Report No: AUS0001235 Poverty and Distributional Analysis of Electricity Poverty and Protection of Vulnerable Customers in Kosovo Poverty and Equity, Social Protection and Labor, and Energy and Extractives Global Practices Europe and Central Asia Region November 2019 1 This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. The financial and technical support by the Energy Sector Management Assistance Program (ESMAP) is gratefully acknowledged. ESMAP―a global knowledge and technical assistance program administered by the World Bank―assists low- and middle-income countries to increase their know-how and institutional capacity to achieve environmentally sustainable energy solutions for poverty reduction and economic growth. 2 Acknowledgements This work was prepared by a multi-sectoral team from the Poverty and Equity Global Practice (GP), Social Protection and Labor GP, and Energy and Extractives GP at the World Bank Group. The lead authors are Monica Robayo-Abril (Economist, Poverty and Equity GP), Stefanie Brodmann (Senior Economist, Social Protection and Labor GP), and Rhedon Begolli (Senior Energy Specialist, Energy and Extractives GP) and the team includes Boryana Gotcheva, Florentin Philipp Kerschbaumer, and Jayne Yoo. The report benefitted from comments from peer reviewers Sandu Cojocaru and Nithin Umapathi, as well as Ereblina Elezaj, Siddarth Hari and Jamele Rigolini. The report was prepared under the guidance of Linda Van Gelder (Country Director, Western Balkans), Marco Mantovanelli (Kosovo Country Manager), Oscar Calvo (Practice Manager, Europe and Central Asia, Poverty and Equity GP), Cem Mete (Practice Manager, Europe and Central Asia, Social Protection and Labor GP) and Sameer Shukla (Practice Manager, Europe and Central Asia, Energy and Extractives GP). The team would like to thank the working group on the protection of energy vulnerable customers, led by Mentor Morina (Director of the Department for Social Policies and Family, Ministry of Labor and Social Welfare), for excellent collaboration and valuable inputs to the report. 3 Acronyms and Abbreviations CNG Compressed Natural Gas ECA Europe and Central Asia ERO Energy Regulatory Office EPOC Environmental Policy Committee ESMAP Energy Sector Management Assistance Program EU European Union Eurostat The Statistical Office of the European Union FPL Food Poverty Line GWh Gigawatt hour GoK Government of Kosovo HBS Household Budget Survey HH Household HPP Hydropower Plants IPA Instrument of Pre-Accession Assistance KEDS Kosovo Energy Distribution Services KEK Kosovo Energy Cooperation KESCO Kosovo Electricity Supply Company KLA Kosovo Liberation Army KWh Kilowatt hour LIHC Low Income, High Cost LPL Lower Poverty Line MED Ministry of Economic Development MLSW Ministry of Labor and Social Welfare MW Megawatt PM Particulate Matter PMT Proxy Means Test PPT Purchasing Power Parity PSM Propensity Score Matching Method PV Solar Photovoltaic SA Social Assistance SAS Social Assistance Scheme SDG Sustainable Development Goal SILC Statistics on Income and Living Conditions Survey TOD Time-of-day UPL Upper Poverty Line WHO World Health Organization VAT Value Added Tax 4 Contents Acknowledgements ....................................................................................................................................... 3 Acronyms and Abbreviations ....................................................................................................................... 4 Objective of the Report ................................................................................................................................. 8 1. Kosovo Electricity Sector Overview .................................................................................................. 10 2. Energy Affordability and Patterns of Energy Consumption .............................................................. 12 3. Poverty and Distributional Impact of Tariff Increases ....................................................................... 21 A. Simulated Impact on Electricity Budget Share ............................................................................... 22 B. Simulated Impact on Incidence of Electricity Poverty ................................................................... 24 C. Simulated Impact on Poverty and Inequality .................................................................................. 25 D. Simulated Household Welfare Losses: Who Loses and by How Much? ........................................ 28 4. Current Social Protection Scheme for Energy Vulnerable Consumers .............................................. 30 A. Background of Current Electricity Benefit and Eligibility Criteria ................................................ 30 B. Simulations of the Current Electricity Benefit Scheme .................................................................. 35 5. Reforming the Social Protection Program for Energy Vulnerable Customers ................................... 39 A. Short-term: Redesigning the electricity benefit.................................................................................. 39 B. Medium-term: Reforming the electricity benefit in the context of SAS Reform ............................... 41 6. Conclusions and Policy Recommendations ....................................................................................... 46 References ................................................................................................................................................... 50 Methodological Appendix .......................................................................................................................... 51 Annex A1. Methodology to Estimate Electricity Consumption ................................................................. 51 Annex A2. Methodology to estimate welfare impact of energy tariffs....................................................... 51 Annex A3. Description of National Framework of Vulnerable Consumers ............................................... 53 Annex A4. Program to support vulnerable customers in North Macedonia and reduce energy poverty .... 56 Annex A5. Energy/electricity poor or vulnerable customers in selected EU member States, Western Balkans and other countries: definitions and support measures ................................................................. 58 Annex A6. Simulating electricity consumption thresholds for the subsidy ................................................ 62 Annex A7. Additional Figures and Tables.................................................................................................. 67 5 List of Figures Figure 1: Electricity prices for household consumers, second semester 2018 (Euro cents/kWh) .......... 10 Figure 2. Energy Affordability in a Broader Context: Kosovo versus Other Europe and Central Asia Countries ................................................................................................................................. 12 Figure 3. Nearly all household are connected to electricity, although access to district heating among the bottom quintile and vulnerable groups is lacking .............................................................. 14 Figure 4. Vulnerable groups cannot afford to keep their house warm .................................................... 14 Figure 5. Electricity Is the Largest Component of Household’s Energy Expenditures .......................... 15 Figure 6. Households in the Bottom Quintile Spent a Larger Fraction of Their Expenditure on Electricity than Those in Higher Quintiles, but Less on Gas .................................................. 16 Figure 7. Who Are the ‘Electricity’ Poor? .............................................................................................. 17 Figure 8. Are the poor and the electricity poor the same households? ................................................... 19 Figure 9. Households’ budget shares spent on energy by different tariff increase scenarios and elasticities ................................................................................................................................ 23 Figure 10. Under the hypothetical case of a 50 percent tariff increase, the share of households’ budget going to electricity increases, particularly for the poor, households in the bottom quintile b20 not covered by social assistance, and single-elderly households ............................................ 23 Figure 11. Incidence of Electricity Poverty Rises More among the Single Elderly and households in the Bottom Quintile Not Covered by Social Assistance ............................................................... 25 Figure 12. Simulated Impact on Poverty Impacts, alternative elasticity scenarios................................... 26 Figure 13. Welfare losses are larger among poorest households since they spend a larger share of their budget on electricity ................................................................................................................ 28 Figure 14. Initial Electricity Shares and Simulated Welfare Losses of Rising Electricity Tariffs, vulnerable groups (as % of household budget) ....................................................................... 29 Figure 15. Structure of the social assistance system of Kosovo ............................................................... 30 Figure 16. Main types of short-term and medium-term protection measures ........................................... 34 Figure 17. Eligibility and Coverage of Electricity Benefit by Quintile .................................................... 37 Figure 18. Eligibility and Coverage of Electricity Benefit for vulnerable groups .................................... 37 Figure 19. Targeting: Share of energy transfer going to each consumption quintile................................ 38 Figure 20: Adequacy: Share of Monthly Household Expenditure constituted by Electricity Benefit, by Quintile .................................................................................................................................... 38 Figure 21: Coverage of households with reformed electricity subsidy..................................................... 42 Figure 22: Cost and overall coverage of reformed subsidy ...................................................................... 43 Figure 23: Coverage of current and reformed electricity benefit, Selected groups .................................. 44 Figure 24. Targeting and adequacy of the reformed electricity subsidy ................................................... 45 6 List of Tables Table 1. Feed-in tariffs for various technologies of renewable energy ................................................. 11 Table 2. Benchmarking Kosovo with Other Countries in Europe and Central Asia Undergoing Energy Reforms ................................................................................................................................... 27 Table 3. Current electricity subsidy – beneficiaries and fiscal cost, 2017............................................. 31 Table 4. Advantages and disadvantages of the present electricity subsidy model ................................ 32 Table 5. Beneficiaries of Electricity Benefits, by Category– Data vs Estimate .................................... 36 Table 6. Budget allocated to Energy Program – Data vs Estimate ........................................................ 36 Table 7: Adjusted bill deduction size .................................................................................................... 40 7 Objective of the Report The objective of the Report is to inform the preparation of a Government Program for establishing the status of customers in need, the scope of their rights, and protection measures to help them meet demands for electricity in case of electricity tariff increases. The Report analyzes patterns of electricity consumption, affordability, poverty and distributional impacts of hypothetical scenarios of tariff increases. It also reviews the current protection mechanism with an electricity subsidy, models the impact of short- and medium-term reform options, and recommends concrete mitigation measures. In accordance with Law No. 05/L-085 on Electricity of 2016, the Government of Kosovo should develop a detailed Program for establishing the status of customers in need, the scope of their rights, and protection measures to help them meet demands for electricity. Electricity accounts for almost 90 percent of the total energy expenditures of households in Kosovo, and over 97 percent of the energy expenditures of poor households.1 According to the Law on Electricity2, the Ministry of Labor and Social Welfare (MLSW) in cooperation with the Ministry of Energy, Ministry of Finance and in consultation with the Energy Regulatory Office (ERO) and other stakeholders of the electricity sector should develop a social protection program (Program), providing social benefits to ensure the necessary electricity supply to customers in need, or providing for support for energy efficiency improvements, to address energy poverty. Such measures shall not impede the effective opening of the electricity market and its functioning. Customers in need are household consumers, who, due to social status, enjoy some special rights regarding the supply with electricity, to be provided in exceptional cases. Commitment to the development of this Program is taken also with Kosovo’s Energy Strategy, and with the requirements of the Treaty of Energy Community as defined by the Third Energy Package 3 . The Program should propose an approach to identifying the electricity poor who are vulnerable4, and outline short- and medium-term measures for their protection. 1 Household expenditures on electricity and other energy sources are calculated using data from the Kosovo Household Budget Survey (HBS) 2017. 2 The Law on Electricity puts forward a number of characteristics of the future protection scheme (Article 49) with definition of the problem, mitigation measures, outputs and outcomes, stakeholders, institutional setup for implementation and timeframe, financing sources, process monitoring and evaluation of impacts. Specifically, “the Ministry in charge for social welfare shall develop, in cooperation with the Ministry in charge for energy, Ministry of Finance and in consultation with the Regulatory and other stakeholders of the electricity sector, a detailed program for establishing the status of socially customers in need, the scope of rights, as well as measures aimed at protecting the socially customers in need in order to meet their electricity demand.” 3 Kosovo, like the rest of the Western Balkan countries, is signatory of the Treaty for Establishing the Energy Community of South East Europe which creates the legal framework for an integrated energy market between the European Union and South East European partners. The Treaty mandates the application of the EU model of energy regulation and the acquis Communautaire, and implies de facto EU membership in the energy sector. The Treaty establishing the Energy Community in South East Europe was signed in Athens on October 25, 2005, building on the Memorandum of Understanding on the Regional Electricity Market in South East Europe and its integration into the European Union Internal Electricity Market (“the Athens Memorandum”) of November 15, 2002. The Treaty calls for the establishment of the legal framework for an integrated energy market, and with it a single regulatory space for trade in energy between the EU and 9 South East European countries, including the United Nations Interim Administration Mission in Kosovo (UNMIK) pursuant to the United Nations Security Council Resolution 1244 on behalf of Kosovo. 4 Not all households classified as electricity poor households are poor/vulnerable. For instance, single elderly households are not among the poorest, but are much more vulnerable to rising electricity prices. With respect to protection of vulnerable customers, the Contracting Parties should define the concept of vulnerable consumer at national level, adopt measures to protect such customers, and address energy poverty. There is no common EU-wide definition in the Third Energy Package. The protection of vulnerable consumers is to be implemented at national levels, and definitions are required in the national legislation (For country- level examples see Annex 1). They should reflect national circumstances, but at the same time take into consideration some general rules and characteristics. 8 The Energy Community Treaty Ministerial Council has endorsed a proposal for a regional definition of vulnerable customers for Contracting Parties in October 2013. According to this proposal, a socially vulnerable customer in the electricity sector should meet the following four criteria: (i) uses energy for supplying his/her permanent housing; (ii) does not exceed the maximum energy consumption per person: when defining electricity consumption level per person, Contracting Parties shall consider total consumption of up to 200 kWh/month for a family with up to 4 members, and reflect seasonality; (iii) belongs to a category of citizens with lowest income: for the definition of low income, beside the income of all available assets shall be taken into account; (iv) have her/his electricity consumption supplied through single-phase meter with a connection not exceeding maximum power. Furthermore, the definition shall not include more than a minority of population. Market prices of the electricity should be cost reflective and consumption of vulnerable customers should be financed by social allowances.’ The Program to be developed will also serve to mitigate the upward pressure on electricity prices as a result of forthcoming significant investments in the energy sector and structural reforms. Significant investments are planned for (i) new mining equipment and lignite mine expansion; (ii) Kosovo B refurbishment for life extension and emission control systems, estimated to be around €300 million starting in 2021 and continuing after commissioning of a new power plant; and (iii) construction of new generation capacities as a replacement for the planned decommissioning of Kosovo A power plant. Further investments are required by clean energy commitments to scale up renewable energy to 25 percent by 2020 and more than 30 percent until 2030. Apart from tariff impacts driven by investments in the sector, regulatory reforms are expected to have further impact on end-user electricity tariffs, especially residential tariffs. Elimination of tariffs cross subsidies between residential and commercial customers and passthrough of renewable energy costs will undoubtedly put upward pressure on residential tariffs. Other reforms such as deregulation of wholesale electricity prices and integration into regional market, considering that Kosovo has the lowest electricity tariffs in the region and in Europe, will also have an impact on increasing current residential tariffs. The current electricity subsidy has a simple design, is easy to implement, and can only be used for its core purpose - paying for electricity consumption. However, its design also has several weaknesses which could undermine its effectiveness in protecting the electricity poor, especially in the case of significant tariff adjustments. While some of these weaknesses are generic, i.e. typical for electricity subsidy schemes in general, others are program-specific, i.e. stemming from specific design characteristics and implementation rules of Kosovo’s electricity subsidy. The GoK has initiated the establishment of an inter-institutional working group to draft a government program for mitigating the adverse effects of energy tariffs increases. On February 25, 2019, the Ministry of Labor and Social Welfare (MLSW) established the working group on protection of energy vulnerable customers, composed of representatives of the Ministry of Finance, Ministry of Economic Development (MED), MLSW, Energy Regulatory Office (ERO), Kosovo Energy Distribution Services (KEDS), and Kosovo Electricity Supply Company (KESCO). The program is expected to provide a definition of energy / electricity poverty; to specify who are the energy / electricity vulnerable customers; to decide on a mix of most relevant and effective protection measures for the energy / electricity vulnerable customers in Kosovo in the short and medium term. The findings of the report will inform the work of the working group in defining a program for the protection of energy vulnerable customers in need. 9 1. Kosovo Electricity Sector Overview Two aging coal-fired power plants continue to dominate domestic power generation. Approximately 97 percent of Kosovo’s domestic power generation comes from two outdated coal-fired power plants, Kosovo A (commissioned in the 70s) and Kosovo B (commissioned in the 80s). These power plants provided 5,008 GWh in 2018 and 5,121 GWh of generation in 2017. Hydropower made up most of the remaining 3 percent of generation, with one medium-sized hydropower plant (Ujman/Gazivoda) and other renewable sources contributing about 303 GWh in 2018 compared to 179 GWh in 2017. In October 2018, the country’s first larger wind park became operational with an installed capacity of 32.5 MW. Power generation has increased in recent years with increased output from Kosovo A and B power plants and investments in new renewable energy generation and prices for consumers have remained low. Due to the increased power output from the rehabilitated Kosovo A and B units, total power generation including hydro power plants has increased steadily from 4,527 GWh in 2008 to 5,311 GWh in 2018, an increase of 17 percent. This increase has come from technical improvements at the existing power plants, especially Kosovo A. The coal-fired power plants have generated electricity at a very low cost of about 29 €/MWh in 2018, which contributed to keeping household electricity tariffs at the lowest levels in the region. According to the latest available Eurostat data on electricity prices (see Figure 1), Kosovo had the lowest average electricity prices in the region for household consumers, followed by Serbia and North Macedonia. Figure 1: Electricity prices for household consumers, second semester 2018 (Euro cents/kWh) 10.3 9.1 8.64 7.87 7.09 6.38 Albania Bosnia and Kosovo Montenegro North Macedonia Serbia Herzegovina Source: Eurostat. While distribution losses have been reduced in recent years, they remain high by regional standards. Distribution losses fell from 43 percent of power entering the distribution network in 2008 to 29 percent in 2017. In 2017, losses comprised about 17 percent commercial and 12 percent technical losses 5. These distribution losses are factored in the end user tariffs and the regulator sets loss reduction targets for the distribution company. The total cost of losses has slightly increased since 2013 due to the rising average price of electricity paid for losses as it is set by the energy regulator. While the volume of losses has decreased further since 2013, the average price has risen from 30.21 €/MWh in 2013 to 35.43 €/MWh in 2017. Increases in wholesale prices have a double impact on end user tariffs because of the actual price increases on one hand and the increased cost of losses that are factored in the tariff on the other hand. 5 These high grid losses are a result of years of underinvestment in the distribution network and legacy of high commercial losses due to illegal grid connections. 10 Kosovo’s energy strategy includes significant investments in new power generation as a replacement for the closure of Kosovo A and upgrade of Kosovo B power plants. The GoK had selected an investor for the construction of a new 450MW coal-fired power plant as a replacement for the decommissioning of Kosovo A, but the project did not materialize. . The aging Kosovo B coal-fired power plant will require significant investments in reducing emissions according to EU standards and extend its lifecycle. According to an EU-funded study, about €300 million will be required for reducing Particulate Matter (PM), Sulfur Oxides (SOx) and Nitrogen Oxides (NOx) emissions to comply with the EU’s Industrial Emissions Directive and prolong the power plant’s operational lifetime for additional 30 years. The EU has committed IPA (Instrument of Pre-Accession Assistance) grant funds to bring the PM and NOx emissions in compliance with EU directives. As a signatory of the Energy Community Treaty, Kosovo has committed to achieve by 2020 a renewable energy target of 25 percent as a share on energy consumed. In line with this target, the MED has set renewable targets for each technology that would be supported through the support scheme to incentivize investments in renewable energy generation. These targets translate into allocated quotas for each technology that will be supported through a fixed feed-in tariff and period of guaranteed offtake. See table 1 summarizing the targets, feed in tariffs and period of support. The feed-in tariffs for the various renewable technologies are more than double the cost of current generation from the existing coal-fired power plants (~ 28 €/MWh). To date, 7 small hydropower plants (HPP) with a total capacity of 36MW have been put into operation, while 2 wind power plants with a combined capacity of 33.75MW are now operational. Additional 5 solar photovoltaic (PV) plants of combined installed capacity of 7MW have been connected to the distribution grid. Kosovo will need to set new renewable energy targets for 2030, in line with the Energy Community commitments and EU energy and climate plans. These targets will vary for each country, but they will likely exceed 30 percent of energy consumed. This implies that Kosovo will need to put further effort to scale up investments in renewable energy. Table 1. Feed-in tariffs for various technologies of renewable energy Technology Quota for Feed-in tariff Period of support support scheme €/MWh scheme (years) in MW Small hydro 240 67.47 10 Wind 150 85 12 Biomass/Biogas 20 71.3 10 Solar 30 136.4 10 Source: Energy Regulatory Office. In addition to new power generation, additional investments will be needed in coal mining equipment and mine expansion. During the next 5 years, Kosovo Energy Corporation (KEK) will need to invest an estimated €400 million to replace and expand the capital equipment base and ensure compliance with the requirements of the local mining regulator in terms of technical, environmental and social performance (land acquisition) of the mining operation. These mining investments will impact the cost of coal leading to additional costs for coal-fired generation. 11 2. Energy Affordability and Patterns of Energy Consumption In Kosovo, electricity prices are the lowest in the region. However, the share of a households’ spending going to electricity in 2017 is comparable to Bosnia and Herzegovina, Albania, and Serbia, all of whom have significantly higher electricity prices. In general, higher prices for electricity (the most important energy source for households) are generally associated with higher burdens of electricity spending in household budgets in the Europe and Central Asia (ECA) countries (see Figure 2). The evidence also suggests that, when looking at changes over time for a particular country, households seem to have limited abilities to keep their electricity expenditures constant in an environment of rising electricity prices (by substituting for cheaper sources of energy), resulting in households cutting down other types of consumption such as food, health, or education. In Kosovo, energy sector investments and reforms may lead to rise in electricity prices. Given the high share spent on electricity in the household budget, it is critical to assess, at the micro level, the potential impact of rising energy prices both on affordability as well as distributional impact on households. Figure 2. Energy Affordability in a Broader Context: Kosovo versus Other Europe and Central Asia Countries Electricity Price and Electricity Share of Total Household Expenditures (excludes health, rent, and durables) 25 2013 electricity price (US cents/kwh) Estonia 2004 20 Poland 2008 Hungary 2007 Lithuania 2004 Turkey 2013 Romania 2013 15 Montenegro 2011 Moldova 2013 Bosnia and Herzegovina 10 2011 Albania 2012 Macedonia 2010 Georgia 2014 Serbia 2013 Azerbaijan 2008 Armenia 2014 Russia 2002 5 Ukraine 2014 Kosovo 2017 0 0 2 4 6 8 10 12 14 16 electricity as a share of total household expenditures (%) Source: World Bank estimates based on Europe and Central Asia Poverty Database (ECAPOV) standardized household surveys that cover most countries in the Europe and Central Asia region. 12 This section analyzes energy affordability in Kosovo, by relying on the 2017 Household Budget Survey (HBS), complemented with administrative data on energy tariff structure and consumption. The HBS collects detailed expenditure information for all household (HH) categories, including energy and electricity expenditure with disaggregation by energy source and allows for indirectly approximating the amount of electricity consumed. The HBS also allows disaggregation among some vulnerable groups, including but not limited to beneficiaries of social assistance, single-elderly households, and those with unemployed ages 55 and above. See Annex A1 for more detail on data and methodology. Traditional indicators show that energy affordability is a salient issue in Kosovo, affecting relatively more the poor and vulnerable groups. Energy affordability usually includes monetary and non-monetary dimensions. Non-monetary measures include connectivity (percentage of households with access to a specific source of energy), bill arrears, and comfort (the inability of households to keep the house warm)6. Connectivity of household to other energy sources different from electricity is low, due to lack of alternatives in the country such as natural gas or district heating. Comfort measures show a significant fraction of households unable to keep their home warm during winter, with poor households being particularly affected. Monetary measures include the electricity share of the overall household expenditures and incidence of energy and electricity poverty. Energy expenditures represent about 7.9 percent of total household expenditures in Kosovo, and electricity itself accounts for about 6.4 percent, reflecting the high dependency of households on electricity. Even though nearly all households are connected to at least one centrally provided energy source in Kosovo, access to district heating among the poor and in rural areas is lacking.7 As in most ECA countries, access to electricity is nearly universal (higher than 99 percent in both rural and urban areas). Access to alternative energy sources, such as district heating is lacking and there are important differences across the welfare distribution and localities and between recipients and non-recipients of social assistance. District heating is only available in some central neighborhoods of Pristina, and small parts of Gjakova and Mitrovica. Households in the bottom quintile, those that are extremely poor (below the food poverty line [FPL])8, and households in rural areas are less likely to be connected to district or central heating9. Natural gas is only available as compressed natural gas (CNG) in lower volumes, which is typically expensive, therefore access to gas is also low across the welfare distribution.10 According to a survey on household energy consumption collected between 2010 and 2012, wood is the primary source of heating for the majority of households.11 6 Bill payment of electricity bills is not available in the HBS, since the only question in the survey on arrears does not differentiate utility bills from others (loan payments, rent, utility bills, etc.). A cross-country comparable measure of comfort is usually typically available in the Statistics on Income and Living Conditions (SILC) survey. Since 2018 SILC data is not yet available for Kosovo, we construct this measure using the following question in the HBS: Can the household afford to keep the home adequately warm during winter? The main difference from the question in the SILC is that in the HBS the question refers to the winter time. 7 In the HBS, we use as a proxy for connectivity households reporting positive energy expenditures. 8 For information on how poverty is measured in Kosovo, see Box 2 in Chapter 3. 9 In the HBS 2017, about 10 percent of households in rural areas reported to be connected to district heating, which seems high as rural households typically are not connected to district heating. We assume that what is meant is not ‘district heating’ but ‘central heating’, which refers to a household level heating system using a boiler within a house/apartment that distributes the heat through water filled radiators throughout the home. 10 In the HBS, we have information on electricity and central heating connectivity but not gas, so to identify households that are connected to gas, we use information on gas expenditures (if positive, household is connected). 11 Source: Bowen et al, 2013: “Kosovo Household Energy Consumption Facts and F igures.” 13 Figure 3. Nearly all household are connected to electricity, although access to district heating among the bottom quintile and vulnerable groups is lacking Access to electricity and Central Heating/District Heathing(%) 100 80 Access to Percent 60 Electricity 40 20 Access to 0 District Heating Source: World Bank estimates based on Kosovo 2017 HBS. Note: Q1 to Q5 refer to quintiles of the consumption distribution (Q1 = lowest quintile and Q5 = richest quintile). SA refers to social assistance scheme. Extreme poor and poor refer to those households living below the national Food Poverty Line (FPL) and the Consumption Poverty Line (CPL). B20 refers to households in the bottom quintile of the consumption distribution, either covered by social assistance or not. Relatively poor and vulnerable households are less likely to keep their houses warm during winter season. About 41.3 percent of households in Kosovo reported to be unable to keep their home warm during winter (Figure 4, panel a). This is much higher than the average for EU countries, also higher than in Serbia and North Macedonia. The percent of households which cannot afford home heating decreases across welfare quintiles, with half of the households in the poorest quintile unable to keep their home adequately warm. This pattern is more evident in rural areas (46.1 percent of households) compared to urban areas (34.2 percent). More than 60 percent of households which receive social assistance are vulnerable to cold weather and this is even higher (close to 70 percent) among the poorest households on social assistance. (Figure 4, panel b). Figure 4. Vulnerable groups cannot afford to keep their house warm Panel a. Inability to keep home adequately warm, EU Countries and Kosovo 45.0 41.3 40.0 35.0 30.0 Percent 25.0 20.0 15.0 7.8 10.0 5.0 0.0 Denmark Czechia Malta Sweden Austria Germany Croatia Latvia Slovenia Slovakia Spain Portugal Romania Cyprus North Macedonia Luxembourg Poland Greece Lithuania Norway France Belgium Hungary EU28 EU27 Kosovo Finland Netherlands Ireland Italy Estonia Serbia United Kingdom Bulgaria 14 Panel b. Inability to keep home adequately warm during winter, Kosovo (%) 100 90 80 70 60 Percent 50 40 30 20 10 0 Source: For EU countries, published estimates based on Statistics on Income and Living Conditions (SILC) survey. For Kosovo, World Bank estimates based on Kosovo 2017 HBS. Note: The figure is obtained by calculating the share of households who can afford to keep their house warm. Electricity is the largest component of household energy expenditures, accounting for about ninety percent of overall energy expenditures (Figure 5).. Households in the bottom quintile spend a largely higher fraction of their energy expenditure on electricity than those in the upper quintile of the consumption distribution (97.4 percent versus 80.8 percent, respectively) and less on firewood (0.7 versus 15.8 percent, respectively). Firewood spending is likely underreported in the survey as it is oftentimes collected from the woods. Firewood represented the main source of heating for the majority of households in Kosovo according to a survey conducted in 2010-2012. Nevertheless, the high share of electricity expenditures in households’ budget means that any substantial impact in electricity prices may have a major impact on household energy expenditures, especially among households at the bottom whose consumption levels may be close to subsistence levels. Figure 5. Electricity Is the Largest Component of Household’s Energy Expenditures Household Energy Mix (%), by Quintile 100 0.7 4.4 5.2 7.9 9.4 15.8 80 60 Percent 89.5 97.4 93.7 92.2 40 87.8 80.8 20 0 All households Q1 Q2 Q3 Q4 Q5 Electricity Gas Oil Firewood Propane Other Fuel Source: World Bank estimates based on Kosovo 2017 HBS. Note: Other heating includes consumptions on district heating, liquefied gas, coal, firewood, and other heating sources. 15 Energy represents a substantial share of household consumption in Kosovo: energy expenditures account for 8.2 percent of overall household expenditures, and among these electricity accounts for 6.6 percent (Figure 6). Who consumes the most energy in Kosovo? The share of the budget spent on electricity tends to decrease with welfare. Among those at the bottom quintile, electricity accounts for 7.5 percent. As income increases, households tend to spend a smaller portion of their incomes on electricity and a larger fraction on gas (propane and butane). This implies that rising electricity tariffs are regressive, because those at the bottom of the distribution are more ‘vulnerable’ due to higher electricity shares.12 Among the vulnerable groups, single-elderly and moderate poor households tend to spend a larger share of their budget on electricity (about 9.9 percent and 7.4 percent, respectively). Similarly, female-headed households and households in the bottom quintile not covered by social assistance also have relatively higher-than-average budget shares of electricity expenditure, 7.2 percent and 8 percent, respectively. Other vulnerable groups have similar or slightly higher-than-average electricity spending. Figure 6. Households in the Bottom Quintile Spent a Larger Fraction of Their Expenditure on Electricity than Those in Higher Quintiles, but Less on Gas Household Budget Share of Electricity Spending (%), by Quintiles 7.5 7.5 8 6.6 6.8 6.5 6 5.3 Percent 4 2.7 1.4 1.6 2 0.9 1.0 0.1 0 All Q1 Q2 Q3 Q4 Q5 households Electricity Gas Oil Firewood Propane Other Fuel Source: Staff estimates based on Kosovo 2017 HBS. Affordability of energy is often considered a critical issue when spending on energy surpasses a certain threshold. We analyze the incidence of households spending more than 10 percent of the overall budget on electricity (or energy),13 in other words, the ‘electricity poor’ or ‘energy poor’ (see Box 1). About 15.8 percent and 25.4 percent of the households in Kosovo are estimated to be ‘electricity and energy poor’ using these definitions, respectively. Consistent with electricity consumption patterns across the welfare distribution, electricity poverty is dominant among households in the bottom quintile compared to the percent among richest households (30.6 percent versus 7.5 percent, see Figure 7, panel a). The share of electricity poor in urban areas is higher than rural areas (about 8 percentage points higher). The incidence of electricity poverty is much higher than average in households in the bottom quintile that do not receive the social assistance scheme (SAS), roughly 60 percent. Among vulnerable groups, the incidence of energy and electricity poverty is higher among poor and single elderly household (Figure 7, panel b). About one third of poor households in Kosovo spend more 12This is also the pattern observed in other countries, such as Serbia (see Nguyen and Robayo-Abril 2017). 13There is no consensus regarding the concept of ‘energy poverty’. This 10 percent is an ‘ad hoc’ threshold currently used for developing countries in the literature. Other concepts may include alternative criteria, like minimum energy consumption or minimum expenses to maintain an adequate temperature, as recommended by the World Health Organization (WHO). 16 than 10 percent of their budget on electricity. Other vulnerable groups present higher incidence of electricity poverty compared to the average household in Kosovo. The incidence of electricity poverty among households in the bottom quintile that are not social assistance recipients is higher than average, about 34 percent (Figure 7, panel b). Other groups, such as households with young children (younger than 5 years) and disabled members, as well as households stating as their main source of income war-related pensions are less likely to be electricity poor when compared to the average household in Kosovo. Figure 7. Who Are the ‘Electricity’ Poor? Panel a. The Incidence of Electricity Poverty Is Higher among households in the Bottom Quintile (%) Incidence of Electricity and Energy Poverty 60% 50% 40% 30% 20% 10% 0% Poorest 2nd 3rd 4th Richest All Urban Rural HH with B20 with B20 w/o Quintile Quintile SA SA SA Energy Poverty Electricity Poverty Panel b. Incidence of Energy and Electricity Poverty Is Also Higher among Poor and Single-Elderly Households Incidence of Electricity and Energy Poverty by Groups (%) All 15.8 Single elderly HH 45.7 Bottom Quintile - Without Social Assistance 33.3 Bottom 20 30.6 Female-Headed HH 20.1 HH with Unemployed 55+ 16.6 Social Assistance Recipient Households 17.7 Bottom Quintile - With Social Assistance 21.1 HH with Child <5 13.0 HH with Disabled members 12.4 Social Assistance Recipients (>700 Euros) 19.0 War-related Pension Recipient HH 13.9 Poor 31.0 Extreme Poor 32.5 0 10 20 30 40 50 60 70 Energy Poverty Electricity Poverty Source: World Bank estimates based on Kosovo 2017 HBS. Note: A household is considered electricity/energy poor if it spends more than 10 percent of its overall budget on electricity/energy. ‘Bottom 20’ refers to the poorest quintile of the national consumption distribution. 17 Box 1. Measuring energy and electricity poverty to design better programs In the European context, particularly in countries where heating needs are significant, researchers have long sought to isolate the needs of those whose energy expenditure represent a major burden on their budget. A relatively easy way to proxy such burden is provided by setting a threshold in terms of energy share devoted to energy costs, customarily set at 10 percent (see for example World Bank 2013) and identifying energy poverty. By analogy to such measure, source specific thresholds have also been set, thereby identifying for example electricity poverty as spending more than 10 percent of one’s budget on electricity. Two recent developments are contributing to refining the way such indicators are designed. One has been the establishment of an Observatory on Energy Poverty, tasked with monitoring the situation in the EU. While the European Commission has reiterated that a common definition cannot be developed at the European level, given the specific nature of energy poverty in different countries, this effort can benefit from some common data. The EU Survey of Income and Living conditions which is run by all EU Member States and increasingly others, for example, includes self-reported information on not managing to heat appropriately and having arrears on utility bills. The Commission to highlight the importance of energy poverty is reflected in innovation at the country level. One influential development has occurred in the UK. The country has been at the forefront of the effort to measure and tackle energy poverty since 2001, when it introduced a fuel poverty measure identifying those who would spend more than 10 percent of their income to keep their home at 21 degrees. A new report has reassessed that measure and suggested two new fuel poverty measures capturing respectively the incidence and the severity of the phenomena have been proposed. The indicator for the incidence of energy poverty, known as LIHC (Low Income, High Cost) is defined as having energy costs higher than the median, and having an income net of energy costs below the official poverty line. The indicator of depth of energy poverty is more complex to measure and involves measuring the gap between household energy needs and a reasonable threshold. A different international effort which provides some new ideas on how to look at electricity affordability is provided by the Multi-Tier Framework for Measuring Energy Access developed by the World Bank, and adopted to monitor the energy access Sustainable Development Goal (SDG). Such a framework, which integrates access and quality issues in one set of measures, also provides basic thresholds on expenditure on electricity for different purposes. Those are identified as 5 percent for heating and 5 percent for cooking adopted by the multi-tier monitoring framework. All these different efforts, which are ultimately aimed at measuring more accurately a very specific facet of deprivation, suggest that there is scope for experimentation in designing new measures and programs which help address the multiple causes of energy poverty itself: low incomes, poor housing conditions and particularly low energy efficiency, and high energy spending. The incidence of poverty is the highest among SAS-beneficiary households, whereas electricity poverty mostly affects single elderly households. Overall, almost 60 percent of SAS beneficiaries are either poor (39 percent), electricity poor (8 percent), or both poor and electricity poor (12 percent) (Figure 8, panel A). Single elderly households, on the other hand, are at relatively low risk of poverty (5 percent) but face very high incidence of electricity poverty (46 percent). Among other vulnerable categories, households with a disabled member have a high incidence of poverty (30 percent) and female-headed households have a relatively high incidence of electricity poverty (20 percent). Among poor households, about a third are electricity poor (31 percent) (Figure 8, panel B). None of the poor among the vulnerable groups is at a particularly elevated risk of electricity poverty. 18 Figure 8. Are the poor and the electricity poor the same households? Panel a. The majority of SAS beneficiaries are poor and/or electricity poor and almost half of single-elderly households are electricity poor. Composition of vulnerable groups by poverty and electricity poverty status 100% 80% 41 54 62 73 66 71 72 60% 8 79 12 40% 8 41 4 11 8 20% 11 39 5 5 16 9 5 26 18 4 5 11 16 8 8 0% 5 All SAS Households Single Elderly HH with HH with HH with child <5 Female-headed War-related HH disabled unemployed HH pension member 55+ household Only poor Poor & electricity poor Only electricity poor Neither poor nor electricity poor Panel b. Poor households are twice as likely to be electricity poor as the overall population, but among poor households, no one category stands out as particularly electricity-poor Incidence of electricity poverty among poor households All 30.96 HH with Disabled members 14.42 HH with Unemployed 55+ 23.45 HH with Child <5 24.46 Female-Headed HH 32.66 0 5 10 15 20 25 30 35 Source: World Bank estimates based on Kosovo 2017 HBS. Given the burden of electricity spending in Kosovo and the likelihood of rising electricity tariffs, it is important to understand patterns of electricity consumption to assess household vulnerability to price shocks. The characteristics of household electricity consumption in Kosovo may also inform the design of policy interventions to protect vulnerable populations against high energy tariffs (for instance, using tariff schemes that depend on electricity consumption levels, or energy subsidies that are conditional on consuming below a certain threshold 14 ). Household surveys often do not directly record energy consumption levels. In the 2017 HBS, we observe self-reported consumption data for many household items but not for electricity. Therefore, electricity consumption must be estimated using reported after-tax electricity expenditures. We estimate the mean electricity consumption for each household given the observed electricity expenditures, the tariff structure in place in 2017, and the VAT rate. The mean electricity consumption is, therefore, estimated with some measurement error (for more specific details on 14 These schemes have been used in other countries. 19 the methodology, see Annex A2). Then, we compare our survey-based estimates with data from the regulator as a robustness check, as well as with electricity consumption estimates from other countries. Average electricity consumption in Kosovo is relatively high when compared with other countries in the broader Europe and Central Asia region. Based on the 2017 HBS and this methodology, the estimated residential mean (monthly) electricity consumption per household in Kosovo is 491 kWh and the median is 472 kWh (see distribution of electricity consumption in Figure A7-1 in Annex A7). This consumption estimate is higher than estimates for the Western Balkans, ranging from 250 kWh per month to 400 kWh per month15. These estimates are much higher compared to other estimates in Environmental Policy Committee (EPOC) countries, where alternative heating sources are available to most households, where average electricity consumption is closer to basic needs (from 120 kWh per month to 200 kWh per month). The residential end-use sector has a seasonal variance in electricity use, with high demand in the winter months, but weaker seasonality patterns are observed among poorer households. The estimated electricity consumption follows a seasonality pattern that one would expect from heating needs: households consume more electricity on average during the first quarter of the year (January–March), regardless of their position in the welfare distribution. The pattern of seasonality differs slightly across the welfare distribution. While the bottom quintile has less variation across quarters, in the upper three quintiles, electricity consumption is considerably higher in the first quarter (see Table A7 -3 in Annex A7). How reliable are our results based on reported electricity expenditures in the HBS survey? We use data from the Kosovo Energy Regulatory Office (ERO) to validate the information collected in the survey and to test underreporting of electricity consumption. In the validation, we check how our mean estimates of electricity consumption using HBS data compared with those provided by ERO (by region and month and overall). The robustness check exercise shows that we slightly overestimate average electricity consumption using the HBS, but the size of the bias is relatively small with the exception of some regions. When we look at seasonal variation, our results are based on HBS underestimate electricity consumption by a relatively ‘small’ amount, and we capture some of the seasonality pattern. When looking at estimates by region, we find that in some regions, the survey estimates are quite close to the ERO estimates. Even though we are underestimating electricity consumption in Ferizaj and Peje, in terms of regional differences, our HBS estimates capture a large share of the geographical heterogeneity (see Table A7 – 1 in Annex A7). 15 Source: Balancing Act. 20 3. Poverty and Distributional Impact of Tariff Increases In this study, we use both national and international poverty lines to measure poverty in Kosovo, and we simulate the poverty and distributional impact of hypothetical electricity tariff increases over the period 2017-2030. To assess the poverty and distributional impact of the electricity price change, given the uncertainty in electricity prices, we simulate four hypothetical scenario of real tariff increases (25, 50, 75, and 100 percent). Our goal is to examine (a) changes in the electricity budget share for the whole population and for vulnerable groups; (b) changes in the incidence of electricity poverty; (c) changes in consumption poverty rate, poverty gap, and inequality (disaggregated by region, urban/rural, quintiles, and vulnerable groups); and (d) welfare effects across the consumption distribution to determine winners and losers. Box 2. How is poverty measured in Kosovo? In Kosovo, consumption is used as the measure of individual well-being or welfare. Household consumption is calculated as the total value of a household’s expenditure on food and nonfood items as recorded in the Household Budget Survey (HBS), a nationally representative survey conducted each year, including imputed values of any home-produced food items that were consumed by the household. In keeping with past practices in Kosovo, expenditures on consumer durable items and rent are excluded from the consumption measure. The welfare aggregate used to officially measure poverty and inequality in Kosovo is expressed in consumption per adult equivalent, and this measure of living standards is assessed against a poverty threshold that is held fixed in real terms over time and space; the monetary value of the poverty line is updated annually to account for changes in prices. Two poverty lines are used in the country: a poverty line that is considered adequate to meet basic needs and a lower extreme poverty line (food poverty line). These poverty lines reflect the cost of purchasing food and non- food items, so as prices rise, nominal poverty lines increase. After adjusting for inflation, the poverty line and extreme poverty lines in current prices in the year 2017 are €1.85 and €1.31 per adult equivalent per day The ‘poor’ are defined as those with consumption per adult equivalent below the total poverty line whereas the ‘extremely poor’ are defined as those with consumption per adult equivalent below the FPL. The World Bank also monitors poverty using the international poverty lines (US$1.9, US$3.2 and US$5.5 per day (in 2011 purchasing power parity [PPP]), and the consumption aggregate to measure international poverty is expressed in per capita terms. International poverty lines are useful for international comparisons and cross-country benchmarking, while national poverty lines should be used to closely monitor development progress in the country and coordinate policies around poverty reduction. Source: World Bank (2019) Facing possible volatility in electricity prices, consumers may decide to adjust their demand and reduce electricity expenditures, but there is little evidence regarding the ability of households to adjust electricity consumption to rising electricity prices in low- and middle-income economies. How hard is it for households to adjust their consumption to higher prices? Given the scarce evidence, we simulate different sensitivity scenarios in this analysis. Several factors suggest the own-price elasticity is low in the short run 16 . Evidence suggests also that these elasticities are lower than for industrialized economies. This is due to a higher share of electricity consumption in low- and middle-income countries being attributed to essential uses such as cooking or heating the house during the winter months. On the contrary, in high-income countries, household consume a higher share of electricity for non-essential purposes and can thus adjust their consumption easier when faced with a price shock. Here, we present 16 A low elasticity implies that households’ electricity consumption is not very sensitive to changes in the price of electricit y. That is, if electricity prices increase, electricity consumption changes only marginally. 21 results assuming a price elasticity of −0.5, which corresponds to a medium-run scenario where households have some ability to adjust, but it is limited (inelastic range).17 A caveat of this analysis exists: we quantify only the direct welfare impact on household welfare, without considering indirect or general equilibrium effects, because an input-output matrix is not available for Kosovo. These first-order effects depend on the share of electricity in the total budget and its substitutability with other fuels and with other goods (see Annex A2 for details). The magnitude of the effects, therefore, will be higher in the short run than in the medium run. We do not quantify indirect impacts because there is no available input-output matrix in Kosovo, necessary to perform this analysis18. Second- order effects may include macro effects, both general price effects and the effects of higher prices or reduced subsidies on government expenditures and revenues, as well as more efficient and cleaner energy due to higher prices (and investment), which benefit consumers. A. Simulated Impact on Electricity Budget Share The share that households spend on electricity depends both on the size of the tariff increase and the expected elasticity (Figure 9) and reaches up to 6.6 percent in the case of a 100 percent tariff increase and zero elasticity (meaning that households cannot adjust their consumption of electricity and bear the full impact of the tariff increase). Simulated tariff increases for the benchmark case of 50 percent tariff increase and the baseline elasticity (-0.5) are expected to increase the share of the household budget spent on electricity by about 0.82 percentage points, on average, ranging from 0.9 percentage points among the poorest to 0.7 percentage points among the richest quintile.19 For the baseline case (elasticity of −0.5), the impact on the electricity share is highest for the bottom quintile (0.9 percentage points), whereas for households in the top consumption quintile, it is lower (0.7 percentage points) (figure 9, panel a). ‘Electricity-poor’ households would be most affected by likely tariff increases, facing a simulated electricity budget increase of 1.6 percentage point. Among vulnerable groups, single-elderly, households in the bottom quintile not covered by social assistance and the ‘extreme poor’ (below the FPL) would be more affected than the average household (figure 9, panel b). Differences between households in urban versus rural areas are relatively small: urban households would face an increase of 0.87 percentage points compared with 0.8 percentage points for rural households. (see Table A7 – 2 in Annex A7). 17 Estimating price elasticity requires an estimation of a system of demand functions, which is difficult with the data available. Therefore, for robustness checks, in our analysis, we consider four different elasticity scenarios: (a) households cannot adjust their consumption of electricity and bear the full impact of the tariff increase (price elasticity of electricity demand of 0); (b) households can adjust their electricity consumption by a small amount given that electricity is a necessity (elasticity of −0.25); (c) households can adjust slightly more in the medium run (elasticity of −0.5); and (d) households can adjust even more in the long run (ela sticity of −1.0). The sensitivity results are presented in the Methodological Appendix. 18 Input-Output matrices of countries with similar economic structure to Kosovo are not directly available to use, and therefore the magnitude of those effects is unknown. 19 A medium scenario of tariff increase (50 percent) is assumed for all estimates presented in this section. Results for other tariff increase scenarios can be found in Table A7 - 2 in Annex 7. In the medium run (ε = −0.5), households may be able to react to rising electricity prices by reducing somehow their energy consumption. The increase in electricity prices is expected to have a larger impact on household electricity expenditures in the sho rt run (ε = 0), when households are not as flexible in switching energy sources as with some other types of consumption. With larger elasticities, the substitution effect will dominate and consumers will reduce their consumption of electricity enough to offset some of the price change. 22 Figure 9. Households’ budget shares spent on energy by different tariff increase scenarios and elasticities e=0 e=-0.25 e=-0.5 7 Perccentage Points (%) 6 5 4 3 2 1 0 Tariff Increase=25% Tariff Increase=50% Tariff Increase=75% Tariff Increase=100% Source: World Bank estimates based on Kosovo 2017 HBS. Figure 10. Under the hypothetical case of a 50 percent tariff increase, the share of households’ budget going to electricity increases, particularly for the poor, households in the bottom quintile b20 not covered by social assistance, and single-elderly households Panel a. Simulated change in electricity budget share (percentage points), by Consumption Quintile e=-0.5 e=-0.25 e=0 Q5 0.7 1.7 2.6 Q4 0.8 2.0 3.2 Q3 0.9 2.1 3.4 Q2 0.9 2.3 3.8 Q1 0.9 2.3 3.7 All Households 0.83 2.07 3.31 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 % points 23 Panel b. Simulated change in electricity budget share (percentage points), by selected household groups Simulated Change in electricity budget share for selected household groups (% points) All households e=0 e=-0.25 e=-0.5 3.3 Poorest quintile and Electricity Poor 6.6 Single elderly HH 4.9 Poorest quintile - Without Social Assistance 4.0 Poorest quintile 3.7 Female-Headed HH 3.6 HH with Unemployed 55+ 3.5 HH with Disabled 3.2 HH with Child <5 3.2 Social Assistance Recipients (>700 Euros) 3.2 War-related Pension Recipient HH 3.1 Social Assistance Recipients 3.0 Poorest quintile -With Social Assistance 2.7 Extreme Poor 3.2 Poor 3.7 0 1 2 3 4 5 6 7 % points Source: World Bank estimates based on Kosovo 2017 HBS. Note: Simulation results for a cumulative 50 percent tariff increase, for a different range of elasticities. B. Simulated Impact on Incidence of Electricity Poverty Based on the benchmark case of a 50 percent tariff increases and the associated growing electricity budget shares of households, the overall incidence of electricity poverty is estimated to increase from 16 percent to 23 percent based on the baseline elasticity estimate (ε = −0.5). The impact is roughly the same in magnitude for various household categories. For instance, there is little difference in terms of impact across households that rely on different heating methods, except for households that heat with oil, for which the simulated increase in electricity poverty incidence is significantly higher (see Table A7 -2 in the Methodological Appendix).20 Among vulnerable households, those most affected are households in the bottom quintile not covered by social assistance and single-elderly households, for which the incidence of electricity poverty increases by 11.7 percentage points and 9.4 percentage points, respectively. For the other groups, the magnitude of increases in electricity poverty is closer to the national average. It is worth noting that after the rise in electricity poverty, the levels of electricity poverty in these two groups remain by far the highest (Figure 11). As we will see later, these are groups that, despite being more likely to be vulnerable to rising energy prices, are less likely to be covered by current mitigation measures. 20When agents are not able to adjust their electricity consumption after a price change (ε = 0), electricity poverty increases roughly by 30 percentage points. We can interpret this as a short-run estimate or as an upper estimate of the long-term impact on electricity poverty. 24 The impact on the incidence of electricity poverty depends on the assumed tariff increases, with larger increases exacerbating electricity poverty even further. Table A7 – 2 in Annex A7 shows the simulated increase in electricity poverty for the hypothetical tariff scenarios. Figure 11. Incidence of Electricity Poverty Rises More among the Single Elderly and households in the Bottom Quintile Not Covered by Social Assistance Impact on electricity poverty for selected groups, (% points) Simulated Baseline All 22.9 Single elderly HH 55.6 Poorest quintile - Without Social Assistance 45.0 Poorest quintile 40.7 Poorest quintile -With Social Assistance 25.8 Social Assistance Recipients 23.6 Female-Headed HH 26.2 HH with Disabled 20.5 HH with Unemployed 55+ 21.5 HH with Child <5 19.9 Social Assistance Recipients (>700 Euros) 25.4 War-related Pension Recipient HH 19.3 Poor 40.4 Extreme Poor 35.2 0 10 20 30 40 50 60 % points Source: World Bank estimates based on Kosovo 2017 HBS. Note: Simulation results for a cumulative 50 percent electricity tariff increases from 2017, assuming a medium price elasticity (ε = −0.5). C. Simulated Impact on Poverty and Inequality The simulated hypothetical tariff increases would result in a moderate increase in consumption poverty, as measured by the national and international poverty lines. To quantify the first-order poverty impact of a cumulative increase in electricity prices, following Freund and Wallich (1995),21 we adjust the welfare aggregates to account for the loss in purchasing power because households have to spend a larger share of the total expenditure on energy. Following the methodology used by Freund and Wallich (1995) and the Balancing Act (2013), the impact of the change in tariffs on household welfare (welfare loss) (can be calculated as the change in the consumer surplus as share of total household expenditures: ΔCS/E = (S0 (P1 − P0) / P0) (ε + ε(P1 − P0) / P0 +1), where CS is consumer surplus, E is total household expenditure, S0 is the initial budget share before the price change, P0 and P1 is the price before and after the tariff change, respectively, and ε is the price elasticity of demand (more details in Annex A2).22 21 Cross-price effects and income effect are assumed negligible. We analyze the change in welfare only under different assumed values of the price elasticity. For details, see Annex A2. 22 Freund and Wallich 1995, 25. 25 The impact on the poverty rate depends significantly on the ability of households to substitute electricity for other sources of energy; impacts can be sizable, particularly in the short-run. If households substitute significantly for other sources of energy (ε= -0.5), likely in the medium term the maximum impact on national poverty is about 3.7 percentage points (Figure 12, panel a). If households substitute little for other sources of energy (ε= -0.25), the maximum impact on national poverty is about 4.4 percentage points (Figure 12, panel b). If households are not able to substitute for other sources of energy (ε=0), likely in the short run, the maximum impact on poverty is about 5.3 percentage points (Figure 12, panel c). Figures below also show the results using the US$5.5 per day poverty line. The impact on the international poverty rate is not small either. Significant increases in the poverty gap are expected 23 . Surprisingly, we found small effects on consumption inequality (measured by the Gini and the Theil indexes). Figure 12. Simulated Impact on Poverty Impacts, alternative elasticity scenarios Panel a. Actual and Projected Poverty rates (% population), Medium Price Elasticity scenario (ε= -0.5) Actual and Projected Poverty Rates 26.7 27.6 30 23.9 24.6 25.7 21.1 22.9 21.1 21.1 Poverty Rate (%) 25 20.5 21.3 18.0 19.0 19.8 20 15.8 15 10 5.1 5.3 5.7 6.3 6.4 5 0 Upper Middle-Income Poverty National Extreme Poverty National Poverty Electricity Poverty ($5.5/day PPP) Baseline 2017 Simulated 2030 (Increase=25%p) Simulated 2030 (Increase=50%p) Simulated 2030 (Increase=75%p) Simulated 2030 (Increase=100%p) Panel b. Actual and Projected Poverty rates (% population), Low Price Elasticity scenario (ε= -0.25) Actual and Projected Poverty Rates 50 45.6 41.1 Poverty Rate (%) 40 34.2 27.3 28.3 25.9 30 23.9 24.6 25.9 21.9 18.0 19.1 19.9 20.7 15.8 20 10 5.1 5.3 6.0 6.3 6.6 0 Upper Middle-Income Poverty National Extreme Poverty National Poverty Electricity Poverty ($5.5/day PPP) Baseline 2017 Simulated 2030 (Increase=25%p) Simulated 2030 (Increase=50%p) Simulated 2030 (Increase=75%p) Simulated 2030 (Increase=100%p) 23The poverty gap calculated among the poor population indicates poverty shortfall, that is, it shows the extent to which the average (or consumption) of the poor falls below the poverty line. 26 Panel c. Actual and Projected Poverty rates (% population), Low Price Elasticity scenario (ε= 0) Actual and Projected Poverty Rates 80 66.6 Poverty Rate (%) 57.5 60 45.6 30.6 40 23.9 24.7 26.2 27.6 29.2 18.0 19.1 20.1 21.3 22.4 15.8 20 5.1 5.4 6.1 6.4 7.0 0 Upper Middle-Income Poverty National Extreme Poverty National Poverty Electricity Poverty ($5.5/day PPP) Baseline 2017 Simulated 2030 (Increase=25%p) Simulated 2030 (Increase=50%p) Simulated 2030 (Increase=75%p) Simulated 2030 (Increase=100%p) Source: World Bank estimates based on Kosovo 2017 HBS. Note: Simulation results for a cumulative 25,50, 75 and 100 percent electricity tariff real increase from 2017, assuming different price elasticities. %p = percentage points. Our results are more in line with other countries’ experiences in the ECA region that are engaged in energy reforms. This significant impact on the overall poverty rate is primarily explained by two factors: (a) the simulated tariff increases are relatively large compared to other countries in the region and (b) the share of electricity expenditures in the overall household budget is not small, slightly larger than in other countries such as Armenia, Albania and Georgia. Table 2. Benchmarking Kosovo with Other Countries in Europe and Central Asia Undergoing Energy Reforms Country Electricity (as % of Percentage Increase in Simulated Impact on Overall Household Electricity Tariff Poverty Rate (% point) Budget) Kosovo 6.62 25-100 0.05-5.3 Serbia 6.80 16.25 1.00 Armenia 6.0 10.6 0-0.3 Albania 5.90 15.7 0.56 Georgia 3.99 4.50 0.03 Source: World Bank estimates based on Kosovo 2017 HBS, Serbia 2013 HBS, Georgia 2015 HBS, Albania 2015 HBS, and Armenia 2016 HBS. 27 D. Simulated Household Welfare Losses: Who Loses and by How Much? Behind these aggregate numbers, there is still a hidden distributional impact. As we would expect from looking at the share of electricity consumption by various welfare groups, the welfare losses of higher electricity prices are greater for the poor households (bottom 40 percent) than for the non-poor, because the poor have larger electricity shares (Figure 13). Assuming an elasticity of demand of −0.5, the bottom quintile’s welfare declines by 3.3 percent while that of the richest quintile declines by 2.3 percent (see Figure 13). In other words, about 3.3 percent and 2.3 percent of their household budget must be given to the households for them to have the same welfare they used to have before the price change.24 When looking at different groups, the single elderly and households in the bottom quintile not covered by social assistance face the highest welfare losses (Figure 14). Figure 13. Welfare losses are larger among poorest households since they spend a larger share of their budget on electricity Initial electricity shares and simulated welfare losses of rising electricity tariffs, by quintile (as % of household budget) Q5 2.3 5.3 Q4 2.8 6.5 Q3 3.0 6.8 Q2 3.3 7.5 Q1 3.3 7.5 All households 2.9 6.6 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Electricity Shares Welfare Loss Source: World Bank estimates based on Kosovo 2017 HBS. Note: Simulation results for a cumulative 50 percent tariff increase from 2017 to 2030, and a medium price elasticity (ε= -0.5). Welfare loss is purchasing power parity losses, measured as the change in consumer surplus relative to household expenditures. 24If the bottom quintile has a lower elasticity of demand (e = 0), because they may have less ability to adjust consumption, the losses for them amount to 0.72 percent. 28 Figure 14. Initial Electricity Shares and Simulated Welfare Losses of Rising Electricity Tariffs, vulnerable groups (as % of household budget) All 2.9 6.6 Poorest quintile and Electricity Poor 5.8 13.2 Single elderly HH 4.3 9.9 Poorest quintile - Without Social Assistance 3.5 8.0 Poorest quintile 3.3 7.5 Consumption Poor (LMPL) 3.2 7.4 Consumption Poor (IPL) 2.8 6.3 Female-Headed HH 3.2 7.2 Social Assistance Recipients (>700 Euros) 2.8 6.3 War-related Pension Recipient HH 2.7 6.1 HH with Unemployed 55+ 3.1 7.1 HH with Child <5 2.8 6.3 HH with Disabled 2.8 6.4 0 2 4 6 8 10 12 14 Percent Welfare Loss Electricity Shares Source: World Bank estimates based on Kosovo 2017 HBS. Note: Simulation results for a cumulative 50 percent electricity tariff real increase from 2017 to 2030, and a medium price elasticity (ε= -0.5). Welfare loss is purchasing power parity losses, measured as the loss in consumer surplus (as a percentage of overall household expenditures) due to rising electricity tariffs. 29 4. Current Social Protection Scheme for Energy Vulnerable Consumers A. Background of Current Electricity Benefit and Eligibility Criteria Kosovo’s social assistance system is built around one main poverty targeted program – the Social Assistance Scheme / SAS. Along with the SAS, Kosovo’s social assistance system includes a limited number of other programs (Figure 15), among them an electricity subsidy 25 for consumers in need or vulnerable customers 26 which is an associated right for the recipients of SAS and certain war-related pension recipient households (excluding pensions for war veterans). Figure 15. Structure of the social assistance system of Kosovo Electricity Social Assistance Assistance in Material support subsidy Scheme / SAS case of for families with Material support - for SAS exceptional children (of age (Last resort 0-18 years) with for foster beneficiaries, needs (ad hoc or income support permanent families - for certain war- one-off social program) assistance) disability related categories The design of the electricity subsidy is determined by the Ministry of Labor and Social Welfare (MLSW). It currently provides a direct subsidy to the electricity bills of two categories of families identified on two principles: (i) based on poverty status (recipients of the SAS scheme, defined by the respective laws27), and (ii) based on merit / recognition for service during the war (recipients of four war-related pensions, namely pensions for close family members of martyrs, Kosovo Liberation Army (KLA) invalids, missing members of the KLA, and civilian victims of war, as defined by the law28). On a yearly basis the GoK, through the budget of the MLSW, allocates a budget transfer of around EUR 4.5 million to the Kosovo Electricity Supply Company (KESCO)29 for payment (fully or partially depending on whether electricity consumption is below or above the preset by KESCO threshold) of the electricity bills of the above categories of customers. 25 The terms ‘electricity subsidy’ and ‘electricity benefit’ are used in the report interchangeably for the current protection s cheme in Kosovo. 26 The terms ‘energy/electricity vulnerable consumer’ and ‘energy/electricity vulnerable customer’ are used interchangeably. 27 Law No.2003/15 on the Social Assistance Scheme and Law No.04/L-096 on Amending and Supplementing the Law No.2003/15 on the Social Assistance Scheme. 28 Law No. 04/L-054 on the Status and the Rights of the Martyrs, Invalids, Veterans, Members of Kosovo Liberation Army, Civilian Victims of War and their Families. 29 The Kosovo Energy Distribution Services (KEDS) is a joint stock company, established in 2009 and in operation since 2013, which has the exclusivity of electricity distribution in the territory of Kosovo. KESCO (Kosovo Electricity Supply Company) was established in January 2015, in line with the reforms in the energy sector calling for unbundling suppliers and energy market liberalization. KESCO serves around 500 thousand household, commercial and industrial consumers. 30 The subsidy is disbursed once a year for the preceding calendar year. MLSW sends KESCO the lists of eligible recipients and transfers the funds. The amount of the subsidy is determined per eligible family, and not per metering point, and varies depending on the amount of funding (which was stable during the past few years), the overall number of beneficiaries, their electricity consumption and the household’s distribution of daytime/nighttime30 use of electricity. Based on these parameters, KESCO determines the ceiling of the electricity consumption per family that the subsidy will cover 31 . The families whose consumption is below the ceiling get their electricity bills written off. The bills of the families with higher consumption are subsidized partially, up to the ceiling. Thus, each eligible household receives a subsidy of up to 186 kWh for the year 2018. The number of direct beneficiaries (families) of the electricity subsidy increased by 13 percent between 2015 (31,107 families) and 2017 (35,102 families). With the increase in beneficiaries, the average annual amount of the subsidy per family dropped from EUR 145 in 2015 to EUR 129.4 in 2017. Table 1 provides details on the beneficiaries and amounts transferred in 2017. Table 3. Current electricity subsidy – beneficiaries and fiscal cost, 2017 Direct beneficiaries Yearly fiscal cost, Average size of (families) 2017 subsidy per family, 2017 Total Share EUR Share Yearly Monthly (%) (%) (EUR) (EUR) SAS recipients 23,299 66.4 2,995,554 66 128.6 10.7 War-related pensioners 11,803 33.6 1,547,120 34 131.1 10.9 Total 35,102 100 4,542,674 100 129.4 10.8 Source: MLSW administrative data for 2017. Two thirds of all beneficiaries (66.4 percent) were SAS recipients . There is a notable divergence between the number of SAS recipient families - 26,302 in 201632 - and the number of families/households who actually received the electricity subsidy in 2017 – 23,299, amounting to 3,003 families/households or 11.4 percent less than eligible SAS recipients. In theory, the take-up of the electricity subsidy should be close to 100 percent because the benefit is given to SAS recipients without a separate application procedure and because KESCO pays the subsidy to each family and not to each metering device 33 . Other circumstances, such as outstanding debt to KESCO or electricity consumption above the consumption limit, do not disqualify households from receiving the electricity subsidy, and should hence not restrict the number of actual recipients. This take-up gap might be due to SAS families having incentives to install their own electricity meters or sharing their meter with other SAS-receiving households. Furthermore, lack of enforcement of electricity bill payment might disincentivize some households from claiming the subsidy as 30 To incentivize off-peak electricity consumption, households receive a lower subsidy the higher their share of daytime electricity consumption. 31 For 2018, this ceiling was determined as 186 kWh per eligible family per month. 32 The comparison of the number of electricity subsidy recipients in 2017 is with the number of SAS beneficiary households in 2016 because, according to the delivery rules, the electricity subsidy in 2017 is received by SAS eligible of the previous year. 33 Meaning that if two or more families who are eligible for the electricity subsidy live in the same building and use the same electricity metering point, they can receive a cumulative bill discount or write-off which would be higher than the maximum kWh that are granted per family but not exceeding the total consumption registered on that meter. This practice is different from most of other electricity benefit delivery models where only one subsidy is extended per holder (title) of electricity metering device; if two or more eligible direct beneficiaries share a single metering device, they will be eligible for only one subsidy. 31 they do not face a penalty for not paying their electricity bill34. Finally, anecdotal evidence suggests that in some cases the owner of the metering point (i.e. the family member on whose name the device is recorded) is not the same person as the direct SAS beneficiary. On the positive side, the current electricity subsidy has a simple design, is easy to implement, and can only be used for its core purpose - paying for electricity consumption. However, its design also has several weaknesses which could undermine its effectiveness in protecting the electricity poor, especially in the case of significant tariff adjustments. While some of these weaknesses are generic, i.e. typical for electricity subsidy schemes in general, others are program-specific, i.e. stemming from specific design characteristics and implementation rules of Kosovo’s electricity subsidy. The main advantages and disadvantages of the present electricity subsidy model are summarized in Table 4. Table 4. Advantages and disadvantages of the present electricity subsidy model Strong points / advantages Weaknesses / disadvantages Simple way of beneficiary identification – Narrow definition of electricity vulnerable – the design reliance on eligibility criteria which have been excludes certain groups of socially disadvantaged and already applied to identify SAS beneficiaries physically vulnerable individuals and households. (poverty test), and war-related pension recipients (categorical identification). Transparent identification of beneficiaries – the Imprecise definition of electricity poor – those eligible for eligible categories are distinct and have clear the subsidy are not necessarily electricity poor. rules of identification. The subsidy cannot be used for any other Exclusion of electricity poor and low coverage due to the purpose than paying for electricity low coverage of SAS (strict eligibility criteria, including consumption in the past year. specific exclusionary filters in the process of eligibility determination35). Easy to implement, no administrative Low benefit adequacy. complexity (delivered as an automatic top up to other benefit). No need for outreach and communication to Level of financial support is not linked to family or increase take up. household size and other characteristics, and is not based on an estimate of a minimum level of affordable electricity consumption depending on family size and – possibly - composition. Low costs of delivery – the calculation and The level of the subsidy is also not linked to electricity payment of the subsidy takes place once a year. tariff increases. Simplicity of planning and budgeting: due to an The subsidy is only paid in the year following the year of absence of a legal commitment to a minimum actual energy consumption. This significant time lag is amount of the subsidy or indexation with tariff most important for first time (first year) recipients. 34 According to anecdotal information, KEDS does not enforce payment of outstanding electricity bills which are not covered by the subsidy. 35 The eligible are identified with categorical criteria followed by a poverty test. Category I eligible are the families where all members are considered dependent and not able to work. Dependent are: (i) persons over 18 years of age with permanent and severe disabilities rendering them unable to work, (ii) persons 65 years of age and older, (iii) full-time caregivers of person(s)or of children under the age of five, (iv) persons up to 14 years of age, (v) persons between the age of 15 and 18 (inclusive) who are in full-time education, and (vi) single parents with at least one child under the age of 15). Category II eligible families can have one able to work member who is registered as unemployed, the rest should be dependent. In addition, an eligible family should be parenting at least one child under 5 years of age or an orphan up to the age of 15. The poverty test assigns scores based on sources of income, household composition, ownership of certain type of assets, and observable living/housing characteristics (Law No.2003/15 on the Social Assistance Scheme and Law No.04/L-096 on Amending and Supplementing the Law No.2003/15 on the Social Assistance Scheme). 32 increase, the per household allocation can be adjusted yearly based on budget availability. The compensation scheme has no built-in mechanism for monitoring the regular payment of electricity bills and mechanism for sanctioning of non-payment. Cash transfers instead of a dedicated energy subsidy is not an option in the current context. An alternative way to protect vulnerable households from the future increases in electricity tariffs is to replace the energy subsidy with a direct and unconditional cash transfer. However, such a reform is not being considered by the GoK at present. In fact, the Law on electricity requires that the household customers benefiting from financial support for payment for electricity supply services shall not be allowed to use such funds for other purposes. Further, one of the advantages of the current program is flexibility provided by the absence of a legal commitment to a minimum amount of subsidy. International experience suggests that electricity poor and vulnerable consumers should be identified among individuals and households at risk of energy poverty, but also within broader groups of consumers who may be at a disadvantage in the purchasing and use of energy in the electricity and gas retail markets36. These include socially disadvantaged (who do not have sufficient funds to pay the cost of energy / electricity) and also physically vulnerable (such as sick, disabled and elderly). The Kosovo scheme is characterized with a rigid approach to beneficiary identification. It does not allow extending the subsidy to electricity poor who fall outside one of the two pre-defined categories of SAS and war-related pensions recipients. However, in 2017, as shown before, the poorest 20 percent of households dedicated on average 7.47 percent of their household expenditures to paying for electricity, whereas the richest group (the top quintile) dedicated close to 5 percent (5.3 percent). Many of these households receive neither SAS nor war-related pensions and are thus excluded from the electricity subsidy. This also implies that any substantial adjustment in electricity prices will have a major impact on household energy expenditures. The rigid approach to beneficiary identification results in, for example, the exclusion of the group of single elderly households despite them having an incidence of electricity poverty that is 3 times higher than the average for the country (45.7 percent compared 15.8 percent). Furthermore, the design excludes households from the bottom quintile who are not SAS recipients despite high incidence of electricity poverty in this group (33.3 percent). At the same time, one third of the electricity subsidy is allocated to recipients of war- related pensions which are merit-based. The families receiving them are generally not among the most socially disadvantaged and do not appear electricity poor or vulnerable. International experience also suggests that financial support measures are complemented by other methods of protection against electricity poverty in the short- and medium run. The current electricity subsidy scheme consists only of financial support and is not complemented by other methods of protection. Figure 16 outlines these additional protection measures in line with international good practices, which are often referred to as “multi-layered approach’’ to protection against electricity poverty. For example, a number of EU Member States and Western Balkan countries, explicitly protect vulnerable consumers 36 See for example: INSIGHT_E, 2015 report, which makes a point that that broader population groups may be at a disadvantage in the purchase and use of electricity and gas. 33 against disconnection in two particular cases - during winter (low temperature) periods37 or when the life or health of electricity consumers would be endangered in case of disconnection38. Figure 16. Main types of short-term and medium-term protection measures Financial support Additional consumer protection Instrument for short-term protection. Most often, Most often this takes the form of protection against social assistance systems are used for identification disconnection from the electricity grid, and applies of recipients to reduce the administrative to electricity vulnerable whose health or life would complexity of identification of the energy-poor. be endangered in the absence of access to The payment can take place either through the electricity. Such measures are primarily welfare system or through the direct transfer of coordinated by the regulators and energy supply public resources to the suppliers of electricity / companies. energy to cover the bill or part of it. Energy efficiency measures These measures are key parts of strategies for Information provision, outreach / combating energy poverty in the medium and long communication and awareness raising term through reducing the consumption needs of These include measures related to price poor households. They vary across countries in comparison and transparent billing. Such measures terms of funding sources, extent of targeting, can be found mostly in countries with liberalized implementing institutions, etc. There is energy markets. Where there is a strong civil considerable scope for improving the targeting of society, the scope of awareness raising and such measures based on a good understanding of information provision is wider. Greater awareness the energy poor. Best practices include putting in of energy poverty and how to tackle it can also place incentives for take-up by low-income come through the wider use of smart metering. households. The proposed reform concept for the electricity subsidy and its proposed / simulated parameters are in line with the multi-layered approach to protection against electricity poverty which involves as first layer, protection within the last-resort income support program where eligibility thresholds adequately reflect the share of cost of utilities as part of cost of a minimum consumption basket. In this context, the clear indexation rules for the SAS benefit are recommended. The main problem here is that the expected cost of electricity cannot be factored out as part of a minimum consumer basket, given the uncertainty of the tariff adjustment scenarios. follow up discussion on possible adjustments to the minimum consumption basket, which adequately reflects the share of utilities in household consumption. A second layer of support - support embedded in the energy sector itself – is also considered in the suggested reform options by proposing a smoothing mechanism for spreading the bill payment burden throughout the year (instead of the current one-off annual payment), thus minimizing delays in subsidy transfers, as well as subsidizing energy efficiency programs. The third 37 For example, in Finland, Spain and Greece, those who are disconnected during the winter due to lack of payment must be reconnected. In Spain this protection is available for the whole year, not only winter period, but only for extremely electricity vulnerable. In Belgium there are a number of steps that need to be taken before a household is disconnected, which include the account being taken from a commercial supplier to the distribution system operator and the installation of a budget meter. 38 For example, customers who are dependent on the use of medical devices and equipment which is connected to the electricity grid. Such provisions exist in other countries in the Western Balkans and beyond, for example Bosnia and Herzegovina, Serbia, Slovenia. 34 layer - temporarily subsidizing within the tariff structures – is where Kosovo’s energy subsidy falls in (separate targeting of energy poverty). Finally, a fourth layer - protection against disconnect and mechanisms for electricity debt relief – is also discussed and policies and measures to that effect feature in the recommendations. B. Simulations of the Current Electricity Benefit Scheme This section analyzes the coverage of the electricity subsidy - overall and for vulnerable populations – based on simulating the design features of the electricity subsidy using the Household Budget Survey (HBS). The 2017 HBS does not directly ask respondents whether they receive the electricity subsidy. However, the survey asks households about whether they receive other benefits – most notably the social assistance scheme. As eligibility for the electricity subsidy depends on receiving either SAS benefits or receiving a war-related pension (excluding war veterans themselves), we use these two benefits as a starting point for estimating electricity subsidy eligibility. The survey slightly underestimates the number of beneficiaries for SAS and significantly underestimates the beneficiaries of war-related pensions39 To simulate additional SAS beneficiaries, we use a propensity score matching method (PSM). Following the methodology by Souza, Osorio, and Soares (2011)40, the imputation of beneficiaries results in 95.5 percent of the beneficiary households recorded in administrative data to be identified in the survey, and a total 7.4 percent of all households in Kosovo as beneficiaries of this social assistance program in the survey.41 Using the simulated beneficiaries for SAS and the households identified in the survey as beneficiaries of war-related pension, we then assign the electricity subsidy to all households who receive one of these benefits. We assume that households are eligible for the electricity if they either (i) receive poverty 39 In the HBS 2017 it is only possible to identify households who state that their main source of income is a war-related pension, i.e., those who received recognition for service during the war (recipients of four war-related pensions, namely pensions for close family members of martyrs, Kosovo Liberation Army (KLA) invalids, missing members of the KLA, and civilian victims of war). 40 This method is commonly used to adjust for underestimation of beneficiaries of a particular program. The method consists in imputing beneficiaries that did not report to be recipients of the SAS program in the survey, but are “likely beneficiaries”. First, we run a Probit model of program participation against household per capita consumption, possession of various household assets and consumer durables, number of children and other sociodemographic variables. Then, we randomly sampled households out of the beneficiary households and match these beneficiary households to non-beneficiary households with the closest propensity scores. Program benefits of SAS are then imputed to the matched households, where the amount of benefit imputed is equal to the amount received (reported in the survey) by the household’s matched beneficiary household. The idea is to match the number of beneficiaries in the survey to the national accounts as closely as possible. However, in our case, we do not match exactly the number of beneficiaries in administrative data due to some restrictions with the method: namely, the number of beneficiaries who need to be randomly sampled and matched to non-reporters has to be lower than the number of beneficiary households. The original number of recipient SAS households in the survey was 23,299. Applying the method, we simulate around 22,238 beneficiaries, still less than the 1,061 recipients in administrative data (which represent about 4.6 percent of the SAS recipients). 41 To identify the beneficiaries of war-related pensions, we use a survey question about the household’s main source of income (rather than a question asking the respondent directly whether the household receives a war-related pension). This method results in about 34 percent of the number of beneficiaries according to administrative records. To identify the beneficiaries of war-related pensions, we use a survey question about the household’s main source of income (rather than a question asking the respondent directly whether the household receives a war-related pension). This method results in about 34 percent of the number of beneficiaries according to administrative records. Due to a lack of information regarding the characteristics of war-related pension beneficiary households we decided against performing Propensity Score Matching (PSM) for additional beneficiary identification. 35 targeted cash transfer from the SAS42 or (ii) receive war-related pensions as main source of income43 . Each group receives approximately 10.7 euros and 10.9 euros per month, respectively. The size of the subsidy does not depend on household size. We assume that all of those who are identified as eligible for the subsidy in the survey, i.e. perfect take-up. This is a reasonable assumption given that households that meet the eligibility conditions should be automatically enrolled to the subsidy. Overall, we estimate 22,238 households receiving the energy subsidy – 7.6 percent of all households in Kosovo. This number is reasonably close but slightly underestimates the number of beneficiaries of the electricity subsidy according to administrative records (Table 5). Table 6 compares the estimate for the budget envelope of the energy subsidy based on survey data with administrative budget data. Again, while slightly underestimating the overall budget envelope our simulations are within reasonable margins of error of actual spending data. Table 5. Beneficiaries of Electricity Benefits, by Category– Data vs Estimate Estimate based on HBS Administrative Data SAS 22,238 23,299 War-related pension recipient 2,232 11,803 household Source: World Bank estimates based on 2017 HBS and MLSW. Table 6. Budget allocated to Energy Program – Data vs Estimate Estimate based on Administrative Data HBS Total Budget (Monthly) for energy SAS Euro 238,320 Euro 249,299 subsidy beneficiaries War-related pension Euro 24,379 Euro 128,927 recipient household Total Budget (Annual) for energy SAS Euro 2,859,171 Euro 2,995,554 subsidy beneficiaries War-related pension Euro 292,504 Euro 1,547,120 recipient household Source: World Bank estimates based on Kosovo 2017 HBS and MLSW. Across the welfare distribution, simulated coverage of the electricity benefit among households in the bottom quintile is low, but much higher than in other quintiles. Approximately 22.8 percent of the poorest households are covered by the benefit44, compared to 3.1 percent among the top quintile (Figure 17). Coverage of certain vulnerable groups, such as single elderly households and energy/electricity poor households, is low (Figure 18). 42 Since the minimum SAS that a (one-person) household can receive is 60 Euros a month or 720 Euros a year, we include only households who indicate receiving more than 700 Euros from social welfare benefits as SAS recipients. 43 For households that fulfill both criteria in the survey, we assume they are beneficiaries under the SAS scheme. This is to avoid double counting or assigning them two electricity subsidies given that they fulfill both criteria. 44 Results based on HBS 2017 show that between 20 and 22.3 percent of poorest households in Kosovo are covered by SAS (with and without PSM approach). 36 Figure 17. Coverage of Electricity Benefit by Quintile households in quintile 25.0 22.8 Percentgae of all 20.0 15.0 9.8 10.0 4.8 5.0 3.0 3.1 0.0 Q1 Q2 Q3 Q4 Q5 Source: World Bank simulations based on Kosovo HBS 2017. Figure 18. Coverage of Electricity Benefit for vulnerable groups Single elderly HH 1.4 45.7 Extreme Poor 32.5 42.0 Poor 25.1 31.0 Female-Headed HH 20.1 21.2 HH with Unemployed 55+ 10.4 16.6 HH with Child <5 10.313.0 HH with Disabled members 12.4 49.1 Energy Poor 13.3 Electricity Poor 15.6 Bottom20 Electricity Poor 20.0 War-related pension recipient households 14.0 Social Assistance recipient households 20.1 0 10 20 30 40 50 60 incidence of electricity poverty Percentage covered by electricity benefit Source: World Bank simulations based on Kosovo HBS 2017. According to the simulation, the electricity benefit seems to be moderately targeted and progressive, leakage is relatively high, however. We estimate that households in the bottom quintile receive about 49.8 percent of the benefits provided and 73 percent of benefits go to the bottom 40 percent (Figure 19). Leakage seems to be relatively high, given that the households in the top quintile receive about 10 percent of the benefits. Considering the fact that one third of the energy subsidy is merit-based for war-related pension recipient households, the targeting performance is rather good compared to other countries in the ECA region45. In regional comparison, the overall share of energy/electricity-poor households covered by the energy benefit in Kosovo (about 18 percent of the energy/electricity poor in the bottom quintile) is only slightly below the coverage of bigger subsidy programs of Moldova and Ukraine. 45See “Balancing Act” for other estimates for the ECA region. For instance, in Albania, 36.6 percent of the benefits go to the top quintile and in Moldova more than two thirds of spending goes to the non-poor. Bulgaria, on the other hand, allocates a larger share of the energy benefit to the poor. 37 Figure 19. Targeting: Share of energy transfer going to each consumption quintile Q5 10.0% Q4 6.5% Q3 10.5% Q2 23.2% Q1 49.8% 0% 10% 20% 30% 40% 50% 60% Source: World Bank simulations based on Kosovo HBS 2017 Adequacy of the electricity benefit in Kosovo seems comparable to other countries in the ECA region, but is at the lower end. Electricity benefits typically contribute from 5 to 10 percent of total expenditures among poor households 46 , and for Kosovo we estimate that they contribute to 5.5 percent of total expenditures for the beneficiary households at the bottom quintile of the welfare distribution. Figure 20: Adequacy: Share of Monthly Household Expenditure constituted by Electricity Benefit, by Quintile 6.0% 5.5% 5.0% 3.8% 4.0% 3.0% 2.4% 2.3% 2.0% 1.6% 1.0% 0.0% Q1 Q2 Q3 Q4 Q5 Source: World Bank simulations based on Kosovo HBS 2017 The simulations show that the electricity benefit leaves out certain groups prone to electricity-poverty and will not be able sufficiently mitigate the impacts of potential tariff reforms. Because the electricity benefit is granted primarily to SAS beneficiaries it is a good instrument to target poor households. This is also reflected in the high share of benefits accruing to poor households. However, the energy subsidy excludes certain groups at high risk of electricity poverty, such as single elderly households. Furthermore, if energy tariffs were to increase, the limited size and extent of the subsidy would obviate its mitigation impact. The next section will discuss reforms to the electricity benefit that would strengthen its mitigation potential. 46According to the “Balancing Act” this share varies between 4 and 8 percent for Poland, Hungary and Bulgaria, while Tu rkey provides a more generous allowance (close to 14 percent) 38 5. Reforming the Social Protection Program for Energy Vulnerable Customers A. Short-term: Redesigning the electricity benefit This section presents options for reforms to the energy benefit that can be implemented in the short term.47 The options presented in this section can feasibly be introduced without a major overhaul of the social assistance system currently in place. The recommendations to improve the design and efficiency of the energy subsidy scheme in the short run are the following: • Introduce a scaling mechanism for the electricity subsidy so it increases with household size and reflects better the poverty and electricity poverty status of larger households. The current energy subsidy design does not adjust the amount of the subsidy based on household size. Thus, households with many members are inherently disadvantaged by the current subsidy design as they have higher electricity consumption needs. Tying the energy benefit to household size would account for the larger energy needs of bigger households. 47Additional simulations that model the impact of introducing consumption thresholds to exclude high-electricity consuming households form the subsidy are presented in Annex A6. 39 Table 7 presents four adjusted benefit sizes that depend on the size of the beneficiary household48. The current budget size is 186 kWh for all beneficiary households. Option 1 gives a monthly bill deduction of 169 kWh to 1-3 member households. The subsidy would then increase to a maximum of 192 kWh per month for households with 4 or more members. Overall, option 1 maintains a neutral budget49 but, as a result, the nominal amounts of consumption proposed for subsidizing in the simulations are lower than the current actual one for households with less than 4 members and higher for households larger than 3. Normative amounts have not been calculated for Kosovo; however, the amounts in the simulations in option 1 are higher than the normative assessment of electricity consumption basic needs in other countries in the region (e.g. Bulgaria and Serbia50). Option 2 gives a monthly bill deduction of 229 kWh to 1-3 member households and increases to a bill deduction of 274 kWh for households with 4 or more members. Option 2 would entail an increase the overall budget of the electricity subsidy by about 42 percent, or about €1.94 million per annum and has been added to the simulations upon request of the MLSW. Option 3 assumes that beneficiaries of the four war-related family pensions (for “close family members of martyrs, missing of the KLA, KLA invalids and civilian victims of war”) will lose access to the subsidy. The resulting savings are reallocated to increase the subsidy size for the remaining beneficiaries, with 1-3 member households receiving 250 kWh, 4-5 member households receiving 275 kWh, and 6 and more member households receiving 300 kWh. 48 To do so, the adjustment infers the household consumption of beneficiary households from the HBS 2017. 49 In the simulations (Table 7, option 1), the different levels of consumption and presumed economies of scale are based on the patterns of electricity consumption across households of different size, identified with HBS data (calculations are based on the 25th percentile of electricity consumption, divided by household size and further divided it by half to get the subsidy amount). 50 Normative assessments in these countries arrive at a minimum consumption requirement of 120 kWh per month for 1-person households, excluding heating. 40 Table 7: Adjusted bill deduction size Option 1: Budget Option 2: Budget Increase Option 3: Budget Neutral by Neutral Monthly Bill Monthly Bill Deduction Reallocation from War-Related Deduction (in kWh) (in kWh) Categories Monthly Bill Deduction (in kWh) HH with 1 Member 169 229 250 HH with 2 or 3 Member 169 229 250 HH with 4 or 5 Member 192 274 275 HH with 6 or more Member 192 274 300 • Disburse the subsidy monthly instead of yearly to assist poor households in their consumption smoothing efforts. Poor households face particular constraints as they often lack the ability to save, even in the short term, and thus cannot easily smooth their consumption. Having benefits paid monthly increases the ability of poor households to cope with utility payments. • Link the payment of the subsidy to the payment of the household’s remaining bill and disconnect non-paying household to enforce compliance with bill paying. Under KESCO’s current policy, subsidy-eligible households are not disconnected if they fail to pay their bills. This disincentivizes households from paying their bills as they face no potential penalty. Furthermore, it amounts to a de facto 100 percent discount of the electricity bill for some household who might not be in need of such a large subsidy. A communication campaign and support to vulnerable customers to manage payment of past bills would be needed in parallel. • Enforce one meter per household. Each eligible beneficiary household should have their own account and electricity metering device. Meter-sharing between households not only complicates correct attribution of the subsidy but also makes it impossible to enforce individual households’ bill payments. • Based on a need’s assessment, introduce more stringent eligibility criteria for beneficiaries of the four types of war-related family pensions (for “close family members of martyrs, missing of the KLA, KLA invalids and civilian victims of war”) to receive the electricity subsidy. There are currently no indications that beneficiaries of these pensions are at particular risk of energy and electricity poverty or belong to the most vulnerable households in Kosovo. Thus, the government should either eliminate their automatic eligibility for the electricity benefit or employ a poverty test similar to the one used for the SAS. Single elderly households feature prominently in the analysis with their high risk of energy and electricity poverty but at the same time single elderly face a lower poverty rate compared to other groups. On one hand, there is no doubt that reaching the poor single elderly with electricity subsidy is desirable and would be important for mitigating their high energy poverty risk. On the other hand, reaching out to the poor single-elderly households would require additional targeting efforts, outside the ‘’mainstream’’ targeting system and the proposed reformed poverty targeting. There are important pros and cons to consider before deciding whether single elderly households should be added to the list of energy subsidy recipients subject to consumption thresholds and some kind of poverty test. 41 Expanding the benefit to single elderly households would target households at disproportionate risk of energy poverty. Household survey data shows that single elderly households have a particularly elevated risk of energy and electricity poverty (see Figure 7, Panel b above). Thus, granting the electricity benefit to single-elderly households would be in principle a way to alleviate electricity poverty. Simulations show that this benefit expansion would cost approximately 260,000 Euros a year. However, there are numerous tradeoffs between including and not including single elderly households as recipients of electricity subsidy, and the decision is quite complex. Identifying the electricity poor with a poverty test would not work because the category’s poverty rate is low. Simulations indicate that the poverty test when applied to this category will reject them as ineligible. Moreover, the coverage of the poorest two quintiles will increase only marginally if the single elderly households are included. Reaching out to the poor single-elderly households would require additional targeting efforts, outside the ‘’mainstream’’ targeting system and the proposed reformed poverty targeting, and additional administrative costs associated with the effort to identify the small group of electricity poor single elderly. B. Medium-term: Reforming the electricity benefit in the context of SAS Reform In the medium term, reform of the electricity benefit should be coupled with reform of the SAS. Currently, the SAS is Kosovo’s main poverty-alleviating mechanism and its reform is already on the government agenda. Using the potential synergies arising from this reform would allow the government to also tackle the effects of energy tariff reform. An advanced reform concept for the SAS is already in place to remedy deficiencies in its performance, design and implementation characteristics, and to align its beneficiary identification mechanism with emerging needs of identification of beneficiaries of new benefits, such as the already adopted exemptions from payment of health insurance contributions and from patients’ copayments51, or new benefits such as monthly child benefit52, social pension and reformed energy or electricity benefit. In line with the planned SAS reform, the medium-term refers to the time when the reformed SAS will become operational, expected for the second half of 2021. The reform concept will change the approach to poverty targeting of different benefits, including a common targeting mechanism for all poverty targeted programs. In the new scheme, SAS eligibility would be determined through a means test (examining a household's formal income and eliminating every household whose income exceeds a certain threshold from eligibility), followed by a Proxy Means Test (PMT). The PMT consists of a scoring system that evaluates the living conditions and assets of a household and assigns a corresponding score.53 Each household below a threshold score is granted SAS eligibility. The new poverty test for the SAS will be the starting point for harmonization of targeting methods and instruments across existing and new programs. The PMT’s scoring formula can be used to determine different eligibility thresholds to identify recipients for a variety of benefits. Such a new scoring 51 The exemption from health insurance contributions (premiums and co-payments) based on poverty status for around ten population groups is regulated with the Health Insurance Law of 2014. 52 The child benefit is mentioned in a resolution of the Parliament of Kosovo of November 7, 2018. 53 A draft Operations Manual is already prepared with an outline of procedures, responsible institutions, submission of documentation and data exchange protocols. 42 system would lead to equity and homogeneity in beneficiary identification and also reduce the high administrative cost of using several beneficiary identification methods in parallel. In the context of this reform, the reform of the electricity benefit would entail the following: • Have the electricity (and potentially energy) subsidy accrue to SAS beneficiary households as well as to households whose PMT score just exceeds the threshold for SAS eligibility. The exact cutoff PMT score for energy subsidy eligibility will be determined based on available budget. • The energy poverty of beneficiaries of war-related benefits that record war-related benefits as their main source of income in the HBS 2017 is at the average level (about 25 percent, see Figure 8). The reform could subject these categories to the proxy means test as well and disqualify from electricity / energy subsidy eligibility those who are beyond the chosen cutoff point. Reforming the scheme along these lines would allow to better target the electricity subsidy to vulnerable customers. The simulations are based on the HBS 2017 and do not include the subsidy for war veterans as these are not easily identifiable in the survey. Figure 21 shows the coverage of both electricity poor (Panel a) and electricity and consumption-poor households (Panel b) with different reform options: (i) reforming the SAS scheme and giving the electricity subsidy to all new SAS beneficiaries while maintaining the current budget, (ii) adding an additional Euro 10 million per annum to the SAS budget and giving the electricity subsidy presented in section 6a (size of the subsidy depends on household size) to all SAS beneficiaries, (iii) same as (ii) but in addition to that granting the electricity benefit to those households whose PMT score is just above the SAS cutoff value until a total of 200,000 individuals receive the electricity benefit, (iv) same as (iii) but also giving the benefit to all single-elderly households whose annual pension income is below 1.25 times the poverty line54. Figure 21: Coverage of households with reformed electricity subsidy Panel a. Coverage of all electricity poor households 25 Percentage of households 20.83 19.3 20 15 12.28 8.97 9.35 10 5 0 current SAS reformed SAS reformed SAS plus 200k beneficiaries 200k beneficiaries 10 million for elec. subsidy for elec. Subsidy + single elderly hh 54 This amounts to an income below Euro 1000 per year. 43 Panel b. Coverage of electricity poor and poor households Percentage of households 60 49.2 49.55 50 40 34.5 30 26.23 20.6 20 10 0 current SAS reformed SAS reformed SAS plus 200k beneficiaries 200k beneficiaries 10 million for elec. subsidy for elec. Subsidy + single elderly hh Source: World Bank simulations based on Kosovo HBS 2017. Reforming the electricity subsidy would also be economical. Reforming and expanding the subsidy to 35,000 households (approximately 200,000 beneficiaries) would cost approximately Euro 5.1 million (the current electricity subsidy budget is about Euro 4.5 million). Figure 22, Panel a gives the cost of the reformed subsidy scheme. Covering all households who receive SAS (but excluding the households receiving war-related pensions) under a reformed scheme would cost approximately Euro 2.8 million (as compared with the current budget of Euro 4.5 million). If the SAS budget were to increase by Euro 10 million per year for additional beneficiaries and all of those beneficiaries would also be covered by the electricity subsidy, the cost would be about Euro 3.8 million. Lastly, increasing the energy subsidy to reach about 200 thousand individuals, or 35,000 households, would cost about Euro 5.1 million per year (i.e., an increase of approximately Euro 600,000 over the current budget of Euro 4.5 million per year). Expanding the subsidy in addition to single-elderly households would add another Euro 260,000 to the total annual cost. Figure 22: Cost and overall coverage of reformed subsidy Panel a. Cost of reformed subsidy scheme 6,000,000 5,353,406 5,094,247 5,000,000 Euros per year 3,762,319 4,000,000 2,741,226 3,000,000 2,000,000 1,000,000 - reformed SAS reformed SAS plus 10 200k beneficiaries for 200k beneficiaries for million elec. subsidy elec. Subsidy + single elderly HH 44 Panel b. Coverage of reformed SAS and subsidy schemes (households) 40,000 38,056 35,082 Number of households 35,000 30,000 25,757 25,000 covered 18,858 20,000 15,000 10,000 5,000 0 reformed SAS reformed SAS plus 10 200k beneficiaries for 200k beneficiaries for million elec. subsidy elec. Subsidy + single elderly HH Source: World Bank simulations based on Kosovo HBS 2017. Reforming SAS will lead to an increased coverage of households which are both poor and electricity poor. Figure 25 provides a breakdown of electricity poor households. As shown in Figure 23, Panel a, 30 percent of electricity poor household are poor, and of these 7.9 percent would be covered through the reformed SAS and therewith receive the electricity subsidy. Put differently, the reformed SAS would cover 7.9 percent of all electricity poor households or about 26 percent of households which are both poor and electricity poor. Increasing the SAS budget by 10 million (Figure 23, Panel b), would lead to a coverage of 10.8 percent of all electricity poor households or about 34 percent of households which are both poor and electricity poor. Increasing the electricity subsidy to reach about 200,000 individuals, or 35,000 households would lead to a coverage of 19.3 percent of all electricity poor households or about 49 percent of households which are both poor and electricity poor. Furthermore, the SAS’ reformed targeting mechanism can be used to scale up the electricity benefit without additional administrative effort, thus allowing for an easy way to increase coverage of poor households. Figure 23: Coverage of current and reformed electricity benefit, Selected groups a. Breakdown of electricity poor households b. ... adding €10 million for reformed SAS 7.9 10.8 22.2 20.3 57.1 55.9 5.1 5.2 7.6 7.6 0.2 0.2 Other war-related hh Other war-related hh female-headed hh single elderly female-headed hh single elderly poor hh reformed SAS poor hh reformed SAS Source: World Bank estimates based on 2017 Household Budget Survey 45 Reforming SAS would also lead to better targeting of poor households of the benefit and higher benefit adequacy. Figure 24, Panel a shows the targeting performance of the electricity subsidy allocated through the reformed SAS targeting mechanism. While about 67 percent of current beneficiaries receiving the electricity subsidy through SAS eligibility are poor, this share would increase to 85 percent under a reformed SAS. Expanding the benefit through an expansion of the SAS scheme or by adding additional electricity subsidy beneficiaries using the SAS PMT scores would slightly decrease the targeting accuracy. However, under all simulated scenarios, targeting accuracy is higher under the reformed schemes compared to the current scenarios. Figure 24, Panel b shows the adequacy of the electricity subsidy. Reforming the electricity subsidy through a reformed SAS would increase the benefit adequacy from currently 4.7 percent among SAS beneficiaries to 5.3 percent of beneficiary household consumption. Figure 24. Targeting and adequacy of the reformed electricity subsidy Panel a Targeting of poor households 90 85.1 80.23 80 72.6 72.3 67.43 70 60 50 40 30 20 10 0 current SAS reformed SAS reformed SAS 200k 200k plus 10 million beneficiaries for beneficiaries for elec. subsidy elec. Subsidy + single elderly hh Panel b Adequacy of the subsidy 55 6.0% 5.5% 5.3% 5.2% 5.0% 5.0% 5.0% 4.7% 4.5% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% current SAS reformed SAS reformed SAS plus 10 200k beneficiaries for 200k beneficiaries for million elec. subsidy elec. Subsidy + single elderly hh Source: World Bank estimates based on 2017 Household Budget Survey 55 In this simulation the benefit size adjusted by household size (Table 7, option 1) is used. 46 6. Conclusions and Policy Recommendations Kosovo has planned significant reforms in the energy sector that may bring significant electricity tariff increases for consumers. Two aging coal-fired power plants continue to dominate domestic power generation. The government’s energy strategy includes significant investments in new power generation as a replacement for the closure of one of these plants and the upgrading of the other. These changes are expected to impact electricity prices. This study simulates the poverty and distributional impact of four hypothetical scenario of real tariff increases (25, 50, 75, and 100 percent), analyzes the current energy benefit in place to protect vulnerable customers, and proposes reform options. Energy represents a substantial share of household consumption in Kosovo and a third of poor households spend more than 10 percent of their budget on electricity. Overall, energy expenditures account for 8.2 percent of overall household expenditures, and electricity accounts for 6.6 percent of all household expenditures. Energy poverty, defined as spending more than 10 percent of the overall budget on energy, affects about one third of poor households. Furthermore, a staggering over 57 percent of single elderly households are affected by energy poverty, and close to 46 percent are affected by electricity poverty. The simulations show that under the hypothetical tariff increases, the poverty impacts of energy reform would be substantial and require concerted mitigation efforts. Under the most extreme scenario of an electricity tariff increase by 100 percent, national poverty would increase by about 3.3. percentage points. Even if tariffs were to increase by a lower amount, the impact on poverty would still be substantial. Furthermore, the budget shares the poorest households devote to electricity would increase by between 2 and 6 percentage points. The current electricity subsidy has a simple design, is easy to implement, and can only be used for its core purpose - paying for electricity consumption. However, its design also has several weaknesses which could undermine its effectiveness in protecting the electricity poor, especially in the case of significant tariff adjustments. While some of these weaknesses are generic, i.e. typical for electricity subsidy schemes in general, others are program-specific, i.e. stemming from specific design characteristics and implementation rules of Kosovo’s electricity subsidy. There is room for improving the current subsidy and the GoK should consider reforming the subsidy in the context of the ongoing SAS reform in order to cushion the anticipated impact of energy reform especially on poor households. This section outlines policy recommendations, both in the short run and in the medium run, to increase energy efficiency and diversify energy sources, improve the design and efficiency of the energy subsidy scheme, and help mitigate the effects of energy tariff reform. • The recommendations to increase energy efficiency and diversify energy sources are the following: The government needs to consider sectoral policies to allow households to diversify energy sources and increase their energy efficiency in order to adapt to higher tariffs. An energy efficiency program targeting energy poor households could be a cost-effective subsidy depending on the households’ energy consumption and condition of their homes. Energy poor households with higher electricity consumption, particularly those using electricity for heating, have potential for higher energy savings through building retrofits (building envelope insulation, 47 windows, lighting etc.) and heating system replacement. Energy efficiency measures should also consider cost effective installation of rooftop solar (photovoltaic or water heating). • In terms of policy reforms, the current Feed-in Tariff support scheme for renewable investments needs to be replaced with a market-based support scheme, such as auctions. A market-based support scheme promotes cost competition for procuring renewable energy based on latest market prices and competition. • To diversify energy sources, the Government should take immediate steps to expand district heating in more neighborhoods in and around Pristina and develop natural gas interconnections with neighboring countries. Additional capacity from thermal power plants can be utilized for more efficient supply of heating through the district heating network in multifamily apartment buildings. Since the country still does not have the natural gas pipeline infrastructure to access regional gas networks, GoK should determine the most feasible option to access most diversified gas supply infrastructures in the region. In 2019, there was an initiative to study the feasibility of connecting Kosovo to the North Macedonian natural gas infrastructure, since North Macedonia plans to diversify its natural gas supply through an interconnector with Greece accessing the Southern Gas Corridor. The recommendations to improve the design and efficiency of the energy subsidy scheme in the short run are the following: • Introduce a scaling mechanism for the electricity subsidy so it increases with household size and reflects better the poverty and electricity poverty status of larger households. The current energy subsidy design does not adjust the amount of the subsidy based on household size. Thus, households with many members are inherently disadvantaged by the current subsidy design as they have higher electricity consumption needs. • Disburse the subsidy monthly instead of yearly to assist poor households in their consumption smoothing efforts. Poor households face particular constraints as they often lack the ability to save, even in the short term, and thus cannot easily smooth their consumption. Having benefits paid monthly increases the ability of poor households to cope with utility payments. • Link the payment of the subsidy to the payment of the household’s remaining bill and disconnect non-paying household to enforce compliance with bill paying. Under KESCO’s current policy, subsidy-eligible households are not disconnected if they fail to pay their bills. This disincentivizes households from paying their bills as they face no potential penalty. Furthermore, it amounts to a de facto 100 percent discount of the electricity bill for some household which might not be in need of such a large subsidy. A communication campaign and support to vulnerable customers to manage payment of past bills would be needed in parallel. • Enforce one meter per household. Each eligible beneficiary household should have their own account and electricity metering device. Meter-sharing between households not only complicates 48 correct attribution of the subsidy but also makes it impossible to enforce individual households’ bill payments. • Based on a need’s assessment, introduce more stringent eligibility criteria for beneficiaries of the four types of war-related family pensions (for “close family members of martyrs, missing of the KLA, KLA invalids and civilian victims of war”) to receive the energy subsidy. There are currently no indications that beneficiaries of these pensions are at particular risk of energy poverty or belong to the most vulnerable households in Kosovo. Thus, the government should either eliminate their automatic eligibility for the electricity benefit or employ a poverty-test similar to the one used for the SAS. • Carefully weight pros and cons when consider adding single elderly households to the list of energy subsidy recipients, subject to consumption thresholds and a means-test. More than half of all single-elderly households in Kosovo are electricity poor, making them an easily identifiable group for expansion of the energy subsidy scheme. However, in order to exclude those single elderly households not in need of the subsidy, eligibility should be tied to receiving pension earnings below a certain threshold (a reasonable threshold would be 1.25 times the poverty line or about Euro 1000 a year which would still include about 60 percent of all single elderly households.)56 However, there are numerous tradeoffs between including and not including single elderly households as recipients of electricity subsidy, and the decision is quite complex. Identifying the electricity poor with a poverty test would not work because the category’s poverty rate is low. Simulations indicate that the poverty test when applied to this category will reject them as ineligible. Moreover, the coverage of the poorest two quintiles will increase only marginally if the single elderly households are included. Reaching out to the poor single-elderly households would require additional targeting efforts, outside the “mainstream” targeting system and the proposed reformed poverty targeting, and additional administrative costs associated with the effort to identify the small group of electricity poor single elderly. • Implement the provisions of the Law on Electricity with respect to developing a program for protection of vulnerable consumers, a national definition of vulnerable consumer / customer in need, and secondary legislation that clarify the provisions of the Law on Electricity. A national framework should harmonize government funded support, customer funded assistance and market developments and clearly define the goals of each party involved in the support of vulnerable customers. A detailed description of this recommendation can be found in Annex 3. The recommendations to improve the design and efficiency of the energy subsidy scheme in the medium run are the following: • Reform the SAS along the lines outlined in the Government’s concept document (new means test, new PMT, removing categorical filters) to improve the targeting of the energy subsidy. A reform of the SAS would not only contribute to reducing poverty overall and remove work disincentives built into its current design but also improve the efficacy of the energy subsidy 56 We do not recommend using a PMT on single elderly households as simulations indicate that such a procedure would exclude most of them from electricity benefit eligibility. 49 scheme. Due to the improved poverty-targeting of the reformed scheme, targeting of vulnerable energy-poor households would also increase. • Use the new SAS PMT score as a targeting mechanism for the energy subsidy for households whose score just exceeds the SAS eligibility threshold. In addition to reform of the SAS benefit, the new SAS targeting mechanism should also be used as a targeting mechanism for the energy subsidy. Households who just do not qualify for the SAS because of their PMT score exceeding the scheme’s threshold should be nevertheless made eligible for the energy subsidy conditional on budget availability. This would target the subsidy to vulnerable households without imposing additional cost on the government for beneficiary identification. In addition to that, this targeting mechanism provides a flexible tool to adjust the overall envelope of the subsidy based on evolving energy tariffs. 50 References Barnes, Douglas, Shahidur Khandker, and Hussain Samad. 2010. “Energy Access, Efficiency, and Poverty: How Many households Are Energy Poor in Bangladesh?” Policy Research Working Paper 5332, Washington, DC: World Bank. Bowen, Brian H, James A. Myers, Arzana Myderrizi, Blendi Hasaj and Blerina Halili. 2013. “Kosovo Household Energy Consumption Facts and Figures” Policy Report EDRC (Economic Development and Research Center). 2016. “Public Perceptions of Energy Sector Reform in Kosovo.” Freund, Caroline, and Christine Wallich. 1995. “Raising Household Energy Prices in Poland: Who Gains? Who Loses?” Policy Research Working Paper 1495, Washington, DC: World Bank. INSIGHT_E. 2015. Energy poverty and vulnerable consumers in the energy sector across the EU: analysis of policies and measures. Policy Report, May 205. IPSOS. 2010. “Qualitative Assessment of Vulnerable households – Energy Use and Patterns and Strategies to Cope with Tariff Increases.” Report commissioned by the World Bank. Trang, Nguyen, and Monica Robayo-Abril. 2017. “Energy Affordability and Impacts of Tariff Reforms on household welfare in Serbia.” Mimeo, Washington, DC: The World Bank Group. Velody, Mark. 2003. “A Regional Review of Social Safety Net Approaches in Support of Energy Sector Reform: A Synthesis Report.” United States Agency for International Development. World Bank. 2009. “Impact of Electricity Tariff Reform on households in the Western Balkans.” (Mimeo). World Bank. 2010. “Lights Out: The Outlook for Energy and Eastern Europe and Central Asia.” Washington, DC: The World Bank Group. World Bank. 2012. “Balancing Act: Cutting subsidies, protecting affordability, and investing in the energy sector in Eastern Europe and Central Asia.” Washington, DC: The World Bank Group. World Bank. 2017. “Activation for Poverty Reduction: Realizing the Potential of Kosovo’s Social Safety Nets.” Washington, DC: The World Bank Group. World Bank. 2015. “Improvement of power tariffs and addressing social impacts”, Energy Sector Management Assistance Program (ESMAP), (2015) World Bank Group World Bank. 2019. “Consumption Poverty in the Republic of Kosovo.” Washington, D.C.: World Bank Group 51 Methodological Appendix Annex A1. Methodology to Estimate Electricity Consumption In this analysis, we use the 2017 round of the Household Budget Survey (HBS), which is a nationally representative sample of 2,232 households (representative roughly 299,845 households). The HBS includes a detailed expenditure module, including energy expenditures, allowing us to estimate household electricity expenditures, as well as their relationship with other energy expenditures and total household budget. Importantly, the data allow us to sort households with particular expenditure patterns in the overall welfare distribution and relate them to the national and international poverty thresholds. The electricity expenditure function is as follows: = (1)(1 + ) where Xit is the observed electricity expenditure in the HBS data, P1 is the estimated average tariff and VAT is the VAT tax rate for year t. We observe Xit in the HBS data, know the tariff structure and the VAT rate in year t. However, while in the HBS data we observe consumption quantities for several consumption items, we don’t observe Cit electricity consumption for household i at time t (in kWh). We also do not observe the time of the day when electricity consumption is made by the household, but we have information to infer the distribution of electricity consumption by time of the day in the population. Therefore, we use that distribution to estimate the average tariff each household faces. The estimation of electricity consumption is relatively straightforward. We can estimate electricity consumption Cit using the inverse mapping: (1 + ) = 1 where P1 is the estimated average tariff. We only include in the sample households with positive electricity expenditures. Annex A2. Methodology to estimate welfare impact of energy tariffs The methodology used to estimate the welfare impact of high energy tariffs is explained in detailed below. 1. The first step is to obtain estimates of electricity consumption for each household in the sample. While we observe consumption for several categories, we do not observe electricity consumption in the HBS. The estimation of electricity consumption is relatively straightforward using household electricity expenditures, since the tariff structure is simple (only varying by day or night, not by type of meter or blocks). We estimate current electricity consumption using electricity expenditures, the current tariff scheme for that particular year, and the VAT rates. 2. We calibrate a price elasticity of electricity demand. Our Approach: We do not estimate price elasticities, we calibrate them from previous empirical studies. Since we do not have specific estimates for Kosovo, we assume alternative scenarios and analyze sensitivity of results. We assume the following cases: price elasticity of -1, -0.5, -0.25, 0. An alternative approach involves assuming different elasticities for the different quantiles of the consumption distribution, since presumably electricity demand is more inelastic among households at the bottom quintiles (since they may have 52 less access to other sources of heating, especially in rural areas). As a result, we may expect higher impact on welfare among poor households. 3. Using the price elasticity and the simulated tariffs increase, we estimate future electricity consumption and expenditures. 4. Assuming a constant household budget, we estimate the household budget share of energy spending under the new tariff scheme. Then, we estimate energy/electricity poverty (if more than 10% of the budget is spent in energy/ electricity). 5. To estimate the impact on the welfare distribution, we need to simulate the household per capita consumption distribution under the new tariff scheme. The change in consumer surplus as a percentage of the budget (PPP Loss)57 is calculated as follows: = ℎ ∗ [1 + ( )] 2 where elect_share represents the share of household budget spent on electricity, ε represents the price elasticity of electricity consumption, and t represents the percentage tariff change. This represents the household welfare losses as a percentage of household expenditure, so after the price change household per capita consumption is reduced by the PPP loss. This simulated measure of welfare is used to estimate poverty rates. Notice that when ε is zero, this estimate of the direct welfare impact implicitly assumes that households do not substitute away from electricity, so it should be interpreted as either an estimate of the short-run impact (i.e., before households can adjust electricity consumption for other sources of energy) or as an upper bound of the long-run estimate. We analyze alternative scenarios (under different ε) to see the sensitivity of the results to this parameter. Some caveats applied in this analysis. First, we do not pretend to account for the impact of substitution on the demand-side by using data on energy expenditure only58. We assume that cross-price effects are very small. Second, we do not pretend to account for the impact of income effects that changes in the prices of these goods have.59 We assume that income effects are also very small. Finally, if the other goods bear taxes or subsidies there may be “second round” efficiency effects that should be taken into account. We assume these effects are negligible. 57 This formula is explained in Freund, Caroline, and Christine Wallich (1995). 58 First, as prices of different energy sources have been changing at different rates, households may have been able to change their spending patterns, exploiting relative price changes to minimize the impact of energy price rises on their welfare. Assessing these substitution possibilities requires estimation of a demand system. 59 Some households might be net producers of some of the items whose prices increase. Therefore, for some households, some price increases might result in an increase in welfare (This may be small for energy-requires large infrastructure investment) 53 Annex A3. Description of National Framework of Vulnerable Consumers Kosovo needs a coordinated national framework that draws on the particular roles and areas of expertise of each party currently involved with the support of electricity vulnerable customers. A national framework should harmonize government funded support, customer funded assistance and market developments and clearly define the goals of each party involved in the support of vulnerable customers. There should be clearly defined objectives for a harmonized framework, which focus on equality, fairness and the promotion of economic efficiency. The objectives of the framework should be consulted on with stakeholders prior to finalization. The framework should consider a two-tier approach with short-term actions geared towards improving the current electricity subsidy and medium-term actions for integration of the electricity subsidy in the social assistance system. • Program for protection of electricity vulnerable consumers In 2016, the GOK adopted new Law on Energy and Law on Electricity with commitments to protection of energy vulnerable (customers in need) in compliance with the EU Third Energy Package. The Law on Electricity requires from the GOK to develop a program for protection of vulnerable consumers60 and sets a number of characteristics of the future protection scheme (Article 49) with definition of the problem, mitigation measures, outputs and outcomes, stakeholders, institutional setup for implementation and timeframe, financing sources, process monitoring and evaluation of impacts. Annex 1 provides basic information on the format of a possible Program in Support of Vulnerable Consumers, drawing on the draft program which was recently elaborated in North Macedonia with European Union (EU) expertise. • Definition of electricity vulnerable customer Kosovo is a member of the Treaty for establishing the Energy Community in South East Europe61 and as such has commitments to meet the requirements stemming from the Third Energy Package of the EU with respect to the protection of vulnerable customers. The Energy Community Treaty stipulates that each Contracting Party shall implement the acquis communautaire on energy62. With respect to protection of vulnerable customers, the Contracting Parties should define the concept of vulnerable consumer at national level, adopt measures to protect such customers, and address energy poverty. There is no common EU-wide definition in the Third Energy Package. The protection of vulnerable consumers is to be implemented at national levels, and definitions are required in the national legislation (for country-level examples see Annex 2). They should reflect national circumstances, but at the same time take into consideration some general rules and characteristics. The Energy Community Treaty Ministerial Council has endorsed a proposal for a regional definition of vulnerable customers for Contracting Parties in October 2013. According to this proposal, a socially 60 Specifically, ”the Ministry in charge for social welfare shall develop, in cooperation with the Ministry in charge for energy , Ministry of Finance and in consultation with the Regulatory and other stakeholders of the electricity sector, a detailed program for establishing the status of socially customers in need, the scope of rights, as well as measures aimed at protecting the socially customers in need in order to meet their electricity demand.” 61 The Treaty establishing the Energy Community in South East Europe was signed in Athens on October 25, 2005, building on the Memorandum of Understanding on the Regional Electricity Market in South East Europe and its integration into the European Union Internal Electricity Market (‘’the Athens Memorandum”) of November 15, 2002. The Treaty calls for the establishment of the legal framework for an integrated energy market, and with it a single regulatory space for trade in energy between the EU and 9 South East European countries, including the United Nations Interim Administration Mission in Kosovo (UNMIK) pursuant to the United Nations Security Council Resolution 1244 on behalf of Kosovo. 62 https://www.energy-community.org/legal/treaty.html 54 vulnerable customer in the electricity sector should meet the following four criteria: (i) uses energy for supplying his/her permanent housing; (ii) does not exceed the maximum energy consumption per person: when defining electricity consumption level per person, Contracting Parties shall consider total consumption of up to 200 kWh/month for a family with up to 4 members, and reflects seasonality; (iii) belongs to a category of citizens with lowest income: for the definition of low income, beside the income of all available assets shall be taken into account; (iv) have her/his electricity consumption supplied through single-phase meter with a connection not exceeding maximum power. Furthermore, ‘the definition shall not include more than a minority of population. Market prices of the electricity should be cost reflective and consumption of vulnerable customers should be financed by social allowances.’ • Elaboration of secondary legislation A number of provisions of the Law on Electricity require clarifications which need to be made with respective secondary legislation, as summarized in Table A3.1. Table A3.1. Summary of key legislative requirements re: protection of electricity vulnerable customers Gaps in legal provisions Needed clarifications To reconcile the differences in The Law No. 05/L–081 on Energy of June 16, 2016, Art. 3.1 (1.26) defines a the definitions of ’customer in customer in need as a household customer of energy who due to social status need’ in the Law on and/or health conditions, shall have the right to be supplied with energy under Electricity and Law on certain conditions. The Law No. 05/L–085 on Electricity of June 16, 2016, Art. Energy. 3.1 (1.29) defines a customer in need as a household consumer, who, due to social status, enjoys some special rights regarding the supply with electricity, provided in exceptional cases, according to this law. The definition in the Law on Electricity is more comprehensive and reflective of the regional definition of the Energy Community Treaty. To specify the concept of ’Minority’ could be the poorest 20 percent of the population who spend on ‘minority of electricity energy (electricity and gas) 11.3 percent of their budget (10.8% on electricity). customers in Kosovo’. Other suggestion (made in the 2018 USAID funded consultancy report) is to set ‘minority’ at 15 percent of the population63. To expand the method(s) for The method(s) should consider: (i) socio-economic characteristics, and (ii) identification of customers in physical energy vulnerability considerations. Also, the Law on Electricity need. specifies that the support should be limited to customers with lowest income … where for the definition of low income, besides the income, all available assets shall be taken into account. This legal provision should be clarified further – given the forthcoming reform of poverty testing in social assistance in general, there should be reference to using the new comprehensive poverty test and the Social Registry for the identification of energy / electricity poor and vulnerable. To specify the key parameters Numerous options, among them writing off of bills or parts of bills / certain of financial protection of amounts of free gas or electricity; cash-backs; price discount for a certain quota customers in need. (in kWh) of consumption; preferential or subsidized tariffs, tariff discounts, social tariffs. If special tariffs, what kind of (e.g. progressive, flat lower, time- 63USAID and National Association of Regulatory Utility Commissioners. NARUC ERO PARTNERSHIP. Protection of vulnerable customers. Pristina, December 2018 55 of-use, etc.); for how long (temporary/transitionary versus constant); for what time of the year (winter season versus whole year). To specify the factors Such factors could be: the size of the household; the presence of household determining the size of tariff members in need of special protection; the source of heating (yes/no connection discount, subsidy, etc. to gas supply or district heating); the monthly or yearly amount of consumed electricity; seasonality, etc. To specify schedule of tariff As a starting point, the Law on Electricity prescribes making reference to a discounts. maximum level electricity consumption per person reflecting seasonality. When defining electricity consumption level per person, total consumption of up to 300 kWh per month for a family with up to 4 members shall be considered. Further, regulations (Administrative Instruction) should specify a schedule for how much will be discounted in each case (in kWh) or euro, or both per family size / composition and season. There could be a justification behind the schedule, such as electricity consumption (basic appliances, lightening, and heating for a certain number of rooms, with or without district heating, plus economies of scales). There can also be sanctions in case of excess consumption. To make sure that the energy / The Law on Electricity requires that the household customers benefiting from electricity subsidy is used for financial support for payment for electricity supply services shall not be allowed covering energy expenditure. to use such funds for other purposes. This implies a scheme where the bill subsidy is received by the final supplier of electricity, and not by the household. Define the data collection and KEDS/KESCO should be able to report on the meter holders eligible for reporting obligations with benefit, the household size and characteristics, consumption and price without respect to the electricity and with the discount, status of bill payment, etc. EDS/KESCO should be able vulnerable consumers and the to generate monthly summary data on number and types of beneficiaries, total electricity subsidy. amount of the discount and needed financing. Regulate protection against The obligation regarding protection against disconnection of vulnerable disconnection. customers is regulated with Article 34, paragraph 4 of the Law on Electricity which states: ‘Suppliers with public service obligations to supply final customers which enjoy the right of universal service shall establish mechanisms necessary to support vulnerable customers, upon consultation with the Regulatory, in order to avoid disconnection due to non-payment of electricity bills’. In this regard, the Regulator will require from suppliers with public service obligations a specific rule on protection vulnerable customers from disconnection. The rule will potentially be a common rule applied for all the suppliers in the Kosovo’s electricity market. 56 Annex A4. Program to support vulnerable customers in North Macedonia and reduce energy poverty The North Macedonia program can be useful for Kosovo primarily in informing the scope of issues to be addressed (content of the program). It cannot be that useful for identifying specific mitigation measures and support mechanisms because in North Macedonia the prevailing form of heating in low- income households is firewood, which is different from the energy consumption patterns of households in Kosovo and requires different approaches and solutions. The content of the program includes the following elements: (i) analysis of the legal and institutional framework in support of vulnerable household consumers; (ii) review of the obligations which need to be taken with the Third Legislative Package of the EU on energy and consumer protection; (iii) review of current national policies; (iv) definitions of vulnerable consumers and schemes for energy support in the EU and other countries (Albania, Bosnia and Herzegovina, Kosovo, Moldova, Montenegro, and Serbia); (v) proposal on how to identify energy vulnerable consumers based on analysis of energy poverty and vulnerability; (vi) model for calculation of subsidies for low income households and vulnerable consumers with variations based on the type of fuel used as main heating source; (vii) estimates of the amounts of energy consumed and the costs of prevention of energy poverty; (viii) review of the Integrated Development Program for Fighting Poverty and Energy Protection of Vulnerable Consumers. The entry point for the North Macedonia’s program, which is currently being drafted with EU technical assistance for strengthening the administrative capacity of the Energy Department of the Ministry of Economy and the Energy Agency, is the Program for subsidizing energy consumption which was launched in 2010 and implemented throughout six consecutive years. It is based on the understanding that: (i) the energy (and electricity) market should be liberalized with prices determined by supply and demand; (ii) each household should be entitled to receive certain amount of ‘social’ energy to guarantee a minimum standard of living; and (iii) all types of energy that are available to households are treated equally. According to the general measures to protect consumers and reduce energy poverty set out in the Energy Law of 2011, programs should be annual, and should provide inter alia for subsidizing household energy consumption, more efficient use of energy, ways of implementing the measures, sources of budgetary and other means and to finance the measures and the bodies responsible for implementation. The Energy Regulatory Commission may impose a public service obligation for the supplier to supply small final customers with electricity at reasonable, easily and clearly comparable, transparent and non- discriminatory prices, including at regulated prices for a transition period. The program envisages financial support to energy consumption with ceilings based on energy needs of households. For the calculation of the amounts of support, North Macedonia uses expert assessments of the characteristics of consumption and a model developed by the Energy Institute Hrvoje Pozar, Croatia. The model defined norms for energy needs include heating, cooking, hot water, lightening and appliances for households with different characteristics - number of members, available and used forms of energy (energy infrastructure), structure of energy use, energy prices for households and income (purchasing power). The model defines norms for heating in square meters and minimum heating norms per square meter which can differ by climatic conditions and regional characteristics. The energy norms for cooking and hot water are common for all households in North Macedonia, while the norms for consumption of electricity for appliances and lighting assume that energy poor households consume less than average 57 households (have less devices due to low purchasing power), and is adjusted with a correction factor for lower purchasing power is 0.8. Source: Draft Program for Support of Energy Vulnerable Customers in North Macedonia. EuropeAid / 129822 / D / SER / MK. 58 Annex A5. Energy/electricity poor or vulnerable customers in selected EU member States, Western Balkans and other countries: definitions and support measures Country Definitions Support measures Western Balkan countries Albania Categorical identification – list of categories Direct subsidies from the state budget of consumers which are protected in terms differentiated based on consumption (lower of consumption and energy demands, subsidy if the monthly consumption is below 300 including specific categories of disabled and kWh, and higher subsidy if the monthly sick people, unemployed and low-income consumption is over 300 kWh). Since the subsidy households. increase with consumption, there are no incentives for energy / electricity saving. Bosnia and Categorical identification combined with Federation of Bosnia and Herzegovina: The Herzegovina low income (income test). Federation of benefit is extended when the monthly electricity Bosnia and Herzegovina: Eligible are consumption is beyond certain level. The pensioners with the lowest pensions, Federation provides direct budget subsidies on a beneficiaries of last resort income support. monthly basis. They differ by provider and Differences across cantons. Sarajevo Canton canton. Sarajevo Canton provides special winter provides support for low-income families, subsidy (5 months) credited to the electricity, single pensioners and two-member district heating or gas bill or in cash (in other pensioner households with low income, low cases). income households with a member who receives benefit for care of other person (disability); families where a member is 100% disabled irrespective of income. Republika Srpska: pensioners with lowest incomes, beneficiaries of last resort income support, maternity allowance, child allowance and allowance for support and care. Montenegro Categorical identification: (i) welfare Lower / subsidized tariffs. The difference recipients whose status is determined by the between costs and revenues that occur during the competent authority; (ii) persons with supply of vulnerable consumers are covered by disabilities and persons with disabilities who the government budget. are in poor health or are in danger in case of Supplier of electricity or natural gas to vulnerable interruption of supply (Art. 156 of the customers is a public supplier. Energy Law, 2010). Serbia Three groups of energy vulnerable Direct subsidy financed by the state budget and customers: (i) recipients of financial social delivered as discount of the monthly electricity assistance (categorical identification); (ii) bill. Depends on household size and consumption recipients of monthly child benefit below certain limits. The subsidy/discount covers (categorical identification); and (iii) other – between 60 and 240 kWh depending on identified with income and asset tests. The household size, and is determined based on status is formally granted by a designated consumption of basic appliances, but not for authority in the municipalities. heating. 59 Selected EU Member States Austria Informal definition - energy poverty affected Identification takes place through the social are households with income below a poverty welfare system. threshold and with above average energy costs. Belgium The definition is based on three composite Social tariffs. Vulnerable consumers can be indicators: (i) ‘measured energy poverty’ - entitled also to free energy scan, or social renting median value of the ratio between energy tariffs for energy saving. Different welfare expenditures and equivalent household system parameters are used to specify the eligible income (corrected for the household size); customers. (ii) ‘boundary value’ - twice the value of the median value, and (iii) ‘hidden energy poverty’ - the fraction of households that restrict their energy use to the extent that it might have a negative impact on living conditions and quality of life. “Subjective energy poverty’ (percentage of households with difficulties to adequately heat their dwelling) is also considered. Bulgaria Vulnerable customers are household Targeted heating assistance is extended for 5 customers receiving targeted assistance for months in a calendar year. The basis for electricity, heating and natural gas in calculation is the ‘’guaranteed minimum income accordance with the Social Assistance Act indicator’’ multiplied by a specific for each and the Regulations for its implementation. category coefficient. Criteria are rigorous – no There are 17 categorical groups with access contract for support by other person, no sale of to targeted energy assistance. immobile assets in the past 5 years, no travel abroad on own expenses in the past 12 months, no savings above EUR 250 per family member. A ‘’social tariff’ was designed but not implemented. The social tariff monthly amount is to be determined on the basis of the national currency unit equivalent of 385 kWh electricity, of which 280 kWh day tariff and 105 kWh night tariff, at the retail price of electricity for household users as of 31 October of the current respective year. Croatia Categorical criteria for energy poor or Tariff discount / deduction of electricity bills with vulnerable: social welfare recipients, people about EUR 26 per month irrespective of the bill with disabilities. Art.3(40) of the Energy Act size. Right to supply of specific quantity of defines specifically the term “protected” energy for protected customer. customer – a customer who, in case of partial distortion of energy supply, has the right to supply of specific quantity of energy. 60 Cyprus Energy poor or vulnerable are customers Social tariff. Beneficiaries are identified through who may be in a difficult position because of the social welfare system. low income (from tax statement) due to professional status, marital status and specific health conditions and are, therefore, unable to cover the costs of reasonable needs of electricity, as they represent significant share of their disposable income. Ireland Energy poor or vulnerable are those Beneficiaries are identified through the social spending more than 10 percent of their welfare system. disposable income on energy services in the home. Italy A family is vulnerable when more than 5 Social tariff. Beneficiaries are identified through percent of income is spent for electricity and the welfare system. Other financial indicators are 10 percent for gas. also used. Portugal Energy (for electricity and/or natural gas) Social tariff for electricity and/or natural gas, vulnerable are the following categories / introduced with Decree-Law No.138-A/2010. recipients of: (i) solidarity supplement for With Ordinance No. 178-B/2016, an integrated the elderly; (ii) social insertion income; (iii) system for assigning the social tariff for supply of unemployment allowance; (iv) first step of electricity and natural gas the family allowance, and (v) invalidity social pension. Law n. 7-A/2016 changed the legal regime for social support to energy consumption, looking to create a single model and an automatic entitlement to gas and electricity social tariffs. As a result of these changes, which came into effect on July 1, 2016, the number of economically vulnerable customers that benefited from a 33.8 percent discount in the electricity bill rose to about 800,000. UK A household is considered to be energy poor The welfare system looks at age and health if: (i) income is below the poverty line conditions of household members as well as their (taking into account energy costs); and (Ii) income. An energy poverty assessment is carried energy costs are higher than is typical for out (Low Income High Cost) which also looks at their household type. assets (scoring formula for the property). Other countries Armenia Categorial identification - energy cash FBP beneficiary households are given ARM benefits top up existing Family Benefit 1,000 every month on top of their FBP monthly Programs (FBP), the last resort income benefit. The scheme is easy to administer, relies support scheme. on social assistance targeting system and delivery infrastructure. Moldova Categorical identification of energy Direct subsidies covered by the state budget. The vulnerable consumers, including pensioners, support scheme distinguishes between winter persons with disability, single mothers with season for heating, and electricity and gas for the 61 children younger than 16 years, people year. Subsidy is provided for: electricity costs for living in one-bedroom dwelling and with consumption of 60 kWh per month; or, district monthly income below pre-set threshold. heating or natural gas heating costs of 30 cubic meters; or coal and firewood – 50 percent of the price (ton or cubic meter). The compensation amounts are determined by the Law on Social Protection. Sources: http://www.energycommunity.org/portal/page/portal/ENC_HOME/ENERGY_COMMUNITY/ Legal/ Treaty; Energy poverty and vulnerable customers in the energy sector across the EU: analysis of policies and measures. Policy Report. https://ec.europa.eu/energy/sites/ener/files/documents/vcwg2013_instruments_and_ practices 0.pdf; Examples of instruments and practices. https://ec.europa.eu/energy/sites/ ener/files/documents/vcwg- 2013_instruments_and_ practices_0.pdf; Energy Community Regulatory Board. Treatment of the vulnerable customers in the Energy Community, June 2013; Austria. Electricity Assistance Fund. http://www.verbund. com/cc/en/ responsibility/corporate_citizenship/electricity-assistance-fund; Bulgaria: Definitions and Measures for Protection of Vulnerable Customers, power point presentation by the Ministry of Energy, Sofia, April 27, 2016; France: Subsidies for Low-income Owners: http://www.anah.fr/ habitermieux.html; Hungary: Home Maintenance Support for Families in Need, Ensured by Local Municipalities: http://csaladitudakozo.kormany.hu/download/ 7/5c/60000/lakásfenntartási%20 támogatás%202013. doc; The Netherlands: social support: http://wetten.overheid.nl/ BWBR0020031/ geldigheidsdatum_11- 10-2013; Portugal. POR 26-Social Tariff (electricity and natural gas). Last updated 07 November 2017. 62 Annex A6. Simulating electricity consumption thresholds for the subsidy This section presents the results of the simulations of a redesign of the electricity benefit that includes consumption thresholds. The simulations introduce additional eligibility requirements for electricity consumption to ensure that the benefit only targets energy-vulnerable households. The savings generated through this more stringent targeting procedure are then used to expand the scope of the scheme by expanding it to single-elderly households, a group of households who are at particular risk of energy- poverty. The simulated redesigned electricity subsidy includes the following features: 1) The size of the subsidy depends on household size, to account for higher energy needs among bigger households. The current energy benefit subsidizes all eligible households with 186 kwh per month irrespective of their size. 2) Introducing electricity consumption filters to incentivize energy savings. These consumption filters remove the subsidy from households whose electricity consumption is above a certain threshold. Table shows the proposed thresholds and the size of the energy bill deductions for each type of household structure.64 3) The savings introduced due to the consumption filters (as some households currently receiving the subsidy will not qualify anymore due to their electricity consumption exceeding the threshold) are used to award the subsidy to single-elderly households (who are at particular risk of electricity poverty). If the savings exceed the resources needed to cover all single-elderly households below the consumption threshold, the remaining budget is used to increase the size of the subsidy. Table A6.1: Electricity Subsidy: Consumption Requirements and Bill Deduction Monthly Electricity Monthly Bill Deduction (in Consumption Subsidy Type kWh) Requirement HH with 1 If c<240kWh Full 120 Member if 240 kWh300kWh No subsidy 0 HH with 2 or 3 If c<300kWh Full 150 Member if 300390 kWh No subsidy 0 HH with 4 or 5 If c<360 kWh Full 180 Member if 360470 kWh No subsidy 0 HH with 6 or more If c<390 kWh Full 195 Member if 390530 kWh No subsidy 0 The amount of the simulated benefit depends on electricity consumption and is proportional to household size. For instance, single-member households that consume less than 240kWh per month may be eligible to receive a monthly bill deduction of 120 kWh (“Full” subsidy). If this household consumes more than 240 kWh and less than 300 kWh per month, it may be eligible to receive a deduction of 60kWh (“Partial” subsidy). If this household consumes more than 300 kWh per month, it does not have the right to any decrease in its monthly obligations. The same reasoning applies to larger households, but the consumption thresholds increase with household size. Three different scenarios are simulated. Assuming full take-up as in the baseline scenario, we quantify the impact on the coverage of vulnerable groups, as well as the fiscal costs associated for three different options. (i) Experiment 1: Current Energy Subsidy, with adjusted consumption thresholds depending on household size and electricity consumption, as proposed in Table A6.1. Allocate the remainder of the funds to single elderly households whose electricity consumption is below the threshold. (ii) Experiment 2: Same as in experiment 1 but double the electricity consumption thresholds (full and partial), leaving the amount of the subsidy unchanged. l. Allocate the remainder of the funds to single elderly households whose electricity consumption is below the threshold. (iii) Experiment 3: Same as 1 but eliminate the partial subsidy, allocating to everyone with consumption below the higher threshold the full subsidy. Allocate the remainder of the funds to single elderly households whose electricity consumption is below the threshold. Under all three experiments, overall coverage of the electricity subsidy would decrease as households with high electricity consumption would be excluded from the scheme. Under experiment one and two, coverage would decrease from 24,470 households to about 20,368 households, a decrease by about 1.4 percentage points of all households. Under experiment 2 with higher consumption thresholds, coverage would only decrease by about 0.1 percentage points. However, the decrease in coverage is due to households whose consumption is above the threshold being excluded from the threshold. Furthermore, the decreasing coverage is balanced out by the inclusion of all single-elderly households (5866 households) in the scheme. Table A6.2: Estimated Coverage of Electricity Benefits, Baseline vs Policy Scenario including single elderly households Experiment 2: Alternative Experiment 3: Keep Experiment Baseline Design +Double both Full Subsidy at 1:Alternative Design Thresholds Partial Threshold Total Households 20,368 24,863 Covered 24,470 20,368 As % of Total Population 8.16 6.79 8.29 6.79 Change in Coverage -1.4 0.1 -1.4 Source: World Bank estimates based on Kosovo HBS 2017. 64 While the experiments would decrease coverage for most categories of vulnerable households (excluding single elderly households), the households excluded from the scheme would be the ones with high electricity consumption. Figure A6.1 gives the coverage of the three experiments scheme by category. As expected, all experiments would reduce coverage especially of social-assistance receiving households as well as households who benefit from war-related pensions. Given higher thresholds under experiment 2, coverage would not decrease as much as for the other two experiments. Furthermore, Figure gives the impact on coverage by quintile. Under experiments 1 and 3, coverage of quintile 1 decreases from 22.8 to about 15 percent. Under experiment 2, coverage by quintile 1 decreases to about 16 percent. Figure A6.1: Estimated Coverage of households by category, Baseline vs. Policy Scenarios 74.1588.47 War-related Pension Recipient HHs 74.15 100 16.3 17.2 Poor 16.3 24.5 25.8 26.3 Extreme Poor 25.8 42.0 7.0 9.4 Female-Headed HH 7.0 11.0 25.636.3 HH with Disabled members 25.6 47.7 0.0 1.4 Single elderly HH 0.0 1.4 6.6 7.9 HH with Child <5 6.6 10.3 6.5 7.4 HH with Unemployed 55+ 6.5 10.1 44.8 59.4 Social Assistance Recipient Households 44.8 78.9 58.3 61.6 Bottom Quintile - With Social Assistance 58.3 91.5 2.7 3.0 Bottom Quintile - Without Social… 2.7 3.3 0 20 40 60 80 100 120 Scenario 3 Scenario 2 Scenario 1 Baseline Source: World Bank estimates based on HBS 2017 data. 65 Figure A6.2: Estimated Coverage of households by quintile, Baseline vs. Policy Scenarios 1.3 Q5 2.5 1.3 3.1 1.5 Q4 3.0 1.5 3.0 2.6 Q3 4.2 2.6 4.8 5.7 Q2 8.1 5.7 9.8 15.0 Q1 16.0 15.0 22.8 4.8 Overall 6.4 4.8 8.2 0 5 10 15 20 25 Scenario 3 Scenario 2 Scenario 1 Baseline Source: World Bank estimates based on HBS 2017 data. The savings generated due to exclusion of high-electricity consuming households would be used to increase the generosity of the program. Table A6.3 shows the new subsidy amounts by household size. Under experiment 1, the subsidy would increase by 53 percent for all categories of households. Under experiment 2, the increase would only be 20 percent due to more generous consumption thresholds and thus lower savings. Table A6.3: Electricity Subsidy: Consumption Requirements and Bill Deduction Monthly Bill Monthly Bill Deduction (in Deduction (in Monthly Electricity kWh): kWh): Subsidy Type Consumption Requirement Experiment 1& 3 Experiment 2 HH with 1 If c<240kWh Full 183.6 144 Member if 240 kWh300kWh No subsidy 0 0 HH with 2 or 3 If c<300kWh Full 229.5 180 Member if 300390 kWh No subsidy 0 0 66 HH with 4 or 5 If c<360 kWh Full 275.4 216 Member if 360470 kWh No subsidy 0 0 HH with 6 or If c<390 kWh Full 298.35 234 more Member if 390530 kWh No subsidy 0 0 Source: World Bank estimates based on Kosovo HBS 2017. 67 Annex A7. Additional Figures and Tables A7-Figure 1: Distribution of Estimated Monthly Electricity Consumption (kWh) Source: World Bank estimate A7-Figure 2: Mean Monthly Electricity Consumption for Vulnerable groups (Kwh), 2017 Poorest Quintile - With Social Assistance 308 Poorest Quintile - Without Social Assistance 393 Social Assistance Recipient Households 423 HH with Unemployed 55+ 487 HH with Child <5 532 Single elderly HH 292 HH with Disabled members 455 Female-Headed HH 429 0 100 200 300 400 500 600 Source: World Bank estimates based on Kosovo 2017 HBS. 68 A7-Figure 3: Seasonal Variation in Household Average Electricity Consumption, by Consumption Quintile (kWh/month) 700 600 500 400 300 200 100 0 Poorest Quintile 2nd 3rd 4th Richest Quintile 1st Quarter 2nd Quarter 3rd Quarter 4th Quarter Source: World Bank estimate A7-Figure 4: Average Household Monthly Electricity Consumption (kWh), by Region 2017 600 500 400 300 200 100 0 Gjakove Ferizaj Gjilan Mitrovice Peje Pristina Prizren ENA HBS Source: World Bank estimate based on Kosovo 2017 HBS and ENA. 69 A6-Figure 5: Distribution of pension of single elderly households Source: World Bank estimate based on Kosovo HBS 2017. A7 - Table 1: Mean Electricity Consumption (Kw/Month), 2017 Avg. Electricity Consumption Admin. Data HBS Gjakove 411.14 488.76 Ferizaj 399.68 422.76 Gjilan 394.05 512.04 Mitrovice 354.67 337.08 Peje 372.82 513.35 Pristina 425.72 468.79 Prizren 386.98 435.98 Source: World Bank estimates based on Kosovo 2017 HBS. Notes: Consumption quintiles based on national welfare aggregate. A7 - Table 2: Simulated impact of electricity tariff increase on electricity budget share (percentage points change in electricity share) High Medium Low e=0 e=-0.25 e=-0.5 e=0 e=-0.25 e=-0.5 e=0 e=-0.25 e=-0.5 70 All 5.0 2.8 0.6 3.3 2.1 0.8 1.7 1.1 0.6 Urban 5.2 2.9 0.7 3.5 2.2 0.9 1.7 1.2 0.7 Rural 4.8 2.7 0.6 3.2 2.0 0.8 1.6 1.1 0.6 Gjakove 3.9 2.2 0.5 2.6 1.6 0.7 1.3 0.9 0.5 Gjilan 5.2 2.9 0.6 3.5 2.2 0.9 1.7 1.2 0.6 Mitrovice 4.2 2.4 0.5 2.8 1.8 0.7 1.4 1.0 0.5 Peje 6.0 3.3 0.7 4.0 2.5 1.0 2.0 1.4 0.7 Prizren 4.5 2.5 0.6 3.0 1.9 0.8 1.5 1.0 0.6 Pristina 5.5 3.1 0.7 3.7 2.3 0.9 1.8 1.3 0.7 Ferizaj 5.0 2.8 0.6 3.4 2.1 0.8 1.7 1.2 0.6 Q1 5.6 3.2 0.7 3.7 2.3 0.9 1.9 1.3 0.7 Q2 5.6 3.2 0.7 3.8 2.3 0.9 1.9 1.3 0.7 Q3 5.1 2.9 0.6 3.4 2.1 0.9 1.7 1.2 0.6 Q4 4.8 2.7 0.6 3.2 2.0 0.8 1.6 1.1 0.6 Q5 4.0 2.2 0.5 2.6 1.7 0.7 1.3 0.9 0.5 Source: World Bank estimates based on Kosovo 2017 HBS. 71