Report No. 48993-LB Lebanon Social Impact Analysis - Electricity and Water Sectors June 18, 2009 Social and Economic Development Group Middle East and North Africa Region Document of the World Bank StandardDisclaimer: This volume is a product o f the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views o f the Executive Directors o f The World Bank or the governments they represent. The World Bank does not guarantee the accuracy o f the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part o f The World Bank concerning the legal status of any territory or the endorsement or acceptance o f such boundaries. LEBANON Social Impact Analysis-Electricity and Water Sectors Table of Contents ACKNOWLEDGMENTS EXECUTIVESUMMARY ......................................................................................................................................... i ELECTRICITY ........................................................................................................................................................... 11 WATER ............................................................................................................................................................ IV CONCLUSIONS ........................................................................................................................................................ VI CHAPTER1 INTRODUCTION . .............................................................................................................................. 1 OBJECTIVES............................................................................................................................................................. 1 CONTEXT ..................................................................................................................................... METHODOLOGY ...................................................................................................................................................... 2 REPORT STRUCTURE ................................................. ..................................... .............................................. 3 CHAPTER2 ELECTRICITY . .................................................................................................................................. 4 I INTRODUCTION 4 I1 KEYSECTOR ISSUES .......................................................................................... .. .............................................................................................................................................. -ASUMMARY OVERVIEW 5 I11 HOUSEHOLD ENERGYDEMAND AND CONSUMPTION ............................................................ 7 IV 12 V... PUBLIC ELECTRICITY TARIFFS .................... ......................................................... HOUSEHOLD ELECTRICITY EXPENDITURES................................................................................................ 16 VI VI1 DISTRIBUTIONAL .......................... 20 VI11 CONCLUSIONS............................................................................................................................................... ...WILLINGNESSTOPAY...... .............................................................................. .......................... 19 IMPACT ....................................... RIFFREFORMOPTIONS 26 CHAPTER3 WATER . .............................................................................................................................................. 28 I INTRODUCTION ................................................................. ..................................................................... 28 I1 KEYSECTORISSUES ............................................................................ 111 PUBLIC WATERSUPPLY ................................... ... -ASUMMARY OVERVIEW Iv. HOUSEHOLD ................................................................................... 30 WATERDEMAND AND CONSUMP ................................................................................... 32 V HOUSEHOLD CONNECTIONSTOTHE PUBLIC NETWORK ............................................................................ 33 VI QUALITY OFPUBLIC ............................................................... ...................... 35 VI1 ALTERNATEWATER BYHOUSEHOLDS .................................. VI11 WATERTARIFFS ............................................................................................... IX HOUSEHOLD 38 WILLINGNESS TOPAY.................................................................................................................................. WATEREXPENDITURES.......................................................................................................... X 41 XI XI1 CONCLUSIONS............................................ .... .... PUBLIC WATERMETERING .......................................................................................................................... 42 ............................................................................................ 44 ANNEXES ............................................................................................................................................................ 46 ANNEXI: METHODOLOGY ................................................................................................................................. 46 ANNEX11: THEPRIVATE ELECTRICITY GENERATION BUSINESS ................................ ........................... 52 ANNEX111: WILLINGNESSTOPAY-BACKGROUND .. .................................................................... 55 ANNEXIv: LESSONSLEARNED ENERGY ................................................................... 59 ANNEXV: REFERENCES ..........................................SEC FROM ............................................. 64 FIGURES FIGURE1.1. SHAREOF POPULATIONUNDER THE OFFICIALPOVERTYLINE................................................................... 2 FIGURE2.1. TRANSFERS TO EDLHAVEBALLOONED ................................................................................................... 6 FIGURE2.2. EDLELECTRICITYPRODUCTIONHAS RISEN SIGNIFICANTLYWHILE THE REAL KWHCOST TOHOUSEHOLDS H A S FALLEN............................................................................................................................................. 7 FIGURE 2.3. ELECTRICITY RELIANCEI S HIGHESTFOR WATER HEATING, REGARDLESS OF QUINTILE ............................. FIGURE 2.4. BEIRUTHOUSEHOLDS EXPERIENCETHE SHORTEST DAILY BLACKOUT PERIODINTHE COUNTRY ..............89 FIGURE2.5. USEOF GENERATORSIS INVERSELYCORRELATEDWITH PUBLICELECTRICITY RATIONING...................... 10 FIGURE2.6. USEOF GENERATORSIS CORRELATEDWITHINCOME.................................................................... ...12 FIGURE 2.7. ATFIRST GLANCE, THE INVERTEDBLOCK TARIFF SUGGESTS PROGRESSIVITY .......................................... 13 FIGURE 2.8. WHEN COMPARING EFFECTIVEVERSUS QUOTED ELECTRICITY CHARGES, REGRESSIVITY (FOR CONSUMPTIONUNDER 300 KWH) BECOMES CLEAR ................................................................................. 15 FIGURE2.9. KWHCONSUMPTIONPATTERNSWITHIN QUINTILES (EDLNETWORK), BY TRANCHE ............................... 16 FIGURE2.10. EXPENDITURES QUINTILE (EDLNETWORK) BY ...................................................................................... 16 FIGURE2.11.EXPENDITURES REGION(EDLNETWORK) BY ....................................................................... FIGURE 2.12. AMONGHOUSEHOLDS THAT USEGENERATORS, EXPENDITURESFORTHIS ALTERNATIVE F EXPENDITURESGOINGTO EDL ................................................................................................................ 19 FIGURE 2.13. WILLINGNESS TO PAY IS HIGH ................................................................................................................. 19 FIGURE2.14.WILLINGNESS TO PAYIS GREATER FORTHOSE WITHPRIVATEGENERATORS ........................................... 20 FIGURE2.15.AFFORDABILITY AND PROGRESSIVITY (CUMULATIVE) ........... ....... ...............24 FIGURE3.1. NETWORK AVAILABILITY - DEMANDMOSTLY EXCEEDSSUPPLY ................................................. WATER 32 FIGURE3.2. NETWORK AVAILABILITY ISNOT RELATEDTO WELFARE............................................................. WATER 33 FIGURE3.3. CONNECTION RATESARE HIGHEST INBEIRUT......................................................................................... 34 FIGURE3.4. USEOF PUBLICNETWORKWATER FORDRINKING SHOWS STRONGREGIONALVARIATION ....................... 35 FIGURE3.5. PERCEPTIONSOF PUBLIC NETWORK DRINKING WATER QUALITY VARY BY REGION ................................. 36 FIGURE3.6. HOUSEHOLD CONSUMPTION SOURCE-ALL HOUSEHOLDS............................... WATER BY FIGURE3.7. SHAREOF TOTAL WATER EXPENDITURESINHOUSEHOLDBUDGET ................................... FIGURE3.8. EXPENDITURES OF CONNECTEDHOUSEHOLDS ON ALTERNATIVE SOURCES ...................... FIGURE3.9. WATER EXPENDITURESOFCONNECTEDVS. UNCONNECTEDHOUSEHOLDS.............................................. 39 FIGURE 3.10. WILLINGNESS TO PAY ............................................................................................................................. 42 TABLES TABLE 2.1. THEPRIMARYREASON FORNOT USINGA GENERATOR IS AFFoRDABILITY. ............................. TABLE 2.2. LOW VOLTAGE EDLELECTRICITY TARIFFS- 2008 (PER KWH)................................ TABLE 2.3. BUT EDL'S HIGH FIXED FEESUNDERMINEPROGRESSIVITY (2008) ........................................................... TABLE 2.4. AVERAGEWHANDAMPERELEVELS(BY QUINTILE) K .............................................. TABLE 2.5. AVERAGE ELECTRICITY PRICE (EDLNETWORK) ........................ ................................................. TABLE 2.6. SHARE OF PUBLIC (EDL)ELECTRICITY EXPENDITURESINTOTAL HOLD EWENDITURES ................17 18 TABLE 2.7. MANY CONSUMERS DISTRUSTTHE BILLING SYSTEM ............................................................................... 18 TABLE 2.8. CHANGES TO NOMINAL TARIFF RATES ..................................................................................................... 24 TABLE 2.9. ILLUSTRATIVE SCENARIOS AND THEIR IMPACTS (EDLEXPENDITURESONLY) .......................... TABLE 2.10. IMPLICATIONS OF KEY ELECTRICITY REFORMS FOR DIFFERENTSTAKEHOLDERS ....................... TABLE 3.1. AVERAGEWATERTARIFFSANDDEGREE OF COSTRECOVERY................................................. TABLE 3.2. SUMMARY OF PERFORMANCEINDICATORS THE RWAS(2004) FOR ................................. TABLE 3.3. PUBLIC NETWORK WATERAVAILABILITY ...................................... ..................................... TABLE3.4. DAILYWATER SUPPLYAPPEARSTOHAVEDECLINEDDURINGTHE S MER (HOURS/DAY) ...................... 33 TABLE 3.5. REASONSHOUSEHOLDI S NOT CONNECTEDTO PUBLIC NETWORK ........................................ TABLE 3.6. WATERTARIFFSBYREGION (ANNUAL FEES, LBP) ......................................... TABLE 3.7. VOLUME VS. NON-VOLUME BASED WATER COSTS .......................................... TABLE 3.8. WATER EXPENDITURESBY SOURCE (MILLION us$)....................................... TABLE 3.9. MANY HOUSEHOLDS ASSOCIATEMETERS WITH HIGHER COSTS TABLE 3.10. IMPLICATIONS OF KEY WATER REFORMS FORDIFFERENTSTAKEHOLDERS ............................................. BOXES Box2.1. REALvs.EFFECTIVE ELECTRICITY PRICE ................................. .................................... ... 6 BOX2.2. THEPRIVATE ELECTRICITY GENERATION BUSINESS .................................................................................... 11 BOX2.3. CONSUMERSEXPLAIN WILLINGNESS TO PAY ................................................................................................. 21 BOX 2.4. SCENARIOS ................................................................................................................................................... 23 BOX 3.1.CONSUMERS EXPLAINWILLINGNESS TO PAY ................................................................................................... 43 ACKNOWLEDGMENTS This Report was prepared by Sebnem Akkaya (TTL), Nils Junge (Consultant) and Wael Mansour (Economist) with the support of a team including: M. Ananda Covindassamy (electricity sector and survey instrument), Chadi Bou Habib (fiscal analysis and survey instrument), Mohammed Benouahi (water sector), Nanak Kakwani (distributional impact analysis) and Muna Abeid Salim (processing o f the Report). Lina Fares (procurement specialist) provided valuable guidance in the organization o f the SIA survey. Inputs were received in early draft from Anna Bjerde, (Sector Manager), Demba B a (Country Manager), Haneen Sayed (Lead Operations Officer), Husam Mohamed Beides (Sr. Energy Specialist), Jonathan Walters (Sector Manager), Lizmara Kirchner (Infrastructure Specialist), Mona Ziade (Communications Officer), and Paul Noumba Um (Lead Economist). Hedi Larbi (Country Director) and Farrukh Iqbal (Sector Manager) sponsored the report and provided valuable suggestions on the design. The team would liketo give special thanks to Consultation and Research Group (CRI) who conducted the survey under sometimes difficult circumstances. Their local knowledge and resources were instrumental in bringing the survey to a successful conclusion, while their reliability and inputs with analysis were valuable. The staff o f the various ministries and organizations has provided valuable support and cooperation inthe preparation o f this report particularly those in the Ministry o f Energy and Water, Electricit6 du Liban, Regional Water Authorities, Council for Development and Reconstruction, Ministry o f Economy and Trade, Ministryo f Finance, and Ministryo f Social Affairs. EXECUTIVESUMMARY 1. This study considers implications for the consumer o f current service provision and impending reforms in Lebanon's utility sector. It aims to assess: i)how different categories o f households, specifically the poor, are affected by weak electricity and water service; ii)the potential social impact o f alternative reform scenarios-concerning tariff changes inthe case o f electricity, and metering inthe case o f water; and iii)implications o f better cost recovery measures inboth sectors on household welfare. ii. Thefindingsarebasedlargelyonanalysisofprimarydatacollectedaspartofahouseholdsurvey specifically designed for this Study; the recent technical sector assessments undertaken by the World Bank and by other domestic or external agencies; and information gathered in the field during the preparation o f the Study. Relatively little research on the energy and water sectors in Lebanon has been conducted to date on the householdconsumer perspective. Most sector studies have focused on technical aspects on the supply side, which representsjust one side o f the equation.' ... 111. The electricity and water sectors face major challenges in increasing supply and improving service, and they both are inneed o f significant reforms. One o f the most strikingaspects inthe electricity sector i s the heavy dependence on the informal, private generation sector, which nonetheless operates outside any state supervision or guiding framework. Bearing in mind the potential drawbacks of introducing regulation, the fact that such a large and monopolistic energy sector player operates in the grey economy requires attention. In both the formal and informal water sector quality concerns are of paramount importance-uality i s a public health issue and there are serious additional socioeconomic consequences. The Study points to the following key similarities and differences between the two sectors from a social impact perspective: Both the electricity and water sectors suffer from inadequate supply. Inthe electricity sector this takes the form of highly uneven rationing. Water supply on the other hand, is constrained in every region by limited infrastructure capacity, population density and demand. Both sectors experience high commercial and technical losses. Beirut households enjoy good electricity supply, with rationing limitedto three hours a day, but the city receives the lowest water supply per household inLebanon. A key problem inthe water sector is the disconnect between supply and demand. The absence o f a metering system means households pay a fixed fee for a fixed amount o f water supply (or allotment). Solving this issue would rationalize water consumption to a degree. There appears to be an informal understanding between water companies and households: many households don't receive their water allotment, and the water companies often don't pressure households to pay their bills. The key problem in the electricity sector i s highcost o f alternative supply, which i s utilized by 58 percent o f households. Virtually all households are connected to the electricity network. Connection rates for water are 80 percent. However, given that water supply is inadequate in terms o f both quantity and quality, connected households tend to purchase from alternate sources. Households have limited choice (on quality and cost) when it comes to electricity generation, with most buying from generator companies operating in the grey economy. There is a I The report should be read in conjunction with the "Republic of Lebanon Electricity Sector Public Expenditure Review, Report No. 41421-LB, Washington DC, January 31"; and "World Bank (forthcoming), Republic of Lebanon Water Sector Public Expenditure Review, Washington DC" which provide in-depth analysis of technical and institutional issues. i somewhat larger menu o f options for purchasingwater, available from tanker trucks, to wells, to large gallon bottles, and small bottles. Willingness to pay for improved electricity service is much higher for electricity than for water, reflecting thehighcost and limited choice o f electricity alternatives. ELECTRICITY PublicElectricitySupply iv. The 2008 spike in the international oilprice highlighted the country's fiscal vulnerability, and critical needfor energy sector reform. The fiscal impact o f the sector i s massive, puttingmacroeconomic stability at risk. Government transfers to cover the cash shortfall o f the electricity company-largely due to highly subsidized domestic price o f oil on which the electricity sector depends for generation-are estimated to reach 5.0 percent o f GDP in 2008, a steep rise compared with earlier inthe decade. Although the price o f oil declined sharply by the end o f the year, the volatility highlighted the vulnerability of Lebanon's fiscal position to factors outside control o f the budget process. While there are no quick fixes, measures for reducing the sector's burden on the government budget and, hence, financing requirements, are critical inthe period ahead. V. Whilephysical access to electricity is good, performance of the electricity sector has declined at an accelerating pace over the past decade. Lebanon has a high level o f electrification, with near universal network coverage at 99 percent, but supply remains a serious problem. N o new power generation capacity has been added since the two combined cycle plants were installed inthe 1990s (with the rest dating back to the 1970s and 1980s). The two plants were designed to operate on natural gas, but inthe absence ofaccessto this fuel, Lebanonrelies on expensive gasoil for power generation. vi. Electricity rationing is characterized by inequity. While some regions go without public electricity for 12-13 hours every day, administrative Beirut is subject to 3 hours o f daily blackouts. The electricity company, Electricite du Liban, (EdL) exerts a certain amount o f discretion over how rationing is distributed by region, utilizing a computerized pre-defined schedule for rolling blackouts. However, although there may be good reasons and strong political support for keeping rationing in the capital to a minimum, the inequitable nature o f the situation is clear, given Beirut's considerably higher welfare levels. vii. The real cost of public electricity tariff for consumers has been declining, while reliance on expensiveprivate generation has pushed up household electricity expenditures. Because the electricity tariff has not been raised since 1996 (while cumulative inflation has been 44 percent) the real price of publicly provided electricity for consumers has seen a gradual decline. At the same time, overall public electricity production inLebanon has continued increasing to meet rising demand. Nevertheless, while the cost o f electricity provided by EdL has remained low for consumers, the burden resulting from EdL's service decline (high and increasing frequency o f supply interruptions) has risen due to reliance on back- up generation, damaged appliances resultingfrom power surges and opportunity costs to households. Privately GeneratedElectricity vm. ... As a result of electricity rationing the majority of households rely heavily onprivate generators during blackouts. It is estimated that one third o f all electricity generated inLebanon, comes from private generators. Fifty eight percent o f households use some form o f self generation. This amounts to 20 percent more than in 2004, when 36 percent used generators. The majority o f households using back-up generation have a subscription with a private generator (a booming informal business). Although not .. 11 formally regulated, private generator businesses, which provide electricity through a small network, are tolerated inmost areas inthe country. ix. Privately provided electricity is more expensive than electricity from EdL at present tariff levels. Because privately generated electricity i s not sold by quantity consumed, but by ampere level, direct cost comparisons with EdL charges on a kWhbasis are difficult. However, it i s clear that electricity purchased from private generators i s more expensive. During the SIA survey period (spring 2008) the average bill from a private generator was US$47, compared with US$26 for the EdL bill; but private generators provided just half the number o f hours per day (7.2 vs. 14.3) and less than half as many amperes on average (6.5 vs. 14.5) as EdL. x. The burden resulting from EdL's service decline has increased significantly. This i s the case despite the fact that the cost of electricity purchased from EdL has remained low for consumers. This reflects reliance on back up generation and indirect losses incurred at the household level. xi. Households spent a total of US330 million on privately generated electricity last year. The informal electricity market serves 58 percent o f households with this far more expensive "imperfect substitute." Generator expenditures (for households that use them) are almost double what households spend on EdL electricity. Reliance on Electricity xii. Overall reliance on electricity is signzpcant. Because it i s relatively affordable, the majority o f households (75 percent) rely on electricity for water heating, which is energy intensive and can account for a large share o f a household's electricity consumption. The SIA found that almost half o f households (48 percent) use electricity to heat their houses and 42 percent own air conditioners. xiii. The tariff structure creates little incentive to constrain one's electricity consumption. Average household kWh consumption among the lowest-income quintile ranges from 295 kWh (off peak months) to 488 kWh (peak months). During peak months, one third o f households in the lowest-income quintile consume above 500 kWh and about one quarter consume below 300 kWh. Duringoff-peak months only 6 percent o f households in the lowest-income quintile consume above 500 kWh and more than 60 percent o f them consume below 300 KWh. Tariff and Expenditures xiv. The inverted block tariff does not achieve its objectives. Although designed to be progressive, in practice EdL's inverted block tariff fails to meaningfully benefit households with low electricity consumption. Because it subsidizes all households and because for households consuming very low amounts of electricity, the fixed cost forms a large share o f their bill, the tariff structure does not benefit low electricity consumers-there i s a large difference between the quoted kWh price, and the effective price, which is the electricity bill divided by kWh consumed. Most households currently pay an effective price o f 6 cents/kWh, risingto an average o f only 8.2 cents/kWh for highincome households. xv. The share of electricity expenditures to EdL in household budget is relatively low. The share of electricity purchased from EdL in household budgets duringpeak months ranges from 3.5 percent for the highest-income quintile to 5.O percent for the lowest-income quintile during peak months. These levels fall well below the 10-15 percent electricity share in household budget, a ceiling considered acceptable in many countries. The addition o f private electricity expenditures does not increase the share of expenditures above the ceiling. ... 111 xvi. The vast majority of households would be willing to pay more for electricity. Over half o f respondents to the SIA (household) survey would be willing to pay double their current expenditures, contingent upon receiving 24 hour service. xvii. Illustrative simulations show that the most progressive tariff reform would include a steep increase in blocks and reduction or elimination of the rehabilitation fee. A simulation model that demonstrates the poverty/social impacts o f how households would be affected by tariff changes in terms o f affordability, increase in the household bill, and progressivity showed that even relatively high tariff increases across the board do not lead household expenditures to exceed international norms for any quintile. Use o f private generation increases electricity expenditures inthe budget share. However iftariff increases are introduced gradually over time, with EdL simultaneously increasing hours o f service, households would reduce their need for private generators. Because electricity i s currently relatively affordable, distributional impacts o fthe tariff structure will matter more once the tariffs begin to increase. WATER PublicWater Supply xviii. Lebanon's ample water resources do not translate into sufficient water supply. One o f the few countries in the region benefiting from plentiful rainfall, Lebanon has potentially sufficient water resources to meet domestic demand. Nonetheless, transforming resources into quality drinking water for the entire population has proved difficult despite Government investments in infrastructure. As a result, water supply service is poor and intermittent in most parts o fthe country. xix. I n the absence of metering (outside a few limited areas) there is a disconnect between water supply and household demand. As in any country without a metering system, supply and demand for public network water i s de-linked at the household level: the amount of water provided to each household i s unrelated to the amount o f water a household would choose to consume and pay for. Key demand determinants-such as, price, income, quality and consumption needs-do not play a role, except insofar as a household chooses not to connect to the network at all. In short, because public water supply cannot be regulated to meet demand, households cannot adjust their consumptionpatterns. xx. Lebanon's cost recovery performance is poor compared to other countries. Three o f the four Regional Water Authorities (RWAs) do not have enough revenue to cover operating and maintenance (O&M) costs, and collection rates are as low as 33 percent in some cases. The Government often steps in to pay for operating expenses in additon to financing investments in water infrastructure. Most RWAs also incur highO&M costs arising from inefficient management o f services and degraded water networks. xxi. The effective cost ofpublic water to the households is often much higher than the quoted tarif$ The quoted average cost o f lm3/day in Lebanon is US$ 0.37, which i s equivalent to the average for MENA countries. Many households, however, pay a higher actual, or effective, rate because they receive considerably less water than Im3/day.Furthermore, unreliability o f supply imposes its own costs, because o f storage, and purchase o f backup or alternative water supply. Connectionsto Public Water Network xxii. Compared with four years ago, connection rates are up, but hours of service are down. An estimated 80 percent of households are connected to the public network system which i s an increase from 76 percent from four years earlier. Intermittent water supply is the norm. The average household receives iv 6 hours per day in the summer season and 9 hours in the winter season. Compared with the earlier estimates, this represents an overall decrease in supply, particularly duringthe winter months. Beirut and Mount Lebanon region is an exception, with an increase inwater suppliedduringthe summer. xxiii. Low-income households are less likely to be connected to the public network, but this is primarily due to location rather than the affordability. Inthe lowest-income quintile connection rates are 62 percent compared with 86 percent for the highest quintile. However, this has less to do with affordability o f the water bill (and even connection rates) than with location, given the pattern o f lower connection rates outside o f Beirut and the strong regional dimension o f the poverty in Lebanon. Less densely populated areas are more difficult and costly to serve. xxiv. Few households don't connect because they cannot afford it. Although households are not connected to the public network for a variety o f reasons, the issue o f affordability ranks quite low. Only 3 percent claim they `cannot afford' or `don't want to pay' for a connection. The most common reason given for not being connected i s that the household didn't have a choice: among the 20 percent o f unconnected households, over half report that there i s no public network available intheir area. xxv. There is high regional variation across indicators. Population density confers both advantages and disadvantages on connections in Beirut. Although Beirut households may receive relatively little water, they are also the most likely to be connected. This is in stark contrast to the other regions, where up to half o f households may not be connected. Perception o f drinking water quality likewise varies by region, with reported satisfaction far higher inBekaaValley than elsewhere. AlternativeWater Sources xxvi. Inadequate and unreliable water supply pushes households to purchase waterfrom alternate sources. Only one quarter o f Lebanese households receive water every day. Combined with often low water quality, the use of a wide variety o f alternate water sources is common. They are used by both connected and unconnected households. Households without connections tend to use artesian wells and delivery trucks more than connected households, but an almost equal number o f connected households purchase delivery truck water for service use, and buy water in gallons or bottles for cooking and drinking. xxvii. Combined water expenditures on public network and other sources are in line with World Bank recommendationson utility affordability. World Bank recommends that not more than 3-5 percent o f household budget is spent on water, and most Lebanese households fall within this range. Connected households spend a lower share o f their household budget on water than unconnected households. xxviii. If households could rely entirely on network water, they could cut their water expenditures signzjkantly. Inabsolute terms, reduced water expenditures for the bottom quintile would generate large savings (as much as 220,000 LBP or US$147 on average for connected households), virtually as much as their current average water bill o f 201,000 LBP (US$134). Savings could be even larger for the upper quintiles. The convenience o f having a household connection has not been quantified but would be an added welfare benefit. However, even under the status quo, simply being connected would lower a household's water expenditures. xxix. Given the current conditionsand alternatives, households are reluctant topay morefor better public service. Despite dissatisfaction with quality and availability, when presented with a scenario o f better quality water and sufficient supply, households were generally unwilling to pay more for public V service. Only half said they would be willing to pay more than an additional 21 percent for better service, while one-third would pay 50 percent (approximately US$66) per year more. CONCLUSIONS xxx. The overarching challengefacing Lebanon's public electricity and water sectors will be to establish trust with consumers while simultaneouslyimprovingperformance. Major investments ineach sector targeting infrastructure, management and human resources will need to take place while increasing revenue from consumers who have little faith in the system and want to see concrete results or credible action. The following remarks consider each sector separately. Electricity: Design a more effective and simplified tariff structure. The current tariff structure is regressive and will do little to shield the poor from any future tariff increases. The inverted tariff block i s not progressive-this i s partly because it subsidizes all households and partly because effective kWh prices differ markedly from the quoted prices. This is especially so for low electricity consuming households. Ensure proper sequencing in implementingpolicy measures. The burden resulting from EdL's service decline has increased significantly, despite the fact that the cost o f electricity purchased from EdLhas remainedrelatively low for consumers. However, increasing tariffs to cost recovery levels without moving to quickly introducing service improvements would likely meet with resistance. Consumers must feel that the burden o f reform i s not placed squarely on their shoulders but i s shared with EdL. Improve production and service by reducing rationing, especially in the areas outside Beirut which experience long blackout hours. Among other things, this will have the effect o f readjusting the burden imposed by rationing from the poor to the non-poor. Although there may be good reasons for keeping rationing in the capital to a minimum,it gives rise social inequities. Beirut households have higher welfare levels and they are most able to afford paying for expensive substitutes, yet have the least need for them. Most households in other regions must choose between going without electricity and spending significant amounts on private generation. Identify key operatingguidelinesfor informal electricity sector. The informal electricity sector, served by hundreds o f private generator businesses, provides up to 30 percent o f Lebanon's electricity and i s an indispensible service to many households. Yet the sector lies entirely outside the legal framework and does not pay taxes to the state. It must be acknowledged that private generation will play a significant role in electricity generation for years to come. Hence, identification o f proper operating guidelines that will protect and benefit consumers while ensuring continued availability o f this alternative source should be given consideration and merit additional research work. Without dampening private sector activity, or generating an additional and onerous layer o f bureaucracy, the guidelines could have several objectives, including: J bring private generator businesses into the formal sector so that customers have access to means o f redress for losses or damage incurredby faulty wiring or through other fault; J enable the stateto taxthe sector likeother businesses; J settechnical standards for service; J monitor the sector and either set guidelines on subscription fees (it would be desirable to assess ifthe generator businesses operate as a monopolistic structure). vi Harmonize electricity sector reform with social safety net reform over time. Given the ongoing progress with the design o f a social safety net system with an adequate targeting framework in Lebanon, it will be important to consider electricity reforms, in particular tariff structure adjustments, in the context o f the objectives and tools of the new social safety net system over time. In some countries social objectives are pursued through social policies and not necessarily through energy pricing. This i s an option that should be kept in mindwhen the new social safety system is up and running, at which point the changes to the tariff structure that are currently being debated for implementationfrom 2009 onwards could be reconsidered. Water: Invest in improving water quality. InLebanon, the burden on poor households comes from poor quality and low water supply, rather than high expenditures. Water expenditures by the low income households are in line with World Bank recommendations o f 3 to 5 percent o f household budget, but could be reduced ifhouseholds relied less on alternative sources. Although compared to other problems in water provision, reducing expenditures for consumers is not the top priority, the costs which consumers bear in terms o f quality o f service-poor reliability and potential health risks-are significant. A focus on improving quality and reducing losses, if accompanied by a public awareness campaign, would have a direct effect on welfare. A rapid rollout of metering is needed if wastage and equity issues are to be addressed. Both Regional Water Authorities and consumers lose when supply and demand are delinked by the current flat fee structure. Water companies cannot charge the marginal cost of production, and consumers do not get what they pay for. The pilot metering programs show that metering can be introduced, but unless metering i s either region-wide or metered households are able to pay by volume, the benefits o f meteringwill not materialize. Adopt a clear framework for reform sequencing by balancing costs and benefits between consumers and water companies. While the scale and complexity o fthe improvementsprecludes them from occurring simultaneously, proper sequencing o f reform measures will be crucial to their successful implementation and acceptance by stakeholders. Reform will likely need to be based on negotiating a compromise between water companies increasing supply and quality (representing a gain for households, but additional costs for the companies) and raising the tariff (a loss to households, a gain for the companies). To increase revenues, RWAs will need to address household concerns. RWAs can increase revenues in two ways, through improved bill collection and through tariff increases. In either case, RWAs will need to revise both the informal and the formal contractual agreements with households, raising tariffs and installing meters to link supply with demand, while also investing in improvementsinquality. They will also needto strengthen collection enforcement mechanism. To be successful, the new contractual arrangement will need to spread the benefits and costs between consumers and water companies ina manner acceptable to both. vii CHAPTER 1.INTRODUCTION OBJECTIVES 1. The purpose o f this Social Impact Analysis (SIA) is to probe the social, poverty and equity dimensions o f electricity and water sector reforms' and provide meaningful analysis to policy makers based on recently collected data. With a focus on households, the end users o f utility services, the study complements the recent and ongoing studies on the Lebanon water and energy sectors that deal with more technical and supply side issues. The study assesses how poor and vulnerable households are affected by the current utility service situation and how they may be affected by reform proposals under deliberation. The aim i s to provide policy makers with a deeper understanding o f the social dimensions o f water and electricity consumption as well as tools for estimating the distributional impacts o f reform measures. 2. This Study followed the Poverty and Social Impact Analysis (PSIA) approach, an increasingly common approach, used both within and outside the World Bank to evaluate the distributional impacts o f policy reforms. This approach is a form o f evaluation research, typically conducted ex ante, and uses an array o f economic and social tools and techniques to provide policy makers with a clearer understanding o f how policy design and implementation affects different population groups, particularly the poor and vulnerable. Anticipating inadvance potential negative impacts allows mitigation measures to be built into the policy design. CONTEXT 3. As a middle income country, the entire population o f Lebanese households relies on water and electricity to meet its basic needs. With universal connections to the electricity network, and four fifths o f the population relying on publicly provided water, all households are potentially affected by changes in utility service. Even in the case o f water, many unconnected households will be exposed to reform measures. As access to the network increases, and affordability and quality levels changes, all households will face a new set o f choices. As reform measures are rolled out, the cost, affordability, and quality o f publicly provided utility services will change inrelation to the `back-up' sources that so many households currently rely on. All households will weigh the costs and benefits of using public utility services versus privately provided alternatives. Among middle class and wealthier households, for whom affordability i s less o f an issue, reforms will be experienced primarily in terms o f convenience and quality. For poor and lower income households, however, utility sector reforms will have a more immediate significance. Changes in the availability, reliability and cost o f utility services are more likely to affect consumption behavior, with concomitant social as well as expenditure impacts. 4. A study on distributional impacts must begin with a general description o f the population which stands to be affected. According to the 2004-05 Living Conditions and Household Budget S ~ r v e yjust , ~ under 8 percent o f Lebanese households lived below the poverty line, equivalent to US$2.4 per capita per day. These are households that considered extremely poor and unable to meet their basic food and non- food needs. An `upper poverty line', set at US$4 per capita per day, was also set, and 28.5 percent o f the population fell under this threshold. 2For the purpose of this study, sector reform is broadly understood as improved service delivery (reliability, availability and quality) andmeasuresto move the sectors toward cost recovery. See Poverty, Growthand Inequality inLebanon, UNDP, 2007. 1 5. Poverty in Lebanon has a strong regional dimension. While Beirut had a poverty rate o f less than one percent (below the extreme poverty line) in 2004, the share was much larger in the Bekaa Valley (10.8 percent), the South (11.6 percent) 'and the North (17.8 percent). The significance o f regional disparity i s immediately apparent when considering distributional impacts o f utility service provision. If, as is presently the case with the Lebanese electricity sector, regions are more highly rationed outside the capital, where poor households predominate, the benefits resulting from restoration to full 24 hour service across the country favors the poor. Inother words, being most deprived, they havethe most to gain. Figure 1.1. Share of population under the official poverty line *O.O 1 17.8 Beirut Nabatieh Mount Bekaa South North Lebanon Source: "Poverty growth and inequality in Lebanon: executive summary': CAS, UNDP and MoSA Living Conditions and Household Budget Survey (2004) 6. In terms o f expenditures, households below the poverty line spend significantly more than the non-poor on food (35 versus 23 percent) and slightly more on utilities (34 versus 30 percent). The opposite is true for health and education expenses. It i s therefore to be expected that changes inthe price, as well as availability, o f these categories o f consumption items will have differential impacts on welfare groups. METHODOLOGY 7. Primary data was collected for the study usingboth quantitative and qualitative researchmethods. A quantitative survey was conducted targeting 1,804 households, representative at the Mohafaza (region) level. The sample was also selected to be geographically diverse (urban, rural, central, peripheral, densely populated, semi-deserted, etc.). The survey was conducted from April to June 2008.4 The results were analyzed to determine the consumption patterns and the willingness to pay o f households according to region, welfare category and other variables o f interest. (see Annex I) 8. In addition to the quantitative survey, qualitative in-depth interviews were conducted with selected profiles (households, municipalities, generator owners, EdL collectors and building managers) in order to elicit more in-depth understanding o fthe issues. These interviews were conducted inthree waves: before, during, and after the quantitative survey. 9. Because o f the changing poverty levels, which estimates indicate have risen since 2004, the analysis does not use the 8 percent poverty line. Instead it analyzes distributional impacts by comparing Itwas briefly disruptedby the Maypolitical andsecurity crisis. 2 household expenditure quintiles. Aside from avoiding the debate on who is poor, ranking by quintiles (in contrast to comparisons o f poor and non-poor households) has two advantages: i)it provides a snapshot o f distribution among all expenditure groups; and ii)it allows policymakers to consider the impacts on a larger population segments at the low end o f the socio-economic ladder, for example the bottom 20 or bottom 40 percent. REPORT STRUCTURE 10. The following chapters examine the energy and water sectors. Chapter 2 provides a review o f household electricity supply and demand, private generation, tariffs, expenditures, and willingness to pay. It is followed by a discussion o f the distributional impact o fthe tariff structure, including simulations o f tariff scenarios for illustrative purposes. Chapter 3 reviews the water sector, public water supply, household connections, water quality issues, alternate water sources, tariffs, expenditures and willingness to pay. Both chapters conclude with recommendations with an emphasis on social implications o f key reforms that matters for households. Annexes cover methodology and a description o f the private generator business inLebanon. 3 CHAPTER2. ELECTRICITY "Electricity is the mother of allproblems in Lebanon. Thesize of theproblem is beginning topose a danger topublicfinances. ''-Adohammad Chatah, Minister of Finance (quoted in L 'orientlejour, July 25, 2008) I.INTRODUCTION 11. Lebanon's dysfunctional electricitysector has in recentyears become a heavy burden onpublic finances, the economy and households. For years, the sector has suffered from a serious lack o f investment in operation and maintenance and new generation capacity, which combined with poor governance, resulting in a large penalty on the economy, low quality o f service, high system losses and inadequate cost recovery. At present the sector is unable to supply the reliable electricity needed by industry, commerce and households with more and more consumers turning to back-up generation at much higher costs. The fiscal impact o f the sector is massive, putting macroeconomic stability at risk. Government transfers to cover the cash shortfall o f the electricity company-largely due to highly subsidized domestic price o f oil on which the electricity sector depends for generation-are estimated to reach 5.0 percent o f GDP in2008, a steep rise compared with earlier inthe decade. 12. The sharp increase in international oil price during most of 2008 has highlighted Lebanon's fiscal vulnerability making sector reform critical. Although the price o f oil declined toward the end o f 2008, its volatility highlighted the vulnerability o f Lebanon's fiscal position to factors outside control o f the budget process. While there are no quick fixes, measures for reducing the sector's burden on the government budget and, hence, financing requirements, are critical over the period ahead. The key sector reforms are included in the Government's existing medium-term economic program presented at a donor conference in January 20075 and further developed in the World Bank's recent Energy Sector Public ExpenditureReview (PER). Duringthe summer o f 2008, the need to raise tariffs was frequently touched upon by the new Ministero f Water and Energy and received prominent coverage in the Lebanese media. The Government's 2009 budget plan envisages reducing the transfers to the electricity company through introduction of a set o f cost recovery measures, including through changes inthe tariff structure. 13. The burden imposed by weak public electricity service has not been shared equally. This chapter assesses how different categories o f households, specifically the poor, are affected by weak electricity service, what the potential impact o f better cost recovery measures would be on different household welfare categories and what sequence o f policy options present opportunities for addressing social concerns given supply constraints on the one side, and household consumption behavior on the other. 5 The program was introduced in the months following the devastating summer-2006 hostilities with Israel. It combines fiscal measures neededfor stabilization with structural measures neededfor higher growth. Participants in Paris I11pledged US$7.6 billion to support the implementation of the program. This includes support from the World Bank through policy based operations with major emphasis on key electricity sector reforms. So far, the progresswith implementation ofreforms has beenslow due to long-lasting political impasse and legislative inaction. Over the period ahead, the improved macroeconomic and political environment since May-2008 Doha agreement, recent reversal inthe trend of international prices, andLebanon's relative insulation from the frst-round direct effect of the global financial turmoil, creates a more conducive environment for re-invigorating medium-term economic reformprogram, as highlighted by the policy reforms that now underpinsthe Government's 2009 budget. 4 11. KEYSECTOR ISSUES OVERVIEW -ASUMMARY 14. While access to electricity is good, performance of the electricity sector has deteriorated at an accelerating pace over the past decade. Lebanon has a high level o f electrification, with near universal network coverage at 99 percent, but supply remains a serious problem, with heavy costs borne by households. N o new power generation capacity has been added since the two combined cycle plants were installed inthe 1990s (with the rest dating back to the 1970s and 1980s). The two plants were designed to operate on natural gas, but in the absence o f access to this fuel, Lebanon relies on expensive gasoil for power generation. To reduce losses major investments are also required in the medium- and low-voltage transmission and distribution system. 15. Electricitk de Liban's considerable investment needs are widely acknowledged, but reform inertia has kept investors away. Lack o f political will with consequent prolonged inaction on reform and restructuring, poorly designed policies (such as those relating to fuel sourcing decisions), and insufficient managerial capacity6 and technical staff' at Electricite du Liban (EdL), have left its investmentneeds and regular maintenance underfunded. Technical losses are estimated at 15 percent, and a further estimated 23 percent is consumed through illegal connections. EdL only bills for about 60 percent o f its production, and collects about 90 percent o f annual billings. Until credible reforms are launched, it is difficult to envision Lebanon attracting significant investments in EdL. 16. The electricity sector acts as a hugefiscal drain on the Government budget, crowding out key social and infrastructure expenditures. The Government's tariff policy, coupled with significant operational inefficiencies in the sector, has translated into increasing reliance on budgetary transfers.8 At present, a large portion o f these transfers can be attributed to a "tariff freeze" in place since 1996, which caps EdL's tariff at a level derived from an average oil price o f US$25/barrel-far below cost recovery levels. Transfers are also explained by other factors that contribute to low revenue (such as inefficient tariff structure, low billings), and very highproductioncosts (such as high O&M cost o f power plants due to insufficient regular maintenance and hightechnical losses). Against a background o f sustained high oil prices beginning in 2003, the budgetary transfers to EdL increased dramatically (see Figure 1). When oil prices topped nearly US$lSO/barrel by August 2008, they were estimated to reach highs o f 14.8 percent o f total budgetary spending and 5.0 percent o f GDP in 2008.' While the subsequent steep drop in the oil price in late 2008 provided some relief inthe near term, the size o f the budgetary transfers to EdL clearly limit fiscal space for key public expenditures," includingwhat could have been spent on social protection through targeted assistance if the tariff reflected fuel cost, consumers had been effectively billed, and 6 EdL's legal status as a state owned enterprise does not provide incentives for performance, enabling political interference in day-to-day operations, and inhibiting its ability to collect for consumedelectricity. Indirectevidence o f these problems comes from the difficulties in obtaining reliable information on the sector, such as financial statements, billing and collection data, and technical performance indicators (See Lebanon Energy Sector PER, World Bank, 2007) 7 Hiring of staff at EdL was frozen in the mid-l990s, leading to a dearth of personnel to manage operations and address financial and administrative issues. The average employee age is about 58 and EdL loses staff each year to retirement.As a result, there has been no renewal of skills to EdL. Problems includelack of planning, maintenance and care of assets, and insufficient capacity to handle billing and collection. (See Lebanon Energy Sector PER, World Bank, 2007) 8 The transfers to EdL cover both fuel payments and debt service. There are also accounts payable to EdL which have not beensettled, for instance, Value Added Tax. 9 Under the fixed tariff, the eventual cost to the budget is obviously heavily dependent on the oil price fluctuations over the course of the entire year. The budgetarytransfers to the EdL reflect the actual purchaseprice of oil with a six months lag due to current practice of purchasingoilthrough letters of credit. lo Primary expenditures,excluding transfersto EdL, increasedby 4.3 percentper year on average between2003 and 2008 while transfers to EdL increasedat an average of 38.3 percentyearly over the same period. 5 service had been stronger. The transfers also push up the fiscal deficit, puttingmacroeconomic stability at risk given Lebanon's record highpublic debt ratio (162 percent of GDP in 2008). Finally, they expose Lebanon's fiscal position to factors outside the control o f the budget process. Figure 2.1. Transfers to EdL have ballooned Deficit and Tranfers to EdL I I I I A 12% 48% (3 50 .c1 10% 40% 2v) 8% 32% z 24% I- 6% .-d04% 16% z .c1 8% 2% 1 0% 0% 2003 2004 2005 2006 2007 2008 Source: OfJicial data and WorldBank staff estimates 17. The real electricity tariff for consumershas declined, while growing reliance on expensive Box 2.1. Real vs. effective electricity price back-up generation haspushed up total household electricity expenditures. Since 1996, cumulative The estimated real price o f electricity is inflation- inflation reached 44 percent in Lebanon (showing adjusted and reflects year-on-year changes in the acceleration since late 2007) driven by increases in average price per kWh.It is based on (i) estimated international prices particularly for oil. However, as usage by a median household consuming between noted above, since the electricity tariff has not been 200 and 400 kWh per month, using 15 amperes; and (ii) the effective price per unit. raised over this period, the real price o f electricity for the average consumer has seen a gradual decline The effective price i s the household electricity bill (see Figure 2.2 and Box 2.1.). The impact o f the divided by kWh consumed. This is in contrast to "tariff freeze? was somewhat counter-balanced by the quoted price, which i s the tariff charge (only a the introduction o f a monthly rehabilitation fee o f portion o f the bill) divided by kWh consumed. LBP 5,000 (equivalent to US$3.3) on every bill for From the household perspective, the effective low voltage consumers in 1996." This led to an tariff i s more important than the quoted tariff. For increase in the effective kWh price for low voltage instance, in a restaurant, it i s the total final bill consumers (i.e. households and small businesses)'2 that affects the diner's pocketbook, not how the by 14 percent on average, with a socially regressive bill breaks down according to menu price, tax, effect, as the percentage increase was higher for service, tips, etc. low-income consumers than large-income consumers (see Box 2.1.). Nevertheless, while the cost of l1Originally, a temporary measure addedto electricity charges for a seven year period, but maintained thereafter. 12Of electricity consumed, private low-voltage consumers represent more than half o f the demand in Lebanon (according to EdL statistics, Chubu Consulting report and Bank PER estimates). This high proportion reflects the importance o f small service businesses in the country and is due to the aggregation in this category o f both residential and small business consumption. 6 electricity provided by EdL has remained low for consumers, the burden resulting from EdL's service decline (high and increasing frequency o f supply interruptions) has risen, due to reliance on back up generation, damaged appliances resulting from power surges and opportunity costs to households. Figure 2.2. EdL electricity productionhas risensignificantlywhile the real kWh cost to households has fallen13 125 1 0 I -+- Production (mill kWh) 50 4 96 97 98 99 00 01 02 03 04 05 06 07 08 Year Sources: EdL and WorldBank stafestimates. 18. At least US$330 million was spent by Lebanese households last year on private electricity generation-almost twice the amount that households spend on EdL electricity. This total14 represents the financial cost to residential consumers o f inadequate service delivery. The large amount spent on this imperfect substitute not only results in additional expenditure outlays by these households (indicating abili /willingness to pay for reliable service), but it also represents potentially foregone revenue for EdL. If24 hours o f service were provided, it can be assumed that far fewer households would continue % to subscribe to private generators for backup with much o f the US$330 million16 currently going to the informal sector providers being redirected to EdL. 111.HOUSEHOLD DEMAND CONSUMPTION ENERGY AND 19. Household reliance on electricity is significant. The majority o f households (75 percent) rely on electricity for water heating, which is energy intensive and can account for a large share o f a household's electricity consumption. The SIA survey found that almost half o f households (48 percent) use electricity to heat their houses and 42 percent own air conditioners. This picture reflects the relative affordability o f electricity, which is discussed in sections IV and V below. 13 This graph does not provide proof of correlation between average unit electricity costs and total electricity generation. The simultaneous increase in generation and decline in real unit costs since 2001 does, however, merit contemplation. 14Basedon WB staff calculations using SIA survey data. 15It is qualified as `potential' forgone revenuefor two reasons: i)it is unlikely that all householdswould completely discontinue private generator use and ii)it applies only to an EdL that has achieved cost recovery; at present the company loses money on every kWh sold, so that the less it suppliesthe lower its losses. l6Calculations suggesting at leastUS$330 millionprivate generationexpenditures are based on findings that show that 58 percentof 906,016 Lebanesehouseholdsuse private generation, paying an average of US053 per month, or a total of US$636 per annum. Thus; 0 . 5 7 9 ~906,016 x 53 x 12= 333,760,856 7 Figure2.3. Electricityrelianceis highestfor water heating,regardlessof quintile 1. 1 SO?? 75% 77% 76% I /ilY 1 1 71% n 70% 70% I 62% 60% 5G?? 40% - 30% 4 0% 5 1 2 3 4 Ouintiles 0 Cooking 0 Water Heating HouseHeating W Cooling(ACs) ~ Source: WorldBank SIA survey (2008). 20. Use of solar energy is marginal. Solar energy for water heating is entering the Lebanese market, and now is used by 3 percent o f households. However, it i s primarily the wealthy who are taking advantage o f the new technology-almost half o f all households who reported using solar energy to heat water belong in the top q~inti1e.l~The remaining households use gas, kerosene/gasoil, and wood/coal energy sources, with many usingmore than one source. 21. Electricity supply is constrained, leading to heavy rationing. Lebanon has the highest rate o f daily blackouts in the MENA region." Based on the SIA survey, average electricity supplied by EdL to households i s about 14 hours per day (so that, on average, there are 10 hours o f daily power outage). However, as discussed below, the amount o f electricity supplied to regions (as well as within regions) varies widely. Service interruptions are daily, but not always at the same time or predictable. In one half to one third o f cases, depending on the region, households reported that the timing o f the blackout period varied. 22. Rationing is characterizedby inequity. While the Bekaa Valley and Nabatieh must do without public electricity for 12-13 hours every day, Beirut i s subject to 3 hours o f daily blackouts (see Figure 2.4.). EdL exerts some discretion (withinthe limits o f regional generation capacity) over how rationing i s distributed by region, utilizing a computerized pre-defined schedule for rolling blackouts. However, although there may be good reasons and strong political support for keeping rationing in the capital to a minimum, it represents a clear case o f inequitable distrib~tion.'~ the other regions, many households In must choose between going without electricity for long daily periods, and spending significant amounts on private generation. 17Solar water heating is both cheaper-in the mediumterm-than electricity, and is not prone to interruptions from blackouts. 18See LebanonElectricity PER, World Bank, 2007. 19Beirut householdshave higher welfare levels-median householdconsumption inBeirut was 37.8 percent more than the average for Lebanon and extreme poverty in Beirut was less than 1 percent compared with an 8 percent average for the country as awhole basedon 2004 HHsurvey. 8 23. Living without electricity is a social and economic burden. One o f the feedbacks from this study i s how households enduring regular blackouts face a variety o f inconveniences and negative economic and social impacts. They are unable, duringthese periods, to heat and cool the house or use the elevator, television or lighting for other essential needs (as in the case o f families with school-age children, for studying). They cannot store food inthe refrigerator because o f spoilage (and hence must buy perishable commodities on a daily basis). Several survey respondents described the situation as depressing, noting that the summer heat and winter cold exacerbate the situation. Some households forgo paying for a generator because they don't want to cut into higher priority items (such as children's essential expenditures). IPI Figure2.4. Beirut households experiencethe shortest daily blackout periodinthe country = 12 8dI I c O I 4 0 u I I I Beirut Mount- North Bekaa South Nabatieh Lebanon Source: WorldBank SIA survey (2008). 24. As a result of electricity rationing the majority of households rely heavily onprivate generators during blackouts. It is estimated that one third o f all electricity generated inLebanon, comes from private generators.20Fifty eight percent o f households use some form o f self generation. This i s 20 percent more than in 2004, when 36 percent used generatorsa21More than half o f generator users began using them since 2000,22further evidence o f deteriorating EdL service over the past decade. 25. Private generation is a booming business. The majority (70 percent) o f households usingback- up generation have a subscription with a private generator business.23Although not legally recognized, private generator businesses, which provide electricity through a small network, are tolerated in most areas in the country. The remaining households own their own generator (22 percent), use a generator owned by the building(7 percent) or use a community-run generator (about 1percent). 26. As expected, the share of households that useprivate generators is correlated with number of hours of electricity availableper day. Because administrative Beirue4is provided with 21hours per day by EdL (far more than other regions), a relatively small share o f Beirut households use generators (see 2o SeeLebanonElectricity PER, World Bank, 2007. This covers commerce and industry as well as households. 21 2004 figures are basedon 2004 Household Living Conditions survey o fthe Central StatisticsAdministration. 22 SIA survey (2008). 23 In Beirut, running a small generator from one's home, a common practice during the 1975-90 civil war, is no longer allowed. 24 The population of administrative Beirut was estimated by M P S to be 389,601 in 2004. Households in Greater Beirutincluding bothNorth and South Suburbsare locatedinthe Mount Lebanonregion. 9 Figure 2.5.). Beirut only became subject to rationing at the beginning o f 2007. When households that did not use generators were asked why, a much higher share in the regions-27 percent compared with 10 percent inBeirut-reported itwas because they couldn't afford it. Figure 2.5. Use of generators is inverselycorrelatedwith public electricity rationing South + +North + MountLebanon Nabatieh + Bekaa 0 20 40 60 80 100 Share of households using generator (%) Source: WorldBank SIA survey 2008. 27. Private generators charge a flat fee based on a maximum ampere level, not kWh usage. In contrast to EdL, households subscribing to private generators are charged a fixed monthly fee that varies depending on the ampere level at which they subscribe25(see Box 2.2.) and on gasoline used. The vast majority o f households subscribe at 5 amperes, which enables them to keep the lights, television and refrigerator on. If a consumer does not use electricity during the blackout, the private generator gains by saving on fuel costs. 28. Privately provided electricity is more expensive than electricityfrom EdL at present tarijJ As noted above, because privately generated electricity i s not sold by quantity consumed, but by ampere level, precise cost comparisons with EdL charges on a kWhbasis are difficulthouseholds do not use the maximum amperes for the entire period the generator i s operating. However, even if they did so, electricity purchased from private generators i s at least 8 times as expensive. During the survey period (April-May 2008) the average bill from a private generator was US$47, compared with US$26 from the EdL. Yet although the average expenditures for private generation were almost double those going to EdL, private generators providedjust halfthe number o fhours per day (7.2 vs. 14.3) and less than half as many amperes on average (6.5 vs. 14.5)26as EdL. 29. Affordability is only one of many concerns with respect to reliance on private generators. Among the 42 percent o f households that do not use private generation, the main reason was economic- 23.7 percent mentioned the high cost o f living and bad economic conditions; 9.6 percent mentioned the highsubscription fees andbad service, followed by lack o faccess, since no private generators were '' 25The ampere level, a capacity indicator, places a limit on the amount o f electricity a household can consume at a iventime. Actual amperes are almost always provided in increments of 5, although some cases of 2.5 amperes have been reported by private generator operators. 10 Box 2.2. The Private ElectricityGenerationBusiness Private generators are informal, but well established businesses. Private electricity generators have been around since the 1975-90 civil war and occupy a well-established comer of the utility market. Most private generator businesses employ two to four persons, operate generatorsfrom 100 K V A to 500 K V A (which cost from US$l5,000 to $60,000). They sell to both businesses and households. Costs include fuel oil, maintenance, filters, replacement of cables and breakers, andrental space. They ,typically have severalhundredsubscribers, and may cover an entire town. The owner may have other businesses as well. The generator business has monopolisticfeatures. Although there are a large number of generator businesses (e.g., over 150 in Tripoli alone) the competition on price i s limited to align prices between generators in the same city- whatever manner a private generator uses to secure rights over an area, it is certainly not through an open tender. Thus, households cannot choose their subscriber, but must use the one operating intheir neighborhood or town. The choices available to ahouseholdare simple-it can choose not to subscribeor stop its subscription, or it can change its ampere level. Reducing amperes from 10to 5, e.g., cuts the bill inhalf. Improvement in electricity availability benefts private generators. Private generator businesses do not see themselves as EdL competitors and would, in fact, welcome less rationing. A decrease inblackout hours would save them fuel costs in the short run. This underlines the fact that although private operators are responding to unmet demand, the relationship betweenpublicly andprivately supplied electricity is not one of perfect substitution. Inother words, a gain for the public provider is not equivalent to a loss for the private provider. Private generators do not expect EdL to provide 24 hour service in the short- to medium-term and expect demand to persist based on the assumptionthat customerswill continueto use their services as aback-up for years to come. This also stems from the public's mistrust of the reliability of Ea's services. Like EdL,private generator businesses are hurt by highfuelprices. Generatorsare concernedabout highfuel prices, which ballooned through September 2008, and threatened to squeeze their profit margins of fuel cost (when not passed-through to consumers). As subscription rates went up, many consumers either lowered their ampere subscription to save money, or stoppedusing generatorsaltogether.As fuel prices come down from their earlier highs, both the public utility provider and the substitute provider may standto gain. . For this reason, little resistanceon the part of private generator businesses to improved EdL performance is expected in the short- to medium-term. In political economy terms, the position of generatorbusinessestoward reform is not seen as antagonistic. (See Annex I1 for further discussion). operating intheir area (see Table 2.1.). Some households inthis group use alternative substitutes, such as UPS,27and battery to provide electricity. Mazout and gas are used by a small minority for lighting.28 Table 2.1. The primary reasonfor not usinga generator is affordability Reasons Percent High cost of living, bad economic conditions 23.7 No private generators inthe area 19.4 Blackouts are tolerable 11.4 High subscriptionfees andbad service 9.6 UPS 5.9 Lighting with mazout and gas 2.6 Battery 2.5 Other sources of electricity 1.4 Private generators not allowed inthe area 1.3 Source: WorldBank SIA survey 2008. Note: 21.4percent of respondentsdid not report any reason. 27 UPS i s similar to abattery. 28 Beirut was the only area where some householdsreported private generatorswere not allowed. Inthe 1990s Beirut municipality banned generators that did not have noise controls, as many people were running generators on their balconies. This ordinance was successfully implemented. At the same time, most large private generators without filters to reducepollution were shut down. The response `no private generatorsallowed in area' reflects the belief by some that they are illegal inall cases. 11 30. Low income groups are least able to afford private generators. As noted above, the costs o f privately generated electricity i s high, preventing many low income households from accessing this imperfect back-up source. Figure 2.6. shows that use o f private generators i s positively correlated with household welfare. Still, 42 percent o f the lowest quintile, compared with 65 percent o f the top quintile, use generators, a fairly significant share given the highsubscription costs. Figure2.6. Use of generators is correlated with income 100 - .*,- -,.- __,-- ,-.I , . . a a a , , I I 1 2 3 4 5 quintile Source: WorldBank SIA survey 2008. 3 1. Privately generated electricity isprovided during the daily blackoutperiods, but may not cover the entire blackout. Most generator owners and households interviewed reported that generators do not runthe entire period o f the blackout. For example, households in areas which receive 16 hours o f public electricity per day, and choose to subscribe to a private generator, will typically receive less than 8 hours o f generator service. This may or may not be made clear in the contract, which i s usually verbal. Many private generators save on costs by not providing service after midnight, with some shuttingo f f as early as lOpmor 1lpm. 32. Lack of oversight of private generation business carries costs and imposes a burden on households. It is noteworthy that the private generation businesses, on which so many households and a large share o f the economy depend, operate in an unregulated grey zone (see Box 2). They do not pay taxes to the state (although they may have entered into informal contractual agreements with local municipalities). As noted earlier, despite the large role they play in providing a basic service, private generators can only be depended upon by a certain share o f the households and for very basic appliance usage. Because o f what can be described as quasi-monopolistic nature, consumers are unable to choose between generator companies, and companies are price setters. IV.PUBLICELECTRICITYTARIFFS 33. EdL applies an inverted block tariff to residential and small commercial consumers. The inverted block tariff (see Table 2.2.) i s commonly used because o f its perceived progressivity and cross- subsidy feature, but is often ineffective (depending on how it is administered) in achieving these 12 objectives and Lebanon i s a case in point. The EdL tariff structure uses five tariff tranches*' and rates range from 2.3 U S cents to 8.0 U S cents per kWh for the 'subsidized' rates for consum tion below 500 kWh per month, and 13.3 US cents for any consumption above 500 kWh per month! EdL does not distinguishbetween residential and commercial consumers. Table 2.2. Low voltage EdL Electricity Tariffs 2008 (per kWh) - Residential and commercial Tranche kWhlevel LBP UScents lSf 0-100 35 2.3 2"d 100-300 55 3.7 3rd 300-400 80 5.3 qfh 400-500 120 8.0 5th >500 200 13.3 Public Administration 140 9.3 Handicraft and Agriculture 115 7.7 Source: EdL data 34. The inverted block tariff does not achieve its objectives. Although designed to be progressive3', in practice EdL's inverted block tariff fails to meaningfully benefit households with low electricity consumption. There are three key reasons this tariff structure i s not progressive in practice: (i) the high lifeline tariff for consumption below 500 kWWper month is overly generous and ends up with subsidizing all users (because there i s no "claw-back" clause stipulating that a consumer consuming in a certain category pays the entirety o f the bill in this category); ii)high fixed costs are included in the electricity bill (penalizing small consumers more than large ones an a per kWh basis); and iii)poor households do not consume substantially less electricity than non-poor households. Hence, the visual depiction of the seemingly progressive tariff blocks in Figure 2.7.' with their rising costs per kWh, i s not mirrored in reality, as discussed below. Figure 2.7. At first glance, the inverted block tariff suggests progressivity 100 - 75 0-100 100-300 300-400 400-500 ~ 5 0 0 Tranches (kWh levels) Source: WorldBank SIA survey (2008). 29 Technically, EdL uses six tranches. It considersthe 2"dTranche (in Table 3) two tranches, even though the kWh charge, 55 LL, is the same for both. Therefore, for analytical clarity, this study uses only 5 tranches. 30 With current generationcosts, all consumptionlevels are effectively subsidized. 3 1The more electricity consumedthe more expensive the unit cost. 13 35. A key reason the current rising tariff block is not progressive in its impact is due to the large share of fixed costs in the electricity bill. A breakdown o f the EdL electricity bill shows that fixed costs (those not based on kWh consumption) are equivalent to at least 9,600 LBP3' per month (or about US$6.4) for the median household electricity consumption (see Table 2.3.). More than half o fthis amount is accounted by a flat "rehabilitation fee". The 10 percent VAT tax i s applied to everything but the stamp fee; an ampere charge, according to ampere level, is applied (with sixty-nine percent o f households subscribe at either 10 or 15 amperes; 16 percent subscribe at 20 amperes and the remainder at other levels); and finally, the bill is rounded up to the nearest 1,000 LBP. While connectionfees are a necessary and standard element for most utility companies, and the rehabilitation fee compensates for stagnating real electricity prices, relative to the bill, it appears highand has a distributional impact. Table 2.3. EdL's fee structure* (20081 Low Voltage Fees LBP USD Stamp (fixed) 1,000 0.67 Rehabilitation(fixed) 5,000 3.33 Ampere/phase (variable) 240 0.16 kWh charge (variable) 35-200 0.023-0.133 Roundingup 1- 999 0.0 -0.66 VAT (variable) 10% 10% Source: WorldBank SIA survey 2008. * Excludes a one-timefee paid by new entrants into the system. 36. As a result of high fixed fees, the less electricity a household consumes up to the medium consumption level, the higher the effective kWh charge. Thus, up to kWh consumption levels o f about 300 kWh per month, the lower the amount o f kWh consumed, the higher the effective kWh price paid by the consumer33(see Figure 2.8.). For example, logic dictates that for those households consumingjust 100 kWh, the fixed cost portion o fthe bill is a much larger percent ofthe total than a household consuming at 600 kWh. The effective kWh charge (as noted in Box 2.1., equivalent to the total bill divided by the number o f kWhconsumed) is thus quite different from the quoted kWhcost. The effective kWh cost for consumption above 400 kWh increases quite gradually. There i s consequently little incentive to reduce electricity consumption, all things being equal. This may not be o f pressing concern at present, but once tariffs begin to rise (as they eventually must) the impact will start to be felt. 329,600 = 5,000 (rehabilitation fee)+ 1,000 (stamp fee) + 3,600 (for 15 amperes, the median level) 33 All else equal, it therefore becomes rational to consume at the level where the price per kWh is lowest, i.e. between 300 and 400 kWh per month. This i s where the majority of households are indeed consuming, although there is no evidence that households make their consumption decisions based on such calculations. 14 Figure 2.8. When comparing effective versus quoted electricity charges, regressivity (for consumption under 300 kWh) becomesclear 18.0 Ec10.0 3, t; 8.0 V 6.0 4.0 2.0 0.0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 khhlmonth Source: WorldBank SIA survey (2008). 37. Low welfare households in Lebanon consume electricity at levels similar to the middle class. As seen in Table 2.4, households inthe bottom quintile consume on average from 295 to 488 from peak to o f f peak, while 31dquintile households consume from 338 to 540. This i s inpart because the effective tariff for the average households i s low and has not changed in over a decade. The implication, as noted above, i s that the pricing structure creates little incentive to constrain electricity consumption. Nonetheless, few households think o f their electricity bills as low, given the huge inconvenience associated with the service. Table 2.4. Average kWh and ampere levels (by quintile) Lowest Td 3rd 4th Highest Peak (kWh/month) 488 532 540 585 755 Off-peak (kWh/month) 295 317 338 419 489 Amperes 12.9 13.4 14.0 15.2 16.8 Source: WorldBank SIA survey 2008. 38. During peak months, many poor households consume above 300 kWh. Although during off- peak months 37 percent o f the bottom quintile consumes over 300 kWh per month, duringpeak months, three quarters o f this group does, and about one quarter consume more than 500 kWh (Figure 2.9.). While the large fluctuations are found for every quintile, it is noteworthy how elastic-in response to seasonal changes-demand for electricity i s among the poorest consumers. It suggests that for many households the cost o f electricity does not inhibitthem from usingelectricity-intensive appliances. 15 Figure2.9. kWh consumptionpatternswithin quintiles(EdLNetwork),by tranche Peakmonths Off-peak months >500 u401-500 E 40% t l u301-400I 101-3001 i 21% 1 it i s regre~sive.~~ index o f below 1.0 would indicate regressivity, and greater than 1.0 An progressivity. Note that the outcome i s graphed in deciles (as opposed to quintiles) to give a more disaggregated picture o f the results. 50. Tariffchange scenarios. Inthe simulations, changes are made to the existing tariff structure43to assess four (out o f a possible large number of) tariff change scenarios along each parameter described above. Scenarios 1 and 2 maintain the current tariff structure and apply across the board tariff increases o f 40 percent and 100 percent, respectively along the lines described in the electricity PER.44Scenario 3 changes the tariff structure by raising substantially the tariff rate for consumption above 300 kWWmonth, while Scenario 4 raises the tariff substantially above 100 kWh month level, and also eliminates the regressive "rehabilitation fee" (see Box 4). The choice o f simulations for Scenarios 3 and 4 was based on attaining the same level o f revenue (from households) that would be raised by Scenario 2. 41The index is based on the ratio o f share of electricity expenditure at the mean of selected percentages of sample cumulative distribution to the mean share of electricity expenditure of the entire sample. It i s not necessary that all shares for various deciles or quintiles be equal. The objective i s to comparethe average of the poorestto the average of the whole sample. This formula explains the slope andthe differencebetweenthe curve andthe horizontal line in Figure 2.15 42The measure here provides an additional assessment tool and does not target such unrealistic and utopian goals as equal electricity shares in expenditures for all. It is important to bear in mindthat the key issue is affordability as tariffs are raised, not affordability at current prices. While this study finds that electricity is generally affordable for virtually all households,as tariffs rise it will be important to assess whether lower income groups are being asked to share a disproportionate share of the increase, as measuredagainsttheir overall expenditures. 43The projected revenue increaseswas heldroughly constant for scenarios 2,3 and 4. 44 Although PER oil price assumptions (US$66harrel) are above current international oil prices (at around US$45harrel), they are still significantly below the average oil priceharrel in2008 (US$1001 accordingto the most recentWorld Bank assumptions); inaddition, given recent volatility inprice trends, they still serve to the purpose of providing indicative illustrative simulations. 22 51. The illustrative simulations do not take into account gradual implementation, which is a reasonable policy choice. The gradual or incremental introduction o f tariff increases is prudent and commonly used tariff reform strategy. Especially in the case o f Lebanon, because o f the substantial increases required, tariff changes mightneed to be phased in over a longer period. The average (peak and off-peak) bill o f each quintile is used for the simulations. Box 2.4. Scenarios SCENARIO 1: 40 percent tariff increase. Increase o f 40 percent in tariff rates at all levels (draws on Scenario 1 inthe Electricity PER, designedto break even before debt service andcapital cost). SCENARIO 2: 100 percent tariff increase. Increase o f 100 percent intariff rates at all levels (draws on Scenario 2 inthe Electricity PER, designed to break even after debt service and capital cost). SCENARIO 3: Tariff structure change. Increases the tariff levels for each block, by 50 percent, 75 percent, 100 percent, 150 percent, and 100 percent, respectively. SCENARIO 4: Tariff structure change with removal o f fee: Increases the tariff levels for each block, by 75 percent, 100 percent, 100 percent, 150 percent, and 125 percent, respectively, and eliminates the rehabilitation fee o f 5,000 LBP. Key Findingsof TariffReformSimulations 52. The tables and Figure 15 below present the simulation outcomes with the following main findings: Affordability. Innone o f the scenarios does the average electricity share o f expenditures reach the internationallyaccepted norm o f 10percent. The highest increase for the lowest quintile is observed under Scenario 2 where the electricity share o f expenditures reaches a maximum of 7 percent. The outcomes o f all four scenarios considered inthe study are within the norms for affordability . Increase in electricity bill. Overall, across all 4 scenarios, the average monthly bill for the lowest quintile increases within the range o f U S 6 . 2 9 to US15.72 (or 26 percent to 66 percent). Looking at Scenario 2 (relatively least `affordable' for the poor), we see that the 100 percent increase in tariffs across the board translates into a 66 percent increase in the bill o f lowest quintile. The least regressive scenario i s Scenario 4, where tariffs are increased by 75 percent to 150 percent, yet bottom lowest quintile sees a 50 percent increase intheir bill (with their electricity share o f expenditures below 6.5 percent). This i s largely explained by the elimination o f the rehabilitation fee under this scenario. The highest average increase o f US42.34 in the bill is experienced by the highest quintile under Scenario 4 (with their electricity share o f expenditures at 5.4 percent). However, if electricity rationing is decreased as tarifs are raised in tandem with supply improvement (over aperiod of years), households will be able to transfer what they currently spend on private generation to EdL, which will partially compensatefor increase in their electricity bill. Progressivity index. The progressivity index shows that that Scenario 4 (removing 5,000 LBP rehabilitation fee) comes the closest to moving the tariff structure in a progressive direction. This is largely because removal o fthe rehabilitation fee lowers the average cost per kWh for low electricity consumers considerably, while revenues from the reform o f the tariff structure compensate for forgone revenues from this source. 23 Figure 2.15. Affordability and progressivity (cumulative) Scenario 1 Scenario 2 Scenario 3 ................. -- Base Case - -1c- -A- ......................... Scenario 4 : : - ' .................................................... *-i----_- *-----* ...................................................................... - I 10 20 30 40 50 60 70 80 90 100 Sample Distribution (pcexp) Table 2.8. Changes to nominaltariff rates Tariff block Base case Scenario 1 Scenario 2 Scenario 3 Scenario 4 kWh kWh Change kWh Tariff kWh Tariff kWh Tariff kWh cost cost change cost change cost change cost 500 13.3 40% 18.7 100% 26.7 100% 26.7 125% 30 Source: SIA survey (2008) and WorldBank stuff calculations 24 Table 2.9. Illustrative Scenarios and their Impacts (EDLexpenditures only) * < `SCpiARIO1 Progressi\ ity I I ' inde\: 0.920 Quintile Increase in Increase in bill Share of bill(YO) (US$) household expenditure( O h ) Lowest 26 6.29 5.05 2 28 7.68 4.26 3 28 8.11 3.57 4 29 9.75 4.20 Highest 31 15.46 3.48 SCENARIO2 Progressivity x -* index: 0.927 Quintile Increase in Increase in bill Share of bill (YO) (US$) household expenditure (%) Lowest 66 15.72 6.7 69 19.19 5.7 70 20.26 4.8 72 24.38 5.7 Highest 78 38.65 4.8 SCENARIO3 Progressh it3 index: 0.935 Quintile Increase in Increase in bill Share of bill (YO) (US$) household expenditure (YO) Lowest 55 13.13 5.9 2 66 18.34 5.3 3 69 19.95 4.4 4 75 25.38 5.5 Highest 80 39.64 4.9 .c SCENARIO4 Progrcssi\ ity 0.948 ~ index: Quintile Share of bill (US$) household 11.89 5.3 62 17.1 4.8 64 18.7 4.0 4 73 24.5 5.2 Highest 85 II 42.34 II 4.7 25 VIII. CONCLUSIONS The conclusions focus primarily on household welfare issues, especially among vulnerable groups. They complement the technical and detailed recommendations made in the Electricity PER. 53. Design a more effective and simplified tariff structure. The current inverted tariff block structure i s not progressive-partly because it subsidizes all households and partly because effective kWh prices differ markedly from the quoted prices. This i s especially so for low electricity consuming households. This is partly because fixed costs (including the 5,000 LBP rehabilitation fee, stamp fee and ampere charge) form a large share o f the bill. These high fixed costs currently penalize households that consume low levels o f electricity and reducing andor eliminating them would have positive equity results. 54. Ensure proper sequencing in implementingpolicy measures. The burden resulting from EdL's service decline has increased significantly, despite the fact that the cost o f electricity purchased from EdL has remained relatively low for consumers. However, increasing tariffs to cost recovery levels without moving to quickly to introduce service improvements would meet with resistance. Consumers must feel that the burden o f reform is not placed squarely on their shoulders but i s shared with EdL. While all the reform objectives outlined in Table 2.10. are crucial, they have different implications for consumers, EdL Table 2.: 1. Implications of key elect city reforms for different .akeholders Objective Implicationsfor Implicationsfor EdL Implicationsfor private households generationbusinesses Increasehours of 1 Reducedneedfor 1 Increaseinoperating 1 Short andmedium- service private generation andmaintenancecost term gains as fuel 1 Improvedquality of life 1 Increasedelectricity expenditures decrease 1 Improvement inwelfare consumption by more than as more budget 1 Increasedinvestments subscription costs available for other on generation 1 Potential long-term goods and services. losses as some exit the business Restructure 1 Smaller tariff increase 1 Fewer high-consuming 1 Negligible impact tariffs (reducing for households that households receive fixed costs) consume less electricity cross subsidies Raise tariffs to 1 Higher prices slightly 1 Ability to cover O&M 1 Makes private cost recovery reduce welfare, unless and investments generationmore accompaniedby 1 Greater willingness to competitive as EdL increasedhours of supply households charges converge service with subscription 1 Householdshave less rates, but may also money to spendon dampen demand. private generation Improvebill 1 Nonpaying customers 1 Ability to cover O&M 1 Reduces available collection, lose and investments expenditures for reducinglosses 1 Distributionbecomes 1 Greater willingness to private generation due to theft more equitable as more supply households householdspay Improve Increasedchance of 1 Better servicedelivery 1 Eventual decline in governance meeting previous four 1 More accurate billing business objectives 1 Pathto cost recovery and independence 26 and private generators, and how they are sequenced i s important in terms o f acceptability and short-term gains and losses. 55. Improve service by reducing rationing, especially in the areas outside Beirut which experience long blackout hours. Among other things, this will have the effect o f readjustingthe burden imposedby rationing from the poor to the non-poor. Although there may be good reasons for keeping rationing inthe capital to a minimum, it gives rise to social inequities. Beirut households have higher welfare levels and they are most able to afford paying for expensive substitutes, yet have the least need for them. Most households in other regions must choose between going without electricity and spending significant amounts on private generation. 56. IdentifV key operatingguidelinesfor informal electricity sector. The informal electricity sector, served by hundreds o f private generator businesses, provides up to 30 percent o f Lebanon's electricity and i s an indispensible service to many households. Yet the sector lies entirely outside the legal framework and does not pay taxes to the state. It must be acknowledged that private generation will play a significant role in electricity generation for years to come. Hence, identification o f proper operating guidelines that will protect and benefit consumers while ensuring continued availability o f this alternative source should be given consideration and merit additional research work. Without dampening private sector activity, or generating an additional and onerous layer o f bureaucracy, the guidelines could have several objectives, including: 0 bring private generator businesses into the formal sector so that customers have access to means o f redress for losses or damage incurredby faulty wiring or through other fault; 0 enable the state to tax the sector like other businesses; set technical standards for service; 0 monitor the sector and either set guidelines on subscription fees (it would be desirable to assess ifthe generator businesses operate as a monopolistic structure). 57. Harmonize electricity sector reform with social safety net reform over time. Given the ongoing progress with the design o fa social safety net system with an adequate targeting framework inLebanon, it will be important to consider electricity reforms, inparticular tariff structure adjustments, in the context o f the objectives and tools o f the new social safety net system over time. In some countries social objectives are pursued through social policies and not necessarily through energy pricing. This is an option that should be kept in mindwhen the new social safety system i s up and running, at which point the changes to the tariff structure that are currently being debated for implementation from 2009 onwards could be reconsidered. Given the generally fragmented and non-governmental nature o f social safety net provision in Lebanon, any new arrangements would likely need to deal with political economy issues, which are beyond the scope o f this paper.45 45 See LebanonEnergy Sector PER, World Bank, 2007. 27 CHAPTER3. WATER "The water sector is one of the most important sectors in Lebanon due to its huge impact on the economy,growth, environment andpoverty alleviation...Attentionshould be given to the long term planning and management of urban and rural development to secure sustainable developmentand to ensure that adequateand equitable services are rendered to the population.)'- Water and Wastewater sector: Sector Reform Strategy Workshop. January 15, 2007. GrandSerail- Lebanon I.INTRODUCTION 58. Lebanon's ample water resources have not translated into sufficient water supply. One o f the few countries in the region benefiting from plentiful rainfall, Lebanon has potentially sufficient water resources to meet domestic demand. Nonetheless, transforming resources into quality drinking water for the entire population has proved difficult despite Government investments in infrastructure. As a result, much o f Lebanon's groundwater simply flows out to the sea, and water supply and service remains poor and intermittent in most parts o f the country. Water tariffs do not allow for cost recovery and although low, they are regressive. Furthermore, in the absence o f metering (outside a few limited areas) there is a disconnect between water supply and household demand. 59. Much of the sector's infrastructure is old and deteriorated, and its operation and management are inefficient. Despite reform measures undertaken earlier in the decade, lack o f cost recovery in the sector i s reflected inthe financial deficits runby three o f the four Regional Water A ~ t h o r i t i e s ~inhibiting ~ , their ability to cover operating and maintenance costs. As a result the Government often steps in to pay for operational expenses in additon to directly financing investment in infrastructure. Existingweakneses ininstitutionalcapacity delay adressingthese key problems and improving the management and supply o f water services. 60. Deficiencies in existing water supply services affect some householdgroups more than others. Obviously, not every household is affected to the same degree by the water situation. This chapter assesses impacts o f water provision in Lebanon on different categories o f households and what sequence o f policy options present opportunities for addressing social concerns given supply constraints on the one side and household consumption behavior onthe other. 11. KEYSECTOR ISSUES OVERVIEW -ASUMMARY 61. The water sector reforms initiated in 2000 have only partially been realized. A new Water Sector Law (Water Law 221) was promulgated by Parliament in M a y 2000 leading to the eventual consolidation o f the 22 separate public water servicing authorities into four autonomous Regional Water Authorities (RWAs) in 2005. One o f the key objectives o f this measure was to improve efficiency, but this has provedelusive. According to the law, the new RWAshavethe following responsibilities: operate and maintain the hydraulic systems that are located in the areas under their respective jurisdiction, implement the required investments inline with the Country's Master Plan to be prepared by Ministry o f Energy and Water (MOEW), secure the replacement o f water systems, achieve cost recovery and propose 46 Since RWAs do not have audited financial statements, prepared according to standard accounting practices the fmancial information remains incomplete for most of them andthe financial indicators are only indicative. However, a1 the estimates confirm that with the exception of BMLWA the Authorities have been faced with fmancial difficulties. 28 tariff adjustments. The law states that RWAs should aim for financial sustainability while taking into account the country's general socio-economic conditions and the population living in their region. The reforms have not yet achieved the desired results, in part due to the aforementioned lack o f institutional capacity o f the RWAs, and in part due to the lack o f an enabling environment. Beirut Mount Lebanon Water Authority (BMLWA) is an exception, having achieved a certain measure o f financial soundness and self-sufficiency. However, this achievement needs to be considered within the context o f BMLWA providing the lowest per household water supply among all four RWAs. 62. Most Lebanese households have access to improved water, but this does not mean they rely on it. The UNestimates that 100 percent o f Lebanese households use an improved water source47compared with 88 percent o f households inthe MENA region. However, the SIA survey finds that while access to improved water may be (nearly) universal, the actual number o f households relying on an improved water source i s below 100percent. This i s because many use bottled water and tanker truck water, which are not considered an improved source (because o f concerns about quality and problems with availability). 63. The effective cost of public water to households is often much higher than the quoted tarvj The quoted average cost o f lm3/day public water is US$0.37 in Lebanon, which is equivalent to the average for MENA countries and slightly above the Upper Middle Income Country (UMIC) average (see Table 3.1.).48 Many households, however, pay a higher actual, or effective, rate because they receive considerably less water than lm3/day.A significant amount of the costs of unaccounted for water (UFW), which exceeds 50 percent o f water produced in Lebanon, is, hence, borne by consumers.. Furthermore, unreliability o f supply imposes its own costs, because o f purchase o f alternative supplies, storage, etc. Table 3.1. Average Water Tariffs and Degree of Cost Recovery Percentage of water utilities whose average tariffs appears to be: Too low to Enough for Average Water cover basic Enough to cover O&M and Region Tariffs (US%/m3)* O&M most O&M partial capital UMIC 0.34 39 22 39 LMIC 0.3 1 37 41 22 OECD 1.04 6 43 51 MENA 0.37 58 25 17 non Source: ADB 2004, ADESAS 2005, GWI 2004, NIUA 1999, WBstaffcalculation *Average tariff are based on residential consumption of 15 m3. ** The actual tariff varies by RWA. 64. Water tarvfs do not cover operation and maintenamce costs, while collection rates are low. As shown in Table 3.1., Lebanon's cost recovery performance i s very poor compared to other countries. Three o f the four RWAs-BMLWA is the exception40 not have enough revenue to cover operating and maintenance (O&M) costs, and the Government often steps in to pay for operating expenses4' in 47 Improved drinking water sources include: piped water into dwelling, plot or yard; public taphtandpipe; tube wellhorehole; protected dug well; protected spring; and rainwater collection. Unimproved drinking water sources include: unprotected dug well; unprotected spring; cart with small tankidrum; bottled water; tanker-truck; surface water (river, dam, lake, pond, stream, canal, irrigation channels). Bottled water i s considered an unimproved source due to limitations in potential quantity, not quality. Source: United Nations Millennium Development Goal Indicators. Available online: httD://md~s.un.org/unsdlmdg/Metadata.as~x?IndicatorId=0&SeriesId=7 10. 482008 World Development Indicators. 49 Includingelectricity arrears and insome cases staff salaries. 29 additon to financing investments in water infrastructure. Most RWAs also incur high O&M costs arising from inefficient management o f services and degraded water networks. The collection rates are low due in part to an inefficient yearly billing system and in part to the unwillingness o f some customers to pay for service because o f unreliability or other reasons" (see Table 3.2.). Table 3.2. Summary ofmainperformanceindicatorsfor the RWAs (2004) Water Authority Subscribers Working BillCollection * (000) Employees/100 0 connections Ratio (percent)** (percent) BML 440 2 6 76 73 North N a 3,7 202 49 Bekaa 67 579 547 47 Best Practice* South ** 120 2.2 188 70 Nu <5 70 100 *Subscribersinclude residential, commercial and industrial clients.** International Water Association. **Operational costs divided byrevenues. ***International Water Association. Source:World Bank Water Sector Public ExpenditureReview, (forthcoming). 65. Weak institutional capacity constrains the RWAs' ability to operate on a commercially sustainable basis. RWAs have difficulty in building a skilled employee base, especially given a government-mandated hiring freeze and low remuneration. This, combined with lack o f appropriate resources for operation to an acceptable level, limits their ability to improve services. 66. Dependence of waterpumping stations on an unreliable electricity supply hampers operations and can cause damage. Unreliable electricity supply limits water pumpingto the regions and fluctuations in supply damage equipment, which cannot easily be repaired or replaced under the RWAs tight budget. This affects the efficiency and quality o fthe services offered by the water authorities, imposing a further burden on their finances and weakening overall performance. 111.PUBLICWATERSUPPLY 67. InLebanon,public water is supplied to connectedhouseholdson a subscriptionbasis.As noted above, a fixed annual fee is charged to each household based on the allotted quota. There are no meters (aside from a few urban areas under pilot projects) and the amount delivered i s regulated by a gauge system (restricted flow mechanism). Under this system, the volume i s controlled via a combination of pipe diameter and water pressure: the gauge sets an upper limit on the amount o f water the system can deliver to each household. The SIA survey results show that the vast majority of households have a lm3 gauge connection-less than 2 percent have the larger 2m3gauge.51 68. R WASare obligedtoprovide lm3of waterper dayper connection,an amount designed to cover household needs. Nevertheless, not all households receive the same amount o f water. The amount delivered depends both on the amount o f time water flows through the pipes, and on water pressure. These indicators vary between regions, within regions, and even within apartment buildings, since water pressure tends to be lower on the upper floors. Billing is, on the other hand, based on a contractual fixed 50 Bill collection rates are changing, inboth directions. While BMLWA has achieved a collection rate of above 80 percent, World Bank calculations suggest that the Bekaa Water Authority collected from only 33 percent of customers in2007, a significant decreasefrom 2004 levels (see Table 2). 5 1According to RWAs, households can apply for a 2 m3gauge if they have a dwelling larger than 200 m2but in practice the survey found that gauge size was not related to domicile area 30 consumption o f lm3per household per day, regardless o f the amount actually delivered. Aside from a small number of houses, where new meters have been installedunder pilot programs in Tripoli and Saida, actual household water consumption i s neither monitored nor used to calculate charges. Nevertheless, even those households with meters still pay fixed charges for water (see section XI1below). 69. Many households do not receive water on a daily basis. The SIA survey revealed that, on average, only one quarter o f connected households receive water every day, with 40 percent receiving it every other day and the rest less frequently. Some connected households reportedthat they do not receive water from the public network for several months out o f the year (see Table 3). Given that the gauge system is designed to allow delivery o f lm3per day, it is obvious that, based on the number of days of water supply in some areas, this target is not being met. It i s not possible, for example, to compensate in one day for water not delivered the previous day by delivering twice as much water. The gauge will not permit it. Table 3.3. PublicNetworkWater Availability Households(percentage) Beirut ML North Bekaa South Nabatieh Average Everyday 10 14 59 43 38 20 26 Everyother day 52 51 14 24 44 21 40 Every3 days 37 27 17 14 17 31 25 Once a week 1 7 9 11 1 17 7 Never 1 2 2 8 0 10 3 Source: WorldBank SIA survey (2008) 70. I n the absence of metering system, supply of and demand for public network water is de- linked at the household level. The amount o f water provided to each household i s unrelated to the amount o f water a household would consume and pay for if it had a choice. The key factors behind household demand-such as, price, income, quality and consumption n e e d s 4 0 not play a role, except insofar as a household chooses not to connect to the network at all.52This leads to problems including: (i)Wasteandinefficiency:, Thesystemcannot`redistribute' waterfromthosewhohaveaccessto more water than they need to those who do not receive enough; (ii)Built-in inequities in the system. Aside from the fact that households receive different amounts o f water depending on their location, inequity appears invarious ways: Larger families receive less water per capita, and lower income households tend to be larger . (5.6 members in bottom quintile) than higher income counterparts (4.4 members in top quintile); Since households have no means o f controlling their water expenditures. To increase water consumption beyond what the gauge supplies, they must purchase water from outside the public network. Ifthey choose to consume less, they must still pay the same amount as their higher consuming neighbor. Either choice constitutes an economic cost to households that would be avoidable under a metered system. 52Approximately 10percent ofhouseholdshavenot connected to the system although they have access. 31 Iv.HOUSEHOLD WATER DEMAND AND CONSUMPTION 11. The current levels of household water supply target do not seem to corroborate with the lzouselzold demand. In some sub-regions RWAs have sufficient supply to meet the lm3target, in other areas they do not (especially during the summer). However, the targeted volume per household o f Im3/dayreflects a somewhat arbitrary benchmark set by the RWAs, and does not reflect actual household demand. One cubic meter of water per day per household implies an average per capita consumption of about 220 liters (based on the average household size of 4.4753),which is, in fact, above average for a middle income country. According to an unpublished study conducted by the Beirut Mount Lebanon Water Authority, it is estimated that a Lebanese family o f five requires 700-800 liters o f water per day, or 140-160 liters per person. The lm3target is probably unrealistically high and not the best metric in any case. 72. Estimates show that water demand exceeds effective water supply in most parts of the country. Although, as noted above, volume of water i s difficult to measure, consumers have a general idea of how many hours o f water they require to meet their household needs. Self-reporting o f hours o f water supply per week allows a rough estimation o f differences between water received and water demanded. According to this proxy measure, 80 percent o f connected households report that during summer months (high season), their demand (interms of hours they need per week to meet their needs) exceeds supply. This ratio decreases somewhat inthe low season (winter months), when 60 percent o f households report not having enough water. The greatest water deficit i s found in the Bekaa Valley region, while in the North supply actually appears to exceed demand (see Figure 3.1.).54 Figure 3.1. Networkwater availability demand mostly exceeds supply - 16 '4 -I 12 - 10 7 ML North Bekaa South Nabatieh~ .-- __ ____ -___.---__Beirut ~ &aLow season 4.1 ~ '88 High season I 6.4 109 I 8.2 1 6 1 I 10.0 1 7.1 1 pf-_p-t rn Required to meet needs 7 8 107 6 6 1 142 10.9--p-I 5.7 i __ - Source: WorldBank SIA survey (2008) 73. Hours of service per day is not correlated with welfare levels. Another consequence of the disconnect between supply and demand i s that household public water consumption is not linked to welfare but to regional capacity. As noted above, the Beirut-Mount Lebanon region, where overall welfare levels are highest, receives very restricted hours o f service relative to other regions (see Figure 3.2.). This is in contrast to electricity provision, as noted in Chapter 2, which is strongly correlated with 53ibid. 54Availability has beentranslated into hours per day to harmonizeit with previous analyses. 32 welfare. Note that for Lebanese households, filling the water storage tank has become the key adequacy indicator. Although round the clock service is the norm in developed countries, failure to provide a constant flow o f water through the network appears not to be a major bottleneck for households in Lebanon, where water storage tanks are in common use. What matters i s that households have enough water to keep the tanks supplied. Figure 3.2. Networkwater availability is not relatedto welfare 12.0 1 10.8 average winter -------------_ average summer Bottom Q2 Q3 Q4 TOP quintiles p i G GI 'mwinter I Source: WorldBank SIA survey (2008) v. HOUSEHOLD CONNECTIONS TO THE PUBLIC NETWORK 74. Compared withfour years ago, connection rates are up, but hours of service are down. Based on SIA survey data, an estimated 80 percent o f households are connected to the public network system (see Figure 3 below), which i s an increase from 76 percent according to the 2004 household survey-an additional 1 percent are connected to a private water network55(see section VI1 for further discussion). However, intermittent water supply has become the norm, with an average 6 hours per day inthe summer season and 9 per day in the winter season (based on the SIA survey data). Compared with the earlier estimates, this represents an overall decrease in supply, particularly duringthe winter months. Beirut and Mount Lebanon region is an exception, showing an increase inwater supplied duringthe summer months (Table 3.4.). Table 3.4. Daily water supply appears to have declined during the summer (hourdday) Beirut ML North Bekaa South Nabatieh Average Summer (highseason) 2002-05 3* 22 8 10 N a 6 Winter (low season) 2002-05 13* 22 8 10 N a 13 Source: Diagnostic Analysis of the Water Authorities (Socie'te' Gkne'rale, 2002; COR, 2003b, 2005b ;ONDEO- Liban, 20054 b); WorldBank 2008 SIA survey Note: * Beirut and Mount Lebanon were not disaggregated in the earlier estimates. 55See pages 36-41 for further discussionon supply from private network andother sources. 33 75. Population density confers both advantages and disadvantages on connections in Beirut. Although Beirut households may receive relatively little water, they are the most likely to be connected (see Figure 3.3.). This is in stark contrast to the North, where the share connected i s a little over half. Beirut has an advantage because the high population density and the ubiquity o f water mains make the marginal cost of additional connections to the public network lower than in larger, less populated regions.56On the other hand, high population density means there i s less water per household, given the current supply limitations. Figure3.3. Connectionrates are highest in Beirut 100 -, 96 -- Beirut ML North Bekaa South Nabatieh Source: WorldBank SIA survey (2008) 76. Low-income households are less likely to be connected to the public network, but this is primarily due to location, not affordability. In the lowest quintile 62 percent are connected to the network, compared with 86 percent for the highest quintile. Being unconnected to the public network has less to do with affordability o f the water bill (and even connection rates)57than with region, given the pattern o f lower connection rates outside o f Beirut and the regional poverty profile. Only 3 percent o f unconnected households claimed they `cannot afford' or `don't want to pay' (see Table 3.5.). RWAs that cover rural areas face additional costs because less densely populated areas are more expensive to serve. For example, the Bekaa RWA covers 40 percent of Lebanon, yet has only 12.6 percent o f the p~pulation.~' 77. Absence of a public water network is a key determinantfor household connection. The most common reason given for not being connected i s that the household had no choice: among the 20 percent o f unconnected households, over half report that there i s no public network available in their area. This indicator exhibits strong regional variation, with households in Mount Lebanon, the North, and Bekaa four times as likely to report the absence o f a public water network than in Beirut, the South and Nabatieh. Most o f the remaining unconnected households report using other sources o f water. A complementary regression analysis also finds no statistically significant causal relationship between 56Inaddition, the geographicalareaof administrative Beirut is smaller andtherefore extendingpipes andconnection has a lower cost than other regions where long connection lines needto be made, especially inrural areas. 57 The cost of connecting to the network varies, depending on things like geographical location of the house, whether the building previously had an old connection, whether and whether it involved rehabilitating an existing line which had been cut, etc. The cost may be upwards of 450,000 LL, or about US$300, in Beirut, and somewhat less inother regions. 58 2004 Multi Purpose Survey (MPS), using the lower poverty line calculated in "Poverty, Growth and Income DistributioninLebanon", Executive Summary, UNDP, 2007. 34 household water connection and a host o f factors such as education status, availability o f electricity, regions, asset ownership (to measure wealth) and household demographics. This might be an additional indication o f the de-linkage between demand and supply Table 3.5. Reasonshouseholdis not connectedto public network Reason (percent) No public water network (inarea) 54 Other sources o f water, o f which: 33 Artesian well 27 Private network 6 Other 1 Do not trust public water network 5 Cannot afford, don't want to pay 3 Public water network not frequent 2 Rent (do not ownthe house/apartment) 2 Source: WorldBank SIA survey (2008) VI. QUALITY OFPUBLICNETWORK WATER 78. Quality of public network drinking water is a significant issue in Lebanon. Concerns about quality are widespread among consumers. Of households connected to the public water system, only 53 percent use it for drinking purposes (see Figure 3.4.). Perception o f drinking water quality varies by region, with reportedsatisfaction far higher inBekaaValley than elsewhere (see Figure 3.5.). Figure 3.4. Use of public networkwater for drinking shows strong regionalvariation 100 1 80 -1 77 Beirut ML North Bekaa South Nabathieh Source: WorldBank SIA survey (2008) 79. Two thirds of households avoid using public waterfor drinking out of safety concerns. The most frequently cited reason for not drinking public water i s perception o f safety relating to health and hygiene (66 percent o f respondents) followed by poor taste (47) and service interruptions (24 percent). However, most households resort to buying water from alternatives sources. Only 11 percent use some 35 form o f sterilization-purification method, for which they spend around 247,000 LBP (US$165) per year on average. Figure 3.5. Perceptionsof public network drinking water quality vary by region 80% 60% 40% 20% Beirut ML North Bekaa South Nabatieh irn Poor or very poor Adequate 0 Good or verygood ~ Source: WorldBank SIA survey (2008) 80. Multiple factors contribute to poor water quality. First, Lebanon's water resources are overexploited due to excessive pumping. There are an estimated 40,000 private wells, compared with 3,000 in 1970" which i s factor in contamination o f the groundwater by seawater. Most of these wells are illegal and were dug during the civil war between 1975 and 1990. Second, inadequate provision o f sanitation services has led to polluted streams, rivers and aquifers. Up to 70 percent6' o f all natural sources are affected by bacterial contamination. Lebanon's potable water was ranked below World Health Organization (WHO) standards for drinkingwater.61 VII. ALTERNATEWATER SOURCESUSEDBYHOUSEHOLDS 81. Inadequate and unreliable water supply pushes households to purchase water from alternate sources. There are a wide variety o f alternate water sources available to households, and these are used by both connected and unconnected households (Figure 3.6.). This covers water used for all purposes, including drinking. Households without connections tend to use artesian wells and delivery trucks more than connected households, but an almost equal number o f connected households purchase delivery truck water for service use, and buy water in gallons or bottles for cooking and drinking. Among all connected households, a small share (4 percent) does not use public water at all, whether for service or drinking purposes. 82. The marketfor alternative water supply provides a greater range of price and quality clzoices than the alternative market for electricity. While most urban households seeking back-up electricity supply can only buy from their local private generator business, when it comes to water, households have a larger number o f options. Depending on the importance o f quality, they can buy bottled or gallon water for drinking and cooking, and still use public water or well water for other needs. Still, as with the 59 Source: Water PER report, draft, World Bank. 60Republicof LebanonPolicy Note on Irrigation Sector Sustainability, World Bank,April 2004. WHO. 36 electricity sector, there is little regulation in the alternative water sector, an especially important issue in light ofwater safety withrelated healthrisks. Figure3.6. Householdwater consumptionby source-all households deliwy trucks I 25% Among unconnected 13% households mineral water-gallons Among connected households mineral water-bottles 15% artesian well 25% primte network p 10% I public network 0% 0% 20% 40% 60% 80% 100% 120% Source: WorldBank SIA survey (2008) 83. The quality of most alternative water sources is considered better than public water. Although an alternate source does not necessarily mean better perceived quality than public network water, only artesian wells and community tanks rank relatively lower. Households report that water from artesian wells i s often salty and unfit for drinking. It i s very likely that the comparison figures demonstrate selection bias, given that households are comparing their drinking water source to the alternatives available to them (not all existing alternatives). 84. Over one third of connected households do not use public water for drinking, reflecting the perceivedpoor quality of network water. On the other hand, many households rely on gallon bottles, regardless o f their connection status. A great number o f unconnected households (40 percent) rely on artesian wells which, as noted above, often provide water o f poor quality. VIII.WATERTARIFFS 85. Water tarifJs are set independently by each RWA. Tariff rates were adjusted following the consolidation o f public water establishments into four RWAs in 2005 and have become more uniform (compared to pre-reform structure o f considerable variation). Nevertheless, tariff differences across regions still exist (see Table 3.6.). Tariffs are proposed by the Water Authorities and approved by the government. Bills are paid as a lump sum on a yearly basis. This type o f billing arrangement typically disfavors lower income households, who may struggle to make large single payments. The tariffs are based on the contractual volume o f water o f lm3/day. 37 Table 3.6. Water tariffs by region (annualfees, LBP) Beirut/ML North Bekaa South Based on lm3/day 200,o 00 180,000 140,000 175,000 Gauge maintenance 35,000 30,000 20,000 25,000 Total 235,000 210,000 160,000 200,000 Source: RWASfee structure, 2008 Note: Figures does not include government VAT (I0percent) and stampfee (I,000LBP) 86. Tariff rates reflect historicalpricing policies, not cost recovery levels. Inmost cases tariff rates are insufficient to cover O&M costs, let alone capital investment-ne water authority reported that they would need to more than double annual tariffs in order to cover their operating costs. Among the four RWAs only the Beirut-Mount Lebanon Water Authority has been able to achieve cost recovery. On the other hand, if one takes into consideration the high level o f network leakage throughout the country- unaccounted for water exceeds 50 percent of water produced-consumers are paying substantially more thanthe nominal annual tariff. IX. HOUSEHOLD WATEREXPENDITURES 87. Combined water expenditures on all water sources are in line with affordability norms, but non-connectedhouseholds spend more. The World Bank has set 3-5 percent o f household budget as the recommended share o f income spent on water,62and most Lebanese households fall within this range (see Figure 3.7.). However, it i s also clear that connecting the remaining households would increase water affordability for them. Except for the top quintile, connected households spend a lower share o f their household budget on water than unconnected households. The lower expenditures for the top quintile reflect the weight o f higher income among Beirut households in the sample. A steep drop off in budget share o f water expenditures is found between the top two quintiles. Figure3.7. Shareoftotal water expendituresin householdbudget II 6% a 5.6% 5.5% 5% 4.7% 4,50 4% E c 0 3% ln 2% 1% 0% 3 4 bottom 2 Source: WorldBank SIA survey (2008) 62World Bank (2002), Sourcebookfor poverty reduction strategies, core techniques andcrosscutting issues, Washington DC. 38 88. Expenditures on alternative water sources, however,far exceed those onpublic network water. Public network water represents half or less o f a household's total water expenditures. The bottom quintile spends as much on alternative sources as it does on public water, while the top quintile spends three times as much (see Figure 3.8.).63 Figure 3.8. Expendituresof connectedhouseholdson alternativesources 100% 80% 60% 40% I I I I 1 I I I Public water expenditureI 20% 0% bottom 2 3 4 top quintiles (annual expenditures in `000 LL) Source: WorldBank SIA survey (2008) 89. If households could rely entirely on network water they could cut their water expenditures signzjicantly. In absolute terms, reduced water expenditures for the bottom quintile o f connected households would generate large savings (as much as 220,000 LBP or US$147 on average), virtually as much as their current water bill of 201,000 LBP (US$134) on average (see Figure 8). Savings could be even larger for the upper quintiles. The convenience o f having a household connection has not been quantified but would be an added welfare benefit, and simply being connected can lower household's water bill. As shown in Figure 9, it i s the second and third quintile which would benefit the most (in percentage terms) fiom being connected, interms o f lower overall water expenditures. 1 Figure 3.9. Water expenditures of connectedvs. unconnectedhouseholds 1000 -; 89 888 813 811 8 0 8 r 800 1- A 600 524 525 IConnected 0 Unconnected 400 - 200 0 top bottom 2 3 quintile 4 Source: WorldBank SIA survey (2008). 63It must be kept in mind that expenditures do not reflect volume consumed. While a certain percentage of households will always use alternative water sources, if quality is improved, it is likely that a significant share o f households would shift to using public water. 39 90. Public network water is inexpensive in financial terms, but the cost of unreliability is considerable. Public water might appear relatively cheap, but higher effective price and the indirect costs o f unreliable supply make more expensive options attractive. Lebanese households spend approximately US$208 million total per year on bottled water (in small bottles or gallons) over the past year. The unit cost of water purchased in gallon bottles, ranges from US$lOO to US$220 per m3,while bottled water in small bottles (1.5 liter) costs up to US$ 500 per m3(see Table 3.7.).64 Ifhouseholds could rely entirely on network water it would cost them US$0.29 - 0.42 per 1 m3,but since very few receive 1 m3'day, and many are not satisfied with the quality, they essentially pay considerably more.65 Artesian wells and private networks are already cheaper but, at least inthe former case, quality i s often quite poor. Table 3.7. Estimatedvolume vs. non-volumetric unit costs for water Water source Range of costs (US$/m3) Low High Public network 0.29 -0.42 na Water truck 3 6 Gallonbottles 100 220 Small bottles 440 500 Average annual payment (US $) Public network 143 Artesian well 126 Private network 82 Source: WorldBank SIA survey (2008) and informal market assessment 91. Lebanese households with apublic water connection spend twice as much on alternative water sources than they are billed by RWAs. Estimates based on the SIA survey suggest that connected households spend US$219 million on all kinds o f alternative water sources, far more than the maximum o f US$104 million which RWAs collect annually (on the assumption o f full payment compliance).66This is in addition to the US$SS million spent on alternative, non-network water sources by non-connected households. 64Of course, the comparison must be understood within context - no one consumes bottled water in cubic meters. Still, the potentially `unnecessary' cost adds up to a significant additional householdbudget item. 65The low range for public network water i s based on the best case scenario of a household receiving lm3/dayin Bekaa (the annual cost of 160,000 LBP divided by 365 days) and inBML(235,000 LBP divided by 365 days). The high range cannot be calculated, but can be summarized as: the less water received per day, the higher the volumetric cost. A household that receives no water over the course of the year would, in theory, face an infinite cost per m3.Inpractice, such householdsdo not pay their bills. 66Because non-payment among residential consumers is a widespread problem, the amount RWAs actually collect will be considerably less than this amount. 40 Table 3.8. Water expendituresby source (million US%) Water Source (US%mln/ Year) Connected Unconnected Total A) Total alternative sources 219.5 88 308 Delivery Truck 54 33 87 Gallons 107 35 142 Bottles 54 11 66 Artesian Wells 3 7 10 Private Network 0.5 3 3 B) Public Network 104 0 104 Source: Author calculations based on WorldBank 2008 householdsurvey. Note: Aggregate expendituresare based on estimated906,916Lebanese households, calculatedfrom WDI (2006)population estimate. 92. Improvements to quality and increase in supply could represent significant savings for households.Although US$219.5 million spent on alternative water sources is a relatively small amount for the Lebanese economy, the expenditures represent only the financial costs borne by households. These expenditures are also an opportunity cost for households, since these are expenses that could be reduced, savings that could be spent on other goods and services, with a resulting welfare boost. Lower purchases o f water would reduce the market size for the various businesses engaged inthe alternative water market. This would, in effect, mean a redistribution o f revenue away from private water suppliers, but would be counterbalanced by the social benefits accruing to low income households. Hidden costs on the supply side (water losses, billing inefficiency, explicit subsidies, etc.) would o f course need to be taken into account in a full cost-benefit analysis.67 x.WILLINGNESS TO PAY 93. The much higher paymentsfor alternative water sources (thanfor public water) underlinesthe valueplaced on sufficient and good quality water. Water is an indispensible good and households will make every effort to obtain enough, though compromises will be made in terms o f quality and cost. However, when asked directly whether they would be willing to pay their water company more if service were improved, few were willing to substantially increase their expenditures. 94. Given the current conditionsand alternatives, households expressed reluctance topay morefor better public service. Despite dissatisfaction with quality and availability, when presented with a scenario o f better quality water and sufficient supply, households were generally unwilling to pay more for public service (see Figure 3.11.). Only half said they would be willing to pay more than an additional 21 percent for better service, while 29 percent would pay 50 percent (approximately US$66) per year more.68 61These other costs will be coveredinthe public expenditurereview that i s being preparedby the World Bank for the water sector. Contingent valuation method (CVM) was usedto estimatethe householdresponse to better service. Respondents were first asked ayeslno questionregarding their willingness to pay more under a scenarioof adequate supply and good quality service, choice for a higher rate (equivalentto either 5 percent, 10percent, 25 percent, and 50 percent). They were thenasked a follow-up questionregardingtheir maximumwillingness to pay. 41 Figure 3.10. Willingness to pay 0 10 20 30 40 50 60 Increase on current bill (%) Source: Author calculations based on WorldBank SIA survey (2008. 95. Willingnesstopay is linked with current satisfaction and werfarelevels. The highest percentage o f households who would accept an increase in water bill is in the South, the region with the highest public water availability and the lowest is in Nabatieh. Willingness to accept a higher bill is, as expected, also related to expenditure quintile: 41.7 percent are willing to accept a higher bill in the bottom quintile compared with 57 percent inthe top quintile. 96. Severalphenomenamay be embedded in willingness topayfindings. There are several ways to interpret willingness to pay findings. On the one hand, it may be a protest signal, an expression o f dissatisfaction with the current service. Onthe other, it may also reflect the fact that ifRWAs plan to raise tariffs, households know they can access alternatives. Finally, households may be answering strategically, adopting a `bargaining position' vis a vis the government by expressing low willingness to pay (see Box 3.1.). XI. PUBLICWATERMETERING 97. Metering is a long term goal in current circumstances. Lebanon is very gradually moving toward a metered system (metering programs have been piloted in Tripoli and Saida), which would rationalize water usage and be a key step in improving service. At present, approximately 4 percent o f households have a water meter but they still pay the same annual flat fee as households usingthe gauge system. However, for any water they consume in excess o f lm3per day, they are charged an additional fee based on volume. 98. Water meters would reduce wastage, align demand with supply, and reduce inequity. Households with meters would have the option o f adjusting their consumption. Since in the current system, everyone pays the same amount for public water (within a given region), a comprehensive, region-wide metered system will also reduce existing inequities and allow households to control the amount they use. It would especially benefit households that receive less than their allocation. 42 Box 3.1. Consumers explain willingness to pay When explaining his unwillingness to pay more, a respondent from Saida (with a water meter) said there was `no need for any `favors' from the government.' Another metered Saida resident said that they `already have enough expenses because of [their] high electricity expenditures.' A Mount Lebanon respondent said they are trying to reduce their payments, not raise them. A respondentfrom Mount Lebanon said that his unwillingness to pay i s related to the fact that he i s not connected. H e said he was comfortable with havingno connection and he does not trust the government. When the government gains the people's trust he would accept any increase because he would know that what his money won't be squanderedbut will be "used to serve the people inthe proper way". Reasonsfor willingness to pay more for publicwater were generally related to expected savings. Households would cut down on expenses by no longer havingto buy drinking water, andpart ofthose savings could go toward paying the higher tariff rate. A Tripoli respondentsaid he would pay more because he expectedto save the cost of buying drinking water and arespondentfrom a poor householdin Saida explained hisrationale by saying that he would pay the equivalentto what hishouseholdcurrently pays for drinkingwater. Good quality water seems to be of greater important than regular service (as many already receive a sufficient water supply to fill their storage tanks). Willingness to pay higher rates was often conditional upon having water that was potable, so that they no longer have to buy it elsewhere. A few middle class households from Mount Lebanon and Beirut were willing to pay as much as 200 percent more and still said they would be saving money if the public water were drinkable. A Mount Lebanon respondent, however, noted that he would continue to buy drinking water `no matter what' and therefore would not pay more than25 percent more thanhe currentlydoes. Ofthe meteredhouseholdsinterviewed indepth, all saidthat having ameter didnot affect their willingness to pay. Note: This box on the householdperspective on the electricity sector presents unjltered views of survey respondents with regard to water service. 99. Perceptions of meters depend on whether a household has used one. Among households without meters, the majority expressed skepticism regarding their value. Fifty seven percent of non- metered households reported that they would not like to have a meter. The responses suggest there is poor Table 3.9. Many householdsassociatemeterswith higher costs Reasonwhy respondentdoes notwant a water meter (percent) Highcosts, expected increaseintaxes 36.1 Cannot decreaseconsumption 22.1 Gauge is not bad, we're usedto it 14.3 No needto control water consumption 7.9 Better to have one bill per year 4.6 No needto change 4.6 Don't care. It is all the same 3.8 Don't trust inthe bill accuracy 3.5 Other 3.1 Total 100 Source: WorldBank SIA survey (2008). 43 understanding o f the benefits o f moving to a metered system, specifically how it would affect household consumption and expenditures. Many see them as part o f a new initiative aimed at raising prices. Some believe it will restrict them to a certain consumption level. 100. I n contrast, households with meters consider themselves better of$ Of the small group that has meters,6995.5 percent consider themselves better o f f than before. Those who support meters note that it allows households to control consumptionlevels, and introduce fairness into the system. 101. Increasing water supply and improving its quality has cost implicationsfor RWAS.Unless such reform measures are combined with tariff increases, RWAs have little incentive (from a commercial operations perspective) to increase supply since it will not translate into increased revenue-the flat fee structure means that the water company i s not rewarded for increasing supply. It i s therefore likely that water production and distribution will continue to require subsidies. Meters, by linking supply and demand, would go some way toward addressing this problem, but universal roll-out o f a metering program i s a long term process. XII. CONCLUSIONS 102. Invest in improving water quality to generate positive health and financial returns for households. A focus on improving quality and reducing losses, if accompanied by a public awareness campaign, would have direct effect on welfare. InLebanon, the burden on poor households comes from poor quality and low public water supply, rather than high expenditures. Water expenditures by.the low income households are in line with international norms but could be reduced if households relied less on alternative sources. Although compared to other problems in water provision, reducing expenditures for consumers is not the top priority, the costs which consumers bear in terms o f quality o f service-poor reliability and potential health risks-are significant. Among the sector's reform requirements, this may be a relatively inexpensive step to take. The benefits o f this type o f intermediate measures, while a metering program is rolled out, would accrue to all households, but many o f the poorest would experience the most immediate impact because of the constraints they face in sometimes costly alternative water sources. 103. A rapid rollout of metering is needed if wastage and equity issues are to be addressed. Both Regional Water Authorities and consumers lose when supply and demand are delinked by the current flat fee structure. Water companies cannot charge the marginal cost o f production, and consumers do not get what they pay for. The pilot metering programs show that metering can be introduced, but unless metering i s either region-wide (so that all households under a RWA use the same billing system) or metered households are able to pay by volume, the benefits o f metering will not materialize. 104. To increase revenues, RWAs will need to revisit household concerns. RWAs can increase revenues intwo ways, through improved bill collection and through tariff increases. Ineither case, RWAs will need to revise both the informal and the formal contractual agreements with households, raising tariffs and installing meters to link supply with demand, while also investing in improvements in quality. They will also need to strengthen collection enforcement mechanism. 105. Adopt a clear framework for reform sequencing by balancing costs and benefits between consumers and water companies. While the scale and complexity o f the improvements precludes them from occurring simultaneously, proper sequencing o f reform measures will be crucial to their successful implementation and acceptance by stakeholders. Reform will likely need to be based on negotiating a 69N=66 inthe SIA survey 44 compromise between water companies increasing supply and quality (representing a gain for households, but additional costs for the companies) and raising the tariff (a loss to households, a gain for the companies). On the supply side, institutional and financing constraints need to be taken into account. On the consumption side, household consumption patterns, access to alternatives, andwillingness to pay need to be factored into the equation. 106. Fivefundamental objectives are identified for improving water services. They are outlined in Table 3.1O., together with a breakdown o f their implications for both water companies and households. (Issues of implementation, costs, and financing are beyond the scope of this study.) In the long terms, achieving the objectives will ensure a net benefitto all stakeholders. Table 3.10. I:plications of key water reforms 1 r different stakeholders Objective Implications for RWAs Implications for households Improvingquality 1 Increase inoperationalcosts 1 Less needto purchase from for treatmentand alternative sources for monitoring. households concerned over 1 Increasedwater consumption quality; i.e. lower expenditures. 1 Improvement inwelfare as more budget available for other goods and services. 1 Improvement inhealth outcomes. Increasingsupply 1 Increase incosts for 1 Less needto purchasefrom investmentsininfrastructure, alternate sources for houseswith andto reduceUFW. inadequate supply; i.e. lower expenditures. 1 Improvement inwelfare as more budgetavailable for other goods and services. Adoption of a a Ability to charge for actual 1 Ability to alignpayments with meteringsystem water consumed. consumption, and control expendituresbasedon need. Increasingnetwork 1 Increaseincosts for 1 Will benefit poor households coverage infrastructure disproportionately, becausei) Increase inoperationallosses they are less likely to be for RWAs distributing water connected, and ii)they will have below marginal cost. aless exDensive oDtion. Raisingtariffs 1 Increasedrevenue, lower 1 Increasedcosts reliance on subsidies 1 Unless combinedwith other 1 Improved O&M, measures, decrease inwelfare as a Ability to increasewages less budget is available for other and hire and retain qualified goods and services. staff. 45 ANNEXES ANNEXI:METHODOLOGY I.SUMMARY OFSURVEYMETHODOLOGY 1. Primary data was collected for the study usingboth quantitative and qualitative researchmethods. A quantitative survey was conducted targeting 1804 households, representative at the Mohafaza (region) level. The sample was also selected to be geographically diverse (urban, rural, central, peripheral, densely populated, semi-deserted, etc.). The survey was conducted from April to June 2008 (and was only briefly disrupted by the May crisis). The results were analyzed to determine the consumption patterns and the willingness to pay o f households according to region, welfare category and other variables o f interest. The selected welfare indicator is expenditure per adult equivalent. 2. In addition to the quantitative survey, qualitative in-depth interviews were conducted with selected profiles (households, municipalities, generator owners, EDL collectors and buildingmanagers) in order to elicit more in-depth understanding o f the issues. These interviews were conducted inthree waves: before (lo), during (21) and following the quantitative survey (20). New questions were added for the final wave based on the quantitative analysis. 11.QUANTITATIVE SURVEY 3. The absence o f a sampling base in Lebanon necessitates the development o f a tailored for every study. Thus, for the purpose o fthis WorldBank study, the Consultation and Research Institutedesigned an approach that takes into account lack o f access to the latest data from official sources. i) A target sample size o f 1800 was established. The unito f observation was the household, with the Palestinian refugee camps excluded. Only primary dwellings were targeted. ii)The sample was stratifiedby Mohafaza(region) based on the householddistribution by Mohafaza published in 2004 by the Central Administration for Statistics (CAS). For instance, the Mohafaza of North Lebanon accountsfor 18.45%of households in Lebanon; therefore 332 of the 1800questionnaires were completed in North Lebanon. iii)Within each Mohafaza, the sample was stratified by Caza following the population distribution published by CAS in 1997.70For instance, the Caza o f Akkar accounts for 31.35% o f the residents o fNorth Lebanon; therefore 104 o f the 332 questionnaires were completed inAkkar. iv) At the Caza level, the sample was stratified by Circonscription Foncibre (CF) the smallest administrative unit in Lebanon. Sampling was based on the CAS 1997 population distribution. On the CF level, two steps were followed in order to obtain a good representation o f the population, both in densely populated and in the more `peripheral' areas. The proposed approach allowed the inclusion of peripheral areas in the sample while keeping transportation costs and commuting effort under control. Based on this estimation, the most densely populated CFs were first selected and the number o f questionnaires completed was determined based on their relative weights in terms o f population. In CFs where 3 questionnaires would have completed, the number was increased to 4 questionnaires. In a second step, the remaining questionnaires were distributed ~ ' O2004 survey is limitedonly to Mohafazat level 46 randomly among the remaining CFs (4 questionnaires per CF). Going back to the example o f Akkar, 40 questionnaires were allocated to the 8 most populated CFs (which corresponds to their relative weights). The 64 remaining questionnaires were allocated in groups o f 4 to 16 CFs, chosen randomly. v) This approach resulted in the selection o f a sample that was representative o f households in Lebanon. It ensured that the results were statistically representative at the Mohafaza level and allowed for a more advanced statistical analysis that specifically targeted electricity and water consumption profiles since the resulting sample i s geographically diverse (urban, rural, central, peripheral, densely populated, semi-deserted, etc.) vi) In villages, the questionnaires were allocated to different neighborhoods and specific maps were provided for cities. Within these pre-selected neighborhoods, the surveyor randomly chose the households (primary residences) that were interviewed. 4. The following tables illustrate the samplingmethodology LC-1997 Sample Mohafaza Distr. Pop Distr. Pop 1997 Beirut 403,337 10.07% 93,090 11.O6% 199 Mount Lebanon 1,507,559 37.64% 336,427 39.97% 719 Northern Lebanon 807,204 20.15% 147,088 17.48% 315 Bekaa 539,448 13.47% 106,843 12.69% 228 SouthernLebanon 472,105 11.79% 95,120 11.30% 203 Nabatieh 275,372 6.88% 63,109 7.50% 135 Lebanon 4,005,025 100.00% 841,677 100.00% 1800 *Living conditions survey MPS-2004 Sample Mohafaza Distr. Pop Distr. HH 2004 Beirut 389,661 10.38% 101,695 11.56% 208 Mount Lebanon 1,501,282 39.99% 371,289 42.20% 760 Northern Lebanon 768,709 20.48% 162,344 18.45% 332 Bekaa 471,137 12.55% 102,797 11.68% 210 SouthernLebanon 401,075 10.68% 89,423 10.16% 183 Nabatieh 221,920 5.91% 52,306 5.94% 107 Lebanon 3,753,785 100.00% 879,854 100.00% 1800 5. The sample distribution was made according on households based on the most recent household distribution. Because o f security concerns, the planned sample was slightlymodified. The deviations from the planned survey, underthe effective CF-sample column, are highlighted ingray. 47 6. Lebanon consists o f 1643 CFs. 193 CFs were excluded because they are completely deserted, leaving 1450 CFs. The last two columns present the distribution o f the sample as well as the number of selected CFs. Thus, a total of 180471 questionnaires were completed in 279 different CFs. The sampling rate is 11485 and around 20% o f the CFs are covered. Minieh-Danieh 53 South Total I 208 472,095 1 183 1 184 34 184 34 Grand Total I 1450 II 4,005,028I 1803 I 1814 279 1804 278 71 The total number o f planned questionnaires slightly exceeded 1800, due to the fact that a minimum o f 4 questionnaires were completed in every selected CF, while the fmal sample o f 1804 reflects security issues. The difference i s not significant. 48 111.FIELDLIMITATIONS 7. During the field survey, surveyors were faces with some difficulties. These difficulties were related to two reasons: (a) the security situation that arose during the first half o f May, and (b) the non- responses o f some households. 0 Delays due to security reasons andpolitical instability. The field survey was launched in April 28, 2008. An interruption took place between May 7 and May 19, mainly inBeirut and some areas o f Mount-Lebanon, then later on in Tripoli and its surrounding areas, due to security reasons, where it was not safe to go to the field duringthe period o f armed tensions in those areas. The field surveyors had to wait for a few days after the end o f the armed conflict inorder for the situation settle andthe residentsto have gone back to their normal daily lives. ii) Non-responses. Surveyors faced some problems related to the cooperation of some households. There were five main problems pertainingto this issue: a. Rich areas,: Response rates appeared to be lower inricher areas where the surveyors had sometimes to visit around 10 households before getting a positive response. It should be noted however that the quotas were respected and the households were selected withinthe determined clusters. b. Political instability and high tensions: People were sometime afraidto answer the questions due to security reasons and were hesitant inreceiving the surveyors. c. Electricity issue: This i s a sensitive issue ingeneral inthe country and thus some people were reluctant to go into details inanswering the questions d. Long Questionnaire: The questionnaire was considered too long by some respondents, so they sometimes did not want to continue answering the entire questionnaire. Questionnaires filled by these households were excluded from the sample. IV.EXPENDITUREDATA 8. The survey collected household expenditure data using recall methodology. The questionnaire identified 18 expenditure categories comparable to the 2004 household living conditions survey. These categories are: food, clothing and footwear, rent or mortgage, fixed line telephone, mobile telephone, internet fees, cable fee televisions, gas, house maintenance and cleaning supplies, housekeeping and nanny services, durable goods, healthcare and medicine, public transport, gas for automobile, repairs for automobile, automobile insurance and taxes, education, leisure and travel. The survey asked the household members responsible for family budget to recall spending on each category for the most convenient time period (daily, weekly, monthly or yearly). To improve recall outcomes, assistance was given to interviewed individuals in listing the various components o f each spending category especially when it came to recalling expenditure on durable goods, and to a lesser extent for education and leisure and travel. 9. Data collected was adjusted to reflect yearly household expenditure on a per capita basis. The total sample o f households that recalled all expenditure categories is 1,3 89 accounting for around 77% of the total sample surveyed. This is the sample size that was used to construct per capita expenditure quintiles which are used inthe analysis to rank households according to welfare status. It should be noted that control variables were used to check for the consistency and validity o f the responses obtained. These measures include comparing expenditure declared to questions on asset ownership especially when it comes to spending related to cars; and comparing expenditure on some services such as telephone and internet to current market prices for those services. Regional representativeness o f the survey was not affected as the 1,389 households in the sample maintained a similar distribution on a Mohafaza level as the complete sample o f 1804 households. 49 10. It should be noted that the 2008 SIA might be underestimating household expenditures. First, households tend to under-report spending when using recall, and second, fieldwork found that the highest rate o f non-responses or unwillingnessto participate inthe survey came from richhouseholds. 11. Due to time and budget limitations a more detailed expenditure survey based on expenditure diaries and price adjustments was not possible at this stage. However, comparisons to the Lebanon 2004 Household Conditions Survey, which used such methodology on around 7,000 households, revealed the validity o f the expenditure data collected by the 2008 SIA. Although validity with 2004 was assessedand found to be good, the expenditures are nonetheless not comparable at an absolute level since different methodologies were used (recall vs. expenditure diary). However, households distribution across quintiles was not affected as families in same quintiles had similar characteristics in both surveys. This was validated by looking at household ownership o f various assets in both surveys. 12. It should be noted that the study uses household quintiles rather that population quintiles. While ineither case the bottom quintile includes the poorest members o f society, usingthe household unit as a point o f comparisonmeans that quintiles will not represent equal shares o f the population. Because lower income households are larger (in Lebanon there are on average 5.6 household members in the bottom quintile against 4.4 top quintile) this means that the bottom household quintile comprises a larger share o the total population than the bottom population quintile would. To be precise, there are 1.02 million persons inthe bottom quintile compared with 0.79 million persons inthe top quintile. v.ELECTRICITYCONSUMPTION TARIFF SIMULATION MODEL Database .. Based on data collected from survey o f 1804 households inLebanon, May-June, 2008. Stratified random sample representative at the Circonscription FonciBre (the smallest administrative unit inLebanon) Uses data on electricity expenditures (LBP), consumption (kWh), and ampere levels from 754 households. Generates consumption estimates at representative level. Uses household expenditure data (generated by recall questions on 18 expenditure items) to disaggregate households by welfare category. Description. 13. The model disaggregates fixed and variable costs o f electricity bills. The quantity o f kWh consumed is derived from the price per kWh, the stamp fee, the rehabilitation fee, the ampere charge, and the tax (which applies to everything except the stamp fee). 14. From the household survey, data was collected on the total electricity bill (average peak, off- peak, and last bill paid), amperes and phase (mono or tri).Not all respondents could or would reveal the actual kWh consumed, based on the last month's bill. Thus, it was first necessary to determine how many kWhthe household consumed. The total expenditures and amperes are entered; and quintile andregional category averages are calculated. 15. Two models were used. Model 1 does not include an estimator for elasticity, but allows changing the levels o f individual tariff tranches, as well as other parameters. Model 2 includes an estimator for 72There are 1450 inhabitedCirconscriptionFonciBre inLebanon. 193 are uninhabitedandwere excludedfrom the sample. 50 elasticity based on analysis from the Energy Sector PER. However, it does not allow for changing individual tariff levels; only across the board tariff changes are possible, i.e. all tranches are changed by the same amount. Simulations 16. The model allows: Calculation of household electricity consumption based on expenditure and ampere data Calculation of what tariff tranche household x i s consuming. Calculation of mean tariff (LBP/kWh) paid by household x. Calculation of meantariffpaid by all households inLebanon. Translation of willingness to pay (WTP) from responses in LBPs relative to their total bill into percentage tariff increases. Estimates of how changes in tariffs, either overall or per tranche, impact households in terms of electricity expenditures, for eachwelfare group. 51 ANNEX11: THEPRIVATEELECTRICITYGENERATIONBUSINESS 1. Private electricity generator businesses have been around since the 1990s and occupy a well- established corner o f the energy market. Inthe absence o f reliable electricity, they play an essential role in Lebanon by filling the gap in service provided by EdL. They are far more expensive than public electricity but many individuals and businesses could not manage without them. 2. Most generator businesses employ two to four persons, operate generators from 100 KVA to 350 KVA and more (which cost from US$15,000 to $60,000). They sell electricity back-up service to both businesses and households, with the latter predominating. As with EdL, they cannot always distinguish between business and residential clients since many clients run businesses from their homes. Private generator businesses typically have several hundred subscribers, and may cover an entire village or town. The owner may have other businesses as well. 3. A potential subscriber contacts the generator service inhis or her area and enters a contract for a fixed amount of amperes per month, typically 5 or 10 kVa. A collector comes by to collect the payment. The generator owner installs boxes in subscribing buildings and equips them with breakers. Although these businesses are informal, and not covered by any legislation (there i s no law in Lebanon which either allows or prohibits private generator businesses), they fillan important and recognizedneed inthe market and thus are tolerated. 4. In some cases the private generator businesses pay a modest monthly fee (e.g. US$lOO) to the municipality, as in the case o f Shiah. Several years ago in Saida a proposal was put forward to tax generators, taking a small percentage o f their profits, but when they protested the idea was dropped. In Tripoli a proposal to charge generator owners US$lper kVa was also not followed through. These examples suggest that private generators not only operate outside the legal framework, but have been generally successful at thwarting local government attempts at imposing fairly modest taxes. The generators owners interviewed for this study reported revenues ranging from L B P 10 million to 30 million (USD6,OOO to 20,000) per month although all complained that their profit margins had eroded with the increase in oil prices (during spring and summer 2008) and subsequent drop off in subscriptions as they had to raise the monthly price to consumers. 5. Costs include fuel oil, maintenance, oil, filters, replacement o f cables and breakers, and rental space. In Table 1 information on the business models o f two generators, from Sfeir neighborhood o f Beirut and from Tripoli i s provided for illustrative purposes. 6. According to a generator owner in Sfeir, prices have stabilized: at 65,000 LBPs for every 5 amperes. Two years ago the price was 40,000 LBPs for 5 amperes and the price o f fuel oil was 15,000 LBPsand blackouts only lasted4 hours. Now, blackouts last a minimumo f 8 hours inadditionto the fuel oil price increase. For the last two months, [inthe springo f 20081 the Ministry [of Energy and Water] has been raising prices every week. As a result, "generators are turning into a free service instead o f a profitable business." 7. The generator business has monopolistic features. Although there are a large number o f generator businesses (for example, over 150 in Tripoli alone) there i s virtually no competition on price. (In whatever manner a private generator secures rights over an area, it is not through an open tender.) Thus, households cannot choose their subscriber, but must use the one operating intheir neighborhood or town. The choices available to a household are straightforward - it can choose to subscribe or not, or it can change its ampere level. Reducing amperes from 10 to 5 kVa, for example, cuts the bill in half. Some households even cut their ampere level to 2.5 kVa. 52 Sfeir Tripoli Capital costs Generator 20,000 (300 KVA) 60,000 (350 KVA) Network and cables 15,000 Rentofa store anda 300 5,000 Source: key informant interviews withprivate generator owners. Note that information is incomplete since informants did not alwaysprovide answers to all questions 8. Counter-intuitively, improvement in public electricity availability appears to benefit, not hurt, private generators, at least in the short-term. These businesses do not see themselves as EdL competitors and would welcome less rationing. A decrease in blackout hours would save them fuel costs but they do not expect to see their business hurt. Inthe spring o f 2008, many generator businesses were not providing coverage for the full blackout period. If rationing is reduced, they would be able to cover more o f the (shorter) blackout period, thus improving their function in addressing public sector deficiencies. Furthermore, in the event that generator businesses could reduce the hours they need to operate, they could lower their prices and attract more customers. 9. In any case, generator businesses have low expectations o f reform. As one skeptical Kitirmaya respondent said, "There won't be reform inthis country, where it's every man for himself." The fact that private generator businesses don't feel threatened by a reform that would, in theory, result in shorter blackout periods underlines the fact that although they are responding to unmet demand, the relationship between publicly and privately supplied electricity i s not one o f perfect substitution. In other words, a gain for the public provider i s not equivalent to a loss for the private provider. Private generators do not expect EdL to provide 24 hour service inthe short- to medium-term and thus see demand for their service persisting. Customers are expected continue usingtheir services, at least as a back-up, for many years to come. 10. Generator business owners expressed concern about high fuel prices at the time o f the interviews, when crude oil was trading internationally at almost US$l50 in early 2008, squeezing their profit margins. As generator businesses increased their monthly subscription rates to cover rising costs, many consumers either lowered their ampere subscription to save money, or stopped using generators 53 altogether. On the other hand, as prices come down from their earlier highs, both the public utility provider and the substitute provider stand to gain. 11. The relationship between generator owners and their customers i s not without its tensions. It was hypothesized that because private generators are strictly for-profit businesses, unlike EdLwhich i s a state entity providing a common good, households would have fewer complaints: there tend to be fewer illusions in a transaction based on mutual self-interest. However, although consumers clearly exhibit a higher willingness to pay for privately generated electricity than EdL (indicated by the much higher monthly costs for fewer hours o f electricity and lower ampere levels), this does not translate directly into customer satisfaction. Very few households were completely satisfied with the service they received from the generator owner, and this was mainly because they do not cover the entire blackout period. One rural household complained that the generator he subscribes to rations electricity just like EdL and turns the generator o f f at 10 or 11 at night. A Saida household said that the generator owner exploits people by charging a stiff rate and then decreasing the hours o f service. Another head o f household made the point that he had to cooperate with the generator owner because he had no choice. Even boycotts are contemplated. A woman from a rural area notedthat she has an agreement with other subscribers that they would stop their subscriptions ifprices went up. 12. Generator owners complain about their customers as well. Some customers don't pay regularly. Some pay in installments and others are late paying their subscriptions. There i s also a tendency (among about 10% o f customers) to decrease the subscription size to even 2.5 amperes for 30,000 LBPs. Inearly 2008 the number dropping their subscriptions was increasing month by month. 13. The relationship between generator owners and EdL can be described as one o f mutual tolerance. Unable to provide constant electricity service, EdL must accept that the gap will be filled by another actor. The generators, on the other hand, need EdL to provide the bulk o f electricity because they would be unable to provide full, 24-hour service, except at prohibitive prices that few households would be willing to pay. Typically, cooperation with EdL is limited to the private generator's use o f EdL's electricity poles. The generator owner may even be the first to call EdL to notify the company when an electricity pole i s down - as this disruption affects his business as well. On the other hand, until EdL is reformed, it can be said that private generators are providing a benefit to EdL as well. By filling a huge unmet demand, the pressure on the public company to provide more hours (and incur even greater losses) i s eased. 54 ANNEX111: WILLINGNESS TO PAY-BACKGROUND I.WILLINGNESSTOPAYVARIABLES The SIA survey included three different variables for willingness to pay (WTP) each o f which can be usedto estimate demand for improved services 1. Revealed WTP 0 For electricity - reported expenditures on private generators (Do you subscribeluse a generator and how much i s your monthly payment?) For water - payments for non-network water (What are your expenditures on bottled water and tank water?) 2. Stated WTP closed-ended Direct questioning method - dichotomous choice (would you pay x more if service were reliable? yeslno) 3. Stated WTP open-ended 0 Direct questioning method - maximum amount (What i s the maximum you would pay if service were reliable and constant?) All three variables for WTP are referred to inthe report. 11.HYPOTHESES 1. Stated WTP i s positively correlated with household income/expenditures and with blackout hours (i.e. higher the number o f blackout hours with higher WTP levels). 2. Stated WTP is positively correlated with revealed WTP. (Households that use private generators will be more willing to pay for 24 hour public electricity service, because they have already demonstratedthat they value uninterrupted electricity service). 3. Differences between stated and revealed WTP can be explained by: Different expectations o f private and public service providers, i.e. consumers factor in a profit for private providers or, conversely, expect private providersto be more efficient; Some households in a multi-unitbuilding pay for a private generator as part o f a package o f utilities and cannot opt out without moving to a different building; and their revealed WTP is therefore inflated. 111.LIMITATIONS POTENTIALWEAKNESSES-VALIDITY AND As always, with open-ended questions, potential validity problems arise. The dichotomous question was included to reduce the likelihood o f respondents answering strategically. At the same time, the closed- ended question, based as it i s on the current electricity bill, may not provide `true' WTP since respondents are using current payments as an `anchor point.' A rebuttal to this problem would be that although the answer may be anchored, it also presents a real world scenario, since households will, necessarily, relate (and react to) any future tariff increases based on their current payments. Interestingly, for the pilot-test, several respondents answered that, under the `ideal' scenario, they would actually be willing to pay less than they were paying currently. This could be interpreted to meanthat they felt that their current bill was too highalready, for the service they received. 55 Iv.OPTIONS CONSIDERED AND REJECTED For electricity, we considered asking WTP for an increase o f x hours o f electricity (compared with current availability) as to full 24 hour coverage. The rationale for this proposed formulation was that the scenario would be more `realistic' given the unlikelihood o f attaining reliable, 24 service in Lebanon for many years. The problem with this i s that the length o f electricity service on a 24 hour basis varies widely between regions. This means that for some households an additional 6 hours could mean uninterrupted service, while for others it might meanjust 18 hours o f service, and potentially continued reliance (for some) on generators. The answer would be ambiguous - some households would still have to/want to rely on generators - and comparisons difficult. As it is, comparing WTP between households with different hours o f service will mean controlling for this factor - we expect lower WTP where hours o f service are longer on average. v.SUBSTITUTABILITY BETWEENPUBLIC AND PRIVATE ELECTRICITY Interms of analyzing WTP, how much does a household's private generator subscription ampere level matter? Lower ampere levels mean fewer appliances can be run, and the generator-supplied electricity i s therefore a less perfect substitute. We believe it should not matter, because lower ampere levels are reflected in lower private generator subscription fees. However, we may be missing out on some interesting information. The qualitative research indicates that many households are either reducing their ampere levels (typically from 10 to 5 amps) or have stopped using generator services altogether. (and rising generator fuel costs are leading some generator businesses to consider exiting the business.) This information provides an additional indicator o fthe threshold at which households are no longer WTP. VI. THEUNEVENRELATIONSHIPBETWEENPUBLICLY AND PRIVATELY SUPPLIEDELECTRICITY The relationship between publicly supplied electricity and generators i s not straightforward. Fewer public service hours do not translate directly into more generator service hours. This i s partly due to increasing fuel prices, which appear to have driven a wedge between the one-to-one substitution relationship. Most generator owners and households interviewed reported that generators do not run the entire period o f the blackout. For example, households in areas which receive 16 hours o f public electricity per day, and choose to subscribe to a private generator, will typically receive less than 8 hours o f generator service. This may or may not be stated inthe contract. Generator businesses report that they benefit (inthe short- term at least) when the length o f service interruptions decreases, because it means they can run their generator for fewer hours, reducing their fuel costs. It i s also apparent that as the price o f fuel rises, some households are discontinuing their private generator subscriptions. There are several implications. First, privately generated electricity i s not a perfect substitute; private generated electricity will not fill the gap in an environment o f rising fuel costs -which means even better o f f households will choose to forgo electricity and some private generator businesses will exit the market. Interms o f analysis, however, subscription prices should still be a valid input in determining the cost o f blackouts. This is because revealed WTP can be calculated for electricity at various prices. A household willing to pay x for 5 amps o f privately generated electricity but not x + y is signaling their value for one blackout hour. WI. SOURCEOFELECTRICITYAND EXPECTATIONS It should be noted that WTP is likely to be colored by attitudes toward the electricity provider. (Behavioral economists have identified expectations o f fairness as a factor in economic decision-making.) Qualitative research reveals that customers expect generator businesses to operate on a for-profit basis and EDL to provide a public service. Although customers are not always happy with generator service, 56 complaining that they do not get as many hours o f electricity as they believe they should, they also accept that as businesses the generators will look out for themselves first, and are furthermore squeezed by high fuel prices. Butwith regardto EDL, customers expect decent service as a basic right. Inconclusion, WTP i s likely to be colored by expectations toward the service provider, with the implication that WTP to EDL may be lower than to a private provider. VIII.ANALYZINGREVEALEDWTP Definition: households that use private generators are considered to be `willing to pay' and those who do not `unwillingto pay.' Assumption: all households have the option of usingprivate generators; no households that have zero expenditures on private generators reflect lack o f choice. (We may need to investigate this assumption -andifwecanidentifythesehouseholds, excludethem.) Demandfunction: y-axis - expenditures on private generator as % o f EDLbill x-axis - percentage o fthe sample Correlations:income level, number o f blackout hours Note: For all approaches, without controlling for income and hours o f blackouts we do not expect to see a downward sloping demand curve E.ANALYZINGSTATEDWTP-CLOSED-ENDED Definition: ifanswered `yes' household is WTP, ifanswered `no' they are not Assumption: for those answering `yes' reported WTP reflects minimum WTP under hypothetical scenario Sample characteristics: sample was divided into four parts, each with a different amount o f increase (25%, 75%, 100%' and 150% for electricity) and each household was given a simple yesfno option Demandfunction: y-axis - expenditures on private generator as % o f EDLbill x-axis - percentage o fthe sample answering `yes' under each increase amount Correlations:income level, number o f blackout hours Limitation: not capturing all o f consumer surplus since some households will be WTP more than indicated by their `yes' response x.ANALYZING STATEDWTP-OPEN-ENDED Definition: maximum amount reflects WTP Assumption: reported WTP reflects full WTP under hypothetical scenario Sample characteristics: this question followed the closed-ended question, for both `yes' and `no' responders. Demandfunction: y-axis expenditures on private generator as % o f EDLbill - x-axis -percentage o fthe sample Correlations:income level, number o f blackout hours Limitation: respondents may be anchoring their `maximum' response to the proposed threshold. 57 XI. TRANSLATING WTP INTOPOLICY IMPLICATIONS Care must be taken intranslating WTP results into policy recommendations for tariff increases. To take a simple example, if a household receives 12 hours o f electricity from EDL at $20/month, it would naturally expect to pay EDL more if it received 24 hours, since it would be using more kWh. The first question is, how much more, compared to its current expenditure? The amount would depend on two things-the variable cost (depending on kWh consumed) and the intensity o f usage duringthe additional 12 hours. (Electricity bills include fixed fees for maintenance, tax, etc. and variable fees for usage.) To start, we can make the simple assumption that, for an additional 12hours, the variable cost would double, under 24 hour service will be $10 fixed + $20 variable = $30. i.e. be equal to the current variable cost. Ifthe original bill i s $10 fixed + $10 variable = $20, the new bill Next, the presumed intensity o f usage during the blackout hours needs to be estimated. If any o f those blackout hours are duringthe night, or when no one is home or duringdaylight hours when lighting i s not needed, the intensity will be less than 100%. (The estimation is beyond the scope o f this note.) Thus, for any level o fblackout hours, the expected change inpayment would be as follows: hours, and P i s the coefficient for intensity o f use. Where PEis the expected payment for full, 24 service, PFis the fix fee, PVis the variable fee, TBi s time in The second question i s - should we adjust the stated WTP o f respondents by the amount that they would be paying ifthey had 24 hour service? Ifthe answer i s yes, their WTP could be changed as follows: Equation 2: WTPR - PE = WTPA Where WTPR i s reported WTP and WTPAi s adjusted WTP. In conclusion, simply because a household states they are willing to pay an additional 50% o f their current bill for improved service, it does not mean that tariffs can be raised by 50%. The above adjustment should be made. Note: This discussion has not included looking at the impacts o f changing tariff tranche levels, a separate topic o fanalysis. 58 ANNEXIv: LESSONSLEARNED ENERGY FROM SECTOR REFORMS'~ BACKGROUND About 70 of the 150 developingcountries and transition economies have embarked on reforming their power markets since the early 1990s. The drivers o f this reform movement are disenchanted with the poor performance o f state-owned power utilities, the need for new investments and modernizationto meet rapid growth in demand, and fiscal pressure, along with the desire to protect and help the poor. The reforms have generally been tentative and incomplete, however, particularly in relation to market structure, degree o f private participation, and development o f the regulatory framework. The countries that have embarked on power market reform cover a broad range in physical, economic, and institutional terms. The most advanced countries inreform are located in Latin America and inEastern Europe, where they also have relatively larger power systems and higher levels o f per capita national income compared with other developing countries and transition economies (referred to here collectively as "developing countries"). Strategic elements of power market reform in developing countries should be assessed against three outcomes that reflect the drivers of reform. These outcomes are: 0 better service quality for electricity consumers to support economic growth and welfare, improvement ingovernment's fiscal position, and more affordable access to electricity for the poor. They reflect the main drivers o f reform. The main elements o f reform-restructuring power utilities and markets, regulation, competition, and the roles o f public and private participants-are the means for achieving these outcomes. Inorder to show how implementationaffects design, this section bringstogether the design lessons under the following four strategic elements for implementingreforms: Power market reform: 1. Has many dimensions. 2. Must be adapted to starting conditions. 3. I s a process-not an event 4. I s an opportunity to help the poor. ELEMENT POWERMARKET REFORMH A SMANYDIMENSIONS 1: Many dimensions of power market reform are important in developing countries. Under mounting experience, power market reform in developing countries has increasingly emphasized the social, legal, and political dimensions o freform indefining the techno-economic dimension. Changeto commercially orientedgovernance isfundamental to achieving sustainable reform of power markets. Power market reform in a broad sense can be viewed as a means to improve governance o f the power market and its participants. The traditional model o f governance under ministry departmental organization i s not sustainable in most developing countries. Commercially oriented governance 73 This Note i s based on the World Bank publication Reforming Power Markets in Developing Countries: "What Have We Learned?' Energy and Mining Sector BoardDiscussion Paper No. 19, September 2006 59 irreversibly removes the management and development o f power supply from political and bureaucratic control to achieve commercial standards in management practices, financial performance, and the pricing of products and services. Social and political factors are importantfor all power market reform programs. Government must generate public acceptance and stakeholder consensus for these programs. Power market reform based on market restructuring and private sector participation involves complex social and political issues for market investors, utility employees, and electricity consumers. The complexity o f these issues can sometimes match the complexity o fthe technical issues involved inreforming power markets. Distributional issues are often at the heart of designingpower reformprograms. Reforms must not only offer benefits that substantially outweigh the costs, but also provide the means for compensating losers. Although reforms in power markets have delivered substantial benefits to society overall through efficiency gains, most of these benefits have been shared by power producers, service providers, and commercial businesses, but have not reached other segments o f society, including the poor. The impact of power market reform on the poor is a critical distributional issue. The poor have obtained a low share of the benefits o f power market reform in developing countries, and some have even suffered welfare losses. Some o f the poor who have lost from reform were obtaining some electricity service before reform-albeit illegally and o f poor quality- but were disconnected or now have to pay for their consumption. Other groups o f the poor continued to receive legal service, but at higher tariffs as subsidies and cross-subsidies were removed under the commercial pressure on service providers introduced by reform. On the other hand, some poor have gained from reform by receiving otherwise unavailable connections to electricity supply. ELEMENT2: POWERMARKETREFORMMUST ADAPTEDTO STARTING CONDITIONS BE Starting conditions in the power market are importantfor designingpower reform programs. These conditions include the size o f the country and its power system and market, the country's location relative to other power markets, its income level and macroeconomic condition, its political situation, and the capacity o f its domestic financial markets and institutions. The variety of market structures is one indicator of the range of reforms in power markets. From the pre-reform structure o f a monopoly, market structures can be categorized according to the increasing degree of competition, starting from a purchasing agency- also known as a single buyer-through whom all or most trade in wholesale passes and who therefore manages competition for market share among generators and independent power producers. Indeveloping countries the competitive structures are based on trading arrangements inthe wholesale power market that allow distribution companies and large users o f electricity to purchase electricity directly from generators either ina power exchange or bilaterally. The economic casefor breaking up a vertically integratedpower utility rests on variousfactors. The gains from breaking up (or "unbundling") the utility by separating the generation component from the distribution component are worthwhile when they exceed the costs o f transactions among the separated segments introduced by unbundling.The relevant factors are power system size and country institutional capacity to manage complex trading mechanisms. The case for unbundling i s strongest in large power systems in countries well endowed institutionally. The case for unbundling i s weakest in small systems in countries with undeveloped institutional capacity and.weak economic conditions. The numerous countries whose power systems are too small for a competitivepower market have intermediate reform options. Unbundlingthe generation and distribution segments o f the power supply chain into tiny entities would not make sense in these systems, because economies o f scale and scope 60 would be lost without realizing the benefits o f competition. Even in small power systems, however, separation o f these components helps regulation o f power service providers by revealing information about their costs, and increasing the transparency o f price setting. These systems can adopt a purchasing agency or single buyer model untilthey can reap the benefits from greater separation o fthe supply chain. The variety of waysfor theprivate sector toparticipate in the supply and delivery of electricity services is another indicator of the range of reforms. The role o f private participants should match their capacity to take on investment risks under specific country conditions. Their roles can range from virtually no at- risk investment under management contracts through some investment risk under long-term concessions to accepting all investment risks under divestiture o f ownership to the private sector. Problems-even failures-as well successes, have been associated with these forms o f private participation in power markets. The casefor bringing the private sector into power supply functions rests on how well this would achieve the desired reform outcomes under the prevailing operating conditions. Latin American experience shows that privatization o f power market assets can improve services at reduced costs and with fiscal benefits, provided that stable macroeconomic conditions prevail. However, many developing countries do not offer the necessary conditions for attracting substantial amounts o f private investment in this way to their power markets. Many o f them have attracted substantial investments by independent power producers, but only by giving contractual protection against most noncommercial risks to these producers. Thepublic sector will remain an important source, and often for the medium term the main source, of investment for a power market where country and market risks deter private investors. In many countries, some public investment will be needed to rehabilitate nonviable generation and distribution businesses as a prerequisite for attracting private investment in them or during the early years o f concessions for distribution businesses. The public sector can play a financing or risk-bearing role by means o f investment financing and the provision o f subsidies and guarantees under public-private partnerships through management contracts, leases, and concessions. The range of approaches to establishing the credibility of power market regulation is an indicator of the range of reforms topower markets. Credibility o f regulation is needed to attract long-term private at- risk investment inelectricity services. Itcovers autonomy to carry out duties, transparency inprocedures and processes, and accountability to government and consumers. A means for developing credibility is a designated regulatory agency or separate governmental department that discharges its duties in a neutral and depoliticized manner. Specific contractual arrangements may be needed to provide stability and credibility for private investors. Private investors place importance on the stability and enforceability o f laws and contracts, and they contend that a credible regulatory system (including contracts) requires more than a newly formed regulatory entity ("regulation by contract"). Empirical analysisindicates that a clear thresholdexists among developingcountries in relation to size and incomefor the composition of power market reform. This threshold is formed by a combination o f system size larger than 1,000 MW and national per capita income above US$900. A large middle-income group o f countries i s formed by a combination o f size and income above these threshold values, and a small low-income group i s formed by a combination o f size and income below these threshold values. Country income level has a relatively stronger influence than power system size on the roles o f the public and private sectors and on access and affordability to electricity services. It also has a stronger influence on the regulation o f power markets on the basis that institutional capacity increases with income level. Power system size has a relatively stronger influence on market structure. 61 ELEMENT POWER MARKETREFORMI SA PROCESS-NOTANEVENT 3: Pressurefor rapid results should not obscure the point that reforming power markets is a long-term process that requires patience to achieve the desired outcomes. This is because such outcomes as improving service quality for electricity consumers, strengthening the government's fiscal position, and providing affordable access to electricity for the poor take time to accomplish. This situation applies especially to countries starting with weak governance structures for power utilities and poor investment climates. Power market reforms in developing countries are generally tentative and incomplete, and are still works in progress. To date, most reform programs have reached interim positions-such as the single buyer model o f energy trade-and still need to find ways to attract private investment sustainably and develop their regulatory capacity. The initial transition stage is critical to the success of power market reform and the most vulnerable period for derailment of the reform process by many developing countries. For market structure, transition concerns the separation o f the industry structure into its main components and the adoption of a single buyer trader for wholesale power. For private sector participation, transition focuses on private sector roles that fall short o f full risk taking, such as management contracts and other forms o f private participation, with temporary risk mitigation mechanisms, such as by setting limits on the amount of financial risk initially faced by private operators o f power distribution and generation facilities. Transition arrangements to provide stability and credibility for a new regulatory regime revolve around regulation by contract under which regulatory rules and procedures are incorporated into concession agreements. Sequencingof power market reform shouldfollow a sound strategy: The legal and regulatory framework necessary for creating the new market structure and trading arrangements i s put in place before privatizing power supply entities and setting up new market trading arrangements. Restructuring o f power markets progresses from an integrated structure to partially unbundled structures. 0 Restructuring o f wholesale power trading arrangements progresses from only internal transactions within an integrated power utility to the entry o f IPPs selling their output to a single buyer, then to opening access to power networks by large users o f power, and eventually to bilateral trading between generators and distributors or to a central power pool under competitive trading. 0 Major organizational and financial restructuring precede the creation o f private ownership rights to avoid problems with stranded costs. Some countries have skipped the early stages of thesesequences, and othersmay do so in thefuture. A sequenced process, however, i s less risky and more sustainable than a single-staged ("big bang") process for reforming power markets mthe conditions o f developing countries. Sequencing of power market reform also raises tactical issues. Tactics should be specifically designed to address problematical issues, such as the following. 0 Whether to increase tariffs before or after investments to improve the quality o f service. 0 Whether to try improving the commercial performance o f loss-making utilities and distribution entities before bringinginprivate participation or with private participation. 0 The sequencing o fprivatizationinone or more rounds o ftransactions. 0 Whether to start the privatization sequence for poorly performing power markets with distribution entities before privatizing generation entities. 62 Whether to give investments in new generating capacity lower priority than investments in distribution, especially in a situation o f bulk power shortages. Reform benefits take longer than expected. Consumers usually expect better services from private companies than from state-owned enterprises. Consumers understandably lose patience and blame the regulators if tariffs go up immediately but service improvements lag behind. Therefore, it is not surprising that most regulators, when faced with this situation, will try to find ways not to raise tariffs. The preservation o f protective features, such as "life-line" rates, may be necessary, even if they mean continuing subsidies within income classes, as well as from industryto residentialconsumers. ELEMENT POWERMARKETREFORMI SAN OPPORTUNITY TO HELP POOR 4: THE Extending access to affordable modern energy services-including electricity services-for poor households is one of the most practicable ways of improving their welfare. This is because expanding access to these services helps to increase household incomes and meet basic needs, such as improved health and primary education, as well as support social empowerment and environmental sustainability. The cost o f these services to users is often considerably lower than the corresponding traditional energy alternatives used by poor households without access to these services. Reform provides an opportunity to rectify the policy and regulatory constraints on electricity access and service for low-income households. Reform can overcome entrenched attitudes to providing electricity services and introduce different kinds o f electricity services better suited to the poor. Opening up the main power market to new entrants can stimulate incentives specifically designed to attract new entrants into markets serving poor areas. The establishment o f a new regulatory system for the main power market provides an opportunity to introduce regulations that help the poor. Reforms that place the power market on a sound commercial footing, however, will not automatically improve access and affordability o f electricity services to low-income households. They may make little difference to this situation, or even worsen it. It i s important to ensure that reform does not adversely impact access and affordability. Access and affordable consumption of electricity by poor households can be promoted by various policy instruments. Instruments that promote access require service providers to extend access, reduce connection costs, and increase supply options. Extending electricity service to urban low-income households requires improvement to the existing power system. Instruments that promote affordability protect low income households from general increases in tariffs and costs o f service and facilitate payment o f bills. They stimulate services through nonstandard service delivery mechanisms, service types, and tariff and payment mechanisms appropriate to low-income households. Even under successful power market reform, poor households need help withfinancing the costs of connecting their premises to the network and installing meters at the points of consumption. Well- designed subsidies provide good incentives to service providers- both specifically for serving low- income areas, as well as generally-to attract private sector participation through concessions and asset sales. The substantial empirical evidence, however, questions the effectiveness o f many existing subsidy schemes as a means o f helping low-income electricity consumers. A number o f approaches (mainly input- based) have been developed to improve the targeting and cost-effectiveness o f subsidy delivery. 63 ANNEXV: REFERENCES Chubu Electric Power Company Ltd (CEPCO), Consulting Report in World Bank (2004), Republic of Lebanon HydrocarbonStrategy Study,ReportNo. 29579-LE, Washington DC, June 30. Central Administration of Statistics (2004), HouseholdLiving Conditions Survey, Beirut. United Nations (2008), Millennium Development Goal Indicators, available online at: httu://mdas.un.or~/unsd~md~~Metadata.asux?IndicatorId=O&Se~esId=7 10. United Nation Development Program (2007), Poverty Growth and Inequality in Lebanon, Executive Summary, October. World Bank (2008), Republic of Lebanon Electricity Sector Public Expenditure Review, Report No. 41421-LB, Washington DC, January 31. World Bank (2007), Lebanon Reform Implementation Development Policy Loan, Washington DC, August 2. World Bank (2007), Lebanon Economic and Social Impact Assessment, Chapter 2 -Electricity, Washington DC, January 22. World Bank (2007), Lebanon Economic and Social Impact Assessment, Chapter 3 -Water Supply and Sanitation, Washington DC, January 22. World Bank (2003), Republic of Lebanon Policy Note on Irrigation Sector Sustainability, Policy Note No. 28766, Washington DC, November. World Bank (2002), Sourcebookfor Poverty Reduction Strategies: Core Techniques and Crosscutting Issues,WashingtonDC. World Bank (forthcoming), Republic of Lebanon Water Sector Public Expenditure Review, Washington DC. World Bank (2008), WorldDevelopmentIndicators. World Bank (2006), Reforming Power Markets in Developing Countries: What Have We Learned? Energy and MiningSector Board DiscussionPaperNo. 19, September 2006. 64 IBRD 33433 35°30'E 36°00'E 36°30'E To To To Hims Al Hamidiyah Tall Kalakh Kabir LEBANON Qoubauyat Halba 35°30'N Funaydiq To Tripoli Abou Moussa Al Qusayr Zhgarta Hermel N O R T H Chekka L E B A N O N Sib'il Sib'il Qaa Amioun Ihdin Jaoz Bcharri . s Orontes Douma Qurnat n as Sawda'' Sawda t Al Labwah (3,088 m) Aynata Mediterranean y M t n s . M e l Sea Al Aqurah l To Jbeil n a o V Ibrahim Afqa n a a A n t i - L e b a n o n An Nabk a 34°00'N Jounie b k Baalbeck e 34°00'N e B M O U N T L B E I R U T L E B A N O N B E K A A BEIRUT Antilyas Rayak Baabda Hammana Zahleh To Az Zabadani Damour Litani Beit ed Dine Jieh To Joub Jannine Ad Dimas Saida Awwali Qaraaoun Jezzine El Zahrani 34°30'N Rachaiya 34°30'N Hasbani 0 10 20 30 Kilometers Nabatiyeh 0 10 20 Miles Maryjayoun N A B A T I Y E H Tyre Litani S O U T H LEBANON To Baniyas L E B A N O N To To SELECTED CITIES AND TOWNS Qiryat Tibnine Hula Shemona GOVERNORATE (MOHAFAZAT) CAPITALS Naqoura NATIONAL CAPITAL Bent Jbail This map was produced by the Map Design Unit of The Rmaich To World Bank. The boundaries, RIVERS Gadot colors, denominations and any other information shown MAIN ROADS To on this map do not imply, on Nahariyya the part of The World Bank RAILROADS Group, any judgment on the 33°00' N legal status of any territory, or any endorsement or GOVERNORATE (MOHAFAZAT) BOUNDARIES a c c e p t a n c e o f s u c h boundaries. INTERNATIONAL BOUNDARIES 35°30'E 36°00'E NOVEMBER 2004