W O R L D B A N K W O R K I N G P A P E R N O . 4 0 Power's Promise Electricity Reforms in Eastern Europe and Central Asia Edited by Julian Lampietti THE WORLD BANK W O R L D B A N K W O R K I N G P A P E R N O . 4 0 Power's Promise Electricity Reforms in Eastern Europe and Central Asia Edited by Julian Lampietti THE WORLD BANK Washington, D.C. Copyright © 2004 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing: June 2004 printed on recycled paper 1 2 3 4 06 05 04 World Bank Working Papers are published to communicate the results of the Bank's work to the development community with the least possible delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally-edited texts. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply on the part of the World Bank any judgment of the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. The World Bank encourages dissemination of its work and normally will grant permission for use. Permission to photocopy items for internal or personal use, for the internal or personal use of specific clients, or for educational classroom use, is granted by the World Bank, provided that the appropriate fee is paid. Please contact the Copyright Clearance Center before photocopying items. Copyright Clearance Center, Inc. 222 Rosewood Drive Danvers, MA 01923, U.S.A. Tel: 978-750-8400 · Fax: 978-750-4470. For permission to reprint individual articles or chapters, please fax your request with complete information to the Republication Department, Copyright Clearance Center, fax 978-750-4470. All other queries on rights and licenses should be addressed to the World Bank at the address above, or faxed to 202-522-2422. ISBN: 0-8213-5900-2 eISBN: 0-8213-5901-0 ISSN: 1726-5878 Julian Lampietti is Senior Social Development Economist in Environmentally and Socially Sustain- able Development Sector Unit for the Europe and Central Asia Region at the World Bank. Cover Photo by Otar Jangvaladze. This photo reflects the current electricity situation in Georgia. Library of Congress Cataloging-in-Publication Data has been requested. TABLE OF CONTENTS Foreword vii Abstract .ix Acknowledgments xi Acronyms and Abbreviations xiii Executive Summary xv 1. The Promise of Reform 1 ECA is Different 1 Power Sector Reforms are Urgent 2 Proper Sequencing is Critical 3 Net Political Benefits Explain Variation in Progress 4 Taking Care of Institutions in the Energy Reform Roadmap 6 Identifying Outcomes is Difficult 6 2. Getting the Utilities off the Books 9 Quantifying the Sector Losses 9 Deficits Declined for Different Reasons 11 Relationship Between the Electricity Sector Deficit and the Fiscal Deficit 12 Reform Savings Did Not Go to Social Spending 13 Conclusions 15 3. Creating More Efficient Companies 17 Improve Transparency and Accountability 17 Revenue per kilowatt hour Rose 18 Generation Costs Declined 19 System Losses Held Steady 20 Collection Rates Varied 20 Mixed Results on Operational Efficiency 21 Focus on Service Quality 23 Regional Companies Take on the Challenge 23 Conclusions 25 4. Maintaining Power to the Poor 27 Household Energy Use Patterns Differ 28 Tariffs Rose 29 The Burden Increased 30 Consumption is Low 32 Gas May Be Filling the Gap 33 Enforcement is Necessary 35 Searching for Better Transfers 36 Conclusions 39 5. Local Versus Global Environmental Benefits 41 Did the Reforms Achieve Environmental Benefits? 41 There Were Unintended Environmental Costs 44 Damage from Dirty Fuel Use May Be Large 44 Conclusions 47 iii IV TABLE OF CONTENTS Annexes 49 Annex A: Overview of the Reform Process in Eight ECA Countries .49 Annex B: Proceeds from Privatization of Electric Utility Companies .53 Annex C: Tariff Losses, Commercial and Collection Losses, as Share of Total Losses .55 Annex D: Fiscal Balance and Electricity Sector Financial Deficit (Million USD and share of GDP .57 Annex E: Efficiency Indicators .59 Annex F: More on the Methodology for Estimating Health Effects .63 Annex G: Base Emission Factors .65 Annex H: More on Factors Leading to Low Contribution of Power Sector Toward Health Damages .67 Annex I: Changes in Generation Mix in the Past Decade .69 Annex J: Proposed Energy Issues to be Addressed and Sample Questions in LSMS/HBS Surveys .71 References 73 TABLE OF FIGURES Figure 1.1: Suggested Sequencing of Power Sector Reforms in ECA 4 Figure 2.1: The Electricity Sector's Losses Declined, 1993­2001 .11 Figure 2.2: Power Sector Losses are Weakly Correlated with Fiscal Deficits, 1995­2000 .13 Figure 2.3: Poor Countries Spend Less on Poor People (1997­2000) .14 Figure 2.4: Social Spending on Education, Health and Social Assistance Did Not Increase (as percent of GDP) .15 Figure 3.1: Nominal Revenues Increased Marginally, 1991­2001 .19 Figure 3.2: System Losses Stayed Stable, 1990­2002 .20 Figure 3.3: Collection Rates Differ by Country, 1990­2002 .21 Figure 4.1: Residential Electricity Tariff--Index of CPI Adjusted Real Tariffs .30 Figure 4.2: Enough for Three Light Bulbs and a Refrigerator .33 Figure 4.3: Electricity Tariffs are Higher Than Those for Gas, 1992­2002 .35 Figure 4.4: Burden of Arrears is the Same for the Poor and Non-poor, 2000­2002 .36 Figure 4.5: Alternative Subsidy Scheme .37 Figure 5.1: Fuel Required to Produce 1 Megawatt-hour of Electricity, 1992­98 .42 Figure 5.2: Electricity Contributes in Small Measure to Health Damages .45 TABLE OF TABLES Table 1.1: ECA is Different 2 Table 1.2: Timeline of Reforms in the Electricity Sector in ECA 4 Table 1.3: Independence of Regulatory Institutions: Paper and Practice 6 Table 1.4: Investment Climate Varies Among Reformers 7 Table 3.1: Suggested Indicators of Production Efficiency .18 Table 3.2: Reported Generation Costs Went Down, 1990­2002 (cents/kilowatt hour) .19 Table 3.3: Aggregate Impact of Reform on Collection Rates in Tbilisi .22 TABLE OF CONTENTS V Table 3.4: Service Quality Improved .24 Table 4.1: Urban Network Energy Use in ECA, (percent of households) .28 Table 4.2: Urban Non-network Energy Use in ECA (percent of households) .29 Table 4.3: Shares of Spending on Electricity Went Up, 1993­2002 .31 Table 4.4: Consumer Surplus Fell .32 Table 4.5: Electricity Expenditure as a Share of Total Energy Expenditure, 1993­2002 .34 Table 4.6: Simulation of Subsidy Cost-effectiveness for Tbilisi, Georgia .38 Table 5.1: Exposure to Indoor Air Pollution is High .46 TABLE OF BOXES Box 2.1: Untangling the Quasi-fiscal Deficit .10 Box 2.2: Calculating the Implicit Subsidies (Sector Losses) .10 Box 2.3: Social Spending in Latin America Increased Since the 1990s .16 Box 3.1: Private Sector Improving Household Collections in Georgia .21 Box 3.2: Opportunistic Behavior by the Private Sector? .25 Box 4.1: Simulation of Alternative Subsidy .37 Box 5.1: Reform Measures Expected to Result in Environmental Quality Improvements .42 Box 5.2: Estimating the Power Sector's Contribution to Air Pollution and Health Damage .43 Box 5.3: Methodology for Calculating Damages from Indoor Air Pollution .46 FOREWORD E nergy sector reforms remain among the most controversial development issues in transition economies, as these countries continue to tread the path toward sustainable growth. The legacy of central planning left the sector highly centralized, vertically integrated, and often ineffi- cient and deeply in debt. The reforms promised improved fiscal balances, more efficient resource use, better consumer service, and environmental benefits. This study ventures beyond the immedi- ate fiscal and efficiency impacts of electricity reforms to include the social and environmental impacts. At one level, it can be viewed as a cost-benefit analysis of power sector reforms in transi- tion economies. The countries in this study--Armenia, Azerbaijan, Georgia, Hungary, Kazakhstan, Moldova, and Poland--started reforming their electricity sectors in the 1990s. Different starting points with respect to income and political and institutional environment influenced sector reform implemen- tation; and the adjustment path was a function of both internal and external political and economic pressures. The study is timely because increasing emphasis is being placed on demonstrating the results of our work. It points to the need to improve our understanding of the linkage between sector reform and fiscal outcomes and how we can better monitoring of the fiscal and efficiency impacts of reform. It also suggests how we can design better strategies to mitigate the impact of reform on the poor and the environment. Ultimately the results of this study and the analytic framework it presents should be helpful for countries interested in assessing the full set of tradeoffs between equity and efficiency in power sector reform. Laura Tuck Sector Director Environmentally and Socially Sustainable Development Europe and Central Asia vii ABSTRACT T his study analyzes the fiscal, efficiency, social, and environmental impact of power sector reforms in seven countries in the ECA region. It finds sector deficits have been falling over the last decade and that the savings from lower sector deficits did not translate into higher social spending. More emphasis must be placed on monitoring deficits and tailoring policy reform to country specific circumstances. The impact of reform on utility efficiency, as measured by the cost of generation, system loss collections, and operational efficiency, is ambiguous. While overall rev- enue per kilowatt hour increased in almost all countries, problems continue with losses, collection rates, and staffing. In terms of social impacts, electricity spending as a share of income increased, especially for the poor, while consumption stayed the same. In terms of environmental impacts, reforms did slightly improve energy efficiency in power plants though this has little direct impact on human health because the electricity sector's share of the total health damage from air pollu- tion is negligible. Several lessons emerge from the analysis. Undertaking simple ex ante simulations of reform impacts will allow better identification of potential reform benefits and costs. Placing more emphasis on outcome-based indicators of service quality would help ensure that future oper- ations produce the intended end-user benefits. In many cases, tariff increases can and should be explicitly timed to coincide with service quality improvements. Yet, this may not be always possi- ble. Where it is not, the adverse impact of tariff increases, especially for low-income consumers, should be mitigated by improving access to and efficiency in the use of clean alternatives. ix ACKNOWLEDGMENTS A multi-sectoral team from ECSSD, ECSIE, and ECSPE prepared this report under the super- vision and guidance of Alexandre Marc (Sector Manager, ECSSD) and Lee Travers (Sector Manager, ECSIE). Julian A. Lampietti (ECSSD) managed the task. Task team members in alpha- betical order include Sudeshna Ghosh Banerjee (ECSSD), Julia Bucknall (ECSSD), Peter Dewees (ECSSD), Jane Ebinger (ECSIE), Irina Klytchnikova (ECSSD), Taras Pushak (ECSSD), Gevorg Sargsyan (ECSIE), Sergei Shatalov (ECSPE), Maria Shkaratan (ECSSD), and Katelijn Van den Berg (ECSSD). The team received support from local consultants including V. Valiyev in Azerbaijan, N. Nadiradze in Georgia, E. Nersisyan and A. Marjanyan in Armenia, S. Katyshev and G. Mandrovskaya in Kazakhstan, L. Lengyel in Hungary, and R. Gillecki from ARE in Poland. We are thankful to Ani Balabanyan for overseeing the work of local consultants in Armenia. Thanks are due to Maureen Cropper (DECRG), Kseniya Lvovsky (SASES), and László Varró of the Hungarian Energy Regulatory Office. We also thank Marcin Sasin in the Warsaw office of World Bank and Zsuzsa Szabo of Hungary National Statistical Office for help with household surveys. We are grateful to Bruce Ross-Larson of Communications Development Inc. for editing this report. Peer review comments were provided by Mamta Murthi (ECAVP), Gary Stuggins (EWDEN), Philippe Durand (LCSFE), Vivien Foster (LCSFP), Kseniya Lvovsky (SASES), and Jan Bojo (ENV). Navroz Dubash (World Resources Institute) and Ian Walker (Economic Adviser, Government of Honduras) provided comments on the concept note. Our partners in the local World Bank offices, Farid Mamedov in Azerbaijan, Pawel Kaminski in Poland, Gayane Minasyan in Armenia, Ilia Kvitaishvili in Georgia, provided comments and support throughout the study. Financial support from the Poverty Window of the Norwegian Trust Fund for Environment and Socially Sustainable Development and ECA Vice Presidency is gratefully acknowledged. We partic- ularly want to thank Giovanna Prennushi and Ignacio Fiestas of the Norwegian Trust Fund for Environment and Socially Sustainable Development for their continued support of this project. xi ACRONYMS AND ABBREVIATIONS CHP Combined Heat and Power CIS Commonwealth of Independent States CPI Consumer Price Index DALY Disability Adjusted Life Years DECRG Development Research Group EBRD European Bank for Reconstruction and Development ECA Europe and Central Asia ECAVP Office of the Regional Vice President, Europe and Central Asia Region ECSIE ECA Infrastructure and Energy, The World Bank ECSPE Poverty Reduction and Economic Management unit, Europe and Central Asia Region ECSSD Environment and Socially Sustainable Development unit, Europe and Central Asia Region EU European Union EWDEN Energy Anchor GDP Gross domestic product HBS Household Budget Survey ICRG International Country Risk Guide IMF International Monetary Fund kWh Kilowatt Hour LCSFE Energy Cluster, Latin America and Caribbean Region LCSFP Finance, Private Sector and Infrastructure unit, Latin America and Caribbean Region LPG Liquefied Petroleum Gas LSMS Living Standards Measurement Surveys SASES Environment and Social unit, South Asia Region UN United Nations UNEP United Nations Environment Program USAID United States Agency for International Development VAT Value Added Tax WHAP Winter Heat Assistance Program WHO World Health Organization xiii EXECUTIVE SUMMARY T he socialist legacy left a power infrastructure in Europe and Central Asia (ECA) that was equitable yet extremely inefficient. To speed the region's economies through the transi- tion, substantial emphasis was placed on reforms to increase efficiency in the sector. The promises of reform are improved fiscal balances, more efficient resource use, better consumer service, and environmental benefits. Ultimately, the deadweight loss is eliminated, and public resources are freed up for more productive investments. This study, intended to inform the design of power sector reforms in the region, identifies what has worked in promoting both equity and efficiency. It systematically brings together a broad range of cross-country indicators for the most aggressive reformers in the region over the last decade: Armenia, Azerbaijan, Georgia, Hungary, Kazakhstan, and Poland. Did the outcomes materialize as intended? While it may be too early to say, all of the chapters point to serious data deficiencies. In general it is possible to conclude that better design of moni- toring and data collection systems is required. In addition to enabling the assessment of reform impacts, those systems would improve program design by producing the data to allow simulation of the full range of policy options of interest to the client. Getting the Utilities Off the Books The legacy of central planning was a power sector contributing to unsustainable government deficits. So getting the power utilities off the government's books is the key to reducing fiscal budget deficit and promoting macroeconomic stabilization and growth. The sector losses have been falling over the last decade, with tariffs as the key determinant. Yet, more emphasis should have been put on monitoring those losses. Country-specific cost- recovery tariffs can generate more precise estimates of these losses. This work is under way in a separate regional initiative. Despite the decline in sector losses, the impact of reforms on the fiscal deficit is ambiguous. Why? Because of a combination of factors, including the lack of data, very complex financial and xv XVI WORLD BANK WORKING PAPER budget flows, and different adjustment paths. Only a detailed analysis of the quasi-fiscal accounts would lead to a definitive conclusion about the impact of the reduction in sector losses on the fiscal deficit. On balance, we conclude that assessment of budgetary impact of the reform can be improved by carefully examining the relative share of the energy and non-energy subsectors. Positive outcomes are expected in countries where a higher share of GDP is from the energy sector. However, institutional reform in these countries will also be the most difficult. The savings from lower sector losses did not translate into higher social spending. It may be that savings from reforms went to reducing prior sector obligations. The share of spending on social assistance, education, and health stayed the same or fell. Aggregate social assistance data show that these costs are much smaller than the savings generated by a reduced losses. This implies that a larger share of savings can presumably go toward alleviating the blow from tariff increases. It does not imply that financing consumption (rather than more productive investments in health and education) is the best way to do this. Creating More Efficient Companies Electricity reforms are expected to improve the production efficiency of the power sector, trans- lating into cost savings and service quality improvements for end users. They are also designed to make the sector financially sustainable by increasing the efficiency of resource allocation and the cost effectiveness of sector investments. The efficiency gains come from a profit motive--created by more competition or transparent regulation. While the legacy of central planning suggests large potential production efficiency gains, the data collected here on the cost of generation, system loss collections, and operational efficiency are ambiguous. Despite continuing problems with losses, collection rates, and staffing, overall revenue per kilowatt hour increased in almost all countries. The beneficiary of the additional rev- enues from declining generation costs and rising retail tariffs is not so clear. Limited aggregate data suggest that service quality improved in a number of capital cities. Simple ex-ante simulations would allow better quantification of potential reform benefits and costs. Placing more emphasis on outcome-based indicators of service quality would help ensure that future operations produce the intended end-user benefits. Both the private and the public sector can improve utility performance, suggesting that a broad range of contract arrangements can produce the desired outcomes. Finally, a number of local investors with both region and sec- tor experience are filling investment needs, suggesting alternatives to the traditional definition of a strategic investor. Maintaining Power to the Poor Unlike other regions, the socialist system gave almost all households access to reliable, inexpen- sive electricity. So the welfare gains from increased access--one of the most immediate and tangible benefits of power sector reforms--is not a consideration in most ECA countries. The welfare gains come from improvements in service quality, again reinforcing the need for reform to emphasize outcome-based indicators of service quality. Without disaggregated baseline data on service quality, reform appears closely linked to a fall in welfare. Electricity spending as a share of income increased, especially for the poor, while con- sumption stayed the same. Several lessons emerge. In many cases, tariff increases can and should be explicitly timed to coincide with service quality improvements. Yet, this may not always be possible. Where it is not, the adverse impact of tariff increases, especially for low-income con- sumers, should be mitigated by improving access to and efficiency in the use of clean alternatives. In some locations, especially urban areas where households heat with electricity, natural gas may be a viable substitute. There are serious problems of self-reported electricity data (as well as energy expenditure data in general) collected in the traditional poverty monitoring surveys such as the Living Standard Measurement Studies and Household Budget Survey. The reason is that the questions are con- founded by recall error, under and over reporting, and the presence of arrears, making it POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA XVII impossible to identify current and historical consumption. This also raises general concerns about the treatment of electricity consumption, an important part of household budgets in many coun- tries, in welfare aggregates. Where actual household data on consumption have been collected (directly from the utilities), consumption is very low and sufficient only to satisfy basic subsis- tence needs. In countries with very low consumption, demand for electricity is relatively inelastic, suggesting that there may be large welfare losses associated with future tariff increases. There may be substantial positive social benefits associated with access to electricity, suggest- ing an important continuing role for the public sector. This report is agnostic on the empirical evidence of the effectiveness of alternative instruments that mitigate the blow of tariff increases on the poor, such as income transfers and lifelines. It does, however, advocate helping clients calculate the social and fiscal implications of a full set of alternative mitigating strategies so that they can make fully informed choices. Local versus Global Environmental Benefits Power sector reforms are expected to produce environmental benefits. Higher production effi- ciency, new investment, and environmentally friendly technologies all contribute to lower fossil fuel consumption and lower emissions. This leads to better ambient air quality and presumably to better health outcomes for the local population. Unfortunately, claims about improvements in ambient air quality are difficult to verify because monitoring programs were never established. Better monitoring of ambient environmen- tal quality improvements is necessary in the future. For the long-term sustainability of reforms, it is critical that environmental impacts be systematically followed and recorded. It is appropriate to define mitigation measures for negative environmental impacts, such as those on health. Impact analysis should be on a national scale, not from a sector perspective. Reforms did slightly improve energy efficiency in power plants. This matters for carbon dioxide emissions and global climate impact. In most cases this has little direct impact on human health since the electricity sector's share of the total health damage from air pollution is negligible. It is also possible that reforms have damaged health because households switched to dirty fuels. Household fuel burning has a high impact on ambient urban air quality because households often live in densely populated areas in houses with low chimneys and no pollution abatement equipment. As noted in the previous section, a key solution is improving access to and efficiency in the use of clean alternatives. Survey data indicate that fewer households would use wood and coal if they had access to gas. Of course, in many countries the gas sector is also in need of reform before it can operate on a sustainable basis. As in the previous section, household surveys do not reveal enough information about energy (and other utility) reforms. More data are needed to evaluate the impact of reforms on fuel switching, energy use, substitution effects, and health and social impacts. This calls for including questions about utilities in such surveys and developing models to help predict behavior under a variety of scenarios. CHAPTER 1 THE PROMISE OF REFORM I n Europe and Central Asia (ECA), the socialist legacy left a power infrastructure that was equitable1 yet extremely inefficient. To speed the region's economies through the transition, substantial emphasis has been placed on efficiency-increasing reforms in the power sector--coupled, where necessary, with redistribution through lump-sum transfers. Unlike other regions where reforms promise increased access to electricity, one of the major objectives in ECA is to prevent further service quality deterioration. The long-run promise of power sector reforms in ECA includes improved fiscal balances, more efficient resource use, bet- ter consumer service, and environmental benefits as energy efficiency increases. Ultimately the deadweight loss is eliminated, and public resources are freed for more productive investments. This report tries to determine if the reforms produced the intended outcomes. In a perfectly competitive economy, tradeoffs between equity and efficiency take place along a production fron- tier (Birdsall and Nellis 2003). In principle, ECA economies are well inside the production fron- tier, so a balanced reform strategy could offer potential for growth without sacrificing equity. The report attempts to identify what has worked in promoting both equity and efficiency, recognizing the importance of externalities, information asymmetries, rent-seeking behavior, and other attrib- utes of imperfect markets. This study, intended for policymakers in the ECA region, informs the broader debate on impact of energy reforms. ECA is Different The starting point for energy reforms in ECA is different from the other regions, further compli- cating program design and implementation (Table 1.1). Incomes, measured by GDP per capita, are higher than in other parts of the world except Latin America. Similarly, other development indicators--such as infant mortality rate, illiteracy rate, and access to basic infrastructure such 1. In this context, equitable implies widespread access to electricity. 1 2 WORLD BANK WORKING PAPER TABLE 1.1: ECA IS DIFFERENT Access to GDP per GDP per an Improved Unit of Energy Capita, PPP Adult Infant Water Source: Use: PPP (Current Inter- Illiteracy Mortality % of $ per kg Oil national $) Rate Rate Population Equivalent East Asia & Pacific 3,730 14 35 76 na Eastern Europe & 6,271 3 31 91 2.3 Central Asia Latin America & 7,093 11 30 86 6.1 Caribbean Middle East & 5,188 36 45 88 3.8 North Africa South Asia 2,236 45 72 84 5.5 Sub-Saharan Africa 1,647 39 106 54 2.9 Note: All the figures are for 2000. Source: World Development Indicators, World Bank (2003). as water--suggest that the situation in ECA is better. GDP per unit of energy use is substan- tially lower--due in part to the cold winters, the legacy of central planning, and substantial declines in household incomes following transition. The Soviet legacy resulted in publicly owned, vertically integrated, and highly centralized power infrastructure designed to provide reliable electricity to all households at little or no cost. Studies indicate that access to electricity is substantially higher than in other regions with similar incomes (Komives, Whittington, and Wu 2001; Clarke and Wallsten 2002). Electricity consumption levels are also thought to be higher than in other parts of the world, suggesting substantial potential efficiency gains (Cornille and Frankhauser 2002). Power Sector Reforms are Urgent Providing large numbers of consumers with reliable electricity at little or no cost requires substan- tial power infrastructure. When this infrastructure was built, energy prices in the Soviet Union were well below international prices. With the onset of transition and the end of central transfers, the sector's large investment needs were neglected because revenues were low, if positive at all. Low tariffs, high consumption levels, and low collections characterized the sector. The net result was a power sector supported through fiscal and quasi-fiscal budget operations and asset depreciation, especially in countries of the former Soviet Union. In Armenia, Georgia, and Moldova the power sector deficit was among the largest items in the budget deficit. The energy (electricity and gas) sector's quasi-fiscal deficit has been estimated at 5 percent of GDP in Moldova in 1999, 3.5 percent in Armenia in 2001, and 2 percent in Georgia in 2000 (Interna- tional Monetary Fund 2001a, 2001b, 2001c). Thus, the opportunity cost of subsidizing the energy sector is large. As countries struggle to balance their budgets and reduce quasi-fiscal operations, transfers of this magnitude take limited funds away from other sectors. Public spending on health and education has fallen dramatically since the beginning of the transition. In Azerbaijan, Georgia, and Moldova spending on health is less than a quarter of what it was in the early 1990s--and in Kazakhstan, about half. Spending on education is a mere sixth of the level in the early 1990s in Armenia, and a third in Azerbaijan. As shares of GDP, total public expenditures on education, health, and social assistance and welfare remained stable or fell (Public Expenditure database, 2003). POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 3 Falling service quality, continuing lack of investment, and persistent sector deficits made the reforms urgent in the ECA region, and the donor community has been advising governments on how to undertake them since the early 1990s. The advice of the World Bank, detailed in the 1998 ECA energy sector strategy, includes raising prices to cost-recovery levels, metering and cutting off nonpaying customers, establishing predictable and transparent regulations, introduc- ing competition in generation and supply, and selling industrial assets to private strategic investors (World Bank 1998). International donors supported energy conservation measures by encourag- ing price reforms, providing funding for improving energy efficiency, and providing advice on better targeting of transfers to poor households. This study examines the reform outcomes for countries that have made significant progress. It evaluates their experience along four dimensions: fiscal and quasi-fiscal operations, productive efficiency, social welfare, and environment. Armenia, Azerbaijan2, Georgia, Hungary, Kazakhstan, Moldova, and Poland were selected for this study because they have been among the most aggressive reformers over the last decade.3 Before analyzing these outcomes in subsequent chapters, the rest of this chapter explores factors driving the sequence and pace of reform in these diverse economies. Proper Sequencing is Critical In most ECA countries the electricity sector reforms began almost a decade back, but privatiza- tion started only recently. Figure 1.1 presents the suggested sequence of reforms. However, each country followed its own path based on the domestic political and economic conditions. Some governments, like Hungary and Kazakhstan, have chosen to go as far as to open the generation subsector to foreign investment and privatize, while others deem the generation assets as strategic and retain public ownership. A comprehensive review of the reform experience in ECA reveals that no universal recipe for reforms can fit the specific conditions of each country (Krishnaswamy and Stuggins, 2003). Hungary and Poland engaged in reform the earliest, followed by Kazakhstan, Moldova and Georgia (Table 1.2). Unlike all other countries, Hungary's electricity sector has been open to participation of foreign strategic investors from the outset of reforms, much like other "strategic" sectors in Hungary, such as banking. Because there was no problem with the payment discipline either before or after the beginning of reforms, distribution and generation companies were privatized simultaneously. Though Armenia entered late, significant progress was made in restructuring the sector. It made a number of abortive attempts to sell the loss-making distribution network to foreign strategic investors before the sale finally went through. In 2002, a controlling stake in its power distribution network was sold to an offshore company--Midland Resources Holding. It happened after substantial attempts by the government to reorganize the sector, improve service, and align prices. The pace of energy reforms was comparatively fast in Georgia, which has initiated all reform measures. Georgia's biggest foreign investor, U.S.-based AES Corporation, recently sold its assets after five years of operation and significant problems with tariffs, bill payments, and political disturbances. Kazakhstan privatized 80­90 percent of its generating capacity using asset sales and concessions in 1996, and it completed privatization of the rest of generation assets between 1999­2002 except for a hydro plant (Annex A provides a detailed description of the reform steps in each country). The common theme in the experience of all countries is the central role of institutions and the regulator in a successful reform process. Another common theme is the sequencing of reform steps. Sequencing becomes crucially important in the presence of nonpayment. If generation assets are privatized before distribution, a poor collection rate will result in an undervaluation of 2. Azerbaijan placed its distribution networks under private management contract in 2002. 3. Moldova and Ukraine could also be considered in this group but were not included here because of data collection issues. 4 WORLD BANK WORKING PAPER TABLE 1.2: TIMELINE OF REFORMS IN THE ELECTRICITY SECTOR IN ECA Date of Passage of Energy Law and Creation of an Independent Corporatization Privatization of Privatization Market Regulator and Unbundling Distribution of Generation Liberalization Armenia 1997 1997 2002 none none Azerbaijan 1998 1996 2001­2003c none none Georgia 1997 1999­2000 1998 2000 1999­2000 Hungary 1993­94 1993­94 1995 1996­97 2001 Kazakhstan 1998­99 1996 1996, 1999 1996, 1999­2002 b Moldova 1998 1997 1999 None None Poland 1997 1993 Ongoing Nonea Ongoing a. except for new entry of private strategic investors b. large customers and generators are allowed to enter in bilateral contracts. In 2001 the government set up KOREM (Kazakh market operator for electric energy and capacity) to organize the spot and `day ahead' markets. c. concession contracts Source: Adapted from Krishnaswamy and Stuggins, 2003. FIGURE 1.1: SUGGESTED SEQUENCING OF POWER SECTOR REFORMS IN ECA Regulatory development Monolithic Corporatization Unbundling of Privatization Privatization Market company of company generation, of distribution of generation liberalization transmission and distribution Source: European Bank for Reconstruction and Development (2001). the generation assets, or a buyer may not be found at all. Sector unbundling becomes difficult when nonpayment is pervasive. One of the key lessons from the reform experience is that dur- ing the period of economic turmoil and pervasiveness of nonpayment, it is prudent to focus on the stabilization efforts, removal of legal, political and attitudinal constraints to denial of serv- ice to defaulters, and on other measures to improve collections and recover arrears. When these preconditions were not present, the reform progress was difficult and associated with social upheaval. Net Political Benefits ExplainVariation in Progress Political, institutional, and macroeconomic conditions explain the variability in the pace of reforms and their outcomes. The decision to reform is determined by the net political benefits that accrue to policymakers--the difference between the benefits and costs of sector reform. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 5 There is a mismatch between short-term costs and long-term benefits because costs are usually immediate and benefits accrue in the future (Rodrik 1994, Banerjee and Munger 2003). It is widely acknowledged that while the demise of communist regimes and pressing fiscal concerns were the major factors driving sector reforms in all the economies in the region, non- payment, widespread economic dislocation, and decline of industrial output accompanied this transition. Country-specific conditions--such as energy endowment, European Union (EU) accession, and the accumulation of energy related debt--also determined the reform's timing and intensity. Therefore, the reforms in Kazakhstan and Azerbaijan are analyzed here in the context of resource dependence theory, Hungary and Poland in the context of EU accession, and Armenia, Georgia, and Moldova in the context of energy-related debt. Paying careful attention to these factors and their implications for the speed of governments acting in the initial stages of reform design may well improve long-term performance. Resources as a barrier. Energy resources can either hurt or ease progress with sector reform. They can ease the reform process by providing the resources necessary to prevent fiscal crises, but they can also be a barrier to reform implementation when resource rents are appropriated by the ruling elites (Esamov and others, 2001). Indeed, energy-rich countries (Kazakhstan and Azerbaijan) and energy-poor countries (Armenia, Georgia, and Moldova) differed in their reaction to reform and the pace of alignment with the world energy prices (Saavalainen and ten Berge 2003). While the energy-exporting countries gained from a change in their terms of trade during the transition period, now able to export their energy resources at the higher world price, the poorer countries lost. The latter countries accumulated energy-related debt and did not have resources within the energy sector to mitigate the adverse social impact of reforms. Energy resources may also have slowed the pace of reforms.4 It has been shown that the rul- ing elites benefit from partial reforms in lucrative sectors, gain control of the regulatory process, and prevent the creation of a level playing field (Hellman 1998). As Saavalainen and ten Berge (2003) note, the symbiotic relationship between energy utilities and the government meant widespread corruption and rent-seeking in the pre-reform era. Accession to the European Union. The prospect of EU accession has provided the impetus for fast-paced reforms, especially in developing a regulatory framework and unified gas and electricity markets. The EU accession countries will have to conform with the EU directive on power reforms, which include liberalizing markets, decoupling generation, transmission, and distribution, and establishing regulated third-party access for the power network. In privatization, Hungary is the leader in pursuing major electricity privatizations in the 1990s, with most electric utilities privatized and prices at the world market level. Poland, following the dissolution of the communist regime, embarked on an ambitious "economic transformation program" in 1990. However, the Polish government has been more careful than Hungary to allow entry of foreign investors in the energy sector, deemed by the government as "strategic."5 Energy-related debt. The need to reduce energy-related external debt has been a significant driver of reform in the heavily indebted Armenia, Georgia, and Moldova (Saavalainen and ten Berge 2003). General economic, social, and political malaise in these countries following transi- tion made power sector reforms urgent. Conditional lending was used in an effort to create a stable macroeconomic environment. For Armenia, Georgia, and Moldova, donor conditionalities focused on cost-recovery tariffs, collections, independent regulation, restructuring, unbundling, and privatization. Performance was mixed. During 1993­2002 only 60 percent of IMF energy 4. For suggestive evidence on this in other regions particularly on the philosophical debate and empirical evidence on the inverse relationship between natural resource abundance and economic growth, see Sachs and Warner (1995). 5. For more information on Hungary and Poland's EU requirements, refer to Hungary: On the road to the European Union (A World Bank Country Study) and "Poland--Country economic memorandum: reform and growth on the road to the EU" (World Bank). 6 WORLD BANK WORKING PAPER TABLE 1.3: INDEPENDENCE OF REGULATORY INSTITUTIONS: PAPER AND PRACTICE Armenia Azerbaijan Georgia Hungary Kazakhstan Moldova Poland Separate regulator Fixed-term appointment Industry funding Full tariff-setting power Transparency Redress Source: Kennedy (2003), adapted from the EBRD/World Bank survey of regulators. conditions were implemented (primarily relating to foreign energy debt and categorical privileges). Recognizing this, in the past year the IMF and the World Bank decreased the number of conditions in all countries except Georgia (Saavalainen and ten Berge 2003). Taking Care of Institutions in the Energy Reform Roadmap The biggest challenge in the reform process has been creating the appropriate institutional environment. Before the transition the energy sector in the former Soviet Union was a monopoly under direct state control. One major objective in the reform process is to encourage private interest to promote capital investment and harness technical expertise. For this to succeed, the government has to be committed to reform and have the adequate institutional capacity in place. Beyond doubt, the role of regulator is paramount in creating a level playing field and plausible dispute-resolution system. Most of the countries in this study have progressed with regulatory reform, and all of them have set up regulatory agencies (Table 1.3). However, two unexpected problems occurred at the early stages of the process. The first was a shortage of regulators with technical experience in the region and its politics. This also suggests a role for international aid agencies to step in to provide technical assistance for setting up a system of fair and accountable regulation. The second was (and is) continued political interference in the appointment of regulators and in funding. Even undertaking all the regulatory steps may not ensure successful outcomes, because it takes a while for regulation to become effective. Based on a survey of international power investors in developing countries, Lamech and Saeed (2003) found that investors require "enforceable legal system and credible rules of the game." The governments should be willing and able to respect its commitments and contractual obligations. It may appear that they do: for instance, the Georgian government has always overtly supported AES. Even so, government intervention is highest for Armenia, Azerbaijan, Moldova, and Georgia, and Georgia also tops the list with its entry barriers (Table 1.4). The problems of the government are exacerbated by the difficulty in identifying and evaluating the outcomes of the process, impor- tant for ensuring credibility of the reforms and justifying the price rise to citizens. Identifying Outcomes is Difficult Studying reform outcomes is complicated because it requires a counterfactual--what would have happened without reforms? Creating this counterfactual to evaluate the performance of the entire electricity sector is difficult because it requires many assumptions--not only about a particular company,6 but about the entire domestic political and economic situation (for example, 6. A methodology for this type of analysis is detailed in Galal and others (1994). POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 7 TABLE 1.4: INVESTMENT CLIMATE VARIES AMONG REFORMERS Armenia Azerbaijan Georgia Hungary Kazakhstan Moldova Poland Degree of government 3 3 3 1 2 3 2 interventiona Entry barriersb 11 na 13 8 12 na 11 Degree of property rights 3 4 4 2 4 3 2 protectionc Quality of regulationd 4 4 4 3 4 4 3 Index of political riske 41 40 na 20 31 32 22 Index of judicial system 2 2 2 >1 >1 1 >1 efficiencyf Index of corruptiong 6 6 5 3 6 6 2 a. High values imply high levels of government intervention.This factor measures `government's direct use of scarce resources for its own purposes and government's control of resources through ownership'. Ranges from `1' (government expenditure less than of equal to 15 percent of GDP) to `5' (Greater than 30 percent'. Source: Heritage Foundation (2002). b. High values imply difficult to enter. Source: Number of Procedures Needed to Start a Business (Djankov et al., 1999). Data are for 1999. c. High values imply weak protection.This factor measures the degree of protection to private property and how laws are enforced to protect private property. Ranges from `1' (very high protection) to `5' (very low pro- tection) Source: Heritage Foundation (2002). d. High values imply poor regulation.This factor measures the extent of difficulty in operating a business. It also examines the degree of corruption in government and whether regulations are imposed fairly in all businesses. Ranges from `1' (very low levels of regulation) to `5' (very high levels of regulation) Source: Heritage Foundation (2002). e. High values imply greater risk. Source: ICRG (June 2002). [100-ICRG political risk score]. f. High values imply poor performance. Source EBRD (2001) (4-score on legal effectiveness). g. High values imply greater corruption. Indicator of `perception of corruption in civil service, the business interests of top policy makers, laws of financial disclosure, and conflict of interest, and anti-corruption activities'. Source: Nations in Transit (2001). Source: Adapted from Lieberman and others 2003,World Bank (2003f). alternative scenarios of economic growth, private investment, and political stability). The alterna- tive of trying to compare outcomes before and after reform ignoring the counterfactual is also difficult because of the debate over the appropriate baseline (pre-transition or transition), the fact that reforms are dynamic, and the scarcity of pre-reform data. Therefore this paper analyzes per- formance trends over the last decade using data compiled from a number of sources including local consultant reports, reviews of project documents, Household Budget Surveys (HBS), and Living Standards Measurement Study (LSMS) surveys. There are substantial methodological chal- lenges in identifying outcomes, particularly the social and environmental. The methods and issues raised here can be used in the new Poverty and Social Impact Analysis initiative in the Fund and the Bank to explicitly incorporate the findings into reform design. Caution is required in drawing conclusions because reforms are a dynamic process and the countries examined are at different points in the reform process (Table 1.2). Hungary and Poland started the reforms as early as 1992, while most of the Commonwealth of Independent States (CIS) countries initiated them in 1995 or 1996. Of course, it has to be kept in mind that most of the CIS economies became countries only the beginning of the decade and tasks of nation- building took precedence. It may thus be too early to draw conclusions about outcomes where 8 WORLD BANK WORKING PAPER the process is not complete. In addition, the time elapsed does not imply anything about the quality or pace of reforms. The rest of the report is organized as follows. Chapter 2 evaluates the electricity sector reforms' financial outcome and specifically their impact on the fiscal and quasi-fiscal operations of the government. Chapter 3 evaluates production efficiency, Chapter 4 the social outcomes, and Chapter 5 the environmental outcomes. CHAPTER 2 GETTING THE UTILITIES OFF THE BOOKS R educing fiscal budget and quasi-fiscal deficits is a key measure to promote macroeconomic stabilization and growth in transition economies. Because the power sector is one of the largest contributors to these deficits in ECA, clarifying and then reducing government liabilities has been a driving force for sector reforms. The reforms can relieve budgetary pressure if they improve enterprise efficiency and reduce the need for transfers. They can also generate immediate budget revenues through asset sales during privatization. This chapter analyzes the implications of reforms on the fiscal and quasi-fiscal balance of the government overtime. There are immediate budgetary impacts of reform (through privatization revenues) and companies or governments usually disclose this information (Annex B). Because the immediate budgetary impact of privatization revenues is most likely negligible in comparison with the long-term effect of reducing subsidies, it is not included in the analysis. In other regions the immediate fiscal impact of privatization has also been small. Pinheiro and Schneider (1994) show that in Latin America privatization did not result in substantial short-term fiscal benefits. Quantifying the Sector Losses There has been little systematic monitoring of the fiscal and quasi-fiscal balance of the power sector over the last decade, making it impossible to identify the magnitude of gains and losses. The scarcity of data may be due to the complexity of untangling financial flows between electric utilities and the government (Box 2.1). The lack of systematic monitoring is well recognized, and several new initiatives are under way to redress this situation, including a regional study that will include calculation of implicit subsidies in the power and gas sectors for ECA. This study will also estimate country specific cost-recovery tariffs. This chapter quantifies the electricity sector's losses or implicit subsidy, using primary data and a formula developed by the World Bank (Box 2.2). It circumvents the problem of complex financial flows by defining the sector losses as the sum of three components: commercial losses, 9 10 WORLD BANK WORKING PAPER BOX 2.1: UNTANGLING THE QUASI-FISCAL DEFICIT The fiscal deficit is the difference between revenues and expenditures as recorded in the official government budget. In addition to fiscal deficit, public finance analysis takes into account government obligations that are not reflected in the budget but that result from explicit or implicit government liabilities outside the budget framework.Those could be explicitly recorded in legal documents or result implicitly from the logic of political events, from institutional rules, from social obligations of the government as understood by the public.These obligations, when they need to be financed, might cause substantial increases in government debt and budgetary spending due to required repayment of the debt.They thus constitute major fiscal (or budgetary) risks. When a utility is publicly owned, the government receives taxes and dividends from the utility and provides explicit and implicit subsidies, many of them not transparent. Untangling these financial flows requires detailed systematic data on financial flows that is not readily available.The data and analysis of the electricity sector fiscal and quasi-fiscal deficits are available for the countries for which the IMF or the World Bank undertook detailed studies, such as Armenia, Romania, and Russia (for example, see Petri and others 2002, Frienkman and others 2003, and Saavalainen and ten Berge 2003).There is also some evidence for Georgia and Moldova, but no systematic methodology or time series data have been available to date. BOX 2.2: CALCULATING THE IMPLICIT SUBSIDIES (SECTOR LOSSES) We use the "End-Product Approach" to calculating sector financial deficit, which is based on the quantity of energy sold, the end-user prices, and the collection rates compared with import or export prices. Developed by Infrastructure and Energy Group of ECA region,World Bank (ECSIE), the approach has been used in different modifications (such as Saavalainen and den Berge, 2003; Petri and others 2002; Cosse 2003). The modification used here is the latest one developed and applied by ECSIE in its current work. This approach allows calculation of the following three types of losses: commercial, collection, and tariff.1 Commercial losses are the cost of electricity injected in the transmission system but not metered/billed minus the cost of electricity lost for technical reasons within nationally accepted norms for unavoidable losses. Measured this way, commercial losses include electricity stolen through illegal connections to the network and technical losses above norms due to bad maintenance and deteriorated physical assets. Collection losses are measured as the value of electricity billed but not collected from the consumers. Tariff losses are estimated as the difference between the amount billed (collected and not collected) to the consumers and the cost of the corresponding amount of electricity. The commercial and tariff losses depend on the assumptions about the cost recovery price (which in turn depends in large part on the value assigned to the sector assets). The higher the cost-recovery tariff, the greater the shortfall between this tariff and the actual tariff. Calculations in this study use the cost-recovery tariff of 4.8 US cents a kilowatt hour for all countries (Armenia,Azerbaijan, Georgia, Hungary and Poland) between 1992 and 2001, except for Kazakhstan, where the tariff is 4 cents until 1997 and 3 cents in the subsequent periods. This is an approximation of the long-term marginal cost, based on information provided by the local expert/investor estimates for this study. The cost-recovery tariff is estimated based on an allowance for generation, distribution, and theVAT. More precise estimates of cost recovery tariffs will be available on completion of a new regional study initiated by the World Bank. collection losses, and tariff losses. The results are very sensitive to assumptions about the cost- recovery tariff, emphasizing the importance of the country-specific calculation of cost recovery tariffs if sector gains and losses are to be more accurately quantified. The approach does not enable us to identify the shares of the sector losses accruing to the government and the private sector. This may be moot, however, because the government eventually ends up being responsible for financing most of these losses. For example, in POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 11 Georgia following privatization, the private sector share of these losses quickly fell while the government share grew. It may be that the private sector is transferring part of its deficit to the budget. Deficits Declined for Different Reasons Sector losses declined between 1993 and 2001 in all countries except Azerbaijan and Georgia, where it remained more or less unchanged after 1993 (Figure 2.1 and Annex C). By 2000 there FIGURE 2.1: THE ELECTRICITY SECTOR'S LOSSES DECLINED, 1993­2001 Armenia Azerbaijan 200 700 180 600 160 140 500 120 ml 400 100 ml $ 80 $ 300 60 200 40 100 20 0 0 93 94 95 96 97 98 99 00 01 93 94 95 96 97 98 99 00 01 Georgia Hungarya 600 500 500 400 400 300 300 ml 200 ml $ $ 200 100 0 100 -10093 94 95 96 97 98 99 00 01 0 -200 93 94 95 96 97 98 99 00 01 Kazakhstan 600 Poland 2500 400 2000 200 1500 ml $ 0 ml 1000 93 94 95 96 97 98 99 00 01 $ 500 -200 0 -400 93 94 95 96 97 98 99 00 01 -500 Commercial losses Noncollection losses Tariff losses a. According to the information provided by the local consultants for this study, most of the outstanding bill amount is collected within the first six months after the due date, and a negligible part of it--later, but within a year.Therefore, it is reasonable to consider the total bill amount as collected. Note: Poland and Hungary have close to 100% collection rate Source: Authors' calculations based on data provided by local consultants for this study. 12 WORLD BANK WORKING PAPER was a slight surplus in Hungary. Sector losses were less than 1 percent of GDP in Poland and less than 5 percent in Armenia and Kazakhstan. It still remained high in Azerbaijan (10.6 percent in 2000) and Georgia (8.8 percent in 2000). (See Annex table 4 for details). The decline is attributable to different reasons, suggesting that carefully targeted reform programs addressing specific deficit components work best. In Azerbaijan, targeting tariff losses by aligning actual tariffs with the cost-recovery price is key. In Georgia commercial losses are most important. In Armenia commercial losses and noncollection losses are equally important. In Kazakhstan noncollection losses are key. Relationship Between the Electricity Sector Deficit and the Fiscal Deficit The fiscal objectives of reforms are to reduce fiscal and quasi-fiscal subsidies to the electricity sector and to increase the transparency of the unavoidable subsidies by bringing them in the budget. Yet, the net fiscal gains of these changes may be small or negative, depending on the resources used to address the social implications of sharply rising electricity prices and reduced access to service. In addition, a tariff increase might result in increased non-payment of taxes, increased non-collection of billing, and reduced consumption, which can also hurt the budget. Moreover, the energy sector is fundamental to the economy. Energy reforms have serious implications for the production and profitability of most other sectors, affecting the flow of taxes to the budget. For this reason the impact of reforms on the fiscal deficit is ambiguous, even though we observe a decline in the sector losses. Reform-induced transparency and accountability may reveal a truer picture of fiscal deficits in the short term, when the accumulated quasi-fiscal deficit of the sector emerges on the fiscal balance of the government. In the long term, we expect the reforms to increase the profitability of energy generation and distribution companies, and we expect the government to increase tax enforcement when these companies become privately owned. This means a higher flow of tax revenues from the energy sector to the budget. At the same time, electricity reforms reduce the profitability of the manufacturing subsector due to higher electricity tariffs, which means lower tax revenues for the budget from the manufacturing subsector. On balance, the budgetary impact will depend on the relative share of the energy and non-energy subsectors of the industry. The impact on the budget is likely to be positive in countries where a higher share of GDP is from the energy sector. Consistent with the analysis in Chapter 1, the experience of the countries points to three adjustment paths. The first is the path of the high income energy importers (Hungary and Poland), the second, the oil exporting countries (Azerbaijan and Kazakhstan), and the third, the low income energy importing countries (Armenia and Georgia). Although at first glance the scatter plot of sector losses against country fiscal deficits appears to show a lack of correlation, a closer look reveals the difference in the country-specific adjustment paths (Figure 2.2). Hungary and Poland have low sector losses and fluctuating fiscal deficits. Armenia and Georgia show a decreasing tendency both in the sector losses and the fiscal deficits. The oil exporters Azerbaijan and Kazakhstan do not show any clear tendency, with the sector loss (sensitive to oil price) fluctuating widely. The availability of domestic energy sources and the income of a country are crucial factors linking the sector loss with the fiscal deficit. With appropriate institutional conditions, in the countries with sufficient investment capital an increase in the price of electricity results in a higher level of investment in energy efficient production processes by the manufacturing sector (as in Hungary and Poland). If little investment capital is available for efficiency improvements, an energy importing country may be locked in a low efficiency-high sector deficit trap, when the investment capital is not available to improve efficiency (as in Armenia and Georgia). Energy rich countries (as in Azerbaijan and Kazakhstan) can use oil revenues to finance a high sector loss, because the energy sector is a major source of revenues for the budget. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 13 FIGURE 2.2: POWER SECTOR Losses ARE WEAKLY CORRELATED WITH FISCAL DEFICITS, 1995­2000 18 Az-95 16 Ge-95 14 GDP 12 Az-96 of Az-01 % Az-99 Az-00 10 Ge-99 Az-98 deficit, Az-97 Ge-00 Ge-96 8 Ge-01 Ge-98 Ar-96 Ge-97 financial Ar-97 6 Kz-95 Sector Ar-95 Kz-00 Kz-96 4 Ar-98 Kz-01 Ar-00 Ar-99 Kz-97 Ar-01 2 Kz-99 Hg-95 Hg-96 Pd-99 Hg-97Pd-97 Hg-99 0 Pd-95 Pd-01 Hg-01 Hg-98 -1 Pd-00 Pd-98 0 1 2 Pd-96 3 Hg-00 4 5 6 7 8 9 Kz-98 -2 Fiscal deficit, % of GDP Note: Country names are abbreviated as follows ­ Armenia (Ar),Azerbaijan (Az), Georgia (Ge), Hungary (Hg), Poland (Pd), and Kazakhstan (Kz). Source: Authors' calculations based on data provided by local consultants and on the public expenditure database for transition countries (World Bank 2002). Reform Savings Did Not Go to Social Spending For the government the net budgetary impact of reforms is equal to the savings from deficit reduction minus the costs of social assistance to mitigate welfare losses from higher electricity tariffs.7 Unsurprisingly, countries with a high poverty incidence have very low per capita annual expenditures on social assistance (Figure 2.3). 8 If social assistance costs are lower than the budgetary gains from reducing the sector's losses, there appears to be room to increase the size of transfers to offset welfare losses from higher tariffs and for the donor community to plan an increasing role in facilitating social transfer mechanisms and providing technical assistance 7. This is a purely financial analysis of the budgetary impact. An economic analysis of the welfare impact on consumers would also need to take into account the gains to consumers from better service (if an improvement happens concurrently with a tariff increase). 8. Poverty is measured as the percentage of population below the absolute poverty line of US$2.15 (purchasing power parity) per capita per day. Poverty rates are from World Bank (2000b), p. 35. Reported absolute poverty rates are for 1999 in Armenia, Azerbaijan, and Georgia, 1998 in Poland, 1997 in Hungary, and 1996 in Kazakhstan. Social assistance/welfare expenditures are averages for 1997­2000 (calculated in US dollars). 14 WORLD BANK WORKING PAPER FIGURE 2.3: POOR COUNTRIES SPEND LESS ON POOR PEOPLE (1997­2000) 25 as (%) 20 Poland capita capita per 15 Hungary per 10 Kazakhstan GDP assistance of Georgia Azerbaijan Armenia 5 share Social 0 0 10 20 30 40 50 Poverty incidence (%) Note: Social assistance and GDP per capita are averages over 1997­2000. Source: Public Expenditure Database for the transition countries, World Bank (2002). to improve targeting. Of course this does not mean that transfers must be limited to electricity consumption, because it is possible for the public sector to make more productive investments in such other areas as education and health. Estimating the costs of transfers is straightforward when social assistance is provided through special funds rather than through the general social assistance system. Special funds were created in Hungary and Georgia. In Hungary the justification was that the combined social and economic cost of disconnection would be higher than the cost of providing social assistance. In 1997 the Hungarian government contributed $4.6 million and the power companies $2.2 million to this fund. In Georgia the state provides subsidies through a complicated system of categorical privileges. The estimated annual cost of these subsidies at current tariff levels is $4.7 million a year, excluding administrative costs. In addition, in 1999 a program financed by the United States Agency for International Development (USAID) provided electricity in the winter to vulnerable households (World Bank 2003b, 2003c, 2003d). These costs are small compared with the financial sector deficit in Georgia and the surplus in Hungary.9 In Armenia, Poland, and Kazakhstan compensation for tariff increases is provided through the general social assistance program rather than earmarked funds. That makes identifying incremental social assistance costs a formidable task. They can rise because of an increase in social assistance payments per person that can be attributed to the reform or because of an increase in the total number of recipients qualifying for social assistance.10 The savings from a reduction in sector losses in ECA did not go to finance social protection or towards social expenditures on health and education, as it did in Latin America. 9. Clearly, these numbers are not directly comparable. Instead, the reduction in deficits should be compared with the costs of social assistance to mitigate the tariff increase. But the choice of a year would strongly influence the results. In Hungary the sector financial deficit fell by $205 million from 1996 to 1997 and by only $19 million from 1998 to 1999. In Georgia it fell by $5 million from 1998 to 1999 and by $17 million from 1999 to 2000. 10. This requires creating a counterfactual scenario of the poverty rate increase or reduction in the absence of the tariff increase. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 15 FIGURE 2.4: SOCIAL SPENDING ON EDUCATION, HEALTH AND SOCIAL ASSISTANCE DID NOT INCREASE (AS PERCENT OF GDP) 35 GDP 30 of % 25 as 20 15 Spending 10 5 Social 0 1995 1996 1997 1998 1999 2000 Armenia Azerbaijan Georgia Hungary Kazakhstan Moldova Poland Note: Analysis of public expenditures in real terms reveals a similar trend, with falling or stable expenditure levels. Source: Public Expenditure Database for the transition countries,World Bank (2002). The explanation may lie in the different starting points of the countries in these two regions in the 1990s. The ECA countries accumulated enormous quasi-fiscal deficits in various sectors by the beginning of the transition period, and some of the savings from the reforms were probably spent to reduce these obligations. In Latin America fiscal pressure has been associated with an increase in the social spending (Box 2.3). Conclusions The legacy of central planning resulted in a power sector contributing to unsustainable government deficits. So getting the power utilities off the government books is the key to reducing fiscal budget deficits and promoting macroeconomic stabilization and growth in the transition. Sector losses have been falling over the last decade, with tariffs the key determinant. Lessons include placing greater emphasis on ex-ante monitoring of sector deficits. Furthermore, using country-specific cost-recovery tariffs can generate more precise estimates of these losses. This work is currently under way in a separate regional initiative. Despite the decline in sector losses, the impact of reforms on the fiscal deficit is ambiguous. This is due to a combination of factors, including the lack of data, very complex financial and budget flows in the sector, and the different adjustment paths of countries. Only a detailed analysis of the quasi-fiscal accounts would lead to a definitive conclusion about the impact of reductions in the sector loss on the fiscal deficit. On balance, we conclude that assessment of budgetary impact of the reform can be improved by carefully examining the relative share of the energy and non- energy subsectors of the industry. Positive outcomes are expected in countries where a higher share of GDP is from the energy sector. But institutional reform in these countries will also be the most difficult. The savings from lower sector losses did not translate into an increase in social spending. The share of spending on social assistance, education, and health stayed the same or fell. It may be that savings from reforms went to reducing prior sector obligations. Aggregate social assistance data 16 WORLD BANK WORKING PAPER BOX 2.3: SOCIAL SPENDING IN LATIN AMERICA INCREASED SINCE THE 1990S In contrast to ECA, social spending in Latin America has increased 50 percent since the 1990s in the real terms, from the per capita spending level of an average of US$ 360 to $540 (CEPAL 2001). The highest increase took place in the countries with low and medium initial levels of per capita social expenditures. Interestingly, fiscal pressure has helped make public spending more socially oriented, increasing it as a share of GNP. Countries with high level of social spending in Latin America combine high fiscal pressure with a high priority for social spending, and vice versa. The fact that fiscal pressure led to budget cuts in the areas of public expenditure other than social spending shows that these countries pursued good policies. In ECA, fiscal pressure translates into a reduction in social spending both in real terms and in expenditure shares. Policymakers face a tradeoff between coverage and targeting of social assistance expenditures.An analysis of the composition of social expenditures in a sample of Latin American countries suggests that primary education, healthcare, and childcare programs are well targeted and inequality reducing. Pensions, housing, especially urban, and unemployment insurance mechanisms tend to be less pro-poor and less well targeted. Cash transfers, the most direct and non-distortionary way of providing assistance, require well developed administrative capacity and can be difficult to target. The disadvantage of other programs, such as the provision of infrastructure services and education, as a social assistance mechanism is that they are broad in coverage but not well targeted to the poor. Means testing may be a better way to target utility subsidies than lifeline tariffs. But, as suggested in studies reviewed by Estache (2001), the outcome depends on how well means-testing is implemented. To increase the efficiency of delivering social assistance, social protection must be carefully planned, with programs well targeted, well coordinated, and well designed to meet the specific needs of the population. To reduce the costs of means testing, Chile used the same targeting system for all types of programs, with applicants filling out one form that allowed them to qualify for a variety of programs. This approach proved very cost-effective, in 1996 a mere 1.2 percent of the benefits distributed. The inherent problem of social assistance and targeted spending is its highly cyclical nature. Protection falls during crises, when it is needed the most.Wodon and others (2003) estimate that a one percentage point decrease in GDP led to a reduction of two percentage point in targeted public spending per poor person in Latin America in the 1990s. It is therefore advisable for the governments to act in a long sighted way and to save at least some public funds during expansions in order to better protect the poor during recessions. Source: Adapted from Wodon and others 2003. show that these costs are much smaller than the savings generated by a reduced deficit. This implies that a larger share of savings can presumably go toward alleviating the blow from tariff increases. It does not imply that financing consumption (rather than more productive investments in health and education) is the best way to do this. CHAPTER 3 CREATING MORE EFFICIENT COMPANIES E lectricity reforms are expected to improve the production efficiency of the power sector, ultimately translating into cost savings and service quality improvements for end users. They are also designed to make the sector financially sustainable by increasing the efficiency of resource allocation and the cost-effectiveness of sector investments. In theory the gains in effi- ciency materialize from a profit motive­created by more competition or transparent regulation. While the legacy of central planning suggests large potential production efficiency gains, the data collected here are ambiguous. ImproveTransparency and Accountability Identifying the magnitude of production efficiency gains in ECA countries is confounded by changes in record keeping and accounting methods, by vested interests, and by private sector operators with few incentives to report production efficiency gains. Looking forward, Bank oper- ations can improve transparency and accountability by emphasizing a systematic set of indicators in all sector operations and disseminating this information to the public. A best practice example is the Armenian Natural Monopoly Regulatory Commission, which discloses power sector per- formance indicators on the World Wide Web on a monthly basis.11 And similar to the report cards used to elicit responses from citizens on public services in the Philippines and India (Bhatnagar 2001; Paul 1994, 1996), a system of citizen feedback on service delivery can be instituted. Such a mechanism can help create a more direct link between service quality and tariff increases. Another good example is the 2004 World Development Report, which outlines a system of accountability that connects consumers, government and providers through four interrelation- ships--client power, voice, compacts, and management. The translation of improvements in production efficiency into cost savings and better service quality for end-users depends on assumptions about the counterfactual. One conclusion that can 11. Web address: http://rcnm.am. View indicators under "sector reports" link. 17 18 WORLD BANK WORKING PAPER TABLE 3.1: SUGGESTED INDICATORS OF PRODUCTION EFFICIENCY Stakeholder Outcome Objective Outcome Indicator Examples Consumers Improved service · Reduced number of System average interruption quality · outages frequency index1 · Frequency and voltage Number of deviations from · stability established standards Power sector Improved resource · Increased revenue/ Rise in electricity billed as % (utilities) efficiency · collections of net supply; rise in collections as % of billings · Reduced cost of supply Reduction in cost of generation ($/kilowatt hour) · Improved energy efficiency Reduction on fuel use per kilowatt hour of electricity produced · Reduced losses % reduction in (kilowatt hour lost/net kilowatt hour generated) · Improved operational Rise in sales per employee, Rise in · efficiency consumers served per employee Government Increased financial · Increased sector % increase in investment in independence · investment (third party) generation/distribution/ transmission · Reduced sector % decline in sector financial deficit · financial deficit expressed as a share of GDP Source: Authors, based on review of project documents be drawn immediately is that sector operations would benefit from simulations of realistic coun- terfactual scenarios. Assuming the service quality would have declined in the absence of reforms, the benefits that accrue to end-users are costs avoided. With tariffs well below cost recovery in most study countries, there is little to no expectation that efficiency gains would be passed on to end users through lower tariffs. This reinforces the need to place more emphasis on the design and use of outcome-based indicators of production efficiency as conditions in Bank and Fund operations. (Examples of outcome-based indicators are in Table 3.1.) This chapter examines changes in six production efficiency indicators12 over the last decade: Revenue per kilowatt hour (revenue/net generation). Cost of generation ($ per kilowatt hour generated). System losses [kilowatt hour lost/total kilowatt hour generated + (kilowatt hour imported--kilowatt hour exported)]. Collections (revenues over billings). Operational efficiency (sales per employee, and consumers served per employee). Service quality (frequency, voltage, and average hours of service). Revenue per kilowatt hour Rose While profits cannot be determined from the available data, examining the cost of delivered electricity (revenue divided by net generation) provides a back-of-the-envelope estimate of prof- itability. Despite problems controlling losses, collections, and staffing, the revenue per kilowatt 12. Details of tariffs, capacity and efficiency indicators are presented in Annex E. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 19 FIGURE 3.1: NOMINAL REVENUES INCREASED MARGINALLY, 1991­2001 6 5 4 3 2 cents/kilowatt-hour 1 0 91 92 93 94 95 96 97 98 99 00 01 02 Armenia Azerbaijan Georgia Hungary Kazakhstan Poland Source: Authors' calculations based on data provided by local consultants. TABLE 3.2: REPORTED GENERATION COSTS WENT DOWN, 1990­2002 (CENTS/KILOWATT HOUR) '90 '91 '92 '93 '94 '95 '96 '97 '98 '99 '00 '01 '02 Azerbaijan 3.0 1.1 2.8 3.0 0.6 1.9 2.2 2.5 2.3 1.9 1.8 1.8 1.5 Armenia na na na na 1.5 1.7 2.0 2.6 2.3 2.7 2.4 2.3 1.5 Georgia na na na na na na na na 1.1 1.3 1.2 1.0 0.9 Kazakhstan 0.7 na na 1.4 2.2 2.6 3.2 na na 2.5 2.2 1.7 na Poland na 1.9 1.9 1.9 2.3 2.6 2.7 2.6 2.9 3.0 2.9 3.1 na Note: Generation costs are in nominal terms. Source: Authors' calculations based on data provided by local consultants. hour delivered (revenue/net generation) rose slightly from lows in the early to mid-1990s in almost all countries (Figure 3.1). Generation Costs Declined Reported generation costs (cents per kilowatt hour), one of the most important determinants of end-user tariffs, have exhibited a downward trend on average over the last decade. They have ranged from as low as 0.5 cents per kilowatt hour to as high as 4.3 cents depending on the country, year, and (most important) method used to calculate them (Table 3.2).13 The analysis of 13. The economic costs of generation are often underestimated because initial investment expenditure is treated as a sunk cost and depreciation is not usually reflected in tariffs, the opportunity cost of fuel and electricity is disregarded, provisions for maintenance and capital repair are inadequate, fuel is often subsi- dized, and the reported cost of generation (especially in the early and mid-1990s) was often based on the cash accounting method, which records expenditures only when they are paid, an approach that can under- estimate the real cost. 20 WORLD BANK WORKING PAPER FIGURE 3.2: SYSTEM LOSSES STAYED STABLE, 1990­2002 70% 60% 50% 40% 30% 20% 10% 0% 90 91 92 93 94 95 96 97 98 99 00 01 02 Kazakhstan Armenia* Tbilisi, Georgia Azerbaijan Poland Hungary * Expert estimates for 1992­93. Source: Authors' calculations based on local consultant reports. generation costs is complex: it is a function of international fuel prices, fuel and plan availability, dispatch impacts, and generation mix, factors largely beyond the control of the operator. System Losses Held Steady Technical and commercial losses (kilowatt hour lost/net generation) have held steady in the 10­20 percent range of net generation in four countries of the study, despite being the primary targets of donor conditions. The exceptions are Armenia and Georgia (Figure 3.2).14 In Armenia losses fell by 40 percentage points from highs in the early 1990s. In Georgia total losses have not been quantified across the whole country, but anecdotal evidence suggests that commercial losses range between 50 and 60 percent in the regions. In Tbilisi reported losses since 2000 have ranged between 35 and 50 percent. Losses of 10­15 percent, as observed in Hungary and Poland, are consistent with fully commercialized electricity utilities. Continuing corruption and theft and the use of outdated distribution equipment keep losses above the desired levels in other countries. Theft continues to be a problem particularly because the low-voltage distribution sys- tems that dominate much of the CIS make it simple to set up illegal connections to trunk lines. Collection RatesVaried Collection rates (revenues/billings), a function of both systemwide losses and tariffs, have varied widely. They have stayed close to 100 percent in Poland and Hungary, gone down in Azerbai- jan, Kazakhstan, and most of Georgia (except in Tbilisi and Rustavi), and gone up in Armenia (Figure 3.3). Collection rates in Tbilisi have improved for distribution functions previously operated by AES Telasi. There have also been marked improvements in Rustavi under a pilot sponsored by USAID. It is often postulated that private operators are more successful at improving collection rates than the public sector because they are not subject to the same political influences. While there is 14. According to the local consultants and other anecdotal evidence, actual losses in Azerbaijan and Kazakhstan are substantially higher. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 21 FIGURE 3.3: COLLECTION RATES DIFFER BY COUNTRY, 1990­2002 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 90 91 92 93 94 95 96 97 98 99 00 01 02 Armenia Azerbaijan Georgia Tbilisi Kazakhstan Note: Hungary and Poland had close to 100% collections in 1990­2002. Source: Authors' calculations based on local consultant reports. BOX 3.1: PRIVATE SECTOR IMPROVING HOUSEHOLD COLLECTIONS IN GEORGIA AES Telasi has dramatically improved revenues and cash flow since the beginning of 2000. Revenue from the residential sector increased 91 percent from 2000 to 2001 and another 41 percent from 2001 to 2002.1 While tariff increases account for some of the increase, better collections from customers--as well as increases in the amount of targeted and non-targeted subsidies--have also played a role. AES Telasi has been particularly successful at reducing household arrears. Over time there has been steady improvement in collection rates, rising from 44 percent in 2000 to 86 percent in 2002. At times collection rates have even exceeded 100 percent of current billings, as households settled arrears and transfer payments for subsidies were received from USAID or the government.The key instruments used by AES Telasi to achieve improved collection rates from customers include better service quality and metering.The data suggest that metering and subsidies had a much larger impact on collection rates and revenues than service quality and retail prices.2 little doubt that the private sector is effective in using a variety of instruments to increase collec- tions and revenues (Box 3.1), this does not mean the public sector is ineffective. Of the two countries that made the largest improvements, one had a private operator, AES Telasi (Tbilisi, Georgia) and the other, a public operator (Armenia) until very recently. As suggested earlier, one of the most important determinants of private operator performance may be government commitment. Mixed Results on Operational Efficiency Operational efficiency--measured by the volume of sales per employee per year (megawatt-hours per employee) and the number of consumers per employee--rose in Azerbaijan, Hungary, and Poland and fell or stayed the same in Armenia, Georgia, and Kazakhstan.15 In Armenia demand dropped together with the rate of new connections, while staffing was increased particularly in 15. Data are not available to estimate the number of consumers per employee in Kazakhstan. 22 WORLD BANK WORKING PAPER TABLE 3.3: AGGREGATE IMPACT OF REFORM ON COLLECTION RATES IN TBILISI Change Change 2000 2001 2002 2000 to 2001 2001 to 2002 Telasi received power 2.79 2.38 1.20 -15% -6% (million kilowatt hour) Telasi requested power 3.23 2.76 1.29 -14% -20% (million kilowatt hour) Ratio of received to requested power 86% 86% 93% 0 pp 7 pp Average price (lari/kilowatt hour) 0.093 0.100 0.124 8% 24% Portion of households re-metered 38% 69% 76% 32 percentage 7 percentage points points Consumption (million kilowatt hour) 2.35 2.31 2.49 -2% 24% Billings (thousand lari) 217 232 309 7% 33% Total receipts (thousand lari) 96 186 266 93% 44% Subsidies (thousand lari) 35 44 55 25% 26% Winter Heat Assistance Program 29 37 47 28% 27% Government privileges 6 7 8 11% 21% Payments by customers (thousand lari) 61 142 211 132% 49% Collection rate from households 44% 80% 86% 36 percentage 6 percentage points points Note: Table includes only Tbilisi households in the sample. Requested and received power in 2002 covers January to June only. Two tools were used to identify the role of the different instruments in AES Telasi's performance. First, a multi- variate analysis estimated receipts as a function of service quality (ratio of requested and received energy), price, enforcement (percentage of households that have been re-metered), and subsidies.The analysis also controlled for monthly temperature and the temporary loss of thermal power plants in the winter of 2001. The second tool was focus group discussions, which solicited the views of Telasi customers on a wide range of issues related to improved payment levels, including re-metering, enforcement, and service reliability. The multivariate results indicate that re-metering and price are equally important determinants of receipts, followed by quality and subsidies.The analysis also shows that collections increase at a decreasing rate with re- metering, suggesting that re-metering may yield higher revenues than prices in the initial stages of reform.The cost of meters is not taken into account in this analysis. 1. These figures are for a sample of 1,349 households included in the Georgia Household Budget Survey. In total,AES Telasi has approximately 300,000 customers. Households participating in the HBS were randomly selected, and may be presumed representative of the total population of households in Tbilisi. 2. The cost of meters is not taken into account in the analysis. Source: Lampietti and others, 2003 the distribution sector in preparation for privatization. This reduced operational efficiency. In Georgia staffing levels declined (except in privatized distribution entities) in parallel with a decline in demand and in the rate of new connections. Operational efficiency in Georgia has remained fairly stable over the review period but is low in comparison with other countries of this study. But any conclusion from this indicator depends on the labor exit policy and state actions prior to sale. It is possible that the utility cannot layoff workers without informing the government or that the government might have made a contractual requirement during sale not to layoff employees. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 23 Focus on Service Quality Licenses and share purchase agreements are often based on mandatory investment levels rather than service quality indicators. Some anecdotal evidence suggests that this has encouraged artifi- cial inflation of reported investments. Armenia is one of the few countries that have linked investments to quality standards16 rather than introducing mandatory minimum investment requirements. Even though service quality indicators would be important for end-users, there are no official statistics on service quality for the countries in this study. Unofficially, the number of hours of electricity supply improved in most CIS countries. Voltage fluctuations, frequency, and outages have normalized in all countries (Table 3.4). Establishing cross-border transmission links, allow- ing networks to run in parallel, has increased grid safety and supply quality. Regional CompaniesTake on the Challenge Private investment in the sector has been falling steadily since 1997, and a number of private operators, embroiled in contractual disputes, have withdrawn or threatened to withdraw. In Kazakhstan the Belgian investor Tractabel walked out after tariff disputes with the govern- ment. In Moldova the state initiated a lawsuit against Spanish investor Union Fenosa, argu- ing that the privatization process was flawed. In Georgia US-based AES described its pur- chase of the Tbilisi distribution company as a mistake and recently sold its stake in the company to Russian interests. This behavior suggests that expected profits are not materializ- ing for reasons beyond the operators' control or due to opportunistic behavior by private operators (Box 3.2). Yet, companies with less international and operational expertise--and less financial strength but better local knowledge--have entered the market and demonstrated a good track record so far. It is too early to draw conclusive evidence, but in Armenia, Azerbaijan, and Kazakhstan, various management and concession contracts and other privatization arrangements with regional investors are in place. This suggests that the conventional definition of "strategic investor" may have to be reconsidered. In Armenia the distribution subsector was privatized in late 2002 to Midland Holding Resources (registered in the Channel Islands). Although this company has no previous power sector experience, it worked in the former Soviet Union, and its performance to date has been very satisfactory. In Azerbaijan management contracts are being introduced to improve collections with the aim of achieving a 100 percent collection rate by contract expiry. One example is the con- tract with the Turkish company Barmek, which has assumed management of the Baku Electricity Network. Under Barmek's contract 50 percent of electricity purchased from Azerenerji in 2002 has to be paid through collections. This must eventually reach 100 percent over a four-year period. In Kazakhstan most power plants had been privatized. Large national-scale power plants were sold to strategic investors. High capacity hydropower plants were given for concession. Industrial combined heat and power (CHP) plants were taken over by the enterprises they served. General-purpose low capacity CHP plants were transferred to utility ownership. And medium and high capacity CHP plants were privatized. 16. Recently, the Armenian Energy Regulatory Agency asked the World Bank to finance technical assis- tance for developing these standards and mechanisms to monitor and enforce them. 24 TABLE 3.4: SERVICE QUALITY IMPROVED W ORLD 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Armenia B ANK Voltage fluctuation,V 200­220 000­220 160­220 160­220 160­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 Frequency, Hz 000­050 000­050 0048­50 0048­50 0048­50 000­050 000­050 000­050 000­050 000­050 000­050 000­050 W Hours of service 000­024 000­024 1.5 1.5 000­018 000­024 000­024 000­024 000­024 000­024 000­024 000­024 ORKING Kazakhstan Voltage fluctuation,V 200­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 P APER Frequency, Hz 000­050 000­050 000­050 000­050 000­050 000­049 000­050 000­050 000­050 000­050 Hours of service 000­024 000­024 000­024 00­024 000­024 000­024 000­024 000­024 000­024 000­024 000­024 000­024 Azerbaijan Voltage fluctuation,V 200­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 000­220 Frequency, Hz 000­050 000­050 000­050 000­050 000­050 000­050 000­050 000­050 000­050 000­050 000­050 000­050 Hours of service 000­024 0016­24 0016­24 0016­24 000­024 000­024 000­024 000­024 000­024 000­024 000­024 000­024 Georgia Voltage fluctuation,V 200­230 200­230 190­230 190­230 170­240 170­240 170­240 190­230 190­230 190­230 190­230 190­230 Frequency, Hz 000­050 000­050 0048­52 0048­52 0048­52 0043­52 0043­52 0043­52 0045­52 0045­52 0046­52 0048­52 Hours of service 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a 0000.n/a Note: The hours of service in Azerbaijan and Georgia are for the capital cities only. In rural areas, service interruptions persist. Source: Consultant reports. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 25 BOX 3.2: OPPORTUNISTIC BEHAVIOR BY THE PRIVATE SECTOR? "Almost by definition, certain features of regulated sectors make them more prone to renegotiation. First, regulation constrains the actions that a concessionaire can take, the most important being the setting of tariffs. Second, tariffs are expected to be set so that they allow the concessionaire to earn reasonable profits. When firms are not able to earn expected returns, it is rational for them to expect a change in contract terms. This is the premise behind the so-called financial equilibrium clause implicit or explicit in most concession service contracts and legislation.That is in principle a valid pillar of any concession contract, since private investors should be allowed to earn a fair rate of return on their investments. Having said that, it is also correct that the financial equilibrium ought to be subject to a number of provisos, including a conditioning to efficient operation. Yet the costs of providing services are rarely linked to a benchmark of efficient operations. And when they are, such costs are often disputed. "Strategic underbidding (or overbidding, depending on award criteria), to some extent encouraged by the incompleteness of contracts, also may explain the high proportion of renegotiation. As noted, many firms have won concession contracts by strategically underbidding (or overbidding), with the expectation that they would be able to renegotiate in the future, and governments have often been unable to commit to enforcing these agreements. If all potential bidders account for that possibility, an auction could still elicit the most efficient operator--but with significant underbids or overbids. But there are two problems with that argument. First, because renegotiation is a bilateral negotiation, the final outcome need not be guided by efficiency and welfare concerns, and rents could be transferred. Second, although any potential operator could submit a bid with the expectation of renegotiation, expectations might vary among bidders and not necessarily be correlated with their efficiency. Moreover, some enterprises may possess a systematic advantage in renegotiation and so be more likely to win a concession through underbidding (or overbidding). As a result a firm with high affiliation and high costs could win an auction." Source: Guasch (2004). Conclusions Electricity reforms are expected to improve the production efficiency of the power sector, ulti- mately translating into cost savings and service quality improvements for end users. They are also designed to make the sector financially sustainable by increasing the efficiency of resource alloca- tion and the cost effectiveness of sector investments. The efficiency gains materialize from a profit motive--created by more competition or transparent regulation. While the legacy of central planning suggests large potential production efficiency gains, the data collected here on cost of generation, system losses collections, and operational efficiency are ambiguous. Despite continuing problems with losses, collection rates, and staffing, overall revenue per kilowatt hour has increased in almost all countries. The beneficiary of additional revenues from the growing gap between declining generation cost and rising retail tariffs is not so clear. Limited aggregate data suggests that service quality has improved in a number of capital cities. Simple ex-ante simulations would allow better quantification of potential reform benefits and costs. And placing greater emphasis on outcome-based indicators of service quality will help ensure that future operations produce the intended end-user benefits. Both the private and the public sector can improve utility performance, suggesting that a broad range of contract arrange- ments can produce the desired outcomes. Finally, a number of local investors with both region and sector experience are filling investment needs, suggesting alternatives to the traditional defini- tion of a strategic investor. CHAPTER 4 MAINTAINING POWER TO THE POOR E CA is unlike other regions of the world in that the socialist system gave almost all house- holds access to reliable, inexpensive electricity. So, the welfare gains from increased access, one of the most immediate and tangible benefits of power sector reforms in other regions (Estache and others 2002; Galal and others 1994), is not a consideration in most ECA countries. The electricity sector reforms have confronted households with a tradeoff between prices and service quality, and the perception of high political costs has had an adverse impact on the reform progress. Tariff increases and disconnections are very unpopular, and the public often views the sale of state assets to the private sector with skepticism. This is especially true when the public expects that the privatization of a public monopoly would result in a higher price without any improvement in the quality of service. If there is an improvement in service quality, it takes place a few years after privatization, while the price rises early on in the process. This mismatch between the timing of the gains and costs of reform and the uncertainty about the gains causes consumers to be skeptical. Experience in Latin America shows that privatization can succeed when the social issues are well addressed--and fail when combined with inadequate analysis of social and distributional issues and a lack of political will to support the reform.17 Even though the official statistics and household surveys suggest that access to service is nearly universal, supply is often rationed. The low investment in electricity generation and 17. Some anecdotal evidence from Latin America and the ECA countries illustrates this. One example is a failed privatization of the water and sanitation services of Tucuman in Argentina in 1995. The concessionaire bid an immediate 68 percent tariff increase to fund the required investment program. The tariff increase was to affect all consumer groups equally. In response, consumers organized a nonpayment campaign, forcing the investor out of business (Estache and others 2001). Another example is the failed privatization of the Almaty electricity distribution company in Kazakhstan in 1996. The contract specified a tariff increase in return for investment requirements by the concessionaire. Facing strong opposition to the tariff increase, the government failed to fulfill its contractual obligations, and in response the concessionaire did not make the specified investments and walked away after arbitration in 2000 (news sources). 27 28 WORLD BANK WORKING PAPER TABLE 4.1: URBAN NETWORK ENERGY USE IN ECA (PERCENT OF HOUSEHOLDS) Central Heat Central Gas Electricity Country Poor Nonpoor Poor Nonpoor Poor Nonpoor Armenia (1999) 11 14 4 16 97 99 Croatia (1997) 15 39 19 30 99 100 Kyrgyz Rep. (1999) 17 55 13 33 100 99 Latvia (1997) 70 83 57 68 99 100 Lithuania (1998) 31 46 47 56 85 94 Moldova (1999) 17 57 37 70 65 89 Tajikistan (1999) 1 1 3 6 100 100 Source: Lampietti and Meyer (2002). distribution infrastructure in the last decade resulted in severe deteriorations in service quality. Some countries experience frequent interruptions in electricity supply. Others experience voltage fluctuations that destroy household appliances (Markandya and others 2001). Unless investments are made in rehabilitation and maintenance of the infrastructure, households may experience widespread supply shortages in the future.18 This also leads us to our first conclusion that greater emphasis must be placed on explicitly linking tariff increases with improved service quality in order to minimize negative welfare effects. This is also likely to generate more political will to support the reform. Household Energy Use Patterns Differ There are systematic differences in energy use in urban areas in the ECA region. Separating network and non-network energy use provides insight into energy use patterns (Table 4.1). Almost all households use electricity, with small differences between the poor and the nonpoor. But poor people use much less central heat and gas. Are the nonpoor more likely to use network energy because they have better access to the network, or is it because they make different choices? Although this question cannot be answered with the data from household surveys, it points to the need for country-specific analysis to identify the supply constraints--such as network location and capital equipment (such as gas heaters)--that limit poor people's access to network energy. If poor people are not using network energy, what are they using? Primarily dirty non-net- work energy (Table 4.2). Wood and coal use are consistently higher among the poor--except in Tajikistan, where coal is heavily subsidized for everyone. Except in Latvia, the nonpoor are more likely to use liquefied petroleum gas (LPG), the cleanest non-network energy. The poor may favor dirty non-network energy because it is less expensive or because they do not have the resources to spend on network appliances. Yet as noted, burning dirty fuels has social costs-- mainly air pollution and deforestation--that require careful, country-specific analysis of the economic implications of raising the price of clean energy (Lampietti and Meyer, 2002). 18. Cambridge Energy Research Associates estimates that half of Russia's generation capacity must be retired in the next 20 years as it reaches the end of its productive life, while more than the total installed generation capacity of France needs to be added. If these investments are not made, Russia is expected to suffer from nationwide electricity shortages in the near future. ("Russian Electricity: In Need of Shock Ther- apy." The Economist. August 31, 2002. pp. 50­51.) POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 29 TABLE 4.2: URBAN NON-NETWORK ENERGY USE IN ECA (PERCENT OF HOUSEHOLDS) LPG Kerosene Coal Wood Country Poor Nonpoor Poor Nonpoor Poor Nonpoor Poor Nonpoor Armenia (1999) 17 27 14 11 n/a n/a 47 50 Croatia (1997) 44 45 3 7 1 1 51 26 Kyrgyz Rep. (1999) 24 39 31 17 60 31 46 22 Latvia (1997) 37 28 n/a n/a <1 <1 1 2 Lithuania (1998) n/a n/a n/a n/a <1 <1 1 2 Moldova (1999) 6 7 n/a n/a 9 5 12 9 Tajikistan (1999) n/a n/a <1 1 11 18 47 32 Note: n/a--not available from household survey Source: Lampietti and Meyer (2002). Tariffs Rose In an effort to reach cost-recovery levels, the residential electricity tariffs were increased in all countries19 In nominal dollar terms they nearly doubled in Azerbaijan, Georgia, and Hungary between the mid-1990s and 2002; Poland and Hungary currently have the highest tariffs. In Kazakhstan, Azerbaijan, and Moldova the tariffs first increased or remained stable and then fell in the late 1990s due to the depreciation of local currency and the 1999 devaluation triggered by the financial crisis in Russia. In real terms the tariff increase during this period was the highest in Azerbaijan and Georgia, where the tariffs quadrupled and tripled, respectively. By 2002 the tariffs doubled in Armenia and Moldova, increased by about 160 percent in Hungary, remained unchanged in Poland and fell in Kazakhstan (Figure 4.1). Such substantial tariff increases are unlikely to be welfare neutral unless accompanied by improvements in service quality or cush- ioned by income transfers. A complete analysis of the tradeoffs between tariffs and service quality requires comparing the welfare losses from price increases with the gains from improved service quality (which may take place with a lag). As noted in chapter 3, aggregate service quality is thought to have improved since the early 1990s, the crisis period in Armenia, Azerbaijan, and Georgia, implying welfare gains for consumers.20 In Kazakhstan service quality remained unchanged. But it is not clear what the situation would have been in the absence of privatization of generation facilities, which were in dire need of investments. It is also not clear how much of the improvement, where it did take place, can be attributed to the general political and economic changes, and how much to power sector reforms. So instead of evaluating the welfare gains from service quality improvements, the welfare impact of reforms is first assessed by examining changes in the share of electricity expendi- tures in income (using total expenditure as a proxy), particularly for the poor and nonpoor and then by assessing the loss in consumer surplus. 19. In Kazakhstan tariffs also increased in nominal terms after 1993, but the increase was negligible, unlike the other countries in this study. 20. In Georgia supply improved in Tbilisi after privatization of the distribution company. The company was not always able to provide 24-hour supply for reasons beyond its control, such as interconnectedness of the power grid with the rest of the country. Power supply in other areas, where distribution is still publicly owned, is worse. (Project Appraisal Document for an Electricity Market Support Project in Georgia, World Bank, April, 2001. p. 5) 30 WORLD BANK WORKING PAPER FIGURE 4.1: RESIDENTIAL ELECTRICITY TARIFF--INDEX OF CPI ADJUSTED REAL TARIFFS 400% 350% 300% 250% 200% Index 150% 100% 50% 0% 95 96 97 98 99 00 01 02 Armenia Georgia Hungary Kazakhstan Moldova Poland Note: Nominal tariffs in local currency were adjusted by the CPI. Base year is 1995 (1996 for Georgia). 1995 is the first year for which data for all countries are available. In Poland and Hungary most of the tariff increase occurred before that.Azerbaijan is not included in the graph.Though its nominal tariffs, when calculated in US cents/kWh, is among the lowest in the regions, the CPI adjusted real tariff is the highest. Source: Authors' calculations based on data provided by local consultants and Counterpart International (for Moldova). The Burden Increased Mean expenditures on electricity range from 2 to 10 percent of total expenditure on average, with a high of 14 percent for the bottom quintile in Armenia in 2001 and a low of less than 1 percent for the bottom quintile in Kazakhstan in 2001 (Table 4.3). The share increased slightly in all countries and for all welfare quintiles.21 Bottom quintile annual change in electricity share varied between -1.9 percentage points (Poland, 2000) to 41 percentage points (Moldova, 1999). The results are conditional on the households reporting positive expenditures on electricity.22 So, the quality of electricity expenditure data collected in the household budget surveys and living standards meas- urement studies affects the ability to draw generalizable conclusions. The increasing share of expenditures on electricity can be explained by rising tariffs, falling income, reduced rationing (as service quality improved), or inelastic demand. That the share falls monotonically across the welfare distribution is consistent with findings that the poor devote a 21. Average yearly change in electricity shares was positive for bottom, top, and all quintiles for Georgia (1997­2002), and Poland (1994­2001). For Moldova, it declines from 2000. There was a similar positive change in electricity shares for all quintiles in Armenia (between 1999 and 2001), Hungary (between 1993 and 1998, 1998 and 1999), and Kazakhstan (between 1996 and 2001). 22. The share of households with zero expenditures varies across surveys due to different survey design. Surveys in which respondents are asked about the previous month's expenditures frequently have over half the respondents reporting zero expenditures, a result more likely due to the nature of the question than to the fact that none of them pay for electricity. If the whole sample is included, it is not clear whether the calculated shares of electricity expenditures in income vary due to different survey structures or because the share actually changed. So only the subsample with households with positive expenditures is included. A series of tests reveals that these households are systematically different from the other households in the sample. Household size and total expenditure levels are similar for both groups, while the share of urban households is substantially higher in the group with non-positive expenditures (excluded from the calcula- tions). The sample thus underrepresents urban households. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 31 TABLE 4.3: SHARES OF SPENDING ON ELECTRICITY WENT UP, 1993­2002 (MEAN OF HOUSEHOLD ELECTRICITY SHARES) 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Armeniaa Bottom quintile na na na 18.2 na na 10.2 na 13.9 na Top quintile na na na 3.6 na na 5.3 na 6.5 na Mean na na na 9.0 na na 7.2 na 8.8 na Georgia Bottom quintile na na na 3.1 3.8 5.7 4.4 3.7 3.6 6.3 Top quintile na na na 1.5 1.5 1.6 1.5 1.5 2.3 2.0 Mean na na na 2.0 2.3 2.9 2.5 2.4 2.7 3.4 Hungary Bottom quintile 4.3 na na na na 5.6 6.3 6.5 na na Top quintile 2.2 na na na na 3.6 3.7 3.7 na na Mean 3.4 na na na na 4.9 5.4 5.4 na na Kazakhstan Bottom quintile na na na 0.16 na na na na 0.92 na Top quintile na na na 0.11 na na na na 0.56 na Mean na na na 0.14 na na na na 0.69 na Moldova Bottom quintile na na na na 6.5 7.6 10.8 8.4 7.7 6.3 Top quintile na na na na 2.7 3.2 4.2 4.1 3.7 3.0 Mean na na na na 4.1 4.6 6.1 5.5 5.2 4.3 Poland Bottom quintile 4.5 4.6 4.6 4.5 4.5 6.6 7.2 5.3 5.8 na Top quintile 2.4 2.3 2.4 2.3 2.1 3.5 3.7 2.6 2.9 na Mean 3.4 3.4 3.4 3.3 3.3 3.8 4.1 4.1 4.3 na a. For 1999 the results reported are based on 1999 Household Budget Survey.The results for the 1999 Energy Survey for Armenia are similar: the average expenditure share is 6.2 percent, that of the bottom quintile 8.8 percent, and that of the top quintile 5.0 percent. Source: Authors' calculations from household survey data. higher share of total expenditures to energy (Lampietti and Meyer 2002) and that electricity is a necessary good. It also implies a greater proportionate welfare loss for the poor and a more active search for substitutes when electricity tariffs increase. The overall welfare impact of the reforms can be measured by the change in consumer sur- plus. Consumers gain from an improvement in service quality and the removal of rationing, but lose from an increase in price. Since we cannot measure the gains from the service quality improvement from the existing data, we focus on the consumer surplus change of a price increase. The magnitude of the welfare effect of a price increase depends on the household's dependence on the energy source, measured by its budget share, the price change, and the household's access to substitute energy sources and other goods and services. The third is meas- ured by the elasticity of demand. For electricity the elasticity of demand is typically low: it has been estimated between -0.08 and -0.32 for a range of countries (Hope and Singh, 1995).23 The 23. Electricity price increase of the late 1980s for Zimbabwe, Colombia, and Turkey. 32 WORLD BANK WORKING PAPER TABLE 4.4: CONSUMER SURPLUS FELL Starting Point Lost (gained) Consumer Surplus Real Price Increase Budget Share (Percent ofTotal Household Budget) Country Period % Change % h = -0.1 h = -0.5 h = -0.9 Armenia 1996­2002 98% 9.0 -8.38 -8.39 -8.40 Georgia 1996­2002 267% 2.0 -5.33 -5.33 -5.33 Hungary 1993­2002 57% 3.4 -2.36 -2.36 -2.36 Kazakhstan 1996­2002 2% 0.2 0.00 0.00 0.00 Moldova 1997­2002 43% 4.1 -0.17 -0.17 -0.17 Poland 1994­2002 9% 3.4 -0.14 -0.14 -0.14 Source: Budget shares calculated from household survey data. budget share of electricity in household income varies depending on income and the geographic location of a household. Typically network energy budget shares are inversely related to income, with the poor urban households spending a highest share of their total income compared to all other groups of households (Hope and Singh 1995). This is also the case in these countries in ECA. We present the consumer surplus calculation24 in a range of demand elasticity scenarios. Zero elasticity means that the household is unable to adjust, so this provides an upper bound of the welfare loss. At a price elasticity of -1, the household is able to reduce electricity consumption in response to the price increase, so reducing the magnitude of the welfare loss. Households in Armenia and Georgia experienced the largest welfare loss, expressed as a percentage of total household budget, because of the high magnitude of the price increase and the high shares of electricity expenditures in the total budget (Table 4.4). In real terms the electricity price fell slightly in Kazakhstan, so households did not experience a welfare loss. Consumption is Low Studying consumption using household survey data is confounded by the presence of arrears (nonpayments), which make it impossible to determine whether reported electricity expenditures represent current or historical consumption. But it is possible to examine a sample of household electricity consumption records from distribution companies in Armenia and Georgia.25 These data do not take into account households not included in the utility's database and thus might be an underestimate of household consumption. In Armenia, where service quality has been consis- tently high since 1996 (provided 24 hours a day, seven days a week), mean household consump- tion fell steadily from an average of 160 kilowatt hours (kWh) a month in 1998 to 117 kWh a month in 2001.26 In Tbilisi, where service quality has been improving since 2000, consumption remained constant at about 150 kilowatt hour a month (Figure 4.2). Greater seasonal fluctuations 24. In most cases, the welfare loss or gain calculation includes information on access. For the ECA region, access to electricity is close to 100 percent. The welfare loss calculation should also include informa- tion on disconnection as a result of price rise, but we could not get accurate data. 25. In Armenia, the sample consists of (usable) monthly records for 1,197 households for the period from March 1998 to September 2002, and in Georgia for 288­408 households, depending on the month, for the period from January 2000 to September 2002. Data were provided by electricity utilities. 26. The median consumption for the same period fell from an average of 143 kWh a month to 95 kWh a month. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA33 FIGURE 4.2: ENOUGH FOR THREE LIGHT BULBS AND A REFRIGERATOR 300 250 kWh/month 200 150 100 consumption, 50 Median 0 Jan-00 May-00 Sep-00 Jan-01 May-01 Sep-01 Jan-02 May-02 Sep-02 Tbilisi, Georgia Yerevan, Armenia Chisinau, Moldova Note: Calculations include only households with positive electricity consumption. Source: Authors' calculations using AES Telasi data (reported in Lampietti and others (2003)) and electric utility records for 1,197 households in Yerevan and 2,092 households in Chisinau. in electricity consumption in Armenia than Georgia suggest greater access to inexpensive substi- tutes for heating in Georgia. Household electricity consumption is close to basic minimum needs, sufficient only for light- ing and refrigeration.27 The median consumption was as low as 84­100 kilowatt hour a month during 2002 in Armenia and Georgia. The assumption is that an electricity demand function is kinked, as is characteristic of necessary goods, sloping steeply around the minimum required for basic needs and then rapidly leveling off as the quantity of electricity consumed moves from necessity to luxury. This leads to two conclusions. First, the welfare losses from a price increase are high for households with very low electricity consumption. Second, there may now be little scope for efficiency gains from the household sector in Armenia, Georgia, and Moldova. There was such scope at the beginning of the transition because the former CIS countries have traditionally been energy intensive, but the move toward cost-recovery tariffs and regulatory improvements had little scope for further reductions in consumption. The decline in household incomes associated with transition also added to the decline in electricity use. If the price of electricity increases further, there will only be a small reduction in household energy consumption. Gas May Be Filling the Gap Electricity is the single largest item in the household energy budget in Armenia, Georgia, Moldo- va, and Poland, and the share is largest for the bottom quintile (Table 4.5). Assuming that it consumes less than the top quintile, this implies that electricity is used for the most basic needs for which other fuels are poor substitutes, so demand by the poor is inelastic. 27. A refrigerator (manual defrost, 5­15 years old) consumes about 95 kWh a month and 3 incandescent light bulbs another 30 kWh a month. 34 WORLD BANK WORKING PAPER TABLE 4.5: ELECTRICITY EXPENDITURE AS A SHARE OF TOTAL ENERGY EXPENDITURE, 1993­2002 (MEAN OF HOUSEHOLD ELECTRICITY SHARES) 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Armeniaa Bottom quintile na na na na na na 73 na 96 na Top quintile na na na na na na 57 na 88 na Mean na na na na na na 62 na 90 na Georgia Bottom quintile na na na 83 65 68 68 68 67 na Top quintile na na na 54 38 41 40 40 45 na Mean na na na 65 48 51 53 53 53 na Hungary Bottom quintile 32 na na na na 37 38 39 na na Top quintile 33 na na na na 37 38 39 na na Mean 31 na na na na 36 37 38 na na Kazakhstan Bottom quintile na na na 41 na na na na 47 na Top quintile na na na 30 na na na na 65 na Mean na na na 34 na na na na 56 na Moldovab Bottom quintile na na na na 86 87 92 89 87 86 Top quintile na na na na 80 74 77 81 79 76 Mean na na na na 83 80 84 86 84 81 Poland Bottom quintile 62 61 61 60 60 68 68 67 69 na Top quintile 35 33 33 32 32 39 41 43 44 na Mean 47 45 45 43 44 39 40 43 41 na a.The 2001 survey, unlike the 1999 survey, does not capture well energy sources other than electricity. b.The share is high in Moldova because none of the surveys captures a larger portion of other energy sources. Source: Authors' calculations from household survey data. Given the large price increases and the inelastic demand, it is surprising that expenditures went up only 1.5 percentage points on average. The impact may have been mitigated by improvements in service quality or substitutions of other energy sources. While there are no perfect substitutes for electric lighting, refrigeration, and television, given a choice of substitutes for electricity in heat- ing and cooking, households are likely to choose natural gas because it is clean, convenient, and the low priced. Other alternatives--such as kerosene, coal, and wood--are less convenient. Relative fuel prices and fuel availability influence household energy consumption choices. Even at full import prices gas is substantially less expensive than electricity (Figure 4.3). While there may be additional costs associated with the technology required to use gas (metering and gas-fired appliances), the convenience and savings suggest that, given access, it is the household fuel of choice. Back-of-the-envelope calculations confirm the rising use of natural gas. In Armenia residential consumption of natural gas more than tripled from 1996 to 2001 (from 29,000 tons of oil equiv- alent to 90,000), while monthly electricity consumption dropped from 187,000 tons of oil POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 35 FIGURE 4.3: ELECTRICITY TARIFFS ARE HIGHER THAN THOSE FOR GAS, 1992­2002 5 4.5 4 3.5 3 2.5 2 Cents/kilowatt-hour1.5 1 0.5 0 92 93 94 95 96 97 98 99 00 01 02 Gas tariff Electricity tariff Note:The applied conversion factor was 277.8 kWh per Giga-Joule (GJ) of natural gas (International Energy Agency).Average tariffs were calculated for Armenia,Azerbaijan, Georgia, Hungary, Kazakhstan, Moldova, and Poland. Note that this is a simple average.The number of observations varies by year depending on data avail- ability. Source: Authors' calculations based on data from local consultants, Counterpart International (for Moldova), and ERRANET database equivalent to 106,000.28 For Georgia the number of gas connections in the capital quadrupled from 2000 to 2003.29 Given the potential welfare losses of rising electricity tariffs households cooking and heating with electricity should be encouraged to use clean and inexpensive substitutes, such as natural gas. This can be done through a variety of instruments, as long as the government explicitly com- pensates the utility for any social transfers it provides. An example would be bidding out competi- tive subsidies to encourage the extension of natural gas networks to poor neighborhoods. Enforcement is Necessary Nonpayments, or arrears, are one of the most vexing problems. A key reform objective has been to resolve it, and collection rates indeed appear to rise after privatization (Box 3.1).30 Understanding who accumulates arrears has important implications for the welfare effect of reforms. If it is mainly the poor, affordability may be a problem and special care must be taken by the state to provide adequate assistance to the poor. If it is all households, free-riding may be the problem and stricter enforcement will not disproportionately hurt the poor. 28. Total residential consumption from the energy balance data in Armenia (Ministry of Energy). Con- verted to kilowatt hours using the conversion factor of 1,000 kWh = 0.086 tons oil equivalent, this is equiv- alent to an increase in natural gas consumption from 337 million kWh in 1996 to 1,046 million in 2001, and a reduction in electricity consumption from 2,174 million kWh in 1996 to 1,232 million in 2001 (con- version factor is from the World Energy Council). 29. Tbilgazi's customer base increased from 39,000 households in June 2000 to 164,000 households in January of 2003 (Lampietti and others 2003). 30. Lampietti and others (2003) reach a similar finding, using collection rates (figure 13, p. 30). 36 WORLD BANK WORKING PAPER FIGURE 4.4: BURDEN OF ARREARS IS THE SAME FOR THE POOR AND NON-POOR, 2000­2002 paid 2000 not 1800 has 1600 1400 1200 household 1000 which 800 for 600 kWh 400 200 0 Cumulative 2000-I II III IV 2001-I II III IV 2002-I II III IV Bottom quintile Top quintile Source: Authors' calculations based on AES data. Evidence on the absolute amount of cumulative arrears (in kilowatt hours) in Georgia sug- gests that the problem may be free-riding. In some periods cumulative arrears31 of the bottom quintile exceed those of the top quintile,32 but in others there is no statistical difference between the two groups (Figure 4.4). However, the cumulative arrears of the bottom quintile are higher than those of the top when arrears are measured relative to electricity consumption. The top quintile consumes more than the bottom quintile in 6 of 12 quarters (all quarters of 2002), and this difference is statistically significant.33 Searching for BetterTransfers In addition to improving service quality, the government can mitigate the welfare effects of price increases by providing assistance to vulnerable households and by stimulating income growth. There is much debate about the validity of each assistance measure. Lovei and others (2000) found that instruments that perform well on some criteria tend to perform poorly on others. Further- more, not all subsidy mechanisms are applicable or perform equally well across all countries and utility services. So no single instrument has been identified that would outperforms all others. One of the most contentious debates is between lifeline tariffs, which subsidize an initial block of electricity for all users, and direct income transfers. Proponents of direct transfers argue that lifelines are not targeted and thus encourage inefficient energy use. Opponents claim that 31. Cumulative arrears take into account that some households may carry arrears for many months before being disconnected. 32 According to t-test results, this difference is significant in the fourth quarter of 2000 and in the third and fourth quarters of 2001. In the fourth quarter of 2000 arrears of the top quintile exceed those of the bottom, and in 2001 the reverse is true. 33. According to t-test results, this difference is significant in the fourth quarter of 2000, the fourth quarter of 2001, and in all quarters of 2002. POWER'S PROMISE ELECTRICITY REFORMS IN EASTERN EUROPE AND CENTRAL ASIA 37 BOX 4.1: SIMULATION OF ALTERNATIVE SUBSIDY Energy subsidies to Georgian households are available through a range of programs. One provides all veterans and pensioners between 35 and 70 kWh per household a month (recently increased to 240 kWh a month in the winter and 120 kWh a month in the summer). Refugees and internally displaced persons (including those not living in collective centers) also receive substantial quantities of free electricity. Other government programs provide households 850 m3 of natural gas a year. In addition to the government-funded subsidies, a major donor-financed subsidy to electricity customers--Winter Heat Assistance Program (WHAP)--has been in place for the past five years.WHAP finances the supply of electricity to low-income households in Tbilisi for winter heating during the January­April period.The amount each household receives has varied each year depending on the funding available. It was 850 kWh in 2000 and 1000 kWh in both 2001 and 2002. The proposed subsidy would reach a higher percentage of low income households than either of the existing subsidies (Table 4.6). It would also reach a higher percentage of the other quintiles as well.The absolute subsidy to each household would be substantially lower than in either of the existing programs.The total cost would fall between the WHAP and the government program.The new program would thus be more cost- effective (in Georgian Lari per household) than either of the existing programs. The simulation illustrates an alternative subsidy design, but there are several important caveats. First, the cost of the proposed subsidy would increase as the old subsidy is phased out, reducing some of the fiscal benefits.This is because more households are likely to consume in the 75 to 125 kWh range.At the same time, poverty targeting may well improve as the old subsidy is phased out.With the loss of the existing subsidies, consumption will be based more directly on actual household income. Second, several well organized stakeholders are encouraging the government to keep the subsidies in place, including veterans (who do not wish to lose their benefits) andTelasi (which presumably enjoys the simplicity and predictability of payments associated with the current system).Third, these results are based on data fromTbilisi, and caution must be taken in generalizing them to the rest of the population. FIGURE 4.5: ALTERNATIVE SUBSIDY SCHEME kWh per Subsidy = Q2-Qi if Q1