of, or urban areas along rivers or coastlines, such compliance with, risk‑informed planning as Bandung, Jakarta, Lhokseumawe, regulations and urban design codes. As a Medan, and Semarang. Land subsidence is result, buildings and urban infrastructure are occurring rapidly because of groundwater unable to withstand damaging geotechnical extraction (often illegal) for industrial use, and hydrometeorological forces. Many gas extraction, and natural processes major Indonesian cities suffer from water- (Chaussard et al. 2013). Bandung and shed degradation, reduced hydraulic capac- Jakarta are sinking at an alarming rate ity of drainage systems, and increased of 7.2 centimeters a year (Deltares 2018). nonabsorptive surfaces, prolonging flooding Medan is expected to be below sea (Gencer 2013). Pluvial flooding is particu- level  within the next 60 years, and parts larly exacerbated by urban developments of  Jakarta Utara are already below sea that create additional impervious surfaces, level.  Environmental degradation and which increase water runoff during intense deforestation associated with inadequately ­ rainfall and do not absorb storm water on- managed urban population growth, land site. Jakarta is a vivid example, with its shortage, and commercial activities are urban growth rate outpacing urban livability major contributors to pluvial floods and and disaster resilience (box S1.2). landslides. These problems are typical for BOX S1.2  An inundated capital: The 2007 floods in Jakarta The multidistrict metro of Jakarta, home to airport for three days) and health costs when more than 30 million people, is the political dengue fever and diarrhea broke out. center and economic hub of the nation. Parts The Jakarta provincial government has been of Jakarta lie below sea level, and 10 million working with the support of the World Bank to residents are at risk of urban flooding, coastal remove some of the contributors to the 2007 floods. erosion, and sea level rise. Extensive ground- To date, more than 3 million cubic meters of dredge water extraction by large urban developments, material have been removed from floodways and including residential complexes, shopping canals, and about 52 ­ kilometers of embankment malls, and industrial development, is causing have been constructed or repaired. Although these widespread land subsidence. Large areas of the measures helped address some of the flood risk city flood each year during the rainy season faced by the city, its residents continue to suffer (November to April). Flooding was especially from recurring floods during the rainy season. In severe in February 2007, when 36 percent of both 2013 and 2014, the capital was heavily inun- Jakarta’s metro core, Daerah Khusus Ibukota dated for many days in January, causing combined Jakarta (flood extent map, DKI Jakarta, damages of about US$3.6 ­ billion (2013/2014 price Dinas-PU), was inundated—up to 7 meters in levels). Going forward, there are many opportuni- some parts. This disaster caused 70 fatalities, ties to strengthen the city’s integrated management displaced 340,000 residents from their houses, of the many disaster risks it faces. affected 2.6 million people, and led to about Sources: Data Informasi Bencana Indonesia (database), BNPB, Jakarta, Indonesia US$900 million in damage. a Additional eco- (accessed August 14, 2018), http://bnpb.cloud/dibi/; EM-DAT: The Emergency Events nomic losses were caused by lost workdays Database, CRED (Centre for Research on the Epidemiology of Disasters), Brussels, Belgium (accessed August 14, 2018), https://www.emdat.be/classification; World Bank and schooldays because of citywide disruption 2012. (including the closure of Jakarta’s international a. The damage estimate is based on 2009 prices. S P O T L I G H T   165 cities expanding into adjacent mountain- disaster-resilient, and climate-resilient ous  areas, as observed in Ambon and cities is a key direction under Indone- Manado. The loss of uphill or coastal for- sia’s National Medium‑Term Devel- ests is especially detrimental to a city’s flood opment Plan (RPJMN) 2015­ –19. The resilience. government’s 30‑year disaster manage- Climate change. Expected sea level rise, ment master plan aims to reduce the changing precipitation patterns, and more number of cities with high disaster risk intense storms will increase disaster risks from 75 percent to 40 – 45 percent by across Indonesian metro and urban areas. 2045. To address these goals, provid- Sea  level rise could threaten 42 million ing dedicated resources with a sustain- Indonesians who live less than 10 meters able financing strategy, incentives, and above sea level. A 50-centimeter sea level rise, mechanisms will help increase urban combined with land subsidence in Jakarta resilience to shocks and economic dis- Bay, could permanently inundate densely pop- ruption and create multifunctional resil- ulated areas of Bekasi and Jakarta that house ient green spaces, enhancing urban pros- more than 270,000 residents.7 Where peak perity and livability. precipitation increases because of climate 2. Improve coordination between urban change and large‑scale climate systems (such planning and disaster risk manage- as El Niño), the risk of flooding will increase, ment agencies. Better agency coordi- although the exact impacts are not clear nation is needed for developing city (Deltares 2018). resilience master plans, improving disaster and spatial data, and updating public infrastructure and spatial plans What needs to be done: (see ­c hapter 5). Agencies are encour- A holistic approach to aged to allow transparent access to key risk information, such as hazard urban disaster resilience probabilities, damage estimates, and Over the last decade, Indonesia has made sig- emergency planning. Improving and nificant advances to increase disaster resil- sharing quality information is crucial ience and is currently redefining its national for citywide risk-informed planning disaster risk management priorities in and communicating disaster risk to the long‑term disaster management planning and community. risk financing. Sustainable and risk-informed 3. Invest in institution al and hum an approaches to urban development are ­c apacit y. Subnational govern ment required to bolster and safeguard the prosper- agencies in metro and urban areas ity, inclusiveness, and livability gains of experience challenges in building and urbanization that would follow from the retaining the technical capacity to other policy actions elaborated in this report. accommodate growing populations Improved planning and institutions in metro (see chapter 5). The need is just as and urban areas and more resilient infrastruc- pressing for policies and practices for ture to cope with increasing disaster risks are resilience. Government agencies could needed. The quality of urbanization is critical introduce certification and incentives and can be improved by incorporating disas- for staff to improve skills, particularly ter mitigation across institutions, planning on risk­‑informed spatial planning, and infrastructure, and society and finance enforcement of zoning regulations through the following 10 steps. and building codes, resilient designs, technical engineering, retrofit ting, Institutions and disaster risk management. Pro- 1. Align national development and disas- fessional organizations could also be ter risk management strategies with encouraged to certify designers and dedicated resources. Building green, developers. 166  TIME TO ACT Planning and infrastructure and nonstructural measures (such as r isk-informed land use planning, build- ­ 4. Scale up risk‑informed spatial plan- ing codes, planning policies, hazard ning and urban design and building monitoring, forecasting, early warning codes that incorporate risk reduction systems, disaster risk financing, com- standards. Planning and designing mu n it y awa reness a nd educ at ion , resilient metro and urban areas will and capacity building). For example, help avoid the loss of lives and assets simulations point out that enhanced (including residential housing and crit- f lood defenses in Indonesian cities ical urban infrastructure) and prevent could reduce up to 93 percent of the major disruptions to urban services risk, though such measures are likely and economic activities when a hazard feasible only in high-risk areas (Muis strikes. Flood modeling for Indonesian et al. 2015). The World Bank (2012) cities shows that rigorously enforced estimates that upgrading hydromet ser- spatial planning can reduce f lood vices in developing countries is one of exposure by 50–84 ­ p ercent, being par- the most cost-effective solutions with ticularly effective in cities with rapid benefit-to-cost ratios between 4 to 1 urban expansion (Muis et al. 2015). and 36 to 1. To ­ h arness urbanization As Indonesian cities prepare detailed for improved livability, plans could spatial plans in the coming years, there standardi ze at trac tive, g reen, and is an opportunity to further integrate resilient cities that incorporate water-­ resilience measures into urban plan- sensitive urban design (Soz, Kryspin- ning and development and to ensure Watson, and Stanton-Geddes 2016). that new building code standards stip- These initiatives may have further ulate disaster risk reduction measures benefits such as sustainable tou r- (box S1.3). ism, improved mental and physical 5. Refor m pe r mit planning and con - well-being, and increased com mu- s t r u c t i o n p ro c e s s e s . S ubn at ion a l nity awareness of disaster and climate gover n m e nt s c ou ld b e supp or t e d change risks. to strengthen building permit issu- 7. Address both existing and future vul- ance and compliance processes to nerability through development plan- ensure that development follows risk- ning and investments. A long with informed planning, design, and con- improving planning and practice for struction methods. Clearly assigned future development, it is critical to responsibilities in a transparent sys- address the vulnerability of the exist- tem, with human resources trained and ing urban infrastructure stock. Spe- certified to monitor compliance, are cific activities could include systematic essential to safeguard urban lives and assessment of risk levels in target areas mitigate urban economic losses due to or infrastructure types with prioritiza- damaged houses and public buildings. tion and technical standards, citywide Following disasters, a “build back bet- retrofit ting prog rams (add ressing, ter” approach to reconstruction efforts for example, foundations, structural should be adopted to strengthen resil- framing, external facades, roofing, and ience against future events. interior finishes), and ­i ntegrated build- 6. Bal an c e st r uc tural an d n on st r uc - ing information systems. Early warn- tural investme nts and st and ardize ing and awareness programs through approaches with large co -benefits. community institutions, contingency Comprehensive strategies combine and response arrangements, and a structural measures (such as drain- comprehensive disaster risk finance age i n f ra s t r uc t u re , s ei sm ic re si l- strategy are also key to dealing with i e n c e , a n d g r e e n i n f r a s t r u c t u r e) the impacts of disasters. S P O T L I G H T   167 BOX S1.3  Risk-informed spatial planning in Indonesia Indonesia has a sophisticated spatial planning be prepared and spatial plans updated to include system that relies on a national spatial plan the relevant disaster risk and hazard informa- (RTRWN), provincial spatial plans (RTRWPs), tion (Jha, Bloch, and Lamond 2012); therefore, district-level spatial plans (RTRWs), and detailed cross-institutional engagement is needed among spatial plans (RDTRs).a The RTRWs consider district-level agencies responsible for spatial plan- strategic activities, transportation networks, ning, development planning, public works, hous- population distribution, disaster evacuations, ing settlements, and disaster risk management. green and nongreen areas, and public services. The Ministry of Agrarian Affairs and Spatial The RDTRs provide zoning maps and regula- Planning (ATR/BPN) has a national program to tions. Because of varying l ­evels of subnational support the inclusion of disaster risk reduction government capacity, however, not all cities have principles in district‑level spatial plans. RDTRs. Only a few major cities developed Urban Following the 2006 Jawa earthquake and the Design Guidelines ( Rencana Tata Bangunan 2010 Mount Merapi eruption, the World Bank Lingkungan, or RTBLs) to regulate the design of supported risk-sensitive land use planning and buildings, roads, housing and residential areas, resettlement in Indonesia. Detailed analysis iden- and spatial planning. The RTBLs, which include tified zones and groups of vulnerable populations detailed maps at 1:1,000 to 1:1,500 scale, are to guide resettlement and resilient reconstruction. usually developed to monitor and regulate urban In some zones, development, including housing developments in areas such as central business construction, was prohibited; in other zones, districts or economic zones. partial resettlement was permitted; and, in yet Risk-based land use planning guides people other areas, efforts focused on strengthening the and assets within or away from hazard-prone preparedness of the population. Reconstruction zones by defining the location, type, design, qual- and resettlement were implemented through ity, and timing of development (Jha, Brecht, and community-driven participatory processes, where Stanton-Geddes 2015). Zoning enables cities to community settlement plans were part of regional identify safer areas; regulate land use; set regu- risk-informed land use planning (Jha, Brecht, and lations on location, bulk, height, shape, and use Stanton-Geddes 2015). of structures in each zone; and identify building a. RDTR = Rencana Detail Tata Ruang; RTRW = Rencana Tata Ruang Wilayah; codes by design type and purpose of structure RTRWN = Rencana Tata Ruang Wilayah Nasional; RTRWP = Rencana Tata Ruang (World Bank 2013). Adequate regulations need to Wilayah Propinsi. Society and finance to ­e ncourage developers to design risk-informed investments. Innovative 8. Promote private sector participation municipal financing streams for drain- in risk reduction through clear roles, age infrastructure and storm water incentives, and regulations. The private management (such as developer levies sector, which often drives construction or fees, incentive schemes, and capital and land redevelopment, should bear works schemes in new developments), its fair share of responsibilities toward programs for retrofitting of critical urban resilience. Incentives and regu- infrastructure (such as incentive or lations for development projects (such grant schemes and ­ public–private part- as those for drainage, permeability, nerships), or land value capture oppor- retention storage, and adherence to tunities in floodplain management and building code standards) are needed development could be considered. 168  TIME TO ACT 9. Invest in communities. Metro and urban government and implemented at the city areas offer many economic opportuni- level to initiate resilient urban develop- ties; but many people, facing unafford- ment. Technologies could facilitate such able housing because of congestion and approaches, with social media and smart high prices (see chapters 2 and 7), have technologies supporting risk mitigation, little choice but to live along river banks early warning, and disaster responses or in informal settlements, where disaster (see box S1.4). Geographic Information risk is highest and housing quality is poor. System (GIS) software and new tools This situation contributes to the poor’s such as drone technologies are instru- disproportionate vulnerability to disas- mental in creating and analyzing spa- ters, which undermines the inclusiveness tial data for forecasting, monitoring, of cities (Hallegatte et al. 2017). Com- and communicating risks (World Bank munities are key to finding sustainable 2013). Agent-based modeling, which solutions, especially related to waste man- simulates how individuals interact with agement and environmental degradation. each other and their environment, could National participatory planning processes enable human behavior to be integrated (musrenbang) could further address the into risk assessments, creating more real- varied needs of diverse community stake- istic scenarios of the impact of natural holders. Urban resilience depends on disasters (Jongman 2018). Nonpropri- addressing the key issues of communities etary open-source platforms and tools and the needs of the most vulnerable. enable pooling, sorting, analyzing, and 10. Pilot innovative programs and use sharing risk data for disaster planning, new tools. There is a need for new preparedness, or response (Jha, Brecht, approaches supported by the central and Stanton-Geddes 2015). BOX S1.4  Disaster information crowdsourcing: PetaBencana.id Indonesia’s National Disaster Management analyzing all Twitter feeds, but it avoids some of Authority (BNPB) is working with part- the key challenges such as inaccurate geographi- ners, including the MIT Urban Risk Lab, cal location and the need to filter ambiguous Zurich Insurance, and Twitter Inc., to ­ create or secondhand information in the Twitter mes- ­ c ommunity-led flood maps for Bandung, sage body. PetaBencana.id, like its predecessor Jakarta, Semarang, and Surabaya. The maps PetaJakarta.org, aims to provide a comprehen- draw on a web-based platform that harnesses sive overview of disaster events, enabling more information on location, flood severity, and informed decisions during emergencies, which on-ground descriptions provided by social helps reduce flood risk and assist in relief efforts. media users as well as by government agen- For example, during the rainy season in February cies. Some similar projects—such as the coop- 2015, more than 1,000 flood sites across Jakarta eration between the World Bank, Global Facility could be mapped and were subsequently viewed for Disaster Reduction and Recovery, and more than 160,000 times. Both the International FloodTags—attempt to automate the analysis Federation of the Red Cross and the Federal of Twitter content and integrate the information Communication Commission of the United States into flood maps. have recommended PetaBencana.id as a best PetaBencana.id takes a different approach, practice for disaster information crowdsourcing. prompting Twitter and Telegram users to input standardized information as well as photos once Sources: Holderness and Turpin 2016; PetaBancana.id website, https://info. petabencana.id/about/; Global Facility for Disaster Reduction and Recovery they flag a flood. Requesting additional inputs website, https://www.gfdrr.org/en/publication/challenge-fund-project- leads to a lower data quantity compared to overview-floodtags. S P O T L I G H T   169 Notes Subsidence Due  to Groundwater and Gas Extraction.” Remote Sensing of Environment 1. Data from EM-DAT: The Emergency Events 128: 150–61. Database, CRED (Centre for Research on Deltares. 2018. “Baseline Analysis of Urban Flood the Epidemiology of Disasters), Brussels, Risks and High Priority Investment Gaps in Belgium (accessed August 14, 2018), https:// Indonesian Cities. Technical Report.” World www.emdat.be/classification. Disasters are Bank, Washington, DC. generally categorized as geophysical haz- Gencer, E. A. 2013. The Interplay between ards (earthquakes and tsunamis, volcanic Urban Development, Vulnerability, and Risk activity, and dry landslides), hydrological Management: A Case Study of the Istanbul hazards (floods, flash floods, storm surges/ Metropolitan Area . Heidelberg, Germany: coastal floods, and wet landslides), and Springer-Verlag. meteorological hazards (tropical cyclones Gunawan, I., S. Sagala, S. Amin, H. Zawani, and and storms). R. Mangunsong. 2015. City Risk Diagnostic 2. Data from Data Informasi Bencana Indonesia for Urban Resilience in Indonesia. Washington, (database), BNPB, Jakarta, Indonesia (accessed DC: World Bank. August 14, 2018), http://bnpb​ .cloud/dibi/. Hallegatte, S., A. Vogt-Schilb, M. Bangalore, 3. The damage and loss estimates are based and J. Rozenberg. 2017. Unbreakable: on 2018 prices. An exchange rate of Building the Resilience of the Poor in the US$1:14,583 Indonesian rupiah was applied. Face of Natural Disasters. Washington, DC: 4. The damages estimate was reported in World World Bank. Bank (2011), which provides no information Holderness, T., and E. Turpin. 2016. “How on the exchange rate used to convert from Tweeting about Floods Became a Civic Duty Indonesian rupiah to U.S. dollars. The estimate in Jakarta.” The Guardian , January 25. represents the “Probable Maximum Loss” for https://www.theguardian.com/public-leaders​ the government of Indonesia with a 50-year -network/2016/jan/25/floods-jakarta-indonesia​ return period. The actuarial methods are based -twitter-petajakarta-org. on loss data between 2000 and 2009. Jha, A. K., R. Bloch, and J. Lamond. 2012. Cities 5. Indonesia’s urban population is projected to and Flooding: A Guide to Integrated Urban grow from 115 million people in 2016 to 189 Flood Risk Management for the 21st Century. million in 2055 (77 percent of the total popu- Washington, DC: World Bank. lation). Urban areas are projected to grow by Jha, A. K., H. Brecht, and Z. Stanton-Geddes. 44 percent, from 17,000 square kilometers in 2015. “Building Resilience to Disasters and 2016 to 24,000 in 2055. Climate Change in the Age of Urbanization.” 6. Information comes from the “Asia Pacific” In  Disaster Risk Reduction for Economic page on the Lloyd’s City Risk Index (data- Growth and Livelihood: Investing in Resilience base), Lloyd’s, London, https://cityriskindex​ and Development , edited by I.  Davis, .lloyds.com/explore/. K.  Georgieva,  and K. Yanagisawa, 7–27. 7. Data from the Indonesia Dashboard London: Routledge. (database), World Bank, Washington, DC Jongman, B. 2018. “Effective Adaptation to (accessed August 14, 2018), http://sdwebx​ Rising Flood Risk.” Nature Communications, . w o r l d b a n k . o r g / c l i m a t e p o r t a l b / h o m e​ May 29. https://www.nature.com/articles​ .cfm?page=country_profile&CCode=IDN​ /s41467-018-04396-1. &ThisTab=NaturalHazards. Llewellyn, A. 2018. “Living in the Shadow of Mount Sinabung.” The Diplomat , March 2. https://thediplomat.com/2018/03/living​ References -in-the-shadow-of-mount-sinabung/. Muis, S., B. Güneralp, B. Jongman, J. Aerts, and AHA Centre. 2018. “The 2018 Lombok P. J. Ward. 2015. “Flood Risk and Adaptation Earthquake.” Situation Update No. 8, AHA Strategies under Climate Change and Urban Centre, East Jakarta, Indonesia. https://aha​ Expansion: A Probabilistic Analysis Using centre.org/wp-content/uploads/2018/08/AHA​ Global Data.” Science of the Total Environment -Situation_Update_no_8_M-7.0-Lombok​ 538: 445–57. -Earthquake-final.pdf. Nugraha, Panca. 2018. “More than 1,000 Houses Chaussard, E., F. Amelung, H. Abidin, and Inundated as Flash Floods Hit West Nusa S.-H.  Hong. 2013.  “Sinking Cities in Tenggara.” The Jakarta Post, March 6. http:// Indonesia: ALOS PALSAR Detects Rapid www.thejakartapost.com/news/2018/03/06​ 170  TIME TO ACT /more-than-1000-houses-inundated-as-flash​ World Bank. 2011. Indonesia: Advancing a National -floods-hit-west-nusa-tenggara.html. Disaster Risk Financing Strategy—Options for Soz, S. A., J. Kryspin-Watson, and Z. Stanton- Consideration. Washington, DC: World Bank. Geddes. 2016. The Role of Green Infrastructure ———. 2012. “Indonesia—Jakarta Urgent Flood Solutions in Urban Flood Risk Management. Mitigation Project (Jakarta Emergency Dredging Washington, DC: World Bank. Initiative).” Project Appraisal Document, Willner, S. N., A. Levermann, F. Zhao, and K. Frieler. December 22, World Bank, Washington, DC. 2018. “Adaptation Required to Preserve ———. 2013. Building Resilience: Integrating Future High-End River Flood Risk at Present Climate and Disaster Risk into Development— Levels.” Science Advances 4 (1). http://advances​ T h e Wo r l d B a n k G r o u p E x p e r i e n c e . .sciencemag.org/content/4/1/eaao1914.full. Washington, DC: World Bank. S P O T L I G H T   171 SPOTLIGHT 2 Urbanization for Human Capital The critical role of human capital in realiz- stunted) and full education (14 years of high- ing any country’s development potential is quality schooling by age 18) (Kraay 2018). well established: healthier, better educated Indonesia’s HCI score of 0.53 suggests societies are more productive societies. that, on average, the future earnings potential Indeed, globally, cross-country differences of Indonesian children born today is roughly in human capital explain between 10 and one-half of what it could be if education and 30  percent of differences in per capita health services were of a better quality. income (World Bank 2019). And, as the Indonesia’s overall score is an accurate reflec- frontier of skills continues to move rapidly tion of its human capital deficit accumulated outward and technology continues to drive after decades of underinvestment and ineffi- changes in production structures, human cient spending. capital will inevitably continue to grow in Indonesia’s public health expenditure importance. Human capital, therefore, (1.4 percent of gross domestic product) is must be continually developed through one of the lowest in the world. As a result, investments in nutrition, health care, qual- Indonesia’s population remains far from the ity education, and skill development. Such “frontier” of full health. Indonesia experi- investments in human capital will enable ences the second-highest tuberculosis preva- people to reach their potential as produc- lence in the world at 391 cases per 100,000 tive members of society, which, in turn, will people, and life expectancy at birth is among help to reduce inequality, improve welfare, the lowest in the region. Only 66.4 percent produce intergenerational benefits, and cre- of Indonesian children under the age of 5 are ate more inclusive societies. categorized as not stunted, compared with 73.5 percent for the average of its middle- income country peers. Even though Indonesia Indonesia’s low score on sets aside 20 percent of total public expendi- tures to education, one of the largest alloca- the human capital index tions in the world, it gets very little in return. The World Bank recently developed the On completing secondary school, 55 percent human capital index (HCI), a measure sum- of Indonesians are functionally illiterate, marizing the future productivity of a coun- compared with 14 percent in Vietnam, try’s citizens using prevailing education and which spends a similar amount on e ­ ducation. health outcomes. Ranging between 0 and 1, Without any improvements in the ­ current a country’s HCI score incorporates elements system, it will take about 50 years for of survival and a lack of stunting—and the Indonesian students to reach the  Organ­ quality of education and health. It captures isation for Economic Co-operation and how far a country is from the “frontier,” Development’s average score in math on the where the frontier corresponds to the situa- Programme for International Student tion that would exist if all children in the Assessment (PISA), and for poorer students country were to achieve full health (not it will take 136 years. Human capital across places data on the key components of the index (stunting, school enrollment, quality of The underlying framework of this report learning, and child and adult survival rates) considers the outcomes of prosperity, inclu- are available or can be constructed, the siveness, and livability as emerging out of an quality of data on some of the components interaction between the forces of agglomera- is not entirely certain. For example, cross- tion on the one side and the forces of conges- district differences in the quality of school- tion on the other. That framework is one ing or learning rely on average test scores way to think about how urbanization inter- on national exams at the lower- and upper- acts with human capital. Leaving aside secondary level. High levels of cheating human capital externalities for the moment, have been noted in these exams, and the chapter 3 presents evidence of significant government of Indonesia has been trying to positive agglomeration forces for metro curb cheating by implementing computer- cores, urban peripheries, and single-district based testing in recent years. These first-cut metro areas associated with population size estimates of district-level HCI scores have and density. been constructed and aggregated for These positive agglomeration forces con- Indonesia’s island-regions and its portfolio of tribute to the higher underlying productivity places, limiting attention only to test of these areas, which are also reflected in wel- scores  from schools that implemented fare, as chapter 4 shows. Presumably, higher computer-based testing in 2017 (figure S2.1). ­ productivity and living standards may then The results show quite significant varia- improve the HCI score through, for example, tion in HCI scores across Indonesia’s better survival and health outcomes. Positive island-regions. Jawa-Bali (0.60) stands out agglomeration forces may also lead to skill for having the highest HCI score, followed selective migration and sorting, which—in the by Kalimantan (0.56) and Sumatera (0.54), long run and through several intergenera- whereas Papua (0.48) and Maluku (0.46) tional transmission mechanisms—may lead to come out with the lowest HCI scores. higher human capital levels in more produc- Across the portfolio of places, urban tive parts of the country. Large fixed costs peripheries (0.61) and cores (0.60) of mul- associated with, say, the building of hospitals tidistrict metro areas have the highest HCI may also make certain health services more scores, and nonmetro rural areas (0.53) the economical to provide in urban areas than in lowest. rural areas. Although the differences across island- By contrast, the effect of congestion forces regions appear to be higher than the differ- could go in the other direction because the ences across the portfolio of places, the fact density of urban areas may impair perfor- that metro areas generally have higher HCI mance on the HCI through a variety of chan- scores than nonmetro areas suggests that, nels. For example, urban dwellers may face a for Indonesia, the positive effects of urban- higher probability of the spread of infectious ization on HCI performance trump the neg- diseases and be more likely to be exposed to ative effects. In addition, the differences in harmful air because of exposure to pollution HCI scores across metro and nonmetro associated with traffic congestion, which is areas likely understate the impact on future more prevalent in metro and urban areas (see productivity of a marginal increase in chapter 2). The effect of urbanization on human capital. The reason is that the HCI Indonesia’s HCI score is theoretically ambigu- methodology, which applies a uniform rate ous and thus ultimately an empirical of return on education across all places, question. does not adequately account for the human To investigate this question, we adopt capital externalities and higher social the  HCI methodology and apply it to returns to education in the metro areas Indonesia’s districts. Although district-level reported in chapters 3 and 4. 174  TIME TO ACT FIGURE S2.1  Average human capital index scores vary widely across Indonesia, 2017 a. Island-regions 0.60 0.60 Average human capital index score 0.56 0.55 0.54 0.51 0.50 0.50 0.48 0.46 0.45 Jawa-Bali Kalimantan Maluku Nusa Tenggara Papua Sulawesi Sumatera b. Portfolio of places 0.65 Average human capital index score 0.61 0.60 0.60 0.60 0.57 0.57 0.55 0.53 0.50 Metro core Urban periphery Rural periphery Single-district Nonmetro urban Nonmetro rural metro Sources: Calculations based on various data sources (notably National Exams test score data for 2017 and Indonesia Basic Health Research [RISKESDAS] 2013, National Socio-Economic Survey [SUSENAS] 2017, and Intercensal Population Survey [SUPAS] 2015). Note: These are preliminary results based on ongoing work and, as such, are subject to revisions. The national average human capital index (HCI) score reported in this spotlight is different from Indonesia’s official HCI score because of the differences in the underlying data used. The average value for each island-region or place is the population-weighted average of HCI scores for the districts that constitute the island-region or place. In panel b, places are defined using the methodology described in box 1.5, chapter 1. Urbanization policy •  Increasing access to safe drinking water and sanitation not only improves livability but sensitive to human capital also lowers the risk of childhood stunting. Greater investments in human capital to close •  Increasing access to good quality health HCI score gaps can help ensure that the ben- care, including maternal and child health efits of urbanization are widely shared. At the care, enhances the availability and use of same time, several policies to make urbaniza- critical inputs, such as vaccination and tion work better are also linked to human nutrition, to all children in their forma- capital: tive years, and helps reduce mortality and morbidity risks. •  Augmenting the quality of basic services •• Having schools adequately equipped with in all parts of the country, especially in age-appropriate learning materials and lagging areas, would boost the quality of well-trained teachers working under the the inputs into the human capital produc- right kinds of incentives determines the tion function. quality of education that gets delivered. S P O T L I G H T   175 Augmenting the quality of services is just as become inputs into the human capital important within the urban and urbanizing of children. places as it is across these places. Chapter 4 presents evidence on higher quality of learn- ing in the more urban parts of the country, References but it also shows that the gaps in learning Kraay, A. 2018. “Methodology for a World Bank among children from different socioeco- Human Capital Index.” Policy Research Working nomic backgrounds are widening in these Paper 8593, World Bank, Washington, DC. places. Even metro areas need to close not World Bank. 2019. World Development Report only the learning gap but also the gaps in 2019: The Changing Nature of Work . other key services that directly or indirectly Washington, DC: World Bank. 176  TIME TO ACT PA R T 2 How Can Urbanization in Indonesia Deliver More? T o help urbanization in Indonesia These same policy actions also provide nec- deliver better prosperity, inclusive- essary preconditions for successful policies ness, and livability outcomes, policy that connect, as well as for policies that suc- makers can follow the basic ACT principles cessfully target. Chapter 6 focuses on trans- of augment, connect, and target: port infrastructure investments that can help ­ better connect—and therefore integrate— •  Augment the coverage and quality of local the portfolio of places. It also includes some infrastructure and basic services for all discussion of policies that relate to barriers people in all places. to geographic labor mobility, the addressing •  C onnect to improve integration both of  which is likewise important to better across and within places. ­ connecting and integrating the portfolio of •• T arget any places and any people left places. Chapter 7 considers policies for behind. ­ better integration within metro and urban Part 2 of this report analyzes the main insti- areas, with a focus on affordable housing, tutional issues and policy areas in Indonesia urban transport policies, and the need for that are central to enacting these basic pol- more effective urban and spatial planning. icy principles. Chapter 5 diagnoses key Chapter  8 discusses targeted place-based issues in Indonesia’s system of urban and policies intended to boost the fortunes of subnational governance and financing that, lagging places, and additional urban plan- if addressed, can provide the foundations ning and design considerations important in for augmenting and equalizing the provision ensuring that basic services and the oppor- of local infrastructure and basic services tunities that urbanization brings are acces- across urban and subnational areas. sible to all. Urban Governance, Institutions, and Finance 5 KEY MESSAGES Indonesia’s institutions have made ­ p rogress •  Financing urban development requires in augmenting local infrastructure and the expanded opportunities and capacities for sub- coverage of basic services across places and national finance, results-oriented intergovern- subnational areas, but reforms are needed mental transfer mechanisms, and incentives to to improve the way metropolitan and urban mobilize private financing. Governments at all areas plan, manage, and finance infrastructure levels should ensure that, as urban population and services—to ensure their sustainable and grows, investments increase enough to main- equitable provision across workers and firms. tain per capita support at a level that sustains The suggested reforms focus on expanding infrastructure and basic services. opportunities for subnational financing; build- •  Strengthening subnational government plan- ing capacity for planning, implementing, and ning capacity requires addressing the shortage financing urban development; and improving of staff professionally qualified in urban plan- institutional coordination both vertically and ning and management and the lack of high- horizontally at all levels of government. quality local empirical analysis. Improving tertiary education, expanding relevant profes- •  Three complex issues challenge the ability of sional courses, and retaining highly qualified Indonesia’s institutions to ensure sustainable subnational government staff would help. and equitable financing for urban develop- •  Vertical and horizontal coordination in urban ment: (1) insufficient resources, (2) low capac- development requires collaborative platforms ity, and (3) coordination challenges. The foun- to remedy Indonesia’s fragmented system: dations for urbanization to deliver better development plans are disconnected from each prosperity, inclusiveness, and livability depend other, and governments at different levels and on addressing those issues, including through across jurisdictions do not work together to legal and regulatory reforms. deliver the required infrastructure and services. 179 Solid institutions, good governance, and sus- expanded local access to basic services and tainable financing are necessary to foster has enabled subnational governments to prosperity, inclusiveness, and livability across make more decisions, it has also complicated Indonesia’s portfolio of places. Many aspects the process of urban development. Indonesia’s of how governments plan, finance, and man- institutions are underperforming in their role age urban areas affect outcomes. They include to provide sustainable and equitable infra- transparency; capacity; incentives; clearly structure and services across the population, defined roles and responsibilities; multiple and reforms are needed that improve plan- sources of reliable financing; and effective ning, management, and finance to enhance coordination within levels of government, outcomes. The required reforms focus on across sectors, and across jurisdictional three main areas: boundaries, especially in multidistrict metro- 1. Expanding opportunities for city-level politan areas. financing Indonesia has made great strides in decen- 2. Building capacity for planning, implement- tralizing urban decision making to the local ing, and financing urban development level. Although services have improved across 3. Improving institutional coordination regions, however, overall prosperity for work- vertically and horizontally at all levels of ers and firms has lagged comparable coun- government. tries in Asia (see chapter 2). The economic underperformance of Indonesian metro and urban areas is often attributed to the public sector’s failure to mitigate the negative Urbanization in the impacts of urban agglomeration—congestion context of Indonesia’s pressures on infrastructure, basic services, land and housing markets, the environ- decentralization ment, and increased disaster risk—which, in Indonesia’s current system of urban gover- turn, challenges livability and inclusiveness. nance is characterized by decentralization and The weakness of the public sector in urban local democracy. The country adopted a new management is most evident in the lack of set of governing institutions in 1998, when capacity (Archer and Dodman 2015; Yoseph- the reformasi (reform) movement ended Paulus and Hindmarsh 2018), coordination 32 years of authoritarian rule and promised a failures (Sandee 2016; Sutriadi and Wulandari new era of democracy and decentralization. 2014), chronic underinvestment in urban Indonesia’s original 1945 constitution was infrastructure (World Bank 2017a), and dis- amended to curb executive powers, strengthen connects between spatial plans and invest- the legislature, adopt direct elections at all ment priorities (World Bank 2012b). levels of government, and enable a larger gov- This chapter explores the approaches to gov- erning role for subnational governments, ernance, institutions, and finance that affect the while firmly keeping Indonesia a unitary coverage of urban infrastructure and basic ser- country. vices for everyone in Indonesia. It provides the For the first time, the management of context for the current system of governance; Indonesia’s cities—including urban planning, highlights key bottlenecks affecting the way financing, and development—became the metro and urban areas are planned, managed, domain of subnational governments, not the and financed; and makes policy recommenda- central government. The premise of decentral- tions for the four types of place explored in ization was that bringing power closer to the part 1 (multidistrict metro areas, which are bro- people would improve democracy, community ken down into cores, urban peripheries, and participation, equality, and justice because of rural peripheries; single-district metro areas; non- the potential and diversity of Indonesia’s sub- metro urban areas; and nonmetro rural areas). national regions (see annex 5A).1 The chapter finds that, although decentral- Indonesia currently comprises 34 prov- ization in Indonesia has been associated with inces divided into 514 districts. Districts take 180  TIME TO ACT FIGURE 5.1  Indonesia’s decentralization has gone through three phases • Autonomy • Direct local • Increasing to districts elections role of 2. Local 3. Achieving 1. Big bang • Strong role • Mayors provinces democracy balance 2001–04 of mayors; 2005–13 must work 2014–present • Further weak role of with local regulatory provinces legislature refinements Source: Adapted from Ministry of Public Works and Housing 2017. the form of notionally urban municipalities FIGURE 5.2  Indonesia’s public governance has called kota (meaning “city”) or notionally improved since the start of decentralization but rural counties called kabupaten.2 Almost no remains far from perfect administrative distinction exists between kota and kabupaten and how they should be man- 60 aged.3 Since reformasi, the number of subna- 50 tional governments has increased from 26 provinces, 268 kabupaten, and 73 kota in Percentile rank 40 2000 to 34 provinces, 416 kabupaten, and 30 98 kota in 20184 (Ministry of Home Affairs 2014; Nasution 2016). 20 Since 2001, Indonesia’s decentralization 10 has gone through three phases (figure 5.1), a more detailed description of which can be 0 found in annex 5A.5 In the first phase, local 1996 2002 2005 2008 2011 2014 2017 autonomy applied primarily to the districts, Control of Political Rule of law Voice and with provinces having little authority. In the corruption stability accountability second phase, district leaders gained more political authority as they started to be Source: Calculations based on data from Worldwide Governance Indicators (http://info.worldbank.org​ /­governance/wgi/#home). directly elected by the people. The disparity Note: Percentile rank ranges from 0 (lowest) to 100 (highest) for all countries. between provincial and district governments started to be addressed in the third phase, financial and political crisis. The control of with provinces being given a larger role to corruption, political stability, and rule of law, oversee districts on behalf of the central gov- as well as voice and accountability at  the ernment. In general, decentralization has aggregate national level, have improved mark- occurred alongside an expansion of local edly since the start of decentralization, as access to public services (Lewis, McCulloch, shown by data from the Worldwide and Sacks 2016; World Bank 2017f). An Governance Indicators (­ figure 5.2).6 Despite urban–rural disparity remains, however: this progress, the percentile ranks of Indonesia’s urban districts (in metro and nonmetro areas) governance indicators in 2017 ranged from 29 tend to have higher access to services than do (for political stability) to 51 (for voice and rural districts (again, both in metro and accountability), still far from the highest rank ­ nonmetro areas), as highlighted in chapter 2. of 100. On other measures, Indonesia’s inter- Also, the effectiveness of increased intergov- national governance ranking has improved but ernmental transfers to low-population dis- remains far from satisfactory: 47th of tricts is questionable (see chapter 6). 137 countries on quality of institutions, accord- Governance has also improved since 2001, ing to The Global Competitiveness Report after falling to a low point due to the 1998 2017–2018 (Schwab and Sala-i-Martin 2017) U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   181 FIGURE 5.3  Subnational government financial rural districts consistently remained report audit scores have improved since 2009 the weakest. Across Indonesia’s six major island-regions, there was marked improve- a. Portfolio of places ment among districts in Kalimantan. 5.0 Meanwhile, districts in Jawa-Bali remained 4.5 the strongest and those in Maluku-Papua the weakest. 4.0 Under decentralization, many government Audit score functions shifted to the subnational level, 3.5 leading to higher spending by subnational 3.0 governments. In 2001–02, subnational gov- ernments were responsible for 26–35 percent 2.5 of total government spending, but in 2016, 2.0 they accounted for 45 percent (figure 5.4). 2009 2010 2011 2012 2013 2014 2015 2016 This share is higher than the average of Metro core Urban periphery Rural periphery 19.1  percent for 79  unitary countries and Single-district metro Nonmetro urban Nonmetro rural close to the average of 47.6 percent for 16 federal countries (OECD and UCLG 2016). b. Island-regions Although associated with improvements, 5.0 decentralization has increased administrative 4.5 overhead. The improvement in outcomes dis- 4.0 cussed in chapter 2 are promising but come with a cost (Lewis 2010; World Bank 2008). Audit score 3.5 Between 1994 and 2015, average access to 3.0 basic services increased by 1.5  times, but 2.5 subnational government spending per capita rose by 2.5 times (World Bank 2017f). 8 2.0 Much of the fiscal resources was spent on 1.5 staff salaries and administration, including 2009 2010 2011 2012 2013 2014 2015 2016 constructing new g ­ overnment offices (Lewis Jawa-Bali Sumatera Kalimantan and Oosterman 2011). Between 2010 and Sulawesi Nusa Tenggara Maluku-Papua 2014, personnel costs accounted for more Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia). than half of Indonesian district-level spend- Note: Audit opinions are based on a scale from 1 (adverse) to 5 (unqualified). Audit scores are ing (­figure 5.5),9 compared with the world- the unweighted averages across districts of a given type. In panel a, places are defined using the wide average of about 35 percent (OECD methodology described in box 1.5, chapter 1. and UCLG 2016).10 The gains in extent and quality of services are not commensurate. and 96th of 180 on the Corruption Perceptions Index.7 Consistent with progress at the national level, measures of governance at the subna- Why don’t Indonesian tional level have also improved. For exam- cities deliver more and ple, over the period 2009–16, audit opinions better urban services and of subnational government financial reports improved substantially, with an increasing infrastructure? number receiving the highest, “unqualified,” The ability of government to plan, manage, opinion (figure 5.3). There was no major and finance the delivery of adequate basic urban–rural difference in financial account- services and infrastructure in metro and ability improvement, but it seems that single- urban areas faces complex issues. Those district metro areas and rural peripheries issues are difficult to untangle but can be made the most progress, whereas nonmetro summarized in three key challenges: weak 182  TIME TO ACT FIGURE 5.4  Subnational government expenditures have risen as a share of total government expenditure after decentralization 1,800,000 50 45.1 47.6% 1,600,000 45 Average for Rupiah (billions), constant 2010 prices 41.9 16 federal 1,400,000 37.9 37.3 38.3 38.1 40 36.6 35.6 36.4 countries 35.3 35.6 34.7 33.7 33.6 35 1,200,000 31.6 30 1,000,000 26.1 Percent 25 800,000 20 19.1% 600,000 Average for 15 79 unitary 400,000 10 countries 200,000 5 0 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Subnational government Total public expenditure Subnational government expenditure expenditure (excluding transfers) (including subsidy and interest payments) as a share of total public expenditure (right axis) Sources: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, Subnational Finance Information System. Benchmark data against federal and unitary countries are from OECD and UCLG 2016. Note: Total public expenditure includes subsidy and interest payments to allow comparison with other countries. Previous World Bank publications on subnational spending used a different framing, where total public expenditure is expressed net of pensions and subsidies (World Bank 2017d, 2018d). coordination, low capacity, and limited FIGURE 5.5  Across the portfolio of places, financial resources. personnel account for more than half of subnational government spending, 2010–14 Governments face coordination challenges in urban planning, management, and finance 70 Share of subnational government spending, Decentralized decision making has made 60 55 55 58 55 56 54 urban planning and management a more average for 2010–14 (%) 50 inclusive affair, but it has also increased the complexity of coordinating across differ- 40 ent sectors, levels of government, and jurisdic- tional boundaries. Both vertical and 30 horizontal fragmentation create challenges in 20 integrating public services: (1) the vertical coordination of government agencies working 10 across different tiers or levels of hierarchy (how central government agencies work with 0 subnational governments); (2) the horizontal Metro Urban Rural Single- Nonmetro Nonmetro core periphery periphery district urban rural coordination of neighboring jurisdictions, metro especially within multidistrict metro areas; Personnel Goods and services Capital Other and (3) the horizontal coordination of national ministries and subnational depart- ments across sectors. These three coordination Source: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, Subnational Finance Information System. deficits hamper the provision of infrastructure Note: Shares are calculated as unweighted averages across districts of a given type of place. Places are and basic services in urban and metro areas. defined using the methodology described in box 1.5, chapter 1. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   183 Challenge 1: Inadequate vertical coordination and poor prioritization of subnational gov- ernment capital expenditure have led to dis- Decentralization provides more of a role for appointing increases in the number of homes subnational governments, but it also makes connected to pipes. The Ministry of Public vertical coordination more challenging Works and Housing plans, designs, and imple- because 32 types of activity are shared among ments water infrastructure development proj- the central, provincial, and district govern- ects in subnational government jurisdictions. ments, with each level playing a different role The ministry later hands over these assets to depending on the sector. Districts play the larg- subnational water utilities, but follow-up est role in education, health, and infrastruc- investment in distribution networks and con- ture, whereas the central government plays the nections may not occur, leading to increased largest role in general administration, social idle capacity of the utilities (World Bank protection, and housing and public facilities 2015b). (figure 5.6) (World Bank 2016). Sectoral Indonesia requires national standards to departments at the subnational level, however, assist in vertical coordination, but they need do not map hierarchically to the correspond- to be flexible enough to account for the ing ministries at the national level. geographical and cultural differences The division of responsibility across differ- between subnational regions, which, in ent levels of government within a sector does Indonesia’s case, can be significant. For not have to result in vertical fragmentation, example, central government standards for but in Indonesia, it has led to the classic coor- the self-help home improvement component dination problem in which everybody’s busi- within the National Affordable Housing ness is nobody’s business. Often no level of Program (World Bank 2017c) exclude government—central, provincial, or local— houses on a river that are built on stilts. In takes responsibility for vertical coordination many river-based cities such as Banjarmasin to ensure that activities within each sector (Kalimantan Selatan), however, settlements deliver the intended results. For example, in are traditionally constructed on rivers— urban water, lack of coordination between excluding them from the program would central, provincial, and district governments FIGURE 5.6  Districts account for about half of public spending in education, health, and infrastructure 100 Share of total government spending, 2014 (%) 27 25 33 60 54 48 57 8 14 16 40 19 8 18 67 59 20 51 35 33 28 0 Education Health Infrastructure General Social Housing and administration protection public facilities Central Province District Source: World Bank 2016. 184  TIME TO ACT leave out many households that would be Where the issues of solid waste, disaster risk ideal beneficiaries. There are also issues management, regional water services, and around the matter of land tenure that affect metropolitan transportation span many dis- inclusion in government programs. The tricts, they pose especially prominent coordi- challenge is to develop guidelines that are nation problems, as seen in the Jakarta metro firm enough to ensure results and flexible area (box 5.1). In light of an increasingly frag- enough to be adapted to different local mented metropolitan governance setting, contexts. finding mechanisms for coordination to pro- vide large-scale infrastructure is an important challenge for both existing and emerging mul- Challenge 2: Inadequate horizontal tidistrict metro areas. coordination among neighboring Limited mechanisms exist to facilitate the jurisdictions ability of one district to provide services that As urban areas grow, many increasingly benefit other districts’ citizens: cost-sharing encompass multiple districts. They need larger- arrangements between the respective districts, scale infrastructure and services like urban pricing of services to shift the cost burden to transport networks, sewage treatment, storm- the citizens of other districts, and specific pur- water systems, and solid waste treatment facil- pose funds that finance the specific infrastruc- ities. Providing these on a district-by-district ture and its recurrent operation, if cost basis is not economically efficient. For exam- recovery does not fully meet operating costs. ple, although the Medan multidistrict metro These mechanisms, however, remain little used area has five districts, it does not need five by districts. Considering such limitations, solid waste treatment plants. The challenge is strengthening the provincial government, as how to achieve these economies of scale mandated by the third phase of decentraliza- through cooperation. tion (see figure 5.1), promises to improve hori- For more than 10 years, decentralization zontal coordination. The latest version of the focused on districts, largely bypassing the decentralization law (Law 23/2014) stipulates provinces. The basic premise of Law 23 of that issues with cross-border implications fall 2014 on Regional Administration is that dis- within the authority of the higher-level gov- tricts are responsible for the delivery of the ernment agency, which means that the provin- same set of basic services falling wholly within cial government now has the authority to their boundaries. There is no legal provision coordinate metropolitan issues for multidis- for multidistrict metro areas as administrative trict metro areas within the province. Few pro- entities sandwiched between provinces and dis- vincial governments, however, use this tricts. The unsettled position of metropolitan authority because the implementing regula- management within the administrative system tions have not been updated and are silent on does not provide space for designing specific funding and accountability mechanisms. financial arrangements aimed at strengthening Where metropolitan coordination hap- metropolitan management. pens, it is cumbersome and usually ad hoc Provincial governments have an important and spontaneous. Some neighboring district role in the horizontal coordination of districts governments have developed metropolitan in a metropolitan area. As of 2016, Indonesia coordination mechanisms, often related to a has 21 multidistrict metro areas (see chap- specific issue or sector (table 5.1). Such ter 1), though many consist of only 2 districts mechanisms have had varying levels of suc- within a single province (Jakarta metro is an cess. Some cases where coordination has exception because it includes 14 districts worked are mentioned in the section on within three provinces). The combined opportunities and policy options later in this population of these multidistrict metro areas ­ chapter—but many do not work and remain has increased by almost 30 percent from on paper only, as the case of Jakarta demon- 67.2 ­ million in 2001 to 86.6 million in 2016. strates (box 5.1). U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   185 BOX 5.1  Cross-jurisdictional coordination in the Jakarta metropolitan area The Jakarta multidistrict metro area—which and often illegal development in the upstream straddles 14 districts in three provinces—is an water catchment areas of Kabupaten Bogor in oft-cited example of unresolved issues in gover- the south of Jakarta. This type of development nance (Firman 2009; Firman and Dharmapatni contributed to major floods in the capital in 1995; Firman and Fahmi 2017; Goldblum and February 2007, causing more than 70 deaths, Wong 2000; Rukmana 2015). Interjurisdictional displacing 340,000 residents, and generat- issues include urban sprawl, watershed manage- ing US$900 million in economic and financial ment, solid waste management, and mass public losses (World Bank 2008). Another major flood transportation. occurred in 2013, and recurring smaller but The metropolitan area has long faced chal- often debilitating incidents continue to harm lenges in coordinating services, despite the estab- Jakarta. lishment of a formal metropolitan coordination Metropolitan coordination issues arose more agency in 1976. For example, in 2001, heated recently in the planning and development of disagreements between the Jakarta provincial the Jakarta Mass Rapid Transit system. Despite government and Kabupaten Bekasi government the presence of about 7 million people in Kota resulted in closing Jakarta’s solid waste land- Depok and Kabupaten Bogor (districts that are fill (in Bekasi, West Jawa), leaving hundreds of part of Jakarta’s urban periphery in the prov- thousands of cubic meters of garbage uncol- ince of West Jawa), the last station and depot lected in the capital city for five days. of the transit line are located in Lebak Bulus Increased flooding risk in Jakarta is asso- in South Jakarta, deliberately within the metro ciated with the failure to prevent haphazard core. Challenge 3: Inadequate horizontal president or a coordinating minister, such as coordination across sectors the National Team for Acceleration of Poverty Reduction and the National Coordination Horizontal coordination across sectors Team for the Achievement of Sustainable occurs at both the national and subnational Development Goals. No such high-level coor- levels. Indonesia has numerous national-level dination mechanism exists for urban institutions—34 ministries and 27 nonmin- ­ development. isterial and nonstructural agencies 11 —­ Horizontal coordination for urban devel- understandably leading to intersectoral opment at the national level involves at least coordination challenges. None is specifically eight ministries and agencies, each in charge responsible for urban development. In con- of different sectors (table 5.2). An interminis- trast, many countries, including China and terial steering committee on urban develop- the United States, have a ministry of urban ment under the stewardship of the Ministry development (often along with housing), to of National Development Planning / National oversee sector planning, coordination, and Development Planning Agency (Bappenas) development. includes representatives from relevant minis- Harmonizing sectors is typically the job of tries and agencies.12 This steering committee the chief executive. In cities, this person is the is chaired by a deputy minister and reports mayor and, at the national level, the presi- to  the Bappenas minister, whereas other dent. Indonesia has several high-profile inter- members of the committee report to their ­ ministerial coordination mechanisms for respective ministers, hampering intersectoral topics that generated interest from the presi- coordination. dent, which are often led directly by the vice 186  TIME TO ACT TABLE 5.1  Types of metropolitan governance mechanisms in Indonesia Mode of cooperation Metro area Core city in Formal Joint Coordination (local name) metro area institution secretariat forum Sectors of cooperation Jabodetabek* DKI Jakarta √ Spatial planning, transportation, waste management, flood management Bandung Raya Bandung √ Governance, economy, physical development, social development, and culture Kedungsepur* Semarang √ Spatial planning, industry, trade, transportation, tourism, health, agriculture, culture and education, employment, social development, security Subosuko-Wonosraten* Surakarta √ Trade, tourism, economic development, river management Gerbang-Kertosusila* Surabaya √ Spatial planning, transportation, infrastructure, environment Malang Raya Malang √ Tourism, transportation, infrastructure Mebidangro* Medan √ Infrastructure Sarbagita* Denpasar √ Waste management, transportation, infrastructure Banjarbakula* Banjarmasin √ √ Infrastructure, spatial planning, education, health Mamminasata* Makassar √ Infrastructure Kartamantul Yogyakarta √ Infrastructure Source: Based on Ministry of Public Works and Housing 2017. Note: This table is not an exhaustive list. It includes areas defined by the government of Indonesia and this report, following the methodology outlined in box 1.5 in chapter 1, as multidistrict metro areas. Eight multidistrict metro areas defined in chapter 1 (Banda Aceh, Blitar, Bukittinggi, Magelang, Pasuruan, Pontianak, Probolinggo, and Sukabumi,) are not yet defined as metro areas by the government of Indonesia. Also, the government of Indonesia identified Mojokerto as part of Greater Surabaya (Gerbangkertasusila) and Salatiga as part of Greater Semarang (Kedungsepur). Unlike the definition of multidistrict metro areas used in this report, the government of Indonesia’s definition is not based on commuting flow data. * Identified by both the government of Indonesia and this report as a multidistrict metro area but with different extents. TABLE 5.2  Several plans related to urban development Planning system National agency in charge Key planning products National Development Planning Bappenas • Long-Term Development Plan (RPJPN/RPJPD) • Medium-Term Development Plan (RPJMN/RPJMD) Subnational Finance Ministry of Home Affairs • Annual Budget Plan (APBD) Human Settlements Planning Ministry of Public Works and Housing • Urban Design Guidelines Spatial Planning Ministry of Agrarian Affairs and Spatial Planning • Spatial Plan (RTRW) • Detailed Spatial Plan (RDTR) Coastal Area and Small Islands Planning Ministry of Maritime Affairs and Fisheries • Coastal and Small Islands Strategic Plan Environmental Management Ministry of Environment • Environmental Protection Plan Disaster Management National Agency for Disaster Management • Disaster Management Plan Sustainable Food and Agricultural Land Ministry of Agriculture • Sustainable Farm Land Plan Source: Based on Ministry of Public Works and Housing 2017. Note: Bappenas refers to the Ministry of National Development Planning / National Development Planning Agency. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   187 Similarly, at the subnational level, technical development that has already taken place and sectoral agencies report directly to the pro- (Rukmana 2015). vincial governor or district mayor, often with- Awareness among policy makers of the out a single agency formally appointed to importance of planning and the need for coordinate the different sectors locally. The coordination is growing, particularly after a presence of disparate planning systems leads to series of natural disasters (see spotlight 1 on fragmentation characterized by complexity, “Strengthening the Disaster Resilience of gaps, and overlaps (box 5.2). Conflict between Indonesian Cities”). The multitude of earth- the information provided by the different plan- quakes, tsunamis, and volcanic eruptions in ning systems results at best in piecemeal, incre- the mid-2000s produced a disaster manage- mental, and organic development and at worst ment planning system that requires coordina- in uncertainty and stagnation. tion. And the major flood that inundated Despite having an intricate planning Jakarta in 2007 (see box 5.1) accelerated the ­ s ystem, the norm for many decades in reform of the spatial planning law, mandating Indonesia has been incremental and haphaz- a 30 percent open space share in cities and ard development (Goldblum and Wong ambitiously criminalizing developments that 2000), where business interests with support deviate from the land use plan (Hudalah and from political actors develop new urban areas Woltjer 2007). The formation of a Ministry of without guidance from plans. The business- Agrarian Affairs and Spatial Planning in 2014 as-usual approach has updated statutory in theory combined spatial planning and land plans to incorporate actual and often illegal management functions, but the extent to BOX 5.2  Fragmentation among Indonesia’s planning systems Indonesia has an intricate planning system with hierarchical and jurisdictional levels, and integra- statutory plans mandated by law. Long-Term tion remains ad hoc and cursory. In many districts, and Medium-Term Development Plans, as well plans from one system, if available, are incompat- as spatial plans, are prepared at the national, ible with other plans. Although mechanisms exist provincial, and district levels. However, the in principle to ensure cross-­referencing of the var- system lacks a strategic spatial direction and ious plans, their use is not codified. Coordination linkage to a spatially directed capital invest- often relies on spontaneous effort, which is ment planning and budgeting framework. The unsupported by the political environment and local level Medium-Term Development Plan behavioral norms of working in silos. (RPJMD), with a five-year time horizon, is based Such fragmentation in urban planning has on the vision of the newly elected governor or often led to infrastructure investments at unsuit- mayor. The district-level spatial plan (RTRW), able sites instead of in priority areas. In many which provides the policy direction and strat- places, urban growth is occurring in areas with egy for land use plans, is valid for 20 years and high disaster risk, such as flooding and land sub- is revisited every 5 years. Regulations stipulate sidence (see also spotlight 1 on “Strengthening that the RPJMD and the RTRW should cross- the Disaster Resilience of Indonesian Cities”). reference each other to ensure alignment of the For example, in Kota Semarang, 29 percent of spatial plan with the activity plan. the city’s population lives in land subsidence The preparation of statutory plans lacks zones, which have been abandoned by high- cross-sectoral integration at the city level, cross-­ income populations but have seen an expansion jurisdictional coordination at the metropolitan of pockets of high poverty, exposing the urban level, and vertical coordination among levels of poor to environmental risks. government. Plans for different urban purposes (see table 5.2 in the main text) exist at different Source: Based on World Bank 2017e. 188  TIME TO ACT which such integration has been realized in 66.7 percent between 1999 and 2001 (World practice is unclear. Bank 2003). The total civil service expanded by  25  percent over 2006–17, from about 3.6 ­million to more than 4.5 million (World Capacity constraints hamper urban Bank 2018c). The growth creates opportuni- planning, management, and finance ties, but civil servants must be skilled, knowl- A civil service at any level of government edgeable, and effective at their jobs. needs a capable, motivated, and efficient staff The number of civil servants has grown and reliable and clear information systems to steadily across all types of districts and in all deliver a consistent quality of services to citi- island-regions (figure 5.7). Nonmetro urban zens. On aggregate, the civil service employed and nonmetro rural districts consistently had about 4.5 million people in 2015, equivalent the highest proportion of civil servants per to 1.7 percent of Indonesia’s population population, growing from about 1.3 percent (World Bank 2018c). This share is lower than the average proportion of general government staff to population among East Asia and FIGURE 5.7  Steady growth in the number of Pacific countries, which was 2.5 percent civil servants across places and island-regions, (World Bank 2016). At the subnational level, 2005–11 however, publicly available data on the civil service are lacking. a. Portfolio of places 2.5 National-level data reveal that local insti- Civil servants per population (%) tutional capacity is weak in Indonesia, with corruption ranked first and inefficient bureau- 2.0 cracy second as problems for doing business, according to the World Economic Forum’s 1.5 Global Competitiveness Report 2017–18 (Schwab and Sala-i-Martin 2017). Although 1.0 Indonesia has recorded notable improvements on some governance measures (see figure 5.2), it has a long way to go on many others 0.5 (OECD 2016). Cities suffer disproportion- 2005 2006 2007 2008 2009 2010 2011 ately from the impediments to doing business Metro core Urban periphery Rural periphery relative to rural areas because of substantially Single-district metro Nonmetro urban Nonmetro rural higher business densities. Corruption and b. Island-regions inefficiency, along with inadequate systems 3.5 to plan, manage, and finance urban develop- Civil servants per population (%) ment (such as information systems and 3.0 ­ p roperty tax management) remain major 2.5 bottlenecks. 2.0 Personnel capacity constraints in planning 1.5 and implementation 1.0 The transfer of administrative and fiscal 0.5 responsibility from central to subnational 2005 2006 2007 2008 2009 2010 2011 government under decentralization was car- Jawa-Bali Sumatera Kalimantan ried out quickly without transitional staff Sulawesi Nusa Tenggara Maluku-Papua training. The central government shifted many staff to subnational government pay- Source: Calculations based on data from the National Civil Service Agency. Note: Proportions are calculated as unweighted averages across districts of a given type of place and rolls, increasing the proportion of subna- in a given island-region. In panel a, places are defined using the methodology described in box 1.5, tional  civil servants from 12.2 percent to chapter 1. U r b a n G o v e r n a n c e , I n s t i t u t i o n s , a n d F i n a n c e   189 in  2005 to 2 percent in 2011. Meanwhile, confusion and sometimes fosters conflict. urban peripheries consistently had the lowest It also delays the preparation of many district- proportion of government staff per popula- level spatial plans and detailed spatial plans tion, growing from 0.8 to 1.2 percent in the and, ultimately, prevents the provision of same period. Among the different island- infrastructure and services. Partly because of regions, districts in Maluku-Papua have gained the lack of base maps, only 90 of 1,400 the most in civil servants per population, grow- detailed spatial plans for targeted priority ing from 1.8 to 3.2 percent between 2005 and areas nationwide have been developed and 2011, whereas districts in Jawa-Bali witnessed issued (discussed more in chapter 7). ­ modest growth from 0.8 to 1.2 percent. Uncertainty results for development and The quality of Indonesia’s civil service has investment. also improved because of more meritocratic The lack of empirical evidence, up-to- practices in hiring and promotion, especially date data for urban and spatial planning, after decentralization (World Bank 2018c). and defined priorities for infrastructure However, many subnational government investments also constrain inclusive urban staff remain ill-equipped to deal with issues development in Indonesia. Medium-term of  urban and regional development. plans (RPJMD) and long-term plans For example, most district-level public works (RPJPD) at the district level are rarely departments have too few professionally informed by analytics because of inadequate qualified staff and too many support workers data and a skills gap in district government. (World Bank 2012c). Many staff are also Many subnational plans lack high-quality unfamiliar with calculating the fiscal needs data input, with subcontracting to consul- for complex projects, let alone with how to tants often done without close supervision attract financing and private sector invest- and vision development by city officials. ments. Trainings are available but not enough, Regulatory frameworks for all plans man- and their quality is questionable. Often, when date urban growth analytics, economic subnational staff improve their capacity, they strategy, and integration of land use and are rotated to other departments for political transport planning, but the quality of these and administrative reasons, causing capacity inputs is low, and at times they are entirely gaps in the departments they leave behind. missing (World Bank 2017e). Systemic incapacity also hampers local taxation. Subnational governments under Weak information systems for decision decentralization are enabled and expected to making, urban planning, and management collect more taxes, but they are ill-equipped Limited government personnel skills are exac- with the capacity to manage and collect taxes. erbated by inadequate information systems They also suffer the consequences of the cen- for urban planning and development. tral government’s historically weak tax collec- Consistent and reliable base maps as well as tion prior to decentralization, which include up-to-date data on metro and urban areas are large uncollected taxes and tax registers with the backbone of good planning. Their absence missing, erroneous, and redundant data hurts the production of local spatial plans. (Haldenwang et al. 2015). At the district government level, lack of access Managing property taxes, which currently to reliable land records and spatial data inhib- provide the greatest potential for city tax rev- its infrastructure investments and service enue, presents many issues for subnational delivery and perpetuates the lack of clarity governments because they lack expertise, and transparency in decision making, spatial capacity, and cadastral information from the planning, and resource allocation (World land agency. This problem is particularly per- Bank 2018f). tinent for Indonesia’s multidistrict and single- The existence of different plans from dif- district metro areas because these areas have ferent ministries and agencies without a com- potential to collect far more in taxes than mon framework to integrate them all creates they currently do. Inability to raise tax 190  TIME TO ACT revenue, particularly in multidistrict metro on local leader performance, fails either to cores and single-district metro areas, con- illuminate strengths and weaknesses or to strains the provision of infrastructure and stimulate healthy competition to improve— basic services that are needed to accommo- partly because 93 percent of subnational date population growth, and in turn contrib- governments evaluated in 2015 received a utes to mounting congestion forces (see rating of high or very high (World Bank chapter 7). 2017f). Another measure of subnational govern- ment performance, the Ministry of State Challenges in regulatory capacity and Apparatus Empowerment and Bureaucratic weak monitoring Reform’s Bureaucracy Reform Index,17 uses Regulations to strengthen the monitoring of a more objective approach. In 2016, it gave a subnational government performance were grade of B or above to only 9 of 34 provin- established at the start of Indonesia’s reform cial governments evaluated (27 percent) and in 1999 but were ineffective. Since then, to 20 of 58 district governments (34 ­ percent) minimum service standards,13 specific laws (KemenPAN-RB 2016).18 But it is unclear and regulations on bureaucracy reform,14 how the results translate into incentives, and evaluations of subnational government sanctions, or capacity building (World Bank performance15 have been mobilized. As dis- 2017f). cussed above, public services have The capacity of local residents to partici- improved, but not to the level expected by pate meaningfully in monitoring the perfor- many and not commensurate with the mance of their governments is similarly resources spent (World Bank 2017d). crucial. Under decentralization, participatory Regulations on minimum service standards development planning is required through are complex, inflexible, difficult to monitor, bottom-up community consultations and often unenforced. For example, the (Brodjonegoro 2005), and local democracy quality of piped water is low, even in urban would not work effectively without the peo- areas, so households prefer groundwater ple’s feedback. However, residents are often sources (World Bank 2015b). Consequent inadequately represented in the consultation overuse leads to environmental externali- meetings and do not have the basic informa- ties, including land subsidence. tion required to participate effectively in these Delivering public services in a decentral- meetings or to analyze, pass judgment, or give ized setting requires different levels of gov- feedback on their leader’s performance. e r n m e n t t o w o r k t o g e t h e r. D i s t r i c t governments, however, are not politically accountable to provincial and central gov- Financial resources for cities are ernments, and a lack of accountability insufficient and often poorly targeted, with few opportunities for raising mechanisms has limited institutional per- revenue formance. This weakness of accountability mechanisms partly derives from a lack of To provide the foundations for delivering data that would enable the central and pro- prosperity, inclusiveness, and livability, cit- vincial governments, as well as concerned ies need reliable financing for urban man- citizens, to analyze the performance of dis- agement, basic services, and infrastructure trict governments. development. As discussed in more detail in Subnational government performance chapter 6, Indonesia’s urban transformation evaluations by the central government are is constrained by a large local infrastructure promising but often lack credibility and are deficit: infrastructure investment has not ineffective in pressuring subnational govern- caught up with pre-Asian financial crisis ments to deliver results. For example, the ­ l evels and lags well behind regional com- Ministry of Home Affairs’ evaluation of petitors. After falling off sharply because subnational governments,16 which focuses of the crisis, Indonesia’s infrastructure U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   191 investment has struggled to recover. Total Room for raising more own-source revenue infrastructure investment declined from an Political considerations have led Indonesia to average of 7 percent of gross domestic focus fiscal decentralization on the expendi- product (GDP) in 1995–97 to about 3–4 ture side. The “money follows function” prin- percent in 2011–13. By comparison, neigh- ciple stipulates that subnational governments boring countries such as Thailand, Vietnam, are mainly responsible for spending, whereas and China registered rates of approximately the central government generates most of the 7, 8, and 10 percent, respectively (World revenue. Although rising, own-source revenue Bank 2017b). contributes a small percentage of total subna- The infrastructure financing gap is espe- tional government revenue. Aggregated across cially high in large metro areas. A 2015 all districts, it made up 13 percent of total World Bank market assessment of 14 large subnational revenue in Indonesia in 2015 cities estimated an overall subnational infra- (OECD 2016), compared with 16 percent in structure investment financing gap of Thailand in 2009 (World Bank 2012a) and US$11.1 billion against borrowing capacity 22 percent in the Philippines, also in 2009 of US$1.7 billion (map 5.1) (Joshi et al. (Canuto and Liu 2013). 2015). In other words, the 14 cities face a Urban districts collected more own-source substantial investment gap and have insuffi- revenue than their rural counterparts as a cient borrowing capacity to afford even one proportion of total revenue (figure 5.8, major infrastructure project. panel a). For example, in metro cores and Subnational governments do not spend urban peripheries, respectively, own-source efficiently despite their shortages in infra- revenue contributed 27 and 29 percent of the structure investment (see also the discussion total district revenue in 2016, compared with in chapter 6). They face limitations with own- 12 and 8 percent in rural peripheries and source revenue, inequitable fiscal transfer for- nonmetro rural districts, respectively. Across mulas, and limited opportunities to raise their the island-regions (figure 5.8, panel b), own financing. MAP 5.1  For 14 subnational governments, investment needs for infrastructure projects far exceed borrowing capacity and revenue, 2014 Investment need gap Borrowing capacity Surabaya Revenue (excl. Salary, earmarked 2,954 and contingency fund) Batam 825 Pontianak 361 Banjarmasin Balikpapan Bangka 651 339 Makassar 449 860 Semarang 1,262 Gresik Sidoarjo Bogor 875 521 642 Lombok Surakarta Denpasar Barat 279 606 339 Source: Joshi et al. 2015. Note: Investment needs gaps are expressed in millions of U.S. dollars at 2014 prices. 192  TIME TO ACT districts in Jawa-Bali collected the highest FIGURE 5.8  Urban districts and those in proportion of own-source revenue, at 19 per- Jawa-Bali generated relatively more own-source cent in 2016, which was almost four times revenue than other types of place, 2014–16 that collected by districts in Maluku-Papua (about 5 percent). a. Portfolio of places Taxes contribute the largest share of 30 Indonesia’s state revenue, but the central gov- Own-source revenue as a share 25 ernment collects about 90 percent of it (World of total revenue (%) Bank 2016). Subnational governments have a 20 very limited tax base (Harjowiryono n.d.), which forms part of their own-source reve- 15 nue.19 For district governments, the most 10 prominent tax revenue sources are the prop- erty tax, property ownership transfer tax, 5 hotel tax, and restaurant tax. Because higher business densities raise 0 property prices, urban areas are better posi- Metro Urban Rural Single- Nonmetro Nonmetro core periphery periphery district urban rural tioned to capitalize on the property tax. In metro fact, they collect three to four times as much as rural districts (figure 5.9, panel a). Tax b. Island-regions revenue contributes 9–15 percent of the 25 total revenue in the metro cores, urban Own-source revenue as a share peripheries, and single-district metros, com- 20 of total revenue (%) pared to 1–5 percent in rural peripheries and nonmetro rural districts. In contrast, in 15 the United States, property tax contributed about 30 ­ percent of local government reve- 10 nue in 2014 (Urban Institute and Brookings Institution, n.d.). Nationwide, the property 5 tax makes up 0.6 percent of GDP in Indonesia, but 1.7 percent in China, 0 Jawa-Bali Sumatera Kalimantan Sulawesi Nusa Maluku- 1.4 ­percent in South Africa, and 1.3 percent Tenggara Papua in Brazil (Prakash 2013). Clearly, there is room for Indonesia’s cities to collect more 2014 2015 2016 property tax revenue. Some of the problems Source: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, cities face are low coverage of the cadaster, Subnational Finance Information System. low tax rate for urban areas, and low collec- Note: Proportions are calculated as unweighted averages across districts of a given type of place tion rates. (panel a) and island-region (panel b). In panel a, places are defined using the methodology described in box 1.5, chapter 1. The property tax, which was not decen- tralized until 2011, contributes the bulk of subnational governments’ own-source reve- Inequitable transfer formulas nue. Property tax revenue increased under district government collection compared with Intergovernmental transfers have gradually central government collection, for example, in increased since the start of decentralization Surabaya, the first district to take over prop- (see also chapter 6). They accounted for about erty tax collection in 2011, and in the 17 dis- 38 percent of the total state budget in 2016, up trict governments that took over tax collection from 14.9 percent in 2000. Most of the trans- in 2012 (Haldenwang et al. 2015). These fers aim to reduce fiscal inequality between the increases highlight the progress achieved central and subnational governments (vertical through decentralization of the property tax inequality) as well as across  subnational for cities. ­g overnments (horizontal ­i nequality). 20 U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   193 FIGURE 5.9  Own-source taxes make up a higher especially perverse incentives, have also been proportion of the total revenue in urban areas and in highlighted (Brodjonegoro 2007; Fengler and Jawa-Bali, 2014–16 Hofman 2008; Lewis and Smoke 2012, 2017; World Bank 2005). Empirical findings in a. Portfolio of places chapter 6 also challenge the success of 16 I n d o n e s i a ’s t r a n s f e r s y s t e m u n d e r decentralization. revenue, average for 2014–16 (%) Share of subnational government 14 12 Across place types and island-regions, the roles of the specific purpose fund and the 10 revenue-sharing fund pale in comparison to ­ 8 that of the general purpose fund. Rural dis- 6 tricts (nonmetro rural and rural peripheries) 4 tend to receive more general purpose funds as a portion of their total revenue (about 51–55 2 percent) than metro core and single-district 0 metro areas (39–46 percent) (figure 5.10, Metro Urban Rural Single- Nonmetro Nonmetro panel a). Across the island-regions, districts in core periphery periphery district urban rural metro Maluku, Nusa Tenggara and Sulawesi receive about 60 percent of their revenues from gen- b. Island-regions eral purpose transfers. On average, districts in 10 Kalimantan and Sumatera receive a higher portion of their revenues from the revenue- revenue, average for 2014–16 (%) Share of subnational government 8 sharing fund, compared to districts in other island-regions, at 20.3 and 8.5 percent, 6 respectively, largely because of natural resources (figure 5.10, panel b). 4 Subnational governments play an important role in basic service delivery, but about 87 per- 2 cent of their revenue depended on transfers from the central government in 2015. Few 0 transfers are linked to the delivery of results, Jawa-Bali Sumatera Kalimantan Sulawesi Nusa Maluku- leading to a lack of incentives for performance. Tenggara Papua Results-oriented transfers, where available, are 2014 2015 2016 small. The 2.6 percent of total grants which come from the regional incentive fund are Source: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, Subnational Finance Information System. insufficient to encourage subnational govern- Note: Proportions are calculated as unweighted averages across districts of a given type of place ments to adjust their behavior (World Bank (panel a) and island-region (panel b). In panel a, places are defined using the methodology described 2018b). in box 1.5, chapter 1. Additionally, because transfers were designed to distribute fiscal resources on a These equalization transfers take three forms: per district basis, not on a per capita basis, general purpose funds, specific purpose funds, they benefit areas with small populations to and revenue-sharing funds. In addition to the disadvantage of more densely populated these funds are the regional incentive funds, ones—multidistrict and single-district metro grants for specific programs, and direct trans- areas and nonmetro urban areas. The gen- fers to rural villages known as the village eral purpose, specific purpose, and revenue-­ fund21 (table 5.3). Plenty has been written sharing funds all follow this pattern about Indonesia’s transfer system, generally (­figure 5.11). Per capita transfer revenue is with a positive nuance (for example, it is asso- nearly 40 times higher in the most advan- ciated with public service improvements and taged district than in the least—dramatically equalization across regions), but challenges, above international norms—and the spread 194  TIME TO ACT TABLE 5.3  Indonesia’s intergovernmental transfers Type of transfer Objective Characteristics General purpose • Equalization of recurrent • Accounts for more than half of district revenues and 80 percent of total fund service delivery intergovernmental transfers (2015) • Can be used flexibly by subnational governments; amount varies according to number of civil servants; incentivizes staff hiring Revenue-sharing • Equalization of recurrent • Accounts for 11 percent of district revenues (2015) fund service delivery • Subnational governments with natural resources receive relatively large revenue- sharing fund transfers, incentivizing resource-rich areas to secede and form a distinct administrative entity Specific purpose • Equalization of ongoing • Accounts for 5 percent of subnational government revenues (2015) but 70 percent of fund capital needs their total capital spending (2016) • Addressing national priorities • Because of its small size, often limited to maintenance, rehabilitation, and • Addressing capital backlogs improvement of existing infrastructure Regional incentive • Other objectives • Accounts for 2.6 percent of total intergovernmental transfers and less than 4 percent fund of the general purpose fund • Serves as a “performance bonus” for receiving at least a “qualified” audit opinion of its financial report, passing of the local budget on time, and using the government e-procurement system Hibah (grant) for • Addressing national priorities • Amount varies according to national priority projects. specific programs Other transfers • Other objectives • Include political transfers to certain provinces, direct transfer to villages, and so on Sources: Based on interpretation of Law 33/2004 and data from the Directorate General of Fiscal Balance, Ministry of Finance. FIGURE 5.10  General purpose funds are important throughout all districts but particularly in rural districts and in Maluku-Papua, Nusa Tenggara, and Sulawesi, 2014–16 a. Portfolio of places b. Island-regions 70 70 Share of total revenue, aggregated for 2014–16 (%) Share of total revenue,aggregated for 2014–16 (%) 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Metro Urban Rural Single- Nonmetro Nonmetro Jawa- Sumatera Kalimantan Sulawesi Nusa Maluku- core periphery periphery district urban rural Bali Tenggara Papua metro General purpose fund Speci c purpose fund Revenue-sharing fund Source: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, Subnational Finance Information System. Note: Proportions are calculated as unweighted averages across districts of a given type of place (panel a) or island-region (panel b), in a given year. In panel a, places are defined using the methodology described in box 1.5, chapter 1. U r b a n G o v e r n a n c e , I n s t i t u t i o n s , a n d F i n a n c e   195 in general purpose fund transfers is more metro areas have a greater overall number of than 95 times (World Bank 2017d). poor people, at about 151,000 persons on Urban peripheries within multidistrict average (figure 5.12), and often have a larger metro areas have suffered the most from the population than the metro core. transfer mechanism. Although rural districts Because of their expanding populations have a higher percentage of poor people in (chapter 1), urban periphery districts typically their populations (13.8 and 12.5 percent in have high and increasing infrastructure costs, nonmetro rural areas and rural peripheries, but they receive the least transferred revenue respectively), urban peripheries in multidistrict per capita (figure 5.13, panel a). This low rev- enue undermines these districts’ ability to aug- FIGURE 5.11  More densely populated districts ment the coverage of local infrastructure, received less transfer revenue per capita than less housing, and basic services in line with their densely populated districts, 2016 population growth. Across the island-regions, districts in Jawa-Bali and Nusa Tenggara 19 received the lowest amount of total revenue Transfer revenue per capita (log) per capita, at 3.4  million and 5.9 million 18 rupiah (Rp), respectively (2016 prices), com- 17 pared to Rp 18.3 million and 9.3 million for districts in Maluku-Papua and Kalimantan, 16 figure 5.13, panel b). respectively (­ 15 Another key challenge is that all districts 14 are financed through the general purpose fund to deliver what is presumed to be an 13 identical set of services per district. Expecting 9 10 11 12 13 14 15 16 every district to provide the same set of basic Population (log) services, however, is not realistic. For exam- Sources: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia) and the Ministry of ple, many rural districts produce plenty of Finance, Directorate General of Fiscal Balance, Subnational Finance Information System. water but have few people, whereas the FIGURE 5.12  Rural districts had higher proportions of poor people, but urban peripheries had greater numbers of poor people, 2016 160,000 25 140,000 20 Percent of total population 120,000 Number of poor people 100,000 15 13.8 80,000 12.5 60,000 10 9.1 8.0 40,000 6.4 6.6 5 20,000 0 0 Metro core Urban Rural Single-district Nonmetro Nonmetro periphery periphery metro urban rural Number of poor people Poor people as a share of the total population (right axis) Sources: Calculations based on data from Badan Pusat Statistik’s (Statistics Indonesia) population census, intercensal population data, and forward-looking population projections. Note: Both the number of poor people and the poor people as a share of the total population are calculated as unweighted averages across districts of a given type of place. Places are defined using the methodology described in box 1.5, chapter 1.  196  TIME TO ACT opposite condition holds for many urban dis- FIGURE 5.13  Urban periphery districts and those in tricts. The uniform treatment of urban and Jawa-Bali received less transfer revenue per capita rural districts goes against one of the basic than other districts, 2016 principles of f ­iscal decentralization—that urban and rural areas should be treated dif- a. Portfolio of places ferently (Bahl 1999). 7 Rupiah (millions), current prices 6 Limited scope for cities to raise their own 5 financing 4 Indonesia has lagged in infrastructure devel- 3 opment since decentralization and faces an 2 estimated gap in infrastructure assets of 1 US$1.5 trillion, with public capital per person 0 only one-third of the average of major emerg- Metro Urban Rural Single- Nonmetro Nonmetro ing economies (World Bank 2017a). core periphery periphery district urban rural Collecting more revenue would help, but not metro enough to meet the need, and closing the gap b. Island-regions with other emerging economies requires 20 Rupiah (millions), current prices increased private investment. The government targeted about US$415 billion in additional 15 investments in transport, water, energy, and other key sectors for 2015–19, with an ambi- 10 tious target of having the private sector finance more than one-third of the amount 5 (World Bank 2017a). Although there are questions about how 0 realistic that target is (see chapter 6, box 6.2), Jawa-Bali Sumatera Kalimantan Sulawesi Nusa Maluku- Indonesia’s public financing framework is not Tenggara Papua designed to leverage private financing. Own-source revenue per capita Transfer revenue per capita Inverting best practice, the framework asks first which projects should be publicly funded, Source: Calculations based on data from the Ministry of Finance, Directorate General of Fiscal Balance, then which should be allocated to state- Subnational Finance Information System. Note: Proportions are calculated as unweighted averages across districts of a given type of place owned enterprises, and last whether to use (panel a) and island-region (panel b). In panel a, places are defined using the methodology described public–private partnerships (PPPs) (World in box 1.5, chapter 1. Bank 2018e). Government contracting authorities determine whether projects are decision has been made not to pursue state public or private during the five-year-plan budget or state-owned enterprise options. budgeting cycle at a very early stage of project The private sector faces four key challenges development, with little preparatory work or when looking to invest in infrastructure screening. Similarly, no clear criteria exist to (World Bank 2017a; see box 5.3 for a concise determine which projects should be competi- explanation): tively tendered and which should be assigned to state-owned enterprises. In practice, many 1. A complex legal landscape for PPPs projects that are viable for private finance are 2. A multitude of actors and nonstandard assigned to state-owned enterprises, partly processes for project identification, plan- because those enterprises are seen to deliver ning, and preparation projects faster and partly because they have a 3. State-owned enterprise dominance in dominant market position. Finally, PPPs are providing infrastructure brought in through a strong preparation and 4. A limited local debt and equity market competitive tendering process after the for long-term currency financing U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   197 BOX 5.3  Regulatory complexity in conducting public-private partnerships Aside from weak enforcement, complexity An estimated 158 national laws and regula- also burdens Indonesia’s regulatory environ- tions affect PPPs. Some overlap or are inconsis- ment for public-private partnerships (PPPs). tent, and the interplay between more general PPP For example, Indonesia has many laws, regu- laws and SSLs is often unclear. Compounding lations, and decrees governing PPPs, resulting this issue is the fact that the main PPP regulations in a legal framework that confuses investors. are generally lower in the legislative hierarchy No  ­o verarching law governs PPPs; rather, than most SSLs. Thus, when regulations conflict, multiple regulations control aspects of project ­ PPP projects are delayed until the relevant SSL preparation and procurement cycles. The PPP is amended or a special ruling is issued—or the legal framework comprises main regulations PPP is eventually cancelled. on PPPs, sector-specific laws (SSLs), and other PPP laws. Source: World Bank 2017a. State-owned enterprises have boosted untapped, and only the larger cities or prov- Indonesia’s infrastructure stock but may have inces with high fiscal capacity would be in a crowded out the private sector in delivering position to issue such bonds in the absence of urban infrastructure through preferential a mature municipal bond market. Meanwhile, access to finance and direct assignments. The because of government budgetary rules, private sector accounted for only 9 percent of subnational government budgets can be used ­ core infrastructure spending in 2015 (World only to pay for small-scale projects or mar- Bank 2017a). Poorly designed tariffs for cer- ginal improvements in basic services that tain utilities also dampen private sector inter- take less than one year to complete (World est, reduce incentives, and make projects Bank 2017b). commercially unviable. For example, a dis- Filling the “missing middle” is critical to trict’s water tariffs are decided by its mayor, meeting the local infrastructure gap. There are often based on political considerations. currently limited sources of project financing Consequently, average water tariffs paid by for subnational governments seeking to consumers are low (at US$0.28 per cubic undertake multiyear investments that are meter), which partly explains the limited ­ economically viable, although perhaps not interest of private financiers and operators financially viable, such as water supply, sani- (World Bank 2017a). tation, solid waste, drainage, affordable hous- Available financing instruments for infra- ing, and  urban transport projects (World structure development are laid out in Bank 2017b). At present, no financial institu- ­ figure 5.14, but they are limited and ill-fitting tion in Indonesia provides access to long-term for the nature and scale of the required invest- financing for local public infrastructure ment. There is a “missing middle” of instru- investment. Although no specific regulations ments for medium- to long-term infrastructure hinder commercial banks from investing in finance at the subnational level. In recent such projects, the banks typically focus on years, Indonesia has developed PPP vehicles short-term, corporate balance sheet secured for commercially viable infrastructure, but the financing. More than 85 percent of Indonesian market for PPPs is oriented toward large-scale bank deposits are in products of less than one revenue-generating projects. Regulations have month’s maturity. been amended recently to enable subnational Borrowing is a potential avenue to raise governments to issue bonds for urban infra- financing at the local level. The government’s structure, but municipal bonds remain ability to borrow, however, is limited by a 198  TIME TO ACT FIGURE 5.14  There is a “missing middle” of available instruments for subnational governments to finance infrastructure investment Public-private partnerships for commercial infrastructure Commercially viable, large infrastructure projects (toll roads, airports, and ports) Direct market borrowing Through municipal bond markets for creditworthy large cities and provinces to finance commercially viable projects Larger multiyear investments Small, medium, and large cities for projects with high “Missing middle” economic viability, funded with 5- to 20-year payback (water, sanitation, waste management, drainage, affordable housing, and urban transport) Local budget Small cities for basic services with up to 5-year payback period (internal city roads, parks, and street lights) Source: Adapted from World Bank 2017b. debt rule, under which total central and subnational government debt cannot exceed ­ Opportunities and policy 60 percent of GDP, and by a balanced budget options rule, limiting the consolidated central and Although Indonesia’s decentralization has subnational government budget deficit in any strengthened the ability of cities to plan, given year to 3 percent of GDP (Lledo et al. finance, and manage, considerable challenges 2017).22 Indonesia’s debt-to-GDP ratio was remain to realizing the full benefits of urban- 29.2 percent in 2017, and its budget deficit ization and creating prosperity, inclusiveness, was 2.57 percent of GDP. Because of the size and livability within and across cities. Several of central government debt, borrowing pow- policy priorities will open opportunities ers of subnational governments are strictly through enhanced institutional coordination; limited. Subnational government debt in expanded capacity for planning, manage- Indonesia, which was 0.04 percent of GDP ment, and finance; and better financing as of December 2018, is very low compared opportunities for cities. Implementation will with most countries. In 2013, subnational require bold action and a sustained commit- government debt accounted for 18.1 percent ment from the central and subnational of GDP in federal countries and 6.2 percent governments. in unitary countries (OECD and UCLG Table 5.4 lays out the policy options for 2016), suggesting much scope for additional addressing the key issues of coordination, borrowing at the subnational level in capacity, and finance, and proposed for the Indonesia. different types of places, as well as for provin- The government of Indonesia has imple- cial governments and the central government. mented initiatives to expand access to sub- national debt for infrastructure investment, with limited success. For example, the Enhance institutional coordination investment fund account, regional develop- Institutional coordination would benefit from ment account, and subsidiary loan agree- the connection and integration of urban plan- ment have experienced significant arrears ning and management processes vertically (World Bank 2017b). across levels of government and horizontally U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   199 TABLE 5.4  Policy options to address coordination challenges, build capacity, and expand opportunities for financing Key to table Time horizon S = Short term (next 2 years) M = Medium term (next 2–5 years) L = Long term (> 5 years) “¸” in a cell indicates that a recommendation applies to the type of place “↔” in a row indicates that a recommendation applies to all types of place Leading institution(s)/Champion(s) ATR/BPN Ministry of Agrarian Affairs and Spatial Planning Bappenas Ministry of National Development Planning / National Development Planning Agency BIG Geospatial Information Agency BPK National Audit Agency Core Multidistrict metro core MCIT Ministry of Communication and Information Technology MOF Ministry of Finance MOHA Ministry of Home Affairs MPWH Ministry of Public Works and Housing MSOE Ministry of State-Owned Enterprises SDM Single-district metro area Multidistrict metro Nonmetro Periphery Single- district Leading institution(s)/ Time BROAD RECOMMENDATION Core Urban Rural metro Urban Rural Champion(s) horizon 1. Enhance institutional coordination Interjurisdictional coordination: Engage actively in ¸ ¸ ¸       Core; Provincial S interdistrict cooperation to deliver complementary services governments; MOHA Interjurisdictional coordination: Take anticipatory       ¸ ¸ ¸ SDM; Provincial S coordination measures with surrounding or nearby districts governments; MOHA Vertical coordination: Work with province government to Provincial governments; S–M improve vertical alignment across levels of government Sectoral ministries Intersectoral coordination: Integrate local sectoral plans ATR/BPN; Bappenas S–M into a common data and mapping platform Intersectoral coordination: Establish a national-level Bappenas; Coordinating M platform for urban transformation led by the president Ministry for Economic Affairs 2. Build capacity for urban planning, management, and finance Personnel: Build capacity for planning, geospatial analysis, MOHA; Sectoral ministries; S–M project management, and public finance Universities; Professional associations Table continued on next page 200  TIME TO ACT TABLE 5.4 Continued Multidistrict metro Nonmetro Periphery Single- district Leading institution(s) / Time BROAD RECOMMENDATION Core Urban Rural metro Urban Rural Champion(s) horizon Monitoring: Establish a system that enables subnational MOHA; MCIT; Sectoral M governments to report on their outputs and benchmark ministries their performance Information system: Strengthen common data and ¸ ¸ ¸ ¸ ATR/BPN; M–L mapping platforms for capital investment and property BIG; MOF tax management Information system: Strengthen common data and ¸ ¸ ATR/BPN; M–L mapping platforms for development control (monitoring BIG; MPWH of spatial plan) Monitoring: Enable local internal audit officers to play a BPK M–L larger monitoring role 3. Expand opportunities for city-level financing Own-source revenue: Support vertical development, ¸ ¸ ¸ ¸ ATR/BPN; S–M more collection of property taxes, and use of real estate MOF; MSOE instruments Transfers: Keep supporting through transfers with ¸ ¸ ¸ ¸ MOHA; MOF; Sectoral M increasing portion of results-based transfers ministries Own-source revenue: Allow to keep a larger portion of ¸ ¸ ¸ ¸ MOHA; MOF M–L income tax Borrowing: Increase debt financing from both public and ¸ ¸   ¸ ¸   MOF M–L private sources Private financing: Continue regulatory reforms to ¸ ¸ ¸ ¸ ¸ MOF; Sectoral ministries M–L encourage and support public–private partnerships Transfers: Change transfer formula to place more weight MOHA; MOF L on population across sectors. Interjurisdictional coordina- programs should be indicated in the long- tion would also benefit from the identification term (RPJPN) or medium-term (RPJMN) of different roles for each type of place and development plans at the national level, level of government. and they should be synchronized with the corresponding plans at the subnational level, every five years at the beginning of a Improving vertical coordination newly elected leader’s governing term. Coordination with the central government This is a role for Bappenas. enables subnational governments to increase •  Clarify the extent of role for each level of their capacity and investment. To enhance government (central, provincial, district) vertical coordination, the central government in delivering infrastructure and services. could do the following: Without a joined-up approach, even increased expenditure may fail to improve •  Improve alignment of national programs service delivery (World Bank 2015b). The for large infrastructure with subnational extent of the roles may differ across sec- and local priorities. Large-scale national tors and perhaps also across provinces. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   201 Ministries in charge of different sectors infrastructure, the central government could should clarify this. consider the following: •  Provide clear standards, guidelines, and ref- •  Encourage greater contracting between erences to guide the implementation of subnational governments to deliver com- national directives by subnational govern- plementary services. The central and pro- ments. At the same time, the central gov- vincial governments could share template ernment should be flexible in ­standards to contracts for different types of services allow regional variations according to local with district governments, as well as guid- conditions and preferences. ance on what to look out for when enter- •• Empower provincial governments to play ing into such arrangements. Publicizing a more prominent role in ensuring vertical and promoting good metro governance coordination between the central and dis- practices would build citizen awareness of trict governments. Dealing directly with the need for governments to work together. more than 500 district governments could •  Operationalize coordinating bodies be overwhelming, so provincial govern- (as stipulated by Law 23) in which partici- ments, as the extension of the central gov- pating member district governments plan ernment’s presence in the regions, should and implement projects and reach consen- actively synchronize programs ­ vertically. sus on issues to be managed at a metro- Such empowerment of provincial govern- politan scale. The central government ments should be the role of the Ministry of could develop implementing regulations Home Affairs. and provide technical assistance to prov- The importance of aligning national, provin- inces and districts to establish such bodies. cial, and district-level programs can be seen, •  Promote the establishment of multisubna- for example, in the development of a light rail tional government enterprises for specific transit system in Palembang, a priority of the purposes (such as solid waste disposal or central government. Adopting the light rail transport). These enterprises would have system implies adopting transit-oriented the constituent district governments (and development, because the city would increase provincial governments, where relevant) density along transit stops. That result on their boards. Effectiveness, however, requires revising the city’s spatial plans and will depend on a new regulation to enable perhaps even the city’s medium-term and authorize such enterprises. Financing (RPJMD) and long-term (RPJPD) develop- could come through a mix of user fees, ment plans. The light rail system also stretches contributions from member governments, beyond the boundaries of Kota Palembang, precept powers, central transfers, and making it a multidistrict metropolitan issue. borrowing. The importance of standards and guidelines •  Elevate metropolitan management func- can also be seen in the growing number of tions to provincial governments. This smart city projects at the district level: districts would require a change in the overall are spending public resources without clear decentralization hierarchy and Law 23. public service standards from the central gov- It would also require specific transfers to ernment. An electronic and Internet-based the provincial governments that take on reporting platform would allow information such functions. Finance could be achieved to flow faster across government levels (see through the voluntary agreement of dis- spotlight 4 on “The Potential of Smart Cities”). tricts to allocate some part of their fund- ing to provinces, in accord with the func- tions taken on. Improving interjurisdictional coordination •• Use specific purpose funds to support coor- Interjurisdictional coordination offers links dination across district boundaries on necessary for enhanced urban management. infrastructure. A multiyear specific purpose To enhance coordination for metropolitan fund for infrastructure could be developed 202  TIME TO ACT BOX 5.4  Some models of metropolitan management International experience suggests that regional providing multiple services to the residents of the governmental authorities and other mechanisms metro area. Some examples include can help coordinate urban service provision •  The Community of Madrid and the Grand across district boundaries. No single mechanism Paris Region; guaranteed to work in all sectors and in all is ­ •  The urban wards within the Tokyo prefecture, places. The success of a particular mechanism which deliver services on behalf of the Tokyo also depends on local and national institutional Metropolitan Government, one of many play- factors. ers involved in regional governance, along Among the most popular form of metro- with prefectures, regional ministerial offices, politan governance is a single-sector or limited-­ Japan Railway, and private companies. subject metropolitan agency established by national or provincial laws and regulations. Sometimes a new institution to govern a met- Some examples include ropolitan area is created by aggregating and dissolving the lower-tier governments that previ- •  The Greater London Authority, a form of ously existed. Some examples include metropolitan governance with power mostly over transport and police functions; •  The Toronto metropolitan government, •  The metropolitan areas of Sydney, Australia, established in 1998 and Lima-Callao, Santiago de Chile, and •  The new, enlarged municipal areas of Mon- Bogotá in South America, as well as metropoli- treal and Quebec, created by the province of tan transit agencies in several metropolitan Quebec in 2000 specific areas in the United States, providing ­ •  The consolidation of various local governments services to those metro areas; and into a single “unicity” by Cape Town in 2000 •  The Seoul Metropolitan Government, run by •  The Bangkok Metropolitan Administration, a mayor and an elected assembly, encompass- created by merging Bangkok and Thonburi. ing 25 districts. Associations of local governments also have greater Sources: World Bank (2015a), based on Bahl, Linn, and Wetzel (2013); Sellers and powers than their constituent municipalities, Hoffmann-Martinot (2008). that is apportioned across jurisdictions for from a rural district. Also, looking forward, a single project (which might comprise a horizontal urban development in the form of coordinated set of subprojects). sprawl and development of a new center in a nearby rural district may easily change a Across the portfolio of places, interjurisdic- ­ single-district metro into a multidistrict metro. tional coordination mechanisms should espe- Indonesia does have some working prac- cially be encouraged for multidistrict metro tices of interjurisdictional coordination. For areas. They would apply to the metro core, as example, three districts in the Yogyakarta well as to the urban and rural peripheries of metropolitan area created a joint secretariat multidistrict metro areas, and the role of facili- for such interjurisdictional issues as solid tating interdistrict coordination would apply waste management, the water catchment area, to the provincial government. Even for single- and regional transportation (Hudalah, Fahmi, district metro areas, interjurisdictional coordi- and Firman 2013). The government of Jawa nation with surrounding districts may be Timur province established a PPP to deliver necessary because of cross-border issues such drinking water from Umbulan in Kabupaten as urban water, which is likely to be sourced Pasuruan to five nearby districts including U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   203 those in the Surabaya metro (Soekarwo On the basis of other existing platforms, the 2018). And three districts in the Malang met- most typical structure seems to be either the ropolitan area are establishing cooperation in vice president or a coordinating minister as the tourism sector by synchronizing their spa- the chair, reporting directly to the president, tial plans (Jatim Times 2018). These programs with the members to include all relevant min- represent good practices within Indonesia, istries such as the Ministry of Finance, but they are more the exception than the Ministry of Home Affairs, Ministry of norm. Several good practices from other Agrarian Affairs and Spatial Planning, countries are in box 5.4. Ministry of Public  Works  and Housing, Ministry of Transportation, Ministry of Communications and Information, and the Improving intersectoral coordination National Geospatial Information Agency. Intersectoral coordination at the national Bappenas could have a unique and critical level requires an integrated approach to urban role as the technical secretariat and hub of the transformation. Given the importance of the platform. Such a technical secretariat would urbanization agenda for Indonesia and how need to be supported, giving it sufficient future prosperity, inclusiveness, and livability capacity to fulfill its mandate. In addition, hinge on sustainable urbanization, a clear because one of the key roles of this platform opportunity exists now to establish a national will be to provide implementation support to platform for urban transformation that will give direction to and coordinate the national elevate a holistic urban development agenda sectoral investment programs (such as afford- to the highest level and ensure that challenges able housing, urban water supply, solid waste, are adequately addressed. disaster risk management, and so on), a clear The proposed national platform on urban role at the forefront for the coordinating transformation will enhance the urban agenda Ministry of Economy will be important to in several important ways: ensure sustainability. At the subnational level, the district or •  Elevate: By deriving authority from a pres- province secretariat and the subnational idential decree, the national platform will development planning agency should work gain significance, which will in turn enable together to ensure intersectoral coordination the various facets of its agenda. around certain themes. This coordination •  Empower: It will help cities tackle urban could be achieved through intersectoral and challenges and service delivery through multistakeholder forums, as well as by estab- better urban and spatial planning, coher- lishing a common data and mapping platform ent capital budgeting, and fiscal manage- to integrate various sectoral programs. The ment (including achieving creditworthiness latter will be discussed more in the next that will allow cities to leverage m ­ ultiple section. sources of finance for investments). •  Integrate: As a holistic platform that enhances vertical and horizontal coordina- Building capacity for urban planning, tion, it will integrate several sectoral agendas management, and finance and programs, breaking down silos, and in Decentralization needs capacity for planning, turn enhance coordinated planning. coordination, service delivery, and financing •• Reform: It will enable a shift in the core at the subnational level. District governments approach to urban management in Indone- should be in the driver’s seat of local develop- sia by enabling cities to be the primary driv- ment, but capacity gaps are evident at the per- ers of their futures through robust decision sonnel level as well as in the information making, with the central government as a systems related to planning, management, and facilitator and enabler. monitoring of development outputs. This sec- Implementing the national urban platform tion provides policy options to build those requires an enabling organizational structure. capacities. 204  TIME TO ACT Building and sustaining subnational retain the specific capacities that they have governments’ personnel capacity invested in staff. Staff rotations for adminis- trative reasons are a normal aspect of large Spatial data infrastructure, a key aspect of organizations, but the rate of these rotations planning, requires the capacity to undertake in Indonesia should be managed to enable geospatial data collection, sharing, use, and subnational governments to capitalize on the analysis.23 At the personnel level, capacity capacities of staff and enable transfer of building is necessary especially for subna- capacity to others. tional governments. For this purpose, the cen- tral government could do the following: •  Establish well-designed, hands-on, on-the- Developing an integrated data framework job training programs to benefit data qual- and platform for cities aligned with ity, production, and management in urban implementation and capital investment planning. This training would ultimately boost cities’ capacity to produce, manage, The central government has recognized over and analyze high-quality data and to the past decade the importance of improving undertake evidence-based urban planning data and information quality for development (World Bank 2017e). The trainings could and spatial planning (World Bank 2017e). To be done in partnership with the respective this end, provincial governments. •  The central government should provide •  Improve geospatial education through consistent, digitized, and spatially enabled both professional courses and tertiary base maps for spatial planning and analy- education. Although tertiary education sis. These maps would equip cities with an provides a holistic understanding of geo- integrated platform to manage spatial data spatial technologies, professional courses and integrate them with other planning can enable a wider audience of midcareer data (World Bank 2017e). Ensuring the professionals in city governments to share presence of such an integrated data and use geospatial information. This edu- and mapping platform should be the role cation could be presented in partnership of the Ministry of Agrarian Affairs and with prominent universities at the national ­ S patial Planning and of the National level as well as universities in different ­Geospatial Information Agency. provinces. •  Districts should integrate currently dispa- •• Provide training in tax collection and rate statutory plans into a common spatial management, including the management development framework, which is an of tax registers and cadastral information improved form and approach to the cur- related to property tax, to subnational rent district-level spatial plans, or RTRWs. governments. Such a framework, or “RTRW 2.0” for Results-oriented transfers were mentioned short, would integrate the RTRW and earlier as an effective tool to incentivize sub- development plan, as well as other related national governments. But use of this tool plans, to provide an integrated, long-term should be supported by well-designed capac- spatial interpretation of the city’s develop- ity-building grants so that governments can ment vision. address weaknesses identified by the perfor- •  To realize this vision, districts need to mance system and avoid being left behind. identify strategic capital investments spa- The distinct but smaller capacity-building tially, prioritize them, and monitor their grants should be widely accessible so that all implementation (World Bank 2017e). subnational governments can improve their Linking spatial development plans to the performance (World Bank 2018b). process of capital investment is critical to Long-term progress requires institutional moving from a general wish list of invest- continuity, and cities should be allowed to ment priorities to implementing the plans. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   205 At the same time, investment priorities in peripheries of multidistrict metro areas, single- urban areas should be differentiated from district metro areas, and nonmetro urban priorities in rural areas, with the intention areas. Ministries are mandated to build the of discouraging sprawl into the rural areas capacity of provinces on their respective and providing more living, working, and ­ sectoral issues, but provinces would ultimately recreational space in urban areas. finance the capacity building of districts. •  For property tax administration, priority goes to metro cores, urban peripheries, Strengthening regulatory capacity and ­ single-district metro areas, and nonmetro monitoring urban areas because they have the most Improved regulatory capacity and capacity potential to generate property tax revenue. for monitoring development outputs at both The relevant government agencies should the central and subnational levels will enhance consider fit-for-purpose over high-accuracy the performance and accountability of district approaches. For example, technologies governments. To enable this, the central gov- such as satellite imagery, unmanned aerial ernment could do the following: vehicles, and automated methods to detect cadastral boundaries could be used to pre- •  Set up an accountability system requiring pare effective cadasters at a more reason- subnational governments to report to the able cost. Kigali, Rwanda, used satellite central government on their performance. imagery to create cadasters for improved This system should be in the form of a property tax collection (Ali, Deininger, and web-based system to standardize reporting Wild 2018). Meanwhile, Albania, Kosovo, by subnational governments. and Ho Chi Minh City in Vietnam used •  Verify the reports using an independent unmanned aerial vehicles to achieve 5–10 body such as the government’s auditor or centimeter square spatial resolution. the relevant line ministry, as in the Local •• The central government should support Government and Decentralization project districts in building and enhancing their (box 5.5). current tax information and management •• Align the incentives of local leaders to systems and improve trainings to local tax deliver better services and provide a basis officials on tax collection strategy, espe- for central government intervention, by cially for property tax. conditioning fiscal transfers on the verifi- cation of such performance reporting. The availability of stronger data and informa- tion foundations needs to be matched with Indonesia’s experience with the Local greater institutional capacity for data gover- Government and Decentralization project nance at the subnational level. The municipal shows the effectiveness of involving the spatial data infrastructure—consisting of national audit agency to verify infrastructure institutional arrangements, people, data, and outputs. However, scaling up such a program systems—is currently being developed, tested, nationwide would require a slightly different and refined under the World Bank’s City approach, considering the number of subna- Planning Labs (World Bank 2017e). tional government units in Indonesia. Each Preparing statutory plans and developing a subnational government’s internal audit offi- robust urban planning system will require cial would need to be empowered to verify substantial investment in hardware, software, project outputs according to technical specifi- and staff capacities. Web connectivity, web cations. To ensure the independence of the mapping services, geospatial software, and auditing process, the national audit agency maintenance and development are essential to could do random checks on the internal audi- support the daily operations of spatial data tors of subnational governments. Local infrastructure. Considering the value of these citizens also need to be empowered to moni- ­ systems for cities and urban areas, implemen- tor the performance of their leaders. The cen- tation should start with the cores and urban tral government could require subnational 206  TIME TO ACT BOX 5.5  Improving reporting and accountability through the Local Government and Decentralization project The Local Government and Decentralization payments were linked to use of the systems for project was a World Bank–funded loan proj- reporting. ect worth US$770 million that ran from 2011 To improve accountability, the improved to 2017 in Indonesia. Before the project, no reports submitted by subnational governments reporting or independent verification process are subject to verification by an independent was mandated for subnational government body. The central government’s independent use of specific purpose funds or compliance internal auditor verifies reports on the outputs of completed work with technical guidelines. of specific purpose fund infrastructure proj- Governments only submitted financial reports ects. The auditor was supported with training to the Ministry of Finance stating that the funds on verifying technical infrastructure and on the had been spent, and no technical reporting was Ministry of Public Works and Housing technical required. Because financial reports were sub- guidelines on roads, water, sanitation, and irri- mitted manually, sharing them among relevant gation. Handbooks on output verification were ministries was delayed. To improve transpar- prepared, providing easily accessible infrastruc- ency in the use of specific purpose funds, ture specifications and photographs for compar- improved reporting and thereby accountability ison (see World Bank 2015c). were necessary. Afterward, and pending the auditor’s verifica- To improve reporting, the project provided tion, 10 percent of the specific purpose fund’s technical assistance to establish and develop a value is transferred to the subnational gov- web-based reporting system. The system stan- ernment as an incentive reimbursement for its dardized specific purpose fund reporting for 10 percent match of the grant, which the central infrastructure projects by subnational gov- government requires. ernments to the Ministry of Finance and the Ministry of Public Works and Housing. Incentive Source: World Bank 2018a. governments to report their fiscal and perfor- private financing is underdeveloped. Lessons mance information in an appropriate format from other countries point to opportunities for laypeople, such as through simple info- for expanding Indonesia’s urban development graphics published in the local media. financing. This section provides policy options Ultimately, the central government should to expand opportunities for city-level make use of the audit results to provide local financing. leaders with comparative information about their local government’s performance in Incentivizing cities to increase tax revenue relation to that of other subnational govern- ­ ments to enable benchmarking and spur com- Most subnational government tax bases, such petition (World Bank 2017f). as high-value residential and office buildings, are in metro and urban areas (Nasution 2016). Tax revenue is a more important reve- Expand opportunities for city-level financing nue source in metro and urban areas than in Critical infrastructure and basic service needs rural areas (see figure 5.9); therefore, large in cities require adequate, transparent, and urban areas, rather than rural districts, may predictable financing. The current multiple be more realistically expected to increase their sources of revenue are insufficient, and own-source revenue. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   207 Differentiating tax revenue targets among Making smarter use of transfers to leverage different types of places could work. Urban improved results districts in metropolitan areas—multidistrict Depending on how they are managed, trans- metro cores, their urban peripheries, and fers could be an important driver of improved single-district metro areas—should be encour- ­ subnational government performance. aged and incentivized to generate a certain Transfers can drive flows of performance percentage of their revenue from local taxes. information to the central government (which Incentives could take the form of awards, per- can “turn off the tap” if subnational govern- formance bonuses, and central government ments do not provide reports). Local govern- grants for infrastructure development. ments can be required to undertake upstream Nonmetro urban districts should be encour- planning as to how they will use specific pur- aged to generate more local tax revenue than pose transfers, which provides some assur- nonmetro rural districts, though perhaps not ance of improved quality of spending. as much as the (core and periphery) urban Performance reporting also provides an districts in metro areas. opportunity to identify strong and weak per- However, political considerations, more formance, thereby enabling the rewarding of than capacity restrictions, often prevent sub- strong performance and the targeting of weak national governments from realizing their tax performers for more intensive support. potential. For example, a regulation setting higher tax rates carries a high political cost because it affects large numbers of taxpayers, Reforming general purpose funds including powerful groups, and is highly vis- Increases in general purpose funds at the mar- ible. In contrast, improving tax registers gin are used mostly to increase spending on through data cleaning has low political cost, subnational government staff salaries and so it is more likely to be used by subnational allowances. An additional rupiah in a general governments (Haldenwang et al. 2015). purpose fund increases personnel spending by Good political communication with the peo- Rp 0.86 (World Bank 2014). Additional ple helped Kota Surabaya, which is the core increases may not necessarily bring positive district of metro Surabaya, increase tax reve- impacts, and the fund can be made more nue from Rp 1.97 trillion to Rp 2.27 trillion effective by reforming the basic allocation (2010 constant prices) between 2014 and formula. 2016.24 The development of a land cadaster, The current formula for fiscal needs, which recommended primarily to provide base data is the basis for determining the size of trans- for spatial planning, also permits better man- fers, creates unequal capital transfers with, all agement of the property tax and boosts else being equal, larger per capita transfers collection (World Bank 2017e). ­ going to less populous districts. This result Much work is needed to build the was highlighted in figure 5.11 and will be fur- revenue-generating capacity of subnational ­ ther discussed in chapter 6. All else being governments. To avoid an overreliance on equal, the formula assumes that a district’s property tax, they also need more capacity absolute expenditure needs increase less than for exploring and drawing on alternative proportionally with its population. sources of own-source revenue through According to Law 33/2004, the fiscal need transparent and sustainable urban develop- of a given district i is based on the formula ment practices. Land value capture and below, where TE is the total expenditure of all transfer of development rights are instru- districts in Indonesia, Popi is the population ments subnational governments should of district i, Areai is the land area, HDIi is the explore for raising funds for urban infra- human development index, and IKKi (Indeks structure development. Such instruments Kemahalan Konstruksi) is a price inflation require, in turn, familiarity with the tools of index in district i. Meanwhile, n is the total real estate development. number of districts in Indonesia, and AvgPop, 208  TIME TO ACT AvgArea, AvgHDI, and AvgIKK are the aver- Strengthening specific purpose funds age population, land area, human develop- Increases in specific purpose funds are highly ment index, and price inflation index, likely to stimulate subnational government respectively, of all districts. The weights (that capital spending. An extra rupiah in a specific is, the a terms) on the cost adjustment vari- purpose fund increases capital spending by ables sum to one and are chosen with a view more than Rp 1.5 (World Bank 2014). The to maximizing equalization. Because of this, added capital spending increases local public a1 < 1 and so the fiscal needs of a district are capital assets for health, education, and infra- assumed to increase less than proportionally structure. Increased public assets are, in turn, with its population. strongly associated with improved service delivery. In general, therefore, more funding TE  Popi should be allocated for specific purpose funds Fiscal   Needsi =     ×   α1   than for other types of transfer. For infra- n  AvgPop structure development, incentives for good Areai HDI i IKKi  specific purpose fund spending can follow an + α2 + α 3 + α 4 . AvgArea AvgHDI AvgIKK   output-based disbursement approach (see box 5.5), which would benefit from improved reporting and accountability. Despite the assumption that fiscal needs are The specific purpose fund can be improved largely independent of population, the costs of by reducing the sectoral and geographic cov- district services—such as schools, health facili- erage, consistent with its original intention as ties, water supply, and ­sanitation—are driven a specific purpose grant. The fund expanded mostly by the number of clients. from 3 sectors in 2001 to 19 in 2013, and, A better formula for calculating fiscal although initially allocations were made only needs, and one that conforms more closely to to a small subset of subnational governments, international good practice, would be one now all subnational governments receive at that assumes that, all else being equal, needs least some funding (World Bank 2014). The increase in line with population. To maintain government should focus the fund on priority a redistributive element, the formula could regions and sectors. Moreover, considering keep the human development index in assess- the complexity and size requirement of many ing each district’s fiscal needs. The suggested infrastructure projects, multiyear project revised transfer formula is implementation should be allowed. TE Fiscal   Needsi =     ×   Popi TotalPop Increasing the regional incentive fund and other results-oriented transfers  Areai HDI i IKKi  ×  β1   +   β2 +   β3 .  AvgArea AvgHDI AvgIKK   The regional incentive fund links performance evaluation explicitly to financial rewards. Although the amount of the reward seems too Changing this formula, however, entails a small to be a meaningful incentive, it appears revision of the current Law 33/2004 on fiscal to be correlated with better financial manage- transfers. Regardless of the transfer formula, ment and governance. Since the fund was the general purpose fund remains important, introduced, a growing number of subnational especially in the rural areas (nonmetro rural governments have qualified for rewards. In districts and the rural peripheries of multidis- 2018, 18 of 34 provinces, 235 of 415 kabu- trict metro areas), where it is difficult to gen- paten, and 60 of 93 kota received the regional erate any substantial amount of revenue, incentive fund reward, compared with except perhaps from natural resources, where 54 subnational governments in 2010 (World available. Bank 2018b). U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   209 The regional incentive fund can be relevant subnational governments have suffi- improved in the following manner (World cient capacity to plan, implement, and man- Bank 2018b): age the mass transit system by themselves. Increasing debt financing from both public •  Scale the grant amount to district size and private sources through, for example, district population or revenue figures, so the grant provides a Raising funds through the issuing of munici- meaningful incentive for all subnational pal bonds is also a viable option. For this governments. option, subnational governments need the •  Establish a credible annual performance capacity to prepare proper business plans for assessment to ensure verification of results. the development projects to be financed •  Simplify the indicators and improve trans- through bonds. parency by publishing the guidelines and Borrowing is an underused financing the full annual results. instrument at the subnational level. Because •• Set up a capacity-building system to ensure of the tight controls imposed by the debt rule a level playing field. and the balanced budget rule, subnational government debt in Indonesia is very low at The hibah grant to subnational governments only 0.04 percent of GDP in December 2018; also adopts principles of results-based however, many subnational governments ­ payment, which Indonesia could scale up have considerable borrowing capacity for urban infrastructure programs. For exam- because of growing transfer revenues. The ple, the water hibah uses an output-based 30 largest cities in Indonesia are estimated to grant linked to infrastructure investments. have an untapped borrowing capacity of Participating subnational governments are approximately US$3.8 billion, on the basis of reimbursed for water connections to poor Indonesian norms (World Bank 2017b). households, with connections independently The limited systemic capacity of the verified by a third party. Between 2014 and domestic banking sector needs to be 2016, almost Rp 1.4 trillion from both donor addressed. The small, highly concentrated and government sources were reimbursed, banking sector does not facilitate the develop- and more than 600,000 new household water ment of suitable infrastructure finance instru- connections were made. An important feature ments. The dominance of short (three to five of the hibah is the division of labor between year) tenure and the absence of limited- the line ministry and the Ministry of Finance. recourse financing make it difficult to invest The line ministry, responsible for the technical in infrastructure. Meanwhile, the bond mar- specifications of the project, supervises it and ket is underused (World Bank 2017a). ensures that the achievement of outputs is Recent regulatory changes provide oppor- verified, which then triggers disbursement by tunities to increase subnational government the Ministry of Finance (World Bank 2018a). borrowing.25 PT SMI, a state-owned enter- The National Urban Transport Program prise of the Ministry of Finance, is tasked being prepared by the World Bank and the to provide medium-term debt financing to government of Indonesia also uses a hibah subnational governments, supported with a mechanism, including a hybrid PPP combin- guarantee fund from an intercept of the sub- ing viability gap funding and availability pay- national government’s future general purpose ment. Under this model, the Ministry of fund and revenue-sharing fund revenue. The Finance supports a special purpose vehicle Regional Infrastructure Development Project through viability gap funding, and line minis- provides funding to PT SMI as well as a proj- tries support the city through availability pay- ect development facility to enlarge access to ment. The financial support created under infrastructure finance at the subnational level this program would continue until the (World Bank 2017b). 210  TIME TO ACT Improving project screening for PPPs contracting agencies about which projects are most appropriate for PPPs, disseminating Efforts to improve the legal and regulatory those criteria to build capacity at the line environment for PPPs are under way. ministry or subnational government level, Presidential regulation 38/2015 introduced and implementing systems for hiring profes- key changes to PPP implementation rules in sional advisors to structure projects to attract Indonesia, such as introducing a direct the private sector (World Bank 2017a). appointment mechanism to expand the types This chapter has discussed the many of infrastructure that can be developed achievements in Indonesia’s efforts to aug- through PPP schemes and expanding the ment the provision of services across places types of investment return schemes that can and subnational areas. However, it has also be adopted in infrastructure projects (World highlighted the need for several key reforms Bank 2017a). related to the way metropolitan and urban Despite progress, Indonesia still struggles areas plan, manage, and finance infrastruc- with choosing the optimal route of delivery ture and services. Those reforms—expanding based on project characteristics and central opportunities for subnational financing; government objectives. Adjustments in the building capacity for planning, implementing, project identification and project preparation and financing urban development; and stages could enhance the government’s ability improving institutional coordination both to bring PPP projects to market. Bappenas vertically and horizontally at all levels of and the Ministry of Finance’s PPP unit are government—will require significant political ­ best positioned to assist government con- will and strong leadership to achieve results. tracting agencies with training on screening This deepening of reforms ultimately has the projects, preparing outlines of business cases, potential to have much positive impact in and making initial appraisals so that the achieving goals to promote prosperity, inclu- pipeline develops stronger projects. Further siveness, and livability for citizens across steps include specifying clear criteria for Indonesia. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   211 Annex 5A Evolution of decentralization Indonesia’s decentralization was deep, fast, Aside from the six absolute affairs that and vast—described as a “big bang” (Hofman remain in the hands of the central government, et al. 2004). In 1999, acting urgently to pre- 32 “concurrent affairs” became the joint vent some parts of the country from breaking responsibility of the central, provincial, and off, Indonesia passed two laws that devolved local governments,27 including basic ­ services substantial autonomy and money to subna- such as health, education, social affairs, hous- tional governments.26 Decentralization was ing and settlements, safety and public order, deep because the authority being devolved and public works and spatial planning. covered almost everything except six “abso- Theoretically, decentralization gives Indonesia’s lute affairs”—foreign affairs, defense, justice, cities the autonomy to chart their own devel- finance, religion, and management of natural opment, including attracting residents and resources—which remain in the hands of the businesses with each city’s comparative and central government. It was fast because the competitive advantage. It also lets cities deliver relatively basic 1999 laws were fully imple- the public services that residents want in ways mented in just over a year. And it was vast that they prefer (Bunnell et al. 2013). because implementation occurred at the same Decentralization has been dynamic, pass- time at the beginning of 2001 in all subna- ing through three major phases in 15 years: tional governments throughout the country. the big bang, the adoption of local democracy, With decentralization, Indonesia shifted and the current effort to achieve balanced the responsibility for nearly all public services roles between the provincial and district gov- to the subnational level, with a strong focus ernments (Ministry of Public Works and on the districts and largely bypassing the Housing 2017). provinces. This focus can be advantageous for In the first phase, the big bang, much auton- urban governance, which is by nature local. omy was given directly to district governments Politically, subnational governments are (kota and kabupaten), whereas provinces held accountable to the local population through a weak role (Nasution 2016; World Bank elections, whereas, administratively, they are 2005). Local leaders had access to power and accountable to the central government money with little structure for accountability, through financial reporting. which led to substantial corruption. Often, those who wanted to be local leaders lobbied the parliament to designate their territories as FIGURE 5A.1  Subnational governments new subnational government units by way of proliferated after decentralization secession, leading to a proliferation of subna- tional governments (figure 5A.1) (Fitrani, 600 Hofman, and Kaiser 2005).28 Onset of The second phase was the spread of direct Number of kota and kabupaten 500 decentralization 98 local democracy. Following a successful and 98 400 peaceful direct presidential election in 2004, governments 91 subnational governments started to conduct 300 73 direct elections of chief executives—mayors 54 55 55 62 200 399 413 and governors—and, since 2005, members of 349 the local council. Local democracy also 246 246 241 243 268 100 allowed new leaders to rise and compete for 0 local power. 1980 1985 1990 1995 2000 2005 2010 2013 The third and ongoing phase seeks to balance the roles of the different levels of Kabupaten Kota government.29 With the central government Source: Calculations based on Nasution 2016. struggling to oversee 508 autonomous dis- Note: A kabupaten refers to a rural county; a kota refers to a city. tricts30 throughout the archipelago, the role 212  TIME TO ACT of the provincial governments has been 8. Five indicators of access to locally provided strengthened. Provinces, extensions of the basic services are used: net enrollment rate central government, now have the authority for junior high school, net enrollment rate to coordinate districts, which can be partic- for senior high school, access to an improved water source, access to improved sanitation ularly important for heavily populated facilities, and proportion of births attended urban districts that make up a metropolitan by a skilled health worker. The average of region (see chapter 1). The Jakarta metro- these five indicators is used to produce a sim- politan region (Jabodetabek) is the only ple aggregate measure. metro region to straddle multiple provinces, 9. The published data are at a very aggregated requiring coordination by the national level and do not allow analysis beyond that government. provided in this report. 10. Based on 2013 data covering 83 countries. 11. Based on Ministry of State Apparatus and Bureaucracy Reform (https://www.menpan​ Notes .go.id/). 12. Bappenas Regulation 60/2017. 1. These principles are stated at the beginning 1 3. For example, Government Regulation of the first postreform law on decentraliza- 65/2005 on minimum service standards. tion (Law 22/1999). 14. For example, a 2010 Presidential Regulation 2. In this chapter, “subnational government” Grand Design of Bureaucracy Reform for 2010– refers to both provincial and district gov- 25 aims to achieve a clean government free of ernments; “district government” refers corruption, collusion, and nepotism; improved specifically to kota and kabupaten govern- quality of public services; and increased public ments. Sometimes “local government” is sector performance and accountability. used to refer to international cases or specific 15. For example, Government Regulation 6/2008 projects. on evaluation of regional governments. 3. The smallest administrative units in Indonesia 16. Evaluasi Kinerja Penyelenggaraan Pemerintahan are notionally rural units called desa (villages) Daerah (EKPPI). and notionally urban units called kelurahan. 1 7. Sistem Akuntabilitas Kinerja Instansi A desa is autonomous (the head is directly Pemerintah (SAKIP). elected by residents), and a kelurahan is not 18. The possible evaluation results are AA, A, (the head is decided by the head of the dis- BB, B, CC, C, and D, with AA being the best trict). A kabupaten tends to have more desa, result and D the worst. whereas a kota tends to have more kelura- 19. Law 28/2009 describes the taxes allocated to han. This means that a kota is likely to have subnational governments. more governing authority than a kabupaten 2 0. Transfers are regulated through Law over its jurisdiction. 33/2004 on Fiscal Balance Between Central 4. Decentralization occurs at the district level and Subnational Government. except for Daerah Khusus Ibukota (DKI) 21. Direct transfers to rural villages are a rela- Jakarta, which retains autonomy at the tively new transfer mechanism (established province level. Aside from five kota and one in 2015) which are used to support develop- kabupaten in DKI Jakarta, the remainder ment in rural areas. They are not discussed (508 districts) of Indonesia’s 514 districts are here because they are less relevant for urban autonomous. development. 5. The original Law 22/1999 on decentraliza- 22. The debt rule and balanced budget rule are tion has been revised as Law 32/2004 and stated in Law 17/2003. more recently as Law 23/2014. 23. A significant milestone was the enactment of 6. See the Worldwide Governance Indicators the Geospatial Information Law (4/2011), (database), World Bank, Washington, DC which paved the way for key improvements (accessed September 30, 2018), http://info​ in mapping and geospatial databases. Most .worldbank.org/governance/wgi/#home. recently, Presidential Decree 27/2014 on 7. For more information on the Corruption Per- Spatial Data Infrastructure mandates local ceptions Index, see https://www.transparency​ governments to develop municipal infra- .org/news/feature/corruption_perceptions​ structure to promote spatial data integration _index_2017 for better planning. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   213 24. According to data from the Directorate International Studies Program, School of Policy General of Fiscal Balance’s Subnational Studies, Georgia State University, Atlanta. Finance Information System. Bahl, R. W., J. F. Linn, and D. L. Wetzel 2013. 25. These changes include Minister of Finance Financing Metropolitan Governments in Regulations 174/2016 on guarantees to PT Developing Countries. Cambridge, MA: Lincoln SMI, in its task to provide subnational infra- Institute of Land Policy. https://www.lincolninst​ structure financing; 117/2017 on cumulative .edu/publications/books/financing​-metropolitan​ maximum deficit limit of the subnational -governments-developing​-countries. budget; and 125/2017 on management of the Brodjonegoro, B. 2005. “Participatory Budgeting guarantee funds. in Decentralized Indonesia: What Do Local 26. The original decentralization laws were Law People Expect?” In Participatory Planning and 22/1999 on subnational governance and Law Budgeting at the Sub-National Level, 36. New 25/1999 on fiscal balance between the central York: United Nations Department of Economic government and subnational governments. and Social Affairs. The laws have been amended several times; ———. 2007. “Intergovernmental Fiscal Relations the latest version of the decentralization law and State Building: The Case of Indonesia.” is Law 23/2014, and the fiscal balance law is In  Fiscal Fragmentation in Decentralized Law 33/2004. The latter has been under revi- Countries: Subsidiarity, Solidarity and sion since 2008. Asymmetry, edited by R. M. Bird and R. D. Ebel. 27. The annex to Law 23/2014 presents the Cheltenham, U.K.; Northampton, MA: Edward full functional assignment among the Elgar Publishing. https://www.e-elgar.com/shop​ central, province, and kabupaten/kota /eep/preview/book/isbn/9781781007617/. governments. Bunnell, T., M. A. Miller, N. A. Phelps, and 28. Despite the splitting, Indonesia’s districts are J.  Taylor. 2013. “Urban Development in a on average the most populous subnational Decentralized Indonesia: Two Success Stories?” governments among 85 countries, with about Pacific Affairs 86 (4): 857–76. https://doi.org/ 500,000 people per district. The average info:doi/10.5509/2013864857. country has about 56,000 people per district Canuto, O. and L. Liu, eds. 2013. Until Debt Do (OECD and UCLG 2016). Us Part: Subnational Debt, Insolvency, and 29. This current phase of Indonesia’s decentral- Markets. Washington, DC: World Bank. ization is guided by Law 23/2014 on subna- Fengler, W. and B. Hofman. 2008. “Managing tional governance. I n d o n e s i a ’s R a p i d D e c e n t r a l i z a t i o n : 30. Indonesia has 514 districts, but five kota and A c h i e v e m e n t s a n d C h a l l e n g e s .” I n one kabupaten within the Daerah Khusus Decentralization Policies in Asian Development, Ibukota Jakarta province are not autonomous. edited by S. Ichimura. Singapore: World Scientific. Firman, T. 2009. “Decentralization Reform and References Local-Government Proliferation in Indonesia: Towards a Fragmentation of Regional Ali, D. A., K. W. Deininger, and M. Wild. 2018. Development.” Review of Urban & Regional “Using Satellite Imagery to Revolutionize Development Studies 21 (2–3): 143–57. https:// Creation of Tax Maps and Local Revenue doi.org/10.1111/j.1467-940X.2010.00165.x. Collection.” Policy Research Working Firman, T., and I. A. I. Dharmapatni. 1995. Paper 8437, World Bank, Washington, DC. “The Emergence of Extended Metropolitan http://documents.worldbank.org/curated​ Regions in Indonesia: Jabotabek and Bandung /­e n/347231526042692012/Using-satellite​ Metropolitan Area.” Review of Urban & -­i magery​-to-revolutionize-creation-of-tax​ Regional Development Studies 7 (2): 167–88. -maps-and-local-revenue-collection. Firman, T. and F. Z. Fahmi. 2017. “The Archer, D., and D. Dodman. 2015. “Making Privatization of Metropolitan Jakarta’s Capacity Building Critical: Power and Justice (Jabodetabek) Urban Fringes: The Early Stages in Building Urban Climate Resilience in of ‘Post-Suburbanization’ in Indonesia.” Journal Indonesia and Thailand.” Urban Climate 14 of the American Planning Association 83 (1): (1): 68–78. https://doi​ . org/10.1016/j.uclim​ 68–79. https://doi.org/10.1080/01944363.201 .2015.06.007. 6.1249010. Bahl, R. W. 1999. “Implementation Rules for Fitrani, F., B. Hofman, and K. Kaiser. 2005. “Unity Fiscal Decentralization.” Working Paper 99-1, in Diversity? The Creation of New Local 214  TIME TO ACT Governments in a Decentralising Indonesia.” of State Apparatus (Empowerment) and Bulletin of Indonesian Economic Studies 41 (1): Bureaucratic Reformation.”) Ministry of State 57–79. Apparatus and Buraucracy Reform. Goldblum, C., and T.-C. Wong. 2000. “Growth, Lewis, B. D. 2010. “Indonesian Decentralization: Crisis and Spatial Change: A Study of Accountability Deferred.” International Journal Haphazard Urbanisation in Jakarta, Indonesia.” of Public Administration 33 (12–13): 648–57. Land Use Policy 17 (1): 29–37. https://doi​ Lewis, B. D., N. McCulloch, and A. Sacks. 2016. .org/10.1016/S0264-8377(99)00043-5. “Measuring Local Government Service Haldenwang, C. von, A. Elfert, T. Engelmann, Delivery Performance: Challenges and (Partial) S. Germain, G. Sahler, and A. S. Ferreira. 2015. Solutions in Indonesia.” Journal of International The Devolution of the Land and Building Development 28 (5): 808–17. https://doi​ Tax in Indonesia. Bonn: Deutsches Institut für .org/10.1002/jid.3106. Entwicklungspolitik. Lewis, B. D., and A. Oosterman. 2011. “Sub- Harjowiryono, M. n.d. “Design and Management National Government Capital Spending in of Decentralized Revenues and Transfers: The Indonesia: Level, Structure, and Financing.” Experiences of Indonesia.” Ministry of Finance, Public Administration and Development 31 (3): Republic of Indonesia. 149–58. https://doi.org/10.1002/pad.582. Hofman, B., K. Kaiser, J. Alm, J. Martinez- Lewis, B. D., and P. Smoke. 2012. “Incentives Vazquez, and S. M. Indrawati. 2004. “The for Local Service Delivery: International Making of the ‘Big Bang’ and Its Aftermath: A Experience and Options for Indonesia.” In Political Economy Perspective.” In Reforming Fiscal Decentralization in Indonesia a Decade Intergovernmental Fiscal Relations and The after Big Bang. Ministry of Finance, Republic Rebuilding of Indonesia: The “Big Bang” of Indonesia, Directorate General of Fiscal Program and Its Economic Consequences , Balance. 15–46. Cheltenham, UK; Northampton, MA: ———. 2017. “Intergovernmental Fiscal Transfers Edward Elgar Publishing. and Local Incentives and Responses: The Case Hudalah, D., and J. Woltjer. 2007. “Spatial of Indonesia.” Fiscal Studies 38 (1): 111–39. Planning System in Transitional Indonesia.” Lledo, V., S. Yoon, X. Fang, S. Mbaye, and Y. Kim. International Planning Studies 12 (3): 291–303. 2017. “Fiscal Rules at a Glance.” International https://doi.org/10.1080/13563470701640176. Monetary Fund, Washington, DC. Hudalah, D., F. Z. Fahmi, and T. Firman. 2013. Ministry of Home Affairs. 2014. “Pembentukan “Regional Governance in Decentralizing Daerah-Daerah Otonom Di Indonesia Indonesia: Learning from the Success of S a m p a i D e n g a n Ta h u n 2 0 1 4 .” ( “ T h e Urban-Rural Cooperation in Metropolitan Establishment of Autonomous Regions in Yogyakarta.” In Cleavage, Connection and Indonesia until 2014.”) Direktorat Jenderal Conflict in Rural, Urban and Contemporary Otonomi Daerah, Kementerian Dalam Negeri Asia, 65–82. Springer. Republik Indonesia. http://otda.kemendagri​ Jatim Times. 2018. “Dulu Jalan Sendiri-Sendiri, .go.id/CMS/Images/SubMenu/total_daerah​ Pemerintah Malang Raya Mulai Kompak.” _​otonom.pdf. (“Once Discordant, the Governments of Greater Ministry of Public Works and Housing. 2017. “The Malang Area Have Now Been Collaborating.”) State of Indonesian Cities 2017: Transformation Jatim Times, August 8. https://www.jatimtimes. Towards Livable Cities.” Ministry of Public com/baca/176885/20180808/081400/dulu​ Works and Housing, Republic of Indonesia, -jalan-sendirisendiri-pemerintah-malang-raya​ Jakarta. -mulai-kompak/. Nasution, A. 2016. “Government Decentralization Joshi, R., R. Gayathri, A. Permono, B. Ladda, Program in Indonesia.” ADBI Working Paper A. Parasher, R. Hutagaol, and R. Prastiwan. 601, Asian Development Bank Institute, Tokyo. 2015. “Consultant Services for Market https://dx.doi.org/10.2139​/ssrn.2877579. Assessment and Operating Framework for the OECD (Organisation for Economic Co-operation Indonesia Regional Infrastructure Development and Development). 2016. Economic Survey Fund: Final Report.” Consultant Report for the of Indonesia–October 2016 . Paris: OECD World Bank, CRISIL, Mumbai. Publishing. https://www.oecd.org/eco/surveys​ KemenPAN-RB. 2016. “Laporan Kinerja /economic-survey-indonesia.htm. Kementerian Pendayagunaan Aparatur Negara OECD and UCLG United Cities and Local Dan Reformasi Birokrasi Tahun 2016.” (“2016 Governments). 2016. “Subnational Annual Report on Performance of the Ministry Governments around the World: Structure U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   215 and Finance—A First Contribution to the World Bank. 2003. “Decentralizing Indonesia: Global Observatory on Local Finances.” OECD A Regional Public Expenditure Review Report.” Publishing, Paris. World Bank Regional Public Expenditure Prakash, P. 2013. “Property Taxes Across G20 Review Overview Report 26191-IND, World Countries: Can India Get It Right?” OXFAM Bank, Washington, DC. India Working Paper Series, OXFAM India. ———. 2005. East Asia Decentralizes: Making R u k m a n a , D. 2 0 1 5 . “ T h e C h a n g e a n d Local Government Work. Washington, DC: Transformation of Indonesian Spatial Planning World Bank. after Suharto’s New Order Regime: The Case of ———. 2008. Spending for Development: Making the Jakarta Metropolitan Area.” International the Most of Indonesia’s New Opportunities. Planning Studies 20 (4): 350–70. Washington, DC: World Bank. Sandee, H. 2016. “Improving Connectivity ———. 2012a. Central–Local Government in Indonesia: The Challenges of Better Relations in Thailand: Improving Service Infrastructure, Better Regulations, and Better D e l i v e r y. T h a i l a n d P u b l i c F i n a n c i a l Coordination.” Asian Economic Policy Review Management Report 2012. Washington, DC: 11 (2): 222–38. https://doi.org/10.1111​ World Bank. /aepr.12138. ———. 2 0 1 2 b . I n d o n e s i a — T h e R i s e o f Schwab, K., and X. Sala-i-Martin. 2017. The Metropolitan Regions: Towards Inclusive Global Competitiveness Report 2017–2018. and Sustainable Regional Development . Geneva: World Economic Forum. https:// Washington, DC: World Bank. http:// www.weforum​. org/reports​/ the-global-com- d o c u m e n t s . w o r l d b a n k . o r g / c u r a t e d / e n​ petitiveness​-report-2017-2018. /520931468269430645/Indonesia-The-rise-of​ Sellers, J. M., and V. Hoffmann-Martinot. -metropolitan-regions-towards-inclusive-and​ 2008. “Metropolitan Governance.” In -sustainable-regional-development. Decentralization and Local Democracy in the ———. 2012c. “Investing in Indonesia’s Roads: World: First Global Report by United Cities Improving Efficiency and Closing the Financing and Local Governments , edited by United Gap.” Road Sector Public Expenditure Review Cities and Local Governments and World 73303, World Bank, Washington, DC. Bank, 255–79. Washington, D.C.: World Bank ———. 2 0 1 4 . “ I n t e r n a t i o n a l B a n k f o r and United Cities and Local Governments. Reconstruction and Development Project http://hdl.handle​.net/10986/2609. Paper on a Proposed Additional Loan and Soekarwo. 2018. “KPBU SPAM UMBULAN: Reconstructuring in the Amount of US$500 Praktik Berhasil Kerjasama Pemerintah Badan Million to the Republic of Indonesia for a Local Usaha (KPBU) Di Daerah Sistem PenyediaanAir Government and Decentralization Project Minum (SPAM) Umbulan.” (“KPBU SPAM Additional Financing / Local Government UMBULAN: A Successful Implementation and Decentralization Project Phase II.” Report of Public Private Partnership (KPBU) in 82002-ID, World Bank, Washington, DC. Umbulan Drinking-Water Supply System ———. 2015a. East Asia’s Changing Urban (SPAM).”) PowerPoint presentation presented Landscape: Measuring a Decade of Spatial at the Indonesia Development Forum, Jakarta, Growth. Washington, D.C.: World Bank. http:// Indonesia, July 10. https://indonesiadevelop- hdl.handle.net/10986/21159. mentforum.com/download​/index/1640. ———. 2015b. “More and Better Spending: Sutriadi, R., and A. Wulandari. 2014. “Towards Connecting People to Improved Water Supply a Communicative City: Enhancing Urban and Sanitation in Indonesia.” Water Supply and Planning Coordination by the Support of Sanitation Public Expenditure Review (WSS- Information and Communication Technology. PER), World Bank, Washington, DC. Case Study Bandung Metropolitan Area, ———. 2015c. “Infrastructure Verification Indonesia.” Procedia–Social and Behavioral Handbook: Good and Bad Illustration of Sciences 135 (August): 76–81. https://doi​ Infrastructure.” World Bank, Washington, DC. .org/10.1016/j.sbspro.2014.07.328. ———. 2016. “Size of the Public Sector: Urban Institute and Brookings Institution. Government Wage Bill and Employment.” n.d. “What Are the Sources of Revenue for World Bank, Washington, DC. http://www​ Local Governments?” Tax Policy Center .worldbank.org/en/topic/governance/brief/size​ Briefing Book . Washington, DC: Urban -of-the-public-sector-government-wage-bill​ Institute and Brookings Institution. https:// -and-employment. www.taxpolicycenter.org/briefing-book​ ———. 2017a. “October 2017 Indonesia Economic /what-are-sources-revenue-local-governments. Quarterly: Closing the Gap.” World  Bank, 216  TIME TO ACT Washington, DC. https://www.world​ bank.org​ 500 Million to the Republic of Indonesia for /en/country/indonesia/publication​/­i ndonesia​ the Local Government and Decentralization -economic-quarterly-october-2017. Project (P111577).” ICR00004433, World ———. 2017b. “Indonesia–Regional Infrastructure Bank, Washington, DC. Development Fund Project.” Project Appraisal ———. 2018b. “Getting the Most out of Indonesia’s Document 1579, World Bank, Washington, DC. Performance-Based Grants: Suggestions for ———. 2017c. “Indonesia: National Affordable Reform.” Policy Note, World Bank. Housing Program Project.” Project Appraisal ———. 2018c. “Mapping of Indonesia’s Civil Document PAD1788. World Bank, Indonesia. Service.” ID: P163712. Analytics of Indonesia’s http://documents.worldbank.org/curated​ Civil Service. World Bank, Jakarta. /­en/121201489975262694/Indonesia-National​ ———. 2018d. “Indonesia—Public Expenditure -Affordable-Housing-Program-Project. and Financial Accountability (PEFA): ———. 2017d. “Improving Quality of Spending Assessment Report 2017.” World Bank, in Indonesia—2018 Budget and Beyond.” PER Washington, DC. Update Phase 2, May. ———. 2018e. “Infrastructure Sector Assessment.” ———. 2 0 1 7 e . “ I n t e r n a t i o n a l B a n k f o r World Bank, Jakarta. Reconstruction and Development Project ———. 2018f. “Indonesia–Program to Accelerate Appraisal Document on a Proposed Additional Agrarian Reform (One Map Project).” Project Loan in the Amount of US$50 Million to the Appraisal Document PAD2748, World Bank, Republic of Indonesia for a National Urban Washington, DC. http://documents.worldbank​ Development Project (NUDP) (P163896).” .org/curated/en/393931532143851037​ Project Appraisal Document PAD2646, World /Indonesia-Program-to-Accelerate-Agrarian​ Bank, Washington, DC. -Reform-One-Map-Project. ———. 2017f. “December 2017 Indonesia Yoseph-Paulus, R., and R. Hindmarsh. 2018. Economic Quarterly: Decentralization That “Addressing Inadequacies of Sectoral Delivers.” World Bank, Washington, DC. Coordination and Local Capacity Building ———. 2018a. “Implementation Completion and in Indonesia for Effective Climate Change Results Report (Ln. 7914-ID and Ln. 8438-ID) Adaptation.” Climate and Development 10 (1): on a Loan in the Amount of USD 220 Million 35–48. https://doi.org/10.1080/17565529.2016 and an Additional Loan in the Amount of USD .1184609. U rba n G o v er n a n ce , I n stitutio n s , a n d F i n a n ce   217 Infrastructure and Policies to Connect the Portfolio of Places 6 KEY MESSAGES Integration across Indonesia’s portfolio of places connective infrastructure. Investment in roads is essential to making urbanization more pros- and transport options are picking up slightly, perous and inclusive, but infrastructure policies driven by subnational governments, but prog- and investment have failed to deliver the needed ress has been limited and unequal across spatial integration. Productivity and welfare places. To fill the gap, Indonesia should losses, both overall and across places, have fol- ­ prioritize increasing the efficiency of public lowed. To reverse the losses, Indonesia needs to ­ i nvestments, supplemented by the private invest more in connective infrastructure—and ­ sector, especially for main national corridors. to invest better and in the right places. Doing so •  Key to this efficiency question is the spatial tar- requires setting expectations about public and geting of infrastructure investments under private contributions, both within and across Indonesia’s decentralization, which, in the con- connective infrastructure sectors; clarifying and text of reduced aggregate investments, directed simplifying the relationships between central more resources to remote, less populated areas. and subnational government institutions; and This spatial targeting has failed, because it has defining and enforcing laws and regulations to not bridged preexisting service gaps or reversed support investment and provide alternatives to the adverse trend in economic activity in the usual disordered, informal urban develop- remote places. The failure suggests that the ment and mobility practices. binding constraint in connective infrastructure and service provision for less populous, non- •  Indonesian investments in connective infra- metro, districts is not resources but is related to structure fell sharply after 1997, mostly for local capacity issues for planning and execut- the central government and state-owned ing infrastructure projects, which undermine enterprises, and have never sufficiently recov- the translation of resources into results. It also ered to satisfy the growing demand for implies that rebalancing the distribution of 219 resources toward places with the highest num- •  Improving the efficiency of infrastructure ber of poor people may be more effective in investments to spread prosperity and generate reducing overall inequality. inclusiveness will require adjusting the trade- •  The failure to generate new infrastructure—to off between efficiency and spatial equity by connect places and serve localities through improving coordination among districts, road, maritime, and flight connections—and to provinces, and the central government in upgrade existing infrastructure is probably planning and investment; addressing local responsible for a major part of the large spatial capacity issues; and adjusting the balance of welfare gaps across Indonesia, which are not responsibility and resources across different arbitraged away and create productivity losses. levels of government. Integration across Indonesia’s portfolio of The reasons can be broadly characterized as places is essential to making urbanization more the following: prosperous and inclusive. However, with rapid •  A centralized decision-making culture com- urbanization and the persistent gaps in pros- bined with decentralized decision-making perity, inclusiveness, and livability described in authority and a complex and inadequate part 1, the current spatially ­ connective infra- distribution and coordination of resources, structure is failing to deliver the services responsibilities, and administrative capa- needed.1 This chapter considers s ­ patially con- bilities across levels of government. nective infrastructure: both ­ transport—roads, •• A legal framework in which rules, such as railroads, ports, and ­ airports—and telecom- those governing land use, are inadequate munications n ­ etworks—telephony and or badly enforced, leading to a pattern of Internet. In addition, the chapter analyzes local disordered, informal urban development infrastructure services such as electricity, water, and mobility. and sanitation. The broad diagnosis of disappointing As a result, segmented and complex policy results is supported by the following findings. making suffers from coordination failures First, investment has been consistently low and spatial mismatches. In turn, a general since the end of the 1990s, leaving Indonesia mismatch between local means, needs, and lagging similar emerging countries on several capacities has resulted, generating spatial infrastructure service indicators. Second, the inequalities in access to public goods, particu- deficit in the aggregate figures has been com- larly in urban and interurban transport. The pounded by a spatial imbalance in resources final outcomes are resource misallocations for and, hence, investments following Indonesia’s both households and firms, bottlenecks, “big bang” decentralization reforms. The imperfect access to markets, and lost produc- resulting redirection of resources toward tivity and, therefore, prosperity. remote, less populous districts has not pro- For infrastructure services to meet current duced tangible infrastructure improvements. urbanization and development challenges, Finally, the barriers to mobility induced by Indonesia needs to invest more in connective the slow progress made in connective infra- infrastructure—and to invest better and in the structure are likely to have an important pro- right places. The results in this chapter sup- ductivity cost and explain an important part port the view that resources are unlikely to be of the large remaining welfare gaps across the the binding constraint for connective infra- country that were documented in chapter 4. structure, and, more generally, basic service Chapter 5 addressed the reasons for these provision for less populous, typically non- infrastructure shortcomings and institutional metro and more rural, districts. Instead, policies to address them in more detail. issues of local capacity to plan and execute 220  TIME TO ACT infrastructure projects are likely to be the decreased from only 7.6 percent to 5.4 percent main bottleneck that prevents translating of GDP, which itself shrank by more than resources into results. 13  percent in 1998 with the Asian crisis. This diagnostic, in turn, suggests that the Similarly, subsequent increases in absolute priority should be clarifying and simplifying infrastructure spending barely matched eco- the relationships between central and subna- nomic growth, so recovery as a share of GDP tional government institutions, both prov- has been limited. Further, adjusting for pop- inces and districts; defining and enforcing ulation growth shows a similar pattern: per institutional and legal rules that support and capita real infrastructure spending in 2015 coordinate investments; and possibly rebal- was 1.8 times the 1995 spending, whereas ancing the allocation of resources according per capita GDP was 3.3 times higher. to needs across the portfolio of places identi- Much of the variation has been driven by fied in the first part of this report. energy investments, which fell from 3–4 ­ percent of GDP to less than 1 percent in the 2000s (figure 6.1, panel b). Investment by state- Lower aggregate owned enterprises and the central government infrastructure investment also retreated, leaving a void that has gone since the 1997 Asian crisis largely unfulfilled by subnational govern- ments and the private sector. After hovering above 7 percent of gross domestic product (GDP), total infrastructure investments in Indonesia fell sharply after Indonesia has had limited and uneven progress on transport infrastructure the 1997 Asian crisis and failed to recover, investment except for a small peak in 2008 (figure 6.1, panel a; see annex 6A for data sources) Focusing more specifically on spatially con- (see also chapter 5). By 2015, the share of nective infrastructure, total transport sector infrastructure investment was below investment fell from about 2 percent of GDP 3 ­ percent of GDP. to less than 1 percent at the beginning of the Although real infrastructure investment 2000s, followed by a recovery to pre–Asian was halved between 1997 and 1998, it crisis levels in the 2010s (figure 6.2, panel a). FIGURE 6.1  Infrastructure investment has fallen as a share of GDP, by source and sector, 1995–2015 a. Infrastructure investment by source b. Infrastructure investment by sector 10 10 8 8 Share of GDP (%) Share of GDP (%) 6 6 4 4 2 2 0 0 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Central government Subnational governments Energy Telecommunications Transportation State-owned enterprises Private enterprises Irrigation Water and sanitation Source: World Bank 2015. Note: GDP = gross domestic product. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   221 In an economy highly dependent on road asphalt road as the largest artery in 2014, transport, with road vehicles accounting for leaving close to 25 percent of villages with 70 percent of freight ton–kilometers and only dirt, hardened, or other types of roads. 82 percent of passenger kilometers (World In 12 percent of villages, vehicles with four or Bank 2012), this level of investment is more wheels cannot pass through all year unlikely to meet demand. Both the house- round. Subnational roads, which are critical hold and the productive sectors are creating for local connectivity, are generally in bad pressure. A large increase in vehicle owner- condition, with 40 percent classified as dam- ship mirrors Indonesia’s poverty reduction, aged or poor. Growing evidence based on rig- in which the national poverty and vulnera- orous econometric analysis shows that the bility rate declined from 54 percent to bad state of the road network impedes eco- 31 percent between 2001 and 2017 and the nomic activity and constrains economic trans- share of the population classified as mid- formation (box 6.1). dle  class increased from 5 percent to In parallel, congestion has become a 22  ­percent.2 The number of passenger cars major issue across most urban areas (see increased nearly sixfold between 1995 and chapters 2 and 7). Jakarta ranks as the most 2014, from less than 2.5 million to congested city in the world in the 2014 14.5 ­million, whereas the number of motor- Castrol Magnatec Stop-Start Index cycles increased tenfold, from 10 million to (Wardhani and Budiari 2015) and as the more than 100 million. third most congested in the 2016 TomTom Meanwhile, only 77 percent of villages Traffic Index. 3 In DKI Jakarta (Daerah where land transport is the main mode had an Khusus Ibukota Jakarta), the number of cars BOX 6.1  Road quality and structural transformation in Indonesia Gertler et al. (2015) show that road improve- primary sector activities, such as agriculture and ments significantly boosted consumption and fishing, by manufacturing or services. income in Indonesia over the period 1990- A log unit increase in district road quality led 2007. These positive welfare effects arose in to a 31 percent increase in the probability that a part through increased labor demand from manufacturing establishment existed in the dis- new firms, generating an occupational shift trict, a 43 percent increase in the net number of from agriculture to higher-paying manufactur- firms, and a significant increase in district manu- ing jobs. Their study investigated the impact facturing output (the elasticity of output with of road quality on economic activity across respect to road quality is 3.86). Improved roads Indonesia, using an administrative database led to a substantial reduction in the likelihood of that tracked road-­ segment quality annually for a worker being employed in agriculture, which all Indonesian highways. The proxy for qual- drops 8.5 percent for every log point improve- ity is an international road roughness measure ment in road quality, whereas the likelihood of collected by sensors in vehicles traveling along being employed in manufacturing increases by the roads. Merging this information with both 10 percent. A 1 standard deviation improvement a nationally representative household panel in road quality in a district led to a 5 percent database (the Indonesian Family Life Survey) increase in household consumption per capita and the annual manufacturing census (Statistik and an almost 20 percent increase in total labor Industri ), the researchers established a link earnings. between better road connections and structural transformation—the gradual replacement of Source: Gertler et al. 2015. 222  TIME TO ACT doubled over 10 years to reach 3.2 million in other bureaucratic inefficiencies, they may 2014, and the number of motorcycles tri- also face poor incentives and soft budget pled. 4 Similar congestion is beginning to constraints (World Bank 2018). occur in Bandung, Yogyakarta, Semarang, The private sector share of infrastructure and Surabaya. Underdeveloped mass transit investment remained low throughout the systems are largely responsible, with modal 1990s and 2000s and has declined even fur- shares of public transport at or below ther, from 19 percent in 2006–10 to 9 ­percent 20 percent in most large cities (see chapter 2; in 2011–15 (World Bank 2018). The bulk is see chapter 7 on policies to address the in telecommunications and energy. In trans- issue).5 port, the only private interventions con- Subnational government investments in cerned commercially viable toll roads. More transport infrastructure have grown in impor- than 3,000 kilometers of toll roads have tance since decentralization, from about been, or are being, privately funded in 25 percent of the total in the mid-1990s to Indonesia to date. The private sector accounts more than 50 percent in the second half of the for only 18  percent of the total length of 2000s, whereas the central government share these projects: 33 percent for completed experienced an inverse evolution (­ figure 6.2, roads and less than 12 percent for roads panel b). under construction or road projects that have The shares of state-owned enterprises and been awarded/assigned (World Bank 2018). of the private sector in transport infrastruc- Of the 40 new toll road projects under con- ture investment remain limited, mostly to struction or soon to be started, 28 are in large national corridors (see f ­igure 6.2, Jawa (table 6.1). panel b). State-owned enterprises may take The government has set an ambitious goal on commercially nonviable projects as part of 37 percent private participation in trans- of their developmental mandates. Although port infrastructure investment, but reaching they deliver on infrastructure priorities fast, that goal may be difficult in the short run with high tolerance for regulatory and finan- given Indonesia’s experience and international cial risks, barriers to land acquisition, and experience (box 6.2). FIGURE 6.2  Transport infrastructure investment has rebounded as a percentage of GDP, with a greater share coming from subnational governments, 1995–2016 a. Share of GDP b. Share of total investment 2.5 100 2.0 80 1.5 60 Percent Percent 1.0 40 0.5 20 0 0 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Central government Subnational governments State-owned enterprises Private enterprises Source: World Bank 2015, updated by World Bank and Prospera. Note: GDP = gross domestic product. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   223 TABLE 6.1  Toll road projects, complete and in the pipeline, 2017 Projects awarded/assigned Completed projects Projects under construction (yet to start construction) Region/ownership Number km Number km Number km Jawa region 26 828 23 987 5 143 State-owned enterprises 18 547 17 810 3 55 Private 8 281 6 177 2 88 Non-Jawa region 3 65 8 640 4 458 State-owned enterprises 2 53 8 640 4 458 Private 1 12 0 0 0 0 Total 29 893 31 1627 9 601 State-owned enterprises 20 600 25 1450 7 513 Private 9 293 6 177 2 88 Source: World Bank 2018. Note: km = kilometers. BOX 6.2  Attracting private infrastructure investment Aiming to increase overall investment in infra- Energy is the main sector where meaningful structure, the government of Indonesia has set private participation can be expected. Expanding an ambitious target that 37 percent of total that participation might allow reorienting some infrastructure investment should come from public resources toward other sectors such as the private sector. The government is push- transport in the short term. However, Indonesia ing, in particular, for a larger role for public–­ is the one country in Asia that has eschewed private partnerships (PPPs). Although scope for reforms that would enable private participa- improvement exists (see chapter 5), this goal of tion in the energy sector (Foster et al. 2017). increasing PPPs might be difficult to attain. To reverse this, the institutional and regulatory In low- and middle-income countries, the environment needs to be modernized by allow- PPP share of total infrastructure investments ing for an independent regulatory entity and has historically been 5–10 percent. About half some vertical or horizontal unbundling. of these PPPs rely on public financing sources. So, although attracting private investments in Attracting private financing involves setting infrastructure has an important role to play in regulatory pricing rules to ensure a minimum addressing the infrastructure gap, much of the cost recovery. Because prices enabling sufficient needed progress is likely to come from the public cost recovery may be unaffordable for the poor- sector. est households, they create a trade-off between financial viability and social inclusion. Source: Fay, Martimort, and Straub 2018; World Bank 2018. Maritime transport sector performance is especially for trade. However, the perfor- hampered by regulatory issues mance of port infrastructure and operations still displays ample room for improvement. Indonesia is an archipelago with more than Reviews of the current situation have high- 17,500 islands, making good maritime con- lighted the shortcomings of the sector, which nectivity vital for passenger travel and 224  TIME TO ACT stem from insufficient investment and an chicken was Rp 20,400 per kg in Jakarta, com- inadequate legal and regulatory framework. pared with Rp 30,050 per kg in West Papua.6 The performance of international services is To some extent, this variation is under- summarized by the Liner Shipping Connectivity standable because Indonesia hosts one of the Index (LSCI) computed by the United Nations most complex port systems in the world, Conference on  Trade  and Development with more than 3,672 ports and terminals (UNCTAD) and based on five components of ranging from large port agglomerations the maritime transport sector: number of ships, along the northern coast of Jawa, such as container-carrying capacity, maximum vessel Tanjung Priok/Banten and Tanjung Perak/ size, number of services, and number of com- Gresik, to small jetties in remote areas (Van panies that deploy container ships in a coun- Tuijll 2018). try’s ports. According to this measure, Physical capacity investments have been Indonesia lags most countries in East and limited, especially because most secondary South Asia, a situation that has not changed ports are not, in their current configuration, since 2004 despite a quite marked improve- designed to handle large ships and contain- ment between 2010 and 2017 (­ figure 6.3). erized cargo, which, in turn, has constrained Unsurprisingly, such poor performance also demand (Van Tuijll 2018). The potential for translates into wide differences in national growth is illustrated by Jayapura, a small shipping costs: sending a 6-meter container port near the border of Indonesia and within Indonesia from Tanjung Priok to Papua New Guinea. In 2012, Jayapura, a Jayapura, Padang, and Banjarmasin costs typical local port serving a large captive approximately US$1,000, US$600, and hinterland with a low population density, US$650, respectively, compared with shipping invested in a 130-meter berth with two con- costs from Tanjung Priok to Guangzhou tainer cranes. Larger vessels now visit the (China) and Singapore of about US$400 and port; port congestion was significantly US$185, respectively. This, in turn, leads to eased, with the berth occupancy rate falling large price disparities across the country. For from above 80 percent to 60 percent; and example, in September 2014, onions in Jakarta throughput volumes jumped by about sold for roughly 41,250 rupiah (Rp; about 20 percent (New Frontier Solutions 2017, $3.14) per kilogram (kg), compared with Rp based on Pelindo IV data). Comparing 60,000 per kg in West Papua. The price of inflation rates and cargo traffic for FIGURE 6.3  Indonesia’s Liner Shipping Connectivity Index performance lags neighbors 160 140 120 100 LSCI score 80 60 40 20 0 Philippines Indonesia Thailand India Malaysia Singapore China 2004 2010 2017 Source: Calculations based on data from United Nations Conference on Trade and Development (UNCTAD). I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   225 23 regional ports in Eastern Indonesia, of •  The absence of a clear definition of the which 13 were upgraded, New Frontier role of the central government in strategic Solutions (2017) finds that increased port transport planning and the lack of inde- capacity led to lower and more stable infla- pendence of the port authority, particu- tion, resulting in economic savings. larly in staffing and procurement The financial upgrading requirements of •• Distortions in the tariff structure. Indonesian ports are often very large and, in Several reform proposals have been floated in some cases, would imply moving the ports recent years to address these issues. The latest outside of the current city center locations includes the pendulum proposal and the rec- and buying land for greenfield development. ommendation by the World Bank following Even focusing just on the main ports, of an evaluation of this proposal (World Bank which there are more than 30, would mean 2014). The pendulum proposal would replace investing in a large number of locations. separate single port-to-port services between As a result, the sector suffers from ineffi- the five main ports of Indonesia (Belawan in ciencies. A McKinsey (2012) report stresses the west, Jakarta and Surabaya, and Makassar deficiencies that lead to very high service and Sorong in the east) by a pendulum service costs: linking all five. It would require investments •  The small size of vessels, requiring the fleet so that these ports can accommodate larger to be larger than necessary ships and use the eastern ports of Makassar •  The low fleet use and low service and Sorong as hubs to serve smaller ports in frequency the region. World Bank (2014) recommends •  The large imbalance between inbound and leaving r­ outing decisions for the market to outbound traffic, due to Jawa’s position as organize, using larger ships than are currently a manufacturing hub and import gateway deployed. •• A domestic network that consists of single These proposals are mostly organized port-to-port routes rather than multiport around alternative visions of the physical itineraries. routes and types of ships that would opti- mally serve Indonesia’s trade shipping Since 1992, all Indonesian commercial ports needs. Whatever the final proposal consid- are operated by four state-owned enterprises, ered, the set of administrative rules and reg- known as Pelindo I to Pelindo IV, each cover- ulations would generate specific incentives ing a designated region. In practice, Pelindos to the actors of the maritime transport sec- have acted both as operators and regulators, tor, including port operators, investors, and subcontracting operations to private compa- ­ clients. The first two issues to be addressed nies. In a worldwide context of significant concern tariffs, particularly their adjust- evolution in the shipping sector—with the rise ment, and the definition of commercial of very large ships and containerization ports. volumes tripled over 20 years), as well as the (­ Current tariffs for domestic users are too increase in private sector participation— low to justify investment in the ports. One Indonesian ports engaged in a development obstacle to adjusting tariffs lies in the obli- process. The 2008 shipping law introduced a gation to consult with all port stakehold- new Port Authority to regulate the sector. This ers, associations for domestic shipping lines, law was meant to ease entry of private opera- freight forwarders, importers, and exporters tors through port operations concessions, de (Regulation 6/2013, article 15b). Because these facto ending the monopoly of the Pelindos; tend to resist tariff increases, redefining the however, the transition to private operation perimeter and attribution of the regulator is has been slower than expected. needed. In addition, even the limited revenues The main regulatory issues are the follow- collected are transferred to the Ministry of ing (Van Tuijll 2018): Finance and end up back in the port of col- •  A mismatch between the classification of lection only through regular budget cycles. ports, their economic importance, and the Accordingly, high-revenue-generating ports structure of the administrative authorities do not get the optimal resources to reinvest, 226  TIME TO ACT and the quality of service remains stagnant at Domestic travel is similarly very concen- best, which results in uncertainty to commit to trated in Jakarta, accounting for 56 percent of investments to improve access, a problem espe- passenger arrivals and departures in 2017, and cially for ports within urban areas. along a few routes, with the 12 main connec- Van Tuijll (2018) reports that only about 30 tions accounting for 40 percent of the total (see ports are likely to qualify as commercial in the map 6.1). At more than 140 million seats in current situation. Revisiting their classification 2017, Indonesia’s domestic market is big and would generate challenges related to staffing growing, comparable in size to that of Japan. and the status of other noncommercial ports.7 Indonesian airports are operated by two state-owned enterprises, Angkasa Pura 1 (AP1) and Angkasa Pura 2 (AP2), which rep- Air transport sector would benefit from resent more than 90 percent of seats. AP1 and more competition AP2 act as de facto monopolies in their terri- Passenger travel in Indonesia relies heavily on tories, and the lack of competition adversely air travel (World Bank 2018). International affects safety and service standards, with the travel is heavily concentrated in two hubs, problem compounded by a relatively weak Jakarta and Denpasar-Bali, which in 2017 regulatory enforcement environment. accounted for 46 and 33 percent of seats, Capacity constraints are thus likely to respectively. The main entry points were become more stringent unless significant Singapore and Kuala Lumpur. Between 2008 changes occur in the sector’s governance to and 2016, annual passenger traffic growth update the current regulatory practice and was 7.8 percent at Jakarta Soekarno-Hatta allow for more competition. Given successful International Airport, and 11.3 percent at experiences in other emerging countries, the Bali-Denpasar Ngurah Rai International possibility of allowing private participation in Airport. Overall, the International Air some of the main airports should be consid- Transport Association forecasts 5 percent ered (World Bank 2018). Transferring a full or annual growth in passenger traffic between a majority ownership to the private sector is 2014 and 2034, making Indonesia one of the likely to be the most effective avenue for pri- fastest-growing markets. vate participation (Fabre and Straub 2018). MAP 6.1  Consolidated air passenger flows Source: Elaboration based on 2015 Air Transportation Statistics, Bada Pusat Statistik (Statistics Indonesia, BPS). Basemap credit: BPS, ESRI, and the World Resources Institute, Openflight data (https://openflights.org/data.html), and GeoNames. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   227 FIGURE 6.4  Telecommunications infrastructure investment has fallen as a percentage of GDP, with almost all coming from the private sector and state-owned enterprises, 1995–2014 a. Share of GDP b. Share of total investment 2.5 100 2.0 80 1.5 60 Percent Percent 1.0 40 0.5 20 0 0 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Central government Subnational governments State-owned enterprises Private enterprises Source: World Bank 2015, updated by World Bank and Prospera. Note: GDP = gross domestic product. FIGURE 6.5  Mobile cellular subscription rates in the 2000s to almost 1.5 percent of GDP but are higher in Indonesia than in most comparator has been below 1 percent since 2010. Unlike in countries, 2016 transport, state-owned enterprises and private investors are the sole source of investment in telecom. Although the state-owned enterprise Thailand share has been decreasing, it still represents Singapore about half (figure 6.4, panel b). Indonesia How does telecommunications investment translate into service? On the one hand, rela- Malaysia tive to comparator countries, mobile subscrip- Vietnam tion figures are high, below only those of Brazil Singapore and Thailand and above those of Brazil, China, Malaysia, the Philippines, and Philippines Vietnam (figure 6.5). Internet access, on the East Asia and Paci c other hand, appears to be much less wide- World spread. Indonesia is clearly lagging, with only a quarter of the population enjoying access, China most of them through their mobile phones India (figure 6.6). 0 20 40 60 80 100 120 140 160 180 Indonesia faces several obstacles to expand- Mobile cellular subscriptions (per 100 people) ing Internet access (Freedom House 2016). First, most connections are concentrated on Source: Calculations based on data from World Bank World Development Indicators (https:// data. Jawa: 52 million of the 88.1 million people worldbank.org/products/wdi). online, according to the Indonesia Association Note: East Asia and Pacific excludes high-income economies. for Internet Providers. Second, average connec- tion speed is slow—7.2 megabits per second, Achieving universal Internet access requires tackling mobile Internet access pricing issues ahead only of India and the Philippines in the Asia and Pacific region (Akamai 2017). By Telecom investment dropped sharply after comparison, average speed was 18.7 megabits 1997 (figure 6.4, panel a). It partially recovered per second for the United States, 20.2 for 228  TIME TO ACT FIGURE 6.6  Internet use remains low in Indonesia despite the high number of secure Internet servers 90 100,000 80 70 10,000 Percent of population 60 1,000 Log scale 50 40 100 30 20 10 10 0 0 Singapore Malaysia Brazil Philippines China East Asia Thailand Vietnam World India Indonesia and Paci c Individuals using the Internet, 2016 (left scale) Secure Internet servers per 1 million people, 2017 (right scale) Source: Calculations based on data from World Bank World Development Indicators (https:// data.worldbank.org/products/wdi). Note: East Asia and Pacific excludes high-income economies. Japan, and 28.2 for the Republic of Korea. percent of subnational government revenue. Notably, between 2010 and 2017, the number The transfers, comprising general and special of secure Internet servers per million people purpose grants, as well as shared tax and nat- jumped from 1.6 to almost 1,300. This sug- ural resource revenue, greatly increased sub- gests that, outside Jawa, advancing toward national government revenues (figure 6.7). universal access across Indonesia requires tack- Provinces have about 25 percent of the ling mobile Internet access pricing and not just revenues of districts on aggregate and depend physical bottlenecks. less on fiscal transfers than districts do, because 10 percent of the main category of transfer, the general purpose fund, goes to Problems arising from provinces and 90 percent to districts. the spatial targeting of Subnational government spending infrastructure investment increased from 26 percent of total govern- ment spending to 43 percent between 2001 under decentralization and 2015. Districts spend about 30 percent of Following the Asian crisis of 1997 and the it, provinces about 10 percent. Subnational 1998 departure of longtime president government infrastructure spending has also Suharto, Indonesia undertook a profound increased substantially—by a factor of four in decentralization of political and administra- real terms between 2001 and 2014. By 2014, tive processes and transferred decision-­ it represented about 20 percent of subnational making power on a large array of policies, government spending and 1.5 percent of GDP. including transport infrastructure, to Most of it went to transport—more than ­subnational governments. 83 percent to subnational road development Although tax revenues remained central- between 2005 and 2009, of which ized, a complex system of intergovernmental about  70  percent went to new roads and transfers was put in place to support subna- 30 percent to maintenance. tional governments’ ability to fulfill their new Decentralization clearly had a redistributive mandates (see chapter 5). Since 2001, inter- purpose, in part to reinforce the cohesion of a governmental transfers have made up 70–90 country with regional diversity and, in some I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   229 FIGURE 6.7  Subnational government revenues have grown greatly, 1994–2016 800,000 600,000 Rupiah (billions) 400,000 200,000 0 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Own-source revenue General allocation grant Special allocation grant Total revenue sharing fund Other revenue Source: Calculations based on data from SIKD (regional financial administration system) of DJPK (Directorate general of fiscal Balance), Ministry of Finance. FIGURE 6.8  Transfers per capita skew toward less for equalization, rather than an adjusted “per populated districts, by current population quartiles, capita” basis. The per region distribution is 1994–2016 based on a fiscal gap formula, whereby the gap is computed as the difference between 12 revenues and expenditures, or “needs,” with 10 the needs being equal to total expenditures divided by the overall number of districts, Rupiah (millions) 8 adjusted by an index of cost adjustment vari- ables (Lewis 2014; McLeod and Fadliya 6 2010; see also chapter 5). 4 Because of this practice, citizens in the most populated districts receive significantly 2 lower transfers per capita, and hence have less 0 revenue per capita than citizens in the less 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 populated districts. There is an approximately Q1 Q2 Q3 Q4 13-fold gap in per capita revenues between districts in the first and the last population Source: Calculations based on data from SIKD (regional financial administration system) of DJPK (Directorate general of fiscal Balance), Ministry of Finance. deciles, and the ratio reaches 40 between the Note: Q denotes quartile. least populous district and the most populous district. Because the most populated districts cases, secessionist temptations. As a result, the also concentrate larger numbers of poor, and transfer allocation has strongly skewed toward hence have some of the most critical infra- poorer and less populated districts. Although structure gaps, the transfer rule achieves some progressivity in transfers to subnational a spatial redistributive purpose, but it fails governments is to be expected, as a way to to  achieve a more general redistributive boost spending, improve social services, and purpose.8 stimulate economic activity in lagging places, The actual distribution of transfers Indonesia is an international outlier. between 2001 and 2016 clearly illustrates Transfers flow to the places with the high- these facts (figure 6.8). Transfers to districts in est poverty rates rather than to those with the different population quartiles were on roughly highest number of poor citizens because the parallel trends between 1994 and 2000, but a transfers use an adjusted “per region” basis clear break appeared in 2001 and continued 230  TIME TO ACT until 2016, with districts in the lowest quar- FIGURE 6.9  Spending per capita skews toward less tile (Q1) ending up having about three times populated districts, by current population quartiles, more transfers per capita than those in the 1994–2016 highest quartile (Q4). This disparity was driven mostly by the nonearmarked general 7 purpose grant. The earmarked special pur- 6 pose grant, which represents about 10 percent 5 Rupiah (millions) of total revenues, was slightly more volatile but followed a similar evolution. 4 The revenue pattern translates directly into 3 the evolution of district infrastructure spend- 2 ing. Total spending and infrastructure spend- ing show a similar ratio of three to one 1 between the lowest population quartile dis- 0 tricts and the highest (figures 6.9 and 6.10). 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 The same districts have higher per capita rev- Q1 Q2 Q3 Q4 enues, resources, and spending. Source: Calculations based on data from SIKD (regional financial administration system) of DJPK Infrastructure spending does not automati- (Directorate general of fiscal Balance), Ministry of Finance. cally translate into actual capital (Pritchett Note: Q denotes quartile. 2000). In Indonesia’s case, is the asymmetry in transfers generating more infrastructure in FIGURE 6.10  Infrastructure spending per capita places that receive more resources, thereby skews toward less populated districts, by current addressing imbalances in economic opportu- population quartiles, 2001–16 nity and promoting inclusiveness? Previous literature looking at this question is limited 1.8 (box 6.3). We therefore perform an analysis 1.5 that answers this question in two steps (see Rupiah (millions) annex 6B for a description of the econometric 1.2 methodology). First, it shows how the skewed 0.9 revenues and related spending translate into physical amounts of the most common infra- 0.6 structure items: roads, electricity, water, and sanitation. For roads, the data cover four cat- 0.3 egories from the national level to the local: 0 motorways, principal highways, highways, 2000 2002 2004 2006 2008 2010 2012 2014 2016 and others. Examining both total kilometers Q1 Q2 Q3 Q4 of road per district and kilometers of road per Source: Calculations based on data from SIKD (regional financial administration system) of DJPK district per capita as outcomes allows com- (Directorate general of fiscal Balance), Ministry of Finance. parison across different levels of aggregation. Note: Q denotes quartile. Second, the analysis shows how the skewed revenues and spending affect firm-level out- added between 17 and 31 kilometers of high- comes, including employment, value added, ways per year, with the effect for two years, and labor productivity. 2011 and 2014, statistically significant. Other Less populated districts (those with below categories of larger roads—principal high- median population), boasting higher per cap- ways and motorways—experienced a relative ita spending after decentralization, produce decrease (compared to more populated dis- more kilometers of roads (see annex 6C, tricts) in these same districts. Because of this tables 6C.1 and 6C.2). The effects are, how- offset, the total stock of roads showed no ever, quite limited. The only significant effect effect. Finally, distinguishing between urban regards a single category of roads—highways. and rural districts does not alter this conclu- From 2003 to 2014, resource-rich districts sion (see annex 6C, table 6C.3). I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   231 BOX 6.3  Literature analyzing the impact of decentralization The end of Suharto’s 35-year rule initiated a local capture after decentralization might pre- substantial fiscal reform that transferred public vent effective public goods provision. Therefore, goods provision from the central government to the effect of decentralization on public goods districts. Before the reform, most development provision is ultimately an empirical question. spending was funded through earmarked central Few studies have attempted this analysis government transfers. After the reform, subna- for Indonesia. Kis-Katos and Sjahrir (2017) tional (district) governments received grants show that, after decentralization, investment in from the central government that they could physical and health infrastructure increased in spend according to their own priorities (see also districts where preexisting infrastructure was chapter 5). limited, thus hinting at better investment target- The theoretical literature on the effect of ing after decentralization. In contrast, Pal and decentralization on public goods provision Wahhaj (2017) show that, after decentralization, is ambiguous (see, for example, Bardhan and villages governed by traditional laws decreased Mookherjee 2015). Because of informational their investment in physical infrastructure. advantages, subnational governments might be better equipped to provide public goods, but Source: Jascisens and Straub (2018). Among other types of infrastructure, elec- analysis should largely address this issue tricity coverage in less populated districts because the effects studied are changes within showed an improvement relative to more pop- districts over time, with fixed factors such as ulated districts (figure 6.11, panel a). Electricity cost levels or specific geography controlled for. access, however, does not depend on district A second explanation attributes more expenditures and instead can be traced to the value to local roads if they connect beyond state electricity company’s aggressive policy, district borders to other equally developed which increased average coverage nationwide road networks and, ultimately, to main cities from 66 percent in 1996 to 96 percent in 2016. and ports across the region and nation. Water and sanitation, conversely, fell directly High-resource districts surrounded by simi- under district responsibility after 2001, and the lar high-resource districts may get higher larger resources targeted to low-population returns from road investments than places failed to produce a differential effect high-resource districts surrounded by low- ­ (figure 6.11, ­panels b and c). resource ones. Economists describe this com- Higher per capita spending thus failed to plementarity between district investments as translate into more infrastructure capital in “spillovers,” meaning one district’s invest- any sector except electricity. This conclusion ment has knock-on effects on neighboring also holds when looking specifically at urban districts so that the effect at a higher level of districts receiving more transfers and spending geographic aggregation is greater than the more. Several explanations might be offered. sum of the effects at the lower geographic First, the analysis automatically rules out level. As a result, we would expect spending explanations based on specific fixed character- to be more effective where clusters of high- istics of districts—for example, the fact that resource districts exist. The confirmation of less populated ones may also be more remote such spillovers would contribute to the case and have higher construction costs due to for promoting coordination across district both natural geography and reduced construc- boundaries. tion industry competition driving up markups. The spillover effect among districts may The inclusion of district fixed effects in the receive reinforcement from provinces, which 232  TIME TO ACT have administrative responsibility for roads FIGURE 6.11  Electricity access increased more in crossing several districts. Although provinces high-resource districts, but water and toilet access have limited resources, their tasks include did not, 1996–2015 coordinating investments among districts. They may find it easier to do that when they a. Share of households with access to electricity 0.12 are composed of many high-resource districts, thanks to both the higher aggregate revenues 0.10 Di erence in di erences estimate available and to the greater congruence of 0.08 objectives among district governments. 0.06 This possibility can be tested by comparing the evolution of outcomes between high- and 0.04 low-resource districts grouped by whether 0.02 they belong to “favored” provinces or to 0 “unfavored” provinces. Provinces can be divided into “favored” and “unfavored” –0.02 groups in two ways. The first is based on their –0.04 population, because the postdecentralization 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 transfer system implies that less populated b. Share of households with access to water provinces should be composed of many low- 0.08 population, high-resource districts. A second Di erence in di erences estimate 0.06 way is to group provinces according to their 0.04 share of high-resource districts. For example, provinces in which more than half the dis- 0.02 tricts are high-resource ones would be 0 expected to generate more spillovers, and –0.02 high-resource districts in favored provinces –0.04 would be expected to enjoy more spillovers. –0.06 Under the second way, four groups of dis- –0.08 tricts are studied: high-resource districts in favored provinces (HRD-FP: those with a –0.10 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 majority of similar high-resource districts), high-resource districts in unfavored provinces c. Share of households with access to toilets (HRD-UP: those with a majority of low- 0.03 resource districts), low-resource districts in 0.02 Di erence in di erences estimate favored provinces (LRD-FP), and low- 0.01 resource districts in unfavored provinces 0 (LRD-UP). In all estimations, district- and –0.01 province-level cutoff points are set at the –0.02 median, and the LRD-UP group is excluded and acts as a reference group. Grouping prov- –0.03 inces in this way produces 72 HRD-FP and –0.04 73 HRD-UP, each about 25 percent of the –0.05 total 291 districts, where, in this analysis, dis- –0.06 tricts are defined by their 1996 administrative 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 boundaries (table 6.2). Sources: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia) and Indonesia’s The results strikingly support the spillover National Socio-Economic Survey (SUSENAS). story regardless of the way in which we iden- Note: The bars indicate, for each year, the differential evolution of the point variables of interest tify “favored” and “unfavored” provinces. between districts with population below the median and districts with population above the median. The blue dots are the point estimates, with a positive value denoting that the position of districts with When we define favored provinces as those population below the median improved in relative terms; the blue lines at the ends of the bars show with below median population, we find the 95 percent confidence interval. that  the total stocks of roads increased I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   233 TABLE 6.2  District groups in the analysis of the impact of transfers on road construction Low-resource districts (LRD) High-resource districts (HRD) (population above median) (population below median) Favored provinces (FP) (more than 50 percent of districts are high resource) 17 72 Unfavored provinces (UP) (less than 50 percent of districts are high resource) 129 73 Source: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia) and Indonesia’s National Socio-Economic Survey (SUSENAS). Note: Districts are defined by their 1996 administrative boundaries. FIGURE 6.12  Districts in favored provinces show a significantly after 2001 in districts located in differential increase in highways and total stock of these provinces, because of an increase in the roads total kilometers of highways. From 2003 to 2014, these districts added between 53 and a. Highways 71 kilometers of highways per year. All the 90 effects are statistically significant. Although 80 the effects on other categories of larger roads Di erence in roads added (km) 70 (principal highways and motorways) were 60 still slightly negative, the effect on the total 50 stock of roads, which reaches about 110 kilo- 40 meters in 2014, is now positive and 30 significant. 20 When favored and unfavored provinces 10 are instead divided by whether more than 0 50 percent of their districts are high resource, –10 results are similar, although the magnitudes 1997 2003 2011 2014 1997 2003 2011 2014 1997 2003 2011 2014 are slightly smaller (figure 6.12). High- HRD-FP LRD-FP HRD-UP resource districts located in provinces with a b. Total stock of roads high share of similar districts (HRD-FP) again show significant results of adding between 38 140 and 51 kilometers of roads. 120 Di erence in roads added (km) For high-resource districts located in prov- 100 inces with few similar districts (HRD-UP), 80 however, the spillover impact vanishes almost 60 completely. And, in provinces with a high 40 share of high-resource districts, even districts 20 that are themselves low resource (LRD-FP) 0 added significantly more highways. In addi- –20 tion, the total kilometers of roads being built –40 are higher for LRD-FP districts. 1997 2003 2011 2014 1997 2003 2011 2014 1997 2003 2011 2014 The quantitative results hold very generally HRD-FP LRD-FP HRD-UP when the measure of the stock of roads is nor- Sources: Calculations based on data from Badan Pusat Statistik (Statistics Indonesia) and Indonesia’s malized by population: high-resource districts National Socio-Economic Survey (SUSENAS). increased their per capita kilometers of roads Note: HRD-FP = high-resource districts in favored provinces; LRD-FP = low-resource districts in favored provinces; HRD-UP = high-resource districts in unfavored provinces. Favored provinces are relatively more after 2001, with the effect those with more than 50 percent high-resource districts. The bars show the difference in kilometers being about twice as large for high-resource of road added compared to low-resource districts in unfavored provinces (LRD-UP). Coefficients districts in favored provinces.9 represented by the solid bars are statistically significant. 234  TIME TO ACT Using this per capita measure of infrastruc- FIGURE 6.13  Higher resources do not differentially ture, which is therefore not scale dependent, increase market access, 1997–2014 also allows an alternative test of the presence of spillovers. If spillovers across districts exist, 200 the size of the coefficient should be bigger in Di erence in di erences estimate estimates using the province rather than the 150 district as the unit of observation.10 Indeed, using the 88 province-level observations leads 100 to coefficients for 2003, 2011, and 2014 that are between 20 and 100 percent larger than 50 their district-level counterparts. Do higher resources, which are associated 0 with additional roads, improve a district’s accessibility to markets? With market access –50 measured by how well a district is connected 1997 2003 2011 2014 1997 2003 2011 2014 1997 2003 2011 2014 to the rest of the country, the answer seems to HRD-UP HRD-FP LRD-FP be negative (­ figure 6.13).11 Sources: Calculations based on data from Badan Pusta Statistik (Statistics Indonesia) and Indonesia’s Other infrastructure outcomes are partially National Socio-Economic Survey (SUSENAS). consistent (figure 6.14). Relative to low- Note: The bars indicate, for each year, the differential evolution of market access compared to resource districts in unfavored provinces, elec- low-resource districts in unfavored provinces (that is, compared to LRD-UP districts). The blue dots are the point estimates, with a positive value denoting that treated districts’ position improved tricity coverage increases significantly in in relative terms, and the blue lines at the ends of bars show the 95 percent confidence interval. high-resource districts in favored provinces, HRD-UP = high-resource districts in unfavored provinces; HRD-FP = high-resource districts in favored an effect that extends to low-resource districts provinces; LRD-FP = low-resource districts in favored provinces. in the same provinces, but clearly decreases Finally, additional evidence using aggre- for high-resource districts in unfavored prov- gate consumption, production, and produc- inces. In contrast, no clear spillover pattern tivity measures at the district level shows no exists for water or access to toilets. significant difference between high-resource, Do additional resources affect firm-level low-population districts in different prov- outcomes, including employment, value inces, despite the evidence for spillovers just added, and labor productivity? In data from discussed. the Indonesian manufacturing census Summing up: (Statistik Industri) for 1990–2015, firms in high-resource, low-population districts were •  The additional resources channeled to already on a downward trend before decen- low-population districts after decentraliza- tralization, meaning that they were losing tion in 2001 helped them increase the ground with respect to firms in other dis- amount of monetary investments targeted tricts (figure 6.15). Consistent with the to infrastructure. absence of market access improvement, this •  The actual impact of these additional trend did not notably change after 2001, investments on the availability of the when additional resources increased the stocks of different types of intercity roads, investment in roads. as well as on household access to water Would additional roads matter more to and toilets, is limited. Regarding the total firms in industries that rely more on them? stock of roads, there is no strong evidence In  “triple difference estimates” based on that more spending produced additional comparing, across districts, firms in vehicle-­ kilometers of road in these low-population intensive industries with firms in other indus- districts. Regarding an intermediate cate- tries, there is no evidence of dividends from gory of roads—highways—there is some additional road investment (­ f igure  6.16). evidence of a positive effect. In fact, the relative situation of firms in high- •  This evidence does not mean that there resource, low-population districts deterio- were no improvements in districts favored rated after 2001. by decentralization (also implied by I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   235 FIGURE 6.14  Electricity coverage, but not water or FIGURE 6.15  Additional resources sanitation coverage, increases in districts in favored in low-population districts did not provinces, 1996–2016 improve employment, value added, or productivity, 1990–2015 a. Share of households with access to electricity 0.25 a. Employment 0.25 Difference in differences estimate 0.20 Difference in differences estimate 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0 0 –0.05 –0.05 –0.10 –0.10 –0.15 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 –0.20 HRD-UP HRD-FP LRD-FP 1990 1995 2000 2005 2010 2015 b. Value added b. Share of households with access to water 0.5 0.10 0.4 Difference in differences estimate Difference in differences estimate 0.05 0.3 0.2 0 0.1 –0.05 0 –0.1 –0.10 –0.2 –0.15 –0.3 –0.20 –0.4 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 –0.5 1990 1995 2000 2005 2010 2015 HRD-UP HRD-FP LRD-FP c. Productivity 0.5 c. Share of households with access to toilets Difference in differences estimate 0.10 0.4 Difference in differences estimate 0.3 0.05 0.2 0.1 0 0 –0.05 –0.1 –0.2 –0.10 –0.3 –0.15 –0.4 1990 1995 2000 2005 2010 2015 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 1996 2000 2004 2008 2012 2016 HRD-UP HRD-FP LRD-FP Sources: Calculations based on data from Indonesia’s manufacturing census (Statistik Industri ), Badan Pusat Statistik (Statistics Indonesia) and Indonesia’s National Socio-Economic (SUSENAS). Sources: Calculations based on data from Indonesia’s manufacturing census (Statistik Industri), Badan Note: The bars indicate, for each year, the differential evolution of the Pusat Statistik (Statistics Indonesia) and Indonesia’s National Socio-Economic Survey (SUSENAS). point variables of interest between high- and low-resource districts. The Note: The bars indicate, for each year, the differential evolution of the variable compared to low-resource districts blue dots are the point estimates, with a positive value denoting that the in unfavored provinces (that is, compared to LRD-UP districts). The blue dots are the point estimates, with a high-resource districts improved, on average, relative to the low-resource positive value denoting that treated districts’ position improved in relative terms, and the blue lines at the end districts, and the blue lines at the ends of the bars show the 95 percent of the bars show the 95 percent confidence interval. HRD-UP = high-resource districts in unfavored provinces; confidence interval. HRD-FP = high-resource districts in favored provinces; LRD-FP = low-resource districts in favored provinces. 236  TIME TO ACT FIGURE 6.16  Firms that used more ­ hapter 2), although as discussed in the c vehicles in high-resource districts first part of the chapter, the overall drop in did not improve employment, value the level of investments starting in the late added, or productivity, 1990–2015 1990s means that progress was slow. However, the improvement has been no a. Employment faster than in more populous districts that 0.15 were less favored (in terms of transfers) by 0.10 decentralization. Difference in differences estimate 0.05 •  There is evidence of cross-district spill- 0 overs: higher district spending generates –0.05 larger stocks of roads only when it hap- –0.10 pens simultaneously for enough districts in the same province. –0.15 •• There is no evidence that the larger invest- –0.20 ments in infrastructure targeted since –0.25 decentralization to less populous places –0.30 have improved market access or changed 1990 1995 2000 2005 2010 2015 the preexisting deterioration of firm-level b. Value added outcomes. This finding is even true for firms 0.5 in industries more dependent on vehicles. 0.4 Difference in differences estimate 0.3 These findings show that the channeling of 0.2 resources to infrastructure under decentraliza- 0.1 tion has failed to achieve its purpose for at 0 least two complementary reasons. First, –0.1 resources are targeted toward lagging districts, –0.2 defined as those with low population (often –0.3 districts with the highest poverty rate but not –0.4 –0.5 the highest number of poor people), but the –0.6 efficiency of the spending is not considered. 1990 1995 2000 2005 2010 2015 Isolated investments have lower returns, and c. Productivity failing to support interdistrict coordination, 0.4 possibly at the province level, in distributing resources leads to low returns. Spillovers do Difference in differences estimate 0.3 not materialize, defeating the purpose of trans- 0.2 fer structures defined under decentralization. 0.1 Second, the low overall returns imply that 0 investment failed to help lagging districts –0.1 overcome the advantage of more populated places in having growing numbers of jobs, –0.2 value added, and labor productivity. The dis- –0.3 advantage of lagging districts persists even –0.4 when they benefit from positive spillovers. 1990 1995 2000 2005 2010 2015 The key question raised by these results is about the underlying channels explaining Sources: Calculations based on data from Indonesia’s manufacturing census (Statistik Industri ), Badan Pusat Statistik (Statistics Indonesia) and Indonesia’s this lack of impact of additional resources. National Socio-Economic Survey (SUSENAS). Although precise information on the spe- Note: The bars indicate, for each year, the differential evolution of the point cific bottlenecks is lacking, the results in variables of interest between high- and low-resource districts. The blue dots are the point estimates, with a positive value denoting that the this chapter indicate that resources are not high-resource districts improved, on average, relative to the low-resource the binding constraint in terms of infra- districts; and the blue lines at the ends of the bars show the 95 percent structure and basic service provision for less confidence interval. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   237 populous districts. Instead, the binding con- their endowments. In particular, the main straints are likely to be related to low local ­ district-level characteristics that matter for capacity for planning and executing infra- welfare are, by order of importance, market structure projects, which prevents translat- access, penetration of roads, and school ing resources into results. This, in turn, infrastructure. raises questions about simple strategies This chapter helps to make sense of these aimed at raising additional resources for facts, both by documenting the relative role local governments—for example, by extend- of household endowments compared with ing their access to outside finance. infrastructure endowments and by explain- ing why welfare differences have not been arbitraged away. First, the low efficiency of Implications for productivity infrastructure investments in places that Despite a massive direction of resources benefitted from more resources after decen- toward remote, less populated districts, these tralization prevented these districts from places have not experienced a relative catching up in terms of infrastructure improvement in market access—mostly endowments. Second, the lack of mobility because they failed to transform the larger stemming from limited improvements in monetary investments into additional infra- connectivity may have played a role in the structure. As a result, lagging places were also fact that these differences were not arbi- unable to reverse the divergence in productiv- traged away.13 ity that started in the 1990s. The findings resonate strongly with those in chapters 3 and 4 on the determinants of Barriers to migration productivity and welfare across districts. A few recent studies document both the barri- In chapter 3, market access had only a sig- ers to mobility posed by poor connectivity nificant positive impact on productivity for and the resulting aggregate productivity urban districts in the peripheries of the mul- cost.14 These barriers could be eased both by tidistrict metro areas during 2008–15, pos- reducing the cost of migrating and by increas- sibly indicating that connective ing potential benefits. Several recent studies infrastructure had not been placed where it have documented these different aspects in was likely to have a positive effect. In chap- the Indonesian context. ter 4, all the regions, in particular Maluku, Gertler et al. (2015) report a positive and Nusa Tenggara, and Papua, and to a lesser significant relationship between the change extent Kalimantan and Sulawesi and the in road quality between 1995 and 2000 and rest of Jawa-Bali, had lower welfare on several measures of local immigration. average than DKI Jakarta (that is, than the Locations with more road improvements core of the Jakarta multidistrict metro experienced higher population growth and a area).12 Other metro cores also appeared to higher share of the recent (less than five do better than Jakarta in 2015. Urban years) migration. periphery areas did almost as well as the Bazzi (2017) estimates the extent to which Jakarta core, whereas nonmetro rural areas, liquidity constraints limit migration for poor rural periphery areas, and other nonmetro rural Indonesian households. Although his urban areas did much worse. The welfare estimates relate to international migration, differences corresponded much more to looking at typical destinations for local variation in returns to characteristics (pro- households, they provide a useful bench- ductivity) across districts than to variation mark. The estimated costs range from two in the level of these characteristics. In other years of total expenditures for rural village words, the different characteristics of places households to up to five years in the poorest, led ­individuals with otherwise comparable more remote regions. The migration flows endowments to enjoy different returns to from villages with a majority of relatively 238  TIME TO ACT poor, small landholders are shown to increase significantly following positive rain- Infrastructure, the fall shocks, indicating binding income decentralization constraints. conundrum, and the need These constraints to migration are likely to have a large productivity cost. Bazzi et al. for coordination (2016) study the 1979–88 Indonesian The current state of Indonesian transport Transmigration Program, which relocated infrastructure and the consequences described two million voluntary migrants from Jawa in this chapter largely result from the coun- and Bali to the Outer Islands (Kalimatan, try’s history of policy making. During the Maluku, Nusa Tenggara, Papua, Sulawesi, decades of Suharto’s “New Order” (1966–98), and Sumatera). Bazzi et al. first show that, public administration decision making was when migrants were assigned plots with a top-down, hierarchical, and strongly central- higher agroclimatic similarity to their envi- ized. Several of the successive national five- ronment of origin, the productivity of rice cul- year plans ( Rencana Pembangunan Lima tivation was significantly enhanced. They Tahun-Repelitas) made infrastructure a prior- then estimate that optimizing the origin–­ ity: repelita II (1974–79) targeted infrastruc- destination match between migrant skills and ture on islands outside Jawa and growth in plots would have generated a 27 percent primary industries; repelita V (1989–94) higher aggregate rice yield. focused on growth in telecommunications Bryan and Morten (forthcoming) estimate and transportation infrastructure; and repel- the aggregate productivity loss from barriers ita VI (1994–98, left unfinished in the reform) to mobility, consisting both of movement shifted the focus to infrastructure supporting costs and differences in amenities, using data foreign investment and free trade. for 1995, 2011, and 2012. Using a structural Within that framework, the central gov- model, they first find that migration costs in ernment in Jakarta defined development Indonesia are substantial—about 2.5 times projects and directed funds through presi- those in the United States. They then estimate dential instructions. Central government the implications of policies that would reduce institutions managed planning and person- migration costs, equalize amenities across nel. Two programs targeted transport devel- space, or both. In their framework, reducing opment on the basis of road length, traffic migration costs to the U.S. level would lead to density, and ­ m aintenance cost: the Road an overall productivity gain of 7.5 percent, Grant for Municipalities and Districts and equalizing amenities across places to a gain of the Road Grant for Provinces (Nasution 12.7 percent, and doing both simultaneously 2016). Execution of these policies was to a sweeping gain of 21.7 percent. Finally, assigned to subnational governments at the Bryan and Morten (forthcoming) also show village, district, and province levels without that these average gains hide significant het- giving them any control over either design or erogeneity across origin locations. For exam- resources.15 ple, the upper bound of their estimated The Indonesian decentralization was productivity gain for reducing migration costs probably one of the most dramatic ever is 68 percent; for equalizing amenities, implemented. It was, however, accompanied 88 ­percent; and for combining both policies, by inertia in the functioning of central pub- 104 percent. lic institutions so that the new decentral- Although not all the costs captured in these ized decision-making process coexisted with studies are due to suboptimal infrastructure, the pre-1998 centralized decision-making the estimates show that the deadweight loss culture. Even today, the old central plan- from the inefficient spatial targeting of infra- ning process is still in place, with annual, structure investments discussed above may be five-year, and long-term plans; and a large substantial. fraction of the administrative staff across I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   239 the country is centrally managed by the problems, provincial governors and the Ministry of State Apparatus Empowerment Ministry of Agrarian Affairs and Spatial and Bureaucratic Reform. Decision points Planning have no direct authority over dis- have multiplied, as witnessed by the eight trict or city governments. As a result, dis- parallel planning processes described in trict and city governments tend not to give chapter 5 and their less-than-perfect coor- priority to land acquisition of provincial dination. This fragmentation has hampered or national projects, whereas provinces the ability of subnational governments to give lower priority for district projects. adequately respond to citizen needs, some- •  For mass-transit systems, provinces are in times resulting in the poor quantity and charge, whereas, for feeder buses, local quality of local service delivery. governments are in charge. A critical noncoordination of public insti- •• The Ministry of Transportation has often tutions and levels has followed: extended in-kind bus grants to cities to support local public transportation; how- •  Roads are a city or district responsibility ever, the buses offered are often mis- when confined to local boundaries, a prov- matched to local conditions. ince responsibility when crossing more than one district, and a central govern- Addressing the coordination issues is critical ment responsibility when crossing more to boosting the efficiency of infrastructure than one province. investments. Rethinking resource distribu- •  Pedestrian ways are under the responsibil- tion across space is necessary, both skewing ity of the Ministry of Public Works and the allocation more toward very populated Housing, whereas roads belong to the places with critical bottlenecks and redefin- Ministry of Transportation. This clear ing responsibilities for types of infrastructure instance of fragmented responsibility to improve coordination across locations across transport modes threatens the effi- and modes. Because most of the coordina- ciency of investments in either of them. tion issues result from the successive addi- •  Incentives for coordination among stake- tion of institutional layers on top of existing holders involved in land acquisition for ones, the first step lies in identifying redun- infrastructure projects are grossly mis- dancies and simplifying the governance aligned (World Bank 2018). Among other structure. 240  TIME TO ACT Annex 6A Aggregate infrastructure investment and stock data World Bank (2015) provides infrastructure that have reported their budget spending investment and stock data by source and sec- to the ministry, compiled in the Regional tor from 1995 to 2012, covering core infra- Financial Information System structure, defined as infrastructure that serves •  State-owned enterprises: Reported capital the public interest in the transport, energy, expenditure in their financial statements telecommunications, irrigation, and water •• Private sector: World Bank Private Partici- supply and sewerage sectors. Infrastructure pation in Infrastructure (PPI) database. investment is defined as investment made by Although gauging data accuracy is the general government (through central and ­ ifficult for aggregate infrastructure spend- d subnational government budgets), state- ing figures, the numbers used here compare owned enterprises, and private-sector entities well with those computed using Gross (generally through PPPs) in core economic Fixed  Capital Formation of General public infrastructure. The specific data sources Government (GFCF-GG) from national are the following: accounts. With PPI figures added in both •  Central government: The government’s cases to account for private sector activity, annual audit report or, if not available, aggregate investment reported here for original or revised budget data 2011, for example, is 2.6 percent of GDP, •  Subnational governments: Ministry of compared with 3.1 percent of GDP for Finance for districts, cities, and provinces GFCF-GG figures. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   241 Annex 6B Econometric methodology How do Indonesia’s policy making transfers allowing cheaper and faster movement dependent on population affect district-level between districts, the analysis hypothesizes outcomes? And how does that effect vary that high-resource districts surrounded by depending on where the district is located? As similar high-resource districts may get higher explained in the main text, after the reform, returns from road investments than high- districts with a relatively small population resource districts surrounded by low-resource received larger per capita transfers compared ones. As a result, spending would be more with districts with a large population. To test effective in a context where clusters of high- how this policy affected various outcomes of resource districts exist (for example, in a interest at the district level, Jascisens and “favored” province with many such Straub (2018) estimate the following districts). equation: How does the effect at the district level vary depending on whether the district is yit = ∑β d + η + γ τ i i t + ε it , located in a favored province or not? Again, some care is needed to define what a favored τ province means. The analysis experiments where i indexes districts and t years, h i with two procedures (here, too, results from denotes district fixed effects, and g t denotes the two exercises are very similar). The more time fixed effects. Our outcome of interest is basic procedure defines a province as favored yit, and di equals 1 for districts affected by the if its population in 1995 is below the 1995 reform and zero otherwise. Some care is median or below the current population needed to define a district “affected by the median, and as unfavored otherwise. This reform” (“high resource”). We define a dis- procedure, however, might not capture the trict to be affected by the reform if its 1995 policy change at the province level if the prov- population is below the 1995 median or if its ince with a very low population is composed current population is below the current of a relatively few large-population districts. median (the results from the two exercises are To overcome this challenge, the analysis remarkably similar). To interpret results as experiments with the second procedure, causal we need to make the classical “parallel which defines a province as favored if more trends” assumption that absent the reform the than 50 percent of its districts are high outcome of interest would have evolved in the resource, and as unfavored otherwise. Thus, same manner in both high- and low-resource to understand how the effect on the district districts. Readers should interpret obtained varies depending on whether the district is estimates as an effect of the policy in year t located in a favored or an unfavored prov- (notice that t spans years before the policy, ince, the analysis estimates the following during which we should not expect any dif- regression: ferential effects, an indirect test of the parallel trend assumption). The inclusion of district fixed effects takes yit = ∑α d + ∑β d + ∑ξ d τ i2 τ i3 τ i 4 + ηi + γ t + ε it , τ τ τ care of factors not changing for a given dis- trict over time (for example, its specific geog- where all the notation corresponds to that raphy, distance from Jakarta, or other used in the previous equation and di2 denotes conditions affecting the construction costs high-resource districts located in unfavored that are fixed over time), whereas the inclu- provinces; di3, high-resource districts located sion of time fixed effects controls for factors in favored provinces; and di4, low-resource common, at a given time, to all districts (such districts located in favored provinces. as macroeconomic shocks). Comparing obtained estimates a t and b t Next, because roads are typical connective explains whether it matters where the high- infrastructure whose value depends on resource district is located. 242  TIME TO ACT Annex 6C Results of the econometric model TABLE 6C.1  Differential effect of 1995 population on roads Total Motorway Principal highway Highway Other (1) (2) (3) (4) (5) Treated x 1997 1.08 –0.41 –2.05** 4.93** –1.39 (0.83) (0.27) (0.82) (2.39) (2.14) Treated x 2003 –5.39 –0.67** –4.22 10.07 –10.58** (7.27) (0.32) (4.96) (11.38) (4.71) Treated x 2011 –2.00 –3.60*** –16.41*** 22.08** –4.06 (9.94) (0.80) (5.52) (10.18) (7.80) Treated x 2014 –5.49 –4.71*** –11.69** 20.90* –9.99 (18.21) (1.08) (5.18) (10.93) (12.82) Fixed effects: District Y Y Y Y Y Year Y Y Y Y Y Observations 1,408 1,408 1,408 1,408 1,408 Adjusted R2 0.780 0.686 0.629 0.652 0.625 Note: Differential effect of population in 1995 on the development of roads. Treated (or high-resource) districts are those with population below median in 1995. All dependent variables are measured in kilometers. Standard errors in parentheses, robust and clustered at the district level. Level of significance: *** p<0.01, ** p<0.05, * <0.1. TABLE 6C.2  Differential effect of current population on roads Total Motorway Principal highway Highway Other (1) (2) (3) (4) (5) Treated –14.04 2.30*** 19.26*** –24.44** –11.16 (20.26) (0.60) (5.02) (9.88) (16.03) Treated x 1997 3.62 –0.58* –3.43** 10.00** –2.38 (3.11) (0.29) (1.35) (4.66) (2.24) Treated x 2003 –1.35 –0.87** –5.25 16.72 –11.95** (7.39) (0.35) (5.01) (11.26) (4.89) Treated x 2011 1.25 –3.88*** –16.64*** 31.30*** –9.53 (9.92) (0.84) (5.23) (10.17) (7.64) Treated x 2014 –7.55 –5.24*** –9.21 27.43** –20.53 (18.44) (1.11) (5.64) (11.40) (12.96) Fixed effects: District Y Y Y Y Y Year Y Y Y Y Y Observations 1,420 1,420 1,420 1,420 1,420 Adjusted R2 0.779 0.688 0.629 0.653 0.627 Note: Differential effect of the population on the development of roads. Treated (or high-resource) districts are those with below median population. All dependent variables are measured in kilometers. Standard errors in parentheses, robust and clustered at the district level. Level of significance: *** p<0.01, ** p<0.05, * <0.1. I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   243 TABLE 6C.3  Differential effect of current population on roads—triple difference results, urban versus rural Total Motorway Principal highway Highway Other (1) (2) (3) (4) (5) Treated x 2003 3.34 –1.18 6.94 –11.77 9.35 (10.63) (0.87) (7.53) (16.01) (7.82) Treated x 2011 0.69 –4.10* 16.82** –24.63* 12.60 (14.40) (2.25) (7.79) (14.11) (11.57) Treated x 2014 16.83 –2.40 2.61 –20.84 37.46* (27.32) (2.90) (9.10) (17.17) (19.45) Fixed effects: District Y Y Y Y Y Year x Urban Y Y Y Y Y Year x Treated Y Y Y Y Y Urban x Treated Y Y Y Y Y Observations 1,145 1,145 1,145 1,145 1,145 Adjusted R2 0.793 0.699 0.583 0.668 0.624 Note: Triple difference estimation, urban versus rural. Treated (or high-resource) districts are those with below median population. All dependent variables are measured in kilometers. Standard errors in parentheses, robust and clustered at the district level. Level of significance: *** p<0.01, ** p<0.05, * <0.1. Notes population, the Indonesian traffic-related death rate is below the world average of 17.4 1. As described in chapter 1, there has been a and the rate elsewhere in the region, includ- slowdown in urbanization since the turn of ing China (18.8), India (16.6), Malaysia the century. (24.0), and Vietnam (24.5) but above the 2. Poverty is defined as the proportion of people rate for Singapore (3.6) and the Philippines with per capita household consumption below (10.5). It has, however, risen steadily from the national poverty line, and vulnerability less than 5 in 2004. is defined as the proportion of people with 6. Information Center for Strategic Food Price consumption above the poverty line but less (PIHPS) data, cited in Sukaesih (2016). than 1.5 times the poverty line. Individuals, 7. Revisiting the definition of commercial meanwhile, are considered to belong to the ports will most likely require stipulation in middle class if their per capita household con- the shipping law (UU 17/2008) itself or at sumption is more than 3.5 times the poverty least in the government regulation on ports line. See chapters 2 and 4 for a more detailed (PP 61/2009). analysis of the evolution of poverty since the 8. Transfers to villages suffer from the same beginning of the 1990s. bias, with those in the lowest population 3. For more information on the TomTom Traffic decile getting 30 times more revenue per cap- Index, see https://www.tomtom.com/en_gb​ ita than those in the highest decile. /trafficindex/. 9. In this case, the effect for favored provinces 4. Statistik Transportasi Jakarta, quoted in is only there when they are defined as having Hanna, Kreindler, and Olken (2017). mostly high-resource districts (as opposed to 5. Congestion has been associated with more having above-median population). traffic accidents. In 2014, about 100,000 10. The comparison of econometric estimates of reported accidents resulted in 28,000 the effect of infrastructure across geographic fatalities (Jusuf, Pulung Nurprasetio, and areas of different size as a way to quantify Prihutama 2017, using National Traffic spatial spillovers goes back at least to Holtz- Police Corps data). At 12 per 100,000 Eakin (1994). 244  TIME TO ACT 11. Market access measures how well connected Reforms across the Developing World.” Policy a district is to areas of population concentra- Research Working Paper 8235, World Bank, tion across the country through the national Washington, DC. road and domestic ferry system. See annex Freedom House. 2016. Freedom on the Net 2016: 3A in chapter 3 for technical details. Silencing the Messenger: Communication Apps 12. Welfare is computed by adjusting household under Pressure. Washington, DC: Freedom expenditures by the ratio of the poverty line House. https://freedomhouse.org/sites/default​ of a given region to the poverty line of DKI /­files​/FOTN%202016%20Indonesia.pdf. Jakarta. Gertler, P., T. Gracner, M. Gonzalez-Navarro, and 13. Of course, other factors may also limit mobil- A. Rothenberg. 2015. “Road Quality and Local ity, such as difficulties in obtaining identity Economic Activity: Evidence from Indonesia’s cards (see chapter 4). Highways.” Working Paper, University of 14. Note that, as documented in chapter 1, net California, Berkeley, CA. rural–urban migration accounted for only a Hanna, R., G. Kreindler, and B. Olken. 2017. relatively small share of Indonesia’s urban “Citywide Effects of High-Occupancy Vehicle population growth over the period 2000–10. Restrictions: Evidence from ‘Three-in-One’ in 15. The process is described in more detail in Jakarta.” Science 357 (6346): 89–93. Nasution (2016) and World Bank (2014). Holtz-Eakin, D. 1994. “Public-Sector Capital and the Productivity Puzzle.” Review of Economics and Statistics 76 (1): 12–21. References Jascisens, V., and S. Straub. 2018. “Decentralization That Delivers? Estimating the Infrastructure Akamai. 2017. Q1 2017 State of the Internet Impact of Indonesia Big-Bang Decentralization.” Report. Cambridge, MA: Akamai Technologies. Background paper prepared for this report, https://content.akamai.com/uk-en-pg9141​ World Bank, Washington, DC. -q1-soti-connectivity.html. Jusuf, A., I. Pulung Nurprasetio, and A. Prihutama. Bardhan, P., and D. Mookherjee. 2015. 2017. “Macro Data Analysis of Traffic “Decentralization and Development: Dilemmas, Accidents in Indonesia.” Journal of Engineering Tradeoffs, and Safeguards.” In Handbook and Technological Sciences 49 (1): 132–43. of Fiscal Federalism, edited by E. Ahmed and Kis-Katos, K., and B. S. Sjahrir. 2017. “The Impact G. Brosio, 461–70. Northampton, MA: Edward of Fiscal and Political Decentralization on Local Elgar. Public Investment in Indonesia.” Journal of Bazzi, S. 2017. “Wealth Heterogeneity and the Comparative Economics 45(2): 344–65. Income Elasticity of Migration.” American Lewis, B. 2014. “Twelve Years of Fiscal Economic Journal: Applied Economics 9 (2): Decentralization: A Balance Sheet.” In Regional 219–55. Dynamics in a Decentralized Indonesia, edited Bazzi, S., Gaduh, A., Rothenberg, A., and M. Wong. by Hal Hill, 135–55. Singapore: Institute of 2016. “Skill Transferability, Migration, and Southeast Asian Studies. Development: Evidence from Population McKinsey. 2012. “The Archipelago Economy: Resettlement in Indonesia.” American Unleashing Indonesia’s Potential.” McKinsey Economic Review 106 (9): 2658–98. Global Institute. Bryan, G., and M. Morten. Forthcoming. “The McLeod, R. H., and Fadliya. 2010. “Fiscal Transfers Aggregate Productivity Effects of Internal to Regional Governments in Indonesia.” Working Migration: Evidence from Indonesia.” Journal Paper 2010/14, Department of Economics, of Political Economy. Australian National University, Canberra. Fabre, A., and S. Straub. 2018. “The Economic Nasution, A. 2016. “Government Decentralization Impact of PPPs in Infrastructure, Health and Program in Indonesia.” Working Paper No. Education: A Review.” Unpublished paper, 601, Asian Development Bank Institute, World Bank, Washington, DC. Tokyo. Fay, M., D. Martimort, and S. Straub. 2018. New Frontier Solutions. 2017. “Assessment “Funding and Financing Infrastructure: The of Benefits of Improving Ports in Eastern Joint Use of Public and Private Finance.” Policy Indonesia.” Report prepared for the World Bank. Research Working Paper 8496, World Bank, New Frontier Solutions, Singapore. Washington, DC. P a l , S . , a n d Z . Wa h h a j . 2 0 1 7 . “ F i s c a l Foster, V., S. Witte, S. Banerjee, and A. Moreno. Decentralization, Local Institutions and Public 2017. “Charting the Diffusion of Power Sector Good Provision: Evidence from Indonesia.” I n frastructure a n d P olicies to C o n n ect the P ortfolio of P laces   245 Journal of Comparative Economics 45 (2): Post, February 5. https://www.thejakartapost​ 383–409. .com/news/2015/02/05/jakarta-has-worst​ Pritchett, L. 2000. “The Tyranny of Concepts: -traffic-world.html. CUDIE (Cumulated, Depreciated, Investment World Bank. 2012. “Investing in Indonesia’s Roads: Effort) Is Not Capital.” Journal of Economic Improving Efficiency and Closing the Financing Growth 5 (4): 361–84. Gap.” Road Sector Public Expenditure Review Sukaesih, M. 2016. “Reducing Regional Price PER 73303, World Bank, Washington, DC. Disparity.” The Jakarta Post, September 21. http:// ———. 2014. “Assessment of Pendulum Service www.thejakartapost.com/academia/2016/09/21​ Concept and Initial Scenarios.” World Bank. /reducing-regional-price-disparity.html. ———. 2015. “Technical Note: Estimating Van Tuijll, D. 2018. “Lost in Transition. Infrastructure Investment and Capital Impediments to Complete Implementation of Stock in Indonesia.” Note, World Bank, Port Reform in Indonesia.” Unpublished paper. Washington, DC. World Bank, Jakarta. ———. 2018. “Indonesia Infrastructure Sector Wardhani, D. A., and I. Budiari. 2015. “Jakarta Assessment Program.” Report, World Bank, Has ‘Worst Traffic in the World.’” The Jakarta Washington, DC. 246  TIME TO ACT Connected and Integrated Cities: A Focus on Housing and Transport 7 KEY MESSAGES For metro and urban areas to benefit from urban- spatial planning. To take steps in this direc- ization and achieve greater prosperity, inclusive- tion, the government should continue efforts ness, and livability, they must promote connected to clarify property rights and improve the and integrated growth. People must be connected efficiency of land management; scale up cur- to jobs, opportunities, and services within urban rent efforts to improve data, tools, and capac- areas. Spatially coordinated urban planning can be ity at the local level for better evidence-based the difference between cities where these connec- urban planning; and ensure the operational- tions materialize and cities where people and busi- ization of plans by building better links from nesses are left disconnected from each other and spatial and sectoral plans to socioeconomic from markets. This chapter highlights challenges development outcomes. in providing affordable housing and sustainable •  Inadequate access to affordable housing transportation systems, two sectors critical for the remains a barrier for intraurban connectivity development of connected metro and urban areas. and integration. Unaffordability has contrib- uted to sprawling urban areas, unsatisfactory •  With Indonesia’s decentralization, most services, and proliferating slums. Key policies urban and spatial planning responsibilities for housing would improve technical capacity were transferred to provincial and district/ at the local level for coordinated and spatially city governments. However, urban and spa- appropriate land and housing planning, and tial planning aimed at fostering more con- for better implementation of permitting pro- nected and integrated growth has remained cesses and construction quality and resilience weak. Building cities that are better con- standards; promote a holistic agenda based nected, and thus more prosperous, inclusive, on the specific needs of different places and and livable, will require better integrated households; and reform the housing finance 247 sector to target unserved groups and to •• The diverse needs of different types of place are crowd in the private sector. key to improving infrastructure and services •  Insufficient urban transport is an important for better urban connectivity and integration. cause of poor accessibility and mobility For multidistrict metro areas, mechanisms to within cities, contributing to strong conges- coordinate planning, investment, and manage- tion forces, especially in metropolitan areas. ment of housing and transport infrastructure Indonesia has focused on expanding roads can integrate pockets of poverty, especially in for private vehicles, making limited invest- the peripheries, into the city. For single-district ment in public and nonmotorized transport. metro areas, guiding development, services, For transport, policies should strengthen the and transport infrastructure to encourage national government’s role in guiding urban density rather than sprawl can prevent growth ­ transport policy and national coverage of patterns that leave people disconnected from investment gaps; enhance local capacity to services and jobs. For growing nonmetro urban plan, operate, and maintain urban transport areas, planning ahead, ensuring rights of way, systems; and promote transit-oriented devel- coordinating housing and land use, and opment that encourages densification in making appropriate transport decisions can ­ transport corridors and reduces travel keep urban areas connected and integrated as demand and private vehicle use intensity. they grow. As stressed in previous chapters, the ACT places, need to be connected through the basic policy principles—Augment, Connect, provision of affordable and ­ w ell-located and Target—provide a three-pronged housing, transport infrastructure, and basic approach toward guiding policies for better services. urbanization outcomes. Just as the previous Instead, Indonesia’s urban growth has chapter focuses on connectivity and integra- given rise to congestion forces that have tion between different places within the port- resulted in fragmentation, and thus high folio of places, this chapter takes a look inequality and persistent spatial disparities within urban areas to identify key bottlenecks within cities. As described in chapter 2, even to connectivity and integration. Why is a if urban areas are more prosperous than rural focus on intraurban connectivity and integra- areas, they remain poor compared to urban tion relevant for achieving prosperity, inclu- areas in more developed countries, and there siveness, and livability? is also considerable heterogeneity across dif- Cities can foster prosperity, inclusiveness, ferent types of urban area. For example, and livability through opportunities, ser- important disparities in welfare and basic vices, and amenities for urban residents and infrastructure between the core and peripher- firms. When well managed, the density in cit- ies of multidistrict metro areas persist. As provides firms with easy access to mar- ies ­ mentioned in chapter 2, more than 20 percent kets and suppliers. Density can also improve of all households and nearly 40 percent of the worker access to job opportunities and ser- poor in the urban peripheries of these metro vices, and it can facilitate the flow of capital areas lack access to safe drinking water. and ideas, leading to innovation and positive Disparities are also seen when comparing the agglomeration forces that promote produc- metro cores: whereas 7 percent of households tivity and prosperity (see box I.3 in the intro- in Jakarta’s core, DKI Jakarta (Daerah duction, and chapter 3). To realize such Khusus Ibukota Jakarta), do not have good potential and take full advantage of the ben- sanitation, 19 percent of households in other efits of urbanization, however, people and metro cores face this issue. Addressing these firms, within all areas in the portfolio of gaps requires strong investments in 248  TIME TO ACT infrastructure: recent estimates put the infra- Tata Ruang Wilayah or RTRW) and develop- structure financing gap above US$11 billion ment plans (such as Rencana Pembangunan (2014 prices).1 Institutional structures, espe- Jangka Menengah Daerah or RPJMD) often cially good urban and spatial planning sys- exist merely as statutory documents because tems, are a precondition to ensuring that they are not integrated with each other or resources are spent efficiently, which results with the respective sectoral plans. Even where in effective decisions that lead to greater con- mechanisms to promote coordination and nectivity and integration within urban areas. integration exist, implementation is weak and Building on the discussion of the institu- relies on spontaneous efforts. Further, plans tional challenges related to urban governance are constrained by gaps in the quality of stra- and financing in chapter 5, the first part of tegic inputs and by ineffective prioritization this chapter looks at the planning challenges of capital investments. to promote connected and integrated growth This section focuses on three challenges of faced by Indonesian cities: inefficient and the urban planning systems for fostering more complex land management systems, limited connected and integrated growth: (1) com- coordination across sectors and levels of gov- plex and unclear land management systems ernment, and limited local capacity to plan, that limit the availability of land for housing manage, and finance infrastructure provision. and infrastructure; (2) a sectoral approach to The second part of the chapter highlights the planning that has ignored the spatial dimen- bottlenecks in two key sectors that could fos- sions of urban areas and led to uncoordinated ter greater connectivity and integration— actions across sectors and levels of govern- housing and transport—and discusses policy ment; and (3) limited capacity of subnational options to improve their delivery and ensure governments for planning, managing, and full inclusion of all places and all people. financing infrastructure provision in cities. Planning for connected Complex and unclear land management systems limit the availability of land for growth housing and infrastructure Effective planning of cities, including for Effective urban and spatial planning for future growth, can make the difference accommodating urban growth requires, to between achieving greater connectivity and start with, strong institutions to manage land. leaving people and firms disconnected from The policies, regulations, and information each other and from jobs and services. It can used to plan land use and manage its transfor- make the difference between facilitating inter- mation will determine how cities grow and action and creating enclaves that segregate whether infrastructure and services will be people. Connectivity and integration are provided to all people within the portfolio of essential for inclusion, prosperity, and livabil- places. Indonesian cities present challenges in ity in urban areas. In Indonesia, however, land administration and complex manage- poorly coordinated and ineffective urban and ment systems that impose artificial controls spatial planning has kept cities from meeting over land available for development, thereby the needs and demands of their growing pop- contributing to congestion pressures in both ulations, thus undermining connected growth. land and housing markets. The government of Indonesia has tried to A complex and often contradictory legal strengthen planning by transferring authority framework constrains the availability of land and responsibilities from the national govern- for housing and infrastructure and unneces- ment to district governments. Currently, how- sarily restricts the fluidity of land markets. ever, Indonesia’s planning system lacks Indonesia’s land management system is char- strategic direction and is constrained by its acterized by competing laws and a lack of complexity (box 7.1; see chapter 5 for a gen- established legal hierarchy. As a result, land eral discussion of planning challenges); disputes and overlapping claims of control ­ district-level spatial plans (such as Rencana and ownership of land are common. Further, C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   249 BOX 7.1  Two key planning systems for Indonesian cities As chapter 5 stressed, the main context of plan- Second is the spatial planning system. With the ning in Indonesia is the decentralization that Spatial Planning Law of 2007 (Law 26/2007), the started in 2001. Decentralization shifted respon- planning system acknowledged the role of spatial sibilities and resources, including some related planning in managing rapid urbanization. The to urban and spatial planning, to district govern- law distributed spatial management authority ments. Two systems of planning are key for this to different levels of government, which develop chapter. First is the development planning sys- spatial plans for periods that vary according to tem, based on the National Development System the level of government. The plans set policies for Law (Law 32/2004) and subsequent develop- zoning, public transport, pedestrian networks, ment plans. The national development plan settlement expansion and density, and the alloca- (RPJPN) sets a 20-year vision and strategy for tion of green open spaces, among others. Plans the country, and local development plans set the are supposed to be elaborated with the support vision for the city or province. Bappenas (the of citizens, local stakeholders, and the private sec- Ministry of National Development Planning) tor. According to 2017 data from the Ministry is responsible for developing and implementing of Agrarian Affairs and Spatial Planning, 33 of national development plans, including medium- 34 provinces have provincial spatial plans, 375 term development plans (RPJMN), which are of 415 kabupaten have district spatial plans, and revised every five years, when a new adminis- 88 of 93 kota have city spatial plans. Although the tration comes into power. Locally, the Local numbers may look encouraging at first, detailed Development Planning Agency (Bappeda) is spatial plans have been falling behind. Only 90 responsible for local development plans and of about 1,400 detailed spatial plans for targeted medium-term development plans. These plan- areas have been developed because of the lack of ning documents, which parallel the spatial plan- technical capacity and suitable planning data at ning process, set the agenda for governance, both national and subnational levels. social services, infrastructure priorities, and socioeconomic development. Source: Dias and Chandra 2017; Rukmana 2015. the land administration system, which falls that add to property transaction costs. Delays under the Ministry of Agrarian Affairs and in permitting procedures, which are common, Spatial Planning, is uncoordinated and ineffi- can cost developers as much as 20 percent of cient. Vagueness and coordination issues the total building cost over 12 months (Bank regarding the cost and length of time needed of Indonesia 2017). This additional cost dis- for registration create frequent inconsistencies courages private investment in infrastructure, (Hudalah and Woltjer 2007), and poor land especially for the development of affordable management and the ambiguity of land status housing. In August 2016, the government of affect the implementation of spatial planning Indonesia released its 13th economic policy by making it hard to find land for infrastruc- package with a focus on reducing the permits ture expansion. As a result, infrastructure required for constructing affordable housing investments either get directed to land far from 33 to 11. This reform was intended to from city centers, contributing to fragmented decrease the permitting time from 981 to 44 urban growth, or are neglected altogether working days and lower costs for developing (Dias and Chandra 2017). housing by up to 70 percent (Indonesia Moreover, low capacity at the local level Investments 2016). However, implementation and lack of transparency have resulted in has been slow and has not yet yielded the cumbersome and bureaucratic procedures desired results.2 250  TIME TO ACT A sectoral approach to urban planning has Subnational governments have limited ignored the spatial dimensions of urban capacity for planning, managing, areas and led to uncoordinated actions and financing infrastructure service across sectors and levels of government provision in cities A coordinated multisectoral and flexible Reliable financing and strong capacity to approach to planning is necessary to manage effectively plan, manage, and coordinate the growth of urban areas and meet the funds are requirements for effective infra- increasing needs of their populations. structure development. District governments Planning must combine sectoral objectives need consistent sources of financing to pro- across levels of government and develop poli- vide the services and infrastructure that are in cies that fit the needs and spatial dimensions their mandate. And they need strong capital of a city. In many metropolitan and urban investment planning tools and capacity to areas in Indonesia, inflexible processes sepa- combine their strategic vision with existing rate planning for individual sectors and levels plans for land use and city expansion and to of government, fragmenting the planning pro- manage development within existing budgets cess itself. Further, spatial plans often lack and financial constraints. Such planning can strategic inputs, clear performance indicators, also enhance coordination across government and clear links to the goals stated in strategic levels, thereby better matching financial planning documents. resources with existing and future needs. For example, a recent assessment of As explained in detail in chapter 5, however, Semarang’s spatial plan found that it was major infrastructure gaps result because dis- inconsistent and incompatible with the trict governments lack the technical capacity dynamics of the city’s development and that it and the spatial planning tools to plan and was detached from other master plans, such manage funds and generate revenue from their as those for utilities and transport (Singh, own sources. Limited local government leader- Raghupathy, and Aurora 2017). 3 Thus, ship and capacity have exacerbated challenges although in theory sectoral master plans must in providing services and prompted the align with spatial plans, the reality is that such national government to guide local decisions plans, where they exist, tend to be misaligned. on infrastructure and planning. In 2009, to Part of the misalignment is caused by the dif- stimulate district revenue generation and build ferent timelines for preparing these plans. district government financial autonomy, Law Further, the spatial planning process has a 28/2009 provided local governments with new hierarchy. District plans must align with local tax instruments;5 however, as of 2017, higher-level (provincial or national) planning percent of the transfers still accounted for 36.7 ­ documents and meet certain (often unrealis- total state budget, and in 2015, own-source tic) guidelines, such as a mandated 30 percent revenue made up only 13 percent of total dis- allocation to open space, unaligned with city trict revenue (OECD 2016). needs. Many regional spatial plans are there- fore still under revision or have not yet been enacted by the government (Dias and Housing and transport: Chandra 2017 ). Key sectors for urban Other Indonesian cities can learn from the connectivity and experience of Balikpapan, Denpasar, and Semarang, which are taking steps to develop a integration more coordinated and multisectoral approach The development of well-located affordable to planning.4 The three cities are improving housing connected to transport infrastructure the technical capacity of staff for coordinated is essential to promote spatially connected city planning, policy, and infrastructure deci- and integrated cities that foster prosperity, sions and strengthening data collection and inclusiveness, and livability. Lack of afford- analysis for strategic and evidence-based deci- able housing can push the poor to live far sion making. from opportunities, services, and amenities; C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   251 when combined with lack of good transport promote the proliferation of slums or the networks, this distance can result in long, development of housing in peripheral loca- expensive commutes and the proliferation of tions, increasing urban sprawl. informal settlements, affecting especially the By contrast, urban transport systems and most vulnerable groups. The focus here is services have the power to shape urban growth thus on identifying the key challenges in the and are increasingly important for connectiv- housing and transport sectors and their ity, especially for large multidistrict metro effects on achieving better intraurban con- areas. They ensure peoples’ access to jobs, ser- nectivity and integration. For each sector, vices, and amenities within acceptable times policy recommendations are provided as a and costs, and connect firms to markets for road map that can guide the way toward both inputs and outputs. Further, they can encouraging more connected growth within have important environmental impacts, includ- urban areas. ing effects on air quality, climate change, and Housing and its location affect individuals’ open space preservation. Achieving sustainable welfare in many ways. Availability of afford- urban transport systems7 requires strong insti- able housing can impact access to basic tutional capacity to integrate transport policy services such as health and education, and ­ with land use planning. Failure to do so can it  may also affect access to opportunities. have significant bearings on accessibility and International experience suggests that hous- mobility, and through that contribute to rising ing influences the overall distribution of inequalities and strong congestion forces such people and, through that, economic and social ­ as increased travel times and costs, greenhouse mobility as it interacts with access to jobs, gas emissions and local pollutants, and traffic- services, and amenities. In a study of 12 major related accidents. European cities, Musterd et al. (2015) find that the spatial organization of housing is a key determinant of socioeconomic segrega- Limited access to well-located affordable tion, and thus has a bearing on poverty and housing has disconnected many from jobs and services inequality. For the United States, Chetty et al. (2014) find that poorly located housing, with The right to adequate housing is enshrined in relatively low access to public services and the Indonesian constitution and Law 1/2011 jobs, is associated with lower intergenera- on housing and settlements: “Every tional economic mobility. Chetty, Hendren, Indonesian citizen should live in a decent and and Katz (2016), meanwhile, find that chil- affordable settlement within a healthy, safe, dren from low-income families in the United harmonious, organized, integrated, and sus- States are most likely to succeed in neighbor- tainable environment.” The government has hoods that have lower concentrations of devoted considerable efforts to housing in p overty, lower income inequality, better ­ recent decades, including the ambitious One schools, lower crime rates, and a larger share Million Houses Program (Satu Juta Rumah) of two-parent families. Similarly, in Indonesia, introduced by President Jokowi in 2015 the residential segregation results presented in (box 7.2). Nonetheless, the growth of urban chapter 4 suggest that more segregated multi- areas has created congestion pressures on the district metro areas generate weaker human land and housing markets, exacerbated by capital externalities (that is, weaker spillovers weak or badly enforced urban planning. of knowledge from more skilled to less skilled As discussed in chapter 2, Indonesia’s cities workers). Providing affordable and well- face significant quantitative and qualitative located housing is therefore one of the most housing deficits, which constrain intraurban critical challenges in fast-growing cities. 6 connectivity and integration. Housing unaf- However, demand-side pressures, deficient fordability remains a major constraint for land-use management, and weak institutional improving housing access and quality, driven coordination often limit the availability of in part by high land and property prices in well-located quality housing and thus well-serviced areas and limited penetration 252  TIME TO ACT BOX 7.2  Government of Indonesia housing policy interventions The Indonesian government has focused recent Million Houses Program, SSB subsidizes the housing interventions around four primary difference between the prevailing interest rate initiatives—the Self-Help Housing Stimulus on a commercial loan and the subsidized flat ( Bantuan Stimulan Perumahan Swadaya, 5 percent rate for the duration of loan tenure. BSPS), credit-linked subsidy programs that Unlike FLPP, capital funding for SSB is the include the Housing Loan Liquidity Facility responsibility of the participating lenders. (Fasilitas Likuiditas Pembiayaan Perumahan, Because the Indonesian government funds FLPP) and the Interest Rate Buy-Down Subsidy only the current year’s interest rate subsidy (Subsidi Selisih Bunga, SSB), and public rental (so future contingent liabilities for SSB need housing (Rusunawa). Specific programs include not be accounted for until disbursed), SSB the following: increased the volume of subsidized housing units while maintaining the fiscal budget. In 1. Self-Help Housing Stimulus (BSPS) : Since 2016, SSB supported 110,000 units (in 2006, the national government has provided addition to the FLPP units). assistance to incremental self-built housing, 4. Public rental housing (Rusunawa) : The targeting rural and urban periphery areas. To Rusunawa program supports multifamily public be eligible, households must earn below the rental housing. The national government provincial minimum wage with a nationwide allocates a capital subsidy for construction, and income limit of Rp 2.5 million (US$187) a district governments are responsible for month. Depending on the province, they identifying and servicing program sites and receive a grant of Rp 10 million to 30 million managing units at completion. Tenants must (US$751–2,253) for home improvement or earn below the provincial minimum wage, with new construction. From 2010 to 2013, the a nationwide qualifying income limit of Rp 2.5 program reached 565,000 beneficiaries, with million (US$187) a month. Rents are set at 30 an average grant of Rp 20 million (US$1,502). percent of the provincial minimum wage. In 2. Housing Loan Liquidity Facility (FLPP) : 2014, the Ministry of Housing allocated Rp 1.3 Introduced in 2011, FLPP provides trillion (US$97 million) to the public rental concessional funding to lenders who provide program, and the Ministry of Public Works mortgages at fixed interest rates—currently added Rp 1.1 trillion (US$82 million). The two 5  percent a year for 20 years. Liquidity is ministries have since merged, so the program funded 90 percent by the government of sits under a single directorate. Rusunawa has Indonesia (at 0.5 percent for 20 years) and been characterized by low use of its budget, 10  percent by the participating banks. underachievement of targets, high vacancy rates, Between 2011 and 2015, FLPP served on and poor maintenance and implementation. average 88,000 households a year. 3. Interest Rate Buy-Down Subsidy (SSB): To Source: Based on data from Ministry of Public Works and Housing provided to the help meet the volume target set by the One World Bank on a request basis. of  the mortgage market. As a result, only (World Bank/Government of Indonesia 20 ­percent of households can afford housing 2015). in the formal market and 40 percent of house- Especially in multi- and single-district holds can afford the same housing only with a metro areas, where demand pressures are government subsidy. For the bottom 40 per- highest, access to well-located land is one of cent of the population, housing in the formal challenges for developing afford- the biggest ­ market is unattainable, thus spurring urban able housing. Within the Jakarta multidistrict slums and poor urban service indicators metro area, land prices in the core (that is, C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   253 DKI Jakarta) increased, on average, by Moreover, Indonesia’s limited mortgage 16 percent a year between 2010–11 and market constrains housing affordability and, 2013–14, while those in the surrounding kota through that, the growth of better connected districts of Bekasi, Depok, Tangerang, and and integrated cities. Mortgage lending is Tangerang Selatan increased at rates of slightly less than 3 percent of gross domestic between 20 and 37 percent a year (table 7.1). product, a small share compared with that of The increase in land prices has contributed to neighboring countries such as Thailand land banking practices that exacerbate the (20  percent), Mongolia (15 percent), and challenges related to land availability and India (6 percent) (World Bank 2016). have pushed affordable housing develop- Mortgage lending is constrained by three ments further from city centers to places key factors. First, most Indonesians cannot where land is cheaper. In the Jakarta metro afford a mortgage without a subsidy or are area, at least 19,707 hectares (roughly 60 per- not eligible for one because they are among cent of the metro area) are controlled by a the large share of the workforce that is infor- small number of developers (Elmanisa et al. mal.9 Second, current credit-linked subsidy 2017). This land could be developed into as programs exacerbate the constraint by failing many as 985,400 housing units,8 but many to leverage private markets. Instead, both the prime areas are still vacant because of land Housing Loan Liquidity Facility (Fasilitas speculation. Similarly, in Tangerang Selatan, Likuiditas Pembiayaan Perumahan or FLPP) private developers own 80  percent of the and the Subsidi Selisih Bunga (SSB) interest area; in Depok, developers own 40 percent rate buy-down for mortgages subsidy pro- (Ministry of Public Works and Housing gram crowd out the private sector, which can- 2017). Land banking by private developers not compete with the 5 percent fixed interest means that high land prices are being driven rate loans offered through government sup- to rise by artificial constraints on the avail- port.10 Moreover, private sector mortgages ability of land rather than by actual shortages are mostly concentrated in six top commer- in land. Property taxation has been used in cial banks that generally target the high- and other cities to regulate land markets and upper-middle-income segments (the top incentivize development in city centers. For 30 percent). BTN Bank is an exception: given example, in Seoul, Republic of Korea, land its corporate strategy to be the premier mort- left vacant for two years is considered idle gage market lender in Indonesia, it has built and is subject to a 5 percent property tax, a  strong sales network focused on the instead of the normal 2 percent; in Yucatan, government’s credit-linked subsidy program. ­ Mexico, vacant land is taxed at 2.5 times the Beyond commercial banks, other lender rate of land that has been developed ­s egments in  Indonesia—multifinance (Amirtahmasebi et al. 2016). companies, microfinancial institutions, and ­ TABLE 7.1  Annual land price increases within the Jakarta multidistrict metro area, 2010–14 (percent) Area 2010–11 2011–12 2012–13 2013–14 Average DKI Jakarta 18 21 15 11 16 Tangerang 29 22 64 28 36 Tangerang Selatan 16 19 25 22 20 Bekasi 29 23 49 19 30 Depok 32 52 25 40 37 Source: Bank of Indonesia data from Elmanisa et al. 2017. Note: DKI Jakarta refers to Daerah Khusus Ibukota Jakarta; the remaining areas are kota districts. 254  TIME TO ACT cooperatives—have a limited risk appetite for promoting low-density growth, urban sprawl, lower-income segments. Third, the mortgage greater land needs, and increased basic infra- market lacks product diversity to fulfill the structure costs. needs of different segments of the urban pop- Given the national government’s emphasis ulation. Most commercial mortgage loans on increasing the housing supply by allocating have a term of 10–15 years and start with a subsidies, city governments have often been left fixed teaser rate of 6 to 9 percent in the first in charge of important but underfunded pro- 1–5 years, followed by a floating interest rate grams for promoting c ­ onnectivity—such as indexed to the Bank of Indonesia’s policy self-help housing upgrading, public rental rate—currently about 11–14 percent.11 housing management, and slum upgrading. In recent years, national government efforts These governments often lack technical and to improve housing outcomes have primarily financial capacity and have poor organiza- focused on increasing supply through credit- tional structures to support implementation linked subsidies (FLPP and SSB) that support and expansion of such programs. ownership of developer-built homes. These interventions have failed to achieve the There are several key policy actions that expected results. Poor design and limited plan- can help to increase access to affordable ning and capacity to enforce location and con- and well-located housing struction quality guidelines have resulted in inadequate quality housing located far from To improve housing outcomes, especially for urban centers, which may be responsible for low-income groups, and, through this, foster the high 36 percent vacancy rate.12 Further, more connected and integrated urban growth, subsidies for new affordable housing are not a comprehensive housing agenda is essential. being channeled toward areas with the great- Both the capacity and planning challenges est needs. Chapter 2 shows that the highest proportion of households living in slums is in FIGURE 7.1  In 2017, few housing loan liquidity urban periphery and nonmetro urban areas; facility subsidies went to urban areas yet, of the FLPP units newly developed in 2017, only 12 percent and 26 percent were 60 built in urban periphery and nonmetro urban 52 areas, respectively, whereas 57 percent were 50 Share of subsidies (%) built in rural areas (figure 7.1). 40 Zooming into cities, the analysis suggests 31 that most subsidies are directed away from 30 26 city cores and into socially suboptimal areas 23 20 18 19 (box 7.3). In Medan, 88 percent of subsidized units for 2016 and 2017 were located 10 kilo- 12 10 meters or more from the city center. In 4 2 5 5 3 Surabaya and Bandung, the percentage was as 0 high as 99 and 98 percent, respectively. Based Metro Urban Nonmetro Nonmetro Rural N/A on the evidence in chapter 4, the location of core periphery urban rural periphery new affordable housing in faraway locations 2016 2017 might be perpetuating welfare differences and residential segregation within cities because Source: Calculations based on data from Ministry of Public Works and Housing and Google Maps API. human capital and skills tend to be concen- Note: Data include addresses of housing units purchased under the Fasilitas Likuiditas Pembiayaan trated in core areas, whereas the poor, with Perumahan ( FLPP) subsidy program for 47,412 recipients in 2016 and 21,806 recipients in 2017 (more than 97 percent of total recipients for each year). The World Bank team geolocated these addresses lower levels of human capital, are being using the Google Maps Place Autocomplete API and Google Maps Geolocate API services, cross- pushed to the peripheries. Moreover, 99 per- checked for accuracy using urban community and village-level shapefiles from Badan Pusat Statistik cent of developer-built subsidized housing is (Statistics Indonesia). Geolocation is expected to be accurate within urban community/village borders in more than 95 percent of cases, a further 2–3 percent to be accurate within 100 meters (“rooftop” in  single-family housing developments, accuracy), and a small remainder to be erroneous. N/A represents the households for which no location which limit connectivity and integration by information was available. Places are defined using the methodology described in box 1.5, chapter 1. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   255 BOX 7.3  Affordable housing land suitability perspective in Semarang The World Bank’s City Planning Labs recently Results show that most of the subsidized houses developed a suitability tool to identify opti- in the sample perform relatively low in the suit- mal affordable housing locations within a ability index (22 points), given that they are city.a The tool estimates a “Suitability Index,” built in the outskirts of the city and in other sub- which ranges from 0 to 100 points, and assigns optimal areas (map B7.3.1; colors closer to blue the highest score to locations that maximize reflect better scores).b It is important to note, the social benefits of a housing development. however, that this is not the case for all housing It determines these scores by evaluating the developments and that improvements in loca- current status of neighborhoods across the tion have been observed between housing built city, according to the availability of infrastruc- in 2017 and that of 2016. ture and the distance to services and amenities. Using bid-rent theory, which predicts that Used in Semarang, the tool analyzed a sample the price of real estate falls as the distance from of 99 FLPP (Fasilitas Likuiditas Pembiayaan the  central business district increases, one might Perumahan or Housing Loan Liquidity Facility) argue that affordable housing projects are per- housing developments built in 2016 and 2017. forming poorly on the suitability index because MAP B7.3.1  FLPP housing developments according to the suitability index in Semarang Year 2016 2017 Suitability index 60 40 20 Source: Singh et al. 2019. Note: FLPP = Fasilitas Likuiditas Pembiayaan Perumahan (Housing Loan Liquidity Facility Program). Box continued on next page 256  TIME TO ACT BOX 7.3 Continued of high land prices in the city center. To respond When comparing “less or equal to 40” and “more to this argument, the City Planning Labs team than 40,” however, the cost of land practically gathered more than 600 observations of com- doubles. It follows that there is room to improve mercial land values in the Semarang metro area the location of affordable housing to a certain and found that the farther from the city center degree without significantly affecting costs. the lower the cost of land. However, the biggest decline is observed between the 0–5 kilometer and Source: Singh et al. 2019. a. The suitability tool is an urban planning tool designed to perform agile 5–10 kilometer bands, after which land prices do and robust assessments to support evidence-based decision making. It is not vary dramatically, which means that, even used to identify socially optimal housing locations. What are considered if there is a concern about land prices, cities are “socially optimal housing locations” can vary depending on which of the priorities is selected as the main one: proximity to economic activity, likely to have a broad range of distance to work proximity to urban services, proximity to health facilities, proximity to with. Indeed, when analyzing the cost of achiev- public transportation, proximity to education facilities, reasonable costs ing a higher suitability index score, no significant of land, and risk-free areas. The chosen main priority receives twice the weight, while all other priorities are weighted once. difference in the cost of land is found when com- ­ b. This suitability index score reflects the average score using reasonable cost of paring locations in the index’s 10–40 points range. land as the main priority. and the needs of the diverse types of place Long-term efforts to improve the quality of in the portfolio of places must be addressed. land administration are necessary to supply Key  policy actions include (1) scaling up well-located land for affordable housing and ­ current efforts such as those in Semarang to infrastructure development. Subnational gov- improve technical capacity at the local level ernments should review existing regulations for coordinated and spatially appropriate to remove artificial constraints on well- land and housing planning, and for improved located land, simplify procedures for land implementation of permitting processes, con- acquisition, accelerate and reduce costs for struction quality, and resilience standards; building registration and permits, and imple- (2) promoting a holistic agenda based on the ment occupancy certificate processes. specific needs of the portfolio of places and A one-stop shop has been introduced for households; and (3) reforming the housing reducing transaction costs by expediting finance sector to target unserved groups and housing development procedures, including to crowd in the private sector. land titling, registration, transfers, and ­ permits; however, the benefits of this effort are yet to be seen. International examples Strengthening local capacity for coordinated have shown that these instruments can be and spatially appropriate land and housing successful—Australia, for example, was able ­ planning, and for improved implementation to cut regulatory procedures and speed up of permitting processes, construction quality, permit approvals by two months, all while and resilience standards ensuring construction quality (McKinsey Improve the availability of well-located land Global Institute 2016). Further, new technolo- through better land management . In gies have the potential to increase the fluidity Indonesia, a complex and bureaucratic system of land markets through reductions in trans- for land administration has resulted in limited action costs and better information or available land in well-located areas for infra- enhanced security of tenure (see also structure development. A well-functioning spotlight 4 on “The Potential of Smart land market, however, is key to fostering inte- Cities”). Ghana and Kenya are testing the use gration and harnessing the benefits of agglom- of disruptive technology such as blockchain eration because it is the mechanism for to speed the formalization of informal land allocating land to its most productive uses. and housing (box 7.4). C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   257 BOX 7.4  Land tenure: Streamlining property rights with blockchain technology In December 2016, Kenya decided to pilot land disputes and register undocumented areas. blockchain technology for keeping track of Bitland securely and transparently records trans- land transactions. Blockchain is an immutable actions in a decentralized network, i ­ncorporating distributed digital ledger across a decentralized ­ such details as global positioning system coordi- web of computers that is used to record and nates, satellite photos, and descriptions. Through process transactions. It can create a clear and Bitland, landowners and government officials can permanent record of land ownership and trans- sign land documents that are simultaneously reg- fer of ownership, thus minimizing property istered to the blockchain l ­edger. So far, 24 com- fraud. Issuing land title deeds through block- munities in Ghana have begun working with it. chain safeguards them from human edits and The technology evades corruption and record corruption. tampering through secured cryptography: the In Ghana, roughly 78 percent of land property data on any given block cannot be altered retro- is not in current, accurate government records, actively without altering all subsequent blocks, opening the door to corruption, tax evasion, and which requires a consensus of the network’s abuse of eminent domain. In this context, the majority users. Bitland platform uses blockchain to create a reli- able digital registry in a centralized ledger to settle Source: United Nations Economic Commission for Africa 2017. Develop comprehensive information sys- (Herbert, Belsky, and DuBroff 2012). The tems to address the information gap. Poor development of the government of Indonesia’s information systems have resulted in lack of planned Housing and Real Estate Information transparency and coordinated action in the System can be a good step in this direction housing sector. Effective tools for developing because the system aims to collect and analyze and disseminating information will improve comprehensive data on housing supply and coordination. An integrated housing data plat- demand.13 form that combines housing and land market Strengthen subnational government information can improve accountability and capacity to develop and enforce regulatory efficiency across all levels of government and systems that support connected and inte- support transparent sharing of market infor- grated growth in a resilient manner. Many mation about demand and supply of housing, district governments lack capacity to enforce location, prices, finance, and mortgages. and implement construction building and Mexico has developed the National Housing resilience standards and to incentivize devel- Registry, a robust platform that consolidates opers to build quality housing close to urban data on all aspects of Mexico’s housing sector. centers. District governments thus need to be It gathers information on home construction able to improve and enforce regulations, such and sales, including property value, progress as on construction quality and location, to and location of construction, and housing ensure that new housing developments sup- quality and characteristics. It monitors hous- port the vision of an integrated city. This ing supply and demand and sets construction capacity should also include robust monitor- standards and price benchmarks. Most impor- ing and evaluation processes to take actions tant, this information platform allows policy to redress quality issues. makers to define areas of greater suitability Support mixed-use, mixed-income devel- for development, where access to service and opment in well-located areas to promote opportunities is highest and exposure to risk integrated and well-serviced neighborhoods. lowest, and thereby assign higher subsidies to Most subsidized affordable housing projects better located projects of affordable housing now cater to a homogeneous income group, 258  TIME TO ACT creating segregated communities that do not master plan that integrates mixed-use design benefit from a diversity of income levels with place-specific considerations. It also inte- and that weaken human capital externalities. grates a public transportation system to con- To promote a more connected growth, nect the project with Bekasi’s train station ­ subnational governments can encourage the and a bus terminal, and it has an educational development of mixed-use, mixed-income facility, shopping center, and marketplace.14 developments that can help control urban sprawl while also helping to balance costs and Promoting a holistic housing agenda based increase desirability (box 7.5). Vida Bekasi— on the diverse needs of places—and that an innovative mixed-use, mixed-income responds to the needs of different income development in Bekasi, Indonesia—shows segments that these types of developments are feasible with strong government support. It is a new Expand rental housing to increase the supply sustainable urban development of 15,000 res- of affordable housing for low-income house- idents. The project has been designed with a holds. Indonesia currently has no functional BOX 7.5  Key principles for mixed-use, mixed-income developments Mixed-use, mixed-income developments are •  Strong local leadership and capacity. Cities characterized by a combination of diversified need strong local leadership, capacity, and land uses including housing (mix of market- coordination to carry the vision through. rate and low-income housing), retail activities, Subnational governments need to create an and private businesses, and proximity to pub- enabling institutional and regulatory lic transportation and other services. If well framework to ensure viability and financial designed and implemented, these develop- sustainability. They can use a combination of ments can help increase intracity connectivity regulatory and financial tools to incentivize and promote more compact, socially cohesive, the provision of affordable housing, such as and public transport–oriented urban devel- bonus floor area ratios, expedited permitting, opment. However, mixed-use developments and robust investment partnering agreements pose several fundamental challenges that make with the private sector. them difficult to materialize. These challenges •  High-quality design based on collaborative include higher development costs, large financ- decision making. The design should be based ing gaps, and implementation difficulties given on a participatory process that involves all the complexity of design, financing mecha- stakeholders to accommodate diverse needs. nisms, and management. Although approaches Developments must be integrated into vary, successful examples in cities like Chicago, neighborhoods, ensure access to quality open Singapore, Stockholm, and Tokyo provide key spaces and services, and evenly distribute the principles and lessons that can help guide future different tenures throughout the development. developments in Indonesia. These include the •  Phasing. A phased approach can be used for following: large projects when resources are not enough to undertake all the development at one time. •  A clear, long-term, and strategic vision of the It is also useful when coupled with good mon- city. A clear, long-term vision of how the city itoring and evaluation systems. Lessons from will grow, including careful analysis of future early phases can help modify the course of housing and transport demand and integrated subsequent ones. planning, is critical to identify and plan for successful mixed-use, mixed-income developments. Source: Chatman, Rayle, and Gabbe 2016; Litman 2017. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   259 rental sector to address the affordability and for a set period. In South Africa’s Social mobility challenges that many households Housing Program, for example, the subsidy face. The Rusunawa public rental program is divided in two parts: the Restructuring lacks strategic direction because of poor Capital Grant and the Institutional Subsidy. accountability and lack of adequate funding Both subsidies are granted to accredited for maintenance of ongoing projects as well Social Housing Institutions, which can be as new capital investments. Private rentals by for-profit firms or operate on a not-for- small-scale landlords (kos kosan), common in profit basis. They are responsible for devel- multi- and single-district metro areas, repre- opment and management of projects and sent an affordable solution for low-income must accommodate families earning below a households. Rental housing has remained a certain threshold. Municipalities are often neglected area of the Indonesian national responsible for providing the land, and all housing policy, which has had a strong focus operations are regulated by the Social on home ownership and upgrading. Housing Regulatory Authority. 15 Between Rental housing policies can be strength- 2007–08 and 2013–14, 59 projects in 10 ened to increase production levels and municipalities produced about 18,000 units improve the quality and sustainability of in a variety of housing types, including high- apartments while crowding in private par- rise, walk-up, and s ­ ingle-story buildings ticipation. As a first step, the government (The Government Technical Advisory Centre needs to conduct a comprehensive rental 2015). study to assess rental market demand, sup- Rental housing can also consider demand- ply, challenges, and opportunities. With side assistance. Such instruments often esti- regard to supply, in order to attract accred- mate the subsidy needed to cover the ited actors (investors, landlords, nongovern- affordability gap between monthly rental mental organizations, or employers) that payment capacity and market rental rates can help provide quality rental housing to through a voucher system. Indonesia could target groups, the government could con- analyze the lessons from international experi- sider designing a capital subsidy program ence as to the pitfalls and strengths of such (such as infrastructure, land, or tax credits, programs (box 7.6). where appropriate). These direct subsidies Support alternative housing typologies could include provisions to ensure that and cost-effective construction methodolo- developers commit to maintaining units at gies. Housing quality gaps remain high in affordable levels for the target population Indonesia because of high housing and BOX 7.6  Rental housing voucher programs In the United States, the Housing Choice In Chile, the Rental Housing Subsidy is aimed Voucher Program, administered by public at low- and moderate-income young families. housing authorities, provides rental assistance The program consists of a flat-rate, limited-time to very low-income households (to be eligible, subsidy, which offers a degree of administrative households must have income 50 percent below simplicity and payment flexibility to tenants fac- the median of their area). Beneficiary tenants ing income volatility. Ministry of Housing and pay 30 percent of their income toward rent, Urban Development regional officers conduct and the program pays the difference between applications, enrollments, and housing inspec- the tenant contribution and the actual rent tions. The administration of the subsidy relies up to “fair market rent.” Currently, the pro- heavily on private banks. gram reaches about 2.2 million households, or 5 ­million individuals. Source: Ross and Pelletiere 2014. 260  TIME TO ACT construction costs, with 22 percent of the Reforming the housing finance sector urban population living in slums and an to target unserved sectors and crowd estimated 70 percent of the housing stock in the private sector self-built.16 It is important to identify more Continue efforts to improve the efficiency of cost-effective and innovative supply-side the credit-link housing subsidy. The current solutions to meet the heterogeneous needs subsidy program (FLPP and SSB) has proven of households. The use of innovative mate- to be economically unsustainable for the gov- rials and techniques that are simple to fabri- ernment, and its design has crowded out the cate, transport, and assemble can lower the private sector and focused mostly on middle- costs and increase the quality of self-built income households. Accordingly, the new housing. Currently, local private developers, Mortgage-Linked Down-Payment Assistance such as the Ataca Cohousing Group, are (Bantuan Pembiayaan Perumahan Berbasis experimenting with alternative construction Tabungan, BP2BT) was recently developed to materials and technologies that are expected improve the targeting and sustainability of to reduce construction costs and facilitate housing subsidy programs, and continuing the mass production of quality affordable efforts in this direction will be key. The pro- h o u s i n g . 17 R e c e n t t e c h n o l o g i e s c a n gram will not only enable the heretofore strengthen the links between the various unserved informal income and self-built sec- stakeholders in housing development. For tors access to subsidies but it will also crowd example, a new construction application in private sector funding and participation at platform called iBUILD, launched in Kenya, market-based pricing. Nigeria, and South Africa, and now Jump-start the housing microfinance sec- expanded into India and Indonesia, helps tor to support the housing needs of the infor- households search and apply for financing, mal sector. The housing microfinance sector review quotes and select contractors, man- is underdeveloped despite the significant age budget and construction, and pay sup- needs of financing especially among low- pliers and contractors. 18 Urban areas income households, including those who especially could benefit because 60 percent work in the informal sector. Although there of residents have cell phones and 72 percent are many microfinancial institutions with use the Internet (BPS 2016). broad customer bases, they have not extended Given the diverse needs across the portfo- their market beyond productive loans. The lio of places, priorities may differ: development of the sector may thus require a •  For multidistrict metro areas, coordina- combination of financial and nonfinancial tion across districts will be key to ensur- services, including (1) a wholesale liquidity ing that housing is available in both core funding support for lenders with no or limited and peripheral areas and to integrate dis- deposit base funding; (2) a government-led connected pockets of poverty. credit default guarantee to mitigate unfamil- •  For single-district metro areas and iar risk; (3) financial institution capacity the  cores of multidistrict metro areas, building including tools and mechanisms to where supply-side constraints may be develop lender underwriting and risk man- more problematic than demand-side pres- agement capacity in lending to low- and sures, making serviced land available for informal-income borrowers; and (4) technical development will be important. assistance for assessing and monitoring proj- •• In nonmetro urban areas, where demand ects. The development of a microfinance sec- may be increasing, ensuring that new tor should consider building strong links with development remains connected with the existing housing programs such as the Self- city and that infill development for new Help Housing Stimulus program (Bantuan housing is attractive for private sector Stimulan Perumahan Swadaya, BSPS) and development will boost housing access upgrading slums to maximize outcomes and and attractiveness. impact for the target sectors. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   261 Even for those with access to housing, infrastructure development. Between 1995 limited transport infrastructure constrains and 2014, the number of cars increased nearly accessibility and mobility sixfold, and the number of motorcycles multi- A key element of good housing is its connec- plied tenfold (see chapter 6), creating substan- tion with the rest of the metro or urban area. tial traffic congestion and lengthy commute When housing is disconnected, people may times and distances, especially in metro areas. have a roof but remain isolated from oppor- A comprehensive transport policy that tunities and services. The case of Mexico is enables greater mobility and accessibility well-known: the growth of large, low-density, must consider all modes in tandem, ensure single-use housing developments over the past that public transport is provided as an alter- two decades fueled urban sprawl and left native for private vehicle use, and price pri- many disconnected from job opportunities. vate modes to reflect the externalities they Mexico began to radically transform its hous- generate (congestion and pollution). In recent ing sector in 2000 but paid limited attention years, the national government has started to to the overall functionality and accessibility of invest more in urban public transport through new developments on the outskirts of cities the development of bus- and rail-based mass that deformed the urban landscape. Low-cost transportation modes. Fourteen cities have housing increased by about 1 million units a invested in elevated bus stops and high-floor year between 2006 and 2011, but, as develop- bus fleets running in mixed traffic, with ers sought to produce more while keeping Jakarta’s TransJakarta the only bus rapid land costs low, they increasingly built on rural transit (BRT) with a dedicated lane.19 The tracts far from city centers. The resulting light-rail transit in Palembang, the first rail- patchwork of dispersed housing develop- based mass transport system in the country, ments substantially increased the marginal commenced operation in October 2018; and cost of providing electricity, water, transporta- mass rapid transit and light-rail transit proj- tion, and other services and exacerbated ects are under construction in the multidis- social exclusion by relegating lower-income trict metro area of Jakarta (that is, in households to poorly planned developments Jabodetabek). The effectiveness of these sys- (Kim and Zangerling 2016). tems has been constrained by planning and In Indonesia, many metro and urban areas coordination challenges that limit their capac- face similar patterns and challenges. The hous- ity to serve the needs of the growing number ing challenges just highlighted have been com- of urban commuters. For example, there is pounded by limited urban transport limited integration across modes and with infrastructure and services, with urban areas feeder services, and there is poor integration unable to achieve greater connectivity and inte- among land use planning and transport devel- gration, contributing to strong congestion forces opment (box 7.7). and accessibility and mobility problems. Not surprisingly, the modal share of public For many decades, the government of transport is declining across all major urban Indonesia has underinvested in sustainable areas. Nationally the proportion of daily urban transport (and especially public trans- commuters using public transport fell from port) and has had no urban transport policy 34.1 percent to 15.2 percent between 2008 to guide the sector and establish priorities. and 2015, with the sharpest declines among The transport sector has followed an unsus- metro core dwellers. Lower public transporta- tainable supply-side model of increasing tion shares of total trips mean that maintain- infrastructure for private vehicles, widening ing a high-quality service (including coverage, roads, expressways, and multilevel intersec- accessibility, and affordability) is difficult, and tions, with little investment in public and non- city officials often face increasing imbalances motorized transport alternatives and a between revenues and operating costs for the generalized disconnect from urban develop- systems they manage. The decline in service ment plans. Moreover, the rapid rise in pri- also affects poor people disproportionately vate vehicle ownership has outpaced because they have limited or no access to 262  TIME TO ACT BOX 7.7  Public transport deficiencies and TransJakarta BRT TransJakarta BRT (bus rapid transit) reveals the generally poor, making commuting difficult. institutional shortcomings that limit the effec- Second, the BRT has not spurred transit-oriented tiveness of public transportation in Indonesia. development in the form of commercial and/or In its current form, TransJakarta BRT was residential development around the stations. inspired by Bogotá’s TransMilenio. It started Third, although most BRT ridership comes from operating in January 2004 along an initial low-income areas, many stations serve higher- 13.6-kilometer north–south corridor and was income neighborhoods. the first BRT system in Southeast Asia. Over A thorough evaluation conducted using data time it expanded to cover 12 corridors totaling up to 2010 found very little impact of the BRT more than 200 kilometers and more than 200 system on transport. It serves only 4.3 percent stations. At present, TransJakarta serves mostly of commuters and has had virtually no impact the DKI Jakarta area, which is the core of the on car and taxi use. Although ridership has gone Jakarta multidistrict metro area, and several up in recent years, its modal share as of 2018 is planned extensions have been slowed by juris- still quite low. Gaduh, Gracner, and Rothenberg dictional issues between the DKI Jakarta gov- (2018) conclude that the only policies likely to ernment and neighboring districts within the directly reduce congestion in the Jakarta metro Jakarta metro area. area would be congestion pricing and increased Three key planning and coordination fail- fuel taxes. ures have thwarted the system. First, pedestrian infrastructure around TransJakarta stations is Source: Gaduh, Gracner, and Rothenberg 2018. private vehicles and must rely on public trans- to 6 p.m. (figure 7.2). Congestion levels vary port services to mobilize. Inefficient public among metro areas, with the highest in transport systems have contributed to the rise Jakarta, where congestion intensity reaches of personalized transport companies such as 59 percent in the evening peak hour, com- Go-Jek, Uber, and Grab, which fill a transport pared with the 27 percent average of all areas. gap in the market yet create regulatory chal- As expected, congestion within core areas is lenges and safety concerns (box 7.8; see also significantly higher, with an average conges- spotlight 4 on “The Potential of Smart tion intensity of 39 percent during the evening Cities”). peak hour. Given these conditions, mobility and acces- Weighted peak hour travel speeds average sibility are limited by congestion forces such 26 kilometers per hour across the 28 multi- as increased travel times, travel costs, traffic- and single-district metro areas, decreasing to related accidents, and greenhouse gas emis- 20 kilometers per hour in the core areas.21 The sions and local pollutants. In Indonesian metro areas with lowest peak hour speeds are metro areas, the intensity of congestion does Jambi and Palembang, where average speeds not follow the international norm of a high of 20 kilometers per hour are found in the eve- morning peak, reduced midday levels, and a ning peak hour between 5 p.m. and 6 p.m. high evening peak.20 Instead, they show a Bandar Lampung, Bandung, and Sukabumi slight morning peak and a gradual buildup of record an average peak hour speed of congestion throughout the day, culminating in kilometers per hour. Large metro areas such 22 ­ a discernible evening peak. Lowest observed as Bogor, Medan, DKI Jakarta, and Surabaya speeds are between 4 p.m. and 7 p.m., with have higher average peak hour speeds, between the average daily peak hour from 5 p.m. 24 and 27 kilometers per hour.22 C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   263 BOX 7.8  Mobility apps and urban transport Recently, several digital platforms have experi- by Grab of Uber’s Southeast Asian business high- enced sustained growth, among them notably lights the risks. In early July, Singapore’s antimo- Go-Jek, a motorcycle transport service platform nopoly agency, the Competition & Consumer created in 2011 in Jakarta, which boasts 12,000 Commission of Singapore, found that the sale drivers across several Indonesian cities. Together had reduced competition and that Grab had with competitors Grab and Uber, Go-Jek epito- already been increasing prices. mizes the growing synergies between transport Second, motorcycles impinge on land and on and telecommunications that are revolutionizing pedestrian ways alongside urban roads, worsen- urban mobility, often capitalizing on the short- ing the interconnection between walking and comings of traditional transportation. mass transit commuting, while generating safety This rapid development has raised several issues. The poor design of traffic regulations and issues. First, the disruptive technologies require failure to enforce existing ones are largely the ­ public authorities to take regulatory and compe- responsible. tition issues seriously if they are to maximize the Source: Ministry of Public Works and Housing 2017 and Competition & Consumer benefits to consumers. The March 2018 purchase Commission of Singapore 2018. FIGURE 7.2  Congestion in Indonesian metropolitan areas peaks in the evening 75 Congestion intensity (%) 50 27 25 25 25 23 21 22 21 22 21 19 20 20 16 17 14 13 10 9 5 5 4 3 2 2 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Midnight to noon Noon to midnight Source: Calculations based on Google Traffic API data. Note: Congestion intensity is the percentage by which observed travel time exceeds free-flow travel time. Each of the 33 lines presents congestion intensities across times of day for each of 28 metro areas and 5 constituent districts of the Jakarta metro area (DKI Jakarta, Bogor, Depok, Tangerang, and Bekasi). See chapter 1 for the 28 metro areas examined, of which 21 are multidistrict metro areas and 7 are single-district metro areas. The white numbers in the blue markers indicate the unweighted average congestion intensity across the 28 metro areas and 5 constituent districts of the Jakarta metro area at a given time of day. The lowest core area peak hour speeds are Pusat), Malang, and Surakarta all have peak found in Yogyakarta (16 kilometers per hour). hour speeds of 18 kilometers per hour. Bandung and Bukittinggi both have peak hour Although the population of a metro area speeds of 17 kilometers per hour. Banjarmasin, has no relationship with peak hour mobility, Denpasar, Depok, central Jakarta (Jakarta increasing population density does have the 264  TIME TO ACT expected inverse correlation with peak hour time), additional vehicle maintenance costs speed, and a positive correlation with conges- due  to additional starts and stops, carbon tion intensity (figure 7.3). dioxide and other greenhouse gas emissions, Jakarta’s metro core, DKI Jakarta, has the public health impacts from emissions, the cost highest congestion intensity. The economic of collisions and traffic-accident-related inju- cost of congestion in DKI Jakarta was found ries and deaths, and losses due to excess freight to be US$2.6 billion a year (figure 7.4). This transport time. However, the costs due to such cost increases to just under US$3 billion a externalities of congestion can be high. Road year for the full Jabodetabek—or Jakarta traffic accidents are among the leading causes multidistrict metro—area. The cost of con- of death by injury: in Indonesia in 2014, there gestion drops rapidly for smaller metro were about 100,000 reported accidents, result- areas, with Bandung having an annual con- ing in 28,000 fatalities.23 Cars, motorcycles, gestion cost of US$222 million and Surabaya and the trucks that carry more than 90 percent a cost of US$236 million. Medan and of freight have made Indonesian cities highly Makassar are the only other metropolitan polluted: the transport sector contributes areas with congestion costs greater than 70 percent of city pollution (Leung 2016). US$100 million a year. The total cost of con- Motorized transport contributes 23 percent of gestion across all 28 of Indonesia’s multi- greenhouse gas emissions. Last, much open and single-district metro areas amounts to space is occupied by private vehicles—parked, just over US$4 billion a year. driven, or jammed in traffic—reducing the The congestion costs just presented can be already limited space available for walking or considered a lower bound for actual conges- biking, and sidewalks have been neglected tion costs, because they do not include the cost (Jakarta Post 2015). Pedestrian facilities, where of travel time uncertainty (additional time that they exist, are much used by food stalls, ven- must be allocated to a trip to ensure that the dors, parked vehicles, and motorcycles cutting vehicle reaches its destination at a specific in and out of traffic. FIGURE 7.3  Population density is correlated with peak hour congestion intensity 70 y = 0.25x + 23.06 Pekanbaru R2 = 0.5487 60 Denpasar Bandung Medan Yogyakarta Bandar Lampung DKI Jakarta Makassar Jabodetabek 50 Surabaya Congestion intensity (%) Palembang Depok Padang Bekasi 40 Malang Banjarmasin Jambi Tangerang 30 Pontianak Bogor Samarinda Banda Aceh Semarang 20 Balikpapan Bukittinggi Blitar Surakarta 10 Sukabumi Magelang Mojokerto Probolinggo Salatiga Pasuruan 0 –20 0 20 40 60 80 100 120 140 160 Population density (persons per hectare) Source: Calculations based on Google Traffic API data. Note: Congestion intensity is the percentage by which observed travel time exceeds free-flow travel time. Peak hour refers to the evening peak hour between 5 p.m. and 6 p.m. See chapter 1 for the 28 metropolitan areas examined, of which 21 are multidistrict metro areas and 7 are single-district metro areas. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   265 FIGURE 7.4  The annual cost of congestion in supply of transport infrastructure and ser- Indonesian metropolitan areas exceeds US$100 vices, with demand management instruments million in six cities, and exceeds US$2.6 billion in to reduce congestion and ensure that the use DKI Jakarta of private vehicles is priced correctly, includ- ing all generated externalities. Key actions are 10,229,032 DKI Jakarta, 2,604 to (1) strengthen the national government’s 8,305,120 Bandung, 222 7,113,428 Tangerang, 98 role in guiding urban transport policy and 6,615,736 Bogor, 69 national coverage of investment gaps; 6,267,951 Surabaya, 236 (2) enhance local capacity to plan, operate, 6,108,920 Bekasi, 134 5,321,383 Medan, 166 and maintain urban transport systems; and 4,375,446 Surakarta, 22 (3) promote transit-oriented development 3,796,348 Semarang, 30 that encourages densification in transport cor- 3,613,549 Malang, 97 2,827,899 Makassar, 108 ridors and reduces travel demand and private 2,762,378 Sukabumi, 8 vehicle use intensity. 2,575,277 Yogyakarta, 25 2,455,999 Denpasar, 44 2,161,182 Depok, 39 Stronger national government guidance of Population 1,786,533 Pasuruan, 2 1,783,141 Pontianak, 8 urban transport policy and national coverage 1,594,110 Palembang, 54 of investment gaps 1,376,717 Probolinggo, 2 1,375,282 Magelang, 3 Develop clear guidelines for transport infra- 1,287,532 Blitar, 1 1,213,937 Mojokerto, 2 structure development and management . 1,196,676 Salatiga, 1 Discussions on an urban transport policy in 1,057,894 Pekanbaru, 18 the Ministry of Transportation have been 990,092 Bandar Lampung, 8 983,269 Banjarmasin, 4 ongoing for several years. The policy has not 909,075 Padang, 10 yet materialized, neglecting the sector at both 824,360 Samarinda, 21 652,566 Banda Aceh, 1 the national and local levels of government 623,398 Balikpapan, 12 and exacerbating coordination challenges due 603,930 Bukittinggi, 2 to the absence of common objectives. As a 581,598 Jambi, 6 first step, creating a national urban transport 0 1 10 100 1,000 10,000 policy is key to developing transport-related US$ million (log scale) guidelines for city and district governments. Source: Calculations based on Google Traffic API data. Such an approach will help ensure that trans- Note: The figure shows the economic costs of congestion in the 28 multi- and single-district metro port system designs are based on robust and areas, where the Jakarta multidistrict metro area is broken down into DKI Jakarta, Bekasi, Bogor, reliable demand forecasts for the medium to Depok, and Tangerang. The economic costs of congestion were determined by extracting the trips to and from each subdistrict in each metro area, in each hour of the day where the congestion intensity long term, and that investments are proposed exceeded 25 percent. The excess travel time and fuel consumption that result from congestion in under a well-defined funding and financing each metro area were calculated. Excess travel time was calculated from the difference between the framework and targeted toward the most observed travel time for each trip and 1.25 times the free-flow speed so that only the travel time above the congestion threshold was counted. Excess fuel consumption was calculated by applying speed/ effective modes of transport for different fuel consumption equations for cars, motorcycles, and buses to the travel speed for each hourly trip areas with different needs. A comprehensive (as extracted from the Google Traffic API) compared with 25 percent less than the free-flow speed. transport policy should outline actions that This analysis used an exchange rate of US$1 = Rp 14,000. Estimates represent a lower bound for costs because they include only the direct costs of excess travel time and fuel consumption. improve supply and manage demand, while discouraging private vehicle use intensity. Policies to increase supply should ensure that public transport systems promote high- Improving and expanding urban transport quality service and urban integration. Public and alleviating traffic congestion requires a transport networks should be designed, devel- multipronged policy approach oped, and upgraded to meet the needs of Relieving traffic congestion will require a commuters and become a more attractive coordinated and spatially driven approach to commuting mode over private vehicles, planning and developing transport infrastruc- with considerations including travel times ture. Such an approach must ensure strong and costs. Key actions along three lines are coordination efforts when increasing the important. First is institutional strengthening 266  TIME TO ACT through transit regulatory and management can help increase efficiency and quality of authorities capable of coordinating public public transport while opening new channels transport services across administrative for infrastructure financing. boundaries. Second is improving operations, National policy, strategy, and guidelines optimized routing and scheduling, together for urban mobility should also propose with prioritizing allocations of space to public actions to manage demand and reduce transport. Third is improving technology, explicit and implicit subsidies to private car such as smart ticketing or fare collection sys- use. Public transport systems will succeed in tems, fleet management systems, and user curtailing traffic congestion only if private information systems to greatly improve the vehicle use is simultaneously contained quality of service. through effective and well-enforced demand- Beyond the availability of physical infra- side interventions. In Jakarta, there has been structure, a focus on soft constraints to the some success in curbing private vehicle use, efficiency of public transport operation will thereby reducing congestion, with demand- be essential, increasing the use of fare collec- side instruments such as high-occupancy vehi- tion systems and enhancing integration cle restrictions (box 7.9). These instruments between different modes. Strong governance require strong policy design and enforcement of the sector to foster cooperation and coordi- to ensure long-lasting and significant effects. nation between the public and private sectors In addition, the reduction of explicit and BOX 7.9  High-occupancy vehicle policy in Jakarta The “3-in-1” high-occupancy vehicle restriction from 28 kilometers per hour to 20 in the morn- was implemented in March 1992. It required ing and from 21 kilometers per hour to 12 in the cars riding in both directions on the major corri- evening. In addition, there were negative knock- dors in Jakarta (Jalan Sudirman–Jalan Thamrin on effects at other times of the day and on roads and Jalan Gatot Subroto) at specific times of outside the regulated areas. The main culprit day to have at least three passengers, including was the increase in the number of cars on the the driver. The penalty for not complying was a road after elimination of the 3-in-1 policy. maximum fine of Rp 500,000 (about US$37.50) Jakarta’s experience with its 3-in-1 policy or two months in prison. reveals several facts. First, drivers were willing Many affluent car users commuting into to pay to access the city center by car—enough Jakarta did not meet the occupancy restriction. to sustain the informal jockey market. Second, An informal market for professional passen- the authorities’ failure to anticipate driver will- gers, nicknamed “jockeys,” emerged. Jockeys ingness to pay, through the payment of jockeys, would ride along from 3-in-1 access points contributed to the limited impact of the policy for about US$1.20. The practice contributed on congestion. Third, faced with this evidence, to the termination of the 3-in-1 policy, tem- the authorities responded by scrapping the pol- porarily in March 2016 then permanently in icy altogether rather than trying to improve it. May 2016. The 3-in-1 experience shows that, in the Jakarta The termination set back the attempt to curb multidistrict metro area, some type of congestion congestion. Despite the policy’s problems, traffic pricing scheme may successfully help address worsened significantly when it was eliminated severe traffic congestion. This is especially so (Hanna, Kreindler, and Olken 2017). During if the proceeds from such a scheme are used to the morning peak period, delays, computed fund urban mass transit or mobility for lower- using the time to travel a kilometer, increased income people. by 39–45 percent and during the evening peak period by 69–85 percent. The average speed fell Source: Hanna, Kreindler, and Olken 2017. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   267 implicit subsidies to private car use can be use the bike-sharing program every day, with considered. In Vienna, strategies such as reor- more than two-thirds of this number having ganizing parking space in the city, eliminating previously made motorized trips. free parking, removing parking from historic A systematic national program to finance places, and instituting street and parking per- urban infrastructure will be needed to missions for residents were key actions for develop public transport networks . The reducing car use (Buehler, Pucher, and national government, under a national urban Altshuler 2017). Congestion charges, which transport program aimed at supporting impose a tariff on car users, can also be con- efforts to improve public transport, can offer sidered. In London, congestion charges cover financial transfers contingent on cities’ meet- a 21-square-kilometer core area, reinforced ing policy, technical, institutional, and by a camera system. The charge is applied financing requirements. An example is hav- Monday to Friday from 7 a.m. to 6 p.m., ing an integrated mobility plan that addresses reducing the incentive to use private vehicles spatial development and proposes demand during peak hours.24 management, transport infrastructure, and Complement demand- and supply-side systems to support the desired city structure. policies by increasing investments in non- Developing such a plan should include a motorized transport and building better comprehensive program of outreach to the links between these forms of transport and public, community groups, and nongovern- the public transport systems. Investment in mental organizations. nonmotorized transportation has not been a Create incentives to support coordination government priority in Indonesia; however, across jurisdictions. Coordination at the met- developing pedestrian- and cycle-friendly cit- ropolitan level is difficult given the diverse ies can contribute to reducing congestion interests and objectives of stakeholders. The and increasing mobility. Walking paths need national government can play a key role in to be as comfortable and as safe as possible incentivizing coordinated planning through so that they attract citizens to switch from the development of legal instruments that private vehicles to walking and public trans- allow for the creation of an authority above port. To ensure success, pedestrian areas the district level to manage services that cross need to be integrated with other transporta- different jurisdictions. tion infrastructure and facilities. The Guangzhou BRT in China has become an Enhance local capacity to plan, operate, and international best practice example of a mul- maintain urban transport systems timodal transport network. In 2010, a 22.5-kilometer BRT corridor was opened Identifying national, subnational, and non- along one of Guangzhou’s busiest roads, government sources of funding to cover Zhongshan Avenue. A greenway was also operating expenses and repayment of capital developed next to the BRT corridor that expenditure debt, under the principle of includes pedestrian walkways and a bike- “who benefits pays,” should be explored as a sharing system with 5,000 bikes across 109 way to ensure high-quality transport services stations. The BRT lines and bike-sharing sta- (Ardila-Gomez and Ortegon-Sanchez 2016). tions are integrated through a ticketing Resources can be collected through, say, ded- ­ system that facilitates transfers across modes. icated taxes and fees on private transport, This BRT system is one of the most success- such as parking charges, local monitoring ful: it transports more than 850,000 people taxes, and fuel levies with proceeds reserved daily. According to an Institute for for public transport investment; employer/ Transportation and Development Policy employee taxes for public transport; reve- study (Hughes and Shu 2011), it has reduced nues from the sale or lease of public assets or traffic congestion by 29 percent and from real estate development; or even land increased the speed of buses and other vehi- value capture instruments, which create cles by 20 percent. More than 20,000 people s table funding not dependent on annual ­ 268  TIME TO ACT BOX 7.10  Tanjung Barat: Leveraging partnerships for transit-oriented development in Jakarta National affordable housing developer the Housing Loan Liquidity Facility ( Fasilitas Perumnas is partnering with the railway Likuiditas Pembiayaan Perumahan, FLPP). state-owned enterprise, PT KAI Commuter According to Perumnas’ representatives, profits Jabodetabek, to create a mixed-income, mixed- from the market-rate apartments will be used to use transit-oriented housing project in the build affordable houses in other regions of the Jakarta multidistrict metro area. The Tanjung country. Barat project will consist of two high-rise tow- Because national laws prohibit the railway state- ers built on previously unused land owned by owned enterprise from selling land, the project will PT KAI Commuter Jabodetabek next to the be registered as a 70-year leasehold for Perumnas. railway station. Apartments will be built above After the Tanjung Barat project, Perumnas intends a dedicated commercial space on the first floor. to develop six more unused, transit-oriented loca- Perumnas will build 650 apartment units, tions around greater Jakarta. 10 percent of which will be affordable hous- ing and eligible for a national subsidy under Source: Tempo 2017. budget cycles. However, few of these instru- coordination and cooperation, modifying cur- ments are easy to implement, even in the rent planning processes, and enlisting com- developed world, so strong local capacity is munity participation. As a first step, district needed; and time will be required before governments can work toward integrating the Indonesia can count on them. strategies outlined in the RTRWs with the transport and housing sectoral plans. Enable development-based land value Promote transit-oriented development capture as a financing and planning mecha- that encourages densification in transport nism for transit-oriented development invest- corridors and reduces the travel demand and ments. City and district governments should private vehicle use intensity seek funding for infrastructure that follows Systematically coordinate urban transport land use planning principles, such as transit- plans with land use plans and related infra- oriented development, and integrates trans- structure plans to promote transit-oriented portation land use strategy with spatial and development.25 Indonesia’s current planning housing plans (box 7.10). In metro areas, cit- system does not establish a clear connection ies can increase land values around transport between transport policy and land use. Land infrastructure through regulatory actions use is controlled by federal regulations that such as higher floor-space ratios, looser height limit the possibilities for local regions to inte- restrictions, or greater density in specific tar- grate, whereas transport plans are usually get zones. Hong Kong SAR, China, and Tokyo local, with policy development done by have used this approach to finance the con- Bappeda and the local Transportation struction, operation, and maintenance costs Department controlling their implementation of their urban rail transit systems. (Singh, Raghupathy, and Aurora 2017). For cities to boost mobility while managing con- gestion and pollution and ensuring good Conclusion quality transport for the all, they must sys- Urbanization creates opportunities to foster tematically coordinate urban transport plans greater prosperity, inclusiveness, and livabil- with land use plans and related infrastructure ity. If badly managed, however, it can create plans. This requires long-term efforts includ- gaps in the provision of adequate infrastruc- ing establishing institutional arrangements for ture including housing, transport, and basic C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   269 services, which can lead to fragmentation and be stressed enough that success depends on widening inequalities within cities. Thus, the effectiveness of the planning processes, strong institutional structures, especially good which must be dynamic, flexible, and urban and spatial planning systems, become responsive to the long-term needs of the cities. ­ increasingly important to build connected A change to the current mindset is required, and integrated cities. Although Indonesia has complemented by changes in the regulatory strengthened its planning efforts in recent frameworks to embed flexibility in the plan- years, this chapter highlights three key areas ning process. Good plans also depend on their of action in urban and spatial planning that effective enforcement, so statutory enforce- are essential for integrated urban growth. ment measures will be necessary to guide and First is improving the transparency and effi- regulate development. Finally, action across ciency of land management systems. Second is all sectors will need to be coordinated, espe- scaling up current efforts for better informa- cially in the housing and transport sectors. tion and data collection at the local level. And third is ensuring the operationalization of plans by building better links between socio- Notes economic development outcomes and spatial 1. A 2015 market assessment on a subset of 15 and sectoral plans. large cities estimated an overall subnational To bring people closer to jobs, opportuni- infrastructure investment financing gap of ties, and services, and firms closer to their US$11.1 billion against a borrowing capac- input and output markets and thus to foster ity of US$1.7 billion (Joshi et al. 2015). real integration, actions in housing and 2. As part of the National Affordable Housing Program, the World Bank is planning to work ­ transport will be key. Three key actions in with the Ministry of Home Affairs, which the  housing sector can help Indonesian oversees local governments, to optimize the cities avoid exacerbating fragmentation: implementation system/process. (1) improve technical capacity at the local 3. The spatial plan is currently under revision level for coordinated and spatially appropri- with inputs from the World Bank’s City ate land and housing planning, and for Planning Lab team to address some of these improved implementation of permitting pro- issues. cesses, construction quality, and resilience 4. With the support of the World Bank’s City standards; (2) promote a holistic agenda Planning Labs initiative. based on the specific needs of the portfolio of 5. Among them, the land and building tax for places; and (3) reform the housing finance urban and rural areas and land and building transfer tax. sector to target unserved groups and to crowd 6. Housing affordability means that a house- in the private sector. Without complementary hold’s demand or purchasing power is suf- actions in the transport sector, however, inte- ficient to access a property, considering the gration within cities will not be possible. This available supply and prices of different hous- chapter has pointed to three key areas for ing products in the market (typologies, sizes, action in this sector: (1) strengthen the locations, and so on). Purchasing power or national government’s role in guiding urban capacity to pay is generally defined as a debt- transport policy and national coverage of to-income ratio of 30 percent or less. A rough investment gaps; (2) enhance local capacity to estimate would compare gross household plan, operate, and maintain urban transport income to the rental payment or installment systems; and (3) promote transit-oriented payment for a mortgage loan or for a home improvement/extension loan. However, a full development that encourages densification in assessment of housing affordability would transport corridors and reduces travel also consider other factors, such as the risk demand and private vehicle use intensity. profiles of individuals. For example, infor- Indonesia has the potential to achieve more mal workers may have higher risk profiles connected and integrated urban areas that given the volatility of their incomes, which can foster greater prosperity, inclusiveness, may in turn lead to a threshold for afford- and livability. Despite that potential, it cannot ability much lower than 30 percent. Lending 270  TIME TO ACT institutions may in fact lower the 30 percent and frequent operations.” For more infor- debt-to-income ratio for riskier consumer mation, see the Institute for Transportation profiles. The capacity to pay should also and Development Policy website at https:// consider characteristics of mortgage mar- www.itdp.org/library/standards-and-guides​ kets, such as interest rates, regulations on /­the-bus-rapid-transit-standard/what-is-brt/. loan-to-value ratios, maturity of loans, and 20. The analysis was carried out for this report. lenders’ risk appetite. ­ Matrixes of travel times, distances, and 7. Sustainable urban transport systems refer to speeds for trips to and from each subdistrict the availability of safe, equitable, efficient, in each metro area were obtained from “Big and climate-responsive transport systems. Data” analysis of Google Traffic API data for 8. Assuming that one hectare could accommo- each hour of a typical weekday. Congestion date 50 houses. indexes were derived from the assumption 9. The share of the informal sector in total that the lowest travel time between each employment in 2017 ranged from a low of origin and destination (typically between 25 percent in the core of the Jakarta multi- 2:00  a.m. and 3:00 a.m.) represented free- district metro area (DKI Jakarta) to a high flow conditions for that trip. of 69  percent in nonmetro rural areas (see 21. Weighting based on the number of daily ­chapter 2). trips to and from each subdistrict in the 10. Although subsidized loans offer a 5 percent metro area, as well as the volume of traffic fixed interest rate, private commercial lend- in each hour of the day. Daily trips to and ers offer between 5.55 and 9 percent interest from each subdistrict were derived from trip rates for 1–2 year fixed loans, between 6.55 production-based generation rates and the and 10 percent interest rates for 3–5 year subdistrict population; trip attractions were fixed loans, and between 11.5 and 13.25 per- derived from the intensity of nighttime lights cent interest rates for floating loans (interest (as a surrogate for economic activity). rate information refers to May 2018). 22. For each multidistrict metro area, the core 11. All interest rates refer to May 2018. district was also analyzed, with the excep- 12. Ministry of Public Works and Housing tion of Jabodetabek (that is, the Jakarta mul- Evaluation Unit sample survey result of tidistrict metro area), which was analyzed 2,016 FLPP and SSB units in 2017. at three levels: the whole of Jabodetabek; 13. With support from the World Bank, the gov- each component, namely DKI Jakarta, ernment is working on the establishment of Bogor, Depok, Tangerang, and Bekasi; and the Housing and Real Estate Information the core area of each component (exclud- System, which will serve as a depository of ing Depok because of its small size), namely housing- and real estate–related data, with Jakarta Pusat, Kota Bogor, Kota Tangerang, indicators collected from areas of housing and Kota Bekasi. supply and housing finance, as well as track- 23. Estimated using data compiled by the ing of home price movements and the imple- National Traffic Police Corps. mentation of public housing programs. 24. Full information for London congestion 14. More information about Vida Bekasi is avail- charging can be found at http://tfl.gov.uk​ able at http://vidabekasi.com/. /­modes/driving/congestion-charge. 15. For more information, see https://www.gov. 25. Transit-oriented development has been a za/about-sa/housing#housing. strategy of locating commercial/mixed-use/ 16. Statistics of Housing and Settlement, 2013, shopping centers or high-density residen- Badan Pusat Statistik (Statistics Indonesia). tial buildings next to a transportation hub. 17. Team interviews with representatives from Developers pay part of the cost of the hub to Ataca Cohousing, including A. Budiman (Pres- have the rights to construct close by. ident Director), Widhi Hartono (Chairman), and Dr. Syadril Majidi, MM (Vice Chairman). Interviews held on July 19, 2017. References 18. More information about iBUILD is available Amirtahmasebi, R., M. Orloff, S. Wahba, and at https://www.ibuild.global/. A. Altman. 2016. Regenerating Urban Land: 19. According to BRT definitions, a BRT system A Practitioner’s Guide to Leveraging Private includes at least the “following components: Investment.  Urban Development. Washington, dedicated bus lanes; off-board fare collec- DC: World Bank. https://openknowledge​ tion in stations; high capacity buses and fast .worldbank.org/handle/10986/24377. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   271 Ardila-Gomez, A., and A. Ortegon-Sanchez. Harvard University Joint Center for Housing 2 0 1 6 .   S u s t a i n a b l e U r b a n Tra n s p o r t Studies, Cambridge, MA. Financing from the Sidewalk to the Subway: Hudalah, D., and J. Woltjer. 2007. “Spatial Capital, Operations, and Maintenance Planning System in Transitional Indonesia.” Financing. Washington, DC: World Bank. International Planning Studies 12 (3): 291–303. h t t p s : / / o p e n k n o w l e d g e . w o r l d b a n k . o r g​ Hughes, C., and X. Shu. 2011. “Guangzhou, /­handle/10986/23521. China Bus Rapid Transit: Emissions Impact Bank of Indonesia. 2017. Residential Property Analysis.” Institute for Transportation and Price Survey for Primary House. Jakarta, Development Policy, New York. https://www​ Indonesia: Bank of Indonesia. .itdp.org/wp-content/uploads/2014/07/GZ​ BPS (Badan Pusat Statistik, Statistics Indonesia). _BRT_Impacts_20110810_ITDP.pdf. 2016. Cellphone Penetration . Jakarta, Indonesia Investments. 2016. “13th Economic Indonesia: BPS. Policy Package Indonesia: Low-Cost Housing.” Buehler, R., J. Pucher, and A. Altshuler. 2017. “Vienna’s August 26. https://www.indonesia-investments​ Path to Sustainable Transport.” International .com/news/todays-headlines/13th-economic​ Journal of Sustainable Transportation 11 (4): -policy-package-indonesia-low-cost-housing​ 257–71. /­item7127. Chatman, D., L. Rayle, and C. J. Gabbe. 2016. Jakarta Post. 2015. “In Jakarta, Sidewalks “Analyzing the Economic Benefits and Costs Have Been Sidetracked.” The Jakarta Post , of Smart Growth.” Paper prepared for the February 22. http://www.thejakartapost.com​ California Air Resources Board. /­news/2015/02/22/in-jakarta-sidewalks-have​ Chetty, R., N. Hendren, and L. Katz. 2016. “The -been-sidetracked.html. Effects of Exposure to Better Neighborhoods Joshi, R., R. Gayathri, A. Permono, B. Ladda, on Children: New Evidence from the Moving to A. Parasher, R. Hutagaol, and R. Prastiwan. Opportunity Experiment. “American Economic 2015. “Consultant Services for Market Review 106 (4): 855–902. Assessment and Operating Framework for the Chetty, R., N. Hendren, P. Kline, and E. Saez. Indonesia Regional Infrastructure Development 2014. “Where is the Land of Opportunity? Fund: Final Report.” Consultant Report for the The Geography of Intergenerational Mobility World Bank, CRISIL, Mumbai. in the United States.” The Quarterly Journal of Kim, Y., and B. Zangerling. 2016. Mexico Economics 129 (4): 1553–623. Urbanization Review: Managing Spatial Competition & Consumer Commission of Growth for Productive and Livable Cities Singapore. 2018. Uber/Grab merger: CCCS in Mexico . Directions in Development: Issues Interim Measures Directions. Press Countries and Regions . Washington, DC: Release: Singapore. https://www.cccs.gov.sg​ World Bank. /media​ - and-consultation/newsroom/media​ Leung, K. 2016. “Indonesia’s Summary Transport -releases​/­uber-grab-imd-13-april-18. Assessment.” Asian Development Bank papers Dias, A., and H. Chandra. 2017. Urban Planning on Indonesia, No. 15, Asian Development in Indonesia. Washington, DC: World Bank. Bank, Mandaluyong, Philippines. Elmanisa, A. M., A. A. Kartiva, A. Fernando, Litman, T. 2017. “Selling Smart Growth: R. Arianto, H. Winarso, and D. Zulkaidi. Communicating the Direct Benefits of 2017. “Land Price Mapping of Jabodetabek, More Accessible, Multi-modal Locations to Indonesia.” Geoplanning: Journal of Geomatics Households, Businesses and Governments.” and Planning 4 (1): 53–62. Victoria, BC, Canada: Victoria Transport Policy Gaduh, A., T. Gracner, and A. D. Rothenberg. 2018. Institute. “Improving Mobility in Developing Country McKinsey Global Institute. 2016. “Housing Cities: Evaluating Bus Rapid Transit and Other Affordability: A Supply-Side Tool Kit for Policies in Jakarta.” Retrieved from https://sites​ Cities.” Briefing Note, McKinsey, New York. .google.com/site/alexrothenberg/research. Ministry of Public Works and Housing. 2016. Hanna, R., G. Kreindler, and B. Olken. 2017. “Konsep Backlog.” Jakarta, Indonesia. http:// “Citywide Effects of High-Occupancy Vehicle ppdpp.id/konsep-backlog/. Restrictions: Evidence from ‘Three-in-One’ in ———. 2017. The State of Indonesian Cities 2017. Jakarta.” Science 357 (6346): 89–93. Jakarta, Indonesia. Herbert, C., E. Belsky, and N. DuBroff. 2012. “The Musterd, S., S. Marcińczak, M. van Ham, and State of Mexico’s Housing—Recent Progress T. Tammaru. 2015. Socio-Economic Segregation and Continued Challenges.” Working Paper, in European Capital Cities: Increasing 272  TIME TO ACT Separation between Poor and Rich. Bonn, suitability p ­ erspective.” 2019 World Bank Germany: Institute for the Study of Labor. Conference on Land and Poverty, March 2019, OECD (Organisation for Economic Co-operation Washington, D.C. Unpublished conference and Development). 2016. Economic Survey paper. of Indonesia: October 2016. Paris: OECD Statistics of Housing and Settlement. 2013. Badan Publishing. Pusat Statistik (Statistics Indonesia). Ross, L., and D. Pelletiere. 2014. “Chile’s New Tempo. 2017. “Perumnas Dan KAI Bangun TOD Rental Housing Subsidy and Its Relevance Pertama Di Tanjung Barat.” Tempo, August 15. to U.S. Housing Choice Voucher Program h t t p s : / / m e t r o . t e m p o . c o / r e a d / 9 0 0 2 0 1​ Reform.” Cityscape: A Journal of Policy /perumnas ​ - dan-kai-bangun-tod-pertama-di​ Development and Research 16 (2): 179–92. -tanjung-barat. R u k m a n a , D. 2 0 1 5 . “ T h e C h a n g e a n d The Government Technical Advisory Centre. Transformation of Indonesian Spatial Planning 2015. “Social Housing.” Government of South after Suharto’s New Order Regime: The Case Africa, Cape Town. https://www.gtac.gov.za​ of Jakarta Metropolitan Area.” International /­perdetail/3.1%20Summary.pdf. Planning Studies 20 (4): 350–70. United Nations Economic Commision for Africa. Singh, G., S. Raghupathy, and D. Aurora. 2017. 2017. “Blockchain Technology in Africa. ” Toward an Integrated Transportation and Land Addis Ababa, Ethiopia. Use Plan for Semarang. Jakarta, Indonesia: World Bank. 2016. Indonesia’s Urban Story. World Bank. Jakarta, Indonesia. Singh, G., T. Hartanto, J. Diaz-Azcunaga, R. World Bank/Government of Indonesia. 2015. Ochoa-Sosa, and E. Perez-Denicia. 2019. Indonesia: A Roadmap for Housing Policy “Affordable housing in Indonesia: A land Reform. Jakarta: Kementerian PPN/Bappenas. C o n n ected a n d I n tegrated C ities : A F ocus o n H ousi n g a n d T ra n sport   273 Targeting Places and People Left Behind 8 KEY MESSAGES Policy makers in Indonesia have tried a wide range establishing special economic zones (SEZs) by of place-based policies, targeting incentives at first focusing on a limited number of places lagging places to reduce spatial inequalities. Place- with real potential to attract investments and based policies to attract firms into lagging districts using success in those places to create a have had no conclusive impact, particularly demonstration effect. Only then should officials outside Jawa-Bali. By contrast, policies to improve try to replicate SEZs in less attractive places, human capital, particularly education and social keeping in mind the large fiscal cost of programs, have had the clearest positive impact ineffective or inoperative infrastructure. across all places. Policy makers therefore need to •  In selecting lagging areas for place-based rethink place-based policies to unleash the policies, policy makers should build human potential of lagging areas and ensure the full capital to leverage the places’ comparative participation of those left behind. advantages. Efforts to build human capital could include health, education, and social •  To date, Indonesia’s place-based policies to protection initiatives. Policies should also allow attract firms have had mixed results, people to migrate to the places that offer the best particularly outside Jawa-Bali. Policies to opportunities and highest returns for their skills. improve local amenities and public goods have •• To ensure the inclusiveness of urbanization alleviated constraints in some districts. Policies throughout Indonesia’s portfolio of places, to improve human capital have had the clearest place-based policies should be complemented positive impact across all types of districts. with urban planning and design that take •  Industrial policies for lagging places should into account the needs of all people in a city, maximize the interaction of people, places, and including women and girls, the elderly, those products. To ensure cost-effectiveness, policy with disabilities, and other groups that are makers should sequence their approach to disadvantaged or discriminated against. 275 Large regional differences in population den- This report has focused on policies to sity, economic activity, and employment have leverage urbanization in Indonesia for better characterized Indonesia’s economic geogra- prosperity, inclusiveness, and livability out- phy for centuries. Although it is difficult to comes: to be inclusive, the urbanization pro- estimate the extent of regional inequalities cess needs to yield improvements through before the colonial period, they almost cer- spillovers to those who continue to live in tainly increased under Dutch rule (Hill 2000). rural areas. Chapter 5 contended that reforms When the Dutch East India Company arrived to the system of subnational governance and in Indonesia in the early 17th century, the finance are fundamental to augmenting the regime followed an uneven development supply of basic services and infrastructure strategy in ruling the archipelago. On across all places. Chapter 6 argued that con- Kalimantan, Sulawesi, and Sumatera, the col- nective infrastructure investments need to be onists created extractive enclaves with planta- expanded and strategically targeted, so that tions, mines, and other sites for obtaining Indonesia’s portfolio of places can be better natural resources; on Jawa, they encouraged integrated, thereby yielding a more inclusive more diversified growth. Possibly because of urbanization process. Chapter 7 focused on path dependence, the mass of economic activ- intracity investments, such as affordable ity remains concentrated on Jawa. housing and public transportation, that could Indonesia’s policy makers have often con- improve livelihoods and mobility in cities. centrated on regional disparities because Although better institutions, improved those disparities threaten national unity. The connective infrastructure, and expanded movement for Indonesia’s independence from access to affordable housing and good urban the Dutch arguably succeeded only because it transport would clearly improve urbanization focused on “Bhinneka Tunggal Ika”—the in Indonesia, by themselves they may not be national motto, which translates to “Unity in enough to cover all places and all people. Diversity.”1 In the 1950s, violent separatist Even if all the policies discussed in chapters movements in Sumatera and Kalimantan, 5–7 were adopted, some areas and people notably in Aceh, threatened to overwhelm might continue to lag. Enhancing the full Sukarno’s early presidency (1945–67) and inclusiveness of the urbanization process in dissolve the Indonesian nation. In 1957–58, Indonesia requires place-based policies, which several regions attempted to secede and, in should be complemented with urban planning the 1960s and 1970s, two distant eastern and design that take into account the needs of provinces (Irian Jaya and East Timor) were all people in a city, including women and girls, incorporated into the republic (Hill 1998), the elderly, those with disabilities, and other only for East Timor to later gain indepen- groups that are disadvantaged or discrimi- dence. Given the changes in territory and nated against. the separatist movements, Booth (1988) won- dered whether it would be possible for Indonesia to survive as a unitary state. Characterizing lagging As in many other countries, Indonesian policy makers have used place-based policies places in Indonesia to try to combat regional differences in eco- We can think of economic development as nomic activity. Place-based policies attempt to having three spatial dimensions: the rural– incentivize growth and development in lag- urban dimension, where the main policy con- ging areas, focusing broadly on attracting cern is urbanization; the leading–lagging firms, attracting workers and human capital, regions dimension, where the main policy and improving local public goods. Such poli- concern is territorial development; and the cies include tax subsidies to attract firms to coastal–landlocked countries dimension, specific regions, school construction pro- where the main policy concern is regional grams to improve local human capital, and integration. Indonesia’s rich portfolio of public investments to improve communica- places ranges from large metropolitan areas tions and transport infrastructure. that span multiple districts (multidistrict 276  TIME TO ACT metro areas) and stand out internationally for nonmetro rural areas provided the worst. their large size to emerging urban areas in Finally, most inequality within Indonesia is otherwise predominantly rural districts. The accounted for by inequality within places, as range calls for rethinking the notion of lead- opposed to inequality between places. This ing and lagging regions, which is central to finding is true whether one looks at island- place-based policies. Part 1 (chapters 1–4) of regions, portfolio-of-place areas, or districts. this report provided analytical background Overall, Indonesia appears to have two for this rethinking. types of leading areas and two types of lag- Chapters 1 and 2 documented the concen- ging areas. The first type of leading area tration of Indonesia’s economic prosperity in includes multidistrict metro cores and single- metropolitan areas, especially Jakarta, which district metro areas: they have per capita GDP accounts for more than 25 percent of the coun- either very close to or above the national try’s gross domestic product (GDP) and average, their productivity is positively corre- 12 ­percent of its population. All types of areas lated with their size, which provides evidence except multidistrict metro cores have per cap- of agglomeration economies, and they are ita income below the national average. The per home to the country’s growing middle class. capita income of Jakarta’s metropolitan core The second type of leading area comprises (Daerah Khusus Ibukota Jakarta) is about five sparsely populated, nonmetro resource-rich times that of Jakarta’s periphery and twice that districts: they have high per capita GDP and of other metropolitan cores. Excluding Jakarta, host highly productive firms and workers in the per capita income of the Jawa-Bali area is the natural resource sector; however, they close to that of other island-regions. employ few workers. The two types of lagging Chapter 3 documented that agglomeration areas are the peripheries of multidistrict metro forces are strongest for multidistrict and areas and nonmetro areas (urban or rural). ­ single-district metro areas and more generally This chapter focuses on place-based policies that more populous places are more produc- to attract firms to lagging districts. Drawing on tive, as are places with higher human capital. Indonesia’s rich experience of place-based poli- Connectivity to both domestic and interna- cies, the chapter urges caution in using tax tional markets through transport n ­ etworks incentives to attract firms. Indonesia, like other also matters, though not as much. Reducing countries, has seen many examples of failure excessive congestion forces in ­ metropolitan with such efforts. Such policies also risk pro- areas can increase productivity and, hence, viding tax giveaways to politically connected prosperity in Indonesia, as can improving firms that do not create productivity spillovers. local business environments. Human c ­ apital Policy makers compelled by political economy appears to have strong externalities across all considerations to adopt such policies should places, but the social returns are highest in proceed cautiously to minimize inefficient sunk urbanized metro districts. In areas of low costs. Place-based policies should maximize the urbanization, human capital investments interaction of people, places, and products in should therefore be central to strategies to lagging districts. improve productivity and prosperity. Efforts to encourage other positive agglomeration Place-based policies in forces should complement such investments. Chapter 4 documented that the cores of Indonesia multidistrict and single-district metro areas Many welfare programs, such as conditional offered the best chances of moving out of cash transfers or unemployment assistance, poverty. By contrast, nonmetro rural and are person based, explicitly targeting individ- urban areas offered the worst chances. Single- uals. Place-based policies instead target geo- district metro areas and multidistrict metro graphic areas for special treatment. The “big cores, closely followed by their rural and push” growth theory (Rosenstein-Rodan urban peripheries, provided the best chances 1943) suggests the need for a massive initial of entry into the middle class, whereas investment to make sure the growth process T argeti n g P laces a n d P eople L eft B ehi n d   277 triggered is sustained beyond the point where (Kline and Moretti 2014b). In the United the marginal return of the initial investment Kingdom, Regional Selective Assistance funds equals its marginal cost. Through this thresh- historically provided substantial grants to old effect, place-based policies to attract suf- firms locating in disadvantaged regions ficient economic activity to generate a “big (Criscuolo et al. 2007). The European push” can cause huge increases in productiv- Regional Development Fund uses tax revenue ity (Azariadis and Stachursky 2005; Murphy, to fund generous business subsidies and pub- Schleifer, and Vishney 1989). The productivity lic investments in lagging regions throughout growth could improve welfare not only in the the European Union. From 2007 to 2013, the areas targeted by these policies but also fund disbursed 49 billion euros a year, roughly throughout the country. Although the needed four times the amount spent by the U.S. fed- subsidies might initially be expensive, their eral government. And China’s SEZ strategy agglomeration economies and productivity strengthened property rights and incentivized spillovers could create a self-sustaining virtu- foreign direct investment in selected areas, ous circle of development that provides which experienced major economic growth growth dividends to cover the initial costs. (Wang 2013). Despite the theoretical promise of place- based policies, their global performance has been mixed, and their relative costs and ben- Relationships between subnational governments and the center have evolved efits can be difficult to calculate (Kline and Moretti 2014a). If firms and workers are per- Under Suharto’s New Order government fectly mobile, all benefits accrue to landown- (1968–98), Indonesia operated as a highly ers and are capitalized in housing prices centralized, unitary state. Local policies were (Roback 1982; Rosen 1979). Even if workers implemented by a hierarchical network of are not perfectly mobile, for a place-based local subnational branches of central govern- policy to have an overall positive welfare ment ministries (Lewis 2017). Although dis- impact, the threshold effect of the “big push” trict governments participated in the national intervention, in terms of additional employ- planning process, they had little control over ment created, needs to be more proportional fiscal expenditures (Lewis 2003; Smoke and than the sunk cost (Kline and Moretti 2014a). Lewis 1996). Local development funds largely Without an agglomeration function dispro- consisted of direct spending by the ministries portionate to the inputs, place-based policies that did not pass through district govern- might just relocate economic activity from ments (Crane 1995). one place to another, making one region bet- In addition, from the mid-1970s to the end ter off at the expense of other regions. At best, of Suharto’s reign, various presidential this result leaves national welfare unchanged, instruction programs (instruksi presiden or and, at worst, it wastes resources consumed INPRES) channeled some of the massive oil by moving costs or transfer payments (Bartik boom and mining windfalls of the era into 1991; Glaeser and Gottlieb 2008).2 regional development. The INPRES grants Nonetheless, place-based policies are polit- were of two types. The first were general pur- ically popular and widely used in low-, pose grants to provinces ( Dana Alokasi ­ middle-, and high-income countries. In the Tier I or DATI-I grants) and districts (Dana United States, place-based policies have taken Alokasi Tier II or DATI-II grants), and block the place of industrial policy, embraced by grants to subdistricts (kecamatan) and vil- governors and mayors from both main politi- lages (desa) and urban communities (kelura- cal parties (Kline and Moretti 2014b). In a han). The second were specific purpose grants typical year, the U.S. federal and local govern- covering a wide range of priority areas, ments spend roughly US$95 billion (in cur- including markets, highways, primary schools, rent prices) on such programs, and it is rare reforestation, and health facilities. Before for a large facility to open in the United States 1998, INPRES was the largest and most sig- without receiving some kind of tax incentive nificant vehicle for place-based policies in or subsidy from state and local governments Indonesia, but the total INPRES revenues 278  TIME TO ACT flowing to the regions were much lower than early in Suharto’s regime gave extended tax direct ministry spending. holidays to firms locating outside Jawa. One When President Suharto resigned in May of the first SEZs, created on the island of 1998, a series of sweeping decentralization Batam in Riau Province, close to Singapore, reforms shifted the balance of power from the prepared the way for several related pro- central and provincial governments toward the grams, including the Integrated Economic districts (see chapters 5 and 6 for detailed dis- Development Zone (Kawasan Pengembangan cussion). By 2001, district governments were Ekonomi Terpadu, KAPET) program and the responsible for administering health, educa- current SEZ program. tion, environmental, and infrastructure services and other important functions, whereas the Early incentives center retained control over justice, police, finance, religion, national defense, develop- Early in Suharto’s New Order government, a ment planning, monetary policy, and interna- series of national investment laws reduced tional relations. The INPRES grants were taxes for firms locating in Indonesia. Firms replaced with a series of equalization grants, locating outside Jawa benefitted from principally from the general purpose fund and extended tax breaks.3 Such incentives did not special purpose fund (see chapter 5). Regions name specific places in the outer islands for exploiting oil and natural resources were special treatment, and their impact on eco- allowed to keep a larger share of their resources nomic activity in lagging regions was limited under the new national resources revenue- (Temenggung 2013). In 1984, these early tax sharing fund. incentives were dropped to increase the effi- Indonesia’s decentralization presents both ciency of tax collection and simplify the tax challenges and opportunities for place-based code (Pangestu and Bora 1996). policy making. Place-based policy initiatives Indonesia established Batam as one of its emanating from the national level may encoun- first SEZs in an effort to boost trade and for- ter local implementation difficulties. District eign direct investment (box 8.1). During the governments have different capacities for mon- first few years, the zone saw foreign and itoring and implementing national initiatives, domestic investments in Batam increase sub- but regional autonomy can provide districts stantially and exports in current prices grew with the opportunity to independently create from US$44 million in 1988 to US$3 billion in place-based policies that attract firms and 1996 (Grundy-Warr, Peachey, and Perry 1999). workers. The proximity of individuals allows Between 1988 and 1998, Batam’s population local governments to learn their citizens’ needs more than tripled, from 88,000 to 294,000. and to base policies on that knowledge. Despite the substantial increase in economic It remains to be seen whether the challenges of activity on Batam, several researchers have decentralization can be overcome and whether argued that the zone did not harness the full it provides Indonesian policy makers with potential of agglomeration economies that more effective place-based policies. could produce nationally beneficial productiv- This chapter focuses on two types of place- ity spillovers. Manufacturers set up on Batam based interventions: policies to attract firms to take advantage of lower labor costs (which and policies to attract workers and human have since increased because of minimum capital. Chapters 6 and 7 focused on connec- wage policies). Most manufacturing at one of tive infrastructure and policies to improve Batam’s largest industrial parks consisted of local public goods. low-skill product assembly (Grundy-Warr, Peachey, and Perry 1999). Manufacturing was not associated with research and development Indonesia has a long history of enacting or idea generation (Wong and Ng 2009). place-based policies to attract firms In general, investments did not spill over out- Many tax incentives and other policies in side the industrial parks and were paralyzed by Indonesia were designed to influence firms’ the shortage of skilled labor (Debrah, location decisions. A series of laws enacted McGovern, and Budhwar 2000). T argeti n g P laces a n d P eople L eft B ehi n d   279 BOX 8.1  Batam and the Singapore–Johor–Riau growth triangle Batam is a city on an island of the same name with Singapore and Johor in Malaysia, as a key in the Riau archipelago. This archipelago lies in part of the Singapore-Johor-Riau growth triangle the Sunda straits south of Singapore, and Batam (map B8.1.1). Batam attracted Singapore’s economic has a long history as a port. In 1970, the area policy makers because it promised to reduce labor was designated as a free trade zone, and later it costs for Singapore-based manufacturers while was established as a logistics base for oil and Batam’s proximity would enable manufacturers to gas activities. To facilitate industrial develop- better monitor and supervise their investments. ment, the central government created the Batam In July 2005, the Batam Industrial Bonded Industrial Development Authority, which focused Zone, Bintan Industrial Estate, and Karimun on infrastructure, light manufacturing, industrial Industrial Cooperation Zone were upgraded to real estate, gas, chemicals, and offshore drill- “Bonded Zone Plus” to give investors more legal ing, sponsored by the state-owned oil company, certainty. Reforms were adopted to improve the Pertamina. In 1978, Batam was further designated island’s investment climate, further supported as a “bonded zone,” or duty-free zone, to support in 2006 by a framework agreement between export-oriented industries. Singapore and Indonesia on economic coop- In 1989, a new law was passed allocating eration in the islands of Batam, Bintan, and approximately 1,700 hectares of land on Batam Karimun. In June 2007, Batam was granted free for the development of eight industrial estates. trade zone status, and Bintan and Karimun were Restrictions on foreign investment in industrial granted enclave status. estates were relaxed, and Batam was named, Sources: Based on Grundy-Warr, Peachey, and Perry 1999; Wong and Ng 2009. MAP B8.1.1  The Singapore–Johor–Riau growth triangle N M A L AY S I A JOHOR STR AIT SO Johoe Bahru FM SOUTH CHINA AL AC CA SEA SINGAPORE BATAM BINTAN RIAU ISLANDS INDONESIA LINGGA ISLANDS SUMATERA Source: Sparke et al. 2004. 280  TIME TO ACT Integrated economic development zones They also intended to encourage creation, expansion, and development of micro, small, In 1993, Suharto’s government developed the and medium-size enterprises; and they KAPET program to accelerate development in thought that manufacturing growth from parts of eastern Indonesia. Eastern Indonesia small, medium-size, and large enterprises refers to the island groups of Kalimantan, would provide jobs, expand economic devel- Maluku, Nusa Tenggara, Papua, and Sulawesi. opment, and reduce unemployment and pov- New businesses in KAPET zones were eligible erty. In total, 35 districts were included in the for substantial tax breaks, including a partial program—27 fully and 8 partially, with spe- tax holiday, a 30 percent reduction of taxes cific subdistricts being affected (map 8.1). on capital, expanded choices for depreciation The KAPET program suffered from imple- and amortization of capital and losses, fiscal mentation issues (Rothenberg et al. 2017; loss compensation for 10 consecutive years, Soenandar 2005; Temenggung 2013). Many and a reduced income tax on dividends for zones were created in early 1998, immediately foreign taxpayers.4 The program also sup- before the Asian financial crisis. Given the ported several nonfinancial incentives, includ- unstable business climate, many potential ing business counseling and assistance investors may have been dissuaded from tak- programs; programs to help micro, small, and ing advantage of the program, despite its tax medium-size enterprise owners apply for breaks. Moreover, because a wide variety of loans; a one-stop shop integrated licensing districts were eligible for the program, their system; and 31 priority programs in facilities initial differences in industrial development and infrastructure, investment facilitation ser- and market access may have increased hetero- vices, and human, economic, and natural geneity in performance.6 Finally, the different resources to reduce the costs of business capacities of the newly empowered district registration.5 governments managing KAPETs and their Policy makers hoped that these fiscal and considerable discretion over using program nonfinancial incentives would attract invest- funds may have weakened the program’s ment to KAPET zones, stimulate manufactur- focus. ing growth, and promote export growth. MAP 8.1  The Integrated Economic Development Zone program included 35 districts KAPET districts Non-KAPET districts 1 250 500 Districts excluded from analysis Kilometers Source: Rothenberg et al. 2017. Note: Blue corresponds to districts included in the Kawasan Pengembangan Ekonomi Terpadu (KAPET) program and light gray to districts not included in the program. Information on which districts were included was taken from the text of various presidential decrees. Because of concerns about data quality, Papua and Biak were removed from the analysis. Map has been generated using the 2000 district boundaries. T argeti n g P laces a n d P eople L eft B ehi n d   281 Box 8.2 summarizes a recent empirical The KAPET program was discontinued and assessment of the KAPET program. Along replaced by the SEZ program after the adop- many dimensions, KAPET districts experi- tion of the SEZ Law in 2009. enced no better development outcomes, and in some cases fared worse, than similar d istricts not included in the program ­ SEZs have become an increasingly (Rothenberg et al. 2017). All the same, the prominent part of place-based policy since the big bang decentralization government appeared committed to it, and further modifications in 2000 streamlined the SEZs are perceived differently throughout the incentive structures and provided more world (box 8.3). An Indonesian law adopted ­ flexibility to firms benefitting from them. in September 2009 defines SEZs as bounded BOX 8.2  Assessment of the impact of the Integrated Economic Development Zone program Rothenberg et al. (2017) investigate the short- nontreated sample to include only districts on and long-run effects of the Integrated Economic the outer islands, which are more comparable to Development Zone (Kawasan Pengembangan KAPET districts than districts on Jawa and Bali. Ekonomi Terpadu, KAPET) program. They use Next, they used an inverse probability weight- a spatial equilibrium model to derive reduced- ing approach that explicitly adjusts for poten- form welfare equations that are estimated using tial ex ante differences between KAPET and the rich spatial variation in treated districts (that non-KAPET districts. This approach reweights is, districts that were part of the KAPET pro- the contribution of nontreated districts to the gram) with high-quality data on demographic counterfactual in accordance with their odds of outcomes, measures of regional output and treatment.a growth, and firm-level outcomes. Average treat- The strongest evidence comes from panel ment effects of KAPET on firm entry, output, data on medium and large manufacturers, from productivity, and wages are derived. They also Indonesia’s Industrial Survey (Survei Industri, estimate the impact of KAPET on demographic or SI). Rothenberg et al. (2017) find that incum- outcomes and measures of regional output and bent medium and large firms may have slightly growth. increased their use of labor in KAPET districts, In estimating the impact of place-based poli- but generally they also reduced their use of capi- cies like the KAPET program, a central concern tal and did not experience any changes in capac- is that omitted variables—that is, variables that ity utilization. If anything, the strongest set of a researcher is unable to include in an analysis results pertain to taxes: firms in KAPET districts because of a lack of data—correlated with treat- paid substantially lower sales, license, building, ment may have influenced both selection into the and land taxes than firms in non-KAPET dis- program and outcomes. Rothenberg et al. (2017) tricts. This result is not surprising, given that document that KAPET zones were established the KAPET program was designed to directly in poorer districts that were more rugged, far- reduce taxes in treated districts. It also suggests ther from Jakarta, and less populated than other that firms locating in KAPET districts were more districts. This endogenous program placement profitable and directly benefitted from the tax creates a negative targeting bias that leads naive subsidy; however, because the program did not treated versus nontreated comparisons— that generate increases in entry or migration, the is, the simple comparison of outcomes between scope for productivity spillovers and growth KAPET and non-KAPET districts—to underesti- impacts was quite limited. mate program impacts. To improve the accuracy Source: Rothenberg et al. 2017. of estimation of the KAPET program’s impacts, a. These odds are constructed from a propensity score estimation, where Rothenberg et al. (2017) first restricted the treatment selection depends on observable, predetermined characteristics. 282  TIME TO ACT BOX 8.3  Special economic zones throughout the world Ask three people to describe a special eco- ­ elevant In Indonesia, two specific forms are r nomic zone (SEZ), and they may offer three for place-based policies: export ­ processing very different images. The first may describe a zones or free trade zones, which focus on fenced-in industrial estate in a developing coun- manufacturing for export, and large-scale try, occupied by footloose multinational cor- SEZs, which usually combine residential and porations enjoying tax breaks, with laborers multiuse commercial and industrial activity. working in garment factories in substandard Export processing zones represent a tradi- conditions. The second may recount the “mira- tional model used in low- and middle-income cle of Shenzhen,” a Chinese fishing village trans- countries for almost four decades. Large-scale formed into a cosmopolitan city of 14 million, SEZs represent a more recent form, originat- with per capita gross domestic product growing ing in the 1980s in China and gaining popu- 100-fold in the 30 years since it was designated larity recently. The models are not mutually as an SEZ. A third may think about Dubai or ­ exclusive—many SEZs include export process- Singapore, whose ports support a wide range of ing zone industrial parks. trade and logistics. SEZs describe all these mod- ern economic zones. Source: Farole and Akinci 2011. zones where firms receive certain facilities and recent establishment, no studies have yet evalu- incentives, including the following: ated impacts; however, emerging findings sug- gest best practices to increase the chances of an •  An income tax holiday, with deductions SEZ’s success (Farole and Akinci 2011). and sunsets depending on the size of the Among them are coordinating the provision of investment hard and soft infrastructure, promoting for- •  Waivers for import value added taxes and ward and backward links with the local econ- sales taxes on luxury goods, as well as omy, and providing business support and value added tax waivers on transactions seamless movement of entrepreneurs and between SEZ investors skilled labor between the zones and the local •  Exemptions on import duties on goods economy.7 These steps represent a shift from made in SEZs for the domestic market relying on fiscal incentives to facilitating a •  Foreign ownership of property in SEZs more effective business environment that pro- •  Value added tax and luxury goods tax motes competition, integrates targeted sectors exemptions for foreigners who own with the rest of the economy, and adequately property protects the environment. Unlike the KAPET •  A deduction of leisure taxes paid on program that focused on eastern Indonesia, the investments in tourism-focused SEZs SEZ program has a national coverage. •  Simplified employment arrangements, including foreign worker arrangements •  Simplified immigration processes Indonesia has achieved some success with •  Improved land acquisition place-based policies to improve human •• Simplified licensing procedures capital Twelve SEZs have been established or are Urban areas in Indonesia with higher average planned in Indonesia (map 8.2). levels of schooling are more productive The government of Indonesia also issued because of strong human capital externalities regulations to improve licensing at SEZs so (see chapter 3). Several historical and current that investors could more rapidly develop busi- initiatives, many with an important place- nesses or expand facilities. Given the program’s based component, seek to improve human T argeti n g P laces a n d P eople L eft B ehi n d   283 MAP 8.2  Twelve special economic zones have been established or are planned Arun Lhokseumawe Maloy Batuta Sei Trans Kalimantan Mangkei Galang Morotai Batang Sorong Bitung Palu Tanjung Tanjung Kelayang Api-Api Tanjung Lesung Mandalika Industry Tourism 0 250 500 Kilometers Source: Information retrieved from National Council for Special Economic Zone website (http://kek.go.id/). capital in Indonesia. Education programs 1973 GDP. The program doubled the number include a large primary school construction of schools, creating more than one per program and an expansion of Indonesia’s 500 children ages 5–14. Lagging districts saw higher education system. Policies to encour- significant increases in school construction, age educational attainment and improve and, once an INPRES school was established, school quality could benefit regions with lag- the government recruited teachers and paid ging human capital performance. In the health their salaries.8 An effort to train more teach- sector, a government-run primary health sys- ers paralleled the INPRES program (World tem was developed, and recent policies have Bank 1990).9 created programs to reduce the cost of health Despite publicly announced rules, the gov- care for the poor. In addition, several initia- ernment built slightly more schools in areas tives support social protection. that already had some students enrolled, so the program was somewhat less equalizing than intended (Duflo 2001). Nevertheless, it Education initiatives to improve education triggered considerable economic returns and outcomes and their impacts resulted in substantially higher wages in the Two main initiatives targeted education: the long run for students once they entered the School INPRES for primary school enroll- workforce (Duflo 2001). The exposure to ment and efforts to expand Indonesia’s uni- each school built per 1,000 children increased versity system. The School INPRES, one of years of schooling by 15 percent for the first the first and largest INPRES initiatives, con- cohort of women fully exposed to the pro- structed more than 61,000 schools between gram when they were children, and by 1974 and 1978 in underserved districts to 26 ­percent for their husbands (Breierova and address large gaps in enrollment. The presi- Duflo 2004). The associated increases in dential instructions listed the exact number of schooling led to higher ages at marriage, schools to be constructed in each district. fewer very early births, and lower child mor- Indonesia spent more than US$500 million tality. The program increased labor force par- (in 1990 dollars), or 1.5 percent of its ticipation but may have reduced wages for 284  TIME TO ACT older cohorts from before the expansion because of poor resources. There are five to (Duflo 2004). Because the program created seven elite universities (most on Jawa) and more educated village heads, governance and many rudimentary colleges that offer few the provision of public goods improved skills and little training. Public spending on (Martinez-Bravo 2017). tertiary education remains low relative to In 1950, Indonesia had 10 tertiary educa- comparator countries (Hill and Wie 2012). tion institutions, all on Jawa, with a total of Academics are paid poorly, and staff qualifi- about 6,500 students.10 Under Suharto, a cations have not consistently improved series of laws enabled the growth of private (Welch 2007). Although more than 8 of universities, and the number of private higher 10 high school students express a desire to education institutions increased from fewer continue their education at the university than 400 in 1975 to more than 1,200 in 1995. level, only 3 in 10 can afford to enroll (Myriad Student enrollment increased from about 2013). High fees and the absence of financial 100,000 in 1975 to more than 1.5 million in aid make it difficult for students to attend. 1995. Today, Indonesia has more than 2,000 Efforts to bolster school quality and institutions, roughly 1,900 private and 100 improve learning, particularly in lagging public, and enrollment has continued to regions, could increase urban productivity increase. Between 2000 and 2016, the num- and growth by leveraging human capital ber of workers with at least some tertiary externalities (Moretti 2004): chapter 3 pro- education increased from 5 million to 15 mil- vides evidence of strong human capital exter- lion. Despite the increased supply of skilled nalities for most types of district in Indonesia. labor, the returns to tertiary education were In lagging regions, primary and secondary still large, and a shortage of high-skilled labor schools need better teachers, lower fees, and in many industries is widely recognized more incentives for learning. Improving (World Bank 2014). access to and quality of tertiary education Although disadvantaged regions in could reduce skilled labor shortages in many Indonesia have made great strides in improv- industries and produce even larger productiv- ing access to schooling, largely because of the ity spillovers. School INPRES, they have seen little progress in learning outcomes and school quality. Health initiatives to improve outcomes In  2015, the performance of 15-year-old Indonesian students in science, mathematics, Using experience from a successful program and reading was one of the lowest among in Bandung in the early 1950s that integrated countries participating in the Programme for preventive and curative medicine, Indonesia’s International Student Assessment (PISA), with Ministry of Health developed a hierarchical an average ranking of 62 out of 69 countries. network of hospitals and health centers The percentage of low performers in science throughout the country, aiming to expand among disadvantaged students was among access to care and increase preventive medi- the highest globally (Pellini 2016). However, cine. At the top of the hierarchy are govern- better PISA scores were obtained in larger ment hospitals located in district capitals and settlements (chapter 4), suggesting that inaccessible for most of the population Indonesia’s education system may be failing (Frankenberg 1995). Next are a network of to serve students in lagging areas and students community health centers, which tend to be from disadvantaged backgrounds. the primary source of care, particularly in Despite increased tertiary education facili- rural areas. Health centers are generally ties, the quality of higher education still located in subdistrict capitals and employ a ­ varies, particularly between private and p ­ ublic doctor, a midwife, and one or more nurses. providers. Buchori and Malik (2004) noted On Jawa, health centers had roughly that, even though there are far more private 15  employees in 1991, but, in the outer than public institutions, most private islands, much smaller staffs (van de Walle ­ i nstitutions have relatively lower quality 1994). Some heavily populated subdistricts T argeti n g P laces a n d P eople L eft B ehi n d   285 have more than one health center. Community According to the 1997 round of the health subcenters and integrated health and Indonesian Family Life Survey (before decen- family planning posts round out the bottom tralization), health care quality varied by type tier of the primary health care network. of provider (public health centers versus pri- Subcenters are staffed either by resident para- vate physicians), and most providers outside medical workers or by itinerant health work- Jawa-Bali gave lower-quality service (Barber, ers, and they are open only once or twice a Gertler, and Harimurti 2007). Within regions, week. the poor and wealthy had access to the same Like education, the health sector under quality, but the poorest women reported Suharto concentrated on expanding facilities receiving fewer prenatal procedures (Barber, in disadvantaged areas. Despite more health Gertler, and Harimurti 2007). facilities, quality still varies—an imbalance that has not improved since decentralization. Social assistance Heywood and Choi (2010) found mixed per- formance of the health system across districts, Indonesia’s first antipoverty program was using data from Indonesian Demographic and INPRES for Left-Behind Villages, which Health Surveys, which monitor health supply began in 1994 and ended in 1998 (the earlier and outcome indicators for 10 districts in transmigration program, discussed below, had East and Central Jawa. They saw little some elements of poverty reduction). improvement in use of maternal and child Although almost 20,000 villages were initially health services, a key indicator of perfor- eligible for grants, only 7,000 were designated mance. Government studies show that, in the as “left behind” (Shah et al. 1994). Each vil- 1990s, the efficiency of health resource use lage received a block grant of 20 million varied widely across districts and that most rupiah (roughly US$9,000 in current prices) a district systems operated at suboptimal levels year for infrastructure projects and job cre- (Heywood and Choi 2010). ation to reduce poverty. By 1998, the program Overall health system performance is operated in all villages in four provinces poorer in Indonesia than in regional neigh- (Nusa Tenggara Timur, East Timor, Irian Jaya, bors. At just 3.6 percent of GDP, overall and Maluku), in addition to other villages health spending in Indonesia continues to be throughout Indonesia. one of the lowest, not only in the East Asia INPRES for Left-Behind Villages was one and Pacific region but also globally (World of the first large-scale community-driven Bank 2016). This is due primarily to low development programs in the world. Villages overall government spending and a relatively were selected according to criteria that low share of government spending going to included the quality of local infrastructure, health. Currently, public spending on health is the quality of housing, rates of livestock own- only 1.5 percent of GDP. To achieve the ambi- ership, ownership of consumer durables, the tious target of extending coverage to all availability of electricity, school enrollment Indonesians, the government needs to increase rates, and indicators of health and infant public health spending to ensure adequacy of mortality (Alatas 2000; Daly and Fane 2002). public financing for health. Increased Once selected, villages would decide how to resources should be focused toward those allocate the annual block grant, and groups of interventions that are (proven) effective, such poor people were invited to submit proposals as increased investments in primary health for using the funds. Grants that generated care, promotive interventions, and preventive income and employment were supposed to be interventions, particularly for vulnerable pop- repaid; however, according to Alatas (2000), ulations living in rural and remote locations. only 60 percent of recipients repaid any part Moreover, investments can be maximized by of the funds they received. Although the cen- focusing on a results-based approach that tral government regulated the spending of maximizes the technical efficiency of the lim- other block grants, village governments and ited resources available. local civic groups had extensive leeway in 286  TIME TO ACT determining how these funds would be spent World Bank. Each village could submit two (Akita and Szeto 2000). proposals, which were reviewed at the subdis- Hill (1998) suggests that INPRES for Left- trict level. Projects were implemented by the Behind Villages was broadly positive in its village, with support and guidance from the impacts, though he criticizes village selection district level. By 2010, the program had as arbitrary and lacking rigor. It was associ- financed projects in roughly 15,000 villages ated with a reduction in intra-provincial (Olken 2010). Its urban equivalent, the urban expenditure inequality between 1993 and poverty program, provides credit for small 1996 (Akita and Szeto 2000). A comparison and medium-sized enterprises and funding for of villages that received funds with similar vil- developing infrastructure in poor urban areas. lages that did not found that the program The transmigration program in the 1970s raised household expenditures, increased and 1980s was funded by Indonesia’s massive employment in rural areas, increased employ- oil windfall (box 8.4). When global oil prices ment of children, and raised the proportion of collapsed in the mid-1980s, declining govern- households that were self-employed (Alatas ment revenues forced dramatic cutbacks in 2000). the budget, greatly reducing the number of How well did the program target the poor households sponsored over the coming years. within selected villages? Yamauchi (2010) As a result, numerous sites selected for the found that wealthier and more unequal vil- program never received any transmigrants. lages did a better job of choosing households Bazzi et al. (2016) used the selected but non- for loans. The positive association between participating villages as counterfactuals to inequality and targeting may reflect estimate the average effects of the program on Indonesia’s political context, where village local economic development. Although the heads were incentivized to follow national participating villages were more populated guidelines for targeting the poor. Villages with and denser than the selected but nonpartici- a high population density and young, edu- pating villages, and although they produced cated heads initially exhibited better target- more agricultural output, they were no more ing, but, as the village heads became less productive and no different in nighttime light involved in monitoring benefit allocation, the intensity—a widely used proxy for economic advantage was lost (Yamauchi 2010). activity. The mixed effects of the transmigra- The Kecamatan Development Program, tion program on economic outcomes partially operating at the village level, had two compo- reflect poor matching of farmers to places, nents: grants for developing infrastructure in because many were sent to new environments the village and loans for small businesses. It where the agricultural skills they had acquired began in 1998, funded by a loan from the on Jawa and Bali were not useful. BOX 8.4  Indonesia’s transmigration program In the 1970s, policy makers were concerned rural areas on the outer islands. Planners hoped about Indonesia’s uneven population distribution. that the program would reduce population pres- The inner islands of Jawa and Bali were thought sures, increase food security, and promote nation to be overpopulated, whereas the outer islands building by integrating diverse ethnic groups. The were relatively unsettled and had untapped agri- program remains one of the largest government- cultural potential. Between 1979 and 1988, the sponsored resettlement programs in history. government relocated nearly 2 million voluntary migrants from rural areas on Jawa and Bali to Source: Bazzi et al. 2016. T argeti n g P laces a n d P eople L eft B ehi n d   287 Rethinking place-based As SEZs proliferate, policy makers must learn from experience and anticipate emerg- policies in Indonesia ing and potential issues. Success will require a Policy makers in Indonesia have tried a range more flexible approach to using SEZs to of place-based policies, targeting incentives to leverage comparative advantages and allow lagging places to address spatial inequalities. them to evolve. This approach will require a Place-based policies to attract firms into shift in mindset from the traditional reliance l agging districts have had no conclusive ­ on fiscal incentives and wage restraint to impact, particularly outside of Jawa-Bali. In facilitating a more effective business environ- contrast, policies to improve human capital, ment that fosters innovation, firm-level com- particularly education and social programs, petitiveness, local economic integration, and have had the clearest positive impact across social and environmental sustainability. The all places, which is consistent with the find- approach will also require proactive, flexible, ings presented in chapter 3 on the existence of and innovative policy approaches to address significant human capital externalities for the impact of Indonesia’s macroeconomic vul- most types of district in Indonesia. nerabilities (which include risk of capital out- flows and raising financing costs) on SEZs, and the many unanticipated challenges as Policies should tap the comparative uncertainties in the global economy persist. advantages of lagging places The main challenges to the second type of The core principle of place-based policies to lagging district—that is, nonmetro urban and attract firms to lagging districts should be rural areas—are the poor coverage and quality adjusted to the spatial challenges of district of basic services (health care, education, and spheres of influence. According to the analysis water and sanitation) despite thorough decen- of earlier chapters, the urban and rural tralization, although chapter 2 shows that peripheries of the multidistrict metro areas these districts have caught up to some extent in face the following challenges: basic services. These challenges limit the agglomeration economies the districts can gen- •  Lack of affordable housing close to eco- erate and thus the districts’ attractiveness to nomic density, leading to slum proliferation firms in general. Place-based policies in such in the cores of the multidistrict metro areas areas should prioritize providing basic services •  Lack of mass transit, leading to high by increasing the efficiency of the spending ­ c ongestion costs and labor market financed by the general purpose and special segmentation purpose grants, by mandating collaboration •• Lack of proper urban planning, leading to among districts if necessary for efficient deliv- poor basic services and infrastructure in ery, or even by merging districts or provinces both the cores and the peripheries to reach mandated basic service standards. Given such challenges, place-based policies to Only then, in a second stage, could specific attract firms in the lagging areas of multidis- measures targeting sectors have a chance to trict metros should focus on providing spa- succeed by identifying and addressing the tially targeted incentives for affordable housing human capital and infrastructure constraints in metropolitan areas; spatially targeted incen- to leveraging a district’s comparative advan- tives for mass transit in multidistrict metro tage. Box 8.5 discusses the case for an ecotour- areas; and urban infrastructure development, ism development strategy in lagging districts. health care, and education in the peripheries of multidistrict metro areas. Policies should also focus on developing a small number of SEZs Targeting people left on industrial lands that could be serviced with backbone infrastructure (water, electricity, behind logistics, and information and communica- Women, men, children, the elderly, and peo- tions technology) and freed from red tape to ple with disabilities experience and use attract foreign and domestic investors. urban spaces in different ways that should be 288  TIME TO ACT BOX 8.5  The potential of ecotourism in Indonesia’s lagging districts Despite the powerful forces of agglomeration, visitors are increasingly concentrated in Bali, regions outside prime metropolitan cores are not and their average spending (and consequently consigned to become lagging regions. Avoiding their economic impact) has recently declined. In this fate, however, depends on how effectively 2016, Bali received more than 40 percent of all areas can integrate, both with regional and foreign visitors to Indonesia (figure B8.5.1), with global markets and with the economic core of Jakarta and Batam a distant second and third, their own country. This integration, in turn, and the remainder thinly spread across the rest depends on leveraging key regional assets and of the archipelago. delivering on specific regional factors that have For Indonesian lagging districts to be able to the greatest impact on trade competitiveness. tap their tourism potential, four key constraints Around the world, there are many noncore need to be addressed: (1) poor interministry/ regions for which global and regional trade inte- interagency, central-local, and public-private gration offers substantial scope for moving to a coordination and weak implementation capa- higher and more sustainable growth path. They bilities for tourism development in general, and include regions with access to key trade gateway for monitoring and preservation of natural and infrastructure, regions with competitive clusters cultural assets in particular; (2) continued poor of economic activity, and regions rich in mineral, access and quality of infrastructure and services agricultural, or tourism resources. for citizens, visitors, and businesses; (3) outside The government of Indonesia has set of Bali, limited tourism workforce skills and ­ several objectives to increase the role of tour- ­ private sector tourism services and facilities; ism in the Indonesian economy in its National and (4) a weak enabling environment for private Medium-Term Development Plan (Rencana investment and business entry. Addressing these Pembangunan Jangka Menengah Nasional, constraints in a comprehensive, integrated, and RPJMN) 2015–19. Despite the strong growth incremental manner is therefore key to unlock- in foreign visitor numbers since 2006, Indonesia ing Indonesia’s potential to develop a world- continues to lag key regional competitors (such class tourism industry. as Malaysia, Singapore, and Thailand) in reach- ing its full potential. Furthermore, foreign Source: World Bank (2018). FIGURE B8.5.1  Foreign travelers to Indonesia and Bali, and Bali’s share of total, 1969–2016 14 45 40 12 Foreign tourist arrivals (millions) 35 10 30 8 25 Percent 6 20 15 4 10 2 5 0 0 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 01 03 05 07 09 11 13 15 Indonesia (left scale) Bali (left scale) Bali's share of total (right axis) Source: World Bank 2018. T argeti n g P laces a n d P eople L eft B ehi n d   289 reflected in the design of key urban infra- and socioeconomic background. Such structure. The cost of leaving some segments groups can experience spatial, social, and of the population behind can be high because economic exclusion in countries like of their restricted labor force participation, Indonesia that are still urbanizing and are reduced human capital due to their con- not able to meet the evolving and specific strained access to social services, and cities’ needs of diverse populations (World Bank deprivation of the talents of people directly 2015). Inclusive urbanization principles or indirectly excluded by an unfriendly recognize that all people have a right to urban environment. Improved intraurban inhabit and work in cities, participating transport to integrate cities, highlighted in fully in urban planning and development chapter 7, includes differentiated design to processes (United Nations 2017), and that reach all urbanites. For example, women, extra efforts are needed by urban planners traditionally the primary caregivers, have and designers to promote the full participa- greater access to economic and educational tion of all city inhabitants. opportunities through urbanization. Their Unfortunately, in many cities around the new, varied role gives many women more world, women and girls often experience ver- complex transport patterns than men as they bal abuse, sexual harassment or assault, and journey between places of employment, even extreme violence in public spaces, par- schools, and homes (Greed and Reeves ticularly after dark and when using public 2005). For more than two decades, labor transportation. In Indonesia, the National force participation among women in Commission on Violence Against Women Indonesia has been low, about 51 percent, reported that 31  percent of 5,002 docu- compared with 84 percent among men. mented cases of gender‑based violence in Indonesia’s female participation is also lower 2015 occurred in the public sphere, and that than in countries at similar levels of eco- figure included mostly cases related to sexual nomic development (for example, Bulgaria, violence. The fear of potential harassment Kazakhstan, and the Russian Federation), and violence can further restrict the move- where transport costs favor greater female ments of women and girls, limiting their labor force participation. Badly congested opportunities to safely and equitably access roads in Indonesia’s major cities, particularly economic and educational opportunities in during peak commuting times, and large cities. Furthermore, the lack of accessible urban transport or commuting costs increase buildings, urban infrastructure, and public the time it takes women to perform daily transportation services also reduces opportu- activities outside the home. If public trans- nities for people with mobility issues (includ- port investments reduced commuting costs, ing the elderly and people with disabilities) the time savings could increase female labor to  move freely about cities, preventing force participation, expand urban mobility, such groups from accessing public services as and make women more productive in their well as economic and educational opportuni- daily lives (Witoelar et al. 2017). ties. Because an estimated 12  percent of To enhance the full inclusiveness of the Indonesians have a disability, there is a clear urbanization process in Indonesia, place- need to promote safe, inclusive, and accessible based policies should be complemented urban infrastructure services in Indonesia. with incentives to target urban services to Indonesian cities are making good progress address the specific needs of different gen- in adopting inclusive practices in planning and ders, ages (including children and the designing cities. For example, Daerah Khusus elderly), and people with disabilities. Ibukota (DKI) Jakarta, which is the core of Inclusiveness could mean many different Jakarta’s multidistrict metro area, and things for vulnerable populations, who Surabaya have allocated dedicated funding to have multifaceted needs when living in and improve footpaths and install tactile indica- visiting urban spaces, depending on their tors to help people with visual impairments to individual experiences, mobility, religion, navigate street networks. In September 2017, 290  TIME TO ACT 14 mayors of Indonesian cities signed the helping to deter criminal or antisocial Charter of the Network of Indonesian Mayors behavior. CPTED principles also encour- for Inclusive Cities, committing to eliminate age the right mix of land uses and activa- discrimination against people with disabilities. tion of identified public spaces at night, Cities including Bandung, Solo (Surakarta), creating attractive and inclusive public and Surabaya address gender and disability spaces to which many people are drawn. issues during the country’s participatory plan- People with disabilities have low rates of ning and budgeting process, known as labor force participation, and, when they Musrenbang ( Musyawarah Perencanaan do participate, they are more likely than Pembangunan) (Rifai, Asterina, and Hidayani others to be unemployed. Some 51 percent 2016). Also, transport companies increasingly of people with disabilities are employed, offer designated “women-only” safe spaces on compared to 70 percent of able-bodied buses and trains. Additional efforts are needed people. Numerous people with disabilities ­ to drive the importance of delivering truly report that the distance to work has an inclusive cities to Indonesians. “Car-centric” impact on their decision making. Many cities still place cars, rather than people, at the work only part time or are in informal center of development, creating busy and dan- employment. Getting around Jakarta is par- gerous streets; insufficient, disregarded, or ticularly challenging for them because the inappropriately designed pedestrian crossings; existing public transport system does not and poorly designed pedestrian pavements adequately cater to their needs (box 8.6). that create major barriers to public transport Busy streets, a lack of suitable street cross- access. Instead, Indonesian cities could adopt a ings, and poorly designed pedestrian pave- more “people-centric” approach to their plan- ments present major barriers to public ning and design processes. Indonesia can also transport access. Many individuals with improve safety in its cities through improve- disabilities require assistance when travel- ments in streetscapes. There are often not ing. Because most modes of public transit enough streetlights to illuminate public streets, are not designed to meet their needs, they and public spaces can be littered with land- find it hard to navigate door to door on scaping and street furniture that serve as hid- their own. ing spots for potential assailants. Indonesian authorities at the national and The physical layout of urban public subnational levels could consider adopting spaces and infrastructure can be planned “place-making” processes, through which and designed to be more socially inclusive. city makers design public places collabora- City planning agencies could integrate tively with communities including women, inclusive urban planning and design prin- children, the elderly, and people with disabil- ciples in the spatial planning process, uti- ities to create unique places that strengthen lizing inclusive design and construction connections between communities and the standards when designing public spaces places in which they live. Cities could adopt and public buildings. For example, crime similar design approaches in community prevention through environmental design consultations for major public projects and (CPTED) is a well‑known concept that facilitate targeted focus group discussions or encourages passive surveillance of public design “charrettes” (collaborative design spaces and streets, creating more “eyes on sketching sessions) with diverse groups to the street” (Jacobs 1961), and could be plan and design cities that cater to all abili- incorporated in local‑level urban design ties, genders, ages, and other diversity groups guidelines ( Rencana Tata Bangunan (see box 8.7 for an example). Safety audits Lingkungan —RTBL). A good application could also be conducted when designing new of this concept is in the design of residen- public spaces to develop recommendations tial uses above the ground level that “look and actions that will improve safety or per- out” onto public spaces and streets below, ceptions of safety. T argeti n g P laces a n d P eople L eft B ehi n d   291 BOX 8.6  The mobility challenge for people with disabilities Getting around an Indonesian city as a pedes- Some major urban areas have been trying to trian is an arduous task, particularly if one has up their game—Jakarta and Surabaya, for exam- a disability; the mobility challenge can some- ple, have both allocated hundreds of ­billions of times be insurmountable. Footpaths rarely slope rupiah in recent years to improve footpaths. down to the road, and when they do, barriers In  2017, Jakarta Utara, which is one of the are placed across entry points to keep motor- districts that forms Daerah Khusus Ibukota cycles out, simultaneously rendering pavements Jakarta, alone dedicated 42 billion rupiah inaccessible to wheelchair users. This lack of (US$2.93 million) to fixing the pavements of just accessibility does not just affect the mobil- five subdistricts. ity of people with disabilities; it also affects The Jakarta Provincial Government is imple- their independence. According to Badan Pusat menting the Law on People with Disabilities by Statistik (Statistics Indonesia), about 12 p­ ercent making public services more accessible: ensur- of Indonesia’s population has a moderate or ing that buildings are equipped with ramps and severe disability—45 percent of this number guiding blocks; that public entrances to build- never graduated from primary school, a sharp ings are wide enough for wheelchairs; that contrast to the 13 percent nongraduation rate lifts, ramps, and audio-visual information are for able-bodied Indonesians. This disadvantage available; and that guiding blocks are properly is compounded because lower levels of educa- installed in footpaths. tion lead to lower rates of employment. Some good examples were introduced by the In 2016, the University of Indonesia found Jakarta governor in 2016. He installed S-shaped that only 51 percent of people with disabilities portals instead of bollards to stop motorbikes are employed, compared with 70 percent of from using pavements while still enabling wheel- able-bodied citizens. Only 20 percent of those chair users to pass through. He also ensured with severe disabilities are currently working. that all newly developed or refurbished foot- Most people with disabilities in Indonesia work paths correctly used yellow guiding blocks for in the informal sector, leaving them less finan- the visually impaired, and he often referred to cially stable with less legal protection. Such bar- Tokyo as the standard to which Jakarta should riers to leaving the house and getting around strive to achieve. disadvantage people with disabilities in a vast range of daily activities. Source: Walton 2018. BOX 8.7  Inclusion along Samoa’s Apia waterfront The government of Samoa is adopting inclu- planning document outlines inclusive p ­ rinciples sive urban development principles in the plan- and strategies to develop the waterfront, par- ning, design, and implementation of its 10‑year ticularly by requiring development to have a waterfront development project in Apia. “people-oriented” focus. A guiding principle During the planning process, a safety audit for the plan’s Urban Design Standards is to was conducted along the waterfront area to “consider the needs of all” by designing spaces identify potential vulnerable areas and spaces abilities, that consider the needs of people of all ­ in which people felt unsafe. Targeted commu- genders, ages, and other diverse groups. These nity consultations were conducted with diverse planning and design guidelines will help to groups including a focus group discussion guide implementation of the Apia waterfront with disability‑interest groups, a design ideas over the coming years. workshop with children, and design charrettes including women stakeholders. The project Source: Government of Samoa 2016. 292  TIME TO ACT Notes real economic  growth. MP3EI was replaced by initiatives linked to the Nawacita Plan 1. Geertz (1971, 19) argued that “[a]rchipelagic when the Jokowi administration came into in geography, eclectic in civilization and het- power in 2014, but the development of SEZs erogeneous in culture, [Indonesia] flourishes continued to be a government priority in when it accepts and capitalizes on its diver- developing regions outside Jawa. sity and disintegrates when it denies and sup- 8. Each school was designed for three teachers presses it.” and 120 pupils. 2. In low- and middle-income countries, cor- 9. Although the program rapidly increased the ruption also often affects these policies. The number of new teachers by 43 percent, the spatially targeted investment subsidies may proportion meeting minimum qualification be given to particular cities or areas where requirements did not worsen between 1971 politically connected individuals or firms and 1978. stand to directly benefit. 10. These institutions included the Bandung 3. Law 1/1967 on foreign investment and Law Institute of Technology, founded in 1920, 6/1968 on domestic investment granted sev- Gadja Madah University, a state-run uni- eral incentives for investors to establish new versity in Yogyakarta founded in 1949, and businesses and to increase investment. These Universitas Indonesia in Depok, founded in incentives included both a five-year holiday 1950. on payments of corporate taxes and exemp- tions on import taxes for capital goods and equipment. Although most incentives applied to all investments in Indonesia, firms located References outside of Jawa benefitted from an addi- Akita, T., and J. Szeto. 2000. “Inpres Desa tional year of tax breaks, as long as projects Tertinggal (IDT) Program and Indonesian met basic size and foreign exchange earning Regional Inequality.” Asian Economic Journal requirements (Ikhsan 2006). 14 (2): 167–86. 4. Other benefits included an exemption on Alatas, V. 2000. “Evaluating the Left Behind Villages income taxes for imported capital goods, raw Program in Indonesia: Exploiting Program materials, and other production-related equip- Rules to Identify Effects on Employment and ment and a 50 percent reduction of building Expenditures.” PhD thesis, Princeton University, and land taxes, including those paid for both Princeton, NJ. new construction and business expansion. Azariadis, C., and J. Stachursky. 2005. “Poverty 5. Rothenberg et al. (2017) provide more Traps.” In Handbook of Economic Growth, background on the one-stop shop licensing volume 1, edited by P. Aghion and E. Durlauf, program and evaluate the extent to which 295–384. Oxford: North-Holland. the program affected levels of business Barber, S., P. Gertler, and P. Harimurti. 2007. registration. “Differences in Access to High-Quality 6. Districts (and subdistricts within those dis- Outpatient Care in Indonesia.” Health Affairs tricts) were selected for the KAPET program 26 (3): 352–66. according to several different, but vague, cri- Bartik, T. J. 1991. Who Benefits from State and teria: favorable geography, economic growth Local Economic Development Policies? potential, potential for large investment Kalamazoo, MI: W. E. Upjohn Institute for returns, and possession of leading sectors Employment Research. capable of boosting growth for hinterland Bazzi, S., A. Gaduh, A. Rothenberg, and M. Wong. areas. Despite the criteria, each KAPET zone 2016. “Skill Transferability, Migration, and was established in a different province in Development: Evidence from Population eastern Indonesia. Resettlement in Indonesia.” American 7. Special economic zones were an important Economic Review 106 (9): 2658–98. part of President Yudhoyono’s Masterplan Booth, A. 1988. “Can Indonesia Survive as a for the Acceleration and Expansion of Unitary State?” Indonesia Circle 58: 32–47. Economic Development of Indonesia Breierova, L., and E. Duflo. 2004. “The Impact (MP3EI) which focused on the development of Education on Fertility and Child Mortality: of economic corridors. This was an ambi- Do Fathers Really Matter Less than Mothers?” tious plan of integrated infrastructure devel- Working Paper No. 10513, National Bureau of opment aiming for expanded countrywide Economic Research, Cambridge, MA. T argeti n g P laces a n d P eople L eft B ehi n d   293 Buchori, M., and A. Malik. 2004. “Asian Greed, C., and D. Reeves. 2005. “Mainstreaming Universities: Historical Perspectives and Equality into Strategic Spatial Planning Policy: Contemporary Challenges.” In The Evolution of Are Town Planners Losing Sight of Gender?” Higher Education in Indonesia, 249–77, edited Construction Management and Economies 23 by P. Altbach and T. Umakoshi. Baltimore, MD: (10): 1059–70. Johns Hopkins University Press. Grundy-Warr, C., K. Peachey, and M. Perry. 1999. Crane, R. 1995. “The Practice of Regional “Fragmented Integration in the Singapore- Development in Indonesia: Resolving Central- Indonesian Border Zone: Southeast Asia’s Local Coordination Issues in Planning ‘Growth Triangle’ against the Global Economy.” and Finance.” Public Administration and International Journal of Urban and Regional Development 15 (2): 139–49. Research 23 (2): 304–28. Criscuolo, C., R. Martin, H. Overman, and J. Van Heywood, P., and Y. Choi. 2010. “Health System Reenen. 2007. “The Effect of Industrial Policy Performance at the District Level in Indonesia on Corporate Performance: Evidence from Panel after Decentralization.” BMC International Data.” Working Paper. https://­ editorialexpress​ Health and Human Rights 10 (1): 3. https:// .com/cgi-bin/conference/­d ownload.cgi?db​ bmcinthealthhumrights.biomedcentral.com​/­arti _name=IIOC2008&paper_id=445. cles/10.1186/1472-698X-10-3. Daly, A., and G. Fane. 2002. “Anti-Poverty Hill, H. 1998. “The Challenge of Regional Programs in Indonesia.” Bulletin of Indonesian Development in Indonesia.” Australian Journal Economic Studies, Taylor & Francis Journals of International Affairs 52 (1): 19–34. 38 (3): 309–29. ———. 2000. The Indonesian Economy . Debrah, Y., I. McGovern, and P. Budhwar. 2000. Cambridge: Cambridge University Press. “Complementary or Competition: The Hill, H., and T. K. Wie. 2012. “Indonesian Development of Human Resources in a South- Universities: Rapid Growth, Major Challenges.” East Asian Growth Triangle: Indonesia, Malaysia, Presentation to the ANU Indonesian Update and Singapore.” International Journal of Human Conference, Australian National University, Resource Management 11 (2): 314–35. Canberra, September 21. https://­ c rawford​ Duflo, E. 2001. “Schooling and Labor Market . a n u ​ . e d u . a u / a c d e / i p / p d f / u p d a t e / 2 0 1 2​ Consequences of School Construction in /Indonesia_Update_2012_1_7_H_Hil_and​ Indonesia: Evidence from an Unusual Policy _Thee_KW.pdf. Experiment.” American Economic Review Ikhsan, M. I. 2006. “FDI and Tax Incentives in 91 (4): 795–813. Indonesia.” Paper presented at the International ———. 2004. “The Medium Run Effects of Symposium on FDI and Corporate Taxation: Educational Expansion: Evidence from a Large Experiences of Asian Countries and Issues in School Construction Program in Indonesia.” the Global Economy, Hitotsubashi University, Journal of Development Economics 74 (1): Tokyo, Japan. 163–97. Jacobs, J. 1961. The Death and Life of Great Farole, T., and G. Akinci. 2011. “Special Economic American Cities. New York: Random House. Zones: Progress, Emerging Challenges, and Kline, P., and E. Moretti. 2014a. “People, Future Directions.” Directions in Development Places, and Public Policy: Some Simple Series. Washington, DC: World Bank. Welfare Economics of Local Economic Frankenberg, E. 1995. “The Effects of Access to Development.” Annual Review of Economics Health Care on Infant Mortality in Indonesia: 6: 629–62. A Fixed Effects Approach to Evaluating Health —— —. 2014b. “Local Economic Development, Services.” Health Transitions Review 5 (2): Agglomeration Economies, and the Big Push: 143–62. 100 Years of Evidence from the Tennessee Geertz, C. 1971. “Report to the Ford Foundation Valley Authority.” Quarterly Journal of Concerning a Program for the Stimulation Economics 129: 275–331. of Social Sciences in Indonesia.” Institute for Lewis, B. D. 2003. “Property Tax in Indonesia: Advanced Study, Princeton, NJ. Measuring and Explaining Administrative Glaeser, E. L., and J. D. Gottlieb. 2008. “The (Under-) Performance.” Public Administration Economics of Place-Making Policies.” Brookings and Development 23: 227–39. Papers on Economic Activity 2008: 155–239. ——— . 2 0 1 7 . “ D o e s L o c a l G o v e r n m e n t Government of Samoa. 2016. Apia Waterfront Proliferation Improve Public Service Delivery? Development Project: Waterfront Plan 2017– Evidence from Indonesia.” Journal of Urban 2026. Apia: Government of Samoa. Affairs 39 (8): 1047–65. 294  TIME TO ACT Martinez-Bravo, M. 2017. “The Local Political Informal Sector.” World Development 80 (C) Economy Effects of School Construction 96–113. in  Indonesia.” American Economic Journal: Shah, A., Z. Quereshi, A. Bagchi, B. Binder, and Applied Economics 9 (2): 256–89. H.  Zou. 1994. Intergovernmental Fiscal Moretti, E. 2004. “Workers’ Education, Spillovers, Relations in Indonesia: Issues and Reform and Productivity: Evidence from Plant-Level Options. Washington, DC: World Bank. Production Functions.” American Economic Smoke, P., and B. D. Lewis. 1996. “Fiscal Review 94 (3): 656–90. Decentralization in Indonesia: A New Approach Murphy, K., A. Shleifer, and R. Vishny. 1989. to an Old Idea.” World Development 24 (8): “Industrialization and the Big Push.” Journal of 1281–99. Political Economy 97: 1003–26. Soenandar, E. S. 2005. “Government Policy Myriad. 2013. Improving Access and Equity to in Solving Uneven Regional Development Indonesian Higher Education for Candidates between West and East Indonesia: Case Study from Disadvantaged Backgrounds. Jakarta, on KAPET.” Economic Journal of Hokkaido Indonesia: Myriad. University 34: 171–92. Olken, B. 2010. “Direct Democracy and Sparke, M., J. Sidaway, T. Bunnell, and C. Grundy- Local Public Goods: Evidence from a Field Warr. 2004. “Triangulating the Borderless Experiment in Indonesia.” American Political World: Geographies of Power in the Indonesia- Science Review 104 (2): 243–67. Malaysia-Singapore Growth Triangle.” Pangestu, M., and B. Bora. 1996. “Evolution of Transactions of the Institute of British Liberalization Policies Affecting Investment Geographers 29 (4): 485–98. Flows in the Asia Pacific.” Policy Discussion Temenggung, D. 2013. “Policies to Promote Paper 96/01, Centre for International Economic Development and Integration of Lagging Studies, University of Adelaide, Adelaide, Regions: The Indonesian Experience.” In The Australia. Internal Geography of Trade: Lagging Regions Pellini, A. 2016. “Indonesia’s PISA Results Show and Global Markets , 209–27, edited by Need to Use Education Resources More T. Farole. Washington, DC: World Bank. Efficiently.” The Conversation, December 15. United Nations. 2017. New Urban Agenda . http://theconversation.com/indonesias-pisa​ New York: United Nations. -results-show-need-to-use-education-resources​ van de Walle, D. 1994. “The Distribution of -more-efficiently-68176. Subsidies through Public Health Services in Rifai, A., N. Asterina, and R. Hidayani. 2016. Indonesia, 1978–87.” World Bank Economic “Making All Voices Count: Improving the Review 8 (2): 279–309. Transparency, Inclusivity and Impact of Walton, K. 2018. “Navigating Indonesia’s Participatory Budgeting in Indonesian Cities.” Streets.” New Naratif , August 20. https:// Kota Kita. www.makingallvoicescount.org. newnaratif.com/journalism/accessibility-in​ Roback, J. 1982. “Wages, Rents, and the Quality -indonesia/share​/xuna/3366297a637d4a3a35 of Life.” Journal of Political Economy 90: 8dfc6faad2fcf5/. 1257–78. Wang, J. 2013. “The Economic Impact of Special Rosen, S. 1979. “Wage-Based Indexes of Urban Economic Zones: Evidence from Chinese Quality of Life.” In Current Issues in Urban Municipalities.” Journal of Development Economics, 74–104. edited by P. Mieszkowski Economics 101: 133–47. and M.  Straszheim. Baltimore, MD: Johns Welch, A. 2007. “Blurred Vision? Public and Hopkins University Press. Private Higher Education in Indonesia.” Higher Rosenstein-Rodan, P. N. 1943. “Problems of Education 54 (5): 665–87. Industrialisation of Eastern and South-Eastern Witoelar, F., A. D. Rothenberg, T. Y. Wicaksono, Europe.” The Economic Journal 53: 202–11. T. Gracner, and B. Sikoki. 2017. How Jakarta’s Rothenberg, A., S. Bazzi, S. Nataraj, and Traffic Affects Labor Market Outcomes for A.  Charim. 2017. “When Regional Policies Women and People with Disabilities: Results from Fail: An Evaluation of Indonesia’s Integrated a Baseline Survey. Jakarta, Indonesia: Australia Economic Development Zones.” RAND Labor Indonesia Partnership for Economic Governance. and Population Working Paper WR-1183, Wong, P. K., and K. K. Ng. 2009. Batam, Bintan and Washington, DC. Karimun: Past History and Current Development Rothenberg, A., A. Gaduh, N. E. Burger, C. Chazali, towards Being a SEZ . Singapore: Asia I. Tjandraningsih, R. Radikun, C. Sutera, and Competitiveness Institute, Lee Kuan Yew School S. Weilant. 2016. “Rethinking Indonesia’s of Public Policy, National University of Singapore. T argeti n g P laces a n d P eople L eft B ehi n d   295 World Bank. 1990. Indonesia: A Strategy for ———. 2018. Indonesia–Integrated Infrastructure Sustained Poverty Reduction. Washington, DC: Development for National Tourism Strategic World Bank. Areas (Indonesia Tourism Development ———. 2014. Tertiary Education in Indonesia: Project). World Bank, Washington, DC. Directions for Policy. Washington, DC: World Yamauchi, C. 2010. “Community-Based Targeting Bank. and Initial Local Conditions: Evidence ———. 2015. “World—Inclusive Cities Approach from Indonesia’s IDT Program.” Economic Paper.” Report No. AUS8539, World Bank, Development and Cultural Change 59 (1): Washington, DC. 95–147. ———. 2016. “Indonesia—Health Financing System: Spend More, Right and Better.” Policy Brief October 28, World Bank, Washington, DC. 296  TIME TO ACT SPOTLIGHT 3 The Invisible Crisis of Wastewater Management in Indonesia Urban population growth generates powerful with the remaining 95 percent sending it into productivity-enhancing agglomeration forces, rivers, agricultural fields, and groundwater but it also gives rise to congestion forces on sources through leakage or unsafe disposal. infrastructure, basic urban services, land and Moreover, centralized public sewerage sys- housing markets, and the environment that, tems exist in only 13 cities, and even those in the absence of appropriate policies, can systems have much idle capacity because of undermine the prosperity, inclusiveness, and low rates of household connections. 1 livability outcomes associated with urbaniza- Compared with many other countries, the tion. These adverse congestion forces are par- condition of Indonesia’s sanitation is dire ticularly evident in Indonesia’s wastewater (figure S3.1). management, with a growing population con- Most urban households use on-site sanita- tinuing to rely almost entirely on nature to tion: flush toilets with septic tanks, pit deal with human waste. latrines, or soak pits. Most septic tanks are The alarmingly low rate of wastewater not built to environmental standards and treatment is becoming obvious. The fetid are not desludged regularly.2 Even in cases smell and gray canals in Jakarta’s metropoli- where septic tanks undergo desludging, the tan core (Daerah Khusus Ibukota Jakarta, or effluent is not disposed of in treatment plants. DKI Jakarta) are stark and vivid reminders The risks caused by unsafe disposal are exac- that only 2 percent of households in a popu- erbated by frequent flooding of drainage sys- lation of 10 million are connected to public tems and many households’ reliance on sewers. Only 5 percent of the Indonesian groundwater for their water supply for drink- population safely disposes of wastewater, ing, cooking, and cleaning. FIGURE S3.1  Access to safely managed sanitation in Indonesia is desperately lagging 80 73 74 70 Percent of population with access 70 to safely managed sanitation 60 50 39 39 40 32 30 24 20 10 5 0 Indonesia Cambodia Iraq Brazil Malaysia China Australia World Source: Calculations based on data from WHO and UNICEF 2017. The consequences Sanitation Program estimated that Indonesia lost US$6.3 billion in 2007 (using 2005 Inadequate wastewater management damages prices) because of poor sanitation and public health, especially that of women and hygiene, equivalent to about 2.3 percent of children, harms the environment, and reduces the country’s gross domestic product, whereas prosperity. improved sanitation would lead to an annual Greater risk of disease, poor child gain of over US$4.5 billion (World Bank health, and stunting. Inadequate wastewa- 2018).3 ter management damages public health, especially that of women and children, undermining both the livability of cities and The causes the inclusiveness of urbanization. Poor The most binding constraints in wastewater groundwater quality contributes to infant management in Indonesia are poor gover- mortality in low-income areas of Indonesia nance, weak implementation of the legal and (212 per 1,000 births), which is more than regulatory framework, and a general lack of 3.5 times the rate in other middle-income citizen demand and political will. countries in Southeast Asia (59 per 1,000). Regulations exist on water quality preser- Indonesia also suffers from a dispropor­ vation and effluents but are not implemented. tionately high incidence of typhoid for its Only about 10 percent of subnational govern- region  and income (Winters, Karim, and ments (SNGs) have adopted the central regu- Martawardaya 2014). Moreover, stunting lations, and the central government has no has become a severe health issue, because way to oblige the other SNGs to adopt them. repeated exposure to fecal pathogens can Even when SNGs pass the regulations, they cause poor absorption and nutrient loss have limited resources for enforcement and through diarrhea and poor gut function, oversight. leading to undernutrition in children. With Under Indonesia’s “big bang” decentraliza- 33 percent of Indonesian children in metro tion, the central government is responsible for and urban areas exposed to poor water and policy making and SNGs for wastewater sanitation, the future opportunity costs are management (see chapters 5 and 6). However, considerable. Poor health attributable to the allocation of roles among local depart- inadequate wastewater management also ments and agencies varies greatly from one reduces the aggregate human capital of city to another. Whereas city or provincial Indonesian cities, undermining a crucial technical units typically carry out construc- driver of urban prosperity and inclusiveness tion, other entities operate and maintain facil- (see chapters 3 and 4). ities and collect for services. Those operating Severe environmental degradation. Unsafe and billing entities generally lack the author- wastewater management pollutes water ity to retain revenue, which flows back to sources by increasing inorganic nutrients and SNG budgets. Thus, operating funds for degrading water quality. As pollutants satu- wastewater management are allocated by rate the oceans, algae will develop, take over SNGs that may be unaware of needs, and the sea, and suffocate it. As an island nation, expenditure is subject to more layers of Indonesia will be severely harmed. It is esti- bureaucracy. This complex and poor gover- mated that as much as 70 percent of the coun- nance arrangement discourages increased try’s groundwater pollution stems from household connections, cost recovery, and leaking septic tanks and septage disposed of better service. in waterways (WASHplus 2010). The risks of Demand-side problems interacting with substantial ecological harm and declines in supply-side unresponsiveness is the main biodiversity and wildlife (such as fish) are driver of substandard service (Winters, Karim, only rising. and Martawardaya 2014). On the supply Economic losses and unrealized potential side, the construction of urban wastewater revenue . The World Bank’s Water and infrastructure is expensive, particularly 298  TIME TO ACT because of the costs of land acquisition and streamlined sanitation priorities in local the challenges of installing new underground government budgeting; cost-reflective and piping in densely populated areas. High affordable tariffs; trained staff with ade- ­ transaction costs deter efforts to get the city quate capacity to carry out their mandates; legislature to accept new governance arrange- and well-defined monitoring and enforce- ments. Other transaction costs arise in seek- ment. Supporting all SNGs to implement ing support from the local constituency for their citywide inclusive sanitation strategies the construction of fecal sludge or wastewater requires a large-scale and long-term techni- treatment plants in urban neighborhoods. cal assistance program and adequate financ- Generally, local government officials see these ing support and incentives from the central transaction costs as higher than the political government. benefits. Moreover, they expect only limited Increasing the demand from citizens and willingness to pay for a new service, so SNGs the willingness of residents to pay for better would have to fund long-term operation and sanitation services is critical to SNGs’ readi- maintenance. ness to invest in infrastructure. National pub- On the demand side, citizen demand is too lic awareness campaigns coupled with local low to catalyze SNG action. Urban residents household sensitization could be part of such appear generally content with the septic tanks efforts. Furthermore, the capacity of commu- they have already invested in and do not link nities should be developed through participa- poor wastewater management with health tory planning processes, so they can help problems. They fear the financial cost of con- prepare and implement citywide sanitation nection and service fees. Citizens do not know strategies. what government agency is responsible for The role of women is particularly impor- wastewater management and do not tant in stimulating latent demand. Not only know where to go if they need help. The con- have women played a critical role in creat- fusion dissipates demand: until it is resolved, ing community awareness, but also they hopes for improved wastewater management have been key in several instances to per- will remain unrealistic. suading households to agree to construction and to help with it (Bader and Ghanem What can be done? 2013; Watts 2004). Because women are the primary household water managers, their Service improvements can be achieved only ability to create demand for wastewater through a complete citywide inclusive sani- management must be pursued in order to tation strategy (Blackett, Hawkins, and increase sustainability. Participatory plan- Heymans 2013), in which everybody bene- ning and consultations, with adequate rep- fits from adequate sanitation service deliv- resentation of women, can engage ery outcomes. The strategy should embrace households as the beneficiaries that govern- a diversity of technical solutions, with ment and service providers are in theory mixed and incremental approaches includ- ultimately accountable to. ing on-site and centralized and decentral- Finally, understanding and engaging the ized sewerage systems, and cover the whole political economy of sanitation in Indonesia is sanitation service chain, with due consider- crucial to sustainable improvement (Mason, ation for effective resource recovery and Batley, and Harriss 2014; McFarlane and reuse. Silver 2016; World Bank 2004). Possible SNGs should be supported to plan and approaches include recognizing windows of implement their strategies. Appropriate opportunity for engagement, identifying sec- governance arrangements need to be in tor champions, and strategically sequencing place to provide sustainable service: local development support to create incentives for government regulation, legislation, and long-term investment and institutional ordinances; administratively and financially reform. Partnerships with civil society also independent sanitation service providers; contribute. S P O T L I G H T   299 Notes Mason N., R. Batley, and D. Harriss. 2014. The Technical Is Political: Understanding the 1. Private sewerage systems also exist in large Political Implications of Sector Characteristics housing and commercial estates, developed for the Delivery of Sanitation Services. London: by private developers, connecting all build- Overseas Development Institute. ings within their enclosed area. These are WASHplus. 2010. “USAID—A Rapid Assessment conventional sewer systems with mechanical of Septage Management in Asia.” Sanitation treatment plants. Updates (blog), April 21. https://sanitation​ 2. See Indonesia’s Standard No. SNI 03-2398- updates.blog/2010/04/21/usaid-a-rapid-assess​ 2002, Design Procedures for Septic Tanks with ment-of-septage-management-in-asia/. Percolation Systems (Tata Cara Perencanaan Watts, S. 2004. “Women, Water Management, and Tangki Septik Dengan Sistem Resapan). The Health.” Emerging Infectious Diseases 10 (11): World Bank and Australian Aid (2013) assess- 2025–26. ment of sanitation in Indonesia found that only Winters, M. S., A. G. Karim, and B. Martawardaya. 8 percent of these septic tanks were built to the 2014. “Public Service Provision under standard. Conditions of Insufficient Citizen Demand.” 3. Economic gains would be realized through World Development 60: 31–42. better health, longer lives, increased produc- World Bank. 2004. World Development Report tivity, reduced water treatment, improved 2004: Making Services Work for Poor People. inland fisheries, improvements to the envi- Washington, DC: World Bank. ronment, and increased tourism. ———. 2018. Economic Impacts of Sanitation in Southeast Asia: A Four-Country Study References Conducted in Cambodia, Indonesia, the Philippines, and Vietnam under the Bader, H., and M. Ghanem. 2013. Women Economics of Sanitation Initiative (ESI) . Empowerment Toward Wastewater Reuse: Research Report 46351. Washington, DC: Relink Practical and Strategic Gender Needs World Bank. in Palestinian Rural Area. Ramallah: Birzeit World Bank and Australian Aid. 2013. “Indonesia University. Country Study.” East Asia Pacific Region Blackett, I. C., P. M. Hawkins, and C. Heymans. Urban Sanitation Review, World Bank and 2013.  Poor-Inclusive Urban Sanitation: An Australian Aid. Overview. Water and sanitation program case WHO (World Health Organization) and study. Washington, DC: World Bank. UNICEF (United Nations Children’s Fund). McFarlane, C., and J. Silver. 2016. “The Political 2017. Progress on Drinking Water, Sanitation City: ‘Seeing Sanitation’ and Making the Urban and Hygiene: 2017 Update and SDG Political in Cape Town.” Antipode 49 (1): Baselines . Geneva, Switzerland: WHO and 125–48. UNICEF. 300  TIME TO ACT SPOTLIGHT 4 The Potential of Smart Cities Indonesian cities are potential engines of Smart city technologies for urban trans- ­ prosperity, inclusiveness, and livability. They port can mitigate traffic congestion and accounted for more than 55 percent of the improve mobility, thus aiding the integration national population in 2018 and nearly of metro and urban areas. Singapore’s elec- 60  ­p ercent of gross domestic product in tronic road pricing mechanism has reduced 2010–16 (see chapters 1 and 2). For some, they traffic congestion by 15 percent since its serve as “escalators out of poverty,” but introduction in 1990 (Woetzel et al. 2018), they remain underleveraged in delivering pros- and hours lost in vehicles in Amsterdam perity and livability for all because of excessive dropped by 10 percent with the introduction congestion forces constraining the benefits of of an intelligent traffic management system agglomeration and inadequate connectivity and regional coordination initiatives.2 Beijing both across and within places, as detailed in works with technology companies to tackle this report. To improve prosperity, inclusive- parking shortages and promote parking opti- ness, and livability, Indonesian cities must, mization through user-oriented mobile appli- among other things, address congestion con- cations. In Japan, near-field communication straints and foster positive agglomeration smart cards are used for digital payments on forces. The smart cities approach, as defined by public transit subways, trains, monorail, and the European Commission, aims to do pre- ancillary services (such as station lockers and cisely this by “focusing on the strategic use of vending machines), allowing for seamless new technology and innovative approaches to intermodal transit and reducing congestion at enhance the efficiencies and competitiveness of transit hubs. Globally, on-demand ride-­ hailing cities.”1 Smart city technologies are intended to services and mobility apps such as Go-Jek, optimize urban infrastructure, resources, and Grab, Uber, and Lyft have provided hyperlo- spaces to respond more effectively, efficiently, cal transport options to areas not served by and dynamically to the needs and demands of traditional public transit (see chapter 7 for a a city’s residents, workers, and visitors (United brief discussion of mobility apps in Indonesia). Nations Task Team on Habitat III 2015). Smart city technologies can remotely col- lect, monitor, and analyze real-time data to improve basic services and mitigate damage Livability to the environment. Smart water technologies Congestion forces are negative effects attrib- have allowed Grand Lyon, France, to reduce utable to the pressure of a city’s population nonrevenue water by 15 percent by detecting on its infrastructure, basic services, land and leakages quickly and directing predictive housing markets, and the environment. Just maintenance so that the city recovers the costs as with previous waves of urban innovation, incurred in treatment and pumping smart city technologies can potentially miti- (Leinmiller and O’Mara 2013). In Fiji, gate negative congestion forces by improving Internet of Things and remote sensing tech- the efficiency of urban management and ser- nologies have been piloted to monitor water vice delivery, thereby facilitating a higher level quality, critical to preventing outbreaks of of agglomeration, and enhancing urban waterborne diseases. In the Republic of livability. Korea, an integrated waste management platform and solar-powered smart waste bins Janeiro’s Smart Operations Center has reduced allowed Seoul’s metropolitan government to emergency response times by 30 percent by reduce waste collection costs by 83 percent allowing more than 30 local and state agencies and eliminate waste overflow in the city and utilities to coordinate and collaborate on (Tomás 2017). monitoring, operating, and managing emer- For a fluid land market, mitigating negative gency data and response (Schreiner 2016). congestion forces is essential. Poor land man- By mitigating negative congestion forces, agement can lead to disputes, land grabs, land smart city applications have improved livabil- degradation, and lost socioeconomic opportu- ity, including the accessibility, affordability, nities. Land-related smart city technologies quality, and safety of urban services and the such as satellite imagery and geospatial data, built environment. A recent study concluded open-source geographic information system that smart city tools can reduce crimes by (GIS) software, digitized land administration 30–40 percent and fatalities by 8–10 percent systems, and national spatial data infrastruc- (measured against baseline starting points), ture can reduce errors and inconsistencies in accelerate emergency response times by land titling and certificates, making land regis- 20–35 percent, shave average commute times try and transfer processes more efficient, trans- by 15–20 percent, lower the disease burden parent, and accessible. For example, Ho Chi by 8–15 percent, cut greenhouse gas emis- Minh City has piloted mobile applications sions by 10–15 percent, and reduce water with geospatial data and machine learning to consumption by 20–30 percent (McKinsey enforce zoning governance (Goldblatt et al. Global Institute 2018). 2018), and Georgia has implemented block- chain technologies to improve land registries. In housing, smart buildings, energy Inclusiveness ­ management systems (EMSs), and district Because the benefits of urbanization should heating mechanisms can help reduce energy be shared both across and within cities, inclu- consumption and carbon dioxide emissions, sion is a key objective for many smart city ini- with beneficial impacts on the environment.3 tiatives. In Bangladesh, blockchain-enabled Building automation systems can control light- peer-to-peer solar energy trading platforms ing, water, security, heating, ventilation, and have networked households with and without air conditioning to provide efficient ­ services rooftop solar systems, enabling communities that directly benefit the health and well-being to earn a direct income from the sun.4 The of inhabitants, the cost and efficiency of oper- mobile phone–based pay-as-you-go energy ations, and the environment. For example, business model in Kenya and Tanzania has Yokohama’s introduction of EMSs in residen- also delivered affordable electricity to tial buildings reduced energy consumption by unserved and underserved populations 20 percent, and EMSs in commercial buildings (Sanyal 2017). Mobile technology has further reduced peak energy consumption also by enabled formal and informal city workers to 20 percent (C40 Cities 2015). access payment schemes, such as M-PESA in Urban safety, security, and disaster risk Africa and Alipay in China. management can greatly improve with smart Furthermore, e-government and open plat- city technologies. Real-time public safety forms allow metro and urban areas to deliver information through sensors, closed circuit services through a two-way channel, enabling video surveillance, facial recognition, crowd stronger citizen and community involvement. data analytics, and automated information For example, Barcelona’s Decidim Barcelona processing have allowed London and Mecca aims to deliver “digital sovereignty” through to enhance security during major events like its digital and democratic platform for citizen the Summer Olympics and the hajj. Japan’s participation, and Seoul’s e-Government aims nationwide warning system, called the J-alert, for social innovation led by citizens. quickly informs the public of threats of earth- In Honolulu, the city has developed a mobile quakes, tsunamis, volcanic eruptions, severe application, Honolulu 311, which allows citi- weather, and special emergencies. Rio de zens to use personal smartphones to report 302  TIME TO ACT abandoned vehicles, broken streetlights, exchange, stimulating human capital exter- ­ i llegal dumping, and other urban issues. nalities. Smart city coordinating platforms Jakarta’s Smart City Program has incorpo- bring together cities, industries, investors, rated a similar mechanism for solid waste researchers, policy makers, small and medium management. The new channels engage citi- enterprises, and other actors. The platforms zens and tap into their knowledge, so that provide a real testing ground for prototype metro and urban areas improve citizen own- projects in development. Barcelona’s Smart ership and, ultimately, address some of their City Barcelona Platform and the European more critical and complex challenges. C o m m i s s i o n ’s E u r o p e a n I n n o v a t i o n Some countries have rolled out national Partnership on Smart Cities and Communities digital citizen/e-government platforms and are examples. services. In Estonia, 99 percent of public ser- vices are available online on a one-stop-shop government portal called e-Estonia. The por- Implications for Indonesia tal provides hundreds of services such as Indonesia’s growing consumer class and e-Residency, i-Voting, e-Tax, e-Business, younger generation will drive innovation e-Banking, e-Ticket, e-School, e-Ambulance, and growth through new ideas, technologies, telemedicine, and more. In India, more than and opportunities. Urban areas are home to an 1.2 billion residents are enrolled in Aadhaar, expanding middle class (see chapters 2 and 4). the world’s largest national digital/biometric By 2045, when Indonesia will celebrate the identity platform, which supports various centenary of its independence, the share of the public services. country’s population living in towns and cities is expected to reach more than 70 percent (see chapter 1), and the continued urbanization Prosperity will raise incomes and increase consumer In addition to fostering fluid land markets, spending on discretionary items and services efficient urban services, and strong citizen par- (Razdan, Das, and Sohoni 2014). A key factor ticipation, cities must do more to achieve pros- in economic growth in Indonesia is its young perity. They must resolve barriers that population—the median age of the population undermine positive agglomeration forces or in 2015 was 28 years (chapter 1)—born amid exacerbate negative congestion forces to doing technological advances and presenting a tre- business such as inefficient processes, lack of mendous opportunity for smart cities. transparency, and corruption. Some barriers Indonesia’s growing mobile penetration must be dealt with at a national level, such as and increasing digital literacy favor smart city labor market regulations and cross-border success. There were 132.7 million Internet trade facilitation. Many, however, are local, users and 69.4 million smartphone users in such as the barriers to starting businesses, Indonesia at the end of 2016, making it one obtaining construction permits, connecting to of the largest smartphone markets in the utilities, registering properties, accessing credit, world. With more than 70 percent of Internet paying taxes, enforcing contracts, and resolv- traffic originating from mobile devices, ing insolvencies (World Bank 2018; see also Indonesia is now the world’s third-largest chapter 3). In some countries and cities, smart mobile commerce market after China and city applications and digital transformation India (Indonesia-Investments 2017). have sped and streamlined cumbersome busi- Furthermore, Indonesia has about 80 million ness processes such as registries, procurement, social media users and is among the biggest and permit applications. Many cities are fur- users of Facebook and Twitter worldwide. ther exploring disruptive technologies to The increasing mobile penetration and digital improve business environments, like Nigeria’s literacy of citizens create an ideal foundation biometric identification to reduce fraud, waste, for smart cities. The use of the web by firms and corruption (Jha 2018). to conduct business, though improving, Smart city initiatives and processes can remains low across Indonesia’s metro facilitate knowledge innovation and areas (chapter 3), as does both the speed of S P O T L I G H T   303 most Internet connections and the reach of the government of Indonesia has highlighted the Internet beyond Jawa (chapter 6). the importance of smart cities in its National Private sector appetite for urban technol- Urban Policy. Under this policy, Indonesia ogy offers Indonesian cities an opportunity. aims to achieve “Sustainable Cities 2045”— The public sector cannot do everything to smart, livable, and resilient cities. Consistent achieve prosperity, inclusiveness, and livabil- with this policy, the Indonesia 100 Smart ity through smart cities. The private sector City Movement began in 2017, bringing and private citizens play an active and funda- together the relevant line ministries— mental role in shaping a city’s performance. Ministry of Communications and As the ride-hailing service Go-Jek proves, Information Technology, Bappenas, Ministry smart city innovations generate revenue for of Home Affairs, Ministry of Public Works private sector companies, which have access and Housing, and the Presidential Staff to funding unavailable to the public sector. Office—municipal governments, and tech- However, such services can also generate neg- nology advisers (box S4.1). The movement ative externalities, which may require regula- aims to systematically supply cities with the tions (see also chapter 7). governance and institutional support, tech- Governance and institutional develop- nical knowledge, and capacity to plan and ment are essential to realizing the prosperity, implement smart city applications to inclusiveness, and livability benefits of smart m itigate negative congestion forces and ­ cities. Recognizing the potential and neces- stimulate positive agglomeration forces in sity of technology in urban development, each city. BOX S4.1  Indonesia 100 Smart City Movement and the Association of Southeast Asian Nations Smart Cities Network In 2017, an expert committee appointed by the and characters of the respective regions. Under Ministry of Communications and Information the guidance and assistance of the Ministry of Technology selected 25 cities and districts to pilot Communications and Information Technology, Indonesia’s 100 Smart City Movement, which the 25 cities and districts have prepared smart aims to achieve 100 smart cities by 2019. The city development plans, which will integrate cities were selected from 70 cities and districts into their broader city master plans, to be imple- identified on the basis of local financial capac- mented in the next 5–10 years.b The ministry is ity, public service performance, participation in targeting another 75 smart cities in 2018. the Green City Program of the Ministry of Public In line with the Indonesian Government’s Works and Housing, and the Sustainability City Movement, the Association of Southeast Asian Index measured by Bappenas. The final 25 cities Nations (ASEAN) Smart Cities Network (ASCN) and districts for the pilot were selected on the Cooperation Program, which was announced at basis of vision and leadership, local government the 32nd ASEAN Summit in 2018, includes three policies and regulations, institutional readiness Indonesian cities—Banyuwangi (in Jawa Timur and human resources, availability of infrastruc- Province), Jakarta, and Makassar (in Sulawesi ture and information and communications tech- Selatan Province). The ASCN is designed to nology applications, and other features.a achieve the shared goal of smart city develop- The 25 cities and districts expressed their ment in the ASEAN region through innovation commitment to the movement through a mem- and a strengthened ASEAN community in the orandum of understanding signed during the midst of very complex urban challenges. Indonesia Smart City Summit 2017. They have been receiving guidance and mentoring from a. Ministry of Communications and Information Technology, Ministry of National Development Planning/National Development Planning Agency, Ministry of Home expert teams on the fundamental aspects of smart Affairs, Ministry of Public Works and Housing, Presidential Staff Office 2017. city development, customized to meet the needs b. Ibid. 304  TIME TO ACT Notes /volume-29/issue-12/water-utility-management​ /smart-water-a-key-building-block-of-the​ 1. The World Bank’s World Development -smart-city-of-the-future.html. Report 2016: Digital Dividends defines a McKinsey Global Institute. 2018. “Smart Cities: smart city as a “city that leverages the lat- Digital Solutions for a More Livable Future.” est in technology and connectivity to make McKinsey & Company. https://www.mckinsey. better decisions and achieve the urban aspi- com/~/media/mckinsey/industries/capital%20 rations of its residents,” which is a slightly projects%20and%20infrastructure/our%20 narrower definition than the European insights/smart%20cities%20digital%20​ Commission’s. solutions%20for%20a%20more%20livable​ 2. For more information, see the Amsterdam Smart %20future/mgi-smart-cities-full-report.ashx. City Homepage, https://amsterdamsmartcity​ Ministry of Communications and Information .com/projects/smart-traffic​-management. Technology, Ministry of National Development 3. In the United States, the building sector con- Planning/National Development Planning sumed 47.6 percent of all energy produced Agency, Ministry of Home Affairs, Ministry and was responsible for 44.6 percent of of Public Works and People’s Housing, ­ carbon dioxide emissions in 2013. Presidential Staff Office. 2017. “100 Smart City 4. For more information, see https://www.me​ Movement.” Jakarta. -solshare.com/. Razdan, R., M. Das, and A. Sohoni. 2014. “The Evolving Indonesian Consumer.” McKinsey & Company. References Sanyal, S. 2017. “‘Pay-as-You-Go’ Solar Could Electrify Rural Africa.” World C40 Cities. 2015. “Cities 100: Yokohama—City- Resources Institute (blog), February  8. wide Rollout of Smart Energy Management.” http://www.wri.org/blog/2017/02/pay-you​ C40 Cities, October 30. https://www.c40.org/ -go-solar-could-electrify-rural-africa. case_studies/cities100-yokohama-city-wide- Schreiner, C. 2016. “International Case Studies rollout-of-smart-energy-management. of Smart Cities: Rio De Janeiro, Brazil.” Goldblatt, R., K. Kaiser, H. T. L. Tran, and K. Vu. Institutions for Development Sector, Fiscal 2018. “Artificial Intelligence for Smart Cities: and Municipal Management Division Insights from Ho Chi Minh City’s Spatial Discussion Paper IDB-DP-447, Inter-American Development.” The Data Blog , March 8. Development Bank, Washington, DC. https://blogs.worldbank.org/opendata/artificial​ Tomás, J. P. 2017. “How Seoul Used the IoT to -intelligence-smart-cities-insights-ho-chi-minh​ Improve Waste Management and Collection.” -city-s-spatial-development. Enterprise IoT Insights, October 9. https://​ Indonesia-Investments. 2017. “Indonesia is the enterpriseiotinsights.com/20171009/smart-cities​ World’s Fastest Growing Mobile-Commerce /how-seoul-improved-waste-collection-via-smart​ Market.” Indonesia-Investments, May 8. https:// -waste-management-tool-tag23-tag99. www.indonesia-investments.com/business​ United Nations Task Team on Habitat III. 2015. /business-columns/indonesia-is-the-world-s​ “Smart Cities.” HABITAT III–Issues Paper 21, -fastest-growing-mobile-commerce-market​ United Nations. /item7802?. Woetzel, J., D.-Y. Lin, M. Sridhar, and S.-E. Jha, A. 2018. “Top 7 Disruptive Technologies Yap. 2018. “Smart Cities in Southeast Asia.” for Cities.” Sustainable Cities (blog), April 12. McKinsey Global Institute Discussion Paper, http://blogs.worldbank.org/sustainablecities​ McKinsey & Company, Singapore. /top-7-disruptive-technologies-cities. World Bank. 2016. World Development Report Leinmiller, M., and M. O’Mara. 2013. “Smart 2016: Digital Dividends. Washington, DC: Water: A Key Building Block of the Smart World Bank. City of the Future.” WaterWorld, December 8. ———. 2018. Doing Business 2018: Reforming to https://www.waterworld.com/articles/print​ Create Jobs. Washington, DC: World Bank. S P O T L I G H T   305 Environmental Benefits Statement The World Bank Group is committed to reducing its environmental footprint. 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 lowered and shipping distances decreased, resulting in reduced paper consumption, 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 recy- cled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed 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. Indonesia has urbanized rapidly since its independence in 1945, profoundly changing its economic geography and giving rise to a diverse array of urban places. These places range from the bustling metropolis of Jakarta to rapidly emerging urban centers in hitherto largely rural parts of the country. Although urbanization has produced considerable benefits for many Indonesians, its potential has only been partially realized. Time to ACT: Realizing Indonesia’s Urban Potential explores the extent to which urbanization in Indonesia has delivered in terms of prosperity, inclusiveness, and livability. The report takes a broad view of urbanization’s performance in these three key areas, covering both the monetary and nonmonetary aspects of welfare. It analyzes the fundamental reforms that can help the country to more fully achieve widespread and sustainable benefits, and it introduces a new policy framework— the ACT framework—to guide policy making. This framework emphasizes the three policy principles of Augment, Connect, and Target: • Augment the provision and quality of infrastructure and basic services across urban and rural locations • Connect places and people to jobs and opportunities and services • Target lagging areas and marginalized groups through well-designed place-based policies, as well as thoughtful urban planning and design. Using this framework, the report provides policy recommendations differentiated by four types of place that differ in both their economic characteristics and the challenges that they face— multidistrict metro areas, single-district metro areas, nonmetro urban areas, and nonmetro rural areas. In addition to its eight chapters, Time to ACT: Realizing Indonesia’s Urban Potential includes four spotlights on strengthening the disaster resilience of Indonesian cities, the nexus between urbanization and human capital, the “invisible” crisis of wastewater management, and the potential for smart cities in Indonesia. If Indonesia continues to urbanize in line with global historical standards, more than 70 percent of its population will be living in towns and cities by the time the country celebrates the centenary of its independence in 2045. Accordingly, how Indonesia manages this continued expansion of its urban population—and the mounting congestion forces that expansion brings—will do much to determine whether the country reaches the upper rungs of the global ladder of prosperity, inclusiveness, and livability. ISBN: 978-1-4648-1389-4 90000 9 781464 813894 SKU 211389