E N E R G Y A N D M I N I N G S E C T O R B O A R D D I S C U S S I O N P A P E R P A P E R N O . 7 M A Y 2 0 0 3 Regulation by Contract: A New Way to Privatize Electricity Distribution? Tonci Bakovic, Bernard Tenenbaum and Fiona Woolf THE WORLD BANK GROUP The Energy and Mining Sector Board AUTHORS DISCLAIMERS Tonci Bakovic (tbakovic@ifc.org) is a Senior Energy Specialist The findings, interpretations, and conclusions expressed in at the International Finance Corporation. Before joining the this paper are entirely those of the authors and should not be IFC, he was the Power Sector Reform Coordinator for the attributed in any manner to the World Bank, to its affiliated World Bank in Bolivia and then Dominion Energy's Business organizations, or to members of its Board of Executive Directors Development Manager for Latin America. Mr. Bakovic co- or the countries they represent. wrote the Bolivian Electricity Law on a leave of absence from Ernst & Young. The material in this work is copyrighted. No part of this work may be reproduced or transmitted in any form or by any means, Bernard Tenenbaum (btenenbaum@worldbank.org) is a Lead electronic or mechanical, including photocopying, recording, or Energy Specialist at the World Bank. Before joining the World inclusion in any information storage and retrieval system, without Bank, he served as the Deputy Associate Director of the the prior written permission of the World Bank. The World Bank Office of Economic Policy at the U.S. Federal Energy encourages dissemination of its work and will normally grant Regulatory Commission. permission promptly. For permission to photocopy or reprint, please send a request with complete information to the Copyright Fiona Woolf (fiona.woolf@cmck.com) is a senior partner in the Clearance Center, Inc, 222 Rosewood Drive, Danvers, MA Energy Practice at CMS Cameron McKenna. She has acted as a 01923, USA, fax 978-750-4470. All other queries on rights and legal and policy advisor on issues relating to power sector reform licenses, including subsidiary rights, should be addressed to the and regulation in more than 30 separate countries or jurisdic- Office of the Publisher, World Bank, 1818 H Street N.W., tions. She recently authored a book titled Global Transmission Washington DC, 20433, fax 202-522-2422, e-mail: Expansion: Recipes for Success (PennWell) that compares trans- pubrights@worldbank.org. mission pricing and expansion policies across several countries. CONTACT INFORMATION To order additional copies please call the Energy Help Desk. 202-473-0652 energyhelpdesk@worldbank.org This paper is available online www.worldbank.org/energy/ The material in this work is copyrighted. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photo- copying, recording, or inclusion in any information storage and retrieval system, without the prior written permission of the World Bank. The World Bank encourages dissemination of its work and will normally grant permission promptly. For permis- sion to photocopy or reprint, please send a request with complete information to the Copyright Clearance Center, Inc, 222 Rosewood Drive, Danvers, MA 01923, USA, fax 978-750-4470. All other queries on rights and licenses, including sub- sidiary rights, should be addressed to the Office of the Publisher, World Bank, 1818 H Street N.W., Washington DC, 20433, fax 202-522-2422, e-mail: pubrights@worldbank.org. E N E R G Y A N D M I N I N G S E C T O R B O A R D D I S C U S S I O N P A P E R P A P E R N O . 7 M A Y 2 0 0 3 Regulation by Contract: A New Way to Privatize Electricity Distribution? Tonci Bakovic, Bernard Tenenbaum and Fiona Woolf The World Bank, Washington, DC THE WORLD BANK GROUP The Energy and Mining Sector Board Copyright © 2003 The International Bank for Reconstruction and Development/The World Bank. All rights reserved CONTENTS FOREWORD ......................................................................................................................3 ABSTRACT.........................................................................................................................4 ACKNOWLEDGEMENTS....................................................................................................4 ACRONYMS AND ABBREVIATIONS..................................................................................4 EXECUTIVE SUMMARY......................................................................................................5 1. INTRODUCTION.........................................................................................................10 A. Recent Events ........................................................................................................10 B. The Wrong Advice?...............................................................................................10 2. REGULATORY INDEPENDENCE ..................................................................................11 A. Why Encourage Regulatory Independence?........................................................11 B. What Went Wrong?...............................................................................................12 3. REGULATION BY CONTRACT .....................................................................................13 A. What Is Regulation by Contract? .........................................................................13 1. Two Definitions ...................................................................................................13 2. Key Characteristics..............................................................................................16 3. What Government Entity Should Administer the Contract?...................................17 4. Power Purchase Agreements: a Useful Precedent?...............................................17 B. Is a Regulatory Contract Different from a Commercial Contract? ......................18 1 C. Should Regulation by Contract Be Just a Transition Mechanism? ......................18 4. REAL-WORLD REGULATORY EXPERIENCES: BRAZIL AND INDIA .............................19 A. Regulation by Contract: The Latin American Approach......................................19 B. The Special Case of Brazil ....................................................................................19 C. What Went Wrong in Brazil?................................................................................20 1. Vagueness in Tariff-Setting Provisions..................................................................20 2. Uncertainty about Pass-Through for Power-Purchase Costs...................................21 3. Low Allowed Prices for Pass-Through of Power-Purchase Costs.............................21 4. Foreign Exchange Risk ........................................................................................22 5. Uncertainty in the Legal Framework....................................................................22 6. Lack of Respect for Contracts ..............................................................................22 D. Regulation by Principles: The Case of India (So Far)..........................................23 E. Regulation by Contract: A Proposal For India .....................................................24 1. Is It Legal?..........................................................................................................24 2. Objections to Multi-year Tariffs............................................................................25 5. THE DETAILS OF THE REGULATORY CONTRACT: WHO BEARS WHAT RISK? ...........26 A. Overview of Risks.................................................................................................26 B. Passing-Through the Cost of Power Purchases ....................................................27 1. Purchases Where the Distributor Does Not Have Discretion .................................29 2. Purchases Where the Distributor Has Discretion ..................................................29 C. Loss-Reduction Targets .........................................................................................34 1. Types of Losses ...................................................................................................34 2. Can Latin America's Success be Repeated in Africa and India? ............................36 D. Foreign Exchange Risks........................................................................................38 1. Convertibility Risk and Exchange-Rate Risk ..........................................................38 2. Indexing .............................................................................................................38 E. Obligation to Supply.............................................................................................40 1. Toward a Precise Definition.................................................................................41 2. Starting Points Matter..........................................................................................41 6. DEALING WITH DISPUTES .........................................................................................44 A. Who Regulates the Regulator?.............................................................................44 B. What Gets Disputed? ............................................................................................44 C. Different Approaches to Dispute Resolution........................................................45 1. The Local Court System.......................................................................................45 2. International Arbitration......................................................................................46 3. Mediation...........................................................................................................48 4. Expert Panels ......................................................................................................48 5. A Specialized Appeals Tribunal............................................................................49 D. Partial Risk Guarantees: A Mechanism for Ensuring Commitment?...................50 1. How Would It Work?...........................................................................................50 2. Why Do It? .........................................................................................................51 3. Is It Feasible? .....................................................................................................51 7. CONCLUDING OBSERVATIONS.................................................................................52 1. Independence is not enough...............................................................................52 2. The regulatory contract must be a political contract.............................................53 3. "Regulation by contract versus regulation by commission" is a false dichotomy. ........54 4. Regulation by contract is a new name for an old paradigm. ................................54 5. Electricity consumers cannot be the forgotten third party to a regulatory contract........55 2 6. Investors must have confidence that the contract will be enforced fairly and efficiently....55 7. The heart of a regulatory contract is a pre-specified, performance-based, multi-year tariff-setting system. ...........................................56 8. A regulatory contract is sustainable only if the underlying economics are viable. ........56 9. A multi-year tariff system can be put into operation even in the absence of high-quality data. ....................................................................57 10. Regulation by contract should be reserved for private distribution companies. ........57 APPENDIX A. Controllable and Non-Controllable Costs: A Proposed Regulatory Framework ............................................................58 APPENDIX B. Obligation to Serve: A Proposed Regulatory Framework .............................60 APPENDIX C. Quality of Service: A Proposed Regulatory Framework ................................62 REFERENCES ...................................................................................................................64 BOXES Box 1. The French Water Sector: Regulation Without a Regulator.......................................15 Box 2. The Delhi Privatizations: A Partial Regulatory Contract ............................................25 Box 3. Automatic Pass Through: It's Not What You Say, It's What You Do ..........................28 Box 4. What Costs Are Controllable? ................................................................................30 Box 5. Adjusting for Losses: Specific Implementation Issues...............................................35 Box 6. Andhra Pradesh: Where There's a Will, There's a Way............................................36 Box 7. When the Gap is Large ..........................................................................................37 TABLES Table 1. Regulation by Contract: Generation Versus Distribution........................................17 Table 2. Alternative Regulatory Models..............................................................................23 Table 3. Major Risks for Distribution Companies and their Lenders ....................................27 Table 4. Trends in Foreign Exchange Rates in Selected Countries Where American and Foreign Power Had Subsidiaries .........................................38 Table 5. Increase in End-User Tariffs with Different Levels of Indexing for Foreign Exchange Fluctuations (%)....................................................40 FOREWORD The worldwide movement towards power sector reform in Regulatory contracts are not a new concept. Regulatory developing countries began more than 10 years ago. contracts have been combined with independent or par- Usually, the reforms have involved some combination of tially independent regulatory commissions in many Latin restructuring, privatization and unbundling. Regulatory American countries, and this combination has generally reform has also been a key element of the overall reform been successful in inducing and sustaining private sector package. And, in almost every instance, the principal rec- investment in more than 60 privatizations of electricity dis- ommendation for regulatory reform has been to create an tribution systems. "independent regulatory agency." The paper goes beyond general principles. It provides a The motivation for this recommendation was straightfor- wealth of details on how specific features of regulatory ward. It was thought that successful power sector reform contracts have been implemented in various countries required depoliticizing the tariff-setting process and that and the lessons that can be drawn from this experience. an independent regulatory commission would be best able In particular, it reviews the real-world regulatory experi- to do this. While there is still general acceptance that ence of several countries, with special emphasis on how implementation of tariff-setting in the power sector must certain key design elements (e.g., pass-through of power be depoliticized (i.e., be more of a technical than a politi- purchase costs, foreign exchange fluctuations, technical cal exercise) if the overall reform is to be sustainable, and commercial loss reductions and obligation to serve) 3 there is also a growing consensus, achieved with the ben- are dealt with in regulatory contracts. It examines why the efit of the 20/20 hindsight, that "independence is not regulatory contracts in Brazil have been less successful enough" to achieve this goal. than those in other Latin American countries. And since disagreements are almost inevitable in any contract, the In particular, there is growing realization that regulatory paper also considers the strengths and weaknesses of sev- independence, while still very important, must be com- eral traditional and non-traditional approaches to dealing bined with a clearly specified tariff framework. Or, to put it with disputes. While the paper's focus is on the design and in another way, independence does not mean that a newly implementation of regulatory contracts, it points out that created regulatory commission should have total discre- regulatory contracts, by themselves, are not a "magic solu- tion in deciding on the substance and process of post- tion." No regulatory contract, no matter how well designed, reform tariff setting. The key lesson of the last 10 years is will be sustainable unless there is a reasonable likelihood that independence, by itself, does not create regulatory that the regulated enterprise's revenues will cover its costs. commitment. Therefore, the principal recommendation of this paper is that regulatory independence should be I believe that this paper makes a major contribution to combined with a clearly specified regulatory contract our understanding of real-world regulatory issues in that must be negotiated by political authorities. This is, in developing countries because it provides important and effect, a recommendation that successful power sector practical insights on how regulation can and should be reform in World Bank client countries requires the combi- practiced. Most people would agree that the sustainability nation of two distinct regulatory traditions--the Anglo- of any regulatory system ultimately depends on its success American tradition of independent regulatory entities with in providing investors with "confidence" and consumers the French tradition of a well-specified regulatory contract or with "protection." This paper provides a number of ideas concession agreement. on how both of these goals can be achieved. I would like to commend the three authors for bringing focus and The specific focus of the paper is on how regulatory clarity to an important debate that will be of interest to contracts can be combined with independent regulatory anyone who is concerned with the future of power sector commissions to promote successful privatization of elec- reform in developing countries. tricity distribution. The essence of the regulatory contract is a pre-specification in one or more formal or explicit Jamal Saghir agreements of the formulas and procedures that deter- Director, Energy and Water mine the prices that a distribution company will be Chairman, Energy and Mining Sector Board allowed to charge for the electricity that it sells. March 2003 4 ABSTRACT In many developing countries, both governments and investors have expressed disappointment with the per- formance of recently privatized electricity distribution companies. Governments complain that tariffs have increased without visible improvements in service. Investors contend that they have not earned reasonable returns on their investments. Both sides often express dissatisfaction with the new independent regulatory commissions established at the time of privatization. In particular, investors argue that the commissions have not lived up to their commitments and almost always side with consumer interests. Some investors claim that the design of the new regula- tory system in many developing and transition economies is fundamentally flawed. They often recom- 5 mend that independent regulatory commissions be sup- plemented or replaced by more explicit "regulation by contract." This paper examines whether regulation by contract or a combination of regulation by contract and regulatory independence would provide a better regulatory system for developing countries that wish to privatize some or all of their distribution systems. The paper: · Describes the key characteristics of regulation by con- tract as it has been implemented in several develop- ing countries · Focuses on how regulatory contracts in several coun- tries handle certain key issues (pass-through of power- purchase costs, foreign exchange fluctuations, loss reduction and the obligation to serve) · Describes the strengths and weaknesses of different approaches for dealing with disputes that inevitably arise in the application of regulatory contracts · Compares and contrasts some recent real-world regu- latory experiences of distribution entities in Latin America and India. Examines some of Brazil's recent problems that may have arisen because Brazil adopt- ed a flawed variant of regulation by contract. The paper concludes with a discussion of some lessons that can be learned from the experience of sev- eral countries. ACKNOWLEDGEMENTS This paper could not have been written without the assistance of many friends and colleagues inside and outside the World Bank. These individuals helped to develop an understanding of power sector and regula- tory reforms in developing and developed countries. They gave freely of their time and insights to make sure that we got the facts right and understood what the facts really meant. They include: Manish Agarwal, Beatriz Arizu, Ian Alexander, Pedro Antmann, José-Maria Bakovic, John Besant-Jones, Laurent Durix, Phillip Gray, Mohinder Gulati, Amit Kapur, Chris Shugart, Alan Townsend, Luiz Maurer and Joseph Wright. Of course, none of them should be held responsible for any errors of fact or interpretation that remain despite their best efforts to educate us. 6 Anne Vroom provided cheerful and timely assistance in the production of the paper while also providing mini- tutorials on the intricacies of MS Word. Chris Marquardt was an excellent editor who constantly reminded us that our goal was to communicate information and ideas to a wide audience and not just to write another report. Finally, we wish to acknowledge the financial assistance of the World Bank's Energy and Water Department in preparing this report. We owe a special debt of grati- tude to the department's director, Jamal Saghir. It was Jamal's suggestion that we look at this topic. We naively said that we would get back to him in a week or two with a short note. When it became clear that the topic merited more than a few pages, Jamal encouraged us to wrestle with the issues and seek out real-world exam- ples even if it meant that the one-week project eventual- ly became a multi-month effort. ACRONYMS AND ABBREVIATIONS ADR Alternative Dispute Resolution disco distribution company FERC Federal Energy Regulatory Commission genco generation company ICC International Chamber of Commerce LCIA London Court of Arbitration MIGA Multilateral Investment Guarantee Agency O&M operations and maintenance PPA power purchase agreement PRG partial risk guarantee SOE state-owned [power] enterprise transco transmission company UNCITRAL U.N. Commission on International Trade Law EXECUTIVE SUMMARY The last few years have not been kind to investors in However, the expected benefit of independent regulato- recently privatized electricity distribution companies in ry commissions following general tariff principles--a developing countries. Governments and consumers commercially viable power sector that benefits both have expressed disappointment with the companies' consumers and investors--has not been realized. The performance, as well as with the new independent regu- reality is that raising tariffs for retail customers is a polit- latory commissions established to regulate the compa- ically charged exercise in the Bank's client countries-- nies. Investors have also been disappointed. They argue usually low-income countries whose existing electricity that the commissions have not lived up to their commit- tariffs often fall far short of covering costs. It has been ments and almost always side with consumer interests. almost impossible for new regulatory systems to operate as planned in the first years of their existence. Following several widely publicized setbacks, many investors are questioning whether private investment in Regulation by Contract electricity distribution is viable in most developing coun- tries. Meanwhile, many consumers are now actively Some observers have recommended regulation by con- opposed to distribution privatization. In fact, the opposi- tract as an alternative to regulatory independence. The tion is so strong that some governments are fearful of essence of regulation by contract is pre-specification, in 7 even using the word privatization. one or more formal or explicit agreements, of the for- mulas that determine prices that a distribution company In light of these events, it is reasonable to ask: is allowed to charge for the electricity it sells. This does What has gone wrong? not mean that the actual prices are pre-specified. What is pre-specified is the regulatory treatment (such as The key lesson of the last 10 years is that regulatory indexing, automatic pass-through or case-by-case independence, by itself, creates neither regulatory com- determination) for the individual cost elements that mitment nor balanced decision-making. The principal together determine the retail tariff. lesson learned is that independence is not enough. Regulatory independence must be combined with a The application of this general concept has yielded dif- clearly specified regulatory contract that must be nego- ferent operational definitions. One definition is "regula- tiated by political authorities. In effect, the authors rec- tion without a regulator." A second definition, and one ommend combining two distinct regulatory traditions-- that is used in this paper, is "a detailed tariff-setting the Anglo-American tradition of independent regulatory agreement administered by an independent regulator." entities and the French tradition of a well-specified regu- Under this definition, the regulatory contract does not latory contract or concession agreement. replace the regulator but substantially limits the regu- lator's discretion. In particular, it forces the regulator to This paper goes beyond general principles. It provides a set tariffs based on specific formulas rather than just wealth of details on how specific features of regulatory general principles. contracts have been implemented in various developing countries and the lessons that can be drawn from this Some private investors have argued that there already experience. Particular emphasis is placed on how cer- exists a working model in the power sector for regula- tain key design elements--pass-through of power pur- tion by contract: the numerous power purchase agree- chase costs, foreign exchange fluctuations, technical ments (PPAs) that currently exist. In fact, regulatory con- and commercial loss reductions and obligation to tracts in distribution are more difficult to negotiate and serve--are dealt with in regulatory contracts. sustain than PPAs because of the large number of cus- tomers, the high visibility of the retail price and the Regulatory Independence need for ongoing investments. Nonetheless, the success of some Latin American countries suggests that regula- The World Bank has been encouraging its client coun- tory contracts for electricity distribution can and should tries to create independent regulatory commissions since be negotiated. the issuance of its power sector policy paper in 1993. To be sustainable, regulation by contract must achieve a form of "regulation by contract" for potential private two goals: it must (1) protect consumers from monopoly distribution companies that would be more akin to what prices and inferior quality of service while also (2) exists in Latin America and elsewhere. As in Latin attracting investors who will make the investments to America, the key elements of the proposed system provide the service at affordable prices. The objective is would be (1) automatic pass-through of cost elements to have the best of both worlds and to define the trade- that are largely beyond the distribution entity's control offs between these two conflicting regulatory objectives. (such as power purchases and taxes) and (2) indexing and The idea is to limit the discretion of the regulator in areas efficiency targets for cost elements that can be controlled that are known to deter investment while at the same (such as losses and labor costs). time using independent regulation to avoid uncertainties for investors created by political micro-management Unfortunately, under most current Indian proposals, and changes of government or governmental policy. multi-year tariffs would be permitted rather than required. However, this would still leave considerable The key component of the regulatory contract is a per- uncertainty compared to other countries. The better formance-based, multi-year tariff-setting system. The solution for India would probably be to amend the exist- concept of independence does not logically require ing state electricity reform acts to mandate the use of that a regulatory commission design the tariff system multi-year tariffs--or, even better, to (1) transfer tariff- 8 that it implements. In many Latin American countries, setting authority back to the government on a one-time independent or quasi-independent regulatory commis- basis for the initial post-privatization period, (2) incorpo- sions have been administering tariff-setting systems that rate the tariff-setting formula directly into the privatiza- were established by governments before the commis- tion agreement (which is the norm in almost every other sions came into existence. country that has successfully privatized distribution) and (3) establish, via amendments to the existing state elec- Real-World Regulatory Experiences: tricity laws, fairly detailed tariff principles and processes Brazil and India that would apply to subsequent multi-year tariff periods. Without such changes, any privatization will take place The most prominent example of regulation by contract under a cloud of legal uncertainty. in developing countries can be found in the more than 60 distribution privatizations that have occurred The Details of the Regulatory Contract, or Who throughout Latin America over the last 15 years. But Bears What Risk? whereas Latin American countries have generally had success with regulation by contract, Brazil has encoun- Many of the disagreements in designing a regulatory tered major regulatory and economic problems with its contract involve disagreements over whether the com- particular approach. The Brazilian experience provides pany, its customers or the government should bear a some lessons on "what not to do" with regulatory con- particular risk. From a potential investor's perspective, tracts: the allocation of risk in the regulatory contract will ultimately affect one of three things: the prices that it 1. Vagueness in tariff-setting provisions can charge, the costs that it can recover and the 2. Uncertainty about pass-through for power-purchase quantity of electricity that it can sell. costs 3. Low allowed prices for pass-through of power-pur- It is generally agreed that the best principle for risk allo- chase costs cation is that a particular risk should be borne by the 4. Foreign exchange risk party that can mitigate or manage the risk at the lowest 5. Uncertainty in the legal framework cost. But although the principle is easy to state, there is 6. Lack of respect for contracts often considerable disagreement over how it should be applied in particular situations. This can be seen in a The Indian regulatory system has suffered from different detailed analysis of four such key risks: flaws. It is based on an annual cost-of-service approach that gives considerable discretion to the regulators. It · Pass-through of power-purchase costs. Partial or has been recommended that Indian regulators move to delayed pass-through of power-purchase costs could bankrupt a distribution company because these costs exchange-rate risk. If the local currency loses value usually constitute about 50 to 80 percent of its total relative to the hard currency, the government allows costs. For this reason most private investors seek total the disco to increase its tariffs by the amount neces- and automatic pass-through of all power-purchase sary to cover the costs incurred in hard currencies. In costs, arguing that such costs are largely beyond their theory, this transfers the risk to the distribution com- control. In contrast, regulators generally fear that pany's consumers. Whether this happens in practice automatic pass-through will lead to corruption and depends critically on the extent of indexing and the inefficiency and, therefore, want the company to bear frequency of adjustments. some risk of non-recovery through a benchmark or some other regulatory mechanism. Any regulatory · Obligation to supply. Obligation to serve goes by mechanism designed to encourage economic pur- different names, including supply obligation and chasing--such as a cap on the prices paid for power public service obligation. The definition of obligation purchases--will inevitably affect the incentives to build to serve cannot and should not be the same across all new generating capacity. Generally, a multi-market countries, and a system's starting conditions must be benchmark (Columbia and the Netherlands) seems considered in defining an appropriate obligation (these preferable to a single, spot-market benchmark. include geographic scope, quality of service, and whether the obligation is absolute or limited). The obli- · Loss-reduction targets. The quantity of power pur- gation to serve has often failed in state-owned systems 9 chases that the regulator will allow the disco to recov- because of lack of money, ongoing political interfer- er in tariffs depends largely on the level of technical ence in operating and investment decisions, and, per- and commercial losses on the disco system that is haps most important, lack of performance-based salary deemed to be acceptable. In India, recent estimates incentives. In designing the regulatory contract for the of overall losses for some of the existing state-owned new private owner, key questions to ask include: Who distribution systems are as high as 50 percent. In must be served? What are the initial and phased in dealing with losses, the two key design questions in a technical and commercial standards for service? What regulatory contract are: What should be the initial are the penalties if the company fails to meet these accepted level of overall losses for tariff-setting pur- standards? Are excuses allowed? poses? And: How quickly can losses be reduced? The answers to these two questions determine how the Dealing with Disputes cost of losses is allocated between the company and its customers. Privatized distribution companies in sev- A distribution utility can be involved in many disputes. eral Latin American countries have been very success- The three principal types of disputes are those between ful in reducing losses. This has largely been due to a the distribution company and its customers, between the high degree of control over the labor force and sup- distribution company and other industry participants, port from government and local police--conditions and between the distribution company and its regulator. that might not exist to the same extent in India, Africa This paper focuses on the last type--disputes between or elsewhere. the distribution and the regulator over either the sub- stance of the regulator's decisions or the process by · Foreign exchange fluctuations. Distribution compa- which the regulator reached these decisions. The various nies receive payments from their customers in local approaches to resolving disputes include the following: currency but often incur costs in hard currencies. This leads to two major risks for private investors. The first · The local court system. Local courts are generally not risk--convertibility risk--is that the government will viable for dealing with regulatory disputes because not give the distribution company access to sufficient they are slow, lack the requisite knowledge, and are foreign exchange to pay for costs incurred in hard sometimes corrupt. currencies. The second risk--exchange rate risk--is that the local currency will lose value relative to hard · International arbitration. This is a necessary and currencies. Local-currency revenues may no longer be appropriate backstop for regulatory disputes in coun- sufficient to cover foreign currency costs. Indexing is tries with no track record for impartial resolution of the most common and transparent way to deal with such disputes. But it is best held in reserve as a last resort for dealing with disputes. Its principal value is Concluding Observations derived from the simple fact that it exists, even if it is never used. Ten key lessons have been learned from the experience of developing and developed countries with regulation · Mediation. Alternative Dispute Resolution (ADR), of by contract: which mediation is the most common form, typically involves the facilitation of structured efforts (e.g., 1. Independence is not enough. Ten years of experi- expert panels and mediation) by the parties to settle ence indicate that true regulatory independence dispute for themselves without going to a local often falls victim to direct or indirect political pres- court. However, no binding resolution can occur sures to avoid actions that a government thinks will unless and until an agreement is reached and com- be politically damaging. Even where regulatory inde- mitted to writing. In general, it does not work for pendence has been achieved, the regulators in devel- regulatory disputes because regulators have little or oping countries will find it difficult to make balanced no incentive to enter into mediation. decisions because starting conditions are bad, transi- tions take longer than expected, the government does · Expert panels. To adopt expert panels for regulato- not pay its bills or enforce law and order, and foreign ry contracts, the distributor must have the unilateral ownership is viewed as a new form of colonialism. 10 right to convene the panel, and there must be an The single most important lesson is that independ- effective mechanism for enforcing the experts' deci- ence must be "backstopped" by a regulatory contract sion. One promising hybrid technique is to create a that goes beyond general principles. And the key ele- standing expert panel that can act as both an expert ment of the regulatory contract should be a well- fact-finding panel and arbitration panel. specified, multi-year tariff-setting system that is required by law and specified in concessions, licenses · A specialized appeals tribunal. To date, the real- and other regulatory instruments. world experience with special appellate tribunals in most countries has generally been positive. They 2. The regulatory contract must be a political contract. tend to produce quick, well-informed, inexpensive Commitment will not be believable unless it comes decisions, and can be created without having to from the country's highest political authorities. reform the existing court system. However, once the regulatory contract is in place (i.e., the political deal has been made), the contract To enforce decisions, an existing World Bank financ- is best administered by an independent regulator. ing instrument known as a partial risk guarantee The underlying principles and initial parameters of (PRG) may prove useful. Other guarantee instruments the regulatory contract should be clearly specified in have been offered by the World Bank's Multilateral the country's primary or secondary electricity laws Investment Guarantee Agency (MIGA) to provide (Argentina, Bolivia, Chile, Peru). A regulatory con- insurance against currency inconvertibility, breach or tract is less likely to survive if it is poorly specified repudiation of contract and expropriation. It has been (Brazil) or exists only within a stand-alone concession recommended that a new type of guarantee be creat- or license agreement (Brazil). ed to backstop the operation of regulatory systems. Specifically, the proposed PRG would guarantee 3. "Regulation by contract versus regulation by com- scheduled payments of principal and interest pay- mission" is a false dichotomy. The real choice is ments on debt if the private investor defaults on or between an independent regulator with substantial delays payments because the regulator fails to honor discretion and an independent regulator with little the terms of the regulatory contract. However, the via- discretion, especially in the first post-privatization tar- bility of such a PRG will depend critically on the clarity iff period. An independent regulator should set and comprehensiveness of the regulatory contract. prices under a well-specified, pre-determined, multi- year tariff-setting system, and recommendations should be transparent. 4. Regulation by contract is a new name for an old 10. Regulation by contract should be reserved for pri- paradigm. Regulation by contract--an independent vate distribution companies. In most developing or quasi-independent regulator administering a well- countries, state-owned enterprises do not respond to specified tariff-setting system that is embedded in normal commercial incentives. Thus there is little to laws, concessions and regulations--has been the be gained in creating a new independent regulator norm for distribution privatizations throughout Latin to regulate a state-owned power enterprise. America for more than 15 years. APPENDIXES 5. Electricity consumers cannot be the forgotten third party to a regulatory contract. The government and Appendixes A, B and C contain text for a proposed regulator must ensure a fair balance between the statute in a South Asian country that contains some ele- interests of consumers and those of investors. If con- ments--governing controllable and non-controllable sumers fail to see any obvious benefits from the reg- costs, obligation to serve, and quality of service--of the ulatory contract (i.e., "early wins"), it will be political- legal foundation of a regulatory contract. ly unsustainable. 6. Investors must have confidence that the contract will be enforced fairly and efficiently. It is preferable 11 to combine the "backstop" of international arbitra- tion with less costly forms of dispute resolution that can prevent one or more smaller disputes from exploding into a big dispute. 7. The heart of a regulatory contract is a pre-speci- fied, performance based, multi-year tariff-setting system. This should include benchmarks or targets for controllable costs and automatic pass-through for non-controllable costs. 8. A regulatory contract is sustainable only if the underlying economics are viable. Regulation by contract will not work if there is a large gap between costs and revenues. The gap must be closed by low- ering costs, increasing revenues, or both. Investors must be protected, and the regulatory contract might need to be combined with transition-period subsidies (with help from World Bank loans or guarantees). 9. A multi-year tariff system can be put into operation even in the absence of high-quality data. Data quality will improve through privatization, especially if "better data" is specified as a performance ele- ment in the regulatory contract. If there is a political concern that investors will be able to earn high prof- its because of poor data quality, then the tariff sys- tem should include a within­tariff-period "sharing" mechanism for profits or revenues between the distri- bution companies and its customers. Regulation by government-mandated rationing program. The company protested that the rationing program put it in the difficult Contract: position of having to tell its customers: "We are asking you not to buy the only thing that we have to sell."2 A New Way to Privatize With these widely publicized setbacks, many investors Electricity Distribution? are now questioning whether private investment in elec- tricity distribution is viable in most developing countries. And many consumers are actively opposed to distribu- 6. "Governments throughout the world engage in tion privatization. In fact, the opposition is so strong that three main activities: they tax, they spend and they some governments are fearful of even using the word regulate. Regulation is the least understood...." privatization.3 In light of these events, it is reasonable to ask: What has gone wrong? 1. INTRODUCTION B. The Wrong Advice? A. Recent Events Investors frequently maintain that these failures stem 12 from flaws in many of the more than 50 new electricity The last few years have not favored private investors in regulatory systems recently established in developing electricity distribution companies in developing coun- countries.4 There has been too much emphasis, they tries. argue, on creating "independent regulatory commis- sions" as the key to successful distribution privatization. In India, the AES Corporation told the Orissa state gov- They contend that this is a naïve recommendation for ernment that it wished to sell its ownership interest in a most developing countries. At a recent privatization con- local distribution company because it saw no way to ference in India, for example, one Indian investor make the company a viable commercial enterprise. A observed with exasperation that "regulatory independ- few months later, BSES, a large Indian power company ence is a fine concept, but it is of little comfort when I that had invested in three other distribution companies don't have enough money to pay my employees and in Orissa, also threatened to leave. In Ecuador, the gov- creditors. I need money, not mantras."5 Other investors ernment announced that it was abandoning its plans to have argued that the danger of an independent regula- privatize 17 state electricity distribution companies after tory commission is that it can easily turn into a "rogue receiving a poll that showed that more than 71 percent regulatory commission" especially if the commission of the general public was opposed to such privatiza- operates under a law that specifies only general tariff- tions. In Senegal, a new government terminated its setting principles. They also contend that a new inde- agreement with Senelec, a consortium of pendent regulatory commission guided only by general Tractebel/Hydro Quebec, after accusing the consortium tariff-setting principles will almost always favor the inter- of failing to improve the frequency and duration of ests of consumers over those of investors because it will blackouts. In Brazil, AES experienced major financial have too much discretion. In their view, the better alter- problems for Electropaulo, its distribution company in native is "regulation through contract" rather than Sao Paulo--problems caused in part by a significant "independent regulatory commissions."6 drop in sales and revenues following in the wake of a Scott Jacobs, Building Regulatory Institutions: The Search for Legitimacy and Efficiency (OECD, Center for Cooperation with Economies in 1 Transition, Paris, 1994). "Power Policemen: Electricity Rationing Roils Brazil , Leaving U.S. Utility in a Spot," Wall Street Journal, July 30, 2001, p. 1. Other private 2 distribution companies have also had problems in Brazil. In August 2002, CEMAR, a subsidiary of the U.S. company PPL, declared bank- ruptcy. In addition, there have been reports of several multinational energy companies wanting to leave. See "8 Utilities In Brazil Could Go On The Block," New York Times, August 30, 2002, Section V, p. 1. Governments have become adept at creating euphemisms for privatization such as "disinvestment" (India), "ownership reform" (China), 3 "peopleization" (Sri Lanka) and "equitization" (Vietnam), and "disincorporatization" (Mexico). See Nellis (2002). In a recent survey of private investors, about 50 percent of the respondents said that their worst power sector investment experience was 4 the failure of regulators to respect regulatory commitments. See Lamech and Saeed (2003). Here and in several other places in this paper, we quote from individuals who provided comments "off the record." 5 Similar views can be found in the academic literature. See Levy and Spiller (1994), whose principal point is that a regulatory system must 6 be compatible with a country's executive, legislative and institutional endowments. In practice, this means that many developing countries will need, at least initially, a well-specified and relatively inflexible regulatory contract. Regulation by contract is an appealing concept because 6. Section 6 discusses alternative approaches to deal- it seems to hold the promise of a regulatory system that ing with disputes that arise in the application of reg- operates on "autopilot." Private power investors like the ulatory contracts. idea because it looks familiar. It bears a strong resem- blance to PPAs (power purchase agreements)--direct 6. Section 7 suggests lessons learned. contracts between government enterprises and private generators for long-term bulk power sales--that are usu- Although the focus of this paper is on "good" and ally directly negotiated and implemented between the pri- "bad" features of real-world regulatory systems, we do vate investors and government. A typical PPA contains not believe that flawed regulation is the principal expla- detailed formulas that specify the prices generators will nation for the disappointing performance of private be allowed to charge over the life of the contract. companies in electricity distribution. Companies often Regulators usually have little or nothing to do with either use regulation as a convenient scapegoat to deflect the design or the implementation of the PPAs. Not surpris- attention away from their own strategic mistakes. For ingly, some private investors argue that if PPAs have been so example, in Brazil and certain other countries it is now successful in promoting private investment in generation, why reasonably clear that some private companies paid too can't the same regulatory technique be applied to distribution? much for their investments and allowed themselves to become too highly leveraged. Although regulatory sys- There is always a danger in replacing one mantra (inde- tems can and should be improved, such improvements 13 pendence) with another (regulation by contract). In this paper, will not save investors from their own mistakes. we take a close look at whether "regulation by contract" is a viable regulatory approach for developing countries, either by itself or in combination with regulatory independence; and, if 2. REGULATORY INDEPENDENCE so, how it could be used to create new private distribution entities. The paper is organized as follows: A. Why Encourage Regulatory Independence? 6. Section 2 examines why regulatory independence The World Bank has been encouraging its client coun- was recommended by the World Bank and others. It tries to create independent regulatory commissions also explores why regulatory independence, when since the issuance of its power sector policy paper in combined with nothing more than general tariff prin- 1993.7 The reason for this was fairly straightforward: it ciples, seems to have failed in developing countries. was largely a reaction to the failures of a system of non-independent regulation that was the historic norm 6. Section 3 describes the key characteristics of regula- for most state-owned power enterprises (SOEs). Under tion by contract for electricity distribution companies this system, one or more government ministers would as it has developed in several countries. It also dis- set both the level and structure of tariffs for SOEs. The cusses how regulation by contract differs from nor- final decision--usually made with the formal or infor- mal commercial contracts and considers whether mal approval of the country's president or prime minis- the use of regulation by contract should be limited ter--was often not published and rarely explained. Not to just an initial, post-privatization transition period. surprisingly, this ministerial model of regulation usually produced tariffs that failed to cover costs because there 6. Section 4 compares the recent real-world regulatory were few, if any, political benefits to be gained by rais- experiences of selected distribution entities in Latin ing electricity prices. America and India. It also takes a close look at recent problems in Brazil that may stem from a When the Bank began to encourage privatization in the flawed variant of regulation by contract. early 1990s, there was a consensus that private investors would not show up unless the system for set- 6. Section 5 examines the details of actual regulatory ting tariffs--the core regulatory task--was "de-politi- contracts in several countries, especially with respect cized" and "made independent." (These were to the sharing of key risks such as pass-through of euphemisms for a tariff-setting system that would bal- power-purchase costs, foreign exchange fluctua- ance consumer and investor interests.) Therefore, the tions, loss reduction and obligation to serve. Bank encouraged policy makers to create regulatory World Bank (1993). 7 entities whose decisions would be both transparent and basic problem seems to be a "weak governance envi- independent of government political authorities. The for- ronment."10 This, in turn, has meant that new commis- mer meant that the new regulatory commission would sions have often failed to achieve independent and publicly explain the reasons for its decisions and the latter technical decision-making. Although new regulatory meant that the commission would make tariff decisions institutions have been created, it appears that in some under specified legal standards without getting the prior countries "institutional change... changed nothing" or approval of a minister or prime minister.8 It was hoped at least very little.11 There seem to be at least three rea- that this new tariff-setting system would produce better sons for this failure: results because its decisions would be more technical than political. But it was also recognized that the new sys- 1. Many commissions never really became independ- tem would not work unless political authorities were will- ent. They may have looked independent on paper ing to give up their existing control over electricity tariffs (i.e., had legally independent decision-making (i.e., give up political power to get electrical power). authority), but in reality many commissions have con- tinued to operate as if they were still branches of one To be fair, the Bank's recommendation of regulatory or more ministries. The behavior of the commissioners independence was not categorical. A close reading of may have been influenced by the hope of a future the 1993 power-sector policy statement shows that the higher-level job elsewhere in government (with the 14 Bank recommended "independence" only if the enter- consequent need not to antagonize a minister when he prise was first corporatized (allowed to make operating called to discuss a pending tariff request), fear of los- and investment decisions independent of political ing their present jobs, continued government control authorities) and commercialized (allowed to charge over their budget, or a continuing antagonism to pri- tariffs that recovered operating and capital costs). vate ownership in a sector that had previously been Although very few state power enterprises were ever suc- largely public. Although one could say that the new cessfully corporatized and commercialized, it appears commissions were "captured" by their governments, it that the regulatory recommendation was widely adopt- is probably more accurate to say that the commission- ed. Fortunately, in actual operational work, Bank task ers never really tried to become independent. managers rarely proposed independent regulatory com- missions as a "stand-alone" option. More frequently, 2. Some commissions were granted only limited legal they recommended an independent regulatory commis- "independence." In Hungary, Latvia and Lithuania, sion combined with a pre-specified tariff-setting formula for example, the tariff-setting authority of the regulato- (a type of regulation by contract).9 ry entity is formally limited to giving advice to a minis- ter. In such a situation the regulator has, in effect, the B. What Went Wrong? legal right to give his opinion about tariffs but not much else. Elsewhere, regulators thought that they With the benefit of close to 10 years of experience, we had independent and exclusive decision-making find that the expected benefit of independent regulato- authority over tariff setting, but then discovered that ry commissions following general tariff principles--a the legislature, cabinet or president disagreed or did commercially viable power sector that benefits both not care what the law said. In Georgia, after the regu- consumers and investors--has not been realized. The latory commission issued a controversial tariff order, it Although the Bank has tended to emphasize decision-making independence from government, full independence would also include 8 independence from the regulated companies that are regulated and the customers that they serve. See Smith (1997). This has been the Bank's recommended approach in Bolivia, Georgia, Moldova, Peru, Uganda and, more recently, India. 9 This is discussed in Levy and Spiller (1994). 10 Putman (1993). 11 Reuters Business Briefings, "Parliamentary Commission Criticizing Decision to Increase Electricity Tariff in Tbilisi," November 15, 2001. 12 Reuters News, "Pakistan Cuts Electricity Rates, Ignores Regulator, Sends Bad Signal To Investors," December 11, 2002. 13 La Paz, "Tariffs Are Frozen Until January," La Paz Bolivia, November 1, 2001. In February 2002, the presidential decree was found to be 14 unconstitutional by the Bolivian Constitutional Tribunal. The tribunal, which consists of six members selected by the Bolivian congress, is responsible for hearing formal complaints that a Bolivian government official or agency has overstepped its legal authority. Some observers have described the tribunal's decision as a landmark decision in Bolivia's efforts to create a credible "rule of law." This would also appear to be true in Chile and Uruguay, where the legal norm is that "the commission proposes but the minister approves." 15 But knowledgeable practitioners have observed that in these two countries it is politically difficult for a minister to overturn a commission deci- sion if it is well-reasoned, well-documented, and well-publicized. Of course, there is nothing to prevent a minister from calling the head of the commission and indicate a strong preference for a technical study that produces lower rather higher numbers. This was particularly easy to do in Chile, where the head of the commission was also the Minister of Economy. was informed that the Parliament believed it could stop In 1993 the Bank was probably naïve in believing that the order from going into effect.12 In Pakistan, the cab- it would be possible to create a regulatory system that inet of a new civilian government recently ordered a 5 would be fair to both investors and consumers by creat- percent tariff cut even though the authority to set tariffs ing a new regulatory institution with nothing more than had been transferred by law to NEPRA, the national elec- the formal elements of independence.17 This "techno- tricity regulator.13 In Bolivia, the president issued a decree cratic and legal" model of regulation ignores the fun- freezing electricity tariffs after the regulator issued a damental reality that raising tariffs for retail customers is notice that would have raised tariffs.14 In all of these situ- a politically charged exercise in the Bank's client coun- ations, the commission may have been legally independ- tries--usually low-income countries whose existing elec- ent but discovered that it was functionally irrelevant.15 tricity tariffs often fall far short of covering costs. 3. Some commissions, faced with a legacy of tariffs In such countries, almost everyone views an independent that fell far short of covering costs, were under- regulatory system with fear and trepidation. Ministers standably reluctant to take the politically explosive fear that a genuinely independent, regulatory commis- step of a raising tariffs to cost-recovering levels. sion may raise tariffs too quickly--possibly leading to Many developing or former socialist countries start political attacks from the opposition, riots and even the their reforms with a large gap between revenues and collapse of the government.18 Investors fear the opposite. costs. When the gap is large, it is totally unrealistic to They are fearful that a new regulatory commission will 15 expect that a new regulatory commission will be able not have the courage to raise tariffs quickly enough so to close the gap under the guise of making technical that they can recover their costs and earn a profit. They tariff adjustments, especially when political authorities are also afraid that, once they make their investments, have been hiding from the gap for many years. they run the risk of being held hostage to a form of de When faced with such a difficult and politically sensi- facto expropriation through unfavorable regulatory deci- tive situation, a regulatory commission can almost sions. And finally the new regulators themselves are also always find creative but plausible reasons to rational- afraid. They are fearful that they will be blamed for the ize small or no increases in tariffs. One common jus- regulatory equivalent of the "perfect storm"--significant tification is that there are "insufficient data" to justify increases in retail tariffs, no obvious improvements in higher tariffs. Another is that the company could quality of service and highly visible profits for a new pri- make a reasonable profit if it just met efficiency tar- vate company that is largely owned by foreigners. All of gets set by the regulator (usually ignoring, as in these conflicting fears make it almost impossible for a Orissa, that the targets have been set at impossibly new regulatory system to be a technical and legal exer- high levels). To be fair to the regulators, many private cise in the first years of its existence. companies have shown disappointing early progress in improving performance, though this sometimes can be explained by a government's unwillingness or inability to provide basic "law and order." For exam- ple, it appears that private companies in Georgia, India and South Africa often have no support and sometimes active opposition from the local police when they try to collect non-paying customers or ille- gally connected customers.16 So even if the private distribution company has the approval of the regula- tor to charge tariffs that recover costs, this does not do much good if it cannot collect what it bills. Letter from AES Orissa Distribution Private Limited to Chief Secretary, Government of Orissa, July 25, 2001 and Washington Post, "For 16 South Africa's Poor: A New Power Struggle," November 6, 2001 p. A-1. The formal elements of independence include: clear mandate excluding ministerial discretion established in law; appointed on the basis 17 of professional criteria with restrictions on conflict of interest; protected from arbitrary removal during fixed terms; staggered terms that do not coincide with the election cycle; and earmarked funding. See Smith (1997). "Four Die As Andhra Police Open Fire on Protesters,"The Times of India News Service, August 29, 2000; and "Indian Protesters Burn World 18 Bank Chief In Effigy," Dow Jones Newswire, November 11, 2000. However, government ministers sometimes want higher tariffs. It was recently reported that several ministers in Brazil and Colombia were unhappy with their electricity regulators for keeping retail tariffs too low. The ministers believed that low tariffs would hurt the country's ability to attract foreign investment. They accused their regulators of failing to see the "big picture." 3. REGULATION BY CONTRACT latory contract does not replace the regulator but substantially limits the regulator's discretion. In par- 6. In framing a government to be administered by men ticular, it forces the regulator to set tariffs based on spe- over men, the great difficulty lies in this: you must cific formulas rather than just general principles. first enable the government to control the governed; and in the next place oblige it to control itself. We use the second definition in this paper because we 6. --James Madison, 1787 are not aware of any developing countries where the behavior of a privatized electricity distribution company 6. A good regulator is a boring regulator. is completely controlled by a regulatory contract with no 6. --U.S. power company official, 2002 further intervention by a government or regulatory entity. "Regulation without a regulator" does exist in the A. What Is Regulation by Contract? French water sector but the conditions that make it pos- sible in France are not likely to be replicable in other 1. Two Definitions countries, especially in developing or former socialist countries. (See Box 1.) Some observers have recommended regulation by contract as an alternative to regulatory independence. There seem to be four principal reasons why regulatory The essence of regulation by contract is pre-specifica- contracts for electricity distribution companies require 16 tion, in one or more formal or explicit agreements, of continued administration by some government entity: the formulas that determine prices that a distribution company is allowed to charge for the electricity it sells. 1. Most countries find it politically impossible to allow This does not mean that the actual prices are pre-speci- a company to calculate periodic price adjustments fied. What is pre-specified is the regulatory treatment permitted by a tariff formula without further govern- (such as indexing, automatic pass-through or case-by- ment oversight. In any democratic country, a govern- case determination) for the individual cost elements that ment that accepted such an arrangement would be together determine the retail tariff. The agreements are strongly criticized for abdicating its responsibilities to between the government and the private company.19 protect captive customers.21 It would be especially dangerous if the company is foreign owned. The application of this general concept has yielded at least two different operational definitions. One defini- 2. The values of key parameters of the tariff-setting for- tion of regulation by contract is that it is regulation mula, such as loss targets and power purchase without a regulator.20 Under this definition, the regula- pass-through mechanisms, need to be reset every tory contract is totally self-contained and self-adminis- several years. In general, courts have neither the tered like a commercial contract. Any disputes arising technical expertise nor the inclination to do this. over implementation are handled by a regular court, an Therefore, the government must designate some administrative court or a special expert panel. This is an governmental entity, other than the courts, to admin- appealing concept because it seems to offer the possi- ister the contract and reset new values at the end of bility of putting regulation on autopilot and eliminates the tariff period. the need to create a new regulatory entity. 3. Even with detailed rules, certain regulatory tasks will A second definition is that regulation by contract is a still need to be performed. These include applying detailed tariff-setting agreement administered by an indexing formulas, monitoring the distribution com- independent regulator. Under this definition, the regu- pany's behavior with respect to pricing and quality The U.S. power sector has operated under an informal understanding known as the "regulatory compact." The essence of the compact 19 was that privately owned utilities would be allowed to recover "prudently incurred" operating and capital costs in return for assuming the obligation to meet all the electricity needs, upon demand, in their service areas. The compact was not written down in any one docu- ment. Instead, it was presumed to exist (at least by the private companies) in numerous regulatory practices, approvals, court cases and explicit or implicit understandings. In the late 1990s, a major controversy arose when the private companies argued that under the "regu- latory compact" they were entitled to recover stranded costs when the government decided to introduce retail competition. For argu- ments in favor of this position, see Baumol et al. (1996); for a contrary view, see Michaels (1996). The fact that the two sides could not even agree as to whether a "regulatory compact" even existed suggests that this is not a workable model for developing countries. In referring to long-term concession or leasing arrangements, the World Bank's 1994 World Development Report stated that "these 20 arrangements....do not require the establishment of independent regulatory bodies because regulatory procedures [and regulatory provi- sions in general] are specified in the underlying contract" (p.10). Alternatively, if there is a regulator, it is limited to functioning as a "con- tract monitoring office." Captive customers are those who lack access to alternative suppliers and cannot generate electricity at their own facilities at competitive prices. 21 BOX 1. The French Water Sector: Regulation Without a Regulator The French water sector has operated for many years under regulatory contracts without a regulator. Under this system, implemented through hundreds of municipally granted concessions or leases, private companies provide about 80 percent of the water supply and about 45 percent of the sewage services. However, a recent, in-depth study of the sector* suggests that the combination of conditions that make it possible to have regulation without a regulator in France are not likely to be replicable in electricity distribution in developing countries for the foreseeable future. These conditions include the following: -- Model contracts. The Ministry of Interior has developed model concessions and leases that are widely used by municipalities. Experienced central government officials are available to assist municipal officials in developing and implementing these contracts for local circumstances. -- Well-developed case law and legal doctrines. More than 200 years of highly developed case law and legal doctrines inform the implementation of concessions and leases. In particular, there are well-defined procedures and standards for dealing with three problem areas: actions taken by a public authority that raise the costs of a concessionaire (fait du prince), unexpected 17 material conditions that make construction or operations more costly (sujétions imprévues) and temporary difficulties typically brought on by the increase in the price of an input that clearly exceed any levels expected at the time the contract was signed (imprévision). -- A respected appellate tribunal. Disputes over implementation of the concessions and leases are handled by an experienced and knowledgeable appellate court known as the Conseil d'Etat, which functions as a special administrative tribunal. Unlike its Anglo-Saxon counterparts, the Conseil d'Etat has the authority to make decisions on both substance and process. In this sense, it acts as a "shadow regulator." -- Private operators are French companies. The sector is dominated by two large French com- panies. They are sensitive to "reputational concerns." Moreover, the fact that they are well-known French companies also eliminates any sensitivity about payments to foreign companies. -- Common educational backgrounds. Many of the executives in the private companies and the high level civil servants in the relevant ministries attended the same elite educational institutions (grandes écoles). This produces a sense of "professional honor" and a commitment to the success- ful delegation of a public service. -- Cultural traditions. The principle of moderation is well developed in French society, and exces- sive argumentativeness is considered "vulgar." * Shugart (1998). of service, and making decisions about noncompli- will trigger the need to decide on the price and non- ance and possible penalties. price terms of unbundled distribution and transmis- sion access, obligations for back-up supply, and 4. Even if distribution tariff setting is susceptible to assessments of the degree of competitiveness in detailed specification, there are many other regula- newly opened markets (i.e., market monitoring). tory decisions that simply cannot be specified in Because a regulator is likely to be needed for these advance. This is especially true if the power sector other decisions, it is probably more efficient for the reform includes the phase-in of one or more forms same entity also to administer the regulatory con- of wholesale and retail competition which, in turn, tract for distribution tariff setting. In light of these considerations, it is reasonable to · In the first multi-year tariff-setting period, the regula- expect that some government entity will need to admin- tor will administer the terms of a tariff formula over ister the regulatory contract. which it has little or no discretion. In later tariff-set- ting periods, the regulator will have more discretion 2. Key Characteristics over selecting the values of some of the parameters in the formula. Experience in Latin America and elsewhere suggests that the terms regulation by contract and regulatory con- · The contract will usually specify three types of regu- tract be defined as one or more written agreements latory actions: resetting of the parameters of the tar- between a private distribution company and a govern- iff formula at the end of the multi-year period; peri- ment that have the following features:22 odic adjustments for inflation or "true-up" for the difference between actual and projected values; and · The government pre-commits to a specified regulatory extraordinary adjustments for unexpected or difficult system that establishes how retail tariffs will be set for to predict events that could have a significant effect a multi-year period typically ranging from four to eight on the financial condition of the distribution company. years. In most instances, it is the government rather than the regulator that designs the regulatory system. · The agreement will usually build on a pre-privatiza- tion "clean-up" of the public enterprise's balance 18 · The regulatory contract is specified prior to receiving sheet. privatization bids so that bidders can estimate their likely future stream of revenues. · The agreement may be a stand-alone document such as a license or concession. But more typically it · Even though the government usually designs the tar- will be embedded within or derived from a privatiza- iff system and makes the formal commitment, a sep- tion agreement, secondary regulations or decrees or arate and possibly independent regulatory entity will the power sector reform law. In these latter cases, implement the contract. regulation by contract is not a single agreement but rather a combination of concessions, licenses, · The agreement contains a formula with pre-specified decrees and laws. The regulatory system will not be parameters that determine how annual total revenue or sustainable unless the documents are consistent. average tariff levels will be established by the regulator. The formula will often distinguish between controllable · The agreement may specify that disputes between the and non-controllable input costs. Controllable costs will regulator and the company will be settled through be tied to external indices or benchmarks with perform- normal judicial channels such as the country's exist- ance targets and associated rewards and penalties. ing court system, through a specialized non-judicial Non-controllable costs will be automatically passed channel such as an arbitration panel (within country through on a regular or episodic basis.23 or external), or through a special electricity appellate court or an appellate body that hears appeals of the · The contract will usually specify some process for decisions of all infrastructure regulators. dealing with unforeseen events that can have a sig- nificant effect on the utility's costs or revenues. These · If the gap between revenues and costs is large, the might include damage from severe storms and government will have to commit to a subsidy pro- major changes in taxes, duties and environmental gram. The subsidies can take different forms: direct regulations. subsidies to the distribution company (such as a subsidized price for the power that the distribution · The regulatory contract may have an indefinite life company pays for bulk purchases), tariff subsidies to but with scheduled revisions every four to eight years. poor customers, and capital cost subsidies to con- nect new rural customers.24 Our focus in this paper is on regulation by contract for the principal distribution supplier. Equally important to power sector reform, but beyond 22 the scope of this paper, are the issues involved in regulating off-grid and mini-grid suppliers. The extent of the pass through for a non-controllable cost is usually based on an assessment of whether the price, quantity or both ele- 23 ments of the input cost are beyond the control of the distribution utility. (See Box 4.) Subsidies that are tied to output performance have been described as "output-based aid." For descriptions of how such programs could 24 be applied in electricity, see www.rru.worldbank.org. 3. What Government Entity Should Administer the the form of direct bilateral contracts between a private Contract? investor and a government entity. From the perspective of the private investor, the single most appealing feature In the preceding list of characteristics, it is assumed that of a PPA is that it specifies a formula that determines government will create the regulatory contract, but that how prices will be established over a 10-to-25 year implementation and enforcement of the contract will be period. Because the pricing formula is "locked in" for transferred to an independent regulatory entity. Why the life of the contract, there is no need to come back should there be this split of functions? to the regulator every year for a new tariff decision. In fact, the regulator may never have reviewed the PPA The initial decision to privatize and set up a viable regu- and may be legally precluded from modifying its terms latory system that will support privatization must be a once the contract is signed. government decision because it inherently a political act. When the initial gap between costs and revenues is Although investors almost always like PPAs because they large (as is often the case in developing countries), the reduce risk and help to ensure profitability, many gov- decision to privatize requires more political courage ernments have had second thoughts about the PPAs that than technical expertise. Therefore, it would be unrea- they or previous governments have signed. Their princi- sonable and naïve to try to delegate this fundamental pal criticism is that the prices are too high and too political decision to a group of technical experts resid- much risk has been transferred to final customers. They ing in a newly created regulatory commission. point out that many of the PPAs were negotiated without 19 any attempt at competitive procurement. In Guatemala, Once the political decision has been made, however, it the Dominican Republic, the Philippines and Pakistan, has a better chance of surviving if implementation can new governments (or opposition parties) have contend- be handed over to a regulatory entity that is insulated ed that the pricing formulas in existing PPAs may reflect from ongoing short-term political pressures. After the corruption or incompetence.25 It has also been pointed contract is in place, it is more likely to be administered out the structure of many PPAs, particularly the require- in a fair and impartial way by an independent regulator ment to take power even if means foregoing a lower than by a new minister cost supply source, who probably was not Table 1. Regulation by Contract: Generation Versus Distribution makes PPAs an a member of the CHARACTERISTIC GENERATION DISTRIBUTION impediment to creat- government that Parties to the contract Private investor as a seller and Private investor as a seller and ing organized bulk negotiated the contract. government as a buyer government as a regulator power markets.26 Number of customers One, usually a state-owned Thousands of customers enterprise 4. Power Purchase Apart from these Price visibility Low (at least initially) High Agreements: a criticisms, there is Investment One time Ongoing Useful Precedent? also the reality that Quality of service One-dimensional and easy Multi-dimensional and hard to to measure measure the basic character- Some private Scheduled resetting of Usually none Every four to seven years istics of distribution investors have tariff parameters make it harder to argued that there Regulators role in Limited to non-existent Limited to non-existent design and imple- designing contract already exists a work- ment a regulatory Economic problems Exposure to foreign Exposure to foreign exchange ing model in the exchange fluctuations fluctuations contract in distribu- Price may be too high because power sector for reg- of corruption or incompetence tion than in genera- ulation by contract. in procurement tion. (See Table 1 for In their view, the a listing of some key numerous power purchase agreements (PPAs) signed in differences between distribution and generation.) Three many developing countries represent a regulatory model characteristics in particular make regulatory contracts in that could be replicated in distribution. PPAs usually take distribution more difficult to negotiate and sustain than Similar allegations of corruption in the awarding of French water concessions led to the passage of an "Anti-Corruption Law" in 1994. The 25 law establishes a transparent multi-step process for selection of concessionaires. Its key requirements are "openness (the municipality must publish a notice of its intention to delegate a public service and then review the qualifications of the respondents in order to select a short- list), competition (presentation of fully developed proposals by more than one firm), and transparency (presentation of various reports to the municipal assembly explaining the conclusions reached)." See Shugart (1998), p. 104. The Shugart dissertation contains one of the best descriptions of the French concession system currently available in English. Woolf and Halpern (2001). 26 PPAs: the large number of customers, the high visibility of how convince potential investors that it will impose restric- the retail price and the need for ongoing investments. tions on the ability of future governments to exercise these Despite these characteristics, however, we think that regu- rights. In summary, the two distingushing elements of regu- latory contracts for electricity distribution can and should lation by contract are that a government pre-commits to a be negotiated. In Sections 5 and 6, we discuss some spe- fairly specific form of tariff setting and also agrees to limit its cific features of sustainable regulatory contracts. own ability to modify the tariff-setting system for a specified number of years. B. Is a Regulatory Contract Different from a Commercial Contract? Most investors would prefer that the regulatory contract be viewed as a simply a contract between them and the When most people think of contracts, they think of com- government. But this is a naïve view because regulation mercial contracts. In a normal commercial contract, one by contract does not take place in a vacuum. Although party is a seller and the other is a buyer. The seller pro- regulation by contract has developed primarily to attract vides a product or service in return for a payment from the investment, it will not be sustainable if it focuses exclu- buyer. A privatization agreement is like a traditional com- sively on this one goal. Regulation by contract must mercial contract. The government is selling some or all of achieve two goals: it must (1) protect consumers from the ownership rights to an existing distribution entity and a monopoly prices and inferior quality of service while private company is the buyer. also (2) attracting investors who will make the invest- 20 ments to provide the service at affordable prices. The A regulatory contract bears some resemblance to a com- objective of regulation by contract is to have the best of mercial contract in that it is based on a quid pro quo. The both worlds and to define the trade-offs between these essence of a regulatory contract is that the government two conflicting regulatory objectives. The idea is to limit agrees to commit itself (or a regulatory commission that it the discretion of the regulator in areas that are known will create) to implement a pre-specified tariff-setting system to deter investment while at the same time using inde- to provide a reasonably high degree of certainty as to the pendent regulation to avoid uncertainties for investor company's future revenues over a multi-year period. In created by political micro-management and changes of return, the private company agrees to provide electricity dis- government or governmental policy. tribution service with certain quality attributes to specified customers over a defined period of time. The rationale for C. Should Regulation by Contract Be Just a bringing the two contracts together--the privatization con- Transition Mechanism? tract and the regulatory contract--is that no serious and reputable private company would be willing to buy distribu- It has been suggested that regulation by contract is tion assets unless it knows the services that it will be obligat- needed only to launch the privatization process (i.e., the ed to provide and the prices that it will be allowed to charge. first, multi-year tariff period after privatization). In other words, it is a technique for "jump-starting" the regulato- But a regulatory contract is clearly different from a normal ry and privatization process. The rationale for limiting commercial contract in several important respects. First, one regulation by contract to just this transition period is of the parties to the contract, the government, is performing that this is the period when the contract is most needed: two roles. It is the seller or lessor of the assets, and it is also the new regulatory commission is inexperienced, will not the enforcer of the regulatory contract. Second, there are have a "track record" and is operating in a "weak gov- asymmetric rights in the contract. In normal commercial ernance environment." (The is a polite way of saying contracts, there is a balance of rights between the two par- that the commission may be formally independent but ties. But in a regulatory contract, the government usually neither the commission nor the government can be reserves some extra-contractual rights for itself. For exam- trusted with much discretion over tariffs.) Although these ple, it may reserve the right to an early termination, the right are plausible justifications for using regulation by con- to make unilateral amendments to the contract and the tract in developing countries, it is interesting that regu- right to prohibit early terminations of the contract by the pri- lation by contract has also become the regulatory sys- vate party. These are the traditional rights of a government tem of choice in many developed countries that are not when it is authorizes a private company to perform a public operating in weak governance environments. service. Nevertheless, if the regulatory contract is to be credible, the government that signs the contract must some- In many developed countries, multi-year price or revenue fied tariff-setting formula, with fixed parameters, that is caps, which are a form of regulation by contract, have used to set the distribution company's average retail price become the system of choice in setting retail tariffs both for or overall revenue over a multi-year period.28 Typically, the new regulatory commissions, such as exist in England and formula provides for automatic pass-through of non-con- Wales and the Netherlands, and old regulatory commis- trollable costs and benchmarking or indexing of control- sions, such as exist in the United States. It appears that lable costs. To provide additional comfort to investors, the these countries have decided to commit themselves to a tariff-setting formula is often written into a law, the con- multi-year (as opposed to an annual) tariff-setting system cession or privatization agreement, commission orders, or because they have concluded that a multi-year tariff system sometimes in all of these legal documents. Under this embedded within a formal regulatory contract provides standard Latin American approach, the regulator's initial stronger incentives for the regulated enterprise to be effi- role is quite limited. For the first tariff-setting period, usu- cient. In effect, they have decided to give up regulatory dis- ally four to eight years, the regulator essentially adminis- cretion because they expect that they will get more efficient ters a formula that was written by someone else. The reg- performance from the regulated entity if they commit to a ulator functions, in effect, like a referee at a soccer game: multi-year tariff regime. The U.S. example is particularly rather than writing the rules, he simply enforces them. interesting because none of the U.S. commissions that chose to adopt multi-year tariff-setting systems were legally B. The Special Case of Brazil obligated to do so. In general, most U.S. commissions have operated for many years under very general laws that simply Brazil's approach to regulation by contract is both similar 21 say that the commissions should set tariffs that are "just and to and different from other Latin American countries. It is reasonable" and "not unduly discriminatory." Nevertheless, similar in that Brazil, like the rest of Latin America, uses some U.S. commissions have chosen to restrict their own concession agreements. It is different in that Brazilian con- flexibility and commit themselves to a multi-year tariff-setting cession contracts for power distribution are essentially system because they concluded that it was a better system "stand-alone" regulatory contracts between the granting for consumers. This suggests that a performance-based, federal or state government and the private company that multi-year tariff-setting system, the key component of the becomes the concessionaire. In contrast, the key elements regulatory contract, should be the preferred approach for of distribution concessions in other Latin American coun- regulating private distribution entities in developing and tries are typically tied to fairly detailed tariff-setting formu- developed countries and not just for a transition period. las specified in comprehensive power sector reform law. Although a similar tariff-setting system was recommended 4. REAL-WORLD REGULATORY EXPERIENCES: for Brazil by the government's privatization and restructuring BRAZIL AND INDIA consultants, the country went ahead with privatization and restructuring without the benefit of a general power sector A. Regulation by Contract: The Latin American reform law. It appears that such a law was never passed in Approach Brazil because of the country's inability to reach political agreement on a reform "game plan." As a result, the exec- The most prominent example of regulation by contract in utive branch of the Brazilian government was forced to pur- developing countries can be found in the more than 60 sue major reforms while relying on a "patchwork" of con- distribution privatizations that have occurred throughout cession agreements and regulations that emphasize "princi- Latin America over the last 15 years.27 In general, the reg- ples" rather than "details," while hoping that more-detailed ulatory component of these privatizations is a pre-speci- secondary legislation would be drafted latter.29 One Brazilian Management contracts could also be viewed as a form of regulation by contract. Management contracts are typically used when a govern- 27 ment is unwilling to privatize or a private investor is unwilling to invest capital in a state-owned enterprise. As a lesser alternative to full priva- tization, the government may write a contract with the private company so that the company assumes responsibility for general manage- ment or one or more operating functions such as meter reading, billing, collection and maintenance. In Africa, management contracts have been recommended as a necessary prerequisite (i.e., a cleanup mechanism) if there is to be any possibility of privatization. Those who make this recommendation are arguing that a company needs to be an operator first in order to be an investor later (Block 1998). Such contracts are often difficult to write and implement because they require careful delineation of the responsibilities of the government and private com- pany. It has been argued that a management contract is harder to write than a full privatization agreement. A 1994 survey of management contracts in electricity and other infrastructure sectors found that generally they were not successful (World Bank 1995). To be more precise, the formula is for the "distribution margin," that is, the capital and operating costs associated with the movement of 28 electricity over distribution lines and commercialization activities. In addition to the distribution margin, there is usually some pre-specified benchmark for the pass through of power-purchase costs. See Section V. B. One private power executive recently observed that the Brazilian power sector "is currently governed by dozens of laws, resolutions, 29 decrees and regulations." As quoted in "Enertrade calls for a new private sector law--Brazil," Bnamericas, January 16, 2002. consultant has observed that Brazil's "original sin" was that tracts seemed to collapse. In 2002 the Brazilian govern- "the plane took off even though we hadn't finished all of the ment proposed a series of 33 major policy initiatives char- design work and there was still some welding to be done." acterized as "mid-course" corrections. It remains to be seen whether Brazil's new government will pursue these Initially, the Brazilian approach of regulation through changes--and if it does, whether the changes will pro- stand-alone concession contracts--a system that gives the duce clarity and consistency or just create new complexi- regulator considerable discretion in applying a general ties and inconsistencies.31 tariff formula--seemed to produce remarkable results at least from the perspective of invesment bankers. By the C. What Went Wrong in Brazil? time ANEEL, the new national electricity regulator, was created in 1997, ten distribution companies had already Because other Latin American countries have generally been sold via concession contracts for a total of US$12 had success with regulation by contract, it is worth taking billion dollars.30 And the average price of US$1,400 per a closer look at why Brazil is now encountering major reg- customer obtained by Brazil's distribution privatizations ulatory and economic problems with its particular beat all previous world records. AES paid a 93 percent approach. The Brazilian experience provides some lessons premium to acquire CEEE in October 1997 and Enron on "what not to do" with regulatory contracts. acquired Elektro at a 99 percent premium in July 1998. In less than five years, private investors invested more than 1. Vagueness in Tariff-Setting Provisions 22 US$27 billion in the Brazilian power sector. These impres- sive results may reflect the fact that the Brazilian privatiza- The tariff provisions in most Brazilian concession contracts tion was, in most instances, led by BNDES, the state- cover only the initial tariff-setting period. Even for that owned privatization bank, which may have been more period, however, some key terms are not clearly defined. interested in maximizing privatization revenues than in cre- For example, a typical agreement will give a distribution ating a workable and sustainable power sector. It may company the right to petition ANEEL for "extraordinary" also reflect a perception on the part of investors that they tariff adjustments if such adjustments are needed to main- would be able to "game" the regulatory system. tain "economic-financial equilibrium."32 But the concept is so vague and subjective that no distribution company has Now, with the benefit of 20/20 hindsight, some of the hid- ever succeeded in getting such an adjustment. den weaknesses in the Brazilian regulatory model have surfaced as the stand-alone concession system has been The concession agreements are even more vague about subjected to major macroeconomic and natural shocks. In what happens at the end of the first multi-year tariff peri- 1999 the country had to confront a 56 percent devalua- od. The concession agreements make a general reference tion of its currency. In 2000, the MAE (the wholesale ener- to the "repositioning" of tariffs at the end of the tariff peri- gy market) was not able to go operational because of sev- od. But the guidance as to what repositioning means is eral legal and technical problems. In 2001, a severe typically limited to a single sentence that states that the drought forced the government to mandate a country- regulator wide electricity rationing program. These major shocks produced disputes over interpretation of various conces- shall process the revision of the amounts of rates for sions, contracts, regulations and laws. Without the refer- commercialization of power, altering them upwards ence point of a single electricity-law-of-the-land and or downwards, taking into account the cost of and detailed tariff-setting provisions, there have been frequent market structures of the Concessionaire, the levels of disagreements as to which of several legal provisions took rates charged by similar companies in the nationwide precedence. By the end of 2001, what may have seemed and international context, stimuli for efficiency and to be a very successful, ad hoc web of regulation-by-con- for reasonableness of rates.33 Joao Lizardo R. Hermes de Araujo (2001). 30 Most of these proposals relate to market and tariff design issues. For a perceptive assessment of second generation of regulatory gover- 31 nance and other institutional issues, see Brown and de Paula (2002). See Concession Agreement of ElectroPaulo, Clause 7, Sub-clause 2. Although many lengthy articles have been written about the meaning 32 of the concept, there is no generally accepted operational definition. The concept of an "economic-financial equilibrium" comes out of the French concession tradition. Although Brazil adopted the concept, it did not adopt the more than 100 years of the French case law that gives operational meaning to the concept. The regulatory contracts in many Anglo-Saxon countries use a different approach. Rather than referring to a general concept, they often refer to a "z" factor that is usually defined in terms of specific events (such as damage caused by extreme weather conditions or changes in tax codes and environmental regulations) that provide the distribution company with a right to petition the regulator for an extraordinary tariff adjustment. Distribution Concession Agreement No 12/97, AES Sul, Article 7, Sub-Article 6. It appears that the concession agreements in Colombia are 33 also vague. See Mercados Energeticos (2002). As a consequence, ANEEL now finds itself in the difficult 2. Uncertainty about Pass-Through for Power- position of having to develop some basic rules and Purchase Costs processes for setting tariffs more than five years after some of the distribution companies were created. This has led to Even in the initial period, the details of the mechanism for bitter and unnecessary regulatory disputes that other Latin passing power-purchase costs through to retail customers American countries have avoided. were not well developed. The concession contracts typical- ly make reference to a general formula that has a compo- Perhaps the most contentious issue is the calculation of nent A for "non-controllable "costs"--mainly power-pur- the regulatory asset base--the amount of capital stock chase costs--and a component B for `"controllable that the regulator uses to calculate the overall revenue to costs"--principally distribution O&M costs. But ANEEL did be recovered in retail tariffs. Like most Latin American not develop detailed rules for the pass-through of power- countries, no value was specified for the regulatory asset purchase costs until after most of the privatizations had base at the time of privatization. But unlike other Latin occurred. ANEEL's rules establish valores normativos American countries, Brazil also did not specify a method- (VNs), which are indexed fuel and technology specific ology for calculating the regulatory asset base at the end caps on the level of power-purchase costs that distribution of the first multi-year tariff period. As the first tariff period companies can pass through to their captive customers.36 came to a close for more than 50 distribution companies, The specific rules for indexation of these VN costs have the trade association of Brazilian distribution companies been clarified or amended four times since the system was proposed a methodology that would produce a regulatory created in August 1998. In contrast, the rules for power- 23 asset base with a value that appeared to be two to three purchase cost pass-through have been more stable in times larger than the value that would result from the reg- other Latin American countries (Argentina, Bolivia and ulator's proposed methodology. Because the dispute was Chile) even though they may create other problems.37 so contentious, the President of Brazil found it necessary to appoint a special inter-ministerial task force to try to 3. Low Allowed Prices for Pass-Through of Power- resolve the dispute at the worst of all times--a few weeks Purchase Costs before the country's presidential elections. Potential investors in new thermal generating plants have Brazil could have avoided this unnecessary regulatory bat- complained that because the administratively established tle if it had simply followed the approach of other Latin VN levels for this type of plant reflect a very low return on American countries, which was to specify, prior to privati- assets, they are not economically viable. In theory, although zation, a methodology for calculating the regulatory asset a Brazilian distribution company could pay more than the base for both the initial and later tariff-setting periods.34 specified VN, ANEEL's regulations would prohibit it from passing through to its retail customers any costs exceeding If it had followed this precedent, the debate would have 1.05 times the VN value. It is not surprising, then, that been limited to the details of applying the methodology Brazilian distributors have been reluctant to sign long-term rather than having to grapple with the threshold issue of power purchase agreements (PPAs) with generators for new which methodology should be used in the first place.35 thermal-fired plants. Such PPAs are needed for financing new generation projects in the absence of a workable spot energy market. To address this issue, in February 2002 the For example, the primary or secondary laws in Chile, Colombia, Peru and Uruguay specify that the allowed capital costs of a distribution util- 34 ity must be based on the capital costs of a "model efficient firm" which is a type of reproduction cost calculation that has now become fairly well-defined in these countries. (See Rudnick and Donoso, 2000). In contrast, Brazilian laws and concession agreements are silent on the calculation of this key cost component of tariffs. ANEEL has also proposed a type of reproduction cost calculation. However, ANEEL would perform the replacement cost calculation on an asset-by-asset basis, whereas the other Latin American regulators use a generalized model to calculate reproduction cost. The problem with the ANEEL proposal is that it requires large amounts of data and is costly and time-con- suming for the company to calculate and for ANEEL to review. In February 2003 ANEEL issued technical notes that strongly suggest that it move to the "model efficient firm" approach used almost everywhere else in Latin America. A major, but not widely appreciated, advantage of this approach is that it greatly reduces "information asymmetry" between the company and the regulator. Distribution companies tend to reveal considerable detailed information about their operations and costs when asked to comment on the accuracy of the model-efficient- firm calculation when it is applied to their circumstances. A discussion of the pros and cons of using different methodologies for setting regulatory asset base is beyond the scope of this paper. See 35 Alexander (2001). For an overview of different approaches by different British regulators, see Carne et al. (1999). The consensus among most experts is that the selected method must be consistently applied once it is selected. ANEEL, Resolution 233/99, July 29, 1999. 36 See for example Article 49 (Nodal Prices) of the Bolivian Electricity Law. Also, see the discussion in Section 5B (Passing Through the Cost of 37 Power Purchases). Brazilian government announced a power sector reform consumption across most of the country. This in turn pro- program that includes a proposal to allow future VNs to be duced claims by distributors that they were entitled to determined by market auctions rather than through the compensation (mostly from government-owned genera- administrative calculations of a regulator. The new Lula tors) for the lower revenues produced by the government- government, which came to power in January 2003, will mandated rationing. The generators, in the final turn, have to decide whether or not to adopt this policy. claimed that they were not required to make such payments because the rationing was triggered by a "force majeure." 4. Foreign Exchange Risk Much of the dispute revolved around conflicting definitions Potential thermoelectric investors have also been con- of "force majeure" found in different laws, presidential reso- cerned about what has been referred to as the "dollar- lutions and concession agreements. Because somewhere cost/real-revenue mismatch." The cost structure for pro- between US$2 billion and 4 billion was at stake, the disput- posed thermoelectric plants in Brazil almost always ing parties were willing to spend millions of dollars on requires significant expenditures in U.S. dollars. With fuel lawyers to protect their claims before the dispute was finally costs, project debt, return on equity, and EPC (engineer- resolved in a wide-ranging settlement in December 2001. ing, procurement and construction) costs denominated in Despite this agreement and subsequent discussions, full U.S. dollars, a gas-fired plant will typically incur about 80 payment of the settlement amounts has not occurred as of percent of its costs in dollars. In the VN formulas, the this writing (early 2003). Other Southern Cone countries 24 weights for dollar versus local currency costs are subject appear to be less vulnerable to this type of dispute because to a case-by-case determination by ANEEL. Although the their concession agreements are embedded in reasonably Brazilian government enacted special legislation to reduce well-defined and consistent hierarchies of laws and contracts. this uncertainty, the new law applies only for a limited number of years and to a limited number of emergency A recent multi-sector study of Latin American concessions thermal plants on a government-approved list. found that the extent of "legal coverage" clearly affects the likelihood that the regulatory contract will be sustain- In contrast, the exchange rate pass-through provisions of able. In a survey of 713 Latin American concessions in other Southern Cone countries38 are usually specified in infrastructure industries (water, transport, telecom and their respective Electricity Laws and therefore control the energy), the probability of renegotiation of the concession provisions in concession contracts.39 In these countries, agreement was determined to be 18 percent if the regula- investors face little or no uncertainty and, if there is less tory framework is in law, 48 percent if it is in a decree and than complete pass-through, they can reflect this risk in 61 percent if it is just in a contract and concession.40 When their offered prices--an option that is not available to the regulatory framework is in law, it is almost always in a generators in Brazil because of the VN ceilings. decree and a contract or concession as well. Presumably, this triple coverage gives comfort to investors and makes it 5. Uncertainty in the Legal Framework less likely that the government will try to renege on the agreement. Also, when the regulatory contract is clearly As discussed earlier, Brazil, unlike its neighbors, has a derived from a national law, there is a smaller likelihood of patchwork of laws, resolutions, regulations and conces- inconsistencies that can lead to later legal battles. sion agreements. It is not always clear how they relate to each other and which takes precedence when they are 6. Lack of Respect for Contracts inconsistent. The regulatory framework was never consoli- dated into a single national electricity law, as was envi- All of this suggests that regulation by contract is not likely sioned when the power sector reform was initiated. to be sustainable unless it is embedded in a reasonably consistent legal framework that reflects a generally con- The problem of inconsistency recently arose in what has sensual vision of the new structure. But this is a necessary been referred to the as the "Annex V dispute." This dispute but not sufficient condition. The legal system could be was triggered by a major drought in the second half of designed to be as "tight as a glove" but it will still fail if 2001. To prevent rolling blackouts and brownouts, the the substantive elements of the contract mandate actions government ordered a mandatory 20 percent cut in power or impose requirements that are not commercially viable. The "Southern Cone" countries of Latin America are Argentina, Brazil and Chile. 38 See, for example, the secondary legislation of Bolivia, which went into effect on the same day as the primary legislation (Government of 39 Bolivia, Regulations for Prices and Tariffs, December 21, 1994, Article 18). See Guasch (2000). 40 Moreover, although the contract could be well written, it operate under a U.S.-style regulatory system, which means could just as well be a "work of fiction" if there is a tradi- they are formally independent of the state governments. tion of not honoring contracts. There are no hard statistics, Unlike their Latin American counterparts, however, they but some knowledgeable observers have claimed that there currently administer a tariff-setting system that (1) is keyed is a tradition of not honoring contracts within the Brazilian to general principles, as opposed to a specific formula;42 power sector.41 This was seen most recently in the aftermath and (2) is annual as opposed to multi-year tariff-setting. of the 2001 rationing disputes. Those power enterprises that owed money had a strong economic incentive not to In practice, this has meant that investors are forced to pay their creditors. The incentive exists, in part, because guess what the level of tariff increase will be in any given Brazilian courts are slower in making decisions than courts year. (See Table 2 for an overview of the regulatory mod- in other Latin American countries. It is not uncommon for els used in France, India, Latin America, the United Brazilian courts to take five or more years to resolve major Kingdom and the United States.) commercial disputes. And when the decision is finally resolved, there may be not be an inflation adjustment on Although the Indian regulatory system superficially resem- the amount due. So the debtor has every incentive to delay bles the U.S. system, it differs in three important respects. payment and try to delay the judicial process because, First, the new Indian regulators have uniformly interpreted even if it loses in court, it will benefit by paying much less their statutes to require that they must completely re-exam- when finally forced to honor the contract. ine retail tariffs every year under a traditional (but flawed) cost-of-service tariff system.43 In contrast, a U.S. regulatory 25 D. Regulation by Principles: The Case of India commission will normally review a company's tariffs only if (So Far) the company or its customers complains to the commis- sion. U.S. commissions are under no legal obligation to The regulatory system in India is quite different from the review retail tariffs every year. Therefore, companies such typical Latin American system. The first sub-national elec- as Kentucky Utilities and Florida Power and Light have tricity commission in India was created in the state of operated for many years without having to file new tariffs Orissa. The other new Indian state electricity regulatory with their state regulatory commissions. commissions, following the Orissa commission's lead, Table 2. Alternative Regulatory Models F R A N C E L AT I N G R E AT U N I T E D I N D I A (WATER SECTOR) A M E R I C A B R I TA I N S TAT E S Separate Regulator? No Yes Yes (national) Yes (national and state) Yes (national and state) Specifity of Regulatory Medium High (Bolivia, High Low (general principles Low (general principles Contract Chile, Peru) interpreted in case law) without case law) Low (Brazil, Colombia) Regulatory Decisions Yes No except for Yes (Competition No No, but proposed by Reviewed by Special Bolivia Commission) government Appellate Tribunal? Ownership of Regulated Private Private and public Mostly private Mostly private Mostly public Entities (varies by country) Form of Private Sector Concessions Concessions that Full privatization Full privatization Full privatization Participation and leases are close to full privatization Maurer (2001). 41 For example, the Orissa Electricity Reform Act states that tariffs shall be "just and reasonable...[and] promote economic efficiency in the 42 supply and consumption of electricity." The tariff-setting criteria in other countries are even more vague. For example, the Hungarian Law uses the term "justified costs" without giving a definition. Laws in Central and Eastern Europe use terms like "objectively determined" tar- iffs based on "rationalized" costs without definitions. See Stern and Davis (1999). These include overoptimistic efficiency targets, less-than-full pass-through of non-controllable costs, and underestimates of cash working 43 capital requirements. Generally, governments will often pressure government-owned power enterprises to project high efficiency improve- ments in their tariff filing because this justifies lower subsidy payments from the government or lower tariff increases for consumers. The government power enterprises will usually accede to the wishes of their owners even though they know that the efficiency improvement estimates are unrealistic. Second, with the exception of Orissa and Delhi, the new 1. Is It Legal? Indian regulators are regulating public rather than pri- vate entities.44 When the regulated entity is government- The problem of regulatory credibility arises because, owned, the regulator will usually find itself trying to move under most current Indian proposals, multi-year tariffs prices up to reflect costs.45 Although state-owned power would be permitted rather than required. This means enterprises often desperately want to file for tariff increas- that an existing Indian state regulatory commission would es, they are usually blocked from making such requests by have to voluntarily give up its current legal right to revisit government owners who fear that any tariff increases will tariffs annually and commit itself to accepting a formula antagonize voters. In contrast, the principal regulatory task that specifies a tariff trajectory for several years. This rais- in the United States, where about 75 percent of the sector es the obvious question of whether an Indian state com- is privately owned, is to get tariffs down to cost levels. mission of 2002 can legally bind the commissions of 2003, 2004 and 2005, etc. to a particular tariff-setting Third, state-owned Indian power enterprises generally ignore the regime and, more important, whether investors would directives of their regulators because the regulators have little or believe such a commitment. no ability to impose rewards and penalties on them. As one Indian regulator observed, "My orders are just pretty poetry." Given the current lack of a clear legal foundation for multi-year tariffs, it has been recommended that any 26 E. Regulation by Contract: A Proposal For India Indian regulatory commission that decides to adopt multi- year tariffs must adopt a "belt and suspenders" approach Several Indian states have announced their intention to to create the new system. This would involve putting the privatize their state-owned distribution systems. To achieve elements of the new tariff system in as many commission this goal, the World Bank has recommended that Indian documents (policy statements, tariff orders and licenses) regulators move to a form of "regulation by contract" for as possible so that its commitment is viewed as genuine potential private distribution companies that would be and irreversible by investors.48 It is also been suggested more akin to what exists in Latin America and elsewhere.46 that any new multi-year tariff system be included as a In India, this new regulatory system is called "perform- component of any proposed privatization agreement, as ance-based multi-year tariffs" or "medium-term tariff fixa- has been done in many other countries. However, it is tion." Like Latin America, the key elements of the pro- unclear whether existing Indian regulatory commissions, posed system are (1) automatic pass-through of cost ele- which were set up to be "quasi-judicial" (i.e., like a ments that are largely beyond the distribution entity's con- court), could be parties to such a contract. And if a com- trol (such as power purchases and taxes) and (2) indexing mission cannot commit, can the government commit on and efficiency targets for cost elements that can be con- its behalf once tariff-setting authority has been legally trolled (such as losses and labor costs). Numerous confer- transferred to the regulatory commission? ences and workshops have been held to discuss how multi-year tariffs might be implemented.47 A newly pro- Given this legal uncertainty, the better solution for India posed national electricity law seems to encourage its would probably be to amend the existing state electricity adoption. But even if Indian state electricity commissions reform acts to mandate the use of multi-year tariffs--or, were to replace their current annual cost-of-service system even better, to (1) transfer tariff-setting authority back to with a multi-year price or revenue cap system, there would the government on a one-time basis for the initial post- still remain a major problem of regulatory credibility. privatization period, (2) incorporate the tariff-setting for- Even though several Indian state power enterprises have now been "functionally unbundled" into separate government corporations, 44 these new corporations still find themselves subject to substantial government control. In particular, they do not have control over the tariff applications that they file with the new state regulatory commissions. In the last two years, Indian state governments have ordered new power enterprises not to file tariff applications, to withdraw previously filed applications or to file tariff applications that do not recover their full costs. The former chairman of India's Central Electricity Regulatory Commission recently observed that "the regulated [state-owned] entities defy the commissions by non-compliance with tacit government approval." Rao (2001). The Indian Ministry of Power estimates that revenues cover only 69 percent of costs for the average Indian distribution entity and this represents a 45 decline from 80 percent. Speech by Ed Lim, World Bank Country Director, Conference on Distribution Reforms, October 12, 2001, New Delhi, India. 46 For example, the Andhra Pradesh Electricity Regulatory Commission (2002) issued a consultative paper on long-term tariff-setting princi- 47 ples that, if adopted, would lead to a type of multi-year tariff. The paper is available at www.ercap.org/home.htm. India is not the only country that may face a legal credibility problem. Turkey may encounter similar problems when it tries to privatize 48 some of its government-owned distribution systems in the next year or two. Because the recently passed Turkish electricity law sets out only general tariff setting principles, any multi-year tariff system would have to be developed in secondary legislation. It remains to be seen as to whether investors will have confidence in the credibility of a tariff setting system that is limited to secondary legislation. mula directly into the privatization agreement (which is 2. Objections to Multi-year Tariffs the norm in almost every other country that has success- fully privatized distribution) and (3) establish, via There are likely to be several objections to the multi­year amendments to the existing state electricity laws, fairly tariff approach. The first is that it would compromise the detailed tariff principles and processes that would apply independence of the existing Indian commissions. We to subsequent multi-year tariff periods. Without such think that this is a misplaced criticism. The concept of changes, any privatization will take place under a cloud independence does not logically require that a regula- of legal uncertainty. (The Delhi government has adopted tory commission design the tariff system that it imple- some elements of this strategy. See Box 2.) ments. In many Latin American countries, independent regulatory commissions have been administering tariff- setting systems that were established by governments before the commissions came into existence. BOX 2. The Delhi Privatizations: A Partial Regulatory Contract "We never had any illusion that the whole world was dying to come and distribute power in Delhi." --Jagdish Sagar, Chairman of DVB, PowerLine, June 2002 On July 1, 2002, the Delhi government sold a 51 percent equity interest in each of three new distribution compa- 27 nies that had been created out of DVB, the state-owned enterprise that had served the metropolitan area. At the time of privatization, DVB was a sick enterprise. It had technical and commercial losses of more than 50 percent and receivables of more US$400 million. Consumers were unhappy with the DVB's quality of service and the endemic corruption. For several years the Delhi government had been forced to prop up DVB with annual subsidies of US$200 to 300 million through loans that no one expected would be repaid. As one Indian official observed: "The government was hemorrhaging through the company and was getting absolutely nothing in return." The negotiated sale to BSES and Tata Power, two private Indian companies, was the first major distribution priva- tization after several failed efforts elsewhere in India. It represents what could be the first example of a second generation of distribution privatizations in India. If successful, it will be an impressive accomplishment.* The tariff system created for the Delhi privatization represents a partial regulatory contract. When it appeared that the government's efforts to privatize would fall victim to regulatory uncertainty, the Delhi government decided to issue a "policy directive" to the regulator. The directive obligated the regulator to (1) accept realistic initial values for technical and commercial losses, (2) adjust tariffs based on the loss improvement trajectory proposed by the bidders and accepted by the government, and (3) allow for the automatic pass-through of subsidized prices that the discos would pay for power purchased from the government-owned transco (trans- mission company). The policy directive created a "partial" regulatory contract in that only some of the performance elements (i.e., loss improvements) and cost elements (i.e., bulk supply costs) were specified on a multi-year basis. Other ele- ments, such as operating expenses and capital expenses, will continue to be reviewed and approved by the regu- lator on a year-to-year basis using general criteria rather than by specific formula. In contrast, the tariff-setting systems used in most Latin America (except for Brazil) are more completely specified. Other Indian states are now considering more-complete, multi-year tariff-setting systems that would be closer to the Latin American model. In at least one of these states, there has been discussion of implementing the new system by amending the existing state law to permit the state government to issue a one-time tariff directive for an initial, multi-year, post-privati- zation tariff period. Apart from the incompleteness of the tariff regime, the regulatory contract was incomplete in one other impor- tant respect. The privatization went forward without an explicit license that spelled out the obligations and responsibilities of the new companies. The commission began developing such licenses only after privatization. The absence of a license prior to privatization obviously creates unnecessary uncertainty and disputes that could have been avoided. * For a more complete description of the privatization by one of its principal proponents, see Sagar (2002). A second objection is that it would put government back and revenues, it could be addressed by including a profit or into the business of setting tariffs, which would undermine revenue-sharing mechanism as one element of the multi-year the principal rationale for creating independent regulatory tariff system. The Delhi government recently adopted this commissions in the first place. But because the closing of approach by requiring a 50/50 split of all earned revenues the large revenue-cost gap that currently exists in most above certain annual targeted technical and commercial loss- Indian states is inherently a political decision, it seems reduction levels. Although this is the functional equivalent of an inevitable and necessary that the government must design extra income tax on the distribution companies, it is does and initiate the tariff-setting system for the first tariff-setting enhance the political acceptability of privatization.52 period and then provide detailed guidance on how it should be applied in later periods. Because the govern- ment has a strong incentive to get viable bids from private 5. THE DETAILS OF THE REGULATORY investors, it is more likely to "get it right" than an existing CONTRACT: WHO BEARS WHAT RISK? independent regulator operating under general tariff guide- lines, which is the current norm throughout most of India. A. Overview of Risks The third objection is that the existing data are too unreliable Many of the disagreements in designing a regulatory to support a multi-year tariff system. It is argued that it is contract involve disagreements over whether the company, hard enough to set tariffs on a yearly basis, much less for its customers or the government should bear a particular 28 several years. It is also claimed that a regulatory contract risk. Table 3 shows the key risks that exist for a new pri- that embodies a multi-year tariff is not feasible in India until vate distribution company and its lenders. From a poten- there are significant improvements in data quality.49The tial investor's perspective, the allocation of risk in the reg- problem with this argument is that the data are not likely to ulatory contract will ultimately affect one of three things: get better under government ownership. Indian regulators the prices that it can charge, the costs that it can recover have been demanding improvements in data from the state- and the quantity of electricity that it can sell. owned utilities for several years with very little success.50 Investors and regulators look at risks from different perspec- This argument also fails to recognize that significant gaps tives. Investors ask: What risks am I being asked to bear? How in data have been the norm rather than the exception in much will it cost to bear this risk? Will I be compensated for most distribution privatizations around the world. In a let- bearing this risk? Governments and regulators ask different ter to the Andhra Pradesh regulatory commission, questions: Would it be fair for government or consumers to Professor Stephen Littlechild, the first electricity regulator in bear this risk? Who will get blamed if something goes wrong? England and Wales, observed that "the Government [at the time of privatization] essentially said `Let the compa- Not surprisingly, an almost universal rule of privatization is nies have revenue equal to the present level increased by that everyone wants someone else to bear the major annual inflation plus a small annual amount X to reflect risks. It is generally agreed that the best principle for risk the need for higher capital expenditure'" with the expecta- allocation is that a particular risk should be borne by the tion that the quality of data would improve (which it did) party that can mitigate or manage the risk at the lowest during the first multi-year tariff-setting period.51 cost.53 But although the principle is easy to state, there is often considerable disagreement over how it should be A related concern in India is that private companies will be able applied in particular situations. This can be seen in a more to take advantage of the poor quality of data to earn high prof- detailed analysis of four of the risks listed in Table 3: pass- its that would be a political embarrassment. Although this does through of power-purchase costs, loss-reduction targets, not seem very likely given the large initial gap between costs foreign exchange fluctuations, and obligation to supply.54 In a recent discussion of the new Indian electricity regulatory commissions (ERCs), it was noted that "All these ERCs have appreciated the 49 need [adopting multi-year tariffs] for doing so but have expressed their inability in the matter, particularly due to the limitations and unreliabili- ty of data." See Godbole (2002). In a recent tariff order, the electricity regulatory commission of the Indian state of Haryana observed that "in spite of this being the third filing of ARR 50 [annual revenue requirement] by Haryana Vidyut Prasaran Nigam [state-owned electric utility], the information as required has not been furnished com- pletely and therefore the commission had no option to grant a number of waivers which were granted last year also." Quoted in Godbole (2002). Letter of Professor Stephen C. Littlechild to the Chairman and Members of the Andhra Pradesh Electricity Regulatory Commission, November 4, 2001. 51 For a negative appraisal of profit (as opposed to revenue) sharing, see Mayer and Vickers (1996). 52 This principle is often misinterpreted. For example, a distribution company may be given the responsibility for mitigating a particular risk (e.g., the risk 53 of future price fluctuations in purchased power) because the distribution company can mitigate the risk at lower cost than anyone else. But the costs incurred by the distribution company to bear this risk will almost always be paid for by its customers. The general form (such as a price cap, revenue cap, or a combination of the two) will have a significant effect on the allocation of risk between the 54 company and its consumers. For an excellent discussion of this issue, see Alexander and Shugart (1999). Table 3. Major Risks for Distribution Companies and their Lenders R I S K E X P L A N AT I O N Collection Risk Risk that the company will be unable to collect its allowed revenues. This might occur for one or more of the following reasons: customers refuse to pay their bills, customers tamper or disconnect meters, company employees receive bribes to make illegal connections or under collect metered or billed amounts, and government officials or courts are unable or unwilling to sup- port disconnections or other actions against non-paying customers. Power-purchase Risk Risk that the company will not be allowed to charge tariffs that recover the cost of its power purchases. This could occur if the regulator disallows the prices paid or the quantities purchased. Demand Risk Risk that the quantity of electricity sold is less than the amount projected by the company or the regulator in setting tariffs. Obligation-to-Supply Risk Risk that the company will collect lower revenues and/or pay penalties 29 because it is unable to meet supply obligations specified in its license or con- cession. The company's failure to supply may be caused by its own actions (e.g., poor transformer maintenance), actions of others (e.g., inadequate generation or transmission capacity) or acts of God (e.g., a major drought). Operating Cost Risk Risk that the company will not be able to recover the costs of operating its distribution system (i.e., the "wires" function) or the costs of retailing electricity (i.e., the supply function) either because the regulator disallows certain oper- ating costs or sets unrealistic performance targets. The allowance for some technical and non-technical losses is sometimes included as operating costs. Capital Cost Risk Risk that the company will not be able to recover its capital costs because the regulator sets a low allowed capital base, disallows costs of certain capital expenditures, or sets low rates of return. Inflation Risk Risk that company's tariff will not be adjusted for general inflation. Foreign Exchange Rate Risk Risk that the company will not receive sufficient revenues from its customers to pay for costs incurred in "hard" currencies. Foreign Exchange Risk that the government will not give the company access to sufficient foreign Convertibility Risk exchange to repatriate earnings and to pay for costs incurred in other currencies. Financing Risk Risks related to the financial risks borne by entities that have lent money to the company. Regulatory Risk Risk that the regulator will reinterpret existing regulations or create new ones that will increase costs or reduce revenues. Political Risk Risks of expropriation, nationalization, war, civil disturbances and breach of contracts. Government Subsidy Risk Risk that the government does not pay promised subsidies or pays with con- siderable delay. B. Passing-Through the Cost of Power Purchases because these costs usually constitute about 50 to 80 percent of its total costs.55 Therefore, it is not surprising that most private The biggest risk for any new private distribution company is that investors seek total and automatic pass-through of all power- it will not be allowed to recover the costs of its power purchases purchase costs, notwithstanding high power prices56and some- for captive customers. Partial or delayed pass-through of times diverging definitions of the term automatic (see Box 3). power-purchase costs could bankrupt a distribution company They argue that such costs are largely beyond their control. BOX 3. Automatic Pass-Through: It's Not What You Say, It's What You Do Some Brazilian officials describe their power-purchase pass-through mechanism as an example of "full" or "automatic" pass-through. This normally means that the regulatory contract provides for frequent and complete adjustments for changes in power-purchases costs. But in fact, this has not been the case because Brazilian law allows the Brazilian electricity regulator to make adjustments in retail tariffs only once a year. The general pro- hibition on more frequent adjustments reflects a macroeconomic concern that indexing or any regulatory mech- anism that mimics indexing (e.g., automatic monthly tariff adjustments to reflect changes in power-purchase costs) could lead to a new outbreak of the hyperinflation that Brazil experienced during the 1990s. This economy-wide, legal prohibition on more-frequent price adjustments has created significant financial risk for Brazilian distribution companies. For example, suppose that a distribution company's retail tariff were set 30 on the assumption that its dollar-denominated power-purchase costs will be US$1 million dollars (3 million reales in Brazilian currency) per month. If the real declines in value against the dollar by 10 percent, the dis- tribution company will need to pay 300,000 more reales per month to its power supplier. Because Brazilian law prohibits it from adjusting its retail tariffs until the next scheduled annual tariff adjustment, it will be losing 300,000 reales per month. And when retail rates are finally changed the following year to reflect the higher power-purchase costs, there is no "catch-up" mechanism to recover the money that the company lost in every month since the previous adjustment. In fact, this was the situation in Brazil until October 2001. Brazilian distribution companies paid for a significant share of their power purchases (e.g., imported power and power from the bi-national Itaipu dam) in dollars and were allowed to recover these costs in Brazilian reales in tariffs that could be adjusted only once a year. ElectroPaulo, one of the distribution utilities serving Sao Paulo, has estimated that it lost about US$180 million between June 1999 and October 2001 because its automatic pass-through mechanism permitted only annual adjustments for cost changes. There are two principal solutions to the mismatch between cost changes and tariff adjustments. The first is to allow for more frequent adjustments. This is the system that exists in Bolivia (monthly), most of the United States (monthly) and the Indian state of Haryana (quarterly). The second is to create a tracking account, the approach adopted by the Brazilian regulator in October 2001. This involves depositing the differences (both positive and neg- ative) between projected and actual power-purchase costs into a special internal account. At the time of the next annual adjustment, the amount of money in the tracking account is added to or subtracted from the then-current level of power-purchase costs. If the distribution company is also allowed to earn interest on the money that accu- mulates in the tracking account, it is made financially "whole" (that is, it does not lose money on the lag in tariff adjustment). In the case of Brazil, the tracking account is "emptied" once a year and consequently does not violate the legal prohibition on more frequent adjustments. Tracking accounts do, however, have problems. They can lead to big tariff increases at the time of the next adjustment, and a distribution company could experience a significant cash shortfall until the adjustment is actually made. Nicaragua has tried to deal with the first problem by creating a pre-specified "trigger" that empties the account if the account balance would produce a tariff increase above a specified size. This is exactly what happened in California. In the fall of 2000, PG & E and Southern California Edison, the two largest distribution companies, 55 were buying power that averaged 20 cents per kilowatt-hour and were required to resell that power to their retail customers at about 6 cents per kilowatt-hour. By the time the California regulatory commission decided to raise the average retail rate, it was too late--PG & E was already in bankruptcy and Southern California Edison was teetering on the brink. See Besant-Jones and Tenenbaum (2001). A guaranteed automatic pass through is of little comfort to a distribution company if it is being forced to accept high-priced power. Even 56 though the distribution company may have been granted a clear legal right to pass through the power-purchase costs, it may be politically or economically impossible to exercise this legal right if the price is too high. Such a concern was recently raised by the only bidder for the Uganda distribution system and was also one of the reasons given by AES when it withdrew from Orissa in India. To deal with this concern, the government of Delhi guaranteed a five-year, subsidized price for power supplied to any winning bidder in the recent privatization of the Delhi distribution systems. In contrast, regulators are generally fearful and suspicious So although regulators may have had the best of inten- of automatic pass-through mechanisms. Their concern is tions in imposing a cap (i.e., protecting consumers from that automatic pass-through will lead to inefficient and slop- inefficient purchases), consumers will ultimately be hurt if py buying practices, "sweetheart deals" (i.e., paying above- investors are unwilling to finance new generating plants-- market prices for purchases from affiliated generators or which, in turn, may lead to rationing and blackouts. In marketers) or even intentional overpayments to generators imposing such caps, there is a tendency for new regula- from distributors in return for a hidden "kickback." Most tors to think of themselves as "masters of the universe" regulators believe that the purchases will not be "economical" and to not think through the likely effects of their actions. unless the company bears some risk of non-recovery through As one Eastern European regulator has observed, regula- a benchmark or some other regulatory mechanism.57 tors forget (or perhaps never realize) that "it is impossible to cheat economics: every investment must be feasible."58 Regulation of power purchases does not exist in isolation. Any regulatory mechanism designed to encourage eco- 1. Purchases Where the Distributor Does Not Have nomical purchasing will inevitably affect the incentives to Discretion build new generating capacity. Consider the case of a regulator that sets a low ceiling on the price of power pur- In designing a regulatory contract, the easiest cases involve chases that a distribution company can pass through to its purchases through "vesting contracts" or from "single buy- retail customers. Although the cap may be formally ers." The former refers to power-purchase contracts that applied just to the distributor, it is, in effect, also a cap on are assigned to distribution companies at the time of priva- 31 generators because no rational distributor will sign a con- tization. Typically, such contracts oblige the distribution tract to buy electricity from generators at prices higher company to buy a certain quantity of electricity from one or than the prices that it is allowed to recover from its retail more generators at specified price for a certain number of customers. But the distributor cannot stop buying if it has years. Because such contracts are usually assigned to dis- an obligation to serve. So even if it no longer purchases tributors by the government as part of the privatization power under contracts, it will still have to buy in a spot or package, a new distribution company is not able to affect balancing market that may have even higher prices. the terms and conditions of the contracts (see Box 4). The regulator is then faced with the decision of whether to Therefore, it is reasonable for the privatization agreement allow the spot or balancing market prices to be passed to require the current or future regulator to allow automat- through to retail customers. If it allows pass-through, con- ic pass-through of all vesting contract costs.59 Similarly, if sumers will end up paying the higher prices that the regu- there is a single buyer who has a legal monopoly to buy lator was trying to block, the only difference is that the on behalf of all distributors, the prices of the single higher prices will come through spot rather than contract buyer's purchases (which may have been inherited by the purchases. If the regulator refuses to allow pass-through single buyer or previously reviewed by the regulator) will of the higher spot prices, then it may bankrupt the distri- generally be automatically passed through in the retail tar- bution company (which happened in California). And if iffs of distributors.60 the regulator imposes a power-purchase ceiling price on both spot and contract purchases, generators may refuse In both cases, there is little or no risk to investors unless to generate electricity--if not openly, then indirectly by the pass-through is delayed or incomplete. But there is a finding an unexpected need to perform lengthy mainte- risk of public pressure to renege on the contracts if the nance on their generating units. vesting contract uses a formula that fixes prices at a signif- icantly higher level than observed spot-market prices. In theory, the need for a regulator to review power purchase costs arises only when a distribution company has captive customers. It has sometimes 57 been assumed that there is no need for a regulator to review the reasonableness of power purchase costs when there is full retail competition (i.e., all retail customers have the right to choose their supplier). But even with full retail competition, many small consumers will choose to remain cus- tomers of their existing distribution companies because it is simply not worth the time to sort through competing offers for the amount of money that is likely to be saved. Other customers with poor payment records may find that no supplier wants to supply them because they are deemed to be too risky. In both situations, regulators will find themselves under pressure to protect these customers from inefficient or dishonest purchases by the local distribution company. For a good discussion of these issues, see Hunt (2002). Attributed to the Chairman of the Polish electricity commission in Power in East Europe, June 2, 2000, p. 10. 58 Such a policy has been implemented or proposed in the distribution concession and license agreements in Argentina, Moldova, Georgia and 59 Brazil. This has been the approach in countries or regions such as Poland, Hungary and Orissa (India) that have relied on the single buyer model. 60 Until the early 1990s, it was also the norm in many parts of the United States where privately-owned vertically integrated utilities were the sole suppliers for captive municipal and cooperative distribution systems that were known as "full requirements customers." BOX 4. What Costs Can the Distributor Control? The heart of any regulatory contract is the tariff-setting system. Most regulatory contracts specify a multi-year system that includes a formula that distinguishes between controllable and non-controllable costs. The distinction is usually based on an assessment of a disco's ability to influence a particular cost.* Changes in non-controllable costs are automatically passed through to retail customers in the tariff-setting formula. In contrast, changes in controllable costs are not automatically passed through. Instead, they are benchmarked and the disco may earn a reward or penalty depending on its performance relative to the benchmark. A variety of benchmarks exist, the most common being the performance of other distribution companies (discos), an external index or the company's own past per- formance. (Appendix A describes a regulatory framework proposed in a South Asian country for dealing with con- trollable and non-controllable costs.) A common mistake made in designing a multi-year tariff system is that the government or the regulator will fail to distinguish between degrees of effective control. For a particular input, a disco may have control over the input price, the input quantity, or both. For example, if a disco is assigned a vesting contract at the time of privatization or is required to purchase from a single specified seller, it will have no control over the price of power purchased. But even if it cannot influence price, it will still have control over the quantity purchased because the latter will depend on its ability to reduce technical and commercial losses. In this case, then, it is appropriate to allow automatic pass- through of the price of power, while benchmarking the quantity. For later post-privatization bulk-power purchases, 32 the disco may have control over price and quantity, and both should be benchmarked. The nature of control over a particular cost item may be quite different between developed and developing coun- tries. For example, in developed countries that have introduced bulk power competition, the price that a disco pays for power may depend greatly on its purchasing skills. Even if the bulk power market is highly competitive, this in itself is no guarantee that the disco will be an effective buyer. Therefore, the price of power purchases is often benchmarked to try to encourage better performance. In contrast, the quantity of power purchased is usually a pass- through because the disco is likely to have achieved close-to-optimal technical efficiency and commercial losses will be very small. In most developing countries the situation is quite different. When a private company takes over from a state-owned enterprise, it is not uncommon for the private company to start operations with loss levels of 30 to 60 percent, largely due to theft and corruption. The biggest potential for cost reductions will be in the quantity of power that the discos purchase and not necessarily in its price. Therefore, it makes sense in most developing countries to focus on setting targets for commercial and technical losses (which indirectly establish a target for the quantity of electricity purchased) because the inefficiencies are large and the opportunities for improvement are significant. * A good discussion of controllable and non-controllable costs for Indian distribution companies can be found in Alexander and Harris (2001). 2. Purchases Where the Distributor Has Discretion sonable to use a market benchmark, this can have unintended consequences. The more difficult case involves new, post-privatization power purchases in which the distribution entity has some Ex ante spot-market benchmarks discretion over the purchases made and the prices paid. The regulatory treatment of such purchases will vary In Argentina, distribution companies regulated by ENRE depending on whether or not there is an organized bulk (the national electricity regulator) are allowed to pass power market. through an estimate of future, geographically differentiat- ed spot prices that are referred to as "seasonal nodal a. Where an Organized Market Exists prices." These estimates are made by the system operator (CAMMESA) based on estimates of what the nodal prices In situations where there is organized market, the natural will be six months into the future and are recalculated inclination of the regulator or the government is to write a every six months. One unexpected consequence of this regulatory contract that uses the market price, either esti- regulatory policy is that distribution companies have little mated or actual, as a benchmark against which to judge or no incentive to enter into long-term contracts because the prices paid by the distributors. Although it seems rea- they have the "no risk" option of automatically passing through the estimated nodal prices to consumers. In such a There are three basic problems with creating benchmarks situation, a long-term contract is a risky option for a distri- that rely exclusively on spot or nodal prices. The first prob- bution company because the price in the long-term contract lem, as noted above, is that it creates a strong incentive could turn out to be higher than the nodal prices allowed for distributors to buy in just the spot market. This, in turn, by the regulator. As a consequence, this well-intentioned creates disincentives for new investment in generation. regulatory policy appears to have caused a dramatic Most generators will not be willing to take the risk of decline in the proportion of purchases made under long- building new plants on the basis of frequently revised esti- term contracts by private Argentine distribution companies.61 mates of spot or nodal prices, particularly if the bench- It has been estimated that long-term purchases constituted mark prices are calculated by the regulator or a govern- about 60 percent of distribution company power purchases ment-controlled entity, which is the prevailing situation in at the beginning of the reforms in 1992. By 2000, the aver- several Latin American countries (Argentina, Bolivia, Chile, age percentage had dropped to about 20 percent. El Salvador, and Peru).63 These disincentives have been counterbalanced to some degree by the facts that (1) the The problems are often exacerbated if the regulator or generators are also eligible to receive a "generation some other government or quasi-governmental body is capacity payment" that can provide a second and more responsible for calculating the spot market benchmark. stable source of revenues and (2) the capacity payments For example, in Peru the benchmark for the estimated are automatically passed through to retail customers. future price of power purchases is calculated by COES, the system operator whose decisions appear to be strongly A second problem is that it forces consumers to bear the 33 influenced by the government power enterprises represent- risk of future price fluctuations in the spot market unless ed on its board. It is relatively easy to manipulate the esti- the benchmark is based on multi-year estimates of spot mates because they are based on projections of average market prices (as in Chile and Peru). Although consumers expected spot prices four years into the future in contrast may actually prefer that the distribution company engage to the three-month projections in Argentina. Since 1997, in hedging activities on their behalf, there is no regulatory COES has calculated the allowed power-purchase pass- incentive for the company to do so.64 through price for distribution companies based on its pre- diction that there would be significant new gas-fired gen- The third and more fundamental problem is that it reflects eration because of the completion of a new gas pipeline a naïve view of bulk power markets--one that fails to rec- from the Camisea basin to metropolitan Lima. This predic- ognize that bulk power can be purchased under a variety tion had the effect of lowering the ceiling price for pur- of terms and conditions. Any regulatory pass-through chased power and the prices charged to retail customers. mechanism that presumes that spot prices are the only Although these low retail prices allowed the government "true prices" will distort the behavior of distributors and to claim success for the power reform program, it conflict- generators and lead to bad outcomes. ed with the reality that the pipeline was not completed by early 2003. Within the Peruvian power sector, these non- Ideally, a regulatory contract should give a distribution existent gas fired plants came to be referred to as the company the incentive to acquire a portfolio of purchas- "ghosts of Camisea." But these ghost plants, by lowering es--some long, some short, some firm, some non-firm. the allowed pass-through price for power purchases, had But such a system will work only if the regulator is willing the real-world consequence of discouraging further private to accept that a distribution company must function as investment in new generating plants.62 more than a passive entity that simply passes through spot or nodal prices to its captive customers. The simple truth is In June 2001, the Argentine government proposed a major change in the power-purchase pass-through mechanism. It proposed that distribu- 61 tors be required to purchase a large percentage of their new supplies of electricity through competitive bidding (under guidelines specified by the regulator). The government proposed replacing the estimated nodal price ceilings with automatic pass through of the actual winning bids. This proposal of mandatory competitive bidding, which was one part of larger power sector reform package, was rejected by the Argentine congress. If adopted, it would have been similar to a system that currently exists in Panama. More recently, the Peruvian spokesman for Duke Energy was attributed as saying that the November 2001 calculation of benchmark nodal 62 prices included a hydro plant that was not in service. See "Duke Pulls Out of Egasa and Egusur Sales," Reuters News Service, May 9, 2002. Peru, Chile and Bolivia have tried to overcome this disincentive by imposing an additional requirement on distribution companies that they buy 63 a high percentage of their supply needs under long-term contracts. Bolivia requires that its distribution companies buy 80 percent of their sup- ply needs in contracts that have a duration of at least three years. Chile and Peru require that their distribution companies obtain 100 percent of their supply needs in contracts that are one year or longer. But this regulatory requirement is unenforceable if distributors are unable to find generators are willing to sell to them at prices that will change every three months based on estimates made by a regulator or system operator who is under political pressure to keep the estimates low. For a discussion of the theory and practice of hedging of purchases by electric and gas distribution companies, see Fernando and Kleindoerfer 64 (1997) and Costello (2002). that no distributor will be a motivated buyer unless it is The ex post, multi-market benchmarks of the Netherlands allowed to recover its hedging costs and has a reasonable and Colombia have several advantages: possibility to earn profits on its purchasing activities. · The regulator does not need to pre-specify an optimal Ex ante or ex post multi-market price caps pattern of spot and contract purchases. · The regulator does not need to conduct after-the-fact In 1999, OFFER, the British regulator, established a sup- "prudence" reviews of the distribution company's pur- ply cap (that is, a cap on power-purchase costs) for 12 chasing practices. distribution companies that contained an 11 percent pre- · It creates an incentive for distributors to engage in mium above its estimate of future spot market prices to hedging practices. encourage hedging. It was left up to the distribution com- · It creates a disincentive for distributors to pay above- panies to decide how they would hedge. Although the market prices to affiliated generators and marketers benchmark was based on estimated spot market prices, because they run the risk that they will not be able to the 11 percent premium created an explicit incentive for pass on the inflated costs to their captive consumers. British distribution companies to try to beat the spot mar- ket prices by signing longer-term contracts at pre-specified Benchmarks are not perfect regulatory instruments, howev- prices or using other hedging instruments.65 er. They will not be fair to any company that is always on the "wrong side of the benchmark" for reasons beyond its 34 Similar incentives also existed through an explicit, ex post control.67 Also, they are only feasible in countries where multi-market benchmark created by the Dutch regulator. In there are multiple distribution companies. the Netherlands, 50 percent of a distribution company's allowed power-purchase costs were based, until recently, b. Where No Organized or Functioning Market Exists on a benchmark keyed to the average cost of power pur- chases by all distribution companies over a three-month Where there is no organized market or the market does period.66 The average cost was based on all purchases, not function, other approaches have to be established for not just spot market purchases. determining the reasonableness of the distributor's discre- tionary purchases. These include the following: The Dutch mechanism resembles a similar mechanism that was created by the Colombian regulator in 1997. · Ex ante, administratively set price caps Distribution companies in Colombia are allowed to auto- · Mandated competitive procurement under regulatory matically pass through about 80 to 90 percent of their guidelines power-purchase costs. However, pass-through of the · Ex ante "reasonableness" reviews of PPAs remaining 10 to 20 percent is keyed to a benchmark rather · Ex post "reasonableness" reviews of PPAs. than to actual power-purchase costs. The benchmark is based on average prices paid by all distribution companies Ex ante, administratively set price caps for all purchases (i.e., spot, intermediate and long-term pur- chases). Even though Colombia, unlike other Latin As mentioned in Section 4, in Brazil the concession agree- American countries, has no requirement that a distribution ments for private distribution companies set the allowed company acquire any specified portion of its supplies from pass-through of power-purchase costs to valores norma- long-term contracts, the ex post, all-market benchmark has tivos (VNs). The VNs are ceiling prices that are administra- created a strong incentive for distribution companies to pur- tively established by ANEEL, Brazil's national electricity sue a mix of purchases. In 2000, Colombian distribution regulator. Until recently, ANEEL opted to establish six sep- companies purchased about 10 to 20 percent of their sup- arate price caps for different fuels and technology based plies in the spot market. The remaining supplies were pur- on ANEEL's estimates of the long-term marginal cost of chased under a variety of contract forms and durations. If a supplying electricity from that particular technology. Each Colombian distributor "beats" the benchmark, it can keep approved power purchase is assigned a VN value at the the savings as additional profits. time of approval by ANEEL, and this initial value is then partially indexed over the life of the purchase. OFGEM (1999). 65 DTE (2002). 66 For example, a distribution company may always be above the benchmark if its customer mix is more costly to serve (e.g., it serves cus- 67 tomers that in the aggregate produce a more peaked load curve). Or it may always have to pay a higher purchase price if it considered less creditworthy by generators (e.g., distribution companies serving an area with guerilla activity). For general discussion of some of the problems in the use of benchmarks by regulators, see Shuttleworth (1999). ANEEL's VN approach has been criticized for two reasons. In the United States, the New Jersey regulator, like its First, generators have argued that the cap for thermal Panamanian counterpart, has also mandated competitive generation was too low, did not differentiate between procurement by all distribution companies. New Jersey peak and base load units, and did not adequately adjust has full retail competition so every customer has the legal for foreign exchange risk. The latter is especially important right to choose an alternate supplier. But the reality is that because thermal generators are forced to pay for their very few retail customers have exercised this right. So the fuel (usually natural gas) in dollars but receive payments procurement is essentially a procurement by the distribu- in reales (the Brazilian currency) for the electricity that they tion companies to acquire the supplies needed to serve a supply to distributors. pre-specified portion of the load of the customers who did not exercise their right to choose, or were not given an Second, the details of the formula were initially unclear. offer by, an alternative supplier. The first auction, conduct- When the distributors were granted their concessions, the ed by an independent third party, was completed in April concessions indicated that pass-through of cost of power 2001 and produced prices ranging from 4.86 to 5.81 purchases in the "free market" (i.e., discretionary purchas- cents per kWh. However, because the New Jersey distribu- es) would be based on the VNs that would be determined tion companies are also subject to a retail price cap, they by ANEEL, but there were few details as to how the VNs could not automatically pass through the results of the would actually be set. Since the inception of the VN sys- mandatory procurement to their customers. If their overall tem in August 1998, ANEEL has found it necessary to costs (distribution costs plus power purchases) go above issue four additional regulations to modify or clarify the the statutory retail price cap, the "extra" costs are deferred 35 operation of the system. This suggests that the Brazilian (i.e., put into a tracking account) for possible recovery "regulatory contract" was effectively a "non-contract" sometime in the future. So the New Jersey system is really because the provisions covering the regulatory treatment a combination of three regulatory tools: mandatory of 50 to 60 percent of a distributor's costs were very gen- competitive procurement for one year's worth of supplies, eral with the specifics were to be filled in later. a mandatory retail price cap, and an after-the-fact review of the "prudence" of all costs (including power-purchase Mandated competitive procurement under costs) that exceed the price cap. regulatory guidelines Ex ante "reasonableness" reviews of PPAs Panama, Nicaragua, Guatemala have opted for a differ- ent approach, focusing on process rather than on out- A third approach is for the regulator to make before-the-fact comes. In these countries, the regulators have issued pro- reviews of proposed power-purchase agreements (PPAs) curement guidelines that require distribution companies to between generators and distributors before they go to finan- acquire new supplies through competitive procurements.68 cial closure. For example, the Andhra Pradesh regulatory Under the Panamanian regulations, a distribution compa- commission in India recently questioned a number of provi- ny is allowed to pass through the costs of new purchases sions in a proposed PPA between a private generator and if it has followed the regulator's purchasing guidelines. the state-owned utility.69 As a general rule, this is an undesir- Distributors in Panama have complained that they cannot able approach to reviewing power-purchase costs because get good prices because the regulator has over-specified (1) it introduces considerable uncertainty and slows down the the purchasing guidelines and is therefore acting more as procurement process, and (2) the considerable discretion it a manager than a regulator. gives to the regulator may create enormous temptation for sellers to try to bribe the regulator to get the PPA approved, especially in those countries with a history of corruption. For the Panamanian regulations, see www.enteregulador.gob.pa/electric/default.asp. A similar approach was proposed by the U.S. Federal 68 Energy Regulatory Commission (FERC) in the late 1980s. The FERC issued a proposed competitive bidding rule and stated that if a utility fol- lowed these guidelines it would pre-commit to finding the purchase to be "just and reasonable." However, the rule was never issued in final form because, among other things, the state electricity regulatory commissions, which had jurisdiction over the retail tariffs of the buying utili- ties, complained that FERC did not have the authority to issue such guidelines. But on their own initiative several state commissions issued competitive procurement guidelines similar to the FERC guidelines for the utilities in their states. The U.S. power procurements in the late 80s and early 90s were always for physical contracts (i.e., contracts that gave the buying utility the right to make dispatch decisions) as opposed to the financial contracts (i.e, the seller commits to supply a specified amount of electricity at a specified price without ceding physical control over its supply source to the buyer) that currently exist in Central America. The use of procurements for physical contracts probably reflected the fact that the buyer was a vertically integrated utility who would integrate the purchase into its own portfolio of supply sources and who did not have access to well-functioning spot market. The guidelines issued by the U.S. state regulators usually required selection based on an assessment of both price and non-price factors. In contrast, in a typical procurement for a financial contract, the selection is almost always limited just to price. For a description of the U.S. experience with mandated competitive bidding, see Plummer and Troppman (1990). Andhra Pradesh Electricity Regulatory Commission (2002). 69 Nevertheless, this may be the only option available to a gas distribution utilities serving that state. In the words regulator if it has jurisdiction over a state-owned utility of one gas company official, these reviews were "time that has been pressured into making uneconomic pur- consuming and frustrating, and left everyone angry chases negotiated by politicians and then handed over because no one got what they wanted."72 A long-time to the utility as a fait accompli. If the regulator simply California consumer advocate similarly described the accepts the PPAs, it will saddle the existing state-owned "retrospective reasonableness reviews" as counterpro- company or future private distribution companies with ductive and ultimately harmful to consumers because impossibly high power-purchase costs.70 Acceptance by they created strong incentives for the distribution com- the regulator of uneconomic PPAs can be a major imped- pany to make purchases "to avoid regulatory disal- iment to any future attempts to privatize distribution. lowance rather than trying to minimize costs."73 In gen- eral, ex post reviews have several major weaknesses: Ex post "reasonableness" reviews of PPAs they tend to focus only on extreme examples of incom- petence or inefficiency, they provide no penalties for Regulators may conduct a "prudence" or "reasonable- failure to adopt best practices, they offer no rewards for ness" review after a PPA has been signed and gone to superior performance, and they tend to drag on and on.74 financial closure. This is the worst kind of regulatory review because it creates an enormous amount of uncer- Perhaps realizing that this was a costly and invasive form 36 tainty. If it becomes the dominant regulatory mode, most of regulation that generally did not benefit anyone other investors will consider the country to be a risky place to than lawyers and expert witnesses, the California commis- do business and will demand a risk premium that will sion replaced its ex post review system with an ex ante ultimately be paid for by consumer. But, once again, it price benchmark system in the late 1990s. Under the new may be the only plausible option for a regulator that is benchmarking system, utilities were given the opportunity seriously trying to protect consumer interests if it suspects to earn profits if they were able to purchase gas at aver- that a state-owned utility was pressured into signing a age prices lower than the benchmark prices. The con- PPA by corrupt government officials who may have sumer advocate described the new benchmark system as received bribes from the generator. The trade-off for the a superior regulatory approach because it "created a regulator is between protecting consumers from paying partnership between shareholders and ratepayers by prices that may reflect corruption versus increasing the aligning their interests." perception that the country is risky for infrastructure proj- ects. Unfortunately, such reviews are often conducted by C. Loss-Reduction Targets officials of a subsequent government or politicians from another political party rather than the regulator.71 When Power purchases raise two questions for a regulatory this happens, there is always a suspicion that the review contract. The first issue, discussed in the previous sec- may be motivated more by a desire to discredit one's tion, involves the prices that the regulator will allow for political opponents than a genuine interest in protecting such purchases. The second issue is the quantity of consumer interests. Therefore, such reviews, whether power purchases that the regulator will allow the disco conducted by the regulator or other government officials, to recover in tariffs. This second determination depends should be used as a "last resort" only when there is per- largely on the level of losses on the disco system that is suasive evidence of outright corruption. deemed to be acceptable. Losses can be thought of as "whatever happens in the great unknown middle" After-the-fact "reasonableness" reviews have been a between the quantity of electricity received at the trans- routine regulatory tool even in developed countries. For mission-distribution interface and the quantity of money example, in the early 1990s the California energy regu- received by the distribution company for electricity that lator routinely conducted ex post procurement reviews is metered, billed and collected from its customers. In to assess the "reasonableness" of purchases made by India, recent estimates of overall losses for some of the This seems to have happened in Pakistan, Guatemala, the Dominican Republic and the Philippines. 70 As this is being written, there is a major controversy in the Philippines over a number of IPPs contracts signed by the state-owned utility. A 71 Philippine politician recently called for a Senate inquiry that "should probe deep into the technical and financial specifications of the expensive contracts." See Manila Bulletin, "Five Costliest IPP-Napocor Contracts Listed," August 19, 2002. See Gee (2001). 72 Procta (2001). 73 In the case of one California gas distribution utility, the final regulatory decision on the utility's 1994 purchases was not made until 2000. 74 existing state-owned distribution systems are as high as America have had considerable success in reducing losses. 50 percent.75 This means that for every two kilowatt- In Chile, overall losses were reduced by more than 50 per- hours purchased, the distribution company is able to cent in seven years. In Argentina, similar reductions were collect money from customers for only one.76 The other achieved in even less time.77 (See Box 5 for a discussion of kilowatt-hour somehow just "disappears." implementation issues in setting loss reduction targets.) 1. Types of Losses Reducing losses can lead to significantly higher profits. For example, in one medium-sized Latin American distri- The overall losses on a distribution system comprise both bution company of about 400,000 customers, it was esti- technical and a non-technical losses. Technical losses are the mated that a 1 percent reduction in commercial losses engineering losses that arise because of the design and added about US$400,000 to the company's net rev- physical operation of the distribution grid. For example, a enues. As some private Latin American distribution com- distribution system with longer feeders will usually show high- panies now move into their second post-privatization tar- er losses because more electricity is lost with increases in the iff-setting periods, they are approaching overall loss levels distance that the electricity has to be transported. Non-tech- that are close to the 8 to 10 percent levels observed in nical losses are commercial losses that result from theft some Western European countries. This is an enormous (sometimes with the active assistance of distribution company change from where the loss levels were 10 years ago. employees), absence of metering, inaccurate metering, 37 under-billing and poor collections. These commercial losses 2. Can Latin America's Success be Repeated in are referred to in Latin America as "black losses." The level Africa and India? of commercial losses very much depends on managerial efforts. In many developing countries, commercial losses will It is tempting to predict that such improvements can be often be two to three times larger than technical losses. repeated in India or Africa or elsewhere. But this overlooks Reductions in technical losses reduce the cost of power pur- two important features of the Latin American privatizations chases while reductions in commercial losses increase rev- that may not exist in India or Africa. First, in Latin America enues. The inability of state-owned power systems in Central the new private owners generally were given full control Asia, South Asia, Central America and Africa to achieve over their labor force at the time of privatization or shortly financial viability is largely attributable to commercial losses. thereafter. Consequently, they retained only the people they wanted. In contrast, most of the proposed privatizations in In dealing with losses, the two key design questions in a India require the new owners keep existing employees regulatory contract are: What should be the initial accepted under the previous terms and conditions of their employ- level of overall losses for tariff-setting purposes? And: How ment contract for several years. Therefore, new private quickly can losses be reduced? The answers to these two companies in India will probably be less successful in questions determine how the cost of losses is allocated reducing losses because some of their employees will be between the company and its customers. In some countries able to sabotage any efforts that force them to give up of Latin America, technical and non-technical losses were money currently earned by promoting theft by customers. estimated to be as high as 30 to 40 percent prior to priva- (In some parts of India, this form of theft is euphemistically tization. Most private distribution companies in Latin referred to as "the micro-privatization problem.")78 In India, the reported numbers have gone up over the last several years. It is probably not because losses have actually increased but because 75 reporting has become more truthful as governments try to get the state-owned companies ready for privatization. There is widespread anecdot- al evidence that state-owned power companies routinely hid their actual losses in overestimates of consumption by agricultural customers. This is relatively easy to do because most agricultural consumption has generally not been metered in India since the 1980s. A recent study of the irrigation pumpset consumption in the state of Haryana found that farmers were actually consuming 27 percent less electricity than the state- owned utility attributed to them. If this result, based on a sample of 584 pumpsets, is extrapolated to the entire state, it would imply that Haryana's actual transmission and distribution losses would be 47 percent rather than the officially reported 33 percent. See Monari (2002). For example, if a distribution company pays US$1 for a unit of electricity and then loses half of its purchased units and then only collects on 76 70 percent of what it bills, it will end up receiving revenues of US$.35 for each US$1.00 of power purchase costs. The Argentinian and Chilean governments included technical loss-reduction targets in the concession agreements for newly privatized distri- 77 bution companies. In contrast, the Brazilian government established no loss-reduction targets for newly privatized distribution companies. This means that the distribution companies have been allowed to pass through the full quantity of power purchased. In its February 2003 proposals for the second, multi-year tariff period of two distribution companies, the Brazilian regulator has stated that it now intends to establish loss-reduction targets rather than just accepting the full quantity of power purchases. In a random sample of Indian electricity consumers, about 30 percent reported paying bribes to employees of power enterprises. Usually, 78 the bribes were paid to linesmen, meter readers and billing employees. This is probably an underestimate for two reasons. First, the survey was limited to individuals and therefore does not capture bribes paid by corporations. Second, it probably fails to capture consumer initiated corruption. See Transparency International (2002). BOX 5. Adjusting for Losses: Specific Implementation Issues In the tariff-setting formula, the adjustment for losses is usually done through a "grossing up" mechanism. For example, if a distribution utility meters and bills 100 units of electricity from its customers, the regulator in a developed country would normally allow the company to charge its customers the cost of purchasing 108 to 112 units of electricity from the disco's bulk suppliers. However, in a developing country where theft has been rampant, the regulator (or the regulatory contract negotiated by the government prior to privatization) may include a formula that assumes that the company will have to buy 140 to 150 units of electricity for every 100 units that it bills its customers. "Grossing up" can be used to account for both technical and commercial losses. For technical losses, the gross-up adjustment allows for the recovery of the cost of electricity that is physically lost on the distribution system. For commercial losses, the gross-up allows discos to recover the cost of power that is stolen. The gross-up for commercial losses allows the discos to charge paying customers for electricity stolen by non-pay- ing customers. In most developing countries where meters do not are exist or have been tampered with, there are always dis- putes over the initial assumed level of losses (i.e., the "initializing value") and the required rate of improve- ment (i.e., the "loss-reduction trajectory"). In India, for example, there is widespread anecdotal evidence that 38 the state-owned enterprises systematically underestimated their losses by claiming that their unmetered agri- cultural customers were consuming more electricity than they actually were. Over the last several years, as many state-owned utilities are being prepared for privatization, these utilities have made substantial upward revisions in their overall loss estimates. (In Orissa, the number was raised from 19 percent to 42.6 percent.) The grossing up can be performed on different bases relating to retail consumption: metered sales, billed amounts or collected revenues. The ideal is to do the grossing up on some measure of physical units billed because the power purchased is measured in physical units. But this may not be feasible where there is incomplete or defective metering among retail customers or the regulator is not able to prevent the company from issuing phantom bills for non-existent customers. As one Indian regulator observed: "There are only two real numbers in the Indian electric sector: purchases and collections. The rest is myth." This has led to a proposal that the grossing up for allowed losses be performed on actual revenues collected (i.e., monies deposited by the disco in a bank). A difficulty with this approach is that it requires that the rev- enues be converted into physical units by dividing total revenues by some measure of average tariff. This can become complicated if there is incomplete or inaccurate information of the quantity of electricity sold to each customer class. The trajectory of loss improvement is also contentious. If the targeted trajectory is too ambitious, private investors may not bid or bid very low. This happened recently in Delhi where the government's privatization proposal set a target of 4 percent loss improvement for each year of an initial, five-year tariff-setting period. Faced with lack of interest from potential bidders, the government eventually agreed to reduce the target to 17 percent over five years when it finally signed a memorandum of understanding with two private compa- nies. The Delhi target is very close to the Uganda government's target of 18 percent reduction in losses over five years. The loss improvement targets in Delhi are relatively low in the first two years but then get much higher in years 4 and 5. The following are the three most important lessons in designing a multi-year loss target: 1. Once the multi-year loss-reduction target is set, investors must be convinced that the trajectory will remain fixed for the entire tariff-setting period. If they think that the government will readjust (i.e., tighten) the targets within the tariff period, they are not likely to make the investments or take the actions needed to bring down losses. 2. The loss-reduction targets should not require annual measurements because this will inevitably lead to disputes that reduce the regulatory certainty that is being sought in the regulatory contract. 3. The number chosen as the base for grossing up losses must be measured consistently for the entire tariff-setting period. BOX 6. Andhra Pradesh: Where There's a Will, There's a Way In 1999, N. Chandrababu Naidu, the Chief Minister of the Indian state of Andhra Pradesh, decided that it would be impossible to privatize the state's power enterprises unless power theft was reduced. With the active encour- agement of the Chief Minister, a strict Anti-Theft Law (the first of its kind in India) was passed by the state legis- lature and went into effect on July 1, 2000. The new law provided for: · A minimum mandatory punishment of 3 to 60 months imprisonment for the theft of electricity. · Mandatory financial penalties ranging from a minimum of US$120 to a maximum of US$1,200. · Residents convicted of stealing electricity would be prohibited from receiving electricity for two years. · The establishment of special courts and tribunals to quickly try cases under the new law. Before the law went into effect, AP citizens were given the opportunity to pay back bills and to "regularize" their status (i.e., become legal customers if they were illegally connected or their request for legal service had not been processed). In a state of about 75 million people, about 1.9 million applications were received for "regulariza- tion." Once the grace period ended, the law was vigorously enforced. From July 2000 to April 2002, more than 2800 people were arrested for stealing electricity (including 87 utility staff and two members of the legislative assembly). Over an 18-month period, billings for electricity increased by 34 percent and revenues increase by 44 percent (while average tariffs increase by 15 percent). Nevertheless, the state-owned power enterprise still experi- enced major deficits because even with the increase in collections, a large number of agricultural and domestic 39 consumers continue to be supplied electricity without metering and under tariffs that recovered only a small frac- tion of the cost to serve them. Second, the new private companies in Latin American general- its own electricity bills. If it fails to take these steps, there is ly had the support of local police authorities in collecting from little point in trying to create a regulatory contract, or at non-paying customers or disconnecting illegal connections. It least a regulatory contract that places the total risk of col- is questionable whether similar conditions of "law and order" lection on the private company (see Boxes 6 and 7). exist in India and Africa.79 In Orissa, there were allegations that the police prevented employees of the AES-owned distri- D. Foreign Exchange Risks bution company from collecting bills or disconnecting cus- tomers. In a letter sent to the Orissa government just before it "The power sector pays dearly for the government's pulled out, the AES representative asserted that "the lack of macroeconomic sins." law and order support has inhibited the company from under- --Latin American consultant, 2001 taking its day-to-day activities such as the collection of dues and the control of the theft of electricity." 80 "Foreign exchange risk is not a risk but a certainty." --Latin American power company official The litmus test of whether a government is serious about power sector reform is the day-to-day support that it pro- 1. Convertibility Risk and Exchange-Rate Risk vides the distribution company to reduce theft. A govern- ment must publicly demonstrate an ongoing commitment Distribution companies receive payments from their cus- to basic "law and order" through the passage and tomers in local currency but often incur costs in hard cur- enforcement of anti-theft legislation that allows for discon- rencies. This leads to two major risks for private investors. nection and prosecution of those who steal electricity. (Box The first risk--convertibility risk--is that the government 6 describes recent efforts in this regard in the Indian State will not give the distribution company access to sufficient of Andra Pradesh.) A good sign of serious political commit- foreign exchange to pay for costs incurred in hard curren- ment is if the government successfully prosecutes one or cies.81 The second risk--exchange-rate risk--is that the two rich or politically well-connected individuals who have local currency will lose value relative to hard currencies. If been stealing electricity. The government also needs to pay the local currency loses value, the distribution company In May 2002, several government officials including a police officer and his wife were held hostage by villagers in the Indian state of 79 Haryanna when they came to collect pending electricity bills. Letter to Chief Secretary, Government of Orissa from AES Orissa Distribution Private Limited, July 25, 2001. 80 The World Bank's Multilateral Investment Guarantee Agency (MIGA) will provide guarantees against currency inconvertibility and transfer 81 restrictions. It does not provide guarantees against currency depreciation. See www.miga.org. BOX 7. When the Gap Is Large The Latin American approach to tariff-setting for privatized distribution entities has generally been successful. Since the mid-1980s, more than 60 government-owned distribution entities have been privatized. But it may be a mistake to assume that the Latin American approach will work equally well in the poorer countries of South Asia and Africa that are now considering privatization. The biggest difference between Latin America and these other regions is that the starting conditions are not the same. India, for example, is different from Latin America in that: · The current gap between revenues and costs is larger in India (currently revenues fall short of costs by an average of about 30 to 35 percent among Indian state-owned utilities); · Indian state governments are not in a position to provide credible guarantees of direct or indirect subsidies while the gap is being closed; · Most sales of electricity to agriculture are not metered; · The local police may not support private companies when they try to disconnect non-paying or illegal cus- tomers (the "law and order" problem); · A private operator may have limited control over the composition of its labor force in the critical early years of operation because of government commitments to employee unions. India, like many poor countries, suffers from the "short blanket problem"--the power's sector's revenues are not 40 currently large enough to cover the costs of generation, transmission and distribution. Given these starting condi- tions, India as well as other poor countries will probably need to employ a different kind of regulatory contract. One alternative proposal made in the Indian state of Karnataka is called the "distribution margin" approach. Its three key elements are (1) the private distribution company would be granted "first rights" to the flow of rev- enues collected from retail customers; (2) certain risks (such as collection risk), traditionally borne by a private company, would be explicitly shared with the government; and (3) the government rather than the private com- pany would bear the risk of paying for bulk power purchases and transmission services because the payment received by the state-owned transco and genco would be a residual. The distribution company would commit to handing over to the government whatever revenues remain after the distribution company takes payment for its distribution margin, which would depend on its success in meeting pre-specified performance targets. Because the residual amount would not be adequate, any remaining shortfalls in payments to the transmission entity and generators would become the government's obligation. Among government and power sector officials, the initial reaction to the "distribution margin" approach was largely negative. It was not uncommon to hear such comments as: "What is the point of privatizing if we are going to bear so much of the risk?" or "This is no different than the demands of IPP developers to put their payments into escrow accounts" or "The disco will not care about retail tariff levels because its payments are guaranteed." We think the distribution margin approach is worth exploring further for three reasons. First, it has the flexibility of being consistent with a range of risk allocations despite the fact that the initial proposals in India put most risks on government. And like the Latin American price cap approach, it can be combined with pre-set targets for improvements in losses and quality of service. Second, it could be used for a transition period with a pre-sched- uled switchover to a more traditional regulatory contract like the ones used in Latin America. Third, it has the potential to "jump-start" improvements in distribution service that may provide government with the political capi- tal for closing the overall cost-tariff gap. But it will probably not be sustainable unless it is combined with an explicit commitment to raise overall retail tariffs in the state and require that farmers pay more than a token amount of money for their electricity. Table 4. Trends in Foreign Exchange Rates in Selected Countries Where American and Foreign Power Had Subsidiaries C O U N T R Y C U R R E N C Y U S $ E XC H A N G E R AT E 1 9 4 3 1 9 5 0 1 9 6 0 P R E VA I L I N G AT DAT E S O F M A J O R AC Q U I S I T I O N S Argentina peso $0.424 $0.247 $0.094 $0.012 Brasil, free cruceiro .120 .049 .053 .005 Colombia peso .973 .572 .510 .151 México peso .499 .206 .116 .080 Venezuela bolivia .193 .299 .299 .299 Note: All rates are yearly averages except for the 1960 rates, which are for March 31, 1960. Source: Gomez-Ibanez (1999). will find that the revenues it receives from its local cus- 2002 demonstration against indexing of foreign costs in tomers in local currency will buy smaller and smaller La Paz Bolivia, one protester said: "It is not fair that the quantities of the foreign exchange needed to pay for population is ever poorer, while the international compa- imported materials, to make interest payments to for- nies are ever richer for the same work, just because their eign lenders or to repatriate profits to its investors. The rates are dollarized."84 risk, then, is that the local-currency revenues may no 41 longer be sufficient to cover foreign currency costs. This What the statement fails to recognize is that consumers is not a hypothetical risk. Between 1975 and 1995, the will inevitably pay in some other way even if indexing is currencies of emerging markets declined by an average prohibited or incomplete. If indexing is prohibited, this of one percent per month relative to the dollar.82 Table 4 will discourage domestic and foreign investors from shows the actual levels of devaluation experienced by investing in the sector.85 This, in turn, will eventually hurt one U.S. power company over two decades of opera- consumers through power shortages. Alternatively, if tion in five Latin American countries. indexing is prohibited but the government has a side agreement to provide investors with access to foreign 2. Indexing exchange on subsidized terms, consumers will have to pay higher taxes so that the government can recover the Indexing is the most common and transparent way to foreign exchange subsidy granted to investors.86 In the deal with exchange-rate risk.83 If the local currency loses first case, the shortages may not arise for several years so value relative to the hard currency, the government the problem is transferred to some future government. In allows the disco to increase its tariffs by the amount the second case, the subsidy will be hidden in general taxes necessary to cover the costs incurred in hard currencies. rather being observable in highly visible tariff increases. In theory, this transfers the risk to the distribution com- Both of these alternatives to indexing are appealing to pany's consumers. Whether this happens in practice politicians because it replaces the visible with the invisible depends critically on the extent of indexing and the fre- or it postpones the pain to a later government. quency of adjustments. a. Indexing of Power Purchases Although indexing tends to work reasonably well if the devaluation is small, it usually breaks down when there The two major components of cost for any distribution is a large, sudden devaluation because, if strictly applied, company are power purchases and distribution costs it would trigger large increases in retail tariffs. Not sur- (the cost of sending power over distribution facilities prisingly, indexing is controversial. During an October and selling it to end users). Among the Southern Cone See Gray (2003). 82 This is not the only available technique for mitigating this risk. Other possible options include reducing foreign currency expenses as a pro- 83 portion of total expenses, increasing local currency financing, generating hard currency revenues, buying hedging instruments and obtaining government guarantees for a fixed exchange rate. A fuller discussion of these other techniques will be presented in Wright (2003). BNAmericas, October 2, 2002. 84 It will also discourage investment by domestic investors because they, too, need foreign exchange indexing if they buy equipment or 85 acquire capital from outside the country. Between 1953 and 1961, Rio Light, the American and Canadian owned company that served Rio de Janeiro, was classified as an "essen- 86 tial industry" and therefore "could import equipment, remit interest on foreign debt and transfer profits--all at preferential rates that repre- sented about one-half the cost of exchange to non-preferred sectors." The advantage of subsidizing the exchange rate was that it "provid- ed a less politically sensitive way of returning to the company roughly what the artificially low power rates took away." Tendler (1968). Table 5. Increase in End-User Tariffs Arising from Different Levels of Indexing for Foreign Exchange Fluctuations (%) PROPORTION OF COSTS 10% 50% 100% 300% THAT ARE INDEXED DEVALUATION DEVALUATION DEVALUATION DEVALUATION 20% 2% 10% 20% 60% 50% 5% 25% 50% 150% 100% 10% 50% 100% 300% countries (Argentina, Brazil and Chile) about 45 percent receives a VN value that is indexed over the life of the of final tariffs are attributable to power-purchase costs contract. In the case of natural-gas­fired units, about and 55 percent to distribution costs.87 However, not all 75 percent of the VN value is indexed to the dollar and of these costs will be indexed for loss of purchasing 25 percent to local inflation. The decision on the appro- power relative to hard currencies. priate weights is made by the regulator on a case-by-case basis and ANEEL is allowed to change the weights after Within the power-purchase category, the percentage of ten years and then after every five years. costs that are indexed for foreign exchange fluctuations 42 ranges from 80 to 100 percent for distribution compa- Unlike Bolivia, Chile and Peru, Brazil has a significant nies in the Southern Cone. In general, the indexing is lag in the indexed adjustments because Brazilian law keyed to benchmark prices rather than the actual prices specifies that indexing can be performed only once a paid by the distribution entity. For example, the bench- year in any contract. In October 2001, the regulations mark in Bolivia are "nodal prices" for energy calculated were changed to allow the distribution companies to every six months by a stakeholder committee of the charge interest for the time value of money lost between National Dispatch Center--and these prices must be the yearly adjustments. (See Box 3 in Section 5). More approved by the regulator. The nodal price is an esti- recently, there have been public discussions of replacing mate of the short run marginal cost of supplying at dif- the VN system with power-purchase price caps keyed to ferent nodes on the system. Between calculations, the the outcome of a mandated competitive procurement nodal price is 90 percent indexed to changes in the requirement for distribution companies or some other price of gas (which, in turn, are linked to a dollar index) market benchmarks. If this change were made, then the and 10 percent to an index of local inflation. There is a foreign exchange adjustment mechanism would no different weighting for energy capacity payments--70 longer be administratively determined but would be the percent is tied to the U.S. dollar exchange rate and 30 outcome of a competitive procurement process or a percent to local inflation. The adjustments for both ener- specified market benchmark. gy and capacity are made monthly. The details of the indexing arrangements are written into Bolivian law, b. Indexing of Distribution Costs which makes the indexing more certain from an investor's perspective. The level of foreign exchange indexing is lower for dis- tribution costs than for power costs. This reflects the fact Brazil uses a different type of benchmark. As discussed that a higher proportion of distribution costs are earlier, until early 2002, indexing was applied to valores incurred locally. Peru and Chile allow about 12 to 17 normativos (VN), which are the regulator's estimates of percent of distribution costs to be indexed to the dollar; the long-run marginal cost of generating electricity from this is roughly the percentage of capital costs within different fuels and technologies. Each distributor must overall distribution costs. Neither Brazil nor Bolivia file with ANEEL the proposed long-term contract it wish- allows any indexing of distribution costs to the dollar. es to sign. ANEEL reviews the contract to make sure the Instead, they provide for indexing that is tied to one or percentage of foreign exchange indexing does not more measures of domestic inflation rather than to a exceed its guidelines. Once the contract is accepted, it foreign exchange index. In India, the proportions are quite different. For a typical Indian distribution entity, about 75 percent of the final tariffs are attributable to 88 power-purchase costs and 25 percent to distribution costs. The higher proportion of power-purchase costs reflects the fact that Indian distribution entities have to buy more power to make up for high levels of theft and that the power they purchase is more expensive because they generally lack access to large quantities of hydro power or low-cost natural gas. However, even an index that ostensibly measures just c. The Special Case of Argentina local inflation may, in fact, be heavily influenced by for- eign exchange fluctuations. This seems to be the case in Until recently, Argentina has been an exception to the Brazil. The concession contract of most Brazilian distribu- Latin American rule because of its over-indexing to the tion companies provides that distribution costs will be dollar. For the three privately owned distribution compa- indexed to a Brazilian price index known as the IGPM nies in Buenos Aires regulated by the national electricity that is significantly affected by foreign exchange fluctua- regulator, all costs, whether incurred domestically or tions. About 50 percent of the distribution companies overseas, were indexed to the U.S. dollar until early overall costs is comprised of distribution costs, while the 2002. (In contrast, only about 40 to 50 percent of end- other 50 percent is largely attributable to the cost of user tariffs are indexed to the dollar in other Southern power purchases. Within the distribution cost category, it Cone countries.) This left Argentina vulnerable when the has been estimated that about 80 percent of the costs Argentine peso experienced a major devaluation relative are wages and salaries. In 2002, there was almost no to the dollar in late 2001. (See Table 5 for a simulation inflation in Brazilian wages and salaries but the IGPM of the relationship between end-user tariff increases and index increased by about 20 to 28 percent. percentage of costs that are covered by foreign exchange indices.) If the tariff formula had been applied Brazilian consumers and government officials have argued as specified, the devaluation would have triggered that that it is unfair for distribution companies to be allowed to increases of about 300 percent in retail electricity tariffs. 43 increase the entire distribution cost component of their retail tariffs by 20 to 28 percent when the actual inflation in distri- This never happened, however, because in January bution costs was much lower. However, this criticism ignores 2002 the Argentine government enacted a law that pro- the fact that the phenomenon is cyclical. In other periods, the hibited the continued usage of such formulas. The law foreign exchange rate may be stable, while domestic inflation stated that "any dollar index or other foreign currency is increasing. During these periods, the IGPM index will fail to index clause or any foreign countries price index provid- reflect inflation in distribution costs. Although the ideal would ed for in any contracts signed by the Public be to use an index that is more closely correlated to actual Administration is declared null and void from the enact- distribution cost changes (whether caused by domestic infla- ment of this Act...."89 As this is being written, negotia- tion or foreign exchange depreciation), such indexes do not tions are under way to replace these indices with some exist in most developing countries.88 other mechanism; however, it remains unclear what the new system will be and how it will affect future foreign Adjustment of distribution costs can also occur in other investment in the power sector. ways. In most Southern Cone countries regulators recal- culate the distribution rate base at the end of a four-to- E. Obligation to Supply seven­year tariff period using an estimate of reproduc- tion cost rather than historic cost. Because the calcula- Obligation to serve goes by different names. In some tion is based on an estimate of a new, optimized distri- common law countries, it is referred to as the supply bution network to serve the distribution company's par- obligation. In civil code countries, it is usually described ticular configuration, the effect of foreign exchange as a public service obligation. In most developing depreciation will be picked up in the calculation if the countries, the state-owned utility has always had a pub- optimized network includes the use of imported equip- lic service obligation. This legal obligation usually ment. This leads to a larger rate base that, in turn, pro- flowed from the fact that supply of electricity was duces higher tariffs. defined as a "public service" (i.e., an obligation of the state) in the country's constitution.90 As one Indian regu- In France, the law "expressly forbids the linking of any contract to the consumer price index, to `the general level of prices or salaries,' or to 88 `prices or goods or services having no relation to the objective of [the contract] or to the activity of one of the parties.'" Shugart (1998), p. 111. Some private investors argue that the January 2002 law does not affect the indexing provisions in their concession agreements. They contend 89 that it only repeals a 1991 peso-dollar convertibility law and cannot nullify indexing provisions in specific contracts. This legal argument has yet to be tested in Argentine courts. In developed countries, the public service obligation is defined more broadly than the obligation to supply. An Italian electricity regulator 90 stated that the public service obligation in his country had four elements: universal service, security of supply, environmental protection and promotion of competition. Available at www.iea.org/about/forum/garribba.pdf. lator observed, however, such requirements were often ate obligation. Among the more important starting nothing more than "pretty poetry" because state enter- points are the size of the revenue-cost gap, the extent prises rarely achieved what was required of them. of electrification, and adequacy of overall installed generation capacity. These starting conditions, in turn, The reasons for failure have been well documented: affect the two key dimensions of obligation to serve: lack of money, ongoing political interference in operat- geographic scope and whether the obligation is ing and investment decisions and, perhaps most impor- absolute or conditional. tant, lack of incentives.91 The last point simply refers to the fact that most officials in publicly owned power a. Geographic Scope enterprises do not earn higher salaries if they succeed or lose their jobs if they fail. In contrast, the public serv- In most developed countries, a distribution company's ice obligation of a private distribution company is obligation to serve is typically defined relative to a viewed as real and enforceable. When a private com- specific geographic area. Within this area, the com- pany takes over, one Indian government official pany is obligated to connect every potential customer observed that "excuses are no longer acceptable and and provide retail customers with the same quality of good performance is expected from Day 1." service--unless a customer chooses to pay more for a higher level of service or less for a lower level (for 44 1. Toward a Precise Definition example, paying less if the company has the right to interrupt with notice for a certain number of times In designing the regulatory contract for the new private during the year). owner, a key design question is: what should be the specific elements of the obligation to serve? In most However, this universal obligation to serve is generally countries, it is generally accepted that the loose "univer- not feasible in developing countries where there is sal public service obligation" that was adequate for the less than full electrification. This is explicitly recog- state enterprise (probably because it was neither con- nized in many Latin American countries, where the tested nor enforced) will have to be replaced with a typical concession agreement often distinguishes a more precise definition of obligation to serve for private company's obligation to serve depending on the companies. The new definition must answer the follow- potential customer's location relative to existing distri- ing questions: bution facilities. The concession agreement may require that the company connect any potential cus- · Who must be served? tomer within 100 meters of an existing distribution · What are the initial and phased in technical and com- facility, grant first right of supply but not an obligation mercial standards for service? to serve for potential customers between 100 and · What are the penalties if the company fails to meet 500 meters, and impose no obligation at all if the these standards? customer is located more than 500 meters from an · Are excuses allowed? existing distribution facility. Within the first band of 100 meters, the company is usually obligated to con- The regulatory contract, whether it is a concession or a nect customers for a pre-specified "regulated" con- license, has to give clear answers to these questions.92 nection charge. Beyond this first band, the company is generally permitted to charge customers a connection 2. Starting Points Matter fee based on the company's actual costs. (See Appendix B for a similar approach proposed in a The definition of obligation to serve cannot and should South Asian country.) This approach reflects the not be the same across all countries. A system's starting broader policy decision that the government rather conditions must be considered in defining an appropri- than the private company will be responsible for rural electrification. World Bank (1995). 91 In India, one criticism of the Delhi government's privatization proposal was that the privatization package did not contain a proposed 92 license so that bidders do not know the extent of their legal obligation to serve. In contrast, most other Indian states (Haryanna, Andhra Pradesh, Uttar Pradesh and Karnataka) that are contemplating privatization have issued detailed licenses for their government-owned dis- tribution entities that give potential bidders a clearer idea of the service obligations that will be expected of them. A company's obligation to serve may be further qualified generating capacity or inadequate transmission capacity by the condition that the company is not obligated to pro- to transport the electricity back to the distribution system. vide a customer with electric service unless it can recover This situation arose in the case of Telasi, the AES distri- the costs of serving that customer. For example, a conces- bution company that was serving the capital city of sion or license may specify that a company is not obligat- Tbilisi. During the winter of 2001, Tbilisi experienced ed to connect a new customer unless it receives payment many hours of blackouts. The regulatory commission for some or all of the capital costs of connection from the announced that it would penalize Telasi for a failure to customer, the government, or a combination of the two. supply even though the regulatory contract within the pri- This is the regulatory regime in Guatemala and Chile. In vatization agreement seemed to specify a "best-efforts Guatemala, the government has used the proceeds of standard." Telasi protested and threatened to leave the privatization to create a fund to subsidize the cost of inter- country.95 Among other things, it argued that it was the connecting new poor rural customers. The private com- only distribution company in the country to have paid pany receives the capital cost subsidy only after the con- "hard cash" to generators. It also contended that its fail- nection is made and is found to satisfy pre-specified tech- ure to supply was caused, at least in part, by the fact that nical standards.93 In Chile, the government provides sub- about 25 percent of the power that it had paid for had sidies for both the capital costs and operating costs to been diverted by the government-controlled system oper- serve poor rural customers.94 The subsidy does not come ator to non-paying, state-owned distribution entities.96 for free. The government will not pay the subsidy unless 45 the customer pays his part of the bill. Therefore, in both Without judging the merits of AES arguments, it appears of these countries the obligation to serve new customers that in countries where there is inadequate generation is a limited rather than a universal obligation. Specifically, supply, it is unreasonable and counterproductive for a the obligation to connect and supply new customers is government to impose an absolute obligation to serve contingent upon payment of a subsidy by government. on new distribution companies. During the initial post- privatization years, the standard has to be a conditional, b. Absolute or Limited Obligation? best-efforts obligation. In practice, this means a stan- dard that includes some or all of the following elements: When Argentina initiated its privatization process in 1992, it imposed an absolute obligation to serve on the three new The company should use all reasonable efforts to pro- private distribution companies that served metropolitan vide energy and related services to customers 24 hours Buenos Aires. The concession agreements required the dis- per day using Good Utility Practice provided that the tribution companies to provide service to all customers with- company is allowed to charge customers the cost of in their service areas even if there was an "upstream" failure providing that service or the government makes timely by generators or the system operator. This is, in effect, an subsidy payments for any shortfall between the cost of absolute, "no excuses" obligation to serve. One advantage serving the customer and the tariffs that can be of this approach is that the regulator does not need to sort charged that customer. Moreover, the company shall out who was responsible for the failure because the distribu- not be liable for any failure of the company to supply tion company is always held accountable. But it is a feasible energy that arises from (a) a fault or failure of any part standard only if a country has close to an adequate supply of the network that could not have been prevented by of generation at the time of privatization. Good Utility Practice (taking into account the state of the network that existed on the date of this agreement); It makes little sense to impose an absolute obligation to or (b) the failure of any energy supplier, transmission serve on newly privatized distribution companies in coun- company, system operator to generate, transmit and/or tries that are starting from a base of insufficient available dispatch energy; or (c) any Force Majeure event.97 Harris (2002). 93 For example, in August 2000 the government of the Bolivian Department of Cochabamba sought bids from private developers to extend electrical 94 service to 380 unserved rural communities. The bidders bid the right to receive a one-time capital cost subsidy from various local governments. The winning bidder, ELEFEC, received a subsidy payment of US$7.3 million or 90 percent of the total capital costs of US$8.1 million. The same issue arose in New Delhi when there were widespread blackouts four days after the government-owned system had been priva- 95 tized. A high level executive in one of the new private distribution companies was quoted as saying that the "private Companies have only taken over the distribution of power not the transmission. We are only able to distribute the power that we receive." "India's Power Privatization Leaves New Delhi Blacked Out," The Times of India, July 05, 20002. AES Press Release at www.aes-telasi.com/news.htm. 96 From a draft of a proposed distribution license in an East Asian country considering privatization of distribution.98For a good overview of 97 quality of service regulation, see Foster (1999). c. Quality of Service · Quality-of-service standards and associated penalties and rewards should be phased in over time. Any Although it is counterproductive to try to impose quality- penalties should be related to the disutility experi- of-service standards that cannot be met, this does not enced by the customer and the costs likely to be mean that quality of service should be ignored. incurred by the distribution entity in meeting the stan- Unfortunately, although everyone talks about improving dards. quality of service, in practice technical and commercial · Where it is feasible and efficient, penalties should be quality of service get very little attention in most distribu- paid to individual consumers. Otherwise, penalties tion privatizations. This happens because it is much eas- should be used to provide subsidies to poor cus- ier to specify tariff rules than quality-of-service stan- tomers. Penalties should not be used to support the dards. Basic electricity laws usually make only general budget of the regulator or any other government entity. references to quality of service. And because quality of Penalties and rewards should be capped so that they service gets little attention, consumers often associate do not exceed more than 2 to 4 percent of the distri- distribution privatizations with "higher tariffs and nothing bution entity's overall revenues. else." The danger in ignoring quality of service is that · Any changes in quality-of-service standards should be the political support that exists for privatization will soon synchronized with the regulatory proceeding to update disappear in the absence of some "early wins" on quali- tariffs for a new multi-year period. 46 ty of service. As one high level Indian government offi- · The regulatory entity should have the legal authority cial observed, "We can fight the political battles if the to delegate quality-of-service monitoring and the supply is good." imposition of penalties to a third party. However, the regulatory entity has ultimate responsibility for ensur- Appendix C presents a proposed regulatory framework ing compliance even if it chooses to delegate to a for establishing quality-of-service standards developed third party. for a South Asian country. The framework is based on · The regulatory entity should establish a reliable, objec- the following principles: tive and publicly available monitoring system that com- pares the quality of service provided by different distribu- · Quality-of-service standards should be established for tion entities.98 those dimensions of service that are important to con- sumers, controllable by the licensee and capable of d. Extend the Obligation to Generators? being measured on a reasonably objective basis. · Quality-of-service standards need not be uniform across In most countries where there has been unbundling (i.e., all customer categories or geographic areas. The stan- vertical separation), the obligation to serve, whether dards should be based on customers' preferences and absolute or conditional, remains exclusively with the their willingness to pay for the costs of providing the distribution entity.99 One exception is Chile. Following specified level of quality. Quality-of-service costs money. three major droughts that led to rationing, the Chilean The regulator should not impose quality-of service stan- government in 1999 modified the existing electricity law dards on a distribution company unless its customers to impose an obligation to serve on existing and future are willing and able to pay for the costs associated with generators to serve certain regulated customers or meeting the standards. compensate them for failure to supply (i.e., failure pay- · Quality-of-service standards should be established for ments) even if the failure was caused by drought. In both technical and commercial dimensions of service. other words, drought would no longer be defined as a The quality-of-service standards may be (1) guaran- force majeure for hydro generators. teed standards where the standard must be achieved in all specified cases and (2) overall standards where It appears that the imposition of this "extra-contractual" the standard must be achieved on average across a obligation to serve, when combined with the risk of specified customer category but need not be satisfied "failure payments," has made Chile too risky a place to for all customers in the category. invest for generators. Even if one concluded that it is For a good overview of quality of service regulation, see Foster (1999). 98 For example, Article 2 of the Bolivian Electricity Law defines the electrical output of generators as a "commodity." The rationale for this 99 designation was to make it clear that generators did not have any obligation to serve beyond what they voluntarily agree to in privately negotiated contracts. equitable and efficient to impose the drought risk on A distribution utility can be involved in many disputes. generators, such a policy will be unworkable unless The three principal types of disputes are those between generators are explicitly allowed to recover the expected the distribution company and its customers, between the costs of making "failure payments" to their customers in distribution company and other industry participants, times of drought. But the current Chilean system fixes a and between the distribution company and its regulator. ceiling price that generators can charge distributors Our focus here is on the last type--disputes between which is limited to the regulator's estimate of nodal the distribution and the regulator over either the sub- (i.e., location-specific marginal) costs. The nodal price stance of the regulator's decisions or the process by has no explicit provision for recovering the expected cost which the regulator reached these decisions. of "failure payments." Therefore, it should not be surpris- ing that generators are reluctant to build new facilities to A. Who Regulates the Regulator? serve the incremental needs of Chilean distributors.100 Independence is not infallibility. The fact that a regulator Faced with a looming energy shortage, the government has been given some degree of legal independence has proposed a "solution" that would give it the authori- does not mean that the regulator will always make the ty to order the system operator to conduct a competitive right decision. Thus there must be some mechanism to procurement for new thermal generation whenever the review the regulator's decisions. This is not easy expected supply is deemed inadequate to serve demand because the almost universal reality is that regulators do 47 during the next 30 months. The problem with this "solu- not like their decisions to be reviewed.101 tion" is that it leads to more government intervention and does not address the fundamental problem: impos- If there is a mandated mechanism of review, most new ing an obligation to serve on generators without any regulators in developing countries would prefer that the provision to compensate them for the cost of meeting review take place in a regular court of law. This prefer- this obligation. The basic lesson is that legal obligations ence probably reflects the fact that most courts will gen- impose costs and the costs must be recoverable in tariffs erally review challenges only on points of law or the or there will be insufficient investment. process by which a regulator arrived at a decision, rather than the substance of the decision itself. And because a regulator is usually able to avoid making a 6. DEALING WITH DISPUTES procedural or legal mistake (especially if the agency has created its own procedures), a regulator knows that he "Nearly a decade after Beijing opened its doors to or she is more likely to win if the dispute goes to a tra- foreign investment in the power sector, the industry is ditional court. Regulators also know that a regular court littered with the remains of foreign investments will probably take several years to render a decision. So wrecked by renegotiated or reinterpreted pricing even if the decision is in favor of the licensee and agreements." against the regulator, it will effectively be in favor of the --Far Eastern Economic Review, January 31, 2002 regulator because the court decision may no longer be useful to the company by the time it is issued. "The inability of societies to develop effective, low- cost enforcement of contracts is the most important Given these realities, it is not surprising that international source of both historical stagnation and contempo- investors are becoming increasingly reluctant to invest in rary underdevelopment in the Third World." electricity distribution unless regulatory disputes are dealt --Douglas C. North, Nobel Laureate and author of with outside the regular court system. Many dispute reso- Institutions, Institutional Change and Economic lution mechanisms have been developed for infrastruc- Performance (1990) ture contracts where a government enterprise is the buyer. The question, then, is: Are these dispute resolution mechanisms equally applicable to regulatory disputes? As Business News Americas reported, "This is Chilectra's [a distribution company] fourth attempt to contract a 400­gigawatt-hour supply. 100 Previous attempts were abandoned because no bids were received from generators, who are reluctant to take on new commitments with distributors while they are under the obligation to compensate interruptions in supply whatever the reason, including during drought." Bnamericas.com, "Chilectra extends tender deadlines," April 12, 2002 In an exception to the rule, one former Sri Lankan regulator observed that "...appeals are necessary. It is a discipline that we need 101 because we are exercising discretion." Samarajiva (2001). B. What Gets Disputed? Given the inevitability of disputes, it critical that any attempt at regulation by contract be accompanied by a No regulatory contract can be totally clear or perfectly dispute resolution system that has the confidence of capable of anticipating future events. As one observer both government and investors.104 An effective dispute of U.S. regulation has commented, "at the edges of resolution system must include three principal elements: words there are always interpretations."102 And even if there are no ambiguities in the contract at the time it 1.The means of dispute resolution was signed, the two parties will never be able to predict 2.The types of relief that can be given (such as tariff everything that might happen. Or, in the worst case, adjustments, cost pass-throughs, damages, specific one of the two parties is no longer able willing or able changes to regulatory rules, injunctions and orders for to comply with the contract. It is thus inevitable that specific performance) there will be disputes between new private distribution 3.A mechanism to ensure that the decision will be hon- companies and the government or regulator even where ored in a timely manner.105 there has been a concerted effort to "contractualize" the entire tariff-setting system.103 C. Different Approaches to Dispute Resolution In countries that have recently privatized distribution, The various approaches to resolving disputes include 48 disputes have arisen over the following: the following: · The extent of pass-through of taxes (Georgia) · The local court system · The calculation of benchmarks for the pass-through · International arbitration of the costs of power purchases (Brazil) · Mediation · The reimbursement for lost revenues because of gov- · Expert panels ernment-mandated rationing (Brazil) · A specialized appeals tribunal. · The responsibility for blackouts and brownouts caused by non-performance of generators and system opera- 1. The Local Court System tors (Georgia) · The process for adjusting tariffs to reflect the cost of In an ideal world, any dispute between the licensee and new distribution investments (Georgia) the regulator would go to a local court that makes a · The disallowance of about 30 percent of the distribu- decision in "an informed, sophisticated and low-cost tion company's new investments (Moldova) way."106 In many developing countries, however, the · The methodology for calculating the asset base for local court system rarely exhibits these qualities. At best, setting tariffs in subsequent multi-year periods (Brazil) local courts are slow. At worst, they are corrupt. And it · The tightening of a loss-reduction target by the regu- is almost inevitable that they will have an inherent bias lator that had been contractually fixed in the tariff in favor of local interests, consumers and institutions. methodology specified in the privatization agreement (Moldova) In addition, traditional local courts are likely to exhibit · Whether rates of return in subsequent multi-year peri- two other weaknesses when confronted with regulatory ods should be calculated on a uniform or company- disputes. First, the courts may not be knowledgeable by-company basis (Colombia) about the technical, engineering and financial issues · The allowed technical and commercial loss levels and that underlie the disputes. Second, even if the judges allowed operating and maintenance costs in subse- happen to have the necessary knowledge, they may still quent multi-year periods (Colombia). be limited by law or precedent to examining whether See Howard (1996). 102 Our focus will be on disputes involving the regulator and the distribution company. Disputes between the distribution company and its 103 customers and between the distribution company and other industry participants will not be discussed. In the Lamech and Saeed (2003) survey, about 45 percent of the investors that responded said that "the fair adjudication of tariff adjust- 104 ments and disputes" was a critical to the success or failure of their power sector investment. A good general discussion of the strengths and weaknesses of various dispute resolution mechanisms can be found in Kerf (1998). 105 Williamson (1983). 106 there was compliance with the regulatory process rather to make decisions that are binding on the parties. than examining the substance of the decision.107 International investors almost always prefer international Generally, this means that the court's review will be limited (i.e., out-of-country) arbitration for disputes.109 to examining whether the regulator International arbitration usually takes place in another country but with an obligation placed upon local courts · acted outside the scope of its powers, to enforce the foreign award as provided for under the · did not follow the procedures that it was obliged to terms of an international treaty or convention. Private follow, investors prefer international arbitration because they · breached natural justice or acted unfairly, consider it to be neutral, there are good procedural · acted unreasonably, or rules for the parties to present their case, and it is likely · made an error of fact or law. to produce the fairest result (at least from their perspec- tive). If adopted, arbitration can be invoked at the initia- The propensity for deciding regulatory disputes on nar- tive of either party. It is increasingly the norm in infra- row legal or process grounds is not a phenomenon structure contracts in which a private foreign investor unique to developing countries. In both developed and builds or operates a facility that produces an output or developing countries, the traditional standard of judicial service purchased by a government-owned entity. review for appeals of a regulator's decision "does not allow the court to review the merits of the decision itself; a. Who Is Appointed and What Do They Do? 49 instead, the court can only review the merits of the man- ner in which the decision was made."108 The arbitrators (typically one or three) can be appointed by agreement between the parties or, if the parties can- Limiting a legal review to "process" is generally not very not agree, by an independent and neutral institution comforting to private investors. From an investor's per- such as the International Chamber of Commerce (ICC), spective, it is especially frustrating if a court upholds a the International Centre for Settlement of Investment regulator's decisions because the regulator followed all Disputes or a Bar Association. In some countries, an the correct procedures but the substantive decision was arbitration association has been established and funded patently absurd. And even if the investor "wins" because by all the players in that country for the express purpose the regulator made a procedural or legal mistake, a court of establishing a common set of rules for arbitration of is generally not allowed to change the regulator's initial disputes arising from contracts in that country and the decision. Instead, the court can only set aside the regula- appointment of suitably qualified arbitrators. tor's decision and order the regulator to reconsider the matter. So the "victory" for the private investor may be Panel members can also be appointed before any dis- more theoretical than real. Given these typical conditions, pute arises. For example, some contracts for major most private foreign investors understandably want access infrastructure development (such as IPP projects) have to a dispute resolution system that examines the sub- included provisions where a panel of three arbitrators stance of the regulator's decision more than the process who are well respected by the parties and who have by which it is arrived at. first-class international experience and reputations can be appointed before any dispute arises. The advantage 2. International Arbitration of creating a standing expert panel is that the panel can be used for other forms of dispute resolution that may The term arbitration usually refers to a dispute resolu- eliminate the need for formal arbitration.110 tion mechanism whereby the disputing parties submit their disputes to a non-judicial body that has the power For example, the Indian Supreme Court, in reviewing the decision of a state High Court, made the following observation: "All that the 107 High Court has to be satisfied with is that the Commission has followed the proper procedure and unless it can be demonstrated that its decisions is on the face of it arbitrary or illegal or contrary to the Act, the Court will not interfere." Supreme Court of India, (2002) 3 Supreme Court Cases 711, "Association of Industrial Electricity Users versus State of AP and Others," March 6, 2002. Lawrence (2002) and Green (1999). 108 For a detailed study of national or within-country arbitration applied to regulatory disputes involving water concessions in Manila, see 109 Houston and Bowley (2000). Another advantage is that the panel will be ready to hear disputes as soon as one party makes a formal notification. In the previously 110 cited Manila Water Concession arbitration system, it took six months to find three individuals to serve on the panel after the company filed a formal request for arbitration. See Houston and Bowley (2000), p. 20. In addition to adjudication of the dispute through formal may be necessary to modify the existing process rules to arbitration, the panel can also be used for a less formal provide the appropriate mechanics for an arbitration process designed to minimize the need for formal arbi- involving three or more parties. trations. For example, the panel can be asked to give a preliminary view or report on the facts as presented to c. Pendulum Arbitration them (see below). This finding of fact, because it is made by a panel of well respected and experienced individu- Some contracts contain what is called a "pendulum" or als, can be prepared quickly and is a powerful tool for "baseball" provision. This provision limits the scope of the parties to use to settle the dispute. In effect, the panel the arbitrators to impose their own solutions, which is available to serve two functions: first, as an expert fact- might be different from either of those sought by the finding panel and second, as an arbitration panel. contract parties.111 The pendulum provision limits the arbitrator to choosing a solution or relief sought by one Although this technique has been largely limited to of the parties. The arbitrator is not allowed to choose a infrastructure projects where a government entity is the solution of his or her own crafting. Pendulum arbitration buyer rather than the regulator, there is no obvious rea- is usually proposed to protect the parties from idiosyn- son why the same technique could not be used for reg- cratic decision-making by the arbitrator. It also promotes ulatory contracts. To make it workable for regulatory the prospects of an amicable settlement by forcing each contracts, it would be necessary, because of the dura- party to present a more reasonable position to the arbi- 50 tion of the regulatory contract, to include a mechanism trator. Where there is pendulum arbitration, the two dis- for replacing an arbitrator who, for whatever reason, puting parties will tend not to propose extreme positions became unavailable. In fact, recent proposals in several because doing so increases the risk that the arbitrator countries would create special electricity or infrastructure will choose the other party's position. Its principal disad- tribunals (see below) that operate outside of or as a vantage is that it can limit the arbitrator's ability to complement to regular courts. Such tribunals are, in make decisions that represent a fair compromise to the effect, an attempt to create something akin to a stand- interests of the parties and, perhaps, a better solution for ing panel of experts. There is one important difference the future, given the long-term nature of the contract. between the two proposals, however: whereas both the government and the distribution company would jointly Pendulum arbitration has been proposed in Chile for determine the members of the panel of experts, the gov- tariff disputes involving distribution utilities. At present, ernment alone would determine the membership of the the Chilean regulatory contract provides that distribution special tribunal. For this reason, most international tariffs are reset every four years based on a weighted investors will probably prefer an expert panel over an average of the tariff values proposed by consultants for appellate tribunal. the regulator and the distribution companies. Not sur- prisingly, this system has created incentives for the regu- b. The Arbitration Rules lator's consultant to produce a low number and the companies' consultant to produce a high number. With The arbitration rules are an important ingredient for inter- each successive tariff-setting exercise, the tariffs pro- national arbitration. Very few contract parties make up the posed by the two sides have come to diverge by ever rules and attach them to the agreement (although this was larger amounts. In an attempt to get more-reasonable done in the case of Channel Tunnel concession). Instead, estimates from the regulator and the companies, it has they usually incorporate well established international rules been proposed that the law be modified so that in the such as those developed by the International Chamber of future an arbitrator (perhaps taking the form of a spe- Commerce (ICC), the London Court of Arbitration (LCIA) cial economic tribunal) is required to choose only of the or the U.N. Commission on International Trade Law two presented values. (UNCITRAL). Although in a strict legal sense a regulatory contract may be signed by only two parties (the govern- d. Enforcement of Arbitration Awards ment and the company), consumers are obviously an important shadow party to the contract. Therefore, if inter- Enforcement will be a major concern with the design of national arbitration is included in a regulatory contract, it any international arbitration system for a regulatory con- It is described as baseball arbitration in the United States because it is often used in salary disputes between baseball players and the 111 baseball club's management. The arbitrator must select one of the two proposed salaries and is prohibited from choosing any other val- ues (such as splitting the difference in the proposed salaries). tract. Much will depend on the law of the jurisdiction in recognized that arbitration can be as lengthy as court which the arbitration award must be enforced. The New proceedings (although it may not involve waiting for York Convention on the Recognition and Enforcement of long periods before hearings can be scheduled). Finally, Foreign Arbitral Awards is designed to provide a mech- it can be as expensive as litigation (and sometimes anism for the automatic enforcement of arbitration more expensive), particularly if the case is complex. awards in the countries that are parties to the Convention. However, there are examples where local In our view, international arbitration is a necessary and laws have been in conflict (either inadvertently or delib- appropriate backstop for regulatory disputes in countries erately) with the letter or spirit of the convention, making with no track record for impartial resolution of such dis- it very difficult for international investors to enforce arbi- putes. That said, it is best held in reserve as a last resort tration awards. For example, many countries that are for dealing with disputes. Its principal value is derived signatories to international arbitration conventions from the simple fact that it exists, even if it is never used. reserve the right to limit "the application of an award to The mere fact that it is available to both parties will differences arising out of legal relationships that are often act as an inducement for the parties to settle. It considered as commercial under the national law of the becomes particularly effective if it is not a stand-alone enforcing state."112 Unless the initial agreement is drawn option, but is combined with other forms of dispute res- tightly, a regulatory commission whose decision is over- olution (e.g., an expert panel) that are faster and less turned or modified could, after the fact, argue that the costly to use. When international arbitration is packaged arbitration panel's decision is unenforceable because it with other forms of dispute resolution, smaller disputes 51 raises overriding constitutional or policy considerations. are less likely to grow into big disputes. e. Criticisms of International Arbitration 3. Mediation International arbitration under international rules has Another alternative to adjudication in a local court or been criticized on several grounds. Government officials international arbitration is Alternative Dispute Resolution in developing countries do not like international arbitra- (ADR). ADR has developed in many countries in order to tion because they view it as an affront to their national meet the criticisms of the length and cost of litigation dignity. In particular, they consider it as an attack on and arbitration and, perhaps, to a lesser extent, to meet their legal sovereignty. Some countries have directly or concerns over the unsatisfactory nature of the outcomes indirectly prohibited its use. Just as international in some cases. ADR typically involves the facilitation of investors believe that they will not receive impartial and structured efforts (e.g., expert panels and mediation) by timely justice in a local court system, government offi- the parties to settle dispute for themselves without going cials in developing countries believe (rightly or wrongly) to a local court. that international arbitration will always tend to favor the foreign private investors. As a consequence, it is not Mediation is the most common form of ADR. It involves surprising that they view "international arbitration as a the appointment of an experienced mediator or foreign institution imposed upon them with a heavy mediation team that carries out a process designed to Western bias."113 Nevertheless, international arbitration enable the parties to better understand each other's has become widely used because it is the quid pro quo concerns and positions and to negotiate a settlement for access to international financing. or agreement on a mutual-gains basis. The mediator may be empowered or mandated to seek a negotiated But even practitioners of international arbitration recog- outcome that is acceptable to the parties. When it nize that it has problems. They will usually admit that works well, mediation or any other form of voluntary arbitration is often no simpler than litigation because it ADR can produce a better answer for the contract requires rules of procedure to cover all eventualities and parties, if they take full ownership. It has been success- to ensure that the proceedings are fair, to allow the par- fully used in regulatory adjudication and rule-making ties to state their cases fully and to be heard. It is also proceedings in the U.S.114 Houston and Bowley (2000), Appendix, p. 1. 112 Schwartz and Paulson (1999). 113 Raab (1994). These proceedings have generally been used to negotiate a prospective agreement between a company and its customers 114 or to develop a consensus on a proposed new rule covering technical issues for which the regulator may have limited expertise. If the regulator agrees with the negotiated "settlement," he may formally approve it. The U.S. FERC has made extensive use of ADR for these types of proceedings. See www.ferc.fed.us/public/adr.pdf. In general, ADR in the United States has not been used to mediate disputes between the regulator and an affected company once the regulator has issued a formal order on the matter. However, because of its informal nature, the mediation convened, this will not accomplish very much if the reg- process does not impose any resolution or decision ulator is an unwilling party to the mediation. A regulato- upon the parties. There will be no binding resolution of ry dispute is different from a commercial dispute the dispute unless and until an agreement is reached because in a commercial dispute both the buyer and and committed to writing at the end of the ADR process. seller will usually have an economic incentive to resolve Moreover, a party is free to walk away from the process the dispute. This is not true for regulatory disputes. In at any time up to that point. The success of ADR thus general, regulators will have no incentive to agree to depends on the willingness of the parties to make the mediation or other forms of voluntary ADR. Although a process work, or on their being persuaded, often by the government could try to pressure the regulator to enter mediation team, that the process could or should be into negotiation, such pressures could put the govern- made to work. Where there is more than one party to ment in the awkward position of being seen as trying to the dispute or the issues are complex, this can be a par- compromise the regulator's independence. ticular challenge. 4. Expert Panels Any settlement that is reached is, by definition, outside of the contract terms. The question therefore arises of Expert panels have been used as a means of resolving how such an agreement can be enforced. The regulato- disputes, on either an interim or a final basis, in some 52 ry contract or the enabling legislation could provide that infrastructure contracts. The members of the panel are all settlements reached as a result of mediation or some chosen for their experience and understanding of the other form of ADR be subject to revisions to the contract issues. The perceived advantage of this procedure is that, that will be fully implemented by the parties. However, because it operates outside the regime governing the this may deter the parties from using ADR (if they have conduct of arbitrations or litigation under the rules of a the choice). ADR works best where parties enter into the court, it is possible for the parties to the contract to process willingly in the hope that they will reach a settle- empower the expert to reach a rapid decision. The expert ment. panel adopts an inquisitorial role in investigating the facts and the law regarding the dispute, often with a wide dis- Several distribution concessions in Brazil contain provi- cretion as to how to go about the task. The expert panel sions for mediation of disputes with the regulator. For is usually required to make its decision in a matter of example, the concession agreement for AES Sul pro- weeks rather than months or years. The result may be vides for the establishment of a committee of three spe- "rough justice" but it may be preferable to the uncertain- cialists with the responsibility for "suggesting...[a] nego- ties and expense that can be created by delays commonly tiated solution for the conflict."115 There is, however, a associated with court proceedings and arbitrations. fundamental problem with the Brazilian mediation approach. In general, the mediation panel can be con- Some infrastructure contracts provide for disputes to be vened only if the two parties, the regulator and the referred to experts as part of the overall dispute resolu- aggrieved distribution company, both agree to convene tion process. The experts' decisions can be binding or the panel. But it is highly unlikely that the regulator will non-binding. If they are non-binding they can still be voluntarily agree to enter into dispute resolution because persuasive in encouraging the parties to settle the dis- this would be equivalent to admitting that he or she may pute.116 However, unlike arbitration, expert determination have made a mistake. Regulators, like most human does not normally have any statutory support or sanc- beings, are generally reluctant to admit to mistakes-- tion and there are no international conventions relating especially if it weakens the regulator's legal position if to the enforcement of expert determinations. there is a later formal appeal to a regular court. Several changes would be required to adopt expert In Brazil, the panels are further limited by the fact that panels for regulatory contracts. The most obvious is that they can only "suggest" (i.e., mediate) a solution. Even the distributor must have the unilateral right to convene if a distribution company could require that a panel be the panel. If the panel can be convened only with the Distribution Concession Agreement 12/97, Article 15. 115 At the U.S. FERC, this technique is referred to as "Early Neutral Evaluation." It is defined as "an early and frank evaluation by an objective 116 observer or `evaluator.'" It has been used to encourage the settlement of disputes between regulated utilities and one or more of its cus- tomers. It has not been used for resolving disagreements between the regulated utility and the FERC. If a utility disagrees with a commission order, its only redress is to file an appeal in an appellate court. See FERC ADR News, Summer 2002, p. 1. Available at www.ferc.fed.us. agreement of both parties, it is unlikely that the regula- a specialized court that has operated successfully in the tor would ever agree to convene the panel for the rea- Indian telecommunications sector.117 sons just discussed. In addition, the regulatory contract must provide that the parties will comply with the What are the typical features of such tribunals? The tri- experts' determinations. bunal usually hears appeals of regulators' decisions in one or more infrastructure sectors. It will usually have However, a further procedure will be required, either considerable discretion over the scope of its actions. For through the local courts or through local or international example, in the United Kingdom the Competition arbitration, to provide an effective mechanism for enforc- Commission, which serves as the appellate tribunal for ing the experts' decision. In some jurisdictions, the law some of the regulators, can, in some instances, sup- may not support expert determination on the grounds that plant a regulator's decision and replace it with its own it illegitimately excludes or restricts the jurisdiction of the decision.118 In other countries, however, the tribunal is courts or that it offends against strict legal due process limited to approving or disapproving the regulator's requirements or the principle that a person is entitled to a decision.119 The right of appeal may be granted to the full trial or hearing of any matter affecting his or her civil company, the regulator and the company's con- rights and liabilities. If binding expert determination is to sumers.120 The tribunal may be a standing body (India) be adopted as the means of dispute resolution in the reg- or simply convened on an "as needed" basis (United ulatory contract, these issues would need to be addressed Kingdom) from a pre-selected group of more than 20 53 in any enabling legislation. experts.121 It will normally be composed of technical experts as well as lawyers. As a quasi-judicial entity, it 5. A Specialized Appeals Tribunal has considerable discretion in creating its own proce- dures and processes. In particular, it does not have to A fifth approach is a specialized appellate tribunal to follow the strict rules and procedures of a regular court adjudicate disputes between the regulator and the distri- but it does need a set of transparent and fair rules to bution company. A distribution company with a griev- ensure that the parties can present their cases fully. The ance against its regulator would not need to get the tribunal's decisions may be appealable to the country's concurrence of the regulator in order to file a complaint supreme court (India) or to international arbitration with the specialized tribunal. Specialized tribunals have (Uganda). But if there is an appeal of the tribunal's been created in countries as varied as England, decisions, the appeals are generally limited to narrow Australia, Uganda, Tanzania and Bolivia. In Chile, a procedural issues rather than a complete de novo special presidential commission recommended the cre- review of the substance of the dispute. ation of a multi-sectoral "economic court" as a critical "second generation reform" for the country's infrastruc- To date, the real-world experience with special appellate ture industries. It has also been proposed in a new tribunals in most countries has generally been positive. national electricity law for India based on the model of Among the observed benefits are the following: See Sections 110-125 of Electricity Bill 2001 which is available on the website of the Indian Ministry of Power 117 (http://powermin.nic.in/report/mopi_opt4.pdf ). The Supreme Court of India, in a recent order overturning the decision of a state High Court, strongly recommended that future appeals of state electricity regulatory commissions be reviewed by a special national electricity tribunal rather than the state High Court. The Supreme Court wrote: "Without meaning any disrespect to the Judges of the High Court, we think nei- ther the High Court nor the Supreme Court would in reality be appropriate appellate forums in dealing with this type of factual and technical matters." Judgement and Order of the Hon'ble Supreme Court, Re: CESC Tariff, Civil Appeal No. 4037 of 2002, October 3, 2002. See Lawrence (2002), p. 33. 118 An appellate tribunal reviews the regulator's decisions and essentially says "yes" or "no" to the decision. If the answer is "no", then the tribunal 119 will normally tell the regulator why it thinks that the decision was wrong on technical, legal or both grounds. In contrast, the Competition Commission in the United Kingdom has authority that goes beyond "yes" or "no" decisions. It can say "yes" or "no" or change the regulator's decision for some of the new infrastructure regulators. Because it has been given this wider authority, it is, in effect, the functional equivalent of a higher level regulator for these industries rather than just an appellate body. It seems counter intuitive that the regulator would have to seek review of its own decisions. But this is effectively the case for the British 120 electricity regulator (OFGEM) because it is not allowed to make any changes to a distribution company's existing tariff formulas unless the company agrees. If the company withholds its approval, OFGEM can mandate the tariff change only if it receives formal approval from the Competition Commission. In India, the current proposal for a national electricity appellate tribunal would also grant the right of appeal to individual electricity consumers. Given the propensity of Indian consumers to sue, the granting of this right could easily overwhelm the proposed tribunal. In contrast, distribution companies in England are required to establish a separate dispute resolution mechanism for consumer complaints and the right to go to the Competition Commission is granted only to the regulator and licensees. In the United Kingdom, although the Competition Commission has no standing panels, its secretariat is a full time and permanent organization. 121 · Special tribunals produce faster decisions. For sense to put reform of the power sector "on hold" example, in India the Telecom Disputes Settlement pending an overhaul of the entire judicial system. As and Appellate Tribunal has usually issued decisions in one former chairman of the U.S. Federal Energy six to eight months as opposed to an average of three Regulatory Commission (FERC) observed: "If you can't to four years for the Indian Supreme Court. However, reform it, then just bypass it." this is not true in all countries. The UK Competition Commission seems to average about 9-12 months for D. Partial Risk Guarantees: A Mechanism for some of its price reviews. But this may reflect the fact Ensuring Commitment? that UK commission is essentially rerunning the basic price review in some infrastructure industries rather Although a dispute resolution mechanism may look fair than simply functioning as appellate review body. and efficient on paper, it will be of little value if the gov- · The tribunals are likely to produce more- ernment or regulator refuses to implement the decision informed decisions. Many of the disputes between produced by the court, arbitrator or a special appellate the regulator and a distribution company will be tech- tribunal. Therefore, investors must have confidence that nical or economic in nature. The judges in most regu- governments or regulators will comply with these deci- lar courts, whether in developing or developed coun- sions. Otherwise the regulatory contract, no matter how tries, generally lack the background to deal with com- well designed, will have little credibility. 54 plex technical and economic disputes. Without such knowledge, they tend to make decisions on narrow It has been suggested that some additional financial legal grounds. And even if they have the requisite mechanism is needed to induce a government or regu- technical and commercial knowledge, existing laws or lator to honor the regulatory contract. One recent pro- legal precedents may preclude them from going posal is to adapt an existing World Bank financing beyond a review of whether the regulatory commis- instrument known as a "partial risk guarantee" (PRG) to sion followed appropriate procedures. In contrast, provide a backstop to the regulatory system.122 The PRG special tribunals are specifically required to examine would be written to guarantee scheduled payments of whether the regulator complied with substance (for principal and interest payments on debt if the private example, whether the regulator implemented the tar- investor defaults on or delays payments because the iff-setting formula) as well as process. regulator fails to honor the terms of the regulatory contract. · An appellate tribunal will be familiar with the industry and may have heard similar disputes in 1. How Would It Work? other infrastructure industries if it has jurisdiction over several industries. For example, some mem- If a private distributor believes that the regulator or the bers of the UK Competition Commission have government has not complied with the provisions of the reviewed several price determinations by British regu- regulatory contract, it could initiate a claim under the lators. In contrast, a judge in a regular court is not dispute resolution mechanism provided for in the regu- likely to be familiar with the technical or institutional latory contract. If the dispute resolution entity agrees characteristics of the industry and, if he is, there is no with the distribution company, it would order the regula- certainty that he will be assigned to hearing the dispute. tor to take the regulatory action (e.g., increase tariffs or · The use of special tribunals may avoid the allow the pass-through of a power-purchase cost) expense and delay inherent in international arbi- required by the regulatory contract. If the regulator tration. Special tribunals can be used for more com- refuses to comply with the panel's decision, the World mon disputes that, if unresolved, could lead to the col- Bank would pay holders of the guaranteed debt for any lapse of the regulatory contract. The existence of a spe- losses in principal or interest payments that result from cial tribunal may eliminate the need to go to interna- the failure of the regulator to comply. The World Bank tional dispute resolution. would, in turn, require a counter-guarantee from the · The use of special tribunals precludes the need government in the form of an "indemnity agreement" to try to reform the existing court system. that would obligate the government to reimburse the Although such reforms may be desirable, it makes little Bank for any guarantee payments that the Bank makes See Gupta et al. (2002). A second possible new use for PRGs would be to guarantee the delivery of promised subsidy payment. For example, 122 the Delhi Government recently promised that the three new private distribution companies that it would subsidize their bulk power purchases up to a ceiling of close to USD $700 million for the first five years after privatization. Such subsidies could be backed up by a guarantee. to lenders. In effect, the Bank would be guaranteeing a regulatory system that it created is operating as intend- payment to holders of debt if the regulator fails to honor ed. Moreover, it may simply be in the government's the outcome of the dispute resolution process. financial interest to offer the guarantee. 2. Why Do It? 3. Is It Feasible? The PRG would be a form of insurance to protect This leads us to a more difficult question: Is it realistic to against the risk that the regulator will not live up to the expect that PRGs can be used to guarantee the per- terms of the regulatory contract. A government official formance of a multi-dimensional regulatory system? To might reasonably ask: Why should the government pay date, PRGs have never been used for this purpose, for insurance to protect private investors against the though they have been used in other contexts. For consequences of the regulator's actions? example, the World Bank has issued PRGs to support several privately financed IPPs projects.124 In these There are at least two answers to this question. First, such instances, the Bank issued PRGs to guarantee that a a guarantee is likely to generate more investor interest in government-owned buyer would make payments for purchasing the state-owned distribution enterprise. In power purchased under the terms of power purchase some developing countries, it may make the difference agreement signed with a private developer. In these between getting some bidders versus getting no bidders. cases, the trigger for payments under the PRG was rela- 55 Second, the guarantee is likely to produce higher prices tively straightforward: either the government-owned for the assets that the government is selling. Investors are buyer made payments for the electricity supplied or it did not. willing to pay more when they see less risk. They will also be willing to pay more because the PRG is likely to lower Guaranteeing the performance of a multi-faceted regu- their financing costs. Past experience shows that PRGs latory system is clearly more complicated because regu- allow investors to get debt financing when it might not oth- latory systems are defined by both substance and erwise be available or allows them to get lower interest process. For example, a regulator may decide not to rates on any debt that they acquire. So it may be in the allow a distribution company to raise tariffs to pay for government's self interest to "buy" this insurance if the additional investments, even though the terms of the benefits--getting serious bidders and a higher price from multi-year tariff formula provide for an increase when bidders--exceed the costs--the charge for the premium such investments are made. Rather than directly and and the possible payment to the World Bank to compen- openly violate the terms of the tariff formula, the regula- sate it for any payments to debt holders.123 tor might argue that it cannot raise tariffs because the distribution company did not provide adequate and The same government official might also ask a second timely information to allow it to determine if the invest- question: Why should the government be held responsi- ments were prudent or fell within a pre-approved cate- ble (that is, be forced to make a guarantee payment) for gory. In this situation, the regulator may be hiding the actions or decisions of an independent regulator? behind regulatory processes to avoid raising tariffs. After all, if the regulator is truly independent, it seems Unless the regulatory contract is clear on both process contradictory to argue that the government should still (such as maximum time before the regulator must make held accountable for the regulator's actions. a decision and information that must be provided) and substance (such as the tariff increase that results from a Although this argument seems superficially plausible, it given level of investments), it will be difficult for an arbi- ignores the fact that the regulator is a creature of the trator or specialized court to determine whether the con- government because government established the regu- tract has been violated. lator in the first place and created the regulatory system administered by the regulator. So the government The PRG will be difficult to enforce if the regulatory con- should ultimately be willing to take responsibility that the tract is vague. This is a problem in the tariff sections of A government could determine these costs and benefits by seeking bids with and without the guarantees to see the effect of the guarantees 123 on offer prices. See World Bank (April 2002). PRGs have also been used by MIGA to insure against other forms of political risk such as changes in laws, 124 outbreak of war, breach and repudiation of contracts, expropriation and access to currency convertibility. Information on guarantees is avail- able at www.miga.org and www.worldbank.org/guarantees. For an overview of both World Bank and non-World Bank guarantee instru- ments, see Bubnova (1999). several of the concession agreements for private distri- 7. The heart of a regulatory contract is a pre-specified, bution companies in Brazil. For example, in describing performance based, multi-year tariff-setting system. how the regulator will reset tariffs at the end of the first 8. A regulatory contract is sustainable only if the under multi-year tariff period, the only guidance, as discussed lying economics are viable. earlier, is a single sentence that states that the Brazilian 9. A multi-year tariff system can be put into operation regulator even in the absence of high-quality data. 10.Regulation by contract should be reserved for private shall process the revision of the amounts of rates for distribution companies. commercialization of power, altering them upwards or downwards, taking into account the cost and market 1. Independence is not enough. structures of the concessionaire, the levels of rates practiced by similar companies in the nationwide and · Over the last decade, regulatory "independence" has international context, and stimuli for efficiency and for been a key element of the recipe for successful power reasonableness of rates.125 sector reform recommended by the World Bank and other development organizations. The rationale for The vagueness of this sentence has led to major battles the recommendation was the belief that reform would in Brazil over its interpretation as the first tariff period is fail (especially if it involved privatization) unless tariff 56 coming to a close for more than 50 Brazilian distribu- setting is depoliticized. It was thought that an inde- tion companies.126 The obvious lesson from the Brazilian pendent regulator--a regulator that is free to make experience is that the viability of a creating a PRG for a decisions on tariffs and other regulatory matters with- regulatory system will depend critically on the specificity out first obtaining the approval of political authori- of the language in the underlying regulatory contract. ties--would do a better job of setting cost-recovering tariffs than a government ministry facing day-to-day political pressures. 7. CONCLUDING OBSERVATIONS · Now, with the benefit of more than 10 years of expe- "Investors need confidence. Consumers need pro- rience, it is clear that independence is not enough. tection." The reality in many countries is that independence has --Speaker at a conference on private participation never been achieved. Regulatory independence has in Chinese infrastructure, November 2001 been more theoretical than real. Despite the many safeguards that exist in new laws, it has been difficult The key lessons learned from the experience of develop- to protect new commissions from direct or indirect ing and developed countries with regard to regulation political pressures to avoid actions that a government by contract are the following thinks will be politically damaging. Most new regula- tors in developing countries, when asked whether they 1. Independence is not enough. have the independence provided for in law, will either 2. The regulatory contract must be a political contract. say "no" or avert their glance and change the subject. 3. Regulation by contract versus regulation by · But even in those countries where effective independ- commission is a false dichotomy. ent regulatory decision-making has been achieved, 4. Regulation by contract is a new name for an old commissions are not likely to follow policies that bal- paradigm. ance consumer and private investor interests because: 5. Electricity consumers cannot be the forgotten third party to a regulatory contract. · Starting conditions are bad. The typical starting 6. Investors must have confidence that the contract will conditions for a regulator in a developing country be enforced fairly and efficiently. are that: tariffs do not cover costs; some customers Concession Agreement for AES Sul, Article 7, Sub-Article 6, November 6, 1997. 125 For example, ANEEL and the distribution companies have a major disagreement over the allowed capital base that will be used to determine 126 allowed revenues in future multi-year tariff periods. ANEEL has proposed a methodology which the distribution companies argue would pro- duce a regulatory capital base of US$4 billion for 16 of the largest distribution companies, while the companies contend that they are entitled to a regulatory capital base of about US$17 billion. Although there are also disputes over asset valuation in other Latin American countries such as Chile and Peru, the disputes are over the details of implementing a particular asset valuation methodology rather than the threshold issue of which of several methodologies should be used. (usually industrial users) heavily subsidize the con- tradition of detailed regulatory contracts. In other sumption of other customers (usually residential words, the single most important lesson of the last users); interruptions are frequent; electricity is widely ten years is that independence must be "back- stolen by rich and poor customers (frequently with stopped" by a regulatory agreement that goes the active support of existing employees); and many beyond general principles. And the key element of rural citizens lack access to electricity. Given these the regulatory contract should be a well-specified, starting conditions, it is unrealistic to expect that an multi-year tariff-setting system that is required by independent regulatory commission will be able to law and specified in concessions, licenses and other close the gap between revenues and costs and regulatory instruments (hard law). rebalance tariffs across classes under the guise of simply making some technical tariff adjustments. 2. The regulatory contract must be a political The reality is that the government that created the contract. commission has been running away or hiding from both problems for many years. · A regulatory contract must inevitably be a political · Transitions take longer than expected. contract between the government and the new private Consumers expect much more of private companies owner of the distribution enterprise because the com- than state-owned enterprises. Consumers under- mitment will not be believable unless it is an explicit standably lose patience if tariffs go up immediately commitment of the country's highest political authori- 57 but service improvements lag behind. When this ties. It is naïve to believe that the design of the regu- happens, the regulators get blamed. Therefore, it is latory contract can or even should be depoliticized. not surprising that most regulators, when faced with · However, even though the regulatory contract should this situation, will try to find a way not to raise tariffs, be a political contract, the political authorities that especially if their legal mandate consists of nothing created the contract need not also be responsible for more than principles, goals and objectives (soft law). implementing it on a day-to-day basis. Once the reg- · The regulator cannot do it alone. It is relatively ulatory contract is in place (i.e., the political deal has easy to pass a reform law but the commercial via- been made), the contract is best administered by an bility of a distribution company depends on the independent regulator. This makes it difficult for the government paying its electricity bills, providing government to change its mind after it agrees to the basic law and order so that the company can col- contract. The temptation for a government to renege lect from non-paying customers and, in some on a regulatory contract, either openly or in hidden cases, providing subsidies for a transition period. If ways, will be very high if the transition is not smooth the government is unwilling or unable to take these (and transitions, almost by definition, are never actions, there is little that a regulator can do to cre- smooth). Also, it is unrealistic to expect that a future ate an economically viable enterprise. The fact that government, especially if it is from the opposition the regulator is independent is largely irrelevant. party, will be equally committed to honoring the regu- · Foreign ownership is often viewed as a new latory contract. So it is best to create a regulatory form of colonialism. If tariffs are not raised prior to contract so that a new government can credibly say: privatization, they will have to be raised soon after "It is beyond my control." privatization. Most electricity consumers in countries · A law is the highest expression of a country's political that are former colonies will view this as exploitation commitment. Therefore, the underlying principles and unless there is an obvious and significant improve- initial parameters of the regulatory contract should be ment in quality of service. The situation is especially clearly specified in the country's primary or secondary sensitive if the new owner is a foreign company. electricity laws (Chile, Argentina, Bolivia and Peru). A · The overall experience of the past decade sug- regulatory contract is less likely to survive if it is poorly gests that independence is not enough. specified (Brazil) or exists only within a stand-alone Independence must also be combined with an concession or license agreement with little clear sup- explicit and binding regulatory contract established port in national laws (Brazil). prior to privatization. For most developing coun- · A regulatory contract will not be credible if interna- tries, the Anglo-American tradition of regulatory tional investors believe that its multi-year, tariff-setting independence must be combined with the French provisions are vulnerable to legal challenge because such a tariff-setting system was not anticipated when able to withstand the open or hidden pressure to keep the law was written (as is the case in India). The regu- tariffs low through a variety of techniques (e.g., latory contract will also not be credible unless the process delays, setting allowed revenues based on government is empowered to design the contract and impossible to achieve efficiency targets). If the regula- the regulator is empowered to enforce it. These are tory system in a developing country consists of an unnecessary legal risks that usually can be eliminated independent regulator who operates under broadly by amending existing laws. worded tariff principles, foreign investors will not · The regulatory contract will just be "pretty poetry" invest. unless the government also takes actions to support · The better alternative is to create an independent reg- the contract. The single best test of a government's ulator who sets prices under a well-specified, pre- ongoing political commitment is the day-to-day sup- determined, multi-year tariff-setting system (Bolivia, port that it provides to the distribution company to Chile and Peru). "Well-defined" and "pre-determined" reduce theft. A government can publicly demonstrate mean that specific tariff formulas and target parame- its commitment to "law and order" through the pas- ters are established as part of the privatization sage and enforcement of anti-theft legislation that process. In order to reduce the risk of revenue uncer- allows for disconnection and prosecution of those tainty to a manageable level for investors, the regula- who steal electricity. But the commitment requires tor must have little discretion on both substance and 58 more than just passing a law. A good sign of serious process during the initial post-privatization, tariff-set- political commitment is if the government successfully ting period. In this critical first tariff period, the regula- prosecutes one or two rich or politically well-connect- tor needs to act more as a "referee" than as a ed individuals that have been stealing electricity. "judge." The regulator's role in the first period should Another sign is if the government pays its own elec- be limited to making certain that the tariff formula in tricity bills. If a government is unwilling to take such the regulator contract is correctly applied and that steps, there is little point in trying to create a regulato- true-ups and pre-scheduled tariff adjustments are ry contract. processed in timely way. · Transparency works. Even if a regulator's legal 3. "Regulation by contract versus regulation by authority is limited to making recommendations to a commission" is a false dichotomy. minister, the minister will find it difficult to overturn the regulator's recommendations if they are made publicly · The real choice is between an independent regulator and are backed up with a clear technical analysis. with substantial discretion and an independent regula- tor with little discretion, especially in the first post-pri- 4. Regulation by contract is a new name for an vatization tariff period. old paradigm. · An independent regulator with substantial discretion is one who sets prices under general, broadly worded · Regulation by contract--an independent or quasi- tariff-setting principles (India). This is also the U.S. sys- independent regulator administering a well-specified tem, where the tariff guidance in the law is often limit- tariff-setting system that is embedded in laws, conces- ed to a general statement that tariffs must be "just sions and regulations--has been the norm for distri- and reasonable and not unduly discriminatory." This bution privatizations throughout Latin America for system has worked well in the United States because more than 15 years. of 70 years of legal precedents as to what the princi- · Regulation by contract is more difficult for distribution ples mean in application combined with an independ- than generation because the distribution business has ent judiciary to ensure that the precedents are not many more strands to it that affect the quality of serv- ignored. When this regulatory system is transplanted ice to thousands of customers and the need for sub- to a developing country, it usually will not work for stantial ongoing investments. Even so, such contracts several reasons: there are no legal precedents (or the have been written and honored in several Latin precedents may have little or no legal significance if it American countries for more than a decade. is a civil law country); the country's judiciary is neither · Process is important. The single most common mis- independent or knowledgeable; and the starting con- take in writing regulatory contracts is to focus on the ditions are so bad that that the regulator will not be tariff-setting principles and formulas while ignoring the regulatory processes needed to implement the 6. Investors must have confidence that the con- principles and formulas. With the exception of Brazil, tract will be enforced fairly and efficiently. the Latin American regulatory contracts for distribution have generally succeeded because they have speci- · Contract disputes are inevitable because no regulato- fied principles, formulas and processes. ry contract can envisage all eventualities. · Regulation by contract will not survive a major macro- · In most developing countries, the existing court system economic "meltdown" (Argentina), but neither would is not a viable dispute resolution mechanism for dis- any other regulatory system. putes between the regulator and the company. · It is not feasible to write a regulatory contract for all Existing courts are slow, not likely to have the expert- regulatory decisions. For example, if the government ise to deal with complicated technical and economic decides to unbundle the sector (i.e., create separate disputes, and may be corrupt. generation, transmission and distribution entities), this · International arbitration of regulatory disputes under will inevitably lead to the creation of a centralized or international rules will make a proposed distribution decentralized bulk power market. If the regulator is privatization much more appealing to international charged with monitoring the market, he or she will investors. However, international arbitration is usually need considerable discretion in deciding the informa- expensive and cumbersome. If it is invoked, it usually tion that is needed and the actions that should be means that the regulatory contract is dangerously taken if market participants are found to have market close to complete collapse. It is preferable to combine 59 power. Therefore, a single regulator may need consid- the "backstop" of international arbitration with less erable discretion for some decisions and little discre- costly forms of dispute resolution that can prevent one tion for others. or more smaller disputes from exploding into a big dispute. 5. Electricity consumers cannot be the forgotten · Some countries have had success in creating standing third party to a regulatory contract. expert panels that can be used not only to adjudicate disputes through formal arbitration but also to give a · Although the government or the regulator signs the preliminary view or report on the facts of the dispute regulatory contract, it must not forget that it is acting as presented to them. These findings of fact, because as an agent for consumers. they are made by a panel or well-respected and expe- · The government and regulator must be able to per- rienced individuals, can be prepared quickly and are suade consumers that the initial contract is fair and a powerful tool for the parties to use to settle the dis- that it will be fairly enforced. pute. If such a panel is created, it effectively performs · If the contract leads to higher prices, consumers must two functions: first, as an expert fact-finding panel see improvements in service sooner rather than later. and second, as an arbitration panel. · If consumers fail to see any obvious benefits from the · An expert panel will not be viable if it can be con- regulatory contract (i.e., "early wins"), it will be politi- vened only with the agreement of the regulator cally unsustainable. (Brazil). It is unrealistic to expect that the regulator · Consumers are often willing to pay substantially more would voluntarily agree to agree such a panel for reliable electricity because they recognize that because it is tantamount to admission by the regula- cheap electricity that does not arrive has no value. tor that his decision may have been flawed. If an · The key task for the government and the regulator is expert panel is to be used, the licensee must have the to ensure that there is a fair balance between the legal right to require that it be convened. interests of the consumer and the interests of the · Mediation may be a viable mechanism for disputes investor. This involves attracting the investment that involving the regulated company and its customers. It the consumer needs but at the same time minimizing may also be a workable for designing new rules. the need for new investment by using existing However, it is not likely to be a viable mechanism for resources more efficiently. disputes between a distribution company and the reg- ulator once the regulator has issued a formal deci- sion. In such circumstances, the regulator will have lit- tle or no incentive to enter into mediation. Nor does it make sense to mandate that the regulator enter into mediation when requested by the company. Although · Benchmarks should be used for establishing a maxi- it may be possible to force the regulator to sit in a mum pass-through price for power purchases. A well- mediation room, it will be a wasted effort because the designed benchmark for power purchases should pos- regulator will still have no incentive to negotiate in sess three characteristics. First, it should not be set too good faith. low. If the benchmark is too low (Chile and Brazil), · Another approach, adopted in several developing and generators will be unwilling to sign contracts with dis- developed countries, is to give the distribution compa- tributors for new capacity and this will eventually lead ny the right to appeal the regulator's decisions to a to shortages. If feasible, the benchmark should be specialized appellate tribunal that reviews decisions of based on market measures (Colombia and the one or all infrastructure regulators. The tribunal Netherlands) rather than administrative projections of should be explicitly authorized to examine the sub- supply costs (Brazil and California). Second, any mar- stantive as well as the legal aspects of the dispute. ket benchmark should not be based solely on the · The World Bank should explore the feasibility of back- price of spot energy (Argentina). It must be defined stopping regulatory contracts by guaranteeing that the more broadly so that a distribution company will have outcome of the dispute resolution mechanism will be incentives to make short, intermediate and long pur- honored. chases that allow it to hedge for the risk of future price fluctuations (Colombia and the Netherlands). 60 7. The heart of a regulatory contract is a pre- Third, distributors must have reasonable opportunity specified, performance-based, multi-year tariff- to make profits on purchases so that they will have an setting system. incentive to purchase efficiently. · Pass-through of non-controllable costs should be · The multi-year tariff system should include bench- done frequently and automatically. The tariff-setting marks or targets for controllable costs and automatic system must also include a mechanism for the pass- pass-through for non-controllable costs. The catego- through of costs associated with unanticipated exter- rization of costs may change over time as industry nal events such as natural disasters or major changes structure changes. However, the regulator should be in law, regulations and some taxes. Whenever possi- allowed to change the tariff treatment of a particular ble, the regulatory contract should include specific cost only at the end of a multi-year tariff period. "trigger" mechanisms to adjust tariffs for extraordinary · The benchmarks for controllable costs can be based events. In developing countries, the civil law concept on the company's historic performance, an external of restoring the enterprise's "financial-economic equi- index or the performance of comparable companies. librium" is not a workable approach for dealing with The benchmarks must be credible and set achievable extraordinary events. targets or they will be ineffective as incentives to · The tariff-setting system must also pay particular change behavior. In establishing a benchmark, the attention to the regulatory review of new investments. tariff formula should distinguish whether price, quanti- A regulatory system that requires an investment-by- ty or both elements are under the control of the com- investment review, either on an ex ante or ex post pany. basis, is not workable. The better approach is to set · Benchmarks or targets should be combined with quality-of-service standards (performance regulation) incentive mechanisms. Such incentive mechanisms rather than engaging in regulatory reviews of individ- should be limited to costs or operating parameters ual investments (conduct regulation). that are measurable, material, controllable and pre- · Any tariff-setting system should explicitly assign risk to dictable. At any one time, there should be no more the company, its customers or the government. Private than 3 or 4 incentive mechanisms in operation. investors should only be assigned risks that they can · The two most important benchmarks in many devel- control or mitigate. Private investors must be allowed oping and former socialist countries are the technical to recover the costs of any risks that they bear for their and commercial loss-reduction targets and the price customers. The allocation of risk will need to vary paid for discretionary power purchases. The financial depending on starting conditions. The worse the start- viability of a new private distribution entity will depend ing conditions, the greater the risk that must remain critically on its ability to reduce commercial losses with government or consumers. (i.e., theft). · The regulatory contract cannot be limited to a single · When the revenue-cost gap is large, the principal multi-year tariff period following privatization. At a purpose of the subsidies is to prevent rate shock minimum, the actual formulas should be specified for rather than to subsidize the private company. the first period and the principles and general · The World Bank should consider giving loans for tran- methodologies should be specified for subsequent tar- sition period subsidies combined with guarantee iff periods. mechanisms to ensure that the subsidies are actually · In countries with more than one distribution company, delivered to the intended recipients. the regulator should require the companies to provide comparable data on costs and operating perform- 9. A multi-year tariff system can be put into oper- ance that can be used to establish benchmarks. ation even in the absence of high-quality data. 8. A regulatory contract is sustainable only if the · The existing quality of cost and technical information underlying economics are viable. is usually poor for state-owned enterprises in develop- ing and former socialist countries. · Regulation by contract is not a "magic bullet." It will not · The introduction of performance-based multi-year tar- work if there is a large gap between costs and revenues. iffs should not be delayed until the data "get better." · The gap must be closed by lowering costs (for exam- · Data quality will improve through privatization, espe- ple, reducing technical and commercial losses that cially if better data can be specified as a performance 61 lead to unnecessary power purchases), increasing rev- element in the regulatory contract. enues (for example, metering and billing farmers in · If there is a political concern that investors will be India for the power that they consume), or both. able to earn high profits because of poor data quality, · International and domestic investors will incur some then the tariff system should include a within­tariff- costs in hard currencies. Even if the domestic tariffs period "sharing" mechanism between the distribution are initially high enough to cover these non-domestic companies and its customers. Revenue-sharing mech- costs, this can quickly change if the local currency anisms are easier to implement and less vulnerable to loses value relative to hard currencies. Therefore, "gaming" than profit sharing mechanisms. investors must receive some protection (usually index- ing) against depreciation of the domestic currency. 10. Regulation by contract should be reserved for Indexing should be limited to costs that will be private distribution companies. incurred in foreign currencies (usually about 40 to 60 percent of the retail tariff) rather than indexing of a · In developing countries, regulation by contract rarely full 100 percent of the tariff (as was the case in works for state-owned power enterprises. This is Argentina). However, any indexing system, no matter because most state-owned enterprises do not respond how well designed, will not be politically sustainable if to normal commercial incentives. the local currency experiences a major loss in value. · Thus there is little to be gained in creating a new · In some developing countries, the regulatory contract independent regulator to regulate a state-owned will need to be combined with transition-period subsi- power enterprise. Despite the fact that the new regu- dies to the private company, its customers or both. The lator may have all the paraphernalia and trappings of easiest way to deliver a subsidy to a distribution com- an independent regulatory system, it will accomplish pany is by subsidizing the cost of the power that it pur- little or nothing. Or in the words of one new regulator chases from government suppliers or private suppliers. who was forced to regulate an existing state-owned · A government should always require a quid pro quo enterprise: "I write longer and longer orders but less for any subsidies that it funds. For example, a govern- and less happens." ment may decide to subsidize the cost of connecting · For most state-owned power enterprises, institutional new rural customers and the bills of poor customers change changes nothing. throughout the system. To receive these subsidies, the company must show that its connections of new cus- tomers meet pre-specified technical standards and the customer must show that it paid the unsubsidized por- tion of his bill. APPENDIX A. CONTROLLABLE (2) The Commission shall utilize, whenever feasible, pre- AND NON-CONTROLLABLE COSTS: determined indices and external benchmarks for A Proposed Regulatory Framework costs under the control of the Licensee. Note: Authors' comments are in bolded italics. (3) The Commission shall establish incentive mecha- nisms that provide Licensees with rewards and/or (1) A multi-year tariff-setting formula may distinguish penalties based on their performance relating to between the treatment of input costs based on differ- benchmarks established for particular cost and/or ent degrees of effective control--that is, the ability of operational targets. [In developed countries, the key a distribution and retail supply licensee to signifi- benchmark is usually an overall efficiency factor cantly influence the cost and quality such that: while in developing countries it is likely to be com- mercial and technical loss reduction and sometimes (a) The prescribed tariff treatment may differ new customer connections.] depending on the degree of effective control exercised by the Licensee over the price and/or (4) The design and implementation of the tariff treat- quantity of a particular input. [Suppose that a ment of both controllable costs and non-controllable distribution company is assigned vesting con- costs should be guided by the overriding principles 62 tracts for some or all of its supply needs as part of implementability and reliable verification. of the privatization package. As a consequence it will have no control over the price paid for (5) The division between controllable and non-control- this power over the life of the contract. lable costs in a multi-year tariff formula will produce However, it may have control over the overall an allocation of risk between different stakeholders quantity of power purchased depending on its like the State Government, the Licensee, and the success in reducing technical and commercial Consumers. In determining a fair and efficient allo- losses on its distribution systems. For any new cation of risk, the State Government and purchases, the distribution company will gener- Commission should be guided by the following gen- ally have control over both price and quantity.] eral principles: (b) Automatic cost pass-through should be limited to (a) A risk should be borne by the entity best able to changes in components of an input cost (i.e., control or mitigate the risk. price, quantity or both) that are beyond the con- trol of the Licensee, on a pre-determined perio- (b) In determining the appropriate expected rate of dicity. For any delay in implementing such auto- return for a Licensee, the Commission shall matic cost pass-through beyond 90 days, the ensure that the risk being borne by the Licensee Licensee shall be compensated by allowing inter- is duly taken into account. est on the additional capital/working capital for the period of delay. This interest shall be reflected (6) If an unforeseen extraneous event occurs that has a in the Licensee's tariff as an automatic pass- material adverse impact on controllable costs, a through. Licensee may seek changes in the tariff to offset the material adverse impact of this event. (c) A Licensee's degree of effective control over an input cost may change over time. The (7) For claiming adjustment because of an adverse Commission has the authority to change the tariff material impact of an unforeseen extraordinary treatment of an input cost in the tariff order event, a licensee may file an application to the issued for a new multi-year tariff period but not Commission seeking tariff changes to offset the during a multi-year tariff period. If the material adverse impact. The application shall be Commission changes the tariff treatment for a accompanied by evidence and calculations to particular input cost, it must provide reasons in demonstrate that the event was extraordinary and writing for the change. uncontrollable, and the extent of the material adverse impact on the Licensee's costs and revenue. (b) the incremental price increase required to Such events may include, but not be limited to, damage ensure that the company is no worse-off finan- caused by acts of God, a calculation of and changes in cially than originally planned if this event had taxes, duties, and environmental regulations. not occurred. (8) Where an extraordinary event has produced a posi- If the recovery period extends into the next multi-year tive material impact on the Licensee, the tariff period, the additional price increase should be Commission may undertake a study to determine treated as a ring-fenced revenue stream regulated and whether a material positive impact has occurred. If a should be treated as an automatic pass-through cost. material positive impact has occurred the Commission may propose a suitable tariff change, with the burden of proof for demonstrating the uncontrollable nature of the event and its effect on the Licensee's cost and revenues falling on the Commission. Any Licensee investigated for an extraordinary material positive impact must cooper- ate with the Commission and provide whatever infor- mation is requested in a timely manner. 63 (9) If the Commission considers that the proposed tariff reconciliation or amendment under sub-sections (7) or (8) of a licensee is not permissible it shall, within 60 days (or such other period as may be prescribed by the Commission) of receipt of all the information which it required, and after consultation with the Commission Advisory Committee and the licensee, notify the licensee that (a) the reconciliation or the unforeseen adjustment sought is not permissible, or (b) in case the Commission is convinced that an adjustment (either positive or negative) is justified but the proposed magnitude of the adjustment is not justified, it shall specify an alternative adjust- ment that must be adopted by the Licensee. (10)If the Commission determines that a material adverse impact has been suffered by the licensee arising due to an extraordinary unforeseen event which warrants a tariff adjustment, it shall determine the tariff adjustment taking into account (a) the time period (the `recovery' period) over which the incremental costs should be recovered (capped at the remaining life of the current multi- year tariff period plus an additional multi-year tar- iff period); and (5) Subject to any Transition Arrangement stipulated APPENDIX B. OBLIGATION TO SERVE: under Section X, the Commission shall establish A Proposed Regulatory Framework and/or revise a schedule of charges and proce- dures applicable to connections for different types Note: Authors' comments are in bolded italics. of new customers, transfer of ownership of existing connections, reconnection of customers and cus- (1) The Licensees are obliged to supply consumers tomer-requested changes in service categories for within the pre-defined area of service of a radius of existing customers within the area of service defined up to 400 meters around existing distribution facili- in clauses 1 and 2 above. The schedule of charges ties owned or leased by the Licensee unless certain may allow for customer contributions to the cost of distribution facilities and the customers served by connections. these facilities are specifically excluded from a Licensee's area of service. Distribution facilities (6) The Government shall be entitled to make "subsidy means distribution substations and lines that are policy decisions" for providing such connections. used for the provision of distribution service to cus- tomers under the conditions included in the License. (7) Subject to a decision of the State Government or [This is the approach used in most privatizations of the Commission to introduce competition in retail 64 distribution in Latin America. It uses a "facilities- supply, [The law would be improved if it included based" approach to defining service area. An conditions that the Government or Commission alternative approach would be to impose an obli- must satisfy before introducing retail competition. gation to serve anyone within a defined geograph- For example, should existing licensees be compen- ic area such as a state or a district. This second, sated for stranded costs?] the Licensee shall have more open-ended obligation is riskier for a distri- the exclusive right to provide a bundled bution company if the defined area is rural and Distribution and Retail Supply within its specified includes many non-connected households.] area of service or an exclusive right to provide unbundled Distribution service or a combination of (2) The specific radius for different distribution facilities the two services as may be specified in its License. and regions shall be established by the Government The obligation to provide a bundled Distribution for the Transition Arrangement [In this country, the and Retail Supply or an unbundled Distribution government rather than the regulator will specify service extends to every person living within the the tariff-setting system in the first multi-year, post- area of service (i) who requests the service, and (ii) privatization tariff period. This reflects a policy who agrees to comply with the procedures and decision that the government is more likely to be pays the charges for the connection based on the sensitive to the tariff conditions that will support a schedule approved by the Commission. [This para- sustainable privatization than the regulator.] and graph reflects a policy decision to delay the intro- by the Commission thereafter. The radius may vary duction of retail competition. This reflects the view within a licensee's area of service. that it would be costly and complicated to intro- duce retail competition at the time of privatization. (3) Once the initial bands for area of service are estab- In Latin America, retail competition is being lished by the Government, they cannot be changed phased in--that is, the right to choose is initially through an increase or a decrease unless the given just to larger customers.] Licensee agrees to the modifications. (8) A Licensee may delegate its obligation to provide (4) Within this area of service, Licensees are required to bundled Distribution and Retail Supply services with- satisfy the Quality-of-Service Standards based upon in its area of service to other entities provided that: Schedule X principles [See Appendix C] - by the Government at the beginning of the transition peri- (a) The Licensee enters into a binding arrangement od, or by the Commission in subsequent multi-year with such delegatee to secure Distribution and tariff periods. Retail Supply of electricity to the area, and (b) The Licensee continues to be responsible for such services and for participation in subsidy the Distribution and Retail Supply of electricity schemes/programs. [The reference to "others" therein. implies that such entities do not necessarily require licenses. It would be over-regulation to require, for (9) Beyond its specified area of service, the Licensee is example, that a small mini-grid operator must satis- not obligated to extend bundled Distribution and fy all the terms and conditions contained in a regu- Retail Supply service or unbundled Distribution serv- lar Distribution and Retail Supply License.] ice to new customers. However, (a) A licensee may approach the Commission to seek approval for expansion of its area of serv- ice by filing an appropriate application. (b) The Commission and the Government shall encourage expansion of the coverage to areas that are not under supply. (c) The Commission shall pass appropriate orders 65 allowing or refusing to permit the licensee to expand its area of service based on terms and conditions deemed appropriate within [45 days] of receipt of application under Clause (a) above. In the event that no order is communi- cated by the Commission to the licensee within said 45 days on its application, the permission shall be deemed to have been granted. (d) If the Commission approves the expansion of the Licensee's facilities to provide bundled Distribution and Retail Supply service or unbun- dled Distribution service to additional cus- tomers, the Licensee's specified area of service shall be redefined to cover the new customers. (e) The Commission may grant an expansion of the Licensee's area of service on either an exclusive or non-exclusive basis. [This allows for other entities to compete for the right to supply unserved areas. There is no presumption that the Licensee will have a superior right to serve new customers outside its original service area.] (10)The Government may define and implement schemes/programs intended to extend service to new areas not in the Licensee's area of service, including provision of subsidy funds required for that purpose. Such schemes/programs shall clearly define conditions for participation of existing Licensees, new Licensees and others in extension of APPENDIX C. QUALITY OF SERVICE: the country and elsewhere in the world. The stan- A Proposed Regulatory Framework dards may be specified as guaranteed standards where the standard must be achieved in all specified All customers of the Distribution Licensees are entitled to cases and overall standards where the standard must an economic and reliable supply of electricity commen- be achieved on average across a specified customer surate with the prices that they pay for electricity. category but need not be satisfied for all customers Quality-of-service standards should be established to in the category. ensure that the licensee does not have an incentive to lower costs by lowering quality below the levels sought e. After a phase-in period, sanctions or penalties may by their customers. Therefore, the Commission, or the be imposed for failure to meet pre-specified quality- Government for the transition period described in of-service standards. Any penalties should be related Section X, shall specify quality-of-service standards that to estimates of the disutility experienced by the cus- satisfy the following principles. tomer (based, where feasible, on estimates of the cost to the customer of not being served) and the a. Quality-of-service standards should be established costs likely to be incurred by the licensee in meeting for those dimensions of service that are important to the standards. Rewards may be granted to consumers, controllable by the licensee and capable Distribution Licensees that meet targets associated 66 of being measured on a reasonably objective basis. with one or more pre-specified quality-of-service standards. Any system of rewards and penalties must b. Quality-of-service standards may be established for be efficient, equitable and likely to enhance cus- technical and commercial dimensions of service. tomer welfare. The magnitude of the rewards and Technical standards may include, but not be limited penalties need not be the same across customer cat- to, attributes such as frequency of outages, duration egories or geographic areas. of outages, stability of voltage and frequency relative to targeted levels. Commercial standards may f. Penalties may be paid to individual consumers or to include, but not be limited to, connection time for a general fund, administered by the Commission, new customers, accuracy in meter reading and that can used to provide subsidies to economically billing and response time to customer complaints. disadvantaged customers. Penalties cannot be used Any required standards must be capable of being to support the budget of the Commission or any measured in a cost-efficient manner and amenable other governmental entity. to auditing. The licensee must publicize the stan- dards to customers. g. The total amount of revenues earned from rewards or lost from penalties should be capped at no more c. Quality-of-service standards need not be uniform than 4 percent of estimated total revenues earned across all customer categories or geographic areas. from electricity sales to retail customers. The standards should be based on customers' pref- erences and their willingness to pay for the costs of h. Any changes in the quality-of-service standards providing the specified level of quality as determined should be synchronized with the regulatory proceed- by the Commission, except for the transition period ing to update the tariffs for a new multi-year period. specified in Section X when the standards may be In this proceeding, the Commission will appoint an established by the Government. independent and qualified consultant to produce estimates of the cost of not supplying energy to the d. Quality-of-service standards and associated penal- principal categories of customers. The results of this ties and rewards may be phased-in over time. study should be made available to the general pub- However, standards may not be changed during a lic for their comments before the Commission issues multi-year tariff period unless the changes were pre- revised quality-of-service standards and associated specified at the beginning of the tariff period or are rewards and penalties for the subsequent multi-year agreed to by the licensee. The standards may be period. based on the licensee's own past performance or the performance of other comparable licensees in i. Although the Commission has the ultimate authority to adjudicate disputes between a licensee and its customers over compliance with pre-specified quali- ty-of-service standards, the Commission has the authority to delegate this function to other entities. If the Commission chooses to delegate some or all of this function to another entity, the Commission still has the obligation to monitor the effectiveness and efficiency of the alternative complaint handling system. j. The Commission shall establish a reliable, objective and publicly available monitoring system that com- pares the quality-of-service provided by Distribution Licensees. The focus of the monitoring system should be on the performance of the Licensee with respect to the standards rather than on the expenditures made by the Licensee to try to achieve the stan- dards. The monitoring system can be based on sta- 67 tistically valid samples rather than full coverage to reduce the cost of creating and maintaining the sys- tem. The Commission has the authority to require licensees to provide the information necessary to create and operate such a system provided that the licensees are allowed to recover these costs in their tariffs. The Commission may require independent verification of the information provided by licensees to ensure that it is accurate and valid. REFERENCES Facility, December 2002. Available at http://rru.world- bank.org/energy. Joao Lizardo R. Hermes de Araujo. "Investment in the Brazilian ESI: What Went Wrong? What Should Be Done?" Bubnova, Nina. "Guarantees and Insurance for Re- Universidade Federal do Rio de Janeiro, Brazil 2001. Allocation and Mitigating Political and Regulatory Risks in Infrastructure Investment: Market Analysis." 1999. Alexander, Ian. "Financial Techniques: Accounting Available at Approaches and Asset Valuation." Power Point Presentation www.worldbank.org/riskconference/papers_txt.htm. at SAFIR Core Course, Agra, India, October 11, 2001. Available at www.ppiaf.org/toolkits/safir/agenda11.htm. Carne, Simon et al. The Competition and Policy Implications of Regulatory Depreciation and the Regulatory Alexander, Ian and Clive Harris. "Incentive Regulation and Asset Base (London: London Business School, Regulation Multi-year Price Controls: An Application to Regulation of Initiative Discussion Paper Series, Number 25, February Power Distribution in India." International Journal of 1999). Regulation and Governance, Volume 1, 2001, pp. 1-22. Costello, Ken. "Regulatory Questions on Hedging: The Alexander, Ian and Chris Shugart. "Risk, Volatility and Case of Natural Gas." The Electricity Journal, May 2002, Smoothing: Regulatory Options for Controlling Prices." pp. 43-51. 68 World Bank and European Bank for Reconstruction and Development, unpublished paper, November 1999. Demsetz, Harold. "Why Regulate Utilities?" Journal of Law and Economics, Volume 11, April 1968, pp. 55-65. Andhra Pradesh Electricity Regulatory Commission. "Long Term Tariff Setting Principles." February 2002. Available at DTE. Guidelines for price cap regulation of the Dutch elec- www.ercap.org/tariff/LongTermTariffPrinciples.htm. tricity sector for the period from 2000 to 2003. February 2000. Available at www.nma-dte.nl/en/default.htm. ------. Order No. 527, "On the Application of APTRANSCO for the consent of the PPA with Andhra Power Fernando, C.S. and P.R. Kleindoerfer. "Integrating Financial Ltd (BAPL)." December 13, 2002. Available at and Physical Contracting in Electric Power Markets." In The www.ercap.org/ppp/ppp.htm. Virtual Utility, edited by S. Awerbuch, Kluwer Academic Publishers, 1997. Besant-Jones, John and Bernard Tenenbaum. "Lessons from California's Power Crisis." Finance and Development, Foster, Vivien. "Non-Price Issues In Utility Regulation: September 2001, pp. 24-28. Available at www.world- Performance Standards and Social Considerations." bank.org/html/fpd/energy/calexp.html. Presentation at the International Training Program on Utility Regulation and Strategy, Public Utilities Research Center, Baldwin, Robert and Martin Cave, Understanding University of Florida, June 2, 1999. Regulation: Theory, Strategy and Practice, Oxford University Press, 1999. Gee, Dennis. "Performance-Based Ratemaking and Business Strategy Implications." Presentation at Infocast, Baumol, William J., Paul L. Joskow, and Alfred Kahn. The Inc., conference, Washington, D.C., December 2001. Challenge for Federal and State Regulators: Transition from Available at www.informationforecast.com. Regulation to Efficient Competition in Electric Power. Edison Electric Institute, 1996. Godbole, Madhave. "Electricity Regulatory Commissions: The Jury Is Still Out." Economic and Political Weekly, Vol. Block, Corinne, Delegated Management: A Solution for the XXXVII, No 7, June 2002. South African Electricity Sector. Association of Municipal Technical Electric Utilities, September 1998. Gomez-Ibanez, Jose A. The Future of Private Infrastructure: Lessons From the Nationalization of Electric Utilities In Latin Brown, Ashley C. and Ericson De Paula. Strengthening the America, 1943-1970. Discussion Paper, Taubman Center, Institutional and Regulatory Structure of the Brazilian Power Kennedy School of Government, Harvard University, Sector. Report prepared for the Brazilian Ministry of Mines January 1999. and Energy by the Public-Private Infrastructure Advisory Gray, Philip. "What We Know About Foreign Exchange and Kerf, Michel et al. Concessions for Infrastructure: A Guide Tariff Adjustment in Relation to Macro Shocks." Viewpoint, to their Design and Award. World Bank Technical Paper World Bank Finance, Private Sector and Infrastructure No. 399, March 1998. Available at Network (Forthcoming 2003. See http://rru.worldbank.org/ Toolkits/concessions. www.worldbank.org/html/fpd/notes.) Lamech, Ranjit and Kazeem Saeed. 2002 Survey of Power Green, Richard, "Checks and Balances in Utility Sector Investors in Developing Countries. World Bank Regulation--The U.K. Experience," Viewpoint, Number Energy and Mining Sector Board, Discussion Paper Series. 185, World Bank Finance, Private Sector and Infrastructure February 2003. Available at: www.worldbank.org/energy/. Network, May 1999. Available at www1.worldbank.org/ viewpoint/HTMLNotes/185/185summary.html. Lawrence, Georgina. Who Regulates the Regulator? Occasional Paper, Centre For The Study of Regulated Guasch, J. Luis, Jean-Jacques Laffont and Stephane Industries, University of Bath (England), August 2002. Straub. "Renegotiation of Concession Contracts in Latin America." Research Paper No. C02-22, University of Levy, Brian and Pablo Spiller. "The Institutional Foundations Southern California Center for Law, Economics and of Regulatory Commitment: A Comparative Analysis of Organization, October 2002. Telecommunications Regulation." Journal of Law and Economics, Volume 10, Number 2, 1994, pp. 201­247. Guasch, J. Luis. "Lessons from Ten Years of Concessions: 69 Determinants of Performance." Unpublished paper present- Maurer, Luiz T.A. "The `Virtuous Circle' of Contracting." ed at the World Bank, June 2000. Institute of Americas Conference on Gas and Power Market Convergence, Rio de Janeiro, June 25, 2001. Gupta, Pankaj, Ranjit Lamech, Farida Mazhar and Joseph Wright. Mitigating Regulatory Risk for Distribution Mayer, Colin and John Vickers. "Profit Sharing: An Privatization: The World Bank Partial Risk Guarantee. World Economic Appraisal." Fiscal Studies, Volume 17, Number Bank Energy and Mining Sector Board Discussion Paper 2, pp. 83­101, 1996. Series, No. 5, November 2002. Available at www.world- bank.org/energy/ subenergy/sectorboard_papers.html. Mercados Energeticos. Analysis of Resolution 073/2002 of the Energy Regulatory Comission of Colombia. December Harris, Clive. "Private Rural Power: Network Expansion 2002. In Spanish. Using Output Based Aid." Viewpoint, The World Bank Finance, Private Sector and Infrastructure Network, Note Michaels, Robert. "Stranded Investments, Stranded 245, June 2002. Intellectuals." Regulation, 1996, Number 1. Houston, Greg and Chris Bowley. "The Use of Arbitration Monari, Lucio. "Power Subsidies: A Reality Check on to Resolve Regulatory Disputes: A Case Study." Subsidizing Power for Irrigation in India." Viewpoint, Unpublished paper. Sydney, Australia: September National Number 244, The World Bank, Private Sector and Economic Research Associates, 2000. Infrastructure Network, April 2002. Available at www.worldbank.org/ Howard, Phillip K. The Death of Common Sense: How Law viewpoint/HTMLNotes/244/244summary.html. Is Suffocating the United States. Random House, 1996. Nellis, John. "Effects of Privatization on Income and Wealth Hunt, Sally. Making Competition Work in Electricity. John Distribution." Paper presented at the World Bank, Wiley and Sons, 2002. November 2002. Available at www.rru.worldbank.org. Jacobs, Scott. Building Regulatory Institutions: The Search OFGEM. Reviews of Public Electricity Suppliers 1998 to for Legitimacy and Efficiency. OECD, Centre for 2000, Supply Price Controls, Final Proposals. December Cooperation with Economies in Transition, Paris, 1994. 1999, p. 17. Available at www.ofgem.gov.uk/. Plummer, James and Susan Troppman, eds. Competition In Electricity New Markets and New Structures. Public Utilities Reports, Inc., 1990. Procta, Mark (Program and Project Supervisor, Office of Smith, Warrick, "Utility Regulators: The Independence Ratepayer Advocates, California Public Utilities Debate," Viewpoint, No. 127, World Bank Finance, Private Commission). Presentation at a conference on Sector and Infrastructure Network, October 1997. "Performance-Based Ratemaking and Business Strategy Available at www.worldbank.org/html/fpd/notes/ . Implications," Infocast, Inc., Washington, D.C., December 2001. Available at www.informationforecast.com. Stern, Jon and J.R. Davis. "Economic Reforms of the Electricity Industries of Central and Eastern Europe." Putman, Robert D. Making Democracy Work: Civic Economics of Transition, Volume 6, No. 2, 1999, pp. 427- Traditions In Modern Italy. Princeton, New Jersey: Princeton 460. University Press, 1993, p. 10. Tendler, Judith. Electric Power in Brazil: Entrepeneurship in Raab, Jonathan, Using Consensus Building to Improve the Public Sector. Cambridge, Massachusetts: Harvard Utility Regulation. Washington, D.C., and Berkeley, University Press, 1968. California: American Council for an Energy-Efficient Economy, 1994. Transparency International. Corruption in South Asia-- Insights and Benchmarks from Citizen Feedback Surveys in Rao, S. L. "The Last Resort." PowerLine, December 2001, p. 79. Five Countries. December 2002, p. 25. 70 Rudnick, Hugh and Jorge A. Donoso. "Integration of Price Viscusi, W. Kip, John M. Vernon and Joseph E. Harrington, Cap and Yardstick Competition Schemes in Electrical Jr. Economics of Regulation and Antitrust. Third Edition. The Distribution Regulation." IEEE Transactions on Power MIT Press, 2000. Systems, Volume 15, No. 4, November 2000, pp. 1428­1433. Williamson, Oliver E. "Credible Commitments: Using Hostages to Support Exchange." American Economic Sagar, Jagdish. "DVB Restructuring and Privatisation of Review, Volume 73, 1983. Distribution in Delhi." Unpublished paper given at the World Bank, August 2002. Woolf, Fiona and Jon Halpern. "Integrating Independent Power Producers into Emerging Wholesale Power Markets." Samarajiva, Rohan. "Establishing Regulatory Legitimacy." In The World Bank, Policy Research Working Paper No. Proceedings of the SAFIR Workshop on Regulatory Strategy, 2703, November 2001. Available at www.econ.world- edited by S. K. Sarkar, New Delhi: Tata Energy Research bank.org. Institute on behalf of SAFIR (South Asian Forum For Infrastructure Regulation), 2001. Available at World Bank, Project Finance and Guarantee Department. www.safir.teri.res.in/wkshp/legalproceed.pdf. IDA Guarantee Catalyses Private Finance for Haripur Power Project in Bangladesh. April 2002. Available at www.world- Schwartz, Eric and Jan Paulson. "Confronting Political and bank.org/guarantees/publications. Regulatory Risks Associated with Private Investment in Developing Countries: The Role of International Dispute World Bank. Bureaucrats In Business: The Economics and Settlement Mechanisms." Paper presented at the Politics of Government Ownership. World Bank Policy Conference on Private Infrastructure for Development: Research Report, Oxford University Press, 1995, pp. Confronting Political and Regulatory Risk, Rome Italy, 133­150. September 8-10, 1999. Available at www.worldbank.org/html/fpd/risk/papers.htm. World Bank. The World Bank's Role in the Electric Power Sector. World Bank Policy Paper, Washington, D.C., 1993. Shugart, Chris. "Regulation-by-Contract and Municipal Services: The Problem of Contractual Incompleteness." Wright, Joseph. "Mitigating Foreign Exchange Risk in Ph.D. thesis, Harvard University, May 1998. Private Power Investments in Developing Countries." Forthcoming in the World Bank Energy and Mining Shuttleworth, Graham " `Regulatory Benchmarking': A Way Discussion Paper series in 2003. Will be available at Forward or a Dead End?" Energy Regulation Brief, www.worldbank.org/energy. National Economic Research Associates, October 1999. Available at www.nera.com. THE WORLD BANK GROUP The Energy and Mining Sector Board The World Bank 1818 H Street N.W. Washington, D.C. 20433 USA