Romania Energy Poverty Assessment Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints September 2024 Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints Monica Robayo-Abril, Jonathan Karver, Britta Rude, Ailin Tomio, Alessandro Silvestri, and Kiyomi Cadena Poverty and Equity Global Practice and Mind, Behavior and Development (eMBeD) Unit The World Bank © 2024 International Bank for Reconstruction and Translations—If you create a translation of this work, please Development / The World Bank add the following disclaimer along with the attribution: This 1818 H Street NW translation was not created by The World Bank and should Washington DC 20433 not be considered an official World Bank translation. The Telephone: 202-473-1000 World Bank shall not be liable for any content or error in this translation. Internet: www.worldbank.org Adaptations—If you create an adaptation of this work, please This work is a product of the staff of The World Bank with add the following disclaimer along with the attribution: This external contributions. The findings, interpretations, and is an adaptation of an original work by The World Bank. conclusions expressed in this work do not necessarily reflect Views and opinions expressed in the adaptation are the sole the views of The World Bank, its Board of Executive Directors, responsibility of the author or authors of the adaptation and or the governments they represent. are not endorsed by The World Bank. 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 Third-party content—The World Bank does not necessarily imply any judgment on the part of The World Bank concerning own each component of the content contained within the the legal status of any territory or the endorsement or work. The World Bank, therefore, does not warrant that acceptance of such boundaries. the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from Rights and Permissions such infringement rests solely with you. If you wish to The material in this work is subject to copyright. Because reuse a component of the work, it is your responsibility to The World Bank encourages dissemination of its knowledge, determine whether permission is needed for that reuse and this work may be reproduced, in whole or in part, for to obtain permission from the copyright owner. Examples noncommercial purposes as long as full attribution to of components can include, but are not limited to, tables, this work is given. Any queries on rights and licenses, figures, or images. including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.orgmailto:pubrights@worldbank.org. Table of Contents Profiling sustainable heating transitions: behavioral barriers to and enablers of sustainable heating transitions . . . . . . . . . . . . 58 Acknowledgments . . . . . . . . . . . . . . . . . . . . 10 Key factors correlated with intentions to ABBREVIATIONS AND ACRONYMS . . . . . . . . . . . . 11 upgrade heating. . . . . . . . . . . . . . . . . . . . . . 66 EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 CHAPTER 4 SIMULATING THE EX-ANTE IMPACTS OF ENERGY PRICE INCREASES ON HOUSEHOLDS . . . . . . . . . . . . . . . 71 Why is addressing energy poverty key for economic development? . . . . . . . . . . . . . . . . . . . . . . . . 13 4.1 Direct Impacts . . . . . . . . . . . . . . . . . . . . . . 72 Impact on energy poverty rates . . . . . . . . . . . . 72 Where do Romanian households stand regarding energy poverty and which households are more Impact on welfare measures . . . . . . . . . . . . . . 75 likely to be affected? . . . . . . . . . . . . . . . . . . . . 14 4.2 Indirect Impacts . . . . . . . . . . . . . . . . . . . . . 77 What are the key structural and behavioral barriers 4.3 Robustness Checks . . . . . . . . . . . . . . . . . . . 78 that prevent households from transitioning out of Using a matched dataset and welfare measures energy poverty and adopting cleaner technologies from the EU-SILC . . . . . . . . . . . . . . . . . . . . . 78 and sustainable energy use? . . . . . . . . . . . . . . . . 16 Using an alternative poverty threshold . . . . . . . . 79 What are the welfare implications of rising energy prices? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Varying price elasticities across the income distribution . . . . . . . . . . . . . . . . . . . . . . . 80 What measures are in place and what potential 4.4 Limitations and Caveats . . . . . . . . . . . . . . . . . 81 mitigation measures can be devised to tackle energy poverty and protect the most vulnerable? . . . . 18 CHAPTER 5 EFFECTIVELY TACKLING ENERGY POVERTY AND MITIGATING THE ADVERSE WELFARE IMPACTS OF CHAPTER 1 MOTIVATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 RISING ENERGY PRICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 CHAPTER 2 ACCESS TO AFFORDABLE, RELIABLE, AND 5.1 Legal and Policy Framework . . . . . . . . . . . . . . . 84 SUSTAINABLE ENERGY AMONG ROMANIAN HOUSEHOLDS . . 29 5.2 Recent Policies and Interventions . . . . . . . . . . . 85 2.1. Energy Sector Background . . . . . . . . . . . . . . . 30 Untargeted measures—energy price caps . . . . . . . 86 2.2. Connectivity . . . . . . . . . . . . . . . . . . . . . . . 32 Targeted income-support measures . . . . . . . . . . 87 2.3. Energy Affordability: Monetary and Energy efficiency measures . . . . . . . . . . . . . . 92 Nonmonetary Measures . . . . . . . . . . . . . . . . . . . 35 Do current social protection measures 2.4. Factors Associated with Energy Poverty . . . . . . . 47 cover the groups who are at a higher risk of CHAPTER 3 A BEHAVIORALLY INFORMED APPROACH TO experiencing energy poverty? . . . . . . . . . . . . . 93 SUSTAINABLE ENERGY TRANSITIONS . . . . . . . . . . . . . . . . . . . . . . 51 5.3 How Do We Design Effective Mitigation 3.1 Sustainable Energy Use and Attitudes toward Measures to Protect the Most Vulnerable Energy Poor? . 95 Energy Efficiency . . . . . . . . . . . . . . . . . . . . . . . 52 Overarching principles . . . . . . . . . . . . . . . . . 95 3.2 A Focus on Sustainable Heating Transitions . . . . . 54 Policy simulations . . . . . . . . . . . . . . . . . . . . 96 The current state of heating and intentions to Strengthening energy-efficiency initiatives with upgrade technologies . . . . . . . . . . . . . . . . . . 55 insights from behavioral science . . . . . . . . . . 101 CHAPTER 6 CONCLUSIONS AND POLICY IMPLICATIONS . . . . . 107 Different Approaches to Measuring Energy Poverty . . 141 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Discussion and Conclusion . . . . . . . . . . . . . . . . 143 ANNEX 1 DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 ANNEX 3 CONSTRUCTION OF UNDERLYING INDICATORS . . . 144 Household Budget Survey . . . . . . . . . . . . . . . . . 122 Monetary Measures . . . . . . . . . . . . . . . . . . . . 144 European Union Statistics on Income and Living Nonmonetary Measures . . . . . . . . . . . . . . . . . . 144 Conditions (EU-SILC) . . . . . . . . . . . . . . . . . . . . 124 Consumption Aggregate . . . . . . . . . . . . . . . . . . 144 World Bank Rapid Household Surveys . . . . . . . . . . 125 ANNEX 4 METHODOLOGY: MEASURING DIRECT AND Qualitative Data Collection . . . . . . . . . . . . . . . . 125 INDIRECT WELFARE IMPACTS OF ENERGY INFLATION . . . . . . 146 Qualitative discussion guide . . . . . . . . . . . . . . . 125 Impacts on Energy Poverty . . . . . . . . . . . . . . . . 146 ANNEX 2 THE METHOD FOR MEASURING ENERGY Impacts on Income Poverty . . . . . . . . . . . . . . . . 147 EXPENDITURE AND ENERGY POVERTY IN ROMANIA ANNEX 5 DESCRIPTIVE STATISTICS OF SINGLE-ELDERLY MATTERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 HOUSEHOLDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Main Findings . . . . . . . . . . . . . . . . . . . . . . . . 131 ANNEX 6 SIMULATED POVERTY AND INEQUALITY RATES Different Approaches to Measuring Energy ACROSS VARIOUS SCENARIOS . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 Expenditure and Welfare . . . . . . . . . . . . . . . . . . 131 ANNEX 7 SIMULATION OF POTENTIAL SHORT-TERM The Method of Measuring Households’ Energy MEASURES MITIGATING DIRECT EFFECTS OF RISING Expenditure Matters . . . . . . . . . . . . . . . . . . . . 132 ENERGY PRICES BY ELIMINATING PRICE CAPS FOR ELECTRICITY AND NATURAL GAS Results Using Income-Based Measures . . . . . . . . . 137 (70% INCREASE IN TOTAL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 Boxes Figure 7 Coverage Rates of Electricity by Groups, 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . 33 Figure 8 Coverage Rates of Natural Gas Box 2.1: Measuring Energy and Electricity Poverty to by Groups, 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . 34 Design Better Social Programs . . . . . . . . . . . . . . . . . 36 Figure 9 Type of Heating Methods, 2021 (%) . . . . . . . . . . 34 Box 3.2: Energy and the Elderly in Romania . . . . . . . . . 41 Figure 10 Type of Cooking Methods, 2021 (%) . . . . . . . . . 34 Figure 11 Type of Heating Energy Source by Rural/ Tables Urban, 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . . . 35 Figure 12 Type of Cooking Energy Source by Rural/ Table 1 Regression of Energy Expenditure Shares, 2021 . . . 48 Urban, 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . . . 35 Table 2 Home Heating Aid Parameters . . . . . . . . . . . . 90 Figure 13 Energy Expenditure Share and Energy Poverty, 2021 (Shares) . . . . . . . . . . . . . . . . . . . . . . 38 Table 3 Simulation of Short-Term Measures Mitigating the Direct Effects of Rising Energy Prices under Figure 14 Monetary vs. Energy Poverty, 2021 . . . . . . . . . 38 Current Program Structure . . . . . . . . . . . . . . . . . . . 98 Figure 15 Energy Expenditure Shares by Income Table 4 Simulation of Short-Term Measures Mitigating Quintiles, 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . 39 the Direct Effects of Rising Energy Prices by Figure 16 Energy Poverty by Income Quintiles for Three EliminatinPrice Caps for Electricity and Natural Gas . . . . 100 Measures of Energy Poverty, 2021 . . . . . . . . . . . . . . . 39 Table A.2.1 Regression of Energy Expenditure Shares Figure 17 Energy Expenditure Shares among Households with Heating Systems, 2021 . . . . . . 140 by Groups, 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . 40 Table A.5.1 Descriptive Statistics of Single-Elderly Figure 18 Energy Poverty Rates by Groups, 2021 (%) . . . . . 40 Households Compared to All Other Households, 2021 . . . 148 Figure 19 Energy Spending Shares by Gender Groups, Table A.6.1 Simulated Estimates of AROP Rates under a 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Full Set of Scenarios . . . . . . . . . . . . . . . . . . . . . . 150 Figure 20 Energy Poverty Rates by Gender Groups, Table A.7.1 Simulation of Potential Short-Term 2021 (%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Measures Mitigating Direct Effects of Rising Energy Prices by Eliminating Price Caps for Electricity and Figure 21 Average Energy Spending Share by Type of Natural Gas (70% Increase in Total) . . . . . . . . . . . . . . 152 Energy and by Heating Technology, 2021 . . . . . . . . . . . 43 Figure 22 Energy Affordability in a Broader Context: Romania vs. Other Europe and Central Asia Countries . . . 44 Figures Figure 23 Inability to Keep Home Adequately Warm in the Past 12 Months, 2020 vs. 2022 (%) . . . . . . . . . . . . . 45 Figure ES1 Inability to Keep Home Adequately Warm, Figure 24 Arrears on Utility Bills in the Past 12 Months, 2022 vs. 2020 (%) . . . . . . . . . . . . . . . . . . . . . . . . . 15 2020 vs. 2022 (%) . . . . . . . . . . . . . . . . . . . . . . . . . 45 Figure 1 Natural Gas Prices for Household Consumers, Figure 25 Households with Leakages by Income Romania vs. the EU27, 2008s2–2023s2 (Half-Yearly) . . . . . 24 Quintiles, 2019–20 (%) . . . . . . . . . . . . . . . . . . . . . . 46 Figure 2 Electricity Prices for Household Consumers, Figure 26 Households with Leakages by Groups, Romania vs. the EU27, 2008s2–2023s2 (Half-Yearly) . . . . . 24 2019–20 (%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Figure 3 Energy Mix in Romania, 2020 . . . . . . . . . . . . . 30 Figure 27 Households with Environmental Problems by Income Quintiles, 2019–20 (%) . . . . . . . . . . . . . . . . . 47 Figure 4 Energy Dependency in the European Union, 2020 . 31 Figure 28 Households with Environmental Problems by Figure 5 Total Energy Supply by Source, 1990–2019 (TJ) . . . 31 Groups, 2019–20 (%) . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 6 Electricity Production, 2011–20 (MW) . . . . . . . . . 31 Figure 29 Awareness of Energy Used at Home, Figure 48 Harmonized Consumer Price Index, Total July 2023 (%) . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 and Energy (January 2021–May 2024) . . . . . . . . . . . . . 72 Figure 30 Consideration of Changes in Energy Use, July Figure 49 Simulated Increase in Energy Poverty Rates 2023 (%) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 and Pre-Price-Increase Energy Poverty Rate (P10) by Groups for 2021 . . . . . . . . . . . . . . . . . . . . . . . . . 74 Figure 31 Willingness to Make Financial Sacrifices for Further Development of Renewable Energy, July 2023 (%) . 54 Figure 50 Simulated Increase in Energy Poverty Rates and Pre-Price-Increase Energy Poverty Rate (P10) by Figure 32 Responses to Energy Price Increases . . . . . . . . 54 Income Quintiles, 2021 . . . . . . . . . . . . . . . . . . . . . 74 Figure 33 Heating Systems by Category . . . . . . . . . . . . 55 Figure 51 Sensitivity of Energy Poverty Rates (P10) to Figure 34 Heating Fuels by System and Type . . . . . . . . . 56 Income and Price Elasticity Parameters . . . . . . . . . . . . 75 Figure 35 Romanians’ Intentions regarding a Heating Figure 52 Simulated Welfare Losses and or Insulation Upgrade within the Next Few Years and Poverty Impacts . . . . . . . . . . . . . . . . . . . . . . . . . 76 Time Frame for Upgrade . . . . . . . . . . . . . . . . . . . . 57 Figure 53 Increase in AROP Rates and Pre-Price- Figure 36 Romanians with Traditional Heating Systems Increase AROP Rate by Income Quintiles for 2021— (All Households) Who Fall into a Specific Profile Direct Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 According to Their Intention to Upgrade Systems Figure 54 Increase in AROP Rates and Pre-Price- (Insulation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Increase AROP Rate by Groups for 2021—Direct Effect . . . . 79 Figure 37 Perceived Financial Difficulty According to Figure 55 Increase in AROP Rates and Pre-Price- Profiles (% of Population within Upgrade Increase AROP Rate by Income Quintiles for 2021— Intentions Group) . . . . . . . . . . . . . . . . . . . . . . . . 58 Direct Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 38 Perceived Difficulty in Spending More to Figure 56 Increases in International Poverty Rates Maintain the Same Level of Comfort with a Modern ($6.85 a Day) and Pre-Price-Increase International Heating System by Region . . . . . . . . . . . . . . . . . . . 59 Poverty Rate by Groups for 2021—Direct Effect . . . . . . . . 80 Figure 39 Awareness of Programs or Initiatives to Figure 57 Increases in Pre-Price-Increase International Financially Support Investments, July 2023 (%) . . . . . . . . 60 Poverty Rates ($6.85 a Day) by Income Quintiles for Figure 40 Awareness of Subsidies (% of Population 2021—Direct Effect . . . . . . . . . . . . . . . . . . . . . . . . 80 within Upgrade Intentions Group) . . . . . . . . . . . . . . . 60 Figure 58 Social Protection Spending, Romania vs. EU- Figure 41 Awareness of Programs and Subsidies 27 and Selected CEE Countries . . . . . . . . . . . . . . . . . 88 to Support the Upgrading of Heating Systems or Figure 59 The Role of Social Transfers in Poverty Insulation by Gender and Age . . . . . . . . . . . . . . . . . 61 Reduction, Romania vs. EU Member States, 2021–22 (%) . . 89 Figure 42 Perceived Barriers to Heating and Insulation Figure 60 Social Transfer Programs: Size, Progressivity, Upgrades by Intentions to Upgrade . . . . . . . . . . . . . . 62 and Contributions to Poverty Reduction . . . . . . . . . . . 91 Figure 43 Barriers to Upgrading Either Heating or Figure 61 Social Protection Benefits in Romania— Insulation by Region . . . . . . . . . . . . . . . . . . . . . . 63 Coverage and Leakage Rates, 2019–20 . . . . . . . . . . . . . 94 Figure 44 Trusted Messenger for Information about Figure 62 Share of Energy Poor (P10) Accessing Social Heating and Insulation Upgrades . . . . . . . . . . . . . . . 64 Benefits by Type, 2019–20 . . . . . . . . . . . . . . . . . . . . 95 Figure 45 Perceived Enablers of Heating and Insulation Figure 63 Benefits Paid to the Energy Poor via the Upgrades by Intentions to Upgrade . . . . . . . . . . . . . . 65 Traditional Social System as Proportion of Total Figure 46 Influence of Different Factors on the Household Income, 2019–20 . . . . . . . . . . . . . . . . . . 95 Propensity to Upgrade Heating System . . . . . . . . . . . . 67 Figure 64 Short-Term Effects on Welfare by Figure 47 Influence of Different Factors on the Policy Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Propensity to Upgrade Insulation . . . . . . . . . . . . . . . 68 Figure 65 Coverage of Potential Mitigation Measures by Figure A.2.11 Average Energy Expenditure Shares by Poverty Groups . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Consumption Quintiles, 2021 . . . . . . . . . . . . . . . . . 137 Figure 66 Main Heating Source, 2023 (%) . . . . . . . . . . . 102 Figure A.2.12 Energy Spending Shares (Income-Based) by Income Quintiles, 2021 . . . . . . . . . . . . . . . . . . . 138 Figure 67 Share of Population Able to Control Temperature in Home, 2023 (%) . . . . . . . . . . . . . . . . 102 Figure A.2.13 Energy Spending Shares (Income-Based) by Consumption Quintiles, 2021 . . . . . . . . . . . . . . . 138 Figure A.1.1 Household Budget Survey Questionnaire for Romania . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Figure A.2.14 Energy Spending Shares (Income-Based) by Income Quintiles and Components, 2021 . . . . . . . . . 138 Figure A.1.2 Household Budget Survey Questionnaire for Romania . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Figure A.2.15 Energy Spending Shares (Income-Based) by Consumption Quintiles and Components, 2021 . . . . . 138 Figure A.2.1 Shares of Households with a Car by Income Quintile, 2020 . . . . . . . . . . . . . . . . . . . . . 133 Figure A.2.16 Energy Poverty Rates (Income-Based) by Measurement and Income Quintile, 2021 . . . . . . . . . . 139 Figure A.2.2 Energy Expenditure by Energy Source, 2021 . . 133 Figure A.2.17 Energy Poverty Rates (Income-Based) by Figure A.2.3 Households’ Energy Expenditure by its Measurement and Consumption Quintile, 2021 . . . . . . . 139 Components across Income Quintiles . . . . . . . . . . . . 134 Figure A.2.18 Energy Spending Shares (Income-Based, Figure A.2.4 Energy Expenditure Shares (Expenditure- No Car) by Income Quintiles, 2021 . . . . . . . . . . . . . . 140 Based) by Income Quintiles, 2021 . . . . . . . . . . . . . . . 135 Figure A.2.19 Energy Spending Shares (Consumption- Figure A.2.5 Energy Expenditure Shares (Expenditure- Based, No Car) by Consumption Quintiles, 2021 . . . . . . . 140 Based) by Income Quintiles and Components, 2021 . . . . 135 Figure A.2.20 Average Energy Expenditure Shares Figure A.2.6 Energy Expenditure Shares (Expenditure- and Energy Poverty Rates Using Expenditure-Based Based) by Consumption Quintiles, 2021 . . . . . . . . . . . 135 Measures, 2021 . . . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.2.7 Energy Expenditure Shares Figure A.2.21 Average Energy Expenditure Shares (Expenditure-Based) by Consumption Quintiles and and Energy Poverty Rates Using Income-Based Components, 2021 . . . . . . . . . . . . . . . . . . . . . . . 135 Measures, 2021 . . . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.2.8 Energy Poverty by Income Quintile, 2021 . . . . 136 Figure A.2.22 Average Energy Expenditure Shares and Figure A.2.9 Energy Poverty by Consumption Energy Poverty Rates Using Income-Based Measures Quintile, 2021 . . . . . . . . . . . . . . . . . . . . . . . . . . 136 and No Car-Related Energy Expenditures, 2021 . . . . . . . 142 Figure A.2.10 Average Energy Expenditure Shares by Income Quintiles, 2021 . . . . . . . . . . . . . . . . . . . . . 137 Acknowledgments This report is the work of a team led by Monica Robayo- Practice), Catalin Pauna and Andrei Silviu (Macroeconomics, Abril (Senior Economist, Poverty Global Practice) under the trade and Investment Global Practice), Mariano Gonzalez and guidance of Gallina Vincelette (former Country Director, Melisa Gaitan (Energy Global Practice). The team also included European Union), Marina Wes (Country Director, European Armanda Carcani (Program Assistant, Poverty Global Practice), Union), Salman Zaidi (former Practice Manager, Poverty Global who provided valuable support with processes; Daniel Practice), Ambar Narayan (Practice Manager, Poverty Global McNaughton, who edited the report; and Sara Ochoa who Practice), and Reena Badiani-Magnusson (Program Leader and designed the cover and Kilka Diseño Gráfico who designed Senior Economist). The core team included Jonathan Karver the rest of the report. Ramona Lipara, Iona Alexandra (Economist, eMBeD Unit, Development Impact (DIME)), Britta Irimia, and Corina Mirabela Grigore supported the country’s consultations. Laurin Rude (Young Professional, Poverty Global Practice), Ailin Tomio (Behavioral Scientist, eMBeD Unit, Development The team benefited from consultations in Romania with Impact (DIME)), Alessandro Silvestri (Consultant, Poverty public institutions officials, including the Ministry of Labor Global Practice and eMBeD Unit, Development Impact (DIME)), and Social Solidarity, the Ministry of Energy, and the Ministry and Kiyomi Cadena (Consultant, Poverty Global Practice). of Investments and European Projects. Consultations were also held with persons affiliated with think tanks, UN The authorship of the chapters is as follows: the executive agencies, the Romania Energy Poverty observatory, and summary chapter was written by Monica Robayo-Abril; development partners. chapters 1–3 were written by Monica Robayo-Abril and Britta Rude; chapter 4 was written by Jonathan Karver, Ailin Tomio, The team also acknowledges financial support for the and Alessandro Silvestri; chapter 5 was written by Monica qualitative data collection from the Whole of Economy Trust Robayo-Abril, Kiyomi Cadena, Jonathan Karver, and Ailin Fund. Tomio; and chapter 6 was written by Monica Robayo-Abril. The team would like to express its gratitude for the advice The team also benefitted from comments and discussions and guidance from peer reviewers Alexandru Cojocaru, with Nga Nguyen, Reena Badiani-Magnusson (Poverty Global Mariano Gonzalez, and Michelle Dugas. ABBREVIATIONS AND ACRONYMS ANRE National Energy Regulatory Agency HICP Harmonised Index of Consumer Prices AROP At-Risk-Of-Poverty IDI In-Depth Interview B20 Bottom 20 IHSN International Household Survey Network CAPP Clean Air Priority Program IMF International Monetary Fund CATI Computer-Assisted Telephone Interview ISR Social Reference Indicator CEE Central and Eastern Europe KGOE Kilogram Of Oil Equivalent CEQ Commitment to Equity KTOE Kiloton Of Oil Equivalent CT Cash Transfer KW Kilowatt EC European Commission KWH Kilowatt-Hour ECA Europe and Central Asia LIHC Low-Income High Cost ECAPOV Europe and Central Asia Poverty Database LPG Liquid Petroleum Gas LTRS Long-Term Renovation Strategy EPAH Energy Poverty Advisory Hub MTF Multi-Tier Framework for Measuring Energy ESMAP Energy Sector Management Assistance Program Access EU European Union M/2 Energy Expenditure Share below One Half of the EU EPOV European Union Energy Poverty Observatory Median EUROSTAT European Statistical Office NECP National Energy and Climate Plan EU-SILC European Union Statistics on Income and Living NRRP National Recovery and Resilience Plan Conditions PPP Purchasing Power Parity FGD Focus Group Discussion P10 Energy Expenditure Share above 10 Percent GDP Gross Domestic Product SCF Social Climate Fund GMI Guaranteed Minimum Income SEforALL Sustainable Energy for All GRM Grievance Redress Mechanism 2M Energy Expenditure Share at least Two Times the GWH Gigawatt-Hour Median HBS Household Budget Surveys WHO World Health Organization Executive Summary Why is addressing energy poverty key push more households into monetary poverty. Therefore, con- for economic development? sidering different population groups, a comprehensive analysis of the multidimensional aspects of poverty and vulnerability that Addressing energy poverty is paramount for economic de- considers different population groups is essential. velopment, given its close connection to income poverty. Moreover, global evidence suggests that residing in ener- Research indicates that lower-income households are dispro- gy-deprived circumstances adversely affects overall well-be- portionately affected by energy price increases. Such households ing, human development, and environmental outcomes. lack the financial means to absorb these shocks, which can lead Individuals in energy-poor households face an increased like- to decreased overall welfare. This vulnerability can result in lihood of developing respiratory and cardiac ailments due to households’ refraining from using energy or using less efficient uncomfortable temperatures and exposure to particulate matter and dirtier technologies and sources, particularly during winter, and experience mental health challenges linked to the stress of which poses health risks. Variations in impact exist across welfare paying for energy. Research spanning 50 developing countries distribution and population subgroups, with studies highlighting reveals that a reduction in energy poverty is associated with the heightened vulnerability of elderly populations and lower-in- improved health and education outcomes, with electricity ac- come households to energy poverty. Energy price increases thus cess exerting a more pronounced effect than energy use. The have the potential to exacerbate existing disparities as well as impact of energy poverty on mental health, exacerbated during economic crises, is evidenced by studies showing a correlation low-income households during Romania’s transition to a more with mental health issues, particularly depression. Environmen- efficient and renewable energy system. By gathering quantitative tal consequences, including deforestation and greenhouse gas and qualitative data and utilizing a mix of quantitative methods, emissions, contribute to public health concerns such as indoor including econometric models and microsimulation techniques air pollution and physical injuries during fuelwood collection along with qualitative approaches such as thematic analysis of and combustion. Furthermore, energy poverty disproportion- data collected via fieldwork, the report aims to comprehensively ately affects gender equity and educational prospects, particu- understand energy poverty in Romania and guide effective pol- larly for women who bear the physical risks of fuel collection. icy interventions. The World Bank’s new data collection in light Time-related effects, such as the need to gather fuel and attend of the post-Ukraine crisis provides an accurate and current as- to energy-related tasks, impact school attendance and contribute sessment of the situation, which is crucial for designing efficient to increased absenteeism due to illnesses. support programs that target the most vulnerable households in The imperative to measure and address energy poverty the current context. in Romania is underscored by the potential development benefits and the European Commission’s prioritization of Where do Romanian households this issue within the European Just Transition context. As stand regarding energy poverty and part of the Clean Energy for All Europeans legislative package, which households are more likely to the European Commission (EC) has established the EU Energy be affected? Poverty Observatory (EPOV) to assist Member States (now the Energy Poverty Advisory Hub, EPAH), emphasizing the im- In Romania, many households reported grappling with ener- portance of mitigating energy poverty for improved well-being, gy poverty, with there being a pronounced need for adequate environmental outcomes, and cost savings. In the Romanian warmth, cooling, lighting, and energy for regular appliances. context, addressing energy poverty is crucial, especially given Given the multidimensional nature of poverty, defining and the impact of the energy crisis on vulnerable consumers, ne- measuring it is challenging. However, underlying indicators cessitating prompt implementation of measures outlined in the reveal a high prevalence of energy poverty in Romania, indi- REPower EU package within the Romanian National Recovery cating that many people face challenges accessing and affording and Resilience Plan (NRRP). However, limited evidence exists essential energy services. Approximately 25 percent of the pop- comprehensively grasping the diverse factors influencing energy ulation experienced some form of energy poverty in 2021, with poverty in Romania, highlighting the need for robust research. households dedicating about 8.7 percent of their expenditures This report aims to fill that gap by providing evidence on critical to energy. High proportions of people grappling with this issue questions about the energy status of Romanian households. persisted, with 17.8 percent facing difficulties in paying utility The report focuses on four key topics, relying on existing bills in 2022, ranking among the EU’s highest. In 2022, 15.2 per- official household surveys and new qualitative and quanti- cent of households struggled to maintain warmth, among the tative data collected in June–July 2023, with the aim of pro- highest proportion in the EU, and even more concerning as this viding an updated snapshot of energy vulnerability among was a larger proportion than in 2020 (figure ES1). Cooling also Romanian households. The key topics include (1) Romania’s emerges as an important challenge, with 31 percent of house- standing on access to affordable, reliable, and sustainable en- holds struggling to keep their homes cool during the summer. ergy among households; (2) structural and behavioral barriers Energy poverty disproportionately affects more vulnerable and hindering households from transitioning out of energy poverty, low-income households. These disparities underscore equity with a focus on sustainable heating transitions; (3) the impact concerns across the energy consumption patterns of affluent of the rising energy prices on energy poverty and welfare; and and less-privileged households. (4) promising policy actions to reduce the energy burden on 14 Romania Energy Poverty Assessment Figure ES1 I n ab i l i ty to Keep Home Adeq u a te ly Wa r m , 2 02 2 vs . 2 02 0 ( % ) 25 20 15 10 5 0 Bulgaria Cyprus Greece Lithuania Portugal Spain Romania France EU27 Italy Malta Ireland Latvia Slovakia Croatia Germany Netherlands Belgium Denmark Poland Hungary Estonia Sweden Czechia Austria Slovenia Luxembourg Finland 2022 2020 Source: Eurostat 2023. https://ec.europa.eu/eurostat/databrowser/view/ILC_MDES01__custom_6037156/bookmark/table?lang=en&bookmarkId= 8f6604d8-6581-4f7b-adde-7a9e53a28caf Two critical aspects explored in this analysis of energy monetary and nonmonetary energy affordability measures in poverty in Romania are connectivity and energy affordability. light of issues such as leakages, housing conditions, and expo- The investigation into connectivity reveals that in 2021, almost sure to environmental problems. The concluding sections stress all households were connected to electricity, with negligible the need for targeted policies to address energy poverty efficient- differences across various population subgroups. However, dis- ly and highlight the potential role of energy-efficient renova- parities emerge concerning natural gas, with rural areas show- tions, especially for poorer households in rural and urban areas. ing lower coverage rates. The use of wood as a primary heating The overall assessment underscores the multifaceted nature of source, especially in rural regions, is highlighted, indicating the energy poverty in Romania and provides valuable insights for importance of considering diversified energy sources when as- policy makers. sessing spending patterns and vulnerabilities. The section on To devise effective strategies to tackle energy poverty, pol- energy affordability in this report delves into both monetary icy makers must acknowledge the significant heterogeneities and nonmonetary measures, reflecting the need for a range of in energy spending patterns and energy poverty rates across methodologies used in estimating energy expenditure shares. various population subgroups. Targeted interventions could The income-based energy expenditure share, excluding car-re- prove to be more cost-efficient and effective than a one-size-fits- lated expenses, is the most robust measure for Romania. This all approach because they address the unique circumstances of report employs three energy poverty indicators and finds that different demographic groups. approximately one-fourth of the Romanian population was en- Notably, the overlap between those considered energy ergy poor in 2021. The overlap and distinction between mone- poor and those experiencing monetary poverty is only par- tary and energy poverty underscore the need for targeted policy tial. This observation emphasizes the need for policy makers to interventions. adopt a distinct approach to tackling energy poverty, recogniz- The detailed analysis further reveals that energy spending ing that it is not solely an extension of monetary poverty, so that patterns and energy poverty in Romania are significantly in- targeting the income poor will not be enough to mitigate the fluenced by income levels, with low-income households and issue. Targeted interventions are required that are explicitly tai- other vulnerable groups experiencing higher vulnerability to lored to the challenges faced by energy-vulnerable households, energy price spikes. This report identifies specific population who may or may not be income poor. subgroups, such as single-elderly households, pensioners, and Energy-inefficient housing emerges as a prominent issue, those receiving social aid, as more affected by energy poverty particularly among poorer households and those residing in and thus necessitating tailored policy interventions. Addition- rural areas. Approximately one out of four households in the ally, the findings emphasize the importance of considering both bottom 20 percent of the income distribution (B20 category) Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 15 reported being affected by warm air leakages (from poor insu- Understanding the factors influencing energy expenditure lation), indicating a need for better infrastructure and housing shares, particularly the positive correlations with technologi- maintenance. Rural areas in particular face higher energy inef- cal and sociodemographic elements, has vital policy implica- ficiency rates than urban areas, highlighting the importance of tions. This analysis calls for tailored income-support measures targeted interventions to improve energy efficiency in these re- that can assist vulnerable households in managing their energy gions. Previous evidence suggests that social housing still needs costs effectively. Policy makers should consider targeted inter- to be developed in Romania, which could alleviate energy-re- ventions for households with characteristics such as being single lated challenges for vulnerable populations. elderly, having multiple unemployed members, or being female headed, given their higher energy expenditure shares, particu- What are the key structural and larly in times of crisis and in the short term. Additionally, there behavioral barriers that prevent is a need for support programs to enhance these households’ households from transitioning out of energy efficiency, because such programs can lead to long-term energy poverty and adopting cleaner savings and reduced energy-related financial burdens. Addition- technologies and sustainable energy ally, promoting energy-efficient heating and cooking technolo- use? gies could contribute to household cost savings. The tailoring of policies to address these specific factors represents an opportu- We observe distinct correlations between technological fac- nity to enhance energy affordability and efficiency, particularly tors and sociodemographic characteristics in understanding for vulnerable demographic groups. energy expenditure shares. The variables positively correlated The higher vulnerability to energy poverty of households with energy expenditure shares include technological aspects of the single elderly and pensioners with limited and fixed such as sources of energy for heating (wood pellets, natural incomes also underscores the importance of targeted inter- gas, etc.) and for cooking (electricity, natural gas, etc.). Socio- ventions. Prioritizing energy-efficient housing renovations demographic factors such as households’ being single elderly, and raising awareness among low-income households about having unemployed members, being headed by a female, and environmental issues are crucial components of a comprehen- making use of specific in-house sewage types also show pos- sive strategy. itive correlations with increased energy expenditure shares. Another factor influencing energy expenditure shares and This analysis provides clear insights into the diverse drivers of poverty rates is the ownership of or access to resources capa- energy spending. ble of generating energy, such as wood plantations or illegal The analysis suggests a link between energy expenditure logging. Research suggests that using wood for cooking and shares and the technology employed for cooking and heat- heating is particularly significant in rural areas. Future studies ing, with income levels also playing a role. However, a more should strive to estimate the potential impact of these factors in-depth investigation is required to understand this relation- on the biasing of energy expenditure shares and, consequently, ship fully that considers factors such as connectivity, household energy poverty rates. preferences, the energy efficiency of technologies in use, and This study also investigates residential sustainable energy associated costs. While income has a role, its impact is relatively transitions, which hinge on both technological upgrades and minor. Certain household types, such as single elderly, those attitudinal shifts, as a way to move out of energy poverty. The with many unemployed members, and those headed by a fe- low level of intention to upgrade heating systems in Romania male, were more likely to report higher energy spending shares. underscores the challenges in motivating behavioral change Policy makers addressing energy poverty should prioritize in- that go beyond financial constraints, emphasizing the need for terventions for these households. However, caution is advised in inclusive, targeted policy measures and clear communication to interpreting energy expenditure shares, because they may be in- support renewable energy adoption. Tailored financial support fluenced by factors like connectivity and access to resources that mechanisms that acknowledge non-financial motivations in households do not have to pay for (like firewood collected from their design and implementation are essential to ensure inclu- properties or nearby forests) , potentially biasing the results. sive, sustainable energy transitions. 16 Romania Energy Poverty Assessment We analyze the barriers to and enablers of sustainable especially electricity-based ones, is crucial. Low institutional heating transitions, focusing on financial, information, and trust in government support programs is widespread, which attitudinal factors. Financial concerns stem from the costs of is attributed to a perceived lack of transparency and concerns both the initial investment and ongoing use. Financial barriers, about favoritism in fund allocation. particularly affordability concerns, emerged as the most critical As of July 2023, despite nearly 7 out of 10 Romanians’ among participants. The perceived difficulty in bearing the costs perceiving upgrades as an inconvenience, a similar propor- of upgrading heating systems is high, with 80 percent expressing tion of the population knew of the benefits of making these that financial difficulties would be expected to sustain the costs improvements. This indicates a potential conflict between the of an upgrade. These barriers vary across regions, with South- perceived inconvenience of upgrades and the understanding of East Romania facing the highest financial difficulty. Informa- their positive impact on energy efficiency and cost savings. Rais- tion barriers include limited awareness of programs supporting ing awareness about the advantages of upgrades and dispelling heating upgrades, with only 11.3 percent of those surveyed fully misconceptions could encourage more Romanians to consider understanding the types of help available. Attitudinal barriers adopting energy-efficient technologies and practices. involve negative beliefs, such as the perceived inconvenience of Despite these barriers, there are clear attitudinal enablers. upgrades and the importance of social influence. Survey estimates from the 2023 World Bank rapid surveys indi- Financial and knowledge constraints are significant barri- cate that most Romanians were aware of their energy consump- ers hindering many households from upgrading their energy tion at home. This awareness can contribute to more informed sources. A small share of the population reported knowing of decisions regarding energy usage and expenditure, potentially existing support programs that could offer financial assistance leading to more efficient energy management. Approximately 70 for such upgrades. Enhancing the dissemination of informa- percent of respondents were aware of the benefits of upgrades, tion about these programs could be crucial in incentivizing and such as lower bills, convenience, increased property value, and enabling more households to make energy-efficient upgrades positive environmental impacts. Trusted messengers, such as (mainly as these are related to sustainable heating practices), friends and family and independent technicians, play a crucial ultimately contributing to the reduction of energy poverty and role in influencing decisions. Beliefs in the benefits of modern negative environmental impacts. heating systems and community action, along with trusted mes- Lack of knowledge and awareness of support programs sengers, are identified as vital enablers. and the complexity of participation further hinder sustain- Our evidence points to a concerning trend where poor- able transitions. A lack of awareness about existing subsidy er households show lower awareness and concern about en- programs is evident, leading to sentiments of exclusion among vironmental problems. This suggests that addressing energy potential beneficiaries. Technical challenges in the online appli- poverty and environmental sustainability must be approached cation process further hinder participation, especially by older in a way that considers economically disadvantaged households’ individuals and in regions experiencing rapid fund depletion. unique circumstances and challenges. Simultaneously, certain In July 2023, a survey revealed that approximately half vulnerable groups, such as households with a high number of the population in Romania would consider upgrading to of members, report higher exposure to environmental issues, appliances that are more energy efficient. However, only a making it imperative to implement measures that support these negligible share of respondents expressed interest in upgrading communities’ efforts to achieve energy efficiency and protect the their home’s heating system or insulation. Encouraging more environment in which they live. widespread adoption of energy-efficient technologies and home improvements could significantly contribute to reduc- What are the welfare implications of ing energy poverty and enhancing overall energy sustainability rising energy prices? in the country. Negative beliefs about upgrades include concerns over the The global and European Union surge in energy prices fol- complexity and safety of modern heating systems. Building lowing Russia’s invasion of Ukraine has heightened the ur- trust in the efficiency and sustainability of modern systems, gency to tackle energy poverty effectively. We simulate the Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 17 potential impacts on energy poverty and welfare of the rising understanding of the implications of energy price changes on energy prices. Our baseline simulations show a moderate in- the well-being of diverse demographic groups. crease in energy poverty, with some groups significantly more affected. In our baseline scenario of a 40-percent increase in What measures are in place and what energy prices, a significant impact is expected. On average, in potential mitigation measures can be our simulation energy poverty (P10)1 increases by 2.3 percent- devised to tackle energy poverty and age points across the population. Vulnerable groups, including protect the most vulnerable? households receiving support from municipalities, those receiv- ing disability benefits, and those consisting of a single-elderly In Romania, addressing energy poverty is a key priority, in person, are disproportionately affected compared to other de- line with EC goals. The legal framework, notably Law 226/2021, mographic groups. Moreover, the rise in energy poverty rates is introduces a novel definition of energy poverty, emphasizing the more pronounced at the lower end of the welfare distribution, inability of vulnerable energy consumers to meet “minimum indicating that the most economically disadvantaged segments energy” needs. The National Strategy on Social Inclusion and of the population bear the brunt of these price increases. No- Poverty Reduction focuses on thermal insulation programs and tably, the second-lowest income quintile would see a substan- monthly assistance for heating expenses. Reintroducing regulat- tial increase of 10.1 percentage points in at-risk poverty rates, ed prices has led to implicit energy subsidies. underscoring the severity of the impact on this specific group. The National Long-Term Renovation Strategy (LTRS) The simulation also shows that the direct effects of a recognizes energy poverty, emphasizing an improved legal 40-percent energy price increase could also lead to an over- framework and collaboration with local governments. Mem- all increase in at-risk poverty rates of 2.2 percentage points. ber States, including Romania, within the European Union (EU) Energy-poor households are the most affected, facing higher must address energy poverty in their National Energy and Cli- poverty rates than non-energy-poor households. Additional- mate Plans as part of the European Green Deal. The Renovation ly, rural areas experience more significant impacts than urban Wave initiative and the EU’s legal framework mandate action to areas, highlighting the disparities in energy affordability and combat energy poverty and achieve energy efficiency goals. The poverty between different regions of the country. Social Climate Fund (SCF), which was created in June 2023, Analysis of the results of the simulation using the $6.85 a allocates funds to Romania to address energy and transport pov- day (2017 PPP2) international poverty line shows that house- erty from 2026 to 2032, supporting policy measures including holds with unemployed members are most affected, although temporary income support and investments in energy efficiency the overall impact on poverty rates is less than in the baseline and building renovation. scenario. This underscores the importance of selecting appro- The Romanian government has implemented a multifac- priate poverty metrics when evaluating the effects of energy eted approach to address energy poverty, particularly in re- price fluctuations on vulnerable populations. Sensitivity testing sponse to the recent energy crisis. Key policy measures include with different energy price elasticity assumptions consistent- untargeted energy price caps, income-support initiatives, and ly shows adverse effects on energy poverty rates, emphasizing energy efficiency programs. The untargeted measures involve the issue’s significance for policy makers and social welfare capping electricity and natural gas prices to shield consumers advocates. Using welfare indicators from the European Union from steep hikes. Despite their ease of implementation, discus- Statistics on Income and Living Conditions (EU-SILC) data re- sions persist on their fairness and effectiveness, given their fiscal inforces the validity of the results, which offer a comprehensive impact and need for more targeting. Targeted income-support 1 Energy poverty in this case is defined as the proportion of households who spend more than 10 percent of their household income on energy. 2 The international poverty line is expressed in 2017 purchasing power parity (PPP) dollars to account for price differences across countries and to reflect changes in the cost of living over time. The World Bank uses PPPs to convert different currencies into a common unit and to derive the poverty line. 18 Romania Energy Poverty Assessment measures are implemented via Romania’s social protection and ensuring an equitable transition to sustainable system, assisting vulnerable families through means-tested energy practices. programs. These include heating subsidies for the cold season, 3. Strategic Targeting and Fiscal Viability: Policymak- which, despite being progressive, face challenges in reducing ers should adopt well-targeted measures that balance poverty due to limited benefit adequacy. The government has cost-efficiency with fiscal viability. Prioritizing recipi- also introduced energy cards and vouchers for specific groups, ents based on specific needs rather than a universal ap- such as pensioners and those with disabilities, to mitigate exces- proach prevents the leakage of resources to non-needy sive heating expenditures. individuals and thus makes possible the provision of In addition to income support, energy efficiency measures more generous benefits to the deserving population. play a crucial role in tackling energy poverty. Programs like 4. Adjustment of Benefit Levels: Regularly adjusting the National Multiannual Program for the Improvement of En- benefit levels to match the increasing cost of living is ergy Performance in Blocks of Flats and the Renovation Wave critical. Mechanisms to index government benefits and focus on enhancing the energy efficiency of buildings, especially tax credits should be implemented to maximize their in marginalized areas. These initiatives provide financial sup- impact in reducing poverty and prevent households port for building renovations and prioritizing projects in urban from slipping into poverty due to erosion in the value or rural regions with populations at risk of poverty and social of benefits. exclusion. The government’s commitment to green projects, in- cluding energy efficiency, is evident in NRRP allocations. These The outcomes of the policy simulations provide valuable comprehensive efforts aim to alleviate the impact of rising ener- insights into effective strategies for addressing energy poverty gy costs on vulnerable households and contribute to long-term and safeguarding vulnerable populations in Romania. Overall, solutions for energy poverty in Romania. our simulations show that targeted income-support approaches The Romanian government should adopt a two-pronged are effective and cost-efficient in mitigating the impact of energy strategy based on overarching principles and policy simu- price increases. Across many EU nations, price caps have been set lations to mitigate energy poverty and protect vulnerable in place to protect consumers from rising energy prices, coupled populations. The following overarching principles emerge with certain forms of financial aid. However, this restriction on from the analysis: energy costs often benefits higher-income groups, is less effec- tive than precisely targeted social assistance, and can strain fis- 1. Social Safety Nets and Energy Efficiency: Creating cal resources significantly. By tailoring interventions to address effective social safety nets is crucial for short-term re- the specific needs of different vulnerable groups, policymakers lief from energy poverty. Simultaneously, prioritizing can optimize resource allocation and achieve more-substantial household energy efficiency measures can enhance outcomes in alleviating energy poverty. Among the targeted ap- overall welfare in the medium term. While providing proaches, focusing on single-elderly households and pensioners financial assistance to individuals in the immediate at risk of poverty stands out as particularly efficient and feasible, future is essential to help them cope with current eco- because these households are relatively easy to identify and assist. nomic challenges, it is equally important to implement However, it is important to acknowledge other equity concerns, strategies that will ensure sustainable benefits over a such as intergenerational considerations, when designing inter- longer period. ventions. Striking a balance between targeted support for specific 2. Targeted Policies for Vulnerable Groups: Pol- groups and maintaining a fair distribution of benefits across so- icies should be designed based on an under- ciety remains a critical aspect of policy development. standing of energy expenditure patterns and The simulations explore two distinct policy scenarios, affordability across income levels. Identifying vulner- each offering a nuanced approach to tackling the challenges able groups is essential for formulating policies prior- posed by rising energy prices. itizing socioeconomically disadvantaged individuals Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 19 Policy Scenario 1 – Income Support duction could be achieved by targeting certain vulnerable Measures Financed by External groups. Although income-support measures are slightly more Funding challenging to administer than price caps, they offer a cost-ef- fective and likely more efficient approach to assisting the most vulnerable with managing higher prices. Notwithstanding that In this scenario, income support measures are proposed and eliminating price caps implies that energy prices would increase financed through external funding sources, particularly from even further, targeted measures could mitigate that adverse ef- the SCF. The emphasis is on targeting vulnerable groups, with fect. In particular, focusing on pensioners at risk of poverty is a specific focus on single-elderly households and pensioners expected to impact poverty reduction significantly. If the fiscal at risk of poverty. In households with pensioners, whose in- revenues gained from eliminating price caps were to finance come sources are primarily fixed, social program coverage rates cash transfers for pensioners, at-risk poverty could fall below among the energy poor lag behind the overall population. This pre-price-increase levels. suggests pensioners are not being adequately supported de- The microsimulation results emphasize the advantage of spite the efforts to address energy poverty in various groups. targeted cash transfer programs in mitigating poverty caused The simulations reveal that an approach focusing on vulnerable by energy price increases. Focusing on vulnerable groups, such groups is cost-effective and efficient in mitigating the adverse as single-elderly households and pensioners, proves to be more impacts of rising energy prices. By directing financial assistance efficient and impactful. These programs result in substantial to these identified at-risk groups, the simulations demonstrate reductions in poverty and the poverty gap, underscoring the a substantial reduction in poverty and the poverty gap. This tar- importance of tailoring interventions to the unique needs of geted strategy optimizes resource allocation and ensures that identified at-risk demographics. The findings reinforce the no- the most vulnerable segments of the population receive the tion that a nuanced, group-specific approach to cash transfers necessary support. is essential for achieving meaningful and sustainable poverty reduction in the context of energy price fluctuations. Policy Scenario 2 – Income Support Clear communication is vital to garner public support Measures Financed by the Phasing for policy interventions. The evidence in this report also Out of Energy Price Caps draws attention to the crucial role of government support and effective communication strategies in the success of new The second scenario involves financing income support mea- measures. Insights from a World Bank survey underscore the sures by phasing out energy price caps. Despite the potential significance of clear communication regarding the benefits and for an increase in energy prices, this policy scenario aims to use costs associated with policy interventions to safeguard the en- fiscal revenues gained by eliminating price caps to fund targeted ergy poor. As revealed by the survey, public perceptions provide assistance. The simulations highlight the critical importance of valuable input for policy makers to ensure that modifications in supporting single-elderly households and pensioners. By doing existing protection schemes related to energy poverty garner so, the policy proves effective in mitigating adverse effects, pro- widespread support. This highlights the need for a transparent viding a cushion against rising energy prices, and, remarkably, and well-communicated approach to secure public trust and co- reducing poverty levels beyond the pre-price-increase baseline. operation in implementing energy poverty alleviation measures. This underscores the resilience and effectiveness of targeted in- It is critical to accompany these income support measures, come support measures, even in the face of a more challenging which are more short-term, with some medium-term ener- economic landscape resulting from the removal of price caps. gy efficiency measures. By combining immediate financial aid During the winter of 2022/23, price caps were introduced with longer-term strategies for energy efficiency, we can create as a response to the energy crisis that aimed at mitigating a more resilient and sustainable support system that addresses the impact of rising energy costs on households. If these both current economic pressures and future energy challenges. caps were eliminated and income-support measures were Policy interventions such as energy-efficiency-upgrade financed instead, a more significant impact on poverty re- support programs should also be designed to support sus- 20 Romania Energy Poverty Assessment tainable energy transitions of both energy-poor/vulnerable ensure a holistic and enduring impact. Secondly, incorporat- and non-energy-poor/vulnerable households. While interven- ing insights from behavioral science is critical to enhancing tions directed at stabilizing energy prices are necessary, these energy efficiency initiatives. Focusing on behavioral aspects must be complemented with programs that facilitate sustain- can significantly contribute to addressing energy inefficien- able energy transitions (especially related to heating) among all cies, particularly in heating systems, with a specific emphasis relevant populations in Romania. Such programs are crucial, 3 on low-income households. To achieve sustainable transitions, because of the inefficient use of electricity-based heating in the it is crucial to expand existing renovation programs, ensuring coldest months to supplement inefficient and dirty traditional their coverage extends to energy-poor households and rural methods of heating using solid fuel (for example, firewood and areas, thereby fostering inclusivity and effectiveness in energy coal). There are only a few programs in Romania that subsidize efficiency improvements. energy-efficient upgrades (such as new heating systems and in- In summary, a comprehensive strategy involving targeted sulation). Even when these programs exist, awareness of them income support, strategic policy simulations, and a focus on is limited and trust in the institutions tasked with supporting sustainable energy transitions, guided by behavioral science sustainable energy transitions is even more limited. Programs to insights, can effectively mitigate energy poverty and protect support these transitions holistically are needed; insights from the most vulnerable populations in Romania. Future research behavioral science can inform their design and implementation should address knowledge gaps, particularly in understanding to benefit energy-poor and vulnerable households the most. the dynamics of wood usage, in order to refine strategies to com- Strengthening energy efficiency initiatives involves bat energy poverty in Romania. adopting a comprehensive approach that considers medi- Finally, given the multidimensional nature, addressing um-term strategies and insights from behavioral science. energy poverty requires a multisector approach. This involves First, sustainable energy transitions necessitate medium-term energy, transport, infrastructure, and social sectors, as well as approaches that facilitate shifts in technologies, fuels, and us- collaboration across various geographic levels (central and local age practices. Prioritizing upgrade support programs geared levels). Effective measures should engage diverse stakeholders, toward sustainable transitions is imperative, and these should including marginalized groups, to ensure comprehensive track- be given due consideration alongside short-term measures to ing, analysis, and the development of informed public policies. 3 Sustainable energy transitions can be understood as the shift toward cleaner and more-energy-efficient use of energy through the adoption of modern energy-intensive technologies (for example, modern heating devices) and complementary upgrades (for example, insulation). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 21 Chapter 1 Motivation In Romania, a large share of households experience some nerable and low-income households, indicating a meaningful form of energy poverty conditions, meaning they lack some of distinction in energy consumption patterns between the affluent the most essential features of ordinary life: adequate warmth, and the less privileged, which has implications for equity. cooling, lighting, and energy required to power typical ap- Tackling the problem of energy poverty is one of the most pliances. Though there is no simple definition and measure of urgent challenges that European policy makers are facing, as poverty, given its multidimensional nature, underlying indica- the current energy crisis that emerged after Russia’s invasion tors show the prevalence of energy poverty is high. When using of Ukraine has raised significant concerns about house- expenditure measures, roughly 25 percent of the population in holds’ vulnerability to energy prices and rising energy Romania experienced some form of energy poverty in the year poverty. Figures 1 and 2 show the trajectory of natural gas 2021. During the same year, households in Romania allocated and electricity prices for household consumers biannually about 8.7 percent of their total expenditures to energy. In 2022, between 2008 and 2023 in Romania and the EU-27 region. 15.2 percent of households struggled to maintain warmth within The figures clearly show a steep increase after the outbreak their homes, and 17.8 percent of households faced difficulties of the war for both electricity and gas. Because energy is a with unpaid utility bills, proportions that ranked among the consumption good that is difficult to replace by alternative highest in the EU. These figures were even higher among vul- sources, especially during winter, these price increases most likely imposed significant financial stress on households. In disposable income and lower savings to mitigate these unex- 2023, although energy inflation was declining, the country pected price shocks, and consequently, their overall welfare may continued to face elevated energy prices. In addition, rising decrease. Poor households might even refrain from using energy energy prices and energy affordability remained an import- altogether, which could result in health deprivations, especially ant concern for Romanians in 2023, as shown later in this during winter months. There may be differences along the wel- report. This is important, as gaining insights into how the fare distribution and across different population subgroups. For public perceives energy is crucial for shaping energy policies example, some studies show that elderly populations are often in the future. more affected by energy poverty (Cong et al. 2022), and others Addressing energy poverty is critically important for eco- show that lower-income households bear the greatest burden nomic development for various compelling reasons. First, of rising energy prices (UKONS 2022). Energy price increases energy poverty is closely linked to income poverty. Previous could, therefore, have important equity effects when analyzing research shows that poorer households are often more affect- the problem from a multidimensional viewpoint on poverty and ed by energy price increases because they have less disposable vulnerability. Another concern is that the additional financial income and less savings as a cushion against these increases constraints could push more households into monetary pover- in household expenditure. Previous research on the impact of ty. Based on this rationale, studying the effects of energy price energy price increases shows that poorer households tend to be increases along the welfare distribution and for different popu- more affected (UKONS 2022; World Bank 2019). They have less lation groups is crucial. Figure 1 N atural G as Pri ces for Hous eh o ld Figu re 2 Ele c t r ic ity Pr ice s fo r H o u s e ho ld Consume rs , Roman i a vs . th e EU27, 2 008 s2 –2 023 s2 Co ns u m e rs , Ro m a nia vs . t he E U 27 , (Half-Yea rly ) 2 0 0 8 s2 –2 023 s2 ( H al f-Yea r ly) .120 Price(Euro per Kilowatt-hour) Price (Euro per Kilowatt - hour) .10 .40 .350 .080 .30 .250 .060 .20 .040 .150 .10 .020 .050 .0 .0 2008-S2 2009-S2 2010-S2 2011-S2 2012-S2 2013-S2 2014-S2 2015-S2 2016-S2 2017-S2 2018-S2 2019-S2 2020-S2 2021-S2 2022-S2 2023-S2 2008-S2 2009-S1 2009-S2 2010-S1 2010-S2 2011-S1 2011-S2 2012-S1 2012-S2 2013-S1 2013-S2 2014-S1 2014-S2 2015-S1 2015-S2 2016-S1 2016-S2 2017-S1 2017-S2 2018-S1 2018-S2 2019-S1 2019-S2 2020-S1 2020-S2 2021-S1 2021-S2 2022-S1 2022-S2 2023-S1 2023-S2 EU27 Romania EU27 Romania Source: Eurostat 2023 https://ec.europa.eu/eurostat/databrowser/ Source: Eurostat 2023 https://ec.europa.eu/eurostat/ view/NRG_PC_202/default/table?lang=en. databrowser/view/NRG_PC_204__custom_7279167/default/ Note: Reported Prices include all taxes and levies and are for table?lang=en. consumption band D2 (Consumption from 20 to 199 GJ). Subsidies and Note: Reported Prices include all taxes and levies and are allowances are included in the final price paid by the consumers, as for consumption band 2 (Consumption from 2,500 to 4,999 of reference period 2023s2. For these semestrial prices, the reference kWh). Subsidies and allowances are included in the final periods are from January to June for semester 1 and from July to price paid by the consumers, as of reference period 2023s2. December for semester 2. More recent data are not available. For these semestrial prices, the reference periods are from January to June for semester 1 and from July to December for semester 2. More recent data are not available. 24 Romania Energy Poverty Assessment Second, residing in circumstances of energy deprivation the European Just Transition context, energy poverty has gained brings about adverse effects on overall well-being, human de- significance throughout the EU. Reducing it can lower health velopment, and environmental outcomes. Individuals living care costs, improve the environment, and, most importantly, in energy-poor households tend to have a higher probability of enhance the well-being and comfort of vulnerable households developing respiratory and cardiac ailments or having existing while achieving cost savings and resource protection. As a result, ailments worsen due to uncomfortable temperatures and men- the EC has prioritized the tackling of energy poverty in Europe tal health caused by the stress linked to the inability to cover through the Clean Energy for All Europeans legislative package energy expenses (Liddell and Morris 2010). Lower energy pov- and has established the EU EPOV (now EPAH) to assist Mem- erty was associated with higher health and education outcomes ber States in addressing and mitigating energy poverty. for 50 developing countries from 1990 to 2017 (Banerjee et al. Until now, European member states have implemented 2021), with electricity access having a more pronounced effect various policy measures to tackle energy poverty and safe- than energy use. Finally, further evidence for the EU before guard vulnerable consumers. These measures fall into five and during the 2008 economic crisis shows how energy poverty broad categories: (1) financial assistance initiatives as part of the worsened and its negative impact on health increased during general social support or targeted energy or heating subsidies the economic crisis. This association was stronger and more funded indirectly through social tariffs; (2) measures to protect susceptible to the effect of the economic crisis in the mental consumers’ rights, such as disconnection protection safeguards; health dimension, particularly in depression (Oliveras et al. (3) actions to increase energy efficiency so as to reduce energy 2021). Polimeni et al. (2022) assess the impact of energy pov- consumption; (4) measures related to providing essential infor- erty on health in the EU-27 countries and found that arrears mation and awareness (that is, measures offering improved bill- on utility bills exerts positive long run effects on current health ing information and adhering to utility codes of conduct during expenditures, self-perceived health and on the capacity to keep consumer interactions and one-stop shops); and (5) measures to the home adequately warm. The environmental consequences simplify and streamline the process of energy efficiency building of energy poverty involve deforestation, alterations in land uti- renovations. In Romania, the investments and changes outlined lization, and the release of greenhouse gases. Furthermore, en- in the REPowerEU package within the NRRP must prioritize ergy poverty is associated with serious public health concerns addressing energy poverty, as will be explained in more detail related to indoor air pollution, physical injury during fuelwood in chapter 5. This should involve measures that can be imple- collection, and limited refrigeration and medical care in areas mented promptly to alleviate the impact of the energy crisis, lacking electricity (Sovacool 2012). particularly among the most vulnerable consumers. Moreover, energy poverty has repercussions on gender Nevertheless, there is limited evidence concerning the di- equity and the educational prospects of individuals of all verse factors influencing energy poverty in Romania. The quan- ages. The gender-related consequences mainly revolve around tity and thoroughness of research pertinent to energy poverty is the physical risks of fuel collection and indoor air pollution’s also affected by the challenges in the definition and measurement health effects, particularly for women who shoulder the burden of the phenomenon. As discussed later in chapter 2, there is a lack of fuel and stove expenses. Additionally, there are time-related of consensus on how energy poverty should be conceptualized effects derived from the completion of tasks such as gathering and measured in the EU. In a comparative context, a significant fuel and water, cooking, and attending to the needs of unwell portion of the evidence regarding the fundamental reasons be- children. In terms of education, there are effects on school at- hind energy poverty has originated from research on “fuel” pov- tendance due to time spent on various energy-related activities erty conducted in the United Kingdom (for example, Bridgen and and increased absenteeism resulting from illnesses (Masud et Robinson, 2023). New evidence on energy poverty for the EU al. 2007; Gaye 2007). focused on monitoring and measurement issues, drivers, and the In this context, identifying and understanding the root policy response across member states shows a renewed interest causes and addressing energy poverty through an effective in the topic in recent years (Maxim et al. 2016; Bouzarovski 2014; policy mix is critical in Romania, given the potential devel- Harriet et al. 2017; Ramos et al. 2022; Bouzarovski et al. 2021). opment benefits; the EC has also prioritized this task. Within With respect to Romania in particular, evidence on the incidence Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 25 and drivers of energy poverty is growing (Clodnitchi and Busu Deal? What is the role of sustainable energy pro- 2017; Murafa et al. 2017; Sinea et al. 2018; Sinea et al. 2019; Jiglau grams and policies and how can behavioral science et al. 2020; Sinea et al. 2021, Murafa, 2022; Vornicu-Chira et al. inform these? 2024). Teschner et al. (2020) examine extreme energy poverty in Roma neighborhoods in Romania, mapping norms, policies, To produce this evidence, we mainly relied on official and regulations, analyzing its main characteristics and challenges, household surveys. However, given the importance of updat- and discussing the nexus of infrastructure, planning, and social ed data in the wake of Russia’s invasion of Ukraine, we col- inequality. Vornicu-Chira et al. (2024) identify energy and trans- lected new quantitative and qualitative data in 2023 to give port vulnerabilities in Romanian households, assess the welfare an updated picture of energy affordability and willingness to impact of the EU Emissions Trading System (ETS 2), and propose move toward cleaner and sustainable heating among Roma- targeted measures and recommendations for national Social Cli- nian households. We use the 2021 Household Budget Surveys mate Plans. Understanding the patterns and addressing the fac- (HBS), the 2020 European Union Survey of Income and Living tors associated with energy poverty in the country is essential to Conditions (EU-SILC), and new data collected by partnering with guiding policies that alleviate it and to find the policy mix that (1) a local research firm. Notably, the official household surveys do increases household welfare and enhances households’ ability to not capture the current situation of households amid the ongoing pay their bills, thereby reducing their energy costs (through lower energy crisis. Accurate and updated data on energy poverty are prices or social support); and (2) reduces their energy consump- critical for assisting decision-makers in creating efficient support tion (by improving energy efficiency or through information and programs that focus on the most vulnerable households. There- awareness campaigns). fore, the World Bank undertook the collection of new data in This report aims to fill this gap by producing evidence to June and July 2023 by means of a quantitative phone survey and answer the following four key questions: a qualitative instrument designed to capture beliefs, attitudes, and behaviors related to sustainable energy use (heating in particu- • Where does Romania stand regarding access to afford- lar) among vulnerable groups. We used a mix of quantitative and able, reliable, and sustainable energy? How do the en- qualitative methods, ranging from descriptive analysis, economet- ergy burdens change across Romanian households and ric models, microsimulation techniques (from the quantitative what factors are associated with energy poverty? survey), and thematic analysis (from the qualitative fieldwork). • What key structural and behavioral barriers prevent The report is structured as follows. Chapter 2 takes stock of households from transitioning out of energy poverty where the country stands regarding energy poverty and access and adopting cleaner technologies and sustainable en- to affordable, reliable, and sustainable energy and potential driv- ergy use? ers. Chapter 3 presents the structural and behavioral barriers • How has the recent rise in energy prices due to Russia’s that hinder households from moving away from energy poverty invasion of Ukraine potentially affected energy poverty and embracing cleaner and more sustainable energy practices and household welfare? by investing in modern heating devices and practices. Chapter • What policy actions offer the most significant promise 4 evaluates the potential ex-ante impact of recent energy price for reducing the energy burden of low-income house- increases on energy poverty and welfare. Chapter 5 presents holds as Romania transitions to a more efficient and potential mitigation measures based on identified patterns and renewable energy system under the European Green barriers to tackling energy poverty. Finally, Chapter 6 concludes. 26 Romania Energy Poverty Assessment Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 27 Chapter 2 Access to Affordable, Reliable, and Sustainable Energy among Romanian Households Access, affordability, and energy efficiency among households proving the level of service provided. Improving household en- are critical aspects that significantly influence household ergy efficiency reduces energy bills, enhances affordability, and well-being; understanding and addressing these factors is es- has positive environmental implications by lowering greenhouse sential for designing policies aiming to ensure a sustainable gas emissions. As shown in this report, addressing energy pov- and equitable energy landscape. Access to reliable and modern erty involves implementing a policy mix that promote energy energy sources is fundamental for meeting basic needs and im- efficiency, subsidizing energy costs for vulnerable households, proving the quality of life for households. Energy affordability and providing targeted support to those in need. refers to the ability of households to access energy services with- This chapter characterizes the patterns of energy access out experiencing undue financial hardship. Rising energy prices and affordability for the Romanian households, overall and can disproportionately affect low-income households, pushing across various population segments, along with the welfare them into energy poverty. Finally, energy efficiency measures distribution. First, we briefly describe the energy sector land- aim to reduce energy consumption while maintaining or im- scape, which can affect energy prices, and then look at different dimensions of access and energy affordability among Roma- in 2020. The country mainly relies on natural gas, oil, and petro- nian households. We then look at different measures of energy leum products (excluding the biofuel portion). Its dependency poverty—when people lack access to energy because of poor ratio in gross available energy, 28.2 percent, is among the lowest in infrastructure or excessively high initial costs—and estimate the the European Union (figure 4). However, the dependency ratio is incidence rates of energy poverty from both a monetary and a exceptionally high for oil and petroleum products (excluding the nonmonetary standpoint. To understand their equity implica- biofuel portion) at 64.7 percent4. Russia’s imports account for 17 tions, we explore the different patterns of how the wealthiest and percent of the energy mix5. This might stabilize energy prices, as the poor and other vulnerable groups consume energy. the country might rely less on external energy sources subject to price fluctuations. A diversified energy mix with lower dependen- 2.1. Energy Sector Background cy on external energy sources could reduce energy vulnerability, because greater energy self-sufficiency can enhance the country’s Romania has one of the lowest dependency ratios in gross resilience in the face of disruptions in energy supply, thereby de- available energy in the EU. Figure 3 shows Romania’s energy mix creasing vulnerability to energy-related crises. Figure 3 En erg y M i x i n Roman i a, 2 02 0 100% Other* 90% 80% Electricity (net imports) 70% Nuclear heat 60% Non-renewable waste 50% 40% Renewables and biofuels 30% Oil and petroleum products 20% (excluding biofuel portion) 10% Natural gas 0% Solid fossil fuels European Union Romania 27 countries (from 2020) Source: Eurostat (2023). https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Archive:EU_energy_mix_and_import_dependency#Energy_mix_and_ import_dependency Note: The figure plots the energy mix in Romania as a share of fuels in gross available energy. This shows how much the country depends on gas, oil, coal, nuclear, and renewables in its energy mix. Gross available energy means the overall supply of energy for all activities on the country’s territory. This also includes energy transformation (including generating electricity from combustible fuels), support operations of the energy sector itself, transmission and distribution losses, final energy consumption (industry, transport, households, services, agriculture, and so on), and the use of fossil fuel products for nonenergy purposes (for example, in the chemical industry). It also includes fuel purchased within the country that is used elsewhere (for example, international aviation, international maritime bunkers, and, in the case of road transport, “fuel tourism”). 4 Source: Eurostat (2023), https://ec.europa.eu/eurostat/databrowser/view/sdg_07_50/default/table?lang=en 5 Source: Eurostat (2023). https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Archive:EU_energy_mix_and_import_dependency 30 Romania Energy Poverty Assessment Figure 4 En erg y Dep en den cy i n th e Euro pea n U nio n, 2 02 0 100.0 % 90.0 % 80.0 % 70.0 % 60.0 % 50.0 % 40.0 % 28.2 % 30.0 % 20.0 % 10.0 % 0.0 % Malta Cyprus Luxembourg Greece Georgia Belgium Moldova Lithuania Italy Ireland Turkey Netherlands Spain Portugal Germany North Macedonia Austria European Union (EU27) Hungary Slovakia Croatia Slovenia Latvia Denmark France Poland Finland Czechia Bulgaria Albania Ukraine Sweden Serbia Kosovo (UN SCR 1244/99) Romania Montenegro Bosnia and Herzegovina Iceland Estonia Source: Eurostat (2023). https://ec.europa.eu/eurostat/databrowser/view/sdg_07_50/default/table?lang=en Note: The figure shows the import dependency in the EU, measured as a share of net imports in the gross available energy. It represents how much a country or region depends on imports from abroad. A negative share means the country is a net exporter (for example, oil and gas in Norway). A share of more than 100 percent means the fuel is put in stocks (for example, oil and coal in Belgium). The total energy supply is mainly based on oil, followed by percent). While natural gas and coal now play a less significant natural gas and coal; nuclear power, wind, and solar energy, role in total energy supply than 20 years ago, hydro energy, bio- as well as hydro energy, play a more significant role nowa- fuels, and waste, nuclear, wind, and solar energy play a more days. In 2019, nearly 30 percent of the total energy supply was critical role (figure 5). There was a drop of 12.3 percent in elec- based on oil, followed by natural gas (27.9 percent) and coal (15 tricity production between 2018 and 2020 (figure 6). Figure 5 Total En erg y Sup p ly by Source, Figu re 6 Ele c t r ic ity Pro d u c t io n, 2 011 –2 0 ( MW) 1990–2019 (TJ ) 1,400,000 21,000 1,200,000 20,500 20,000 1,000,000 19,500 800,000 19,000 600,000 18,500 400,000 18,000 200,000 17,500 0 17,000 1990 1995 2000 2005 2010 2015 2019 16,500 Coal Natural gas Hydro Biofuels and waste 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Oil Nuclear Wind, solar, etc. Source: IEA Data Services (https://www.iea.org/countries/romania) based on data from World Energy Balances (2022) (https://www.iea. org/data-and-statistics/data-product/world-energy-statistics-and-balances) and Eurostat (nrg_inf_epc, 2022 https://ec.europa.eu/ eurostat/databrowser/view/NRG_INF_EPC/default/table?lang=en ). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 31 Households account for less than 20 percent of energy use; 2.2. Connectivity energy intensity in Romania has improved over time but is still above the EU average. Electricity consumption per capita Connectivity to and availability of different energy sources dropped between 1990 and 2000, but has shown an increasing could significantly influence energy spending patterns. If trend since then, suggesting the need for continued efforts to households have limited access to energy due to limited infra- promote energy efficiency and sustainable energy practices. structure or high prices, they might report low energy spend- Household consumption accounted for 16.9 percent of energy ing shares. Nevertheless, these low shares do not reflect lower use in 2020, a relatively small share compared to energy trans- exposure to high prices, but rather another form of energy vul- formation, nonenergy use, industrial, transport, and services. nerability, namely low access. Based on this line of thinking, we Back in 2011, this share was equal to 15.6 percent. Energy inten- next analyze the connectivity of households to different ener- sity (measured as the amount of energy needed to produce a unit gy sources. We investigate how connectivity differs along the of GDP), commonly used as one of the major energy efficiency welfare distribution and analyze average connectivity rates for indicators, improved in Romania from 280 to 190 kilograms of different population subgroups. oil equivalent (kgoe) per thousand euros from 2011 to 2020. The assessment of access to energy services involves con- However, in 2020 it was still significantly higher than the EU- siderations at both the extensive and intensive levels, encom- 27 average of 117 kgoe per thousand euros, suggesting there is passing the availability of energy as well as its quality and room for improvement in energy efficiency and implementation reliability. Typically, measures focusing on energy access pri- of new technologies to reach the standard of the EU average. marily examine whether households have access to energy, rep- Although Romania traditionally had among Europe’s resenting the extensive margin. Additionally, the quality aspects lower energy import dependency rate6, it has not escaped the of energy access and its reliability,7 referred to as the intensive current energy crisis, leading to many Romanians worrying margin, contribute to the understanding of energy poverty. Bha- about rising energy prices. A 2022 survey conducted by Avan- tia and Angelou (2015) introduce a comprehensive multitier garde reveals that nearly 2 out of 10 Romanians were worried framework, particularly for electricity, incorporating parameters about the current energy crisis. One-third of those heating with such as reliability (frequency of disruptions per week), legality, gas said they would consider moving to alternative heating health and safety aspects of the energy source, peak capacity, sources. An additional 68 percent of respondents indicated that and duration. Access-based approaches to evaluating energy they would reduce their overall electricity consumption. Four poverty can also extend to the type of energy available, includ- out of 10 respondents expected a complete gas supply cut by ing considerations of access to clean energy, as highlighted by Russia during the winter (Fodor 2022). A more recent house- Ullah et al. (2021). hold survey conducted by the World Bank in June 2023 in Ro- Nearly all households in Romania were connected to elec- mania shows that 6 out of 10 Romanians expected electricity tricity in 2021,8 and differences across groups are negligible. prices to increase even further during the next 12 months. These Figure 7 shows full electricity coverage for nearly all groups. numbers demonstrate the population’s awareness of their vul- While there are some minor differences across groups, these are nerability to the current energy crisis and potential behavioral negligible. Coverage in rural areas is slightly lower.9 One group changes to cushion the impacts of rising prices. that reported below-average coverage rates consisted of house- 6 Source: Eurostat (2023); https://ec.europa.eu/eurostat/databrowser/view/sdg_07_50/default/table?lang=en 7 Some aspects of this are how many hours per day the household receives electricity, the number of blackouts per month, etc. This is more difficult to quantify. 8 Because we do not observe in the household survey a connection to the grid, we proxy for electricity connectivity by identifying the share of households with a nonzero household expenditure on electricity and renewable energy, which are combined in the survey collection process. However, expenditures on renewables are expected to be negligible compared to electricity. Therefore, it is likely that this spending mostly captures electricity expenditures. 9 This is consistent with previous evidence. According to the Atlas or Urban Marginalized Areas in Romania, less than 1% of the urban dwellings are not connected to electricity. Lack of electricity connection in rural marginalized areas tends to be more prevalent. Similarly, the share of the 32 Romania Energy Poverty Assessment holds receiving social aid from municipalities. Still, more than 9 cooking or heating, less than one-fourth of those living in rural out of 10 households in this group reported nonzero electricity areas did so.10 Previous evidence also shows urban-rural differ- spending. Based on these results, we rule out lack of connectivity ences in gas coverage.11 A small share is also observed among to electricity as a driver of our results. households receiving social aid from the municipality, those A different picture emerges when looking at natural gas; with Roma members, and those with more than five household however, the heterogeneity in coverage rates across groups members. However, because we proxy connectivity to natural could be driven by household preferences about energy sourc- gas by determining the share of households using natural gas for es rather than by a lack of access. Figure 8 shows significant dif- cooking and/or heating, these differences could relate to differ- ferences across groups in the coverage rate of natural gas. While ences in household energy preferences rather than limited access most households in urban areas reported using natural gas for due to limited infrastructure or high prices. Figure 7 Coverage Rates of Electri ci ty by G ro u ps , 2 021 ( % ) HH with social aid from municipality 93.6 HH with more than 5 members 97.1 HH with disability recipient 97.1 Household with pensioner 98 Single-elderly HH 98.2 HH with child 98.2 Household with unemployed 98.3 Female-headed household 98.4 Household with Roma 99 Rural 97.1 Urban 99.3 90 91 92 93 94 95 96 97 98 99 100 Source: Own estimates based on HBS (2021). Note: We proxy for electricity connectivity by identifying the share of households with a nonzero household expenditure on electricity and renewable energy, as expenditures on electricity only cannot be identified in the 2021 HBS. This indicator measures access at the extensive margin (binary description) and does not capture the intensive margin aspects (quality and reliability). Expenditures on renewables are expected to be negligible compared to electricity. There is no formal ethnicity identifier in the HBS, but Roma is one of the categories in the nationality question. When analyzing heating and cooking methods, it becomes systems, with the majority using gas as the main fuel. However, evident that energy sources are relatively diversified (figures more than one-third of households still used traditional tech- 9–10). Almost half of the households relied on central heating nologies for heating (wood, coal, or oil stoves).12 population living in dwellings not connected to electricity jumps from 0.6 percent of non-Roma in non-marginalized areas to over 10 percent of Roma people living in marginalized areas, suggesting pockets of deprivations at lower levels of disaggregation and among marginalized groups (Sandu et al., 2016). 10 This is only an estimate of differences in coverage and usage between the urban and rural populations, which does not necessarily imply that the gas network needs to be extended to reach the whole population. 11 A study from the Center for the Study of Democracy (2018) found that, in urban areas, 148 cities (68%) are connected to gas, covering 75% of the urban population. In contrast, 71% of the rural population has access to gas. Localities connected to gas are primarily concentrated in central Romania along an axis that starts in the northwest (including eastern Cluj and Alba counties, Mureș county, and most of Sibiu and Brașov counties), extends southward through Dâmbovița, Prahova, Ilfov, and Bucharest. Conversely, the regions outside the Carpathian Mountains are the least covered by the gas network. 12 The HBS does not collect information on the type of heating technology (modern vs. traditional). Information on the type of heating technology (modern vs. traditional) is important, because without it a comprehensive understanding of energy usage patterns is not possible. Knowing the Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 33 Figure 8 Coverage Rates of Natural G as by Gro u ps , 2 021 ( % ) HH with social aid from municipality 14.9 HH with +1 Roma 22.7 HH with more than 5 members 26.4 HH with +1 unemployed 47.4 HH with disability recipient 47.5 Single elderly (+64) HH 51 HH with pensioner 52.3 HH with child 54 Female-headed HH 57.7 Rural 23.6 Urban 83.3 All 55.2 0 10 20 30 40 50 60 70 80 90 Source: Own estimates based on HBS (2021). Note: We proxy for connectivity to natural gas by identifying the share of households using natural gas for cooking and heating. Figure 9 Ty p e of Heati n g M eth ods , 2 021 ( % ) Figu re 1 0 Type o f Co o k ing Me t ho d s , 2 021 ( % ) Disconnected 0.2 Cooking (other) 0 No heating system 0.1 Cooking (cylinder) 43 Has wood/coal/oil stoves 35.2 Cooking (wood/coal/oil) 20.8 Has natural gas stove 2.5 Cooking (natural gas) 54.7 Has central heating 47.2 Has thermal Power Station Cooking (electricity) 0.3 13.8 00 10 20 30 40 50 00 10 20 30 40 50 60 Source: Own estimates based on HBS (2021). Note: Central heating includes systems using various energy sources to generate heat. In the survey, these included wood, natural gas, electricity, and so on. The last category of “others” is not specified in the survey. A significant share of rural households also relied on or forest plantations owned by households, which might result wood as a heating and cooking energy source. The availabili- in underreporting. Based on this rationale, we analyze whether ty of wood as a confounding factor is especially relevant when energy sources for heating and cooking differs between rural considering spending patterns on energy in rural areas. Other and urban areas. These analyses reveal that wood as a heating potential confounding factors are energy sources that are illegal and cooking energy source plays a significant role, especially in or freely available, such as wood generated from illegal logging rural areas (figures 11–12). heating technology households use is crucial for assessing energy efficiency, environmental impact, and potential health implications. Therefore, we classify the available heating sources into modern and traditional heating technologies. The classification of heating technologies is based on various features, highlighting the diverse methods employed for providing warmth. Traditional technologies involve the use of wood, coal, oil, or natural gas stoves, reflecting a reliance on conventional sources for warmth, while modern heating technologies include options like thermal power stations and central heating systems using various energy sources to generate heat (wood/pellets, natural gas, electricity, and so on). Finally, we also collect this information in the 2023 World Bank survey. 34 Romania Energy Poverty Assessment Figure 11 Ty p e of Heati n g En erg y Source by Figu re 1 2 Type o f Co o k ing Ene rg y S o u rce by Rural/U rban , 2 021 (% ) Ru ral /U r ba n, 2 021 ( % ) rural urban rural urban Disconnected 0.4 90 82.9 0 0.2 80 73.3 No heating system 0 70 9.9 Has wood/coal/oil stoves 63.8 60 Has natural gas stove 1.9 3.2 50 Central heating type: 0 39.6 Other 0 40 Central heating type: 0.6 30 23 Elecricity 0.1 Central heating type: 59.8 20 16.2 Natural Gas 16.5 Central heating type: 2.7 10 4.0 0.2 0.4 0 0 Wood/Pallets 12.5 00 Has central heating 63.1 (electricity) (natural gas) (wood/coal/oil) (cylinder) (other) Cooking Cooking Cooking Cooking Cooking 29.1 Has thermal Power Station 24.1 2.1 00 10 20 30 40 50 60 70 Source: Own estimates based on HBS (2021). 2.3. Energy Affordability: Monetary households’ welfare. While some approaches rely on household and Nonmonetary Measures income, others rely on household expenditure (consumption). The diverging methodologies might result in different pic- Researchers have developed a variety of approaches to mea- tures of overall energy poverty incidence, energy expenditure sure energy expenditure shares. Several measures of energy patterns, and households’ vulnerability to energy prices, im- expenditure shares have been used in the literature (for a com- pacting the design and monitoring of energy poverty policies. plete overview, see Robayo-Abril and Rude (2024). While some Given that energy expenditure shares might vary depending on measures rely on nonmonetary approaches, others use income the underlying measure used, it is crucial to base the chosen and expenditure data. However, even within the literature that measure on empirical considerations. In addition, the compari- uses monetary approaches, significant methodological differ- son of different empirical approaches is recommended to better ences exist in the empirical approaches taken for two reasons. understand the sensitivity of the insights generated based on First, different measures consider different components when resulting energy expenditure shares to the underlying method- estimating household energy expenditure. For example, some ology. Because policy recommendations might differ depending might include car-related energy expenditures (for example, gas- on the estimates that are generated, reflecting on the underlying oline), while others might abstract from this component because methodology is crucial. they relate it more to transport than the energy sector. Especial- In this report, we conduct empirical analyses to choose ly when analyzing households’ vulnerability to fluctuations in the most suitable monetary measure in the context of Roma- energy prices, it is crucial to carefully consider which compo- nia. To this end, we rely on the 2021 HBS data and explore dif- nents to include when measuring absolute energy expenditure. ferent approaches to measure energy expenditure and household Second, the literature has used several different methodologies welfare. More concretely speaking, we analyze how estimates to set households’ absolute energy expenditure in relation to differ across different methodological approaches in the case Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 35 of Romania. We do so by considering different components of four different approaches to measuring energy poverty. Unlike households’ overall energy expenditure. In addition, we explore nutritional poverty, which can be defined by a minimum daily ca- two different approaches to measuring households’ welfare: in- loric intake, there is no absolute reference point for meeting basic come and consumption. Appendix 1 presents the detailed em- energy needs. Therefore, we apply four of the most commonly used pirical approach and the resulting estimates. measures in the EU (for a complete overview, see Robayo-Abril Based on our empirical analysis, the income-based energy and Rude, 2024). The first defines households as energy poor who expenditure share that abstracts from car-related energy ex- have an energy spending share at least twice as large as the median penditure is the most robust measure in the case of Romania. energy spending share in the population (2M). The second relies We define energy expenditure shares as follows: on absolute energy expenditure and defines households who re- port absolute energy expenditure below half the national median as energy poor (M/2). The third measure defines energy poverty as the share of households with an energy expenditure share above where ES is the energy expenditure share for each household, 10 percent (P10). Finally, the last approach, the Low-Income High E is the total expenditure on energy, and I is the household’s to- Cost (LIHC) measure, identifies energy-poor households as those tal income (household budget). Notably, E does not consider who are pushed into energy poverty due to high energy costs and energy-related household expenditure, such as expenditure on low incomes. These different measures of energy poverty capture gasoline or other type of car-related fuels. different aspects of energy poverty (see box 2.1 for details). As Energy poverty lacks a universally accepted metric due to explained in chapter 5, we did not operationalize the official defini- the absence of a standardized definition of what should be in- tion in the legislation due to the lack of clearly defined `minimum cluded under an energy poverty line. In this report, we explore energy’ needs and the lack of data.13 Box 2 .1 : M eas ur i n g En erg y an d Elec tricity Poverty to D es ig n Better Social Prog rams In the European context, particularly in countries where heating needs are significant, development practitio- ners and academics have long sought to isolate the needs of those households whose energy expenditure re- presents a significant burden on their budget. The literature on measuring energy poverty is ample (Hills 2012; Thomson et al. 2017; Trinomics et al. 2017; Romero et al. 2018; Castaño-Rosa et al. 2019; Sareen et al. 2020). A relatively easy way to proxy for such a burden using an expenditure-based indicator is by setting a threshold of budget share devoted to energy costs (it is customarily set at 10 percent; see World Bank 2012) and identifying energy poverty. By analogy to such measures, source-specific thresholds have also been set, thereby identifying, for example, electricity poverty as spending more than 10 percent of the household budget on electricity. This is a fixed threshold, so it is useful to track changes over time and for cross-country comparisons. Several developments have contributed to the refining of the way such indicators are designed. One was the establishment of the EPOV,14 now the Energy Poverty Advisory Hub, tasked with monitoring the situation in the EU. Against the absence of one commonly agreed-upon definition of energy poverty across the EU, the EPOV aims to contribute to a common understanding—but not to finally define energy poverty. It has produced a set of gui- delines to define and measure energy poverty.15 While the EC has reiterated that a standard 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 and definitions. For example, the EU-SILC conducted by all EU Member States includes self-reported information on households’ being unable to heat their homes appropriately and carrying arrears on utility bills. Other expenditure-based indicators based on the household budget surveys are absolute (equivalized) energy expenditure below half the national median (M/2) and the share of (equivalized) energy ex- 13 Estimating “minimum energy needs” is also data intensive, because it requires detailed information on the dwelling (space heating, water heating, lights, and appliances and cooking) and its occupants. 14 https://energy-poverty.ec.europa.eu/energy-poverty-observatory_en. 15 https://energy-poverty.ec.europa.eu/system/files/2021-09/epov_methodology_guidebook_1.pdf. 36 Romania Energy Poverty Assessment penditure (compared to equivalized disposable income) above twice the national median (2M). The M/2 indicators measures abnormally low energy expenditures, capturing the concept of hidden energy poverty, whereas the 2M indicator captures abnormally high energy expenditure. Both indicators are relative per definition and are sensitive to the underlying distribution of the variables used for their calculation, so they are not ideal for tracking differences over time (where a set bar is recommended). The EC highlights the importance of energy poverty as reflected in innovation at the country level. For example, the United Kingdom 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 households who would spend more than 10 percent of their income to keep their home at 21 °C. Other evidence has reassessed that measure and suggested two new fuel po- verty measures to capture the incidence and severity of the phenomena. The indicator for the incidence of energy poverty, known as Low Income, High Cost (LIHC), is defined as having energy costs higher than the median and having an income net of energy costs below the official at-risk-of-poverty (AROP) line. Finally, the measurement of the depth of energy poverty is a more complex task and the indicator for it involves measuring the gap between household energy needs and a reasonable threshold. A different international effort that provides some new ideas on how to look at electricity affordability is pro- vided by the Multi-tier Framework for Measuring Energy Access (MTF).16 The framework was developed by The World Bank’s Energy Sector Management Assistance Program (ESMAP) in consultation with international partners under the Sustainable Energy for All (SEforALL) initiative to improve measures for monitoring energy access under Sustainable Development Goal 7.1 (target on access to affordable, reliable, and modern energy for all). The MTF proposes operational definitions for measuring access to energy, going beyond the binary metrics (whether a hou- sehold has an electricity connection and cooks with fuels like charcoal or dung). Electricity access is measured based on the combination of seven energy attributes across six access tiers with minimum requirements by tier of electricity access. It also provides basic thresholds on expenditure on electricity for different purposes. Those are identified as 5 percent for heating and 5 percent for cooking, which have been adopted by the MTF. All these different efforts, which are ultimately aimed at measuring more accurately a particular facet of mone- tary and nonmonetary deprivations, suggest that there is scope for experimentation and dialogue with the Ro- manian government in designing new measures and programs that help address the multiple causes of energy poverty itself: low incomes, poor housing conditions, low energy efficiency, and high energy spending. We report these values along the income distribution and allows for indirectly approximating the amount of energy con- for population subgroups. We approximate the welfare distri- sumed. The HBS also enables disaggregation among some vul- bution by relying on the income measure of the household sur- nerable groups, including but not limited to beneficiaries of the vey. We first calculate the income per capita and then analyze leading social assistance program and households of the single to which income quintile each household belongs. Based on this elderly, those with unemployed persons ages 55 and above, and distribution, we generate insights on energy affordability and pov- those headed by a single parent. erty along the welfare distribution. Moreover, we generate these We also rely on data from the EU-SILC 2020. To analyze estimates for subgroups of the population, which we believe could affordability among the poor and other income groups across be especially vulnerable, such as households of the single elderly, the distribution (that is, income quintiles), we also use the EU- with female heads, or with many household members. SILC data, the leading survey instrument for measuring official We mainly rely on the 2021 HBS to analyze monetary en- poverty in the country. The EU-SILC also allows us to measure ergy affordability in Romania. The HBS collects detailed ex- several nonmonetary energy affordability and poverty indica- penditure information for all consumption categories, including tors. To combine energy expenditure and income data, we use energy expenditure with disaggregation by energy source, and 16 https://www.esmap.org/mtf_multi-tier_framework_for_energy_access#:~:text=The%20Multi%2DTier%20Framework%20(MTF,vs%20nonsolid%20 fuels%E2%80%9D%20for%20cooking. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 37 the HBS from 2019, because the income reference year of the spent, on average, 8.7 percent of their overall household expen- EU-SILC 2020 is 2019, and match the two datasets. diture on energy.17 The resulting energy poverty incidence rates Finally, the 2023 World Bank rapid survey, conducted in using the P10 and the M/2 are similar and indicate that approx- June–July 2023, provides an updated picture of household imately one-fourth of the population in Romania was affected energy vulnerability. This household survey was purposely de- by energy poverty. The M/2 indicators measure unusually low signed to capture energy affordability and barriers to moving energy expenditures, capturing the concept of hidden energy to cleaner energy and heating sources. It captured information poverty, which is different from other concepts of energy pov- on beliefs, attitudes, and behaviors related to sustainable energy erty that typically focus on high energy costs or lack of access to transitions. The details of this survey are presented in annex 1. energy. Therefore, it is important to capture a full picture of en- Our estimates suggest that, when using expenditure-based ergy poverty. The absolute energy poverty rate was significantly measures, approximately one-fourth of the Romanian popu- smaller when using the 2M and the LIHC measures, 12.3 and lation was energy poor in 2021. In 2021, Romanian households 5.7 percent, respectively (figure 13). Figure 13 En erg y Ex p en di ture Sh are an d Figu re 1 4 Mo ne ta r y vs . Ene rg y Pove r ty, 2 021 Energy Poverty, 2 021 ( S ha re s ) 0.30 0.261 1.0 0.243 0.25 0.8 0.20 0.6 0.15 0.123 0.4 0.10 0.087 0.2 0.057 0.05 0.0 2M M/2 P10 LIHC 0.00 Energy M/2 2M P10 LIHC Energy and monetary poor Only energy poor spending share Only monetary poor Neither nor Source: Own estimates based on HBS (2021). Note: We measure energy spending as the amount of money spent on energy divided by the monthly income of each household. We abstract from car-related expenditure when calculating spending on energy. Energy expenditure includes electricity and renewables, natural gas, liquified energy sources, liquid fuels, solid fuels, and thermal energy. Collected firewood (not traded in markets) is not captured as energy expenditure and may lead to low energy expenditures. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). Monetary poverty is the AROP rate using equivalized household income and the poverty line estimated in the HBS 2021. The AROP rate estimated using the HBS is higher than the official rate estimated with the EU-SILC, standing at 23.8 percent compared to 21.2 percent, respectively. While interrelated, energy poverty and income poverty poor populations emphasizes the difference between money and are distinct. When energy-poverty measures are used in tan- energy poverty. We would want such a share to be small, but dem with the measure of households at risk of poverty, often this is not the case in Romania. When using the 2M and P10 used to evaluate monetary poverty in the EU, between 4 and measures of energy poverty, only half of those at risk of poverty 13.4 percent of households in Romania were energy poor but were also energy poor. The share is lower in the cases of the M/2 not income poor. The proportion of energy poor among non- and LIHC measures (25.5 and 17.6 percent, respectively). On 17 There are many different methodological approaches to measure energy poverty. For a full overview of different measures see Robayo-Abril and Rude (2024). For the chosen methodology in this report and the applied empirical approach, see appendix 1. 38 Romania Energy Poverty Assessment the other hand, when considering all energy poverty measures access energy-efficient technologies or renewable energy sourc- except LIHC, only half of those experiencing energy poverty es. Moreover, energy policies can focus on improving energy were also in monetary poverty. However, when examining the infrastructure in low-income areas to ensure access to reliable LIHC measure, this proportion increases to 73 percent (figure and affordable energy. 14). These results imply that while there is some overlap between Although those at risk of poverty and those who are en- the energy and the monetary poor, a significant share of the ergy poor do not fully overlap, those at the lower end of the population did not face monetary and energy poverty jointly. income distribution report higher shares than those at the This finding has important implications for designing higher end, regardless of the measure used, suggesting high- policies to tackle energy poverty. If people who lack access er energy vulnerability among low-income groups to energy to energy are mostly also poor in terms of income, then re- price spikes. Figure 15 reveals that households in the lowest in- ducing income poverty is essential to reducing energy pover- come quintile spend, on average, 14.7 percent of their household ty. However, suppose energy poverty and income poverty are income on energy expenditures, while those in the upper-in- not necessarily coincident. In that case, energy policies can come quintile only spend 3.4 percent of their household income still play a crucial role in reducing energy poverty, because tar- on these expenditure types. The energy poverty incidence rates geting schemes that target those at risk of poverty might only vary significantly by income quintile, with the lower-income partially address the vulnerabilities faced by those exclusively population experiencing higher rates, regardless of the defini- affected by energy poverty. For example, energy policies can tion used (figure 16). These patterns suggest that energy price provide subsidies or incentives for low-income households to spikes will affect low-income households disproportionately. Figure 15 En erg y Ex p en di ture Sh ares by Figure 16 Energy Poverty by Income Quintiles Income Q ui n ti les , 2 021 (% ) for Three Measures of Energy Poverty, 2021 (%) 16.0 60.0 14.7 14.0 50.0 12 12.0 40.0 10.0 8.3 30.0 8.0 6.0 5.3 20.0 4.0 3.4 10.0 2.0 0.0 0.0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 2M M/2 10P LIHC Source: Own estimates based on HBS (2021). Note: We measure energy spending as the amount of money spent on energy divided by the monthly income of each household. We abstract from car-related expenditure when calculating spending on energy. Energy expenditure includes electricity and renewables, natural gas, liquified energy sources, liquid fuels, solid fuels, and thermal energy. Collected firewood (not traded in markets) is not captured as energy expenditure and may lead to low energy expenditures. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). Some groups are more affected by energy poverty than ventions. Figure 17 shows that energy spending share differs others, which requires targeted and differential policy inter- across population subgroups and that there are some important Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 39 variations. The energy expenditure share ranges from 6.7 to households receiving social aid from municipality reported 14.5 percent. The households who reported the highest shares the highest rates. This hints at the problem of hidden energy were those of the single elderly, those who receive social aid poverty, a phenomenon that occurs when households report from municipalities, and those with older adults or pension- low energy spending shares because they restrict their energy ers. Energy spending shares are also higher in rural than urban usage due to energy poverty. What these results show is that areas. Energy poverty patterns are similar when the 2M and policy makers need to design efficient interventions to target P10 measures are used (figure 18). When the absolute energy those who are most affected instead of implementing universal, poverty incidence rate (M/2) is used, Roma households and undifferentiated interventions. Figure 17 En erg y Ex p en di ture Sh ares by Figu re 1 8 Ene rg y Pove r ty Ra te s by Gro u ps , Groups, 2021 (% ) 2 021 ( % ) HH with child 6.7 HH with social aid from municipality HH with more than HH with disability 5 members 7.5 recipient HH with more than Household with Roma 9.2 5 members Household with Household with unemployed 9.4 unemployed Female-headed Household household 10.3 with Roma Female-headed HH with disability recipient 10.4 household HH with social aid Single-elderly HH from municipality 11.2 Household with Household with pensioner 11.5 pensioner HH with child Single-elderly HH 14.5 Rural Rural 10.9 Urban Urban 6.8 All All 8.7 0 10 20 30 40 50 60 0 2 4 6 8 10 12 14 16 LIHC 10P M/2 2M Source: Own estimates based on HBS (2021). Note: We measure energy spending as the amount of money spent on energy divided by the monthly income of each household. We abstract from car-related expenditure when calculating spending on energy. Energy expenditure includes electricity and renewables, natural gas, liquified energy sources, liquid fuels, solid fuels, and thermal energy. Collected firewood (not traded in markets) is not captured as energy expenditure and may lead to low energy expenditures. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). 40 Romania Energy Poverty Assessment Box 3 . 2 : En erg y an d t h e Eld er ly i n Roman ia One main result of our analysis is that single-elderly households were much more likely to be affected by energy poverty, and related to this, they also reported, on average, higher energy expenditure shares. Previous stu- dies find similar patterns for older adults in the case of other countries. Research in Japan, for example, finds that older adults are more likely to spend time at home during winter and summer than younger people, leading to higher energy consumption (Inoue et al. 2022). Single-elderly households are also less likely to benefit from eco- nomies of scale of energy consumption (Inoue et al. 2022). In addition, data from the United Kingdom shows that older adults are more likely to live in energy-inefficient housing and houses that are more difficult to heat (Harding 2022). At the same time, they might also have less income to cover rising energy bills. These factors make older adults, especially those living alone, more vulnerable to rising energy prices. The higher exposure of single-elderly households to energy prices is problematic, given that they also show stronger health reactions to energy deprivation. A study from the United States, for example, showed that older adults are more likely to die from heat- or cold-related causes during the summer and winter months with extreme temperatures (LIHEAP Clearinghouse 1995). We analyze which factors could drive the higher probability of being energy poor of single-elderly households in the case of Romania. To this end, we use data from the HBS 2021 and describe single-elderly households in terms of observable socioeconomic characteristics, compare the type of energy they used for cooking and heating, and analyze some of their housing and living conditions more broadly. We find evidence of the mechanisms behind the higher vulnerability to energy poverty of older adults identified in previous studies in Romania. Table A.5.1. in Annex 5 shows that single-elderly households are more likely to be at risk of poverty (30 percent vs. 22 percent). Some indicators also hint at worse housing conditions in the case of single-elderly households. For example, they are less likely to have available appropriate sewage systems or hot water. Lastly, there are some systematic differences in the types of energy they rely on for heating and cooking. More-detailed information on the energy efficiency and costs of these different heating and cooking systems is required to understand further the extent to which the latter point drives the higher energy-spending shares for single-elderly people in Romania. Overall, we conclude that single-elderly households require special attention when designing policies to de- crease energy poverty. These policies should consist of income-support measures on the one hand and access to energy-efficient housing and heating and cooking methods on the other. We do not find that energy spending patterns and en- show these groups’ energy spending shares and energy poverty ergy poverty differ by gender. Commonly, energy poverty is incidence rates. The highest energy spending share (and poverty conceptualized and assessed at the household level, potentially incidence rate) was reported by households with no employed overlooking gender disparities in the expenses and advantages adults, while the lowest was that reported by dual-earner house- related to energy and fuel consumption. To this end, we con- holds. These results indicate that income might be more relevant struct some gender typologies to investigate household com- than the gender of household members in determining energy positions beyond the traditional female headship typology and spending shares and energy poverty rates. However, the gender analyze the extent to which different forms of ‘female’ and ‘male’ dimensions of energy poverty may be broader and require a households are affected by energy poverty. Figures 19 and 20 more detailed analysis.18 18 For example, women and children may bear a disproportionate impact from energy poverty, as they tend to spend more time at home (Petrova & Simcock (2021); Clancy et al. (2007)). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 41 Figure 19 En erg y Sp en di n g Sh ares by G e nd e r Figu re 2 0 Ene rg y Pove r ty Ra te s by Ge nd e r Groups, 2 021 (% ) Gro u ps , 2 021 ( % ) Dual-earner No employed adults 13 Female breadwinner Female-headed household 10.3 (no employed male) Female-headed household Female majority 9.6 with children (<15) (no employed female) Male breadwinner 8 (no employed female) Male breadwinner with children (<15) Female majority 7.1 Female-headed household (no employed male) Female-headed household 7 Female breadwinner No employed adults Dual-earner 5.6 0 10 20 30 40 50 0.0 5.0 10.0 15.0 LIHC 10P M/2 2M Source: Own estimates based on HBS (2021). Note: We measure energy spending as the amount of money spent on energy divided by the monthly income of each household. We abstract from car-related expenditure when calculating spending on energy. Energy expenditure includes electricity and renewables, natural gas, liquified energy sources, liquid fuels, solid fuels, and thermal energy. Collected firewood (not traded in markets) is not captured as energy expenditure and may lead to low energy expenditures. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). Among all energy expenditures, households spend the Similarly, households spend, on average, 17.5 percent of their most on electricity . Household energy expenditure alone, 19 overall energy expenditure on natural gas (figure 21, panel a). not the energy sources, is the only factor used to estimate ener- On average, energy expenditure shares are larger when gy poverty. However, the assessment of energy poverty and the traditional heating technologies are used. Overall, households formulation of policy directives to reduce it requires a thorough who used traditional heating technologies (such as a natural gas understanding of the many energy sources that are used. When stove or wood/coal/oil stoves) spent, on average, 12.2 percent of analyzing the energy expenditure patterns in Romania in more their household income on energy expenditures, whereas house- detail, it quickly becomes clear that the largest share of energy holds who used modern technologies (thermal power station spending goes to electricity, followed by solid fuels. Households or central heating) spent 6.6 percent. These suggest the use of spend, on average, 32.8 percent of their overall household en- modern technologies yields cost savings, once the upfront costs ergy expenditure (including car-related fuel) on electricity, 20.3 of installation are recovered. Panel b of figure 21 reveals that percent on solid fuels, and 18.0 percent on car-related fuels. savings are present across income quintiles, although they are lower for the lowest income quintile.20 Even when we control 19 As mentioned earlier, the household budget survey does not separately identify expenditures on electricity and renewable energy. However, renewable energy expenditures are expected to be negligible compared to electricity, so this spending likely mostly captures electricity expenditures. 20 We used alternative energy expenditure shares, such as expenditure-based vs. income-based, including vs. excluding car-related expenditure, and 42 Romania Energy Poverty Assessment for socioeconomic and demographic characteristics, as well as dwelling characteristics, we find significant savings using mod- ern technologies (see table A2.1). Figure 21 Average En erg y Sp en di n g Sh a re by Type o f Ene rg y a nd by H ea t ing Te c hno lo g y, 2 021 Panel a. Average energy spending share by type Panel b. Energy expenditure shares by income of energy quintiles and heating technology 100% 0.16 0.15 0.14 0.14 80% 0.14 0.12 60% 0.11 0.11 0.10 40% 0.08 0.07 0.07 20% 0.06 0% 0.05 0.05 0.04 0.03 Solid fuels Liquid fuels 0.02 Natural gases (share) 0.00 Thermal energy (share) Q1 Q2 Q3 Q4 Q5 Electricity and renewable energy (share) Traditional heating Modern heating Source: Own estimates based on HBS (2021). Note: Panel a shows the data collected by the HBS for the energy spending categories. Collected firewood (not traded in markets) is not captured as energy expenditure, and may lead to low energy expenditures. In panel b, modern heating technologies include thermal power station and central heating and traditional technologies include natural gas stoves and wood/coal/oil stoves. Among the European and Central Asia countries, higher energy), resulting in households cutting down other types of electricity prices (the most critical energy source for house- consumption such as food, health, or education (World Bank holds) are generally associated with higher burdens of elec- 2012). This is particularly the case among the poor. In Romania, tricity spending in household budgets. The evidence also energy shares among the poor and energy poverty rates (P10) suggests that, when looking at changes over time for a partic- are among the highest in the region (figure 22). Therefore, at ular country, households seem to have limited abilities to keep the microlevel, it is critical to assess both the impact in terms their electricity expenditures constant in an environment of of affordability and the distributional impact of rising energy rising electricity prices (by substituting for cheaper sources of prices on households. expenditure vs. income quintiles, all of which yielded consistent results. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 43 Figure 2 2 En erg y Affordab i l i ty i n a B roa d e r Co ntext : Ro m a nia vs . Ot he r Eu ro pe a nd Ce nt ral A s ia Count r ies (Sh ares ) 1.0 0.8 0.6 0.4 0.2 0.0 Türkiye Kazakhstan Kyrgyz Rep. Kosovo Georgia Russian Fed. Ukraine Croatia BiH Albania Montenegro Serbia Poland Romania N.Macedonia Armenia Bulgaria Moldova Share of population with energy spending > 10% Share of energy spending in total consumption, bottom 40% of the population Source: World Bank (2022). Staff estimates based on the Europe and Central Asia Poverty Database (ECAPOV) standardized household budget surveys covering most of the countries in Europe and Central Asia. Energy expenditures are expressed as a share of total consumption, not income, and include all sources of energy spending. Therefore, results are not comparable to the previous measures presented in this section. We also analyze nonmonetary measures of energy af- trends in the region, these rates experienced a significant in- fordability and energy poverty. These measures might be less crease relative to 2020. These results are consistent with World sensitive to measurement errors, such as the ones described Bank rapid survey estimates from July 2023, where 14.9 per- previously, and might also capture slightly different aspects of cent of the population reported that they could not afford households’ vulnerability to energy poverty. to keep their home adequately warm during the past winter. Compared to the citizens of other EU countries, in 2022 Similarly, the proportion of households with arrears on utility a large share of Romanians reported being unable to keep bills was among the largest in the EU, reaching 17.8 percent in their home warm (15.2 percent). Among the poor, the inci- 2022, a trend that was increasing. Figures 23 and 24 show the dence was almost twice (29.5 percent). Consistent with the respective estimates. 44 Romania Energy Poverty Assessment Figure 23 I n ab i l i ty to Keep Home Adequ a te ly Wa r m in t he Pas t 1 2 Mo nt hs , 2 02 0 vs . 2 02 2 ( % ) 30.0 25.0 20.0 15.0 10.0 5.0 .0 Bulgaria Cyprus Greece Lithuania Portugal Spain Romania France EU27 Italy Malta Ireland Latvia Slovakia Croatia Germany Netherlands Belgium Denmark Poland Hungary Estonia Sweden Czechia Austria Slovenia Luxembourg Finland 2022 2020 Source: Own estimates based on Eurostat (2023). Note: This consensual-based indicator is based on self-reported answers to a question in the EU-SILC about affordability (or ability) to keep home warm, regardless of whether the household needs to be kept warm. Figure 24 Arrears on Uti l i ty B i l l s i n th e Pas t 1 2 Mo nt hs , 2 02 0 vs . 2 02 2 ( % ) 40. 30. 20. 10. 0. Lithuania Sweden Denmark Belgium Austria Netherlands Spain EU27 Slovenia Germany Czechia Romania Latvia Bulgaria Hungary Malta Estonia Slovakia Ireland Poland Finland Luxembourg Greece Cyprus Croatia Portugal France Italy 2022 2020 Source: Own estimates based on Eurostat (2023). Note: This consensual-based indicator is based on self-reported answers to a question in the EU-SILC survey about inability to pay utility bills (heating, electricity, gas, water, and so on) on time due to financial difficulties. Recent survey estimates from the 2023 World Bank rapid (Thomson et al. 2019). To understand this issue, we collected surveys indicate that hot temperatures also pose a problem information on the ability of households to keep cool during for Romanians and that their energy consumption patterns summer (as in the Spanish EU-SILC), which is currently not might change if energy prices continue to increase. Energy collected in the Romania EU-SILC. Our results show that about poverty in the EU has traditionally been associated with the one-third of the population indicated they could not keep their inability of households to meet their heating needs during win- home adequately cool in the summer of 2022. This is an import- ter. However, understanding cooling needs is also important, ant finding, as most existing studies solely focus on the harmful particularly given that many countries in the region have expe- health impacts of low indoor temperatures, even though sev- rienced an increase in the frequency and intensity of heatwaves eral European regions are also susceptible to dangerously high Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 45 summertime temperatures (Maxim et al. 2016).21 Interestingly, holds are, on average, more affected by leakages than wealthier over one-third of the population will either heat less or replace households. Nearly one-fourth of households in the lowest in- heating sources with cheaper ones or both. In addition, 60.7 come quintile reported leakages compared to only 3.2 percent percent indicated that they would limit the usage of energy ap- in the highest income quintile. Similarly, we find substantial pliances. These estimates confirm that households in Romania heterogeneity in leakages for households with more than one are vulnerable to rising energy prices and will likely remain so unemployed household member (27.8 percent) and those with in the near future. many members (18.7 percent) (figure 26). Leakages are a more Households who are poorer or with either many or un- persistent problem in rural than urban areas (16.3 percent ver- employed members are more affected by leakages. Another sus 5.2 percent). Based on 2023 survey estimates, 7.7 percent interesting measure to look at in the context of energy efficiency of interviewees experienced a leaking roof; 7.9 percent damp is the share of households reporting leakage by income quintiles walls, floors, or foundations; and 4.7 percent a rotten window and across groups. Figure 25 demonstrates that poorer house- frame or floors. Figure 2 5 Hous eh olds wi th Leakages by Figu re 2 6 H o u s e ho ld s w it h Lea kage s by Income Q ui n ti les , 2 01 9–2 0 (% ) Gro u ps , 2 01 9 –2 0 ( % ) 30 All 10 24.8 25 Rural 16.3 20 Urban 5.2 15 HH with +1 unemployed 27.8 11.7 HH with more than 5 members 18.7 10 7.1 Single elderly (+64) HH 15.3 HH with child 11.5 5 3 3.2 HH with pensioner 10.4 Female-headed HH 9.3 0 Q1 Q2 Q3 Q4 Q5 0 5 10 15 20 25 30 Source: Own estimates based on EU-SILC (2020). Note: Q1 indicates the poorest income quintile and Q5 the richest. Leakages include roof, wall, floor, foundation, and windows. These estimates show that poorer households might need the welfare of such households along several dimensions. renovation measures and other energy-efficient modules; Leakages can also reduce the effectiveness of other energy-ef- rural areas might have more significant needs than urban ficiency measures, such as increased insulation and high-per- ones. The measure above shows that poorer households were formance windows. Investing in energy-efficient renovations more exposed to problems related to their housing conditions. for the poorest could be beneficial in both the medium and These could generate energy inefficiencies, further worsening long term. 21 The WHO recommends keeping indoor air temperatures between 18 and 24 °C, although in Southern and Southeastern Europe, where many households lack air conditioning, this range is frequently surpassed. People who are chronically unwell and the elderly are particularly susceptible to heat stress. 46 Romania Energy Poverty Assessment At the same time, a lower share of poorer households re- households’ needing to be made aware of these problems or not ported exposure to environmental problems. Not even 1 out identifying them as problems due to their more-severe challeng- of 10 households in the lowest income quintile indicated that es. Nevertheless, plotting these estimates for different groups they were exposed to some form of environmental problem. In reveals that vulnerabilities might also play a role (figure 28)— comparison, 14.4 percent of households in the highest income households with at least one unemployed member and those quintile did so (figure 27). These results could hint at poorer with a pensioner reported the highest exposure. Figure 27 Hous eh olds wi th Env i ron mental Figu re 2 8 H o u s e ho ld s w it h Env iro nm e ntal Problems by I n come Qui n ti les , 2 01 9–2 0 ( % ) Pro ble m s by G ro u ps , 2 01 9 –2 0 ( % ) All 12.3 16 14.4 13.4 14 12.4 11.8 Rural 8.5 12 9.8 Urban 15.2 10 8 HH with more than 5 members 10.9 6 Female-headed HH 12.6 HH with child 12.7 4 Single elderly (+64) HH 12.9 2 HH with pensioner 13.3 0 HH with +1 unemployed 19.9 Q1 Q2 Q3 Q4 Q5 0 10 20 30 Source: Own estimates based on EU-SILC (2020). Note: Q1 indicates the poorest income quintile and Q5 the richest. Moreover, previous evidence shows that social housing is 2.4. Factors Associated with Energy underdeveloped in Romania. Only 1.5 percent of the housing Poverty stock is earmarked for state-subsidized housing and most of that is in private hands (Adăscăliţei et al. 2020). Local govern- To identify potential factors associated with energy poverty, ments manage the housing stock, but often lack the capacity we start by analyzing potential factors correlated with high and resources to expand this sector. There are also reports of energy expenditure shares. To this end, we run an ordinary institutional discrimination against the Roma (Adăscăliţei least squares regression and identify the variables significantly et al. 2020). These dynamics could worsen the disparities correlated with energy expenditure shares. We include socioeco- in Romania when it comes to facing problems related to nomic and demographic characteristics, as well as dwelling char- energy inefficiency. acteristics (including the type of energy used for cooking and heating) to understand whether certain technologies correlate with higher energy spending shares. Table 1 reports the results. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 47 Table 1 Re g ress i on of En erg y Ex p en di tu re S ha re s , 2 021 (1) (1) Variables Energy exp. share Variables Energy exp. share (income-based no car) (income-based no car) -0.0105*** -0.0993 Household with children (<15) Subsidy (wood/coal/oil) (0.00136) (0.0874) 0.00245 -0.180** Household with pensioner Subsidy (electricity) (0.00164) (0.0836) 0.0390*** 0.0211** Single-elderly Cooking (electricity) (0.0141) (0.00953) 0.00740*** 0.0360*** Female-headed household Cooking (natural gas) (0.00146) (0.00951) -0.0116*** 0.00352* Household with Roma Cooking (wood/coal/oil) (0.00396) (0.00195) 0.0109*** 0.0290*** Household with unemployed Cooking (cylinder) (0.00265) (0.00533) -0.00758*** -0.0345 Urban Cooking (other) (0.00174) (0.0854) Household with more than five -0.00963*** -0.00763 Cold water, indoors from public supply members (0.00238) (0.00608) -6.98e-07*** -0.00245 Income per capita Indoors, from in-house (2.55e-08) (0.00621) 0.00813 Outside the residence, but inside the -0.0139* Social aid from municipality (recipient) (0.00535) building (0.00833) -0.00792* 0.00317 Disability benefit recipient Outdoors, fountain, pump, well (0.00425) (0.00495) 0.0512*** 0.00668 Connectivity to electricity Hot water: N/A (0.00482) (0.0223) -0.0285 -0.00448 Has a thermal power station Hot water: Public system (0.0200) (0.00497) -0.0512** -0.00494 Has central heating Hot water: In-house system (0.0219) (0.00384) 0.0354*** 0.000694 Central heating type: Wood/pellets Hot water: Disconnected (0.00935) (0.0202) 0.0296*** 0.00405* Central heating type: Natural gas Sewage type: In-house system (0.00952) (0.00232) 0.185*** 0.00699 Central heating type: Other Sewage type: None (0.0527) (0.00510) -0.0174 -0.0102 Has a natural gas stove Natural gas (cooking or heating) (0.0206) (0.00881) -0.00208 0.0433* Has wood/coal/oil stoves Constant (0.0200) (0.0223) -0.0309 Disconnected Observations 15,192 (0.0257) 0.185** Subsidy (thermal energy) R-squared 0.163 (0.0834) Source: HBS (2021). Note: The table reports correlation coefficients from a simple regression on energy expenditure shares. Energy expenditure shares are income based, abstract from energy spending on car-related energy (such as gasoline), and range from 0 to 1. 48 Romania Energy Poverty Assessment We identify several variables that are negatively correlated preferences, the energy efficiency of different technologies cur- with energy expenditure shares. Based on the figures in table rently in use, and the costs related to each technology (fixed 1, the following variables are related to a significant decrease in and variable costs). Income plays a role, but to a lower extent, the energy expenditure share: households with children (<15), given that the effect is close to zero. In addition, a significant households with Roma, those living in urban areas, households correlation is expected, given that the outcome variable uses with more than five members, income per capita, having central income as an input. heating, receiving electricity subsidies, being a disability benefit The analysis also shows that certain household types are recipient household, and having access to cold water outside the more likely to report higher energy spending shares: those of residence, but inside the building (the last two are only signifi- the single elderly, with many unemployed, and those head- cant at the 10 percent level). ed by a female. Policy makers addressing energy poverty rates Several variables are positively correlated with energy should prioritize these households when designing policy inter- expenditure shares. The following variables are positively cor- ventions. Still, energy expenditure shares could be biased, given related with the energy expenditure share: households of a single that shares decrease with lower connectivity or greater access to elderly person or with many unemployed; households headed by resources for which households do not have to pay (for example, females; being connected to electricity; using wood pellets, natu- through ownership of forest plantations or illegal logging). This ral gas, or other sources for heating; and using electricity, natural channel could, for example, be behind the negative coefficient gas, wood/coal/oil, or cylinder for cooking. Having an in-house observed for households with Roma. sewage type also correlates with increased energy expenditure The next chapter adopts a behaviorally informed ap- shares, although this is only significant at the 10-percent level. proach to illuminate the barriers and facilitators of behavior We conclude that energy expenditure shares are part- change in sustainable energy transitions. By delving into the ly related to the technology used for cooking and heating; attitudes and behaviors that contribute to energy poverty, this income levels and socioeconomic factors also play a role. A approach enhances traditional diagnostic methods, offering a more detailed study is required to understand the relationship more nuanced understanding of policy challenges from the per- entirely, because policy conclusions would require additional spective of users or beneficiaries. information that allows controlling for connectivity, household Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 49 Chapter 3 A Behaviorally Informed Approach to Sustainable Energy Transitions Energy poverty is an outcome (or condition) driven by vari- A behaviorally informed approach adds value to tradi- ous factors, including individual and social behaviors. As dis- tional diagnostic approaches by generating a nuanced un- cussed in chapter 2, there are multiple factors correlated with derstanding of policy challenges from the user or beneficiary energy poverty, including the sociodemographic composition perspective, which is particularly relevant to residential sus- of the household, energy-dependent technologies in use, and tainable energy transitions. Standard development policy typi- geographic location. These factors, however, can only be con- cally targets financial resources, incentives, laws, or information fronted per se by first addressing the determinants of behaviors provision—the conventional tools available to policy makers. By that sustain energy poverty, such as the use and choice of en- contrast, a behavioral approach draws on various disciplines in ergy-intensive technologies. An approach informed by insights placing the focus on mindsets, decision-making frames, and the from behavioral science can yield a holistic understanding of the social environment. The result is a more holistic understanding bottlenecks and enablers of behavior change in the sustainable of how decisions are made and actions taken in relation to be- energy transition space. In particular, this approach can help havior change in specific policy contexts. bridge the gap between the condition of energy poverty and the attitudes and behaviors that sustain it. Residential sustainable energy transitions can be un- fordability, awareness, beliefs, and trust and their influence on derstood as the uptake of behaviors and technologies that willingness to transition to more sustainable heating and energy promote the clean and efficient use of energy for household behaviors (captured through a quantitative survey and qualita- purposes. Sustainable energy transitions can center around the tive fieldwork). adoption of modern appliances and technologies that limit the amount of energy consumed in the home (for example, ener- 3.1 Sustainable Energy Use and gy-efficient appliances), conserve energy that has already been Attitudes toward Energy Efficiency generated (for example, insulation), or change the source of energy combustion to a cleaner or more efficient source (for ex- Sustainable energy use revolves around awareness of the ample, away from fossil fuels toward renewable sources or from amount of energy used and the willingness to reduce energy dirtier or more inefficient to cleaner or more efficient fossil fuels, consumption. A recent high-frequency survey conducted in such as from coal to natural gas). For behavioral and technolog- July 2023 shows that most Romanians knew how much ener- ical change to happen at the household and individual levels, gy they used at home. Only a small share indicated that they however, attitudinal change is required, which means that pol- needed to be made aware of how much energy they consumed icies and programs should target both behaviors and attitudes. at home (6.9 percent) and 80 percent of the population were This chapter examines select structural and behavioral aware of peak and off-peak energy use. In addition, nearly 70 determinants of transitions toward sustainable energy prac- percent were concerned about the energy used in their home tices and away from energy poverty. Structural (for example, (figure 29). Nearly three-fourths of Romanians indicated that economic, legal/regulatory, and information) and behavior- they limited their energy use to save energy (figure 30). Still, it al (for example, psychological, social, and contextual) factors must be clarified whether they did so due to energy affordability can explain the transition out of energy poverty and into sus- constraints or environmental concerns. tainable energy practices. As such, it is essential to understand Intentions to upgrade to more-energy-efficient technolo- what correlates with sustainable energy transitions (structural gies are lower, mainly when this concerns heating systems and and behavioral determinants) and to map decision-making insulation. Half of the Romanian respondents were considering processes as individuals consider engaging in more-sustainable upgrading to more-energy-efficient appliances, but only a negli- energy practices. Effectively, the questions to be answered here gible share were considering upgrading their home’s heating sys- are what gets in the way of transitioning out of energy pover- tem or insulation. Based on survey estimates from 2023, half of ty or vulnerability and how these bottlenecks can be addressed the population considered upgrading at least one of their home through sustainable energy policy and targeted programs. First, appliances to more-energy-efficient appliances (figure 30). In this chapter will discuss the behaviors of interest in sustainable contrast, only a negligible share of the population was consider- energy use, supported by the presentation of quantitative evi- ing the upgrading of their home’s heating system (9.4 percent) or dence on attitudes toward energy efficiency (captured through insulation (9.1 percent). Intentions to upgrade heating behaviors a recent survey). The remainder of the chapter will look at (including investing in insulation and complementarities) will sustainable heating transitions, focusing on the barriers of af- be explored in greater detail in the next section of the chapter. 52 Romania Energy Poverty Assessment Figure 29 Awaren ess of En erg y Us ed at H o m e, Figu re 3 0 Co ns id e ra t io n o f Cha nge s in Ene rgy July 2 023 (% ) Us e, J u ly 2 023 ( % ) 100% 4.6 6.1 7.6 100% 4.7 2.2 13 4.7 8 3.7 7.7 9 14.6 15.3 21.8 31.2 29.2 29.6 15.9 33 50% 50% 21.2 57.2 50.6 41.7 39.3 28.1 0% 0% Aware of energy Concerned about Aware of the peak Consider upgrading to Limit my energy use in used at home energy used and off-peak more energy-inefficient order to save energy at home energy us appl. in my home Strongly agree Agree Disagree Strongly agree Agree Disagree Strongly disagree Don't know Strongly disagree Don't know Source: Estimates based on World Bank rapid household survey (July 2023). Despite high concerns over energy use at home, survey ing and could afford to make sacrifices. Nearly 9 percent could responses suggest that most Romanians were unwilling to afford the sacrifices, but were unwilling to make them, and 35 make financial sacrifices to develop renewable energy fur- percent were neither able nor willing to sacrifice for renewable ther. When respondents were asked about their willingness to energy. Meanwhile, the survey shows limited actions are being make financial sacrifices for the societal benefit of sustainable taken in response to energy price increases (figure 32). While 62 energy transitions (in terms of greater use of renewable energy), percent of those surveyed claimed to limit the use of appliances, less than half were willing to make sacrifices, and less than 10 40 percent or less took more-proactive measures such as heating percent reported being able to afford it (figure 31). Nearly 3822 less overall (40 percent) or substituting heating or cooking fuels percent of Romanians were willing to sacrifice for renewable en- (just 37 percent). Over half (55 percent) reported keeping their ergy, but said they could not afford it, while 6 percent were will- habits the same in response to energy prices. 22 A previous survey collected in the city of Cluj- Napoca shows similar results; while most citizens recognize the climate impact of energy consumption, they are unwilling to incur additional costs (Babeș-Bolyai University, 2021). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 53 Figure 31 W i l l i n g n ess to M ake Fi n an ci al Figu re 32 Re s po ns e s to Ene rg y Pr ice Sac r ifices for Furth er Develop men t of I nc reas e s Renewab le En erg y, J uly 2 023 (% ) 100% 40 37.5 90% 34.5 80% 35 70% 30 60% 50% 25 40% 30% 20 20% 10% 15 13.2 0% ho er yu k r h uch y h ergy t a lian st 8.6 erg ea 10 me se e se lia ces ien pp lea my rd om e en e p in o w m 6.1 ou in in o d m en nc ffic e a at ak of th un ch 5 y-e om ade rgy se ou pp sed t h aro mu ne y u erg y h gr s u ou re w en of m er up e e erg d o wa b 0 o pe a h an m a sav en I am willing I am I do not Although I am willing erg ed ff- f d o ed en cern to t my on onsi Ia to but neither know I am able and are yi us cannot able nor to, I am not able to aw n mi Ic is e co afford to willing to willing to I li m I'm Ia to Strongly agree Agree Disagree Strongly disagree Don't know Source: Estimates based on World Bank household survey (July 2023). Findings from qualitative fieldwork conducted in June cussants said the programs had not yet reached their villages or 2023 also point to possible intentions to adopt more-sustain- expressed the belief that the funding would be exhausted when able energy practices, though affordability was once again a people became aware of the benefits. Even with the benefits, the primary concern. During many focus group discussions and cost of installing solar panels was perceived by many as too high. in-depth interviews (see annex 1 for methodology and discus- To install a 3KW power [solar panels] would cost sion guides), when covering the topic of energy efficiency and around 20,000 euros, of which 2000 is your contribu- programs to support energy-efficient investments, participants tion of 10 percent. The problem is that at the moment, highlighted the demand for solar panels for improved residential the cost has risen to about 23,000 euros to install a 3KW energy efficiency. When asked about their attitudes toward en- one; you have to pay around 5000 euros, which is too ergy efficiency upgrades and knowledge of existing support pro- much. (Participant from Vrancea, Reghiu FGD) grams to promote efficient heating practices, solar panels were mentioned with an aspirational tone in many of the discussions, particularly in electricity-based heating solutions (heat pumps 3.2 A Focus on Sustainable Heating for space heating and water heaters). In multiple discussions, Transitions solar panels were mentioned as a way to reduce electric bills. Still, concerns about high upfront installation costs and unreli- Sustainable energy transitions in Romania are very sensitive ability contribute to the use of energy-intensive appliances, such to decisions around residential heating, given the role heating as electric heaters and other appliances, during winter months. plays in energy poverty and vulnerability and the impacts heat- Some discussions unearthed knowledge about energy-efficiency ing can have on climate change and local air quality. For this support programs to install solar panels (most support programs reason, sustainable heating transitions are an essential component mentioned concerned solar panel installation). However, dis- of sustainable energy transitions. Sustainable heating transitions are 54 Romania Energy Poverty Assessment a transition to affordable, reliable, efficient, and low-emission heat- The current state of heating and ing practices (World Bank 2023b). Residential heating in Romania intentions to upgrade technologies varies according to the heating system and appliance type. How- ever, an important share of the population needs more-efficient Most Romanians—around 8 out of 10—are classified as using appliances and systems. While both heating and cooking practices traditional heating technologies, meaning there is substan- are highly correlated with the share of expenditures going to energy tial scope for the adoption of improved and more-sustainable (chapter 2), the energy intensity needed for heating is dispropor- heating practices and behaviors. Traditional (conventional) tionately higher and the ability to substitute energy sources is more heating is defined by heating systems that burn fuels less cleanly limited. By improving heating practices and implementing ener- and efficiently than modern devices. These are old-style stoves, gy-efficient technologies, Romania can reduce its overall energy space heaters, fireplaces, or distribution systems (boilers and consumption and decrease its dependence on fossil fuels. otherwise) that typically burn raw fuels (unprepped firewood, Given the scope of energy poverty and vulnerability in Ro- lump coal, or solid waste) or natural gas (albeit inefficiently). In mania, it is important to map the factors that limit the initia- Romania, just over half of survey respondents reported using tion of the types of sustainable energy transitions that would traditional heat-distribution systems (those that produce heat in overcome those challenges, particularly in the heating space. one home location and distribute this to other areas of the home Understanding the structural and behavioral barriers preventing through pipes and radiators or vents) (figure 33).23 Twenty-eight households from transitioning out of energy poverty and adopt- percent of Romanians heat their homes with traditional stoves, ing cleaner technologies and sustainable energy use is essential, electric space heaters, or fireplaces (without heat distribution because these must be considered in order to effectively inform to other home areas). Most of those with modern heating sys- any policies or programs targeting the challenge. While financial tems (9 percent) use high-efficiency heat distribution systems. barriers, in many cases, are binding constraints, the existence Ten percent of Romanians are connected to centralized (that is, of support programs (albeit limited) signals that other factors heat generated at the building level and distributed to individual influencing decisions need to be explored. This exploration can units) or district (that is, heat generated at the local or municipal be accomplished using a mixed methods approach combining level and distributed to properties) heating system. qualitative and quantitative research methodologies. Figure 33 Heati n g Sys tems by Category Traditional Modern District Heating Heating Heating 50% Percent of the population 40% 30% 20% 10% 0% Traditional stove, Traditional heat High efficiency High efficiency District or Does not space heater, or distribution stove or space heat distribution centralized know fireplace system heater system heating Source: Estimates based on World Bank household survey (July 2023). 23 Respondents were asked what type of heating device was primarily used to heat the home and were prompted to specify whether this was traditional or modern (as well as if it had distribution or not). As such, respondents could report the same type of device (natural gas heat distribution system) but a different class (traditional or modern). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 55 Regarding heating fuels among Romanians with individ- coal pellets, electricity, or other fuels. Similarly, around 80 per- ual rather than central or district heating systems, firewood cent of Romanians used natural gas in heat distribution systems, was used most commonly in traditional and modern heating with low shares of firewood used for traditional systems (16 per- systems, followed by natural gas. Nearly 80 percent of tradi- cent) and modern systems (8 percent) and negligible shares of tional and over 80 percent of modern stoves in Romania burned other fuels (figure 34).24 firewood and fewer than 10 percent used wood pellets, raw coal, Figure 34 Heati n g Fuel s by Sys tem an d Type Percent of population within heating system type 100% Traditional heating Modern heating 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Traditional stove, space Traditional heat High efficiency stove or High efficiency heat heater, or fireplace distribution system space heater distribution system Firewood Wood pellets Raw coal Coal pellets Other solid Natural gas Electricity Source: Estimates based on World Bank household survey (July 2023). Intentions to upgrade heating systems or insulation insulation. Around two-thirds of these planned to upgrade were very low in Romania, with around 1 in 10 respondents within the next two years, while the rest may need up to four. planning to do upgrades. The majority of Romanians were These shares are similar across different genders; in terms of not planning to upgrade heating or insulation within the next age, people over 60 had a lower propensity to upgrade. Some years (figure 35). On average, only 9 percent of those with tra- geographical differences also exist; most notably, respondents ditional heating systems planned to upgrade the heating appli- from the South-East expressed a much lower propensity to ances or systems and just 9 percent planned to upgrade their upgrade heating or insulation. 24 According to World Bank (2023), heating that relies on coal, peat, oil and other petroleum products is classified as unsustainable, whereas natural gas is classified as a better alternative, though only a medium-term solution. Biomass (such as wood) is considered to be sustainable if the management of the resource is “certified and follow[s] a robust regulatory framework. . . . Natural gas can have a role in providing household and business heating solutions in some countries over the medium term as a transition fuel provided it is compatible with a country’s goal of long- term decarbonization. For biomass to be a sustainable fuel in line with the definition cited above, the management of the underlying biomass resource must be certified and follow a robust regulatory framework.” (page 1-2) 56 Romania Energy Poverty Assessment Figure 35 Roman i an s’ I n ten ti on s regard ing a H ea t ing o r I ns u la t io n U pg ra d e w it hin t he N ext Fe w Years and Ti me Frame for Up g rade Plan to upgrade? If yes, in how many years? 100% 50% 80% 40% 60% 30% 40% 20% 20% 10% 0% 0% Heating Insulation Heating Insulation Yes No Unsure Within a year 2 years 3 years 4 years Source: Estimates based on World Bank household survey (July 2023). Capturing intentions to upgrade heating systems and in- few years. Two-thirds of this group were ready to upgrade in sulation allows us to establish a typology of traditional heat- the following two years, while the rest would need more time. ing users and all respondents. Panel a of figure 36 shows the Fifteen percent needed clarification about the upgrades, while proportions of the population with traditional heating systems three-quarters were unwilling to upgrade. Panel b of figure 36 willing, unwilling, or unsure with regard to updating. House- presents the same breakdown for all Romanians regarding in- holds with modern heating systems were excluded from this sulation upgrades. Nearly 80 percent were unwilling to initiate calculation because they had already made the transition. As of a home insulation upgrade. Those ready to upgrade insulation 2023, traditional heating users were around 80 percent of the were around 7 percent of the total population, while the other 3 total population. The analysis shows that only 9 percent were percent were planning to upgrade between 3 to 4 years from now willing to upgrade to a new heating system within the next and 13 percent were still deciding whether they should upgrade. Figure 36 Roman i an s w i th Tradi ti on al H ea t ing Sys te m s ( A l l H o u s e ho ld s ) Who Fal l into a S pe c if ic Profile Accordi n g to Th ei r I n ten ti on to U pg ra d e Sys te m s ( I ns u la t io n) Panel a. Heating system upgrades Panel b. Insulation upgrades Traditional heating users All respondents Willing to Unsure about Unwilling to Willing to Unsure about Unwilling to upgrade (9.4%) upgrades (15.1%) upgrade (75.5%) upgrade (9.1%) upgrades (12.6%) upgrade (78.4%) Ready to Ready to upgrade (5.8%) upgrade (6.5%) Cautiously Cautiously ready (3.6%) ready (2.6%) Source: Estimates based on World Bank household survey (July 2023). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 57 Profiling sustainable heating the most critical among Romanian participants in relation transitions: behavioral barriers to to sustainable heating transitions. Respondents and partic- and enablers of sustainable heating ipants in the survey, focus group discussions, and in-depth transitions interviews highlighted difficulties in upgrading heating prac- tices even if part of the upgrade cost was covered through a Barriers (and enablers) to sustainable heating transitions fall program. While heating appeared to be a priority for many under financial, information, and attitudinal. Financial barriers households in the winter (related to other expenses), the per- refer to those perceived or real affordability concerns that block in- ceived capability to take action on these priorities needs to tentions or actions to change behaviors. Information barriers refer be improved, particularly among economically disadvantaged to knowledge and awareness gaps that limit the target population’s households. understanding of the need to engage in sustainable heating tran- Perceived financial difficulties among the surveyed sitions, the tools at their disposal to support this transition, and population with respect to heating and insulation upgrades how information about technologies and programs is presented. were very high. The share of Romanians who would find it Attitudinal (understood loosely as the beliefs, values, ideas, and difficult financially to cover the costs of upgrading the heating perceptions) barriers and enablers refer to the cognitive and social system is around 80 percent, regardless of any actual plans to factors that influence decision-making in the context of sustain- upgrade either the heating system or insulation. Nonetheless, able heating, including beliefs about the costs and benefits of up- those unwilling to upgrade their heating systems reported a graded heating practices, social norms about sustainable heating significantly higher level of financial difficulty compared to practices, and trust in messengers and upgrade facilitators. In this those willing to do so, suggesting that the affordability of the section, we link these barriers and enablers to intended changes in new heating technologies may play an important role in de- behavior (intentions to upgrade the heating system or insulation). terring people from upgrades. Moreover, only 6 percent of the population indicated they were both willing and able to make Financial barriers financial sacrifices to advance renewable energy further. At the According to the quantitative and qualitative data collec- same time, the majority was either not willing or not able to tion, financial and affordability-related barriers are by far or both (figure 37). Figure 37 Percei ved Fi n an ci al Di ffi cul ty A cco rd ing to Pro f ile s ( % o f Po pu la t io n w it hin U pg ra d e Intent ions G roup ) Financial difficulty - Heating upgrade Financial difficulty - Insulation upgrade 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Willing to Unwilling to Unsure about Already Willing to upgrade Unwilling to Unsure about upgrade upgrade upgrades upgraded upgrade upgrades Very difficult Difficult Easy Very difficult Difficult Easy Very easy Don't know Very easy Don't know Source: Estimates based on World Bank household survey (July 2023). 58 Romania Energy Poverty Assessment Perceived financial barriers vary across regions, with the upgrade. Conversely, respondents in the Bucharest-Ilfov region highest financial difficulty reported in the country’s South-East believed they faced lower financial barriers to meeting the costs region. Exploring the perceived costs of upgrading highlights that of upgrading. The responses did not suggest any gender difference these perceived barriers are higher in certain parts of the country. with respect to perceived financial difficulty, through in terms of A much higher proportion of respondents in South-East Roma- age people over 60 seemed to find upgrading more financially nia expressed the belief that it would be financially difficult to difficult than younger respondents (figure 38). Figure 38 Percei ved Di ffi cul ty i n Sp en d ing Mo re to Ma inta in t he S a m e Leve l o f Co m fo r t w it h a Moder n Heati n g Sys tem by Reg i on Upgrading cost: financial difficulty 70% 60% 50% 40% 30% 20% 10% 0% Bucharest-Ilfov Center North-East North-West South-East South- South-West West Total Muntenia Oltenia Very difficult Difficult Easy Very easy Don't know Source: Estimates based on World Bank household survey (July 2023). Concerns over the affordability of such upgrades also Well, when you install it, it’s difficult, financially came out strongly in the qualitative fieldwork. Financial con- wise. After that, yes, these costs can be greatly reduced cerns over upgrades can relate to the initial investment costs in two to three years. . . . If you install panels and elec- (fixed costs) or the use costs (variable costs). Both are relevant in tric heating with a heat pump, costs should reduce in regard to sustainable heating upgrades (for example, fixed costs two to three years, after which heating comes free. (Par- include purchasing and installing devices and variable costs re- ticipant from Mehedinți Drobeta Turnu Severin FGD) fer to the cost of fuel or maintenance). The most frequently men- tioned concern related to upgrading was the installation price, in Information barriers line with the financial concerns outlined above. Even those who Only a small share of the population were aware of and un- believed it would have long-term benefits mentioned the prob- derstood existing programs that could support them finan- lem of having enough financial resources to face the expenses. cially in upgrading to modern heating systems. The survey The investment would be too high, and I’m not sure shows that half of the population was aware of programs/ini- how many years it would take to recover it, not to tiatives to financially support investments in modern technol- mention that you will require maintenance during all ogies for heating their home more efficiently and keeping it these years, and I don’t think it’s worth it. (Participant warm. Still, only 11.3 percent fully understood them the details from Vrancea Răstoaca FGD) (figure 39). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 59 Figure 39 Awaren ess of Prog rams or I nit ia t ive s to Fina nc ial ly S u ppo r t I nve s t m e nt s , J u ly 2 023 ( % ) 60.0 49.0 50.0 40.0 30.0 19.7 20.0 20.0 11.3 10.0 0.0 Yes, and I fully Yes, and I partly Yes, but I know No, I am not aware understand th know about a nothing about of any prog Source: Estimates based on World Bank household survey (July 2023). Knowledge of support programs varied significantly and 25 percent believed they fully understood how the subsi- depending on intentions to upgrade. Awareness of subsi- dies and programs worked. Two-thirds of those unwilling to dies was much higher in the willing-to-upgrade group com- upgrade had yet to learn of any program. Awareness of subsi- pared to those unwilling to do so. About 80 percent of the dies could also be higher in the unsure about upgrades group willing-to-upgrade group knew about subsidies and programs (figure 40). Figure 40 Awaren ess of Sub s i di es (% o f Po pu la t io n w it hin U pg ra d e Inte nt io ns Gro u p) Subsidies awareness - Heating upgrade Subsidies awareness - Insulation upgrade 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Willing to upgrade Unwilling to Unsure Willing to upgrade Unwilling to Unsure upgrade upgrade Yes, Fully understand Yes, Partly understand Yes, Fully understand Yes, Partly understand Yes, but know nothing No Yes, but know nothing No Source: Estimates based on World Bank household survey (July 2023). Awareness of support programs was sensitive to the de- 41). Geographical differences were also present, with there being mographic characteristics of the population. Men in Romania much higher awareness in the Bucharest-Ilfov and North-East were slightly more informed than women about subsidies and and lower awareness in the Center, South-East, and North-West programs—the same was true for people ages between 30 to regions, where around two-thirds of the respondents were en- 60 with respect to either younger or older populations (figure tirely unaware of subsidies. 60 Romania Energy Poverty Assessment Figure 41 Awaren ess of Prog rams an d S u bs id ie s to S u ppo r t t he U pg ra d ing o f H ea t ing Sys te m s o r Insulat ion by G en der an d Age Awareness (gender) Awareness (age) 60% 70% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% ms d ini and es pro now ms s ram ing ini and es ini f any es d tia d s pro now ms s ram ing ini and es ini f any es ive ve ive gra stan gra stan ini s an tiv tiv tiv tiv tiv gra gra rog oth rog oth i iat t iat k k s s tia tia tia tia tia o o m pro der pro der ly ly h p ow n h p ow n nit nit ms are ms are art art ble un ble un di di gra aw gra aw le le Ip ta dIp or or t s I kn t s I kn an an lab lab ila lly ila lly pro not pro not d ava I fu ava I fu ab s, an ab s, an vai vai t t uc uc ab s, bu ab s, bu am am ta the , and the , and Ye Ye ou ou ou ou Ye Ye ,I ,I No No s s Ye Ye Male Female 30-60 under 30 over 60 Source: Estimates based on World Bank household survey (July 2023). Similar insights about the low awareness of support pro- But anyhow, you don’t have entrance, you or I. / May- grams were identified in the qualitative fieldwork. The need be . . . at a certain point we will. / Only people working for wider awareness of existing subsidies was evident in some at the village hall. / The one that’s higher [in the hier- of the group discussions and interviews where, in most cases, archy]. / I can’t gain access. Maybe you do. (Participant knowledge could be improved. Some interviewees attributed from Vrancea Răstoaca FGD) this need for more awareness to limited transparency and scarce advertisement of existing subsidy programs, such that only those We, poor people, don’t get to see or . . . receive what actively seeking information would find it. they should offer there. Besides, in the back, the di- rector, the engineer, a friend, and others come to fill No, it’s not transparent enough. It’s not known. It was their bags and leave through the back door. . . . Noth- scarcely advertised, but just for interested parties, who ing is done. We, as Roma people, are left as no one’s went read and tried to find out, but for it to become children. (Participant from Mehedinți Drobeta Turnu visible, for you to understand it and take the neces- Severin FGD) sary steps, or for someone to come, to talk about it on a larger scale, it wasn’t accessible. (Participant from Some participants also highlighted technical problems Vrancea Adjud FGD) in the application process that may deter people from ap- plying. Challenges in the application process relate to the ap- In particular, a lack of awareness translated into a sen- plication’s being exclusively online, which might pose some timent of exclusion regarding intended beneficiaries. Lack problems for older people. Also, web pages with connections of transparency was perceived as intrinsic to the procedure timing out too early made it difficult for applicants to upload used to select who gets the benefit. Some respondents believed all the necessary documents. they would not benefit even if they met all the requirements, although it needs to be clarified whether this was linked to a It requires digital abilities that many people do not direct experience or personal belief. own. . . . For example, to know how to scan documents, Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 61 to go onto the portal, to know how to update data and / In Adjud, mainly, there are more older people than information, and to upload documents. And all this young ones, and to put in information very quickly, quickly, to get the chance. . . for someone familiar with for an older person, is harder work. (Participant from working on a computer, it’s ok. The ones who’ve ob- Vrancea Adjud FGD) tained the solar panels know how to use the platform. These results show that even though there was a particular But who doesn’t have any digital skills is excluded. (Par- awareness of existing programs, knowledge of to whom and ticipant from Vrancea, Reghiu FGD) how to apply and friction in the application process can be a There is a platform, and we go on it, and we have lim- barrier to obtaining the subsidies. ited time . . . and we have to get on it, and in 3 to 5 min- Attitudinal barriers utes we have to send the documents . . . and there are some people who do this for many more. (Participant Regarding negative beliefs, fieldwork participants signaled from Vrancea Adjud IDI) concerns over the complexity of upgrades and the inconve- nience associated with them. Concerning barriers to heating or One participant also cited subsidy funds’ being depleted insulation upgrades, around two-thirds of Romanians believed in a few minutes in some regions, so that people were not upgrading heating or insulation involved a great deal of inconve- allowed to apply. nience and that having flexibility in what to burn was essential. Yes, but I want to say that in some regions, the funds Forty-two percent of Romanians would only upgrade if their finished in 2 minutes. For Vrancea, it was 6 minutes. friends and neighbors did the same (figure 42). Figure 42 Percei ved B arri ers to Heati n g a nd I ns u la t io n U pg ra d e s by Inte nt io ns to U pg ra d e Barriers - Heating upgrade Barriers - Insulation upgrade Fuel flexibility is Fuel flexibility is important to me important to me Upgrading involves Upgrading involves a lot of hassle? a lot of hassle? 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% Average Already upgraded Unsure Unwilling to upgrade Average Unsure Willing to upgrade Unwilling to upgrade Willing to upgrade Source: Estimates based on World Bank household survey (July 2023). Differences in beliefs did not vary meaningfully by the different regions. In particular, the inconvenience of upgrad- sociodemographic composition of the Romanian population, ing was perceived more in North-East and West Romania. In but important regional differences were identified through contrast, fuel flexibility was perceived as more important in the the quantitative survey. No clear gender difference was iden- South-West, South, and North-East regions. Social influence tified with respect to these barriers, while Romanians under 30 was much lower in the Center, North-West, and West regions seemed to perceive them as being lower. Nonetheless, some dif- compared to the others. North-West, South-East, and Center ferences in the perception of these barriers was evident across Romania also perceive these barriers are being lower (figure 43). 62 Romania Energy Poverty Assessment Figure 43 B arri ers to Up g radi n g Ei th er H ea t ing o r Ins u la t io n by Re g io n Barriers to heating/insulation upgrades Total West South-West Oltenia South-East North-West North-East Total Center Bucharest-Ilfov 0% 20% 40% 60% 80% 100% Upgrading involves a lot of hassle? Fuel flexibility is important to me More likely to upgrade if others do the same Source: Estimates based on World Bank household survey (July 2023). Negative beliefs, as captured in qualitative fieldwork, ipants said they would trust the system because they had had revealed a need for more trust in the efficiency and sus- positive experiences in past applications. Among the reasons tainability of specific modern heating systems, particularly cited for low trust in institutions was the belief that the process electricity-based ones. The main concerns mentioned in the of assigning subsidies needed to be more transparent, with ad- focus group discussions were a lack of trust in the efficiency of ministration officers favoring themselves or their peers over the modern systems (especially solar panels), safety concerns related general public. to burning or blowing up, the lack of preparation of technicians They take the information and the benefit from all and repair persons, the need for a network development from [such] programs. Some people. (Participant from Me- the administrations, and high costs of installation and service. hedinți Drobeta Turnu Severin FGD) I’ve learned that there are some batteries which can be adapted to this type of heating system, in case elec- I trust that we will receive part of the investment tricity is cut off. . . . There’s a risk of explosion. There’s funds, and most likely, there are a few competent peo- a place where this happened, here in the village. (Par- ple who know what they’re doing, but we don’t trust ticipant from Mehedinți Ilovița IDI) them because we all know that some of that money will end up in hands where it shouldn’t go. . . . Often, The thermostat could break down on the electric these programs are made so someone else could profit heater, and then what? (Participant from Mehedinți from them and not for the normal people. (Participant Drobeta Turnu Severin FGD) from Mehedinți Izvoru Bârzii FGD) Low institutional trust in the sustainable heating tran- Interpersonal and institutional trust can also serve as a sition space was consistent across all fieldwork. Supporting barrier or enabler to sustainable heating transitions, specif- our survey results, trust in administration or government was ically as these relate to preferred messengers and behavior limited at all levels in the hierarchy. Nevertheless, some partic- change facilitators. One area where trust serves as an important Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 63 barrier or enabler pertains to preferred sources of information tion sources for heating or insulation upgrades, technicians were about heating upgrades. Trusted sources of information dif- the most trusted sources by around 50 percent of the population, fered in the different profiles of Romanians. Those unwilling to followed by relatives, friends, and neighbors at around 22 per- upgrade had a much lower propensity to trust technicians and cent (figure 44). Substantially lower trust was given to govern- other similar groups, but a higher propensity to trust family, ment officials and doctors and health workers. Concerning trust, friends, or neighbors. This same group of respondents also had a the responses were similar across different genders and ages. At higher propensity to distrust information sources. Government the same time, respondents from the Center, North-East, and officials and health workers were trusted by a substantially low West regions seemed less prone to trust any of the abovemen- share of respondents in all groups. Regarding trusted informa- tioned sources of information. Figure 44 Trus ted M ess en ger for I n form a t io n a bo u t H ea t ing a nd Ins u la t io n U pg ra d e s Trust in information sources - Heating upgrade Trust in information sources - Insulation upgrade 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Relative, Technician Government Doctor None Relative, Technician Government Doctor None friend official or of the friend official or of the or neighbour health above or neighbour health above worker worker Willing Unwilling Unsure Already to upgrade to upgrade upgraded Willing to upgrade Unwilling to upgrade Unsure Source: Estimates based on World Bank household survey (July 2023). Low institutional trust was closely tied to experience or Attitudinal enablers expectations about involvement in a government support Although nearly 7 out of 10 Romanians perceived upgrades program. Lack of trust also related to the applicant’s percep- as an inconvenience, a similar amount were aware of the dif- tion of an excessive bureaucracy. Some participants stated they ferent benefits of doing so. Sixty-six percent perceived these were asked for additional documents once they presented the upgrades to be an inconvenience. Yet approximately 50 to 60 necessary documentation according to the application form. percent of the population indicated that upgrading would yield But if you go to the town hall and ask, “I want to ask benefits, such as lower heating bills, convenience or time sav- about these European funds,” they say, “This, this, ings, increased property value, better family health, better air this, this, this, this,” they give a list of 20 files, you quality, and a better environment. Around 4 out of 10 Roma- go and collect them and when you go back, and they nians indicated that they would be more likely to undertake say, “You’re missing this, that’s missing, that’s miss- these upgrades if surrounding communities did so as well. ing, that’s missing.” After this, you go with those, and The perceived benefits of sustainable heating upgrades when you go the third time, they say, “The funds are fin- are relatively high in Romania. Regarding the perceived bene- ished.” (Participant from Mehedinți Izvoru Bârzii FGD) fits of upgrading heating systems, Romanians showed a slightly 64 Romania Energy Poverty Assessment increased agreement over present-related benefits such as lower might play a role in the decision to upgrade (figure 45). A slightly bill charges, convenience, and increased property value compared larger proportion of men than women perceived the upgrades to climate- or health-related benefits. However, those who were as having benefits. In terms of age, the proportion of Romanians unsure or unwilling to upgrade showed less agreement with re- over 60 who believed that upgrading would yield benefits was gard to these benefits, indicating that the lack of perceived benefits smaller than the proportion of the younger population. Figure 45 Percei ved En ab lers of Heati ng a nd Ins u la t io n U pg ra d e s by I nte nt io ns to U pg ra d e Enablers - Heating upgrade Enablers - Insulation upgrade More likely to upgrade More likely to upgrade if others do the same if others do the same Helps climate change/ Helps climate change/ environment environment Helps local air quality Helps local air quality Improve family health Improve family health Increase property value Increase property value Save you time or be Save you time or be more convenient more convenient Lower heating bills? Lower heating bills? 0% 20% 40% 60% 80% 100% 0% 30% 60% 90% Average Already upgraded Average Unsure about upgraded Unsure about upgraded Unwilling to upgrade Unwilling to upgrade Willing to upgrade Willing to upgrade Source: Estimates based on World Bank household survey (July 2023). Supporting the findings from the quantitative survey, My husband had a problem with the heating appli- positive beliefs, as captured through qualitative fieldwork, re- ance and went to an authorized shop in Focșani. . . . vealed some of the expected benefits from heating upgrades. He asked for advice from someone in the field, just as During the focus group discussions, benefits such as an increase that man said. So, he knew what he was saying about in comfort and a decrease in time spent heating the home were utilizing and installing heating systems. He offered frequently mentioned by participants. Less frequently raised good advice. (Participant from Vrancea Răstoaca IDI) were the benefits to health and the environment. One participant also suggested that the involvement of But the comfort . . . because you don’t have to take people experienced in navigating the bureaucracy in the sub- out the ash, chop the wood, put wood on the fire. You sidy application process would make it easier to opt for an just click a button and look on your phone. (Participant upgrade. from Mehedinți Drobeta Turnu Severin FGD) It would be good if the ones that know and deal with Evidence from qualitative fieldwork opens the opportuni- installing [such systems] would deal with the bureau- ty to consider trusted messengers and facilitators as enablers. cracy issues. It would be good for them to deal with all In most focus groups, friends and family appeared to be the aspects of bureaucracy so that people can access [these most relied-upon sources of information regarding heating and programs]. I mean, someone does deal from end to insulation. That said, technicians and energy professionals were end with the whole thing. For both a block of flats and cited as reliable sources of information, particularly regarding individual homes, some older people don’t know the heating systems. necessary steps to take. And if the one that installs the Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 65 system could deal with everything, it would be perfect. characteristics, the various behaviors and enablers, and the in- (Participant from Vrancea Adjud FGD) tentions to upgrade heating and insulation. For this, a robust probit model was used to estimate the influence of attitudes, Finally, a generally positive attitude toward considering awareness, trust, and sociodemographic characteristics on the subsidy programs was expressed in the group discussions, propensity to upgrade heating or insulation within the next few mainly on the condition that the governmental contribution years. to the investment was relevant. Considering Romanians and their household characteris- I would make a start towards this change. If the state tics, many factors are associated with intentions to upgrade helps, meaning the government offers us . . . I don’t heating behaviors. Regional differences significantly influence know . . . at least 50 percent of the entire sum of the in- upgrading intentions, with propensity being around 5–8 per- vestment, I would agree to opt for such an investment. centage points higher among respondents from the West, South, (Participant from Mehedinți Ilovița IDIs) South-West, and North-East regions (figure 46). While house- hold size is not associated with intentions to upgrade heating, These findings suggest there are several enablers of the other demographic factors are associated with lower intentions upgrading of heating systems. Beliefs in the benefits of modern to upgrade heating: for example, households with 1 child (rel- heating systems accompanied by community action and trust- ative to those with 0) are nearly 4 percentage points less likely ed messengers available to guide households through upgrad- to want to upgrade, while respondents over 60 are nearly 7 per- ing and maintaining the appliances are vital to consider when centage points less likely to upgrade relative to young adults. persuading households to get involved in upgrading programs. Meanwhile, there is no association between levels of education People also expect the government to reciprocate their efforts by and intentions to upgrade heating. subsidizing them financially. Regarding other factors, many financial, information, and attitudinal barriers and enablers also display an essen- Key factors correlated with intentions tial association with intentions to upgrade heating behaviors. to upgrade heating. Awareness of subsidies is also significantly connected to inten- tion to upgrade, with fully aware respondents being 5 percent- While the previous analysis signaled a variety of factors that age points more likely to be willing to upgrade and unaware could be influencing intentions to upgrade heating behaviors, respondents 3 percentage points less likely to do so. Attitudes a question remains about which factors appear to be binding play a more limited role, with higher propensity found for those in the Romanian context. To answer this question, regression believing upgrades increase property value and those who give analysis was used to isolate the association between respondent importance to fuel flexibility. 66 Romania Energy Poverty Assessment Figure 46 I n fl uen ce of Di fferen t Factors o n t he Pro pe ns ity to U pg ra d e H ea t ing Sys te m Heating upgrades Region: West Region: South-Muntenia Attitude Upgrades: Flexibility is important Region: South-West Oltenia Region: North-East Subsidies: Fully aware Household size: More than two Attitude Upgrades: Saves time Attitude Upgrades: Increase property value Attitude Upgrades: Improve health Region: Center Higher Edu class Region: South-East Region: North-West Attitude Upgrades: Involves a lot of hassle Lower Edu class Gender: Female Age: under 30 Attitude Upgrades: Helps climate change Children: More than one Attitude Upgrades: Lowers heating bills Household size: Couple Trust government Attitude Upgrades: Helps air quality Attitude Upgrades: If others do the same Subsides: Not aware Children: One Age: over 60 -8% -6% -4% -2% 0% 2% 4% 6% 8% Source: Estimates based on World Bank household survey (July 2023). The x-axis plots the percentage point change in intentions to upgrade (the marginal effects from a Probit regression) Note: A solid blue bar means the result is significant. Similar individual and household characteristics are Financial, information, and attitudinal barriers and en- associated with intentions to upgrade insulation. Regional ablers are similarly limited in insulation upgrade intentions differences are also a significant determinant in the propensity related to respondent and household characteristics. Those to upgrade insulation. As with heating upgrade intentions, re- unaware of subsidies (relative to those who are somewhat aware) spondents from the South-West, South, and North-East regions are around 3 percentage points less likely to intend to upgrade have a higher propensity to upgrade insulation (figure 47). The insulation. Other factors show unexpected results, which are only demographic factor associated with insulation upgrade in- likely driven by an already adequate level of insulation in the tentions is the respondent’s age: Romanians over 60 are nearly home: those who believe heating and insulation increase prop- 4 percentage points less likely than young adults to consider erty value are less likely to upgrade. Meanwhile, those who view insulation upgrades. fuel flexibility as necessary are nearly 4 percentage points more likely to upgrade insulation. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 67 Figure 47 I n fl uen ce of Di fferen t Factors o n t he Pro pe ns ity to U pg ra d e Ins u la t io n Insulation upgrades Region: South-West Oltenia Region: South-Muntenia Region: North-East Attitude Upgrades: Flexibility is important Household size: More than two Region: North-West Region: Center Attitude Upgrades: Helps climate change Attitude Upgrades: Helps air quality Region: West Gender: Female Household size: Couple Higher Edu class Region: South-East Subsidies: Fully aware Attitude Upgrades: Saves time Attitude Upgrades: If others do the same Children: One Lower Edu class Attitude Upgrades: Improve health Age: under 30 Attitude Upgrades: Involves a lot of hassle Children: More than one Attitude Upgrades: Lowers heating bills Subsides: Not aware Trust government Age: over 60 Attitude Upgrades: Increase property value -6% -4% -2% 0% 2% 4% 6% 8% 10% Source: Estimates based on World Bank household survey (July 2023). The x-axis plots the percentage point change in intentions to upgrade (the marginal effects from a Probit regression) Note: A solid blue line indicates a significant result. 68 Romania Energy Poverty Assessment Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 69 Chapter 4 Simulating the Ex-Ante Impacts of Energy Price Increases on Households Taking stock of the magnitude of energy inequities is critical affordability, energy poverty, and at-risk poverty rates. In addi- for designing targeted policies to protect the most affected tion to the direct impacts, this section further explores potential groups. Who bears the burden of the recent rise in energy prices, indirect consequences of energy price increases. These indirect given the unequal energy spending patterns shown in chapter 2? effects manifest through the price changes of other essential goods What are the potential impacts on energy poverty, poverty, and and services, thus providing a more comprehensive understand- inequality? This chapter delves into a simulation of the potential ing of the broader socioeconomic ramifications. By examining ex-ante welfare impacts of energy price increases, shedding light these interconnected ripple effects, this research aims to provide a on the heterogeneous impacts on households due to different ex- comprehensive outlook on how energy price fluctuations can trig- penditure patterns. We employ microsimulation methodologies ger price changes throughout the economy, particularly affecting to examine the potential direct impacts of price spikes on energy vulnerable households along the income distribution. 4.1 Direct Impacts ternative energy sources unaffected by price hikes, such as ille- gally sourced wood or self-generated resources. Given the large variability of residential price elasticity of residential electricity Impact on energy poverty rates demand in the literature, we consider various scenarios: a price elasticity of 0, -0.25, -0.5, -0.75, and -1. Moreover, we assume a The impact of rising energy prices on the shares of household uniform energy price elasticity for all households, irrespective expenditures allocated to energy consumption, energy pov- of their position in the welfare distribution. To test the sensitiv- erty, and welfare is felt through two main channels. Firstly, ity of our findings, we also examine variations in energy price there is a direct effect on the share of household expenditure elasticity across income quintiles. Low-income households may dedicated to energy due to the elevated costs associated with have low residential price elasticity, because substantial substi- energy consumption. Second, indirect effects may occur as the tutions away from energy might be challenging, particularly price of energy affects the prices of other consumption goods, during winter. Furthermore, given the uncertainty of future in turn affecting households’ purchasing power. price variations, we incorporate diverse scenarios of average The measurement of price elasticity of energy demand residential energy price increases, encompassing a range of 10, plays a crucial role in determining the direct impact of en- 20, 30, 40, and 50 percent increments. ergy prices or policy changes on welfare; therefore, we con- We incorporate additional scenarios by varying house- sider multiple elasticity scenarios to explore the sensitivity hold income levels to consider the potential income effects. of the welfare of Romanian households to increasing energy We examine a range of income decreases, namely 0, 2.5, 5, 7.5, prices. The importance of price elasticity of energy and elec- and 10 percent. These income variations account for changes tricity demand when evaluating welfare effects is discussed am- in energy expenditure shares resulting from fluctuations in in- ply in several studies (for example, Miller and Alberini 2016; come. While the income effect could potentially influence ener- Burke and Abayasekara 2018). The extent of this increase relies gy shares through alterations in energy consumption, we assume on the price elasticity of energy, which determines the house- that this income effect is already encompassed within the price hold’s responsiveness to changes in energy prices. In response to elasticity. In other words, we interpret the price elasticities as higher energy prices, households may opt out to reduce energy encompassing both the substitution and income effects, as per consumption, adopt energy-efficient measures, or switch to al- the underlying rationale. Figure 48 Harmon i zed Con s umer Pri ce I nd ex , To tal a nd Ene rg y ( J a nu a r y 2 021 –May 2 024 ) 160 150 Index (Base 2021) 140 130 120 110 100 90 80 y2024M11 y2021M11 y2022M11 y2021M09 y2022M09 y2023M09 y2021M05 y2022M05 y2023M05 y2021M01 y2022M01 y2023M01 y2021M03 y2022M03 y2023M03 y2021M07 y2022M07 y2023M07 y2024M1 y2024M3 y2024M5 HICP Index Total HICP Index Energy Source: Eurostat (2024). Link: Harmonised index of consumer prices (HICP). Note: The HICP gives comparable measures of inflation for EU countries and the country groups for which it is produced. The HICP for energy includes electricity, gas, and other fuels. 72 Romania Energy Poverty Assessment In our analysis, we establish a baseline scenario that we late alternate energy price scenarios with increases ranging from consider the most probable situation households encounter 0 to 50 percent. in the current context. This baseline scenario encompasses an Out of the multiple energy poverty metrics discussed in average increase in energy prices of 40 percent, a price elasticity section 2, we opt to use a specific measure (P10) for present- of -0.25, and a reduction in household income of 2.5 percent. ing our results in both this chapter and chapter 5. Under this The chosen price increase is based on the observed price data measure (P10), energy poor households are those who spend from the Harmonised Index of Consumer Prices (HICP) for more than 10 percent of their household income on energy. We energy published by Eurostat (2024), specifically from Janu- choose this measure for several reasons. First, this is the only ary 2021 to May 2024 (40 percent, as depicted in figure 48). absolute measure among the expenditure-based measures, al- It should be noted that we interpret the price increase as an lowing comparisons against a fixed standard. Second, this mea- absolute price change. However, in relation to the overall price sure allows for cross-country comparability. Finally, though increase, the relative price increase, as indicated by the harmo- using the official definition of energy poverty drafted in the Law nized energy and housing price index, is significantly smaller. 226/2021 would be ideal, it is still not clear how this measure Regarding household income decreases, we assume a moderate can be operationalized. to slight reduction, given that real GDP has continued to grow Our simulation results indicate that within the baseline since 2021 (the rates were 5.9 and 4.8 percent in 2021 and 2022, scenario with a 40 percent rise in energy prices, energy poverty respectively).25 Additionally, the unemployment rate remained (P10) may see an average increase of 2.3 percentage points; relatively stable during this period, standing at 5.6 percent in moreover, vulnerable groups (households receiving support 2021–23, according to data from the IMF. Consequently, we 26 from municipalities, those receiving disability benefits, and argue that the decline in household income will likely be slight. those of single elderly persons) are more affected than other While we define a baseline scenario in which energy pric- groups. We analyze the impact of energy price increases on en- es increase by 40 percent, we also consider some alternative ergy poverty rates in our baseline scenario (a price increase of 40 price scenarios, acknowledging both historical trends and the percent, an income change of 2.5 percent, and a price elasticity potential for future uncertainties in the energy market. Es- of -0.25) (see annex 4 for methodological details). Figure 49 il- tablishing a 40 percent price increase as a foundational baseline lustrates the shift in the energy poverty rate (P10), represented for a hypothetical scenario in the future is reasonable, given its by the blue line, alongside the pre-price-increase energy poverty alignment with recent trends in the past two years. Over recent rate shown in orange. The overall expected increase amounts to years, energy prices have fluctuated within this range, making 2.3 percentage points, which would represent 174 thousand more it a plausible starting point for projections. However, it is cru- households in energy poverty. This increase is consistent across cial to acknowledge the energy market’s inherent uncertainty, rural and urban areas, displaying relatively minor fluctuations. due to geopolitical events, supply-demand dynamics, and en- Among the most impacted would be households receiving social vironmental policies. While aiming for a realistic baseline, we aid and disability support. Single-elderly households would also must consider the possibility that energy price inflation may not experience a notable impact compared to other demographic moderate and could persist at elevated levels, albeit not reaching groups. The graph further demonstrates that the steepest in- the peaks seen in the past. Energy prices will likely remain vol- creases are observed among those grappling with energy pov- atile due to geopolitical uncertainty and the transition toward a erty before the price hike. These findings raise concerns because low-carbon energy system, resulting in higher consumer costs they suggest a widening gap in inequality and vulnerability across (Mišík 2022; Pahle et al. 2022). Given this uncertainty, we simu- various population segments due to rising energy prices. 25 IMF, Country Harmonized Indexes and Weights: Romania and the IMF. 26 IMF, Country Harmonized Indexes and Weights: Romania and the IMF. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 73 Figure 49 Si mulated I n creas e i n En erg y Pove r ty Ra te s a nd Pre -Pr ice -Inc reas e Ene rg y Pove r ty Ra te (P10) by G roup s for 2 021 HH with social aid from municipality HH with disability recipient Single-elderly HH Female-headed household Household with Roma Household with pensioner HH with child HH with more than 5 members Household with unemployed Rural Urban All 0 10 20 30 40 50 60 Baseline rate (p10), percent Change (p10), percentage points Source: Own estimates based on HBS (2021). Note: The figure plots energy poverty rates in the pre-price-increase scenario and the change in energy poverty rate in our baseline scenario (an energy price increase of 40 percent, a change in household income of -2.5 percent, and a price elasticity of -0.25) for different groups of the population. For a detailed overview of how we construct the energy poverty rates and the microsimulation, see annex 2 and 4. Consistent with energy expenditure patterns, the rise in population segments, vulnerable households bear a heavier energy poverty rates is notably pronounced among house- burden. The escalation in energy poverty rates is more pro- holds with lower income levels. Our analysis delves into nounced among the lower-income quintiles, contributing to the variations in the impact of energy price hikes across the an uptick in inequality within the realm of energy poverty welfare distribution. Consistent with existing data on diverse (figure 50). Figure 50 Si mulated I n creas e i n En erg y Pove r ty Ra te s a nd Pre -Pr ice -Inc reas e Ene rg y Pove r ty Ra te (P10) by I n come Qui n ti les , 2 021 60 54 50 41 40 30 24 20 09 10 04 03 03 03 01 01 00 Q1 Q2 Q3 Q4 Q5 Change (p10), percentage Baseline rate (p10), percentage points Source: Own estimates based on HBS (2021). Note: The figure plots energy poverty rates in the pre-price-increase scenario and the change in energy poverty rate in our baseline scenario (an energy price increase of 40 percent, a change in household income of -2.5 percent, and a price elasticity of -0.25) by income quintiles. Income quintiles are based on per capita household income; Q1 is the poorest and Q5 is the richest. For a detailed overview of how we construct the energy poverty rates and the microsimulation, see annexes 2 and 4. 74 Romania Energy Poverty Assessment Nonetheless, an important caveat warrants consider- where households cannot adjust their energy consumption ation. The estimates are notably sensitive to the assumed (price elasticity is zero), energy poverty rates surge significantly, price elasticity, albeit to a lesser extent to the assumed ranging between 25 and 30 percent, contingent on the assumed change in household income. Before examining the impact parallel change in household income. Conversely, when keep- of energy price hikes on energy poverty rates, we thoroughly ing the price elasticity fixed and altering the assumed change analyze these estimates’ sensitivity to the parameters outlined in household income, the estimates showcase a comparatively in annex 6. Figure 51 clearly illustrates this sensitivity, particu- lesser degree of variation. This implies that our estimates are larly regarding the price elasticity. In an extreme scenario with less susceptible to shifts in assumptions concerning income a price elasticity of -1, households theoretically cease energy changes. The notable sensitivity to price elasticities highlights consumption entirely—an improbable outcome—reducing a critical limitation of this study—namely, the lack of reliable energy poverty to zero. Conversely, in the opposite extreme, estimates on price elasticities for Romania. Figure 51 Sen s i ti v i ty of En erg y Poverty Ra te s ( P1 0 ) to Inco m e a nd Pr ice Elas t ic ity Pa ra m e te rs 35 Energy poverty (p10), percentage 30 25 0 20 0.025 15 0.05 10 0.075 5 0 0.1 -1 -0.75 -0.5 -0.25 0 Change in price elasticity Source: Own estimates based on HBS (2021). Note: The figure plots energy poverty rates (P10 measures) when underlying assumptions (income change and price elasticity parameters) are varying. For a detailed overview of how we construct the energy poverty rates, see annex 2. The colors present different changes in household income. Impact on welfare measures expected to increase for these groups (figure 52, panel a). The size of the losses can inform the adequacy of the income sup- Our analysis extends beyond the assessment of energy ex- port measures to put in place, so that the losses are somewhat penditure and energy poverty rates. We delve into the broad- mitigated. Policy makers addressing energy poverty rates should er welfare implications of energy price increases, considering prioritize these households when designing policy interventions. an evaluation of income and at-risk-of-poverty changes. We These welfare losses are enough for some groups to ex- approximate the changes in welfare using consumer surplus perience substantial poverty increases and can lead to an variation, following the methodology proposed by Freund and overall rise of 2.2 percentage points in AROP rates in the short Wallich (1997) (see annex 4 for details). term. Energy-vulnerable households experience the most sig- In our baseline scenario, a 40 percent increase in energy nificant increase, particularly those in rural areas, which are prices can lead to significant welfare losses, pushing some more impacted than urban areas, those of the single elderly, households into income poverty. Given the larger budget shares and those with pensioners. The AROP rate exhibits a more spent on energy by people in rural areas, the single elderly, pen- substantial increase in rural than urban areas (3.2 versus 1.4 sioners, and social aid and disability recipients, welfare losses are percentage points). Additionally, households already grappling Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 75 with energy poverty, excluding those receiving social aid from subsequent increase in AROP rates. The data underscore that municipalities likely due to their already elevated AROP rates, the increase in AROP rates is driven mainly by households situ- observe the most substantial surge in energy poverty rates. ated at the lower spectrum of the welfare distribution, especially This includes single-elderly households, households with those within the second income quintile. For these households, pensioners, and recipients of disability benefits, all witnessing the AROP rate nearly doubles. Conversely, the upper-income at-risk-of-poverty rate increases exceeding 4 percentage points quintiles experience minimal impact in terms of the risk of pov- (figure 52, panel b). erty. These findings underscore significant equity implications, Examining the impact across income quintiles highlights demonstrating the interconnectedness of monetary and energy that the second income quintile primarily drives the surge in poverty. However, there is only a marginal increase in the Gini AROP rates. Figure 53 shows the pre-price AROP rate and the coefficient, amounting to less than 1 percentage point. Figure 52 Si mulated Wel fare Loss es an d Pove r ty Im pa c t s 0.055 Panel a. Simulated welfare losses, direct effects 16 0.060 0.044 0.043 0.042 14 0.039 0.039 0.050 0.055 0.033 0.036 0.035 12 Energy Shares Welfare Losses 0.040 0.029 16 0.060 0.026 10 0.025 0.044 0.043 0.042 14 0.039 0.039 08 0.030 0.050 0.033 0.036 0.035 12 Energy Shares 06 Welfare Losses 0.040 0.029 0.020 0.026 10 0.025 04 08 02 0.010 0.030 06 00 0.000 0.020 04 All Urban Rural HH with pensioner social aid from municipality Female-headed household with unemployed Household with Roma HH with more than 5 members HH with child HH with disability 02 recipientrecipient 0.010 Single-elderly Household Household 00 0.000 All Urban Rural HH with pensioner aid from municipality Female-headed household with unemployed Household with Roma HH with more than 5 members HH with child HH with disability HH with Single-elderly HH with social Household Household Energy spending share ( in %) PPP Loss Energy spending share ( in %) PPP Loss Panel b. Simulated increases in AROP rates and baseline AROP rate by groups for 2021— direct effect 07 Single-elderly HH 06 07 HH with disability Poverty Increase Household recipient 05 Single-elderly HH with pensioner 06 HH with disability Poverty Increase 04 -headed Female Household recipient 05 household with pensioner 03 Simulated 04 Female -headed All Household with Roma 02 household HH with more 03 than 5 members Simulated HH with social 01 All Household with Roma aid from municipality 02 HH with child HH with more Household with 00 than 5 members unemployed HH with social 01 00 10 20 30 40 50 60 70 80 90 aid from municipality HH with child Household with 00 At-risk of poverty (baseline) unemployed 00 10 20 30 40 50 60 70 80 90 Source: Own estimates based on HBS (2021). At-risk of poverty (baseline) Note: The bottom figure plots the AROP rates in the pre-price scenario for 2021 and the simulated change in the AROP rate in our baseline scenario (an energy price increase of 40 percent and a price elasticity of -0.25) for different population groups. For a detailed overview of the microsimulation, see annex 4. AROP rates are based on equivalized household income. 76 Romania Energy Poverty Assessment Figure 53 I n creas e i n ARO P Rates an d Pre -Pr ice -I nc reas e A ROP Ra te by Inco m e Q u int ile s fo r 2021—Direct Effect 100 86 80 60 40 17 20 10 01 02 00 02 00 00 00 00 Q1 Q2 Q3 Q4 Q5 At-risk of poverty (pre) Difference (ppt.) Source: Own estimates based on HBS (2021). Note: The figure plots the AROP rates in the pre-price scenario for 2021 and the change in the AROP rate in our baseline scenario (an energy price increase of 40 percent and a price elasticity of -0.25) by income quintiles. For a detailed overview of the microsimulation, see annex 4. AROP rates are based on equivalized household income. Income quintiles are based on per capita household income. 4.2 Indirect Impacts and Darbaz (2021) has established a causal relationship between energy and food prices, highlighting the interconnectedness of Indirect effects can also contribute to changes in energy ex- these variables. penditure shares, energy, and income poverty. Increased en- In most research studies, estimating the indirect effects ergy prices can have ripple effects on income, such as during resulting from energy price increases relies on input-output periods of economic downturns or inflation, resulting in re- tables; however, in Romania, the availability of updated in- duced income for individuals. Moreover, businesses may strug- formation is limited. For instance, a recent study conducted by gle to cope with higher costs due to increased energy prices, Guan et al. (2023) employed multiregional input-output tables potentially leading to closures and subsequent job losses, further and expenditure data from 201 expenditure groups and 116 affecting individuals’ income. This reduction in income can sub- countries to estimate the direct and indirect consequences of sequently impact the numerator of energy spending shares. Ad- the energy crisis triggered by Russia’s invasion of Ukraine. Their ditionally, companies may pass on the increased costs resulting findings revealed significant variations in energy cost burdens from higher energy prices to consumers through hikes in the among households, and they projected that the global impact of prices of their products, thereby contributing to inflation. The the crisis could push anywhere from 78 million to 141 million extent to which higher energy prices translate into price increas- people into extreme poverty. While it would be valuable to ex- es for nonenergy goods depends on the energy intensity of each pand our analysis to encompass indirect effects and incorporate specific nonenergy good (Guan et al. 2023). These price increas- input-output analysis, it is not feasible in the Romania context es in nonenergy goods, in turn, can influence households’ ener- due to the unavailability of updated input-output tables, as the gy consumption patterns, as they now face higher prices across latest input-output matrix for Romania is for 2020. various goods, including essential items like food, resulting in Our previous analysis focused solely on the direct effects a reduced budget allocation for energy (Battistini et al. 2022). of energy price increases. Despite evidence from previous stud- The specific response of households to these changes depends ies suggesting their significance, we did not consider the poten- on factors such as price elasticity, income elasticity, and cross- tial indirect effects. Freund and Wallich (1997) acknowledges in price elasticity. For instance, research conducted by Kirikkaleli their research that their formula only captures the direct effects Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 77 and overlooks the indirect effects of energy price increases. Con- could be pretty sizable. The most recent input-output tables for sequently, our analysis, which relies on the formula proposed by Romania date back to 2020.27 To gain a rough estimate of the Freund and Wallich (1997), only accounts for a portion of the potential magnitude, we refer to estimations from another study total effects associated with energy price increases. This simplifi- conducted by Prasad et al.(2023), indicating that the indirect ef- cation represents a significant limitation, because prior research fects are likely to be approximately double the direct effects, espe- indicates that the indirect effects of energy price increases in the cially when considering the $6.85 a day (2017 PPP) poverty line. EU are likely to be substantial. For instance, a study by Saun- It is essential for future research to update these estimations by ders (2023) reveals that the inflation rates for energy-intensive utilizing more-recent input-output tables. nonenergy goods and services surpassed those of less energy-in- tensive nonenergy goods and services. Furthermore, Ari et al. 4.3 Robustness Checks (2022) estimate that indirect effects contribute to approximate- ly 30 percent of the overall impact of rising energy prices on household budgets. However, it is worth noting that there is Using a matched dataset and welfare currently no consensus within the existing literature regarding measures from the EU-SILC the precise measurement of indirect effects and their magnitude is challenging to quantify accurately (Riksbank 2022). Our comparative analysis using HBS and matched EU-SILC Furthermore, existing studies indicate that the influence of data shows the impact of energy price increases on AROP indirect effects varies across different segments of the welfare rates (2.2 versus 1.5 percentage points); differences in AROP distribution, which emphasizes the importance of incorporat- rates are evident between the two approaches, yet the overall ing these effects into the analysis. In a manner consistent with impact of rising prices remains consistent in magnitude. We the observations on energy expenditure shares, expenditure pat- conduct a comparative analysis of our simulations using the terns for nonenergy goods may also differ across income groups HBS dataset alongside simulations carried out on a matched and household types (Guan et al. 2023). For instance, a study con- dataset incorporating EU-SILC and HBS data. Welfare metrics ducted by Battistini et al. (2023) highlights the severe impact of and AROP rates are typically derived from the EU-SILC dataset. energy price shocks on nonenergy companies, particularly those Subsequently, we investigate whether there are variations in re- operating in energy-intensive sectors such as intermediate goods sults when relying on welfare estimates from the EU-SILC. Giv- and transport services. These shocks have repercussions on labor en that the EU-SILC dataset does not encompass expenditure income and employment. Notably, labor income often constitutes data, we perform statistical matching techniques to integrate in- a more significant portion of the total income for households at formation from both household surveys (HBS and EU-SILC). the lower end of the welfare distribution. Additionally, as energy The approach’s specifics are outlined by Rude and Robayo-Abril price increases can lead to higher food prices, the poor are par- (forthcoming b). Crucially, the overall estimates of the total im- ticularly vulnerable, because they allocate a significant propor- pact of price increases on AROP rates exhibit similarity (2.2 tion of their income toward food expenses (Mahler et al. 2022; versus 1.5 percentage points). Disparities in AROP rates across Menyhért 2022). Consistent with this evidence, recent research by different groups emerge between the two approaches, although Guan et al. (2023) demonstrates that the impact of rising energy the impact’s magnitude remains consistent (figure 54). Upon an- prices tends to be regressive. Furthermore, these authors’ findings alyzing results by income quintiles, the primary finding remains indicate that the distributional effects in low- and high-income robust: the effect is most pronounced for the second-lowest in- countries primarily stem from the indirect effects of energy prices. come quintile. However, within the matched dataset, the impact In contrast, in middle-income countries, the effects are predom- displays a slightly broader distribution across income quintiles, inantly driven by direct effects. with the lowest and third-lowest income quintiles experiencing Another study suggests that the indirect effects of energy more-substantial effects (figure 55) compared to the microsim- price increases stemming from Russia’s invasion of Ukraine ulations solely based on HBS data (figures 52–53). 27 http://statistici.insse.ro:8077/tempo-online/#/pages/tables/insse-table. 78 Romania Energy Poverty Assessment Figu re 55 I nc reas e in A ROP Ra te s Figure 54 I n creas e i n ARO P Rates an d Pre -Pr ice - a nd Pre -Pr ice -I nc reas e A ROP Ra te Inc rease ARO P Rate by G roup s for 2 021 — D ire c t by I nco m e Q u int ile s fo r 2 021 — D ire c t Effec t Ef fe c t 80 100 67.2 67.9 90.5 70 60 80 50 42.1 38.9 40 30 23.4 24.4 23.1 60 19.3 20 11.7 10 1.5 1.1 2.1 0.8 1.5 2.3 3.5 0.0 0.0 40 00 24.7 All Urban Rural HH with child HH with unemployed HH with more than 5 members Female-headed HH HH with pensioner Single-elderly HH 20 2.3 3.8 1.6 1.5 0 0 0 0 00 Q1 Q2 Q3 Q4 Q5 At-risk of poverty (pre) Difference (ppt.) At-risk of poverty (pre) Difference (ppt.) Source: Own estimates based on a matched HBS (2019) and EU-SILC (2020) dataset. Note: The figures plot the AROP rates in the pre-price-increase scenario for 2021 and the change in the AROP rate in our baseline scenario (an energy price increase of 40 percent and a price elasticity of -0.25) by groups (left graph) and by income quintiles (right graph). For a detailed overview of the microsimulation, see annex 4. AROP rates are based on equivalized household income. Income quintiles are based on per capita household income. Using an alternative poverty poverty line). In this scenario, the overall impact diminishes (0.6 threshold versus 1.5 percentage points), with a shift in the most affected group. While single-elderly households were primarily affected The overall poverty effects are comparatively modest when using AROP rates, households with unemployed members expe- employing a comparatively lower poverty line (the $6.85 a day rienced the most substantial increase using the $6.85 a day (2017 2017 PPP international poverty line). The outcomes also vary PPP) poverty line (figure 56). Rural areas see a more significant across demographic groups, highlighting the heightened impact impact in both cases than urban areas. The equity implications on households with unemployed members. Shifting our focus, persist when employing the $6.85 a day (2017 PPP) poverty line, we compare our findings on AROP rates with those under a sig- but the disparities are more pronounced. The rise in poverty nificantly lower threshold ($6.85 (a day 2017 PPP international rates is driven entirely by the lowest income quintile (figure 57). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 79 Figure 56 I n creas es i n I n tern ati on al Pove r ty Figu re 57 Inc reas e s in Pre -Pr ice -I nc reas e Rates ($6.8 5 a Day ) an d Pre- Pri ce- I n creas e I nte r na t io nal Pove r ty Ra te s ( $6 . 8 5 a Day) by Inter nat ion al Poverty Rate by G roup s fo r I nco m e Q u int ile s fo r 2 021 — D ire c t Ef fe c t 2021—Direct Effect 80 60 56.2 67.2 70 60 50 50 40 34.2 40 30 20.1 16.9 20 11.3 30 7.8 10 4.6 3.7 2.0 2.4 0.6 0.3 0.9 0 0.4 0.1 0.4 0 0 20 All Urban Rural HH with child HH with pensioner Single-elderly HH Female-headed HH HH with unemployed HH with more than 5 members 10 2.9 0 0 0 0 0 0 0 0 0 Q1 Q2 Q3 Q4 Q5 Int. poverty rate (6.85) - pre Difference (ppt.) Int. poverty rate (6.85) - pre Difference (ppt.) Source: Own estimates based on a matched HBS (2019) and EU-SILC (2020) dataset. Note: The figures plot the AROP rates in the pre-price scenario for 2021 and the change in international poverty rates at the $6.85 (2017 PPP) line in our baseline scenario (an energy price increase of 40 percent and a price elasticity of -0.25) by groups (left graph) and by income quintiles (right graph). For a detailed overview of the microsimulation, see annexes 4 and 6. International poverty rates are based on per capita household income. Income quintiles are based on per capita household income. Varying price elasticities across the be more significant for those at the lower end of the welfare income distribution distribution. This is because high-income households tend to exhibit more-significant responses to changes in energy prices Existing research indicates that the price elasticity of ener- than low-income households. Consequently, low-income house- gy varies across different segments of the population and holds face a more significant burden and are more adversely household categories, which points up the importance of affected by energy price increases, which has significant implica- disaggregating behavioral response patterns. For instance, tions for energy affordability across the welfare distribution. This Schulte and Heindl (2017) provide evidence that the price elas- regressive impact underscores the equity concerns associated ticities of energy consumption significantly differ among income with rising energy prices, highlighting the greater need for sup- quintiles. Their findings reveal that low-income households are port and assistance among lower-income households compared less responsive to changes in energy prices than high-income to their higher-income counterparts. households. Schulte and Heindl’s study indicates that house- In line with existing scholarly work, we investigate the ro- holds in the top 25 percent of income levels are three times more bustness of our primary findings by accounting for varying price-elastic than those in the lowest 2–5 percent. Moreover, price elasticities across the welfare distribution. Specifically, their research highlights variations in price elasticities of energy we utilize the estimates of energy price elasticities provided by consumption across different types of households. Schulte and Heindl (2017) as a benchmark. We assume an en- The observed patterns of energy price elasticities have no- ergy price elasticity of -0.2 for the lowest income quintile and table implications for welfare, mainly due to the regressive decrease the elasticity by 0.1 for each subsequent quintile. Con- nature of energy price increases. When energy prices rise, the versely, we assign an energy price elasticity of -0.6 to the high- welfare losses experienced by individuals and households will est income quintile. These adjusted elasticities are used in our 80 Romania Energy Poverty Assessment analysis while maintaining the baseline scenario of a 40 percent long-term implications for household wealth. However, we do increase in energy prices. not account for these impacts in our study due to our dataset’s Our findings closely align with the results obtained from lack of savings data. our baseline scenario for both the AROP and international Existing research highlights the variation of income elas- poverty rates. The impact on the Gini coefficient is slightly more ticities across the welfare distribution, similar to the variation pronounced but remains close to zero. Specifically, the overall observed in price elasticities. Studies by Wadud et al. (2009) effect on the AROP rate resulting from a 40 percent increase in and Schulte and Heindl (2017) provide estimates of income energy prices is very similar at 2.05 percentage points (com- elasticities of energy consumption ranging from 0.2 to 0.6. In pared to 2.08) when using the matched EU-SILC-HBS dataset. line with their findings on price elasticities, Schulte and Heindl Moreover, the impact on each income quintile closely corre- (2017) demonstrate substantial differences in income elasticities sponds to previous observations, which can be attributed to the across income groups. Specifically, they report income elastici- shape of the welfare distribution, with the upper two income ties of around 0.2 to 0.3 for the lowest-income group, while the quintiles remaining unaffected by the AROP threshold. upper-income group exhibits elasticities of around 0.5. Howev- er, this level of granularity is not accounted for in the current 4.4 Limitations and Caveats report. It would be valuable for future research to investigate how the results might vary by incorporating different income Previous research indicates that the effects of increasing en- elasticities across the welfare distribution. ergy prices extend beyond the direct impact on household Some additional limitations of and considerations con- income. Households may respond to price shocks by adjusting cerning our methodology need to be acknowledged. Firstly, their spending patterns, such as reducing expenditures or de- it is essential to note that we do not have reliable estimates of ferring payments. A study conducted by Battistini et al. (2022) energy price elasticities for Romania, which is a significant lim- reveals that the distributional consequences of rising energy itation. Secondly, we encountered challenges in obtaining up- prices on households are more extensive than those discussed to-date input-output tables specific to Romania. Consequently, in our current report. The study highlights a substantial reduc- we employ a more constrained approach to estimate indirect tion in spending, particularly among lower-income households, effects. Our analysis focuses exclusively on households and we consistent with liquidity constraints of low-income households, do not account for energy consumption by other entities, such which forces them to cut back on spending when prices rise. as public or private institutions. This uneven effect on savings could potentially have adverse Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 81 Chapter 5 Effectively Tackling Energy Poverty and Mitigating the Adverse Welfare Impacts of Rising Energy Prices In this chapter, a comprehensive examination is conducted chapters. The analysis primarily focuses on two key policy cat- to assess the existing legal and policy framework in Romania egories: income support initiatives aimed at immediate relief for addressing the multifaceted issue of energy poverty. This for vulnerable populations’ energy needs and energy efficiency chapter first evaluates the current legal and policy framework in initiatives geared toward long-term sustainability and prudent Romania addressing energy poverty. Second, it reviews current energy usage. Finally, we explore the importance of potential government measures specifically designed to shield households behavioral interventions, aiming to tackle the behavioral bar- from the impact of rising energy prices and examines their po- riers uncovered in chapter 3. This approach offers a distinctive tential effectiveness. Third, it identifies and simulates policy perspective on combating energy poverty through sustainable packages that can potentially play a pivotal role in the ongoing energy transitions by considering key elements of human behav- efforts to combat and alleviate energy poverty, leveraging the ior. The present chapter aims to inform and guide future poli- diagnostic evidence and insights gleaned from the preceding cy-making initiatives for a more effective and targeted approach to addressing energy poverty by providing a nuanced under- process. However, the government returned to regulated prices standing of the legal landscape, current government measures, in April 2022 and established price caps to respond to the recent and their outcomes. sharp rise in energy prices, leading again to the introduction of implicit energy subsidies. 5.1 Legal and Policy Framework The LTRS33 recognizes energy poverty in single-family homes and apartment units. This strategy explores energy pov- In Romania, “energy poverty” has often been aligned with the erty in buildings and proposes solutions that focus on energy concept of a “vulnerable consumer,” as stipulated in various efficiency and heating assistance interventions. The LTRS em- normative acts. The enactment of Law 226/2021 in September 28 phasizes the necessity for an improved national legal framework 2021 introduced a novel definition of energy poverty: the inabil- to address energy poverty and stresses the importance of robust ity of a vulnerable energy consumer to cover their minimum 29 government programs, accessing EU funds, and exploring other energy needs.30 Despite this being an official definition, how it financial schemes on the private market. Additionally, the LTRS is operationalized for monitoring and evaluation purposes and underscores the crucial role of local governments in securing how the measurement of minimum energy needs is assessed is grants and integrating energy poverty considerations into their not defined in the legislation. 31 local renovation programs. Tackling energy poverty has been a strategic priority for Furthermore, EU Member States are required to assess the Romanian government, as outlined in various govern- and mitigate energy poverty within their National Energy ment strategies, and it aligns with the priorities set by the EC. and Climate Plans (NECPs) as part of the European Green The Romanian National Strategy on Social Inclusion and Deal. The 2021–2030 Integrated National Energy and Climate Poverty Reduction for 2022–2027,32 along with its associated Plan offers details about the share of households experiencing Action Plan, as of April 2022, aims to address energy poverty energy poverty using consensual measures (arrears on util- through two main approaches. First, it seeks to establish public ity bills and inability to keep home warm) and establishes a programs targeting thermal insulation in communities affected nationwide goal to alleviate energy poverty and protect vul- by energy poverty (aiming to address energy poverty through nerable customers. However, it does not specify clear metrics increased energy efficiency) that offer subsidies for rehabilita- and time frames for monitoring and evaluating the impacts tion projects. Second, the strategy involves providing monthly of policies on energy poverty. When a significant portion of assistance to cover part of the expenses associated with home households faces energy poverty, Member States should in- heating during the cold season. Prior to the initiation of the clude distinct national goals and policies in their plans to ad- strategy, a social tariff for electricity was in effect (from 2006 dress and mitigate it. Tackling energy poverty is one of the key to 2018); it was removed as part of the market liberalization focus areas of the Renovation Wave,34 an initiative to boost 28 https://legislatie.just.ro/Public/DetaliiDocumentAfis/246430. 29 A “vulnerable energy consumer” refers to an individual or family requiring social protection measures and additional services due to health, age, insufficient income, or isolation from energy sources to meet at least their minimum energy needs. 30 “Minimum energy needs” encompass the essential energy consumption for lighting, optimal home cooling and heating, cooking, hot water preparation, using plugged-in communication devices, and powering medical devices vital for life support or improving health. The minimum consumption limit is established by order of the Minister of Labor and Social Protection, based on the data made available by the National Energy Regulatory Authority, as well as by the National Institute of Statistics. 31 There is no universally accepted basket of basic energy services. To clearly distinguish the energy poor from those who are not using this definition, one must define a basket of basic energy services and the minimum amount of each service needed (an energy poverty line). Unlike the food poverty line, which can be defined in terms of a minimum daily caloric intake, there is no absolute reference for what fulfills a minimum level of basic energy needs. For a measurable metric of energy poverty, this energy poverty line needs to be estimated. This involves several steps to determine the minimum income needed for an individual or household to meet basic energy needs. 32 https://www.mmuncii.ro/j33/index.php/ro/minister-2019/strategii-politici-programe/6562-sn-incluziune-sociala-2022-2027. 33 https://energy.ec.europa.eu/system/files/2021-04/ro_2020_ltrs_en_version_0.pdf. 34 See more details here: https://energy.ec.europa.eu/topics/energy-efficiency/energy-efficient-buildings/renovation-wave_en. 84 Romania Energy Poverty Assessment energy efficiency of public and private buildings, as outlined vation quality specifications. Due to homeowners’ low prioriti- in the European Green Deal. zation of energy efficiency and reluctance to pay for advice, most The EU legal and institutional framework recognizes en- one-stop shops require public sector support. In Romania, these ergy poverty as a critical and growing problem among EU initiatives haven been integrated into the recovery and resilience Member States. The Recast Electricity Directive and Directive plan, particularly under the REPowerEU chapter, which focuses 2009/73/EC mandates that Member States create national ac- 35 on accelerating the deployment of renewable energy and energy tion plans, employ other suitable frameworks to address energy efficiency renovations. poverty, take appropriate measures to combat energy poverty, Finally, the EC established the SCF in 2021, the first EU protect vulnerable customers, and quantify the number of af- Fund designed specifically to offer financial assistance to vul- fected households. The Energy Efficiency Directive, which sets 36 nerable households, transport users, and microenterprises rules and obligations for achieving the EU’s ambitious energy during the energy transition. The SCF regulation, which was efficiency targets, also requires the addressing of energy poverty approved in May 2023 and became effective in June 2023, is within energy efficiency obligations with a focus on protecting expected to mitigate the social impacts of implementing the vulnerable households. The revised Energy Performance of new emission trading system for building and road transport Buildings Directive, aiming to reach the building and reno- 37 by explicitly targeting vulnerable households, microenterprises, vation goals set out in the European Green Deal, compels the and transport users in the EU who are affected by energy and inclusion of national measures to alleviate energy poverty within transport poverty during the 2026–32 period. Member States long-term renovation strategies. National strategies among EU must submit their Social Climate plans by June 2025. The pro- Member States for long-term renovation and related initiatives jected SCF allocation for Romania is estimated to be around toward achieving 2030 and 2050 energy efficiency goals should 9.25 percent of the total EU funds, amounting to approximately prioritize the protection of energy-poor households. 6 billion euros. This makes Romania the sixth-largest recipient Furthermore, the revised Energy Efficiency Directive and of SCF funds. These funds can finance temporary income-sup- the Energy Performance of Buildings Directive encourage the port measures and investments in energy efficiency and the ren- development of one-stop shops for home energy renovation ovation of buildings, clean heating, and cooling. Up to 37.5% of in all EU Member States. These one-stop shops, highlighted this allocation can be used for temporary direct income support in the directives (with the Energy Efficiency Directive adopted to assist vulnerable households. in September 2023 and the Energy Performance of Buildings Directive pending adoption), are crucial for implementing the 5.2 Recent Policies and Interventions ‘energy efficiency first principle’ in the fragmented residential sector. EU Member States are expected to support the establish- In response to the recent energy crisis, the Romanian gov- ment of these one-stop shops and stimulate both demand and ernment has primarily focused on a combination of energy supply through favorable legislation, regulatory frameworks, price caps and income-support measures to protect house- and other support mechanisms. The concept of one-stop shops holds; some energy efficiency programs are also in place. Di- aims to provide consumers with a single point of contact for rect financial support measures included untargeted price caps information and advice on energy retrofits, making the process and targeted income-support measures, which include a heat- simple and straightforward. This includes pre-renovation audits, ing subsidy for the cold season, a supplementary heating sub- renovation design, tailored financial plans, process coordina- sidy, and energy cards. The existing heating subsidy, reformed tion, access to affordable financing, and energy consumption in 2021, continues to provide financial support to compensate monitoring. However, existing one-stop shops lack standardized for the heating costs of vulnerable customers. Energy efficien- services, energy savings thresholds, technical criteria, and reno- cy measures include a renovation program (already approved), 35 https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:211:0094:0136:en:PDF. 36 https://energy.ec.europa.eu/topics/energy-efficiency/energy-efficiency-targets-directive-and-rules/energy-efficiency-directive_en. 37 https://energy.ec.europa.eu/topics/energy-efficiency/energy-efficient-buildings/energy-performance-buildings-directive_en. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 85 with the ambitious goals of enhancing the energy efficiency of Despite the advantages of the price caps discussed above, buildings and investing in energy-efficient housing, especially 38 discussions persist regarding the fairness and effectiveness of among poorer households, which could have beneficial welfare this measure (Hardy et al. 2019; Philibert et al. 2009; Guo et effects. In addition, Romania’s NRRPs have allocations for green al. 2019), primarily due to their sizable fiscal impact and lack projects, including energy efficiency. 39 of targeting. Moreover, international evidence shows lower fos- sil fuel prices can incentivize fossil fuel consumption. Suranovic Untargeted measures—energy (2013) argues that fossil fuel addiction, similar to cigarette ad- price caps diction, may result in a prolonged period of expressing a de- sire to switch to clean energy without significant action, as the As in numerous other EU Member States, in Romania efforts opportunity cost of switching to cleaner alternatives also rises. have been made to significantly cap energy prices to shield Taghvaee et al. (2022) show that price policies are ineffective in consumers and businesses from steep price hikes. In 2021, reducing fossil fuel consumption, while energy efficiency im- Romania completed the process of fully liberalizing its energy provements are much more effective. market. However, the government reintroduced regulated pric- Evidence presented in this report shows that the welfare es to curb any additional increases in energy bills at the end impacts of rising prices in Romania have been sizable despite of February 2022. Romania established a one-year ceiling on the energy price caps; moreover, the associated fiscal cost is electricity and natural gas prices for households and firms to also high. As shown in chapter 4, the risk of poverty increases is be rolled out from August to December 2022. Customers for expected to be sizable despite the price caps, which is explained gas and electricity were provided a reduced price that fell below by the untargeted nature of this policy measure. Moreover, they the market rate set by companies and a governance ordinance are fiscally costly, given the untargeted nature of the policy and was put in place to address the price differential between the leakage to high-income groups. Estimating the fiscal cost of un- market price and the customer price. Utilities purchased at the targeted price caps in Romania is challenging, but several stud- prevailing market price and sold at a restricted rate, with the ies provide some estimates. Bruegel estimates show the fiscal government providing compensation to cover the difference.40 costs of the initial price cap and subsequent extension amounted The electricity and natural gas prices ceiling was extended until to approximately 2.9 and 1.6 billion euros, respectively, for a August 2023. However, price caps for vulnerable customers have total of 4.5 billion euros, representing a large share of the policy been extended until 2025. These retail price caps were a valu- 41 response (Sgaravatti et al. 2021). Estimates from the IMF indi- able tool in protecting households and enterprises from spikes in cate gas and electricity explicit consumption consumer subsidies energy costs, especially those who allocate a significant portion of 1.1 US$ billion. Implicit consumer subsidies for gas and elec- of their budget to energy expenses. The appeal of price caps for tricity are estimated at 0.6 US$ billion.42 Alternative estimates governments lies in their ease of implementation, providing a from Autoritatea Nationala de Reglementare in Domeniul En- clear and easily monitored framework for energy pricing. ergie (ANRE) indicate the annual fiscal cost of price caps to be 38 The program aims to promote the use of heat pumps as a means of significantly reducing energy demands for heating and transitioning to electrification. 39 For more details, see https://commission.europa.eu/business-economy-euro/economic-recovery/recovery-and-resilience-facility/country- pages/romanias-recovery-and-resilience-plan_en. 40 The government introduced a social tariff of 0.68 leu/kilowatt-hour (kWh) (approximately 0.14 euros/kWh) with VAT included for households with monthly consumption of up to 100 kWh, covering approximately 4 million households. Another 4 million households with monthly electricity consumption between 100 and 300 kWh benefited from a tariff of 0.80 leu/kWh (0.16 euros/kWh), including VAT. 41 The cap of 0.14 euros/kWh for vulnerable customers, including those with a max monthly consumption of 100 kWh, households using medical equipment, families with at least three children, and single-parent households, will continue until March 2025. Additionally, the cap of 0.16 euros/ kWh will continue until March 2025 for domestic customers with monthly consumption between 100 and 255 kWh. 42 Explicit subsidies reflect supply costs’ being greater than the retail prices, whereas implicit subsidies reflect the efficient price’s being greater than the retail price, exclusive of any explicit subsidy, with the efficient price being the monetary supply cost plus all externalities. Costs of externalities include estimates of associated air pollution, greenhouse gas emissions, and road congestion. 86 Romania Energy Poverty Assessment close to 6.6 billion lei (1.4 billion euros).43 Estimates from the programs to assist to individuals with disabilities; and (4) ad- Ministry of Labor suggest that, in 2022, implicit energy sub- ditional programs, such as social pensions, aimed at providing sidies in Romania amounted to 1.177 billion RON for natural comprehensive support to diverse segments of the population. gas and 2.73 billion RON for electricity44. In February 2024, the This comprehensive approach underscores Romania’s commit- Romanian government announced that the cap on energy prices ment to addressing the varied needs of its citizens through a would remain unchanged until March 2025. social benefits framework. In 2021, Romania’s social protection spending, at 13.3 Targeted income-support measures percent of GDP, was below the EU-27 average and some CEE countries, with “old age” pensions and family allowances dom- Understanding the key features of a country’s social protec- inating the allocation. Social protection spending as a share of tion system is essential before delving into the discussion of GDP was significantly below the EU-27 average and some other income-support measures to address energy poverty. The CEE countries from 2012 to 2021 (figure 58, panel a). In Roma- social protection system, the foundational framework within nia in 2021, social protection expenditure totaled 157.8 billion which these measures operate, influences their design, effective- lei, representing 13.3 percent of the GDP and comprising 33.4 ness, and overall impact. By describing the key features of the percent of the overall expenditure. The most substantial category social protection system, we gain insight into the mechanisms within this framework, “old age” (9.7 percent of GDP), primar- and structures that underpin the delivery of support to vulner- ily encompassed pension disbursements. Following closely was able populations. This contextual understanding is crucial for “family and children” (1.5 percent of GDP), the second-largest evaluating the appropriateness and success of income-support category. Expenditure for “sickness and disability” (1.2 percent measures to ensure they align with the broader social welfare of GDP), the third-largest group, primarily involved social framework and effectively address the nuanced challenges posed payments in cash or kind linked to social insurance schemes. by energy poverty. Expenditure attributed to “social exclusion not elsewhere clas- sified,” which constituted 0.5 percent of GDP in 2021, encom- Key features of the social protection passed benefits for socially excluded individuals, such as those system with low income, refugees, or facing substance abuse. Other Romania’s social benefits system is characterized by a com- programs, including “survivors” (0.1 percent of GDP) and “un- bination of categorical and means-tested initiatives designed employment” (0.1 percent of GDP), notably feature pension to support poor and vulnerable families. These programs can payments to survivors and contribute significantly to overall be broadly categorized into four key areas: (1) family support expenditure. The “housing” category, which involves social programs, which focus on promoting the well-being of families; protection payments to households for housing costs and the (2) means-tested programs tailored to low-income households, operation of social housing schemes, accounts for a negligible ensuring targeted assistance to those in need; (3) specialized share of GDP (figure 58, panel b). 43 Source: https://anre.ro/suma-totala-verificata-de-anre-si-transmisa-spre-decontare-a-depasit-205-miliarde-ron/ 44 Source: Ministry of Labour and Social Solidarity, https://mmuncii.ro/j33/index.php/ro/transparenta/statistici/buletin-statistic Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 87 Figure 58 Soci al Protecti on Sp en di n g , Ro m a nia vs . E U -27 a nd S e le c te d C E E Co u nt r ie s 25.00 Panel a. Evolution of social protection expenditures (as % of GDP), Romania vs. comparator countries, 2012–21 20.00 25.00 15.00 20.00 10.00 15.00 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 EU-27 Bulgaria Croatia Poland Romania 10.00 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 EU-27 Bulgaria Croatia Poland Romania Panel b. Social protection expenditures (as % of GDP) by functional categories, Romania vs. EU countries, 2021 30 20 30 10 20 0 10 France Finland Italy Austria Denmark Denmark Belgium Belgium Germany Germany Greece Spain EU 27 Norway Sweden Sweden Luxembourg Portugal Portugal Slovenia Slovenia Poland Netherlands Slovakia Slovakia Switzerland Lithuania Lithuania Croatia Latvia Czechia Estonia Bulgaria Bulgaria Romania Romania Iceland Hungary Hungary Cyprus Malta Ireland 0 France Finland Italy Austria Greece Spain EU 27 Norway Luxembourg Poland Netherlands Switzerland Croatia Latvia Czechia Estonia Iceland Cyprus Malta Ireland Sickness and disability Old age Survivors Family and children Unemployment Housing Social exclusion n.e.c. R&D Social protection Social protection n.e.c. Sickness and disability Old age Survivors Family and Source: Eurostat (2023). Indicator: children TESSI120 Unemployment Housing Social exclusion Note: Social protection spending n.e.c. in Romania R&D was primarily driven Social by protection “old age” Social pensions and family and protection children n.e.c. programs. Romania’s social benefits system, especially its means-test- benefits hinders their effectiveness in reducing poverty and en- ed programs, faces substantial challenges in protecting the hancing program administration efficiency.46 A recent analysis poor and vulnerable; moreover, there are inequalities in fam- of how to improve the social protection system in Romania from ily policies and pensions and as a result social transfers have 2020 revealed several shortfalls, such as a bias toward middle-in- a limited role in reducing poverty. The existing means-tested come families with respect to family policies—at least prior to programs, such as the Guaranteed Minimum Income, Family the pandemic (Adăscăliţei et al. 2020). Support Benefit, and heating benefit, provide limited coverage In line with this assessment, Romania has the EU’s and offer relatively low generosity. Furthermore, despite the in- third-highest AROP rate for children. In 2021, 3 out of 10 troduction of the Social Reference Indicator in 2008, the value children below 18 years old were affected. Moreover, the pen- of social benefits has not been consistently adjusted to infla- sion system is marked by significant inequalities. In 2020, the tion (World Bank 2023a). Fragmentation within means-tested 45 highest income quintile of pensions was more than four times 45 Various benefits have been ad-hoc indexed based on separate laws, resulting in discrepancies. For example, while the value of universal child allowances tripled in the last decade, the guaranteed minimum income barely changed. The social reference indicator (ISR) remains fixed at 500 lei/ month, irrespective of inflation, which reached around 42 percent between 2008 and 2021, while benefit values linked to the index remained stagnant. 46 These benefits differ regarding assistance units, equivalence scales, documentation requirements, income exemptions, asset filters, recertification periods, coresponsibilities, linkage to social services, and incentives for transitioning from benefits to employment or income-generating activities. Each program maintains separate beneficiary records and payment systems, leading to inefficiency and administrative redundancy. 88 Romania Energy Poverty Assessment higher than the lowest income quintile and there is dualization Consequently, the overall impact of social transfers on pover- in the pension system (Adăscăliţei et al. 2020). In September ty reduction in Romania ranks the lowest among EU Member 2020, the guaranteed minimum pension was below the poverty Sates (figure 59). An updated comprehensive fiscal incidence line for a single person (Adăscăliţei et al. 2020), which is one analysis provides a more detailed assessment of how the overall of the reasons why Romania reports one of the highest AROP tax-transfer system can reduce poverty and inequality and which rates for older adults in the EU. In addition, social protection for households bear the burden or receive the greatest benefits along persons with disabilities is fragmented (Adăscăliţei et al. 2020). the income distribution (Robayo-Abril et al forthcoming). Figure 59 The Role of Social Transfers in Poverty Reduction, Romania vs. EU Member States, 2021–22 (%) 70 60 50 40 30 20 10 0 Romania Greece Croatia Portugal Bulgaria Latvia Italy Malta Spain Estonia Lithuania Cyprus Luxembourg Netherlands Slovakia Euro area - 20 countries (from 2023) Hungary Slovenia Poland Sweden Czechia Austria France Germany Belgium Finland Denmark Norway Ireland 2021 2022 Source: Eurostat (2023); indicator: TESPM050. Note: The figure presents reductions in the AROP rate due to social transfers (calculated comparing AROP rates before social transfers with those after transfers; pensions are not considered as social transfers in these calculations). The indicator is based on the EU-SILC. Data reflects the 2021 and 2022 survey years. Heating subsidies ized system, the benefit amount is calculated as a proportion of the heating energy bill. The proportions decrease with income For the protection of energy-vulnerable customers, the home brackets and increase for single persons. The lump sum benefit heating aid subsidy is a means-tested cash support targeted to low-income families during the cold season (November to amount for heat provided by the burning of natural gas, wood, March). The benefit entitlement is subject to income and asset coal, and oil fuel is a lump sum that also decreases with income testing and the benefit amount is differentiated according to (table 2, panel a). The benefit is given monthly during the cold the type of heating source. For heating provided by a central- season, between November 1st and March 31st next year. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 89 Table 2 Ho me Heati n g Ai d Parameters Panel a. Income thresholds and benefit amounts, 2018–21 Percentage compensation in centralized systems Per capita monthly net (lei) Reference value (% reference value) Family Single person Natural gas (lei) Wood, coal, and oil fuel (lei) Up to 155 90 100 262 54 155.1–210 80 90 190 48 210.1–260  70 80 150 44 260.1–310 60 70 120 39 310.1–355 50 60 90 34 355.1–425 40 50 70 30 425.1–480 30 40 45 26 480.1–540 20 30 35 20 540.1–615 10 20 20 16 615.1–786 5 15 - - 786.1–1082 0 10 - - Source: Ministry of Labour and Social Protection of Romania, Law no. 92/2012. Panel b. Income thresholds and benefit amounts, 2022–23 Percentage compensation in centralized Per capita monthly net (lei) Reference values systems (% reference value) Family Single person Natural gas (lei) Wood, coal and oil fuel (lei) Electricity (lei) Up to 200 100 100 250 320 500 200.1–320 90 90 225 288 450 320.1–440 80 80 200 256 400 440.1–560 70 70 175 224 350 560.1–680 60 60 150 192 300 680.1–920 50 50 125 160 250 920.1–1040 40 40 100 128 200 1040.1–1160 30 30 75 96 150 1160.1–1280 20 20 50 64 100 1280.1–1386 10 n/a 25 32 50 1280.1–2053 n/a 10 25 32 50 Source: Ministry of Labour and Social Protection of Romania, Law no. 226 of September 16, 2021. Current fiscal incidence analysis shows that the heating pacts were small due to the limited benefit adequacy. Fiscal aid is quite progressive, but the poverty and inequality im- incidence analysis using 2021 data (Robayo-Abril et al. forth- 90 Romania Energy Poverty Assessment coming)47 shows that the home heating aid is quite progressive rion and was aimed to provide support for heating but for no 2.00 compared to other social transfers, as shown by larger Kak- aallowance State for children other energy expenses. Our analysis in chapter 2 suggests that 1.80 wani index, but the poverty and inequality impacts are small, while heating costs are an important component of household 1.60 to its small size (figure 60). Additionally, this primarily due Size (%Market Income Plus Pensions) energy expenditures, the other components represent a signifi- scheme did not have any energy consumption threshold crite- 1.40 cantly larger share. 1.20 Figure 60 Soci al Tran s fer Prog rams : Si ze, Pro g re ss iv ity, a nd Co nt r ibu t io ns to Pove r ty Re d u c t io n 1.00 Panel a. Size (% market income plus pensions) and progressivity (Kakwani index) 0.80 Child related benefits (raising+incentive) 2.00 0.60 State allowance for children Support allowance for families with children benefitsfor Otherbenefits Other children forchildren 1.80 Other social benefits non-cash 0.40 Unemployment benefit Disability benefits Size (%Market Income Plus Pensions) 1.60 0.20 School related benefits non-cash Minimum guaranteed income Temporary work incapacity 1.40 and maternity benefits 0.00 Scholarships Other social benefits Minimum pension Heating aid 1.20 0.1 0.3 0.5 0.7 0.9 1.1 1.3 Kakwani Index 1.00 Regressive Progressive 0.80 Child related benefits (raising+incentive) 0.60 Support allowance for families with children Other social benefits non-cash benefitsfor Otherbenefits Other children forchildren 0.40 Unemployment benefit Disability benefits 0.20 School related benefits non-cash Minimum guaranteed income Temporary work incapacity and maternity benefits 0.00 Scholarships Other social benefits Minimum pension Heating aid 0.1 0.3 0.5 0.7 0.9 1.1 1.3 Kakwani Index Regressive Progressive Panel b. Marginal contributions to poverty reduction Heating aid Other social benefits Other benefits for children Minimum pension Minimum guaranteed income Scholarships School related benefits non-cash Temporary work incapacity and maternity benefits Support allowance for families with children Other social benefits non-cash Unemployment benefit Child related benefits (raising+incentive) Disability benefits for children State allowance Heating aid Other social benefits 0.0000 0.0050 0.0100 0.0150 0.0200 0.0250 0.0300 0.0350 Other benefits for children Minimum pension et al. forthcoming. Source: Robayo-Abril Minimum guaranteedLatest Commitment to Equity Approach (CEQ) based on 2021 HBS. This figure does not capture income the heating subsidy reform introduced in 2021. Scholarships School related benefits non-cash Temporary work incapacity and maternity benefits Support allowance for families with children Other social benefits non-cash Unemployment benefit 47 A comprehensive distributive analysis of the fiscal system that allows for the observation of the changes in poverty and inequality of each fiscal Child related benefits (raising+incentive) intervention would require the implementation of approaches such as the Commitment to Equity Approach (CEQ) (Lustig 2018). The forthcoming Disability benefits CEQ for Romania is based on 2021 State data before allowance the surge in energy prices. for children 0.0000 0.0050 0.0100 0.0150 0.0200 0.0250 0.0300 0.0350 Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 91 In 2021, changes were implemented in the heating sub- and even pellets. The vouchers carried a monetary value of 1,400 sidy system to improve its progressivity, increase its impact lei benefit (approximately 280 euros), were distributed in two on poverty, and introduce a heating supplement to assist installments, and were usable through December 2023. The el- during the offseason. These changes, implemented under Law igibility criteria encompassed individuals with pensions below 267 (Vulnerable Consumer Law), encompass various measures 2,000 lei per month, those with disabilities and incomes below related to the heating subsidy and a new energy supplement. No- 2,000 lei per month, and families benefiting from the GMI and tably, the amendments involve raising the income cap for benefit family support allowances. The application process for eligibility eligibility of the heating aid, thereby increasing the number of involves multistage verification, including checks by the Nation- beneficiaries (table 2, panel b). Furthermore, there is an elabo- al Pension House and the National Agency for Pensions and ration on the definition of a vulnerable customer. Addition- 48 Social Inspections to identify eligible pensioners and individuals ally, an income-tested supplementary heating subsidy was also with disabilities based on specific income thresholds. Further introduced to provide help for the whole year, including during scrutiny is performed using the Agenția Națională de Admin- the cold season. This is an important step, considering that this istrare Fiscală platform to confirm the total household income report shows cooling has emerged as an important challenge. remains below 2,000 lei and to ascertain whether multiple in- Starting from the 2021–22 cold season, the energy supplement dividuals reside in the same dwelling. If the per capita income serves as an additional benefit for the same recipients. This is under 2,000 lei per month, households qualify to receive the monthly top-up is provided throughout the entire year, not just energy card.49 The continuity of this program beyond 2023 re- during the cold season. The amount of the supplement varies mains uncertain. based on the type of energy utilized for heating: it is 10 lei for While there is no available analysis or data to assess the thermal energy and natural gas, 20 lei for fossil fuels, and 30 lei effectiveness of the energy cards, these vouchers are directed for electricity. However, if electricity is the sole source of energy toward groups identified as particularly susceptible to energy used in the household, the supplement increases to 70 lei. The poverty. This includes pensioners, households receiving GMI, upcoming CEQ analysis is expected to shed light on the effec- and households with members with disabilities. The aim is to tiveness of this reform. support these specific groups, acknowledging their heightened susceptibility to energy-related challenges. Energy cards or vouchers In addition to existing heating subsidies and energy price Energy efficiency measures caps, in 2023 the government introduced a temporary “ener- gy cards” program targeting pensioners, those with disabil- Several energy efficiency measures have been adopted in ities, and beneficiaries of the Guaranteed Minimum Income Romania to tackle energy poverty. The National Multiannual (GMI) program and family allowances. The primary objec- Program for the Improvement of Energy Performance in Blocks tive of the energy cards was to assist low-income and vulnerable of Flats is an ongoing rehabilitation program established in 2009 families facing excessive heating expenditures. The funds could targeting multilevel blocks of flats constructed before 2005, in- be utilized for a range of energy-related services and products, cluding public social buildings, focusing on the building enve- encompassing electricity, centralized thermal energy, gas, wood, lope (exterior walls, foundations, roof, windows, and doors) and 48 A vulnerable energy consumer is a single person/family who, for reasons of health, age, insufficient income, or isolation from energy sources, requires social protection measures and additional services to ensure that at least the minimum energy needs are met. “Energy needs,” however, is not defined in the legislation. 49 For more details, see Emergency ordinance no. 166 of December 8, 2022 https://legislatie.just.ro/Public/DetaliiDocument/262296. 92 Romania Energy Poverty Assessment heating system. Prior to 2019, the grant rate was 80 percent from mal rehabilitation through this initiative, leading to a cumulative the state and local budgets, with flat owner associations covering energy savings of approximately 149 GWh. 20 percent. Since 2019, the state has been covering 60 percent, with the balance coming from flat owner associations or local Do current social protection measures budgets. Priority is given to flat owner associations and those cover the groups who are at a higher exempted from payment include individuals with disabilities, risk of experiencing energy poverty? low-income single persons or retirees, and war veterans with surviving spouses. In 2018, recorded energy savings of imple- As highlighted in chapter 2, energy poverty is more prevalent mented energy efficiency measures were 4.3 GWh (375 ktoe), among individuals in lower-income brackets, those residing compared to 573 GWh (49 ktoe) energy efficiency measures in in rural areas, single-elderly households, pensioners, and re- the residential buildings included in the National Program from cipients of municipal social aid. Consequently, it is crucial to 2011 to 2017. The budget for 2022–24 is 22.5 million euros 50 examine the coverage of these demographic groups by social from local funds. protection programs. Under the Renovation Wave (Component 5) of the NRRP, A significant portion of these groups had access to various there is a grant scheme for energy efficiency and resilience in social protection programs in 2019–20, although a notable multifamily residential buildings. The projects seek to cover proportion still lacked access to some initiatives. We assess 100 percent of the investment costs for moderately renovating the coverage and leakage rates of selected social protection pro- multifamily blocks of flats in rural and urban regions, particu- grams in Romania, including pensions, unemployment benefits, larly in marginalized urban or rural areas, as defined in the na- child and family allowances, and disability benefits, for 2019–20. tional mapping or declared in the Integrated Local Development Figure 61 illustrates that almost 90 percent of elderly Romanians Strategies. The targeted buildings are those constructed before had access to pensions, and all families with children received 2000, with the maximum eligible project value of 200 euros/ some form of family or child allowances. Approximately 70 per- m2 for moderate renovation work (deployed area), excluding cent of families with members with disabilities reported access VAT. This initiative focuses on areas with a population at risk to disability benefits. However, only a negligible proportion of of poverty and social exclusion, allowing local authorities to the unemployed received unemployment benefits. Additional- execute renovation projects through their Local Development ly, only 5 percent of the population at risk of poverty received or Integrated Urban Development Strategies. The total budget means-tested noncontributory benefits. Interestingly, over half for this endeavor is 219 million euros, and as of 2022, 29 sub- of all households at risk of poverty, regardless of whether they mitted projects totaling approximately 100 million euros had were in the targeted group or not, received child and family al- been registered. lowances—a slightly larger proportion than of those not at risk Another call for projects for thermal rehabilitation of of poverty. Conversely, concerning pensions, coverage rates were blocks of flats under the Regional Operational Program higher for those not at risk of poverty compared to their poor (2014–20) is specifically designed to fund investments to en- counterparts. Regarding benefit adequacy, per capita pensions hance energy efficiency in both public and residential build- demonstrated the highest average, followed by disability and ings and public lighting. In 2018, a total of 279 multifamily unemployment benefits. residential buildings, comprising 19,596 flats, underwent ther- 50 Raport de monitorizare a implementarii Planului National de Actiune în domeniul Eficientei Energetice (PNAEE), ANRE 2019: https://www.anre.ro/ ro/eficienta-energetica/rapoarte/rapoarte-de-monitorizare-aimplementarii-planului-national-de-actiune-in-domeniul-eficientei-energetice- pnaee Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 93 Figure 61 Soci al Protecti on B en efi ts i n Ro m a nia— Cove rage a nd Lea kage Ra te s , 2 01 9 –2 0 1.00 1.0 0.87 0.9 0.8 0.67 0.7 0.56 0.6 0.48 0.48 0.5 0.4 0.3 0.23 0.19 0.2 0.13 0.14 0.05 0.02 0.04 0.02 0.04 0.01 0.1 0.00 0.00 0.00 0.01 0.0 Pension Unemployed Child/family allowance Disability benefit Mean-tested non- contributory benefit Coverage Leakage Coverage of the at-risk of poverty Coverage of the not at-risk of poverty Source: Own estimates based on EU-SILC (2020). Note: Social transfers were directly identified in the survey data. Coverage refers to the proportion of the intended target groups covered, while leakage pertains to the proportion of the nontargeted group covered. We cannot fully replicate the target population behind these social protection measures. Consequently, shares of the coverage and leakage rates might also reflect measurement errors. Family and children-related allowances are at the household level and we divide them by the number of household members. All other variables are collected at the individual level. We exclude all observations with zero values. For a detailed description of how we construct the variables, see appendix 1. To assess the coverage of energy-poor individuals by exist- ventional social protection schemes and how it contrasts with ing social protection programs, we employ statistical match- the broader population (figure 62). Approximately one-third of ing techniques to integrate data from the EU-SILC and the the energy-poor have access to pensions and 4 out of 10 ener- HBS. A nuanced understanding of Romania’s social protection gy-poor households receive child or family allowances. Howev- schemes is crucial for evaluating the effectiveness of these mit- er, energy-poor individuals do not benefit from unemployment igation measures. While the HBS provides limited insight, the benefits, and only a negligible fraction receive disability bene- EU-SILC contains necessary information—but it lacks expendi- fits and benefits from means-tested noncontributory programs. ture data, making identification of the energy poor challenging. When narrowing down the energy-poor population to a specif- To address this, we utilize statistical matching techniques, as ic target demographic, 72 percent of the energy poor who live extensively detailed by Rude and Robayo-Abril (forthcoming b), with elderly persons who are covered by pensions (versus 87 to merge and harmonize these datasets. Due to data constraints, percent of the overall population). Similarly, only 34 percent of the analysis is based on 2019 data, as our 2020 EU-SILC data the population who reside with a person with disabilities and captures income information from 2019. Therefore, we use HBS are energy poor receive benefits, compared to 67 percent of all data from 2019 during the matching process. individuals who reside with someone with a disability. Coverage The coverage rates of social programs among the ener- rates for unemployment benefits surpass the general average, gy-poor target population are comparable or higher than the with 18 percent of energy-poor individuals who are unemployed total population, except for pensions and disability benefits, accessing these benefits, compared to only 5 percent across where they lag significantly behind the average; the lower all unemployed individuals. Similarly, the coverage rates for coverage rates for pensions among the energy poor are con- households with children accessing child or family allowances cerning, especially given that pensioners are significantly (universal) remain similar to the total average. Figure 63 under- impacted by energy poverty and subsequent energy price scores that the average benefit amounts received by beneficiaries increases. We now investigate the energy poor’s access to con- are noteworthy. 94 Romania Energy Poverty Assessment Figu re 63 B e ne f it s Pa id to t he Ene rg y Figure 62 Sh are of En erg y Poor (P1 0) Po o r v ia t he Tra d it io nal S o c ial Sys te m as A ccessing Soci al B en efi ts by Ty p e, 2 01 9 –2 0 Pro po r t io n o f To tal H o u s e ho ld I nco m e, 2 01 9 –2 0 1.2 80 71.0 1.00 70 1.0 60 52.0 0.8 0.72 50 39.2 0.6 40 0.41 0.36 0.34 30 0.4 17.0 0.18 20 0.2 0.08 4.4 6.5 0.00 0.04 0.08 10 0.00 0.00 0 0.0 Pension Unemployed Child/family allowance Sickness benefits Mean-tested non- Pension Unemployed Child/family allowance Sickness benefits Mean-tested non- Disability benefit contributory benefit Disability benefit contributory benefit All energy poor (p10) Conditional to specific group Share of total household income (energy poor) Source: Own estimates based on matched 2019 HBS and 2020 EU-SILC dataset. For a detailed description of the statistical matching approach, see Rude and Robayo-Abril (forthcoming b). Note: The left graph shows the coverage rate of all energy poor—independent of whether they are targeted or not—in terms of several social benefits in Romania. The right graph shows the per capita annual social benefit paid to energy-poor households, independent of whether they form part of the target group or not. Family and children-related allowances are at the household level, and we divide them by the number of household members. All other variables are collected at the individual level. We exclude all observations with zero values. For a detailed description of how we construct the variables, see appendix 1. 5.3 How Do We Design Effective air quality, boosting property value, creating employment op- Mitigation Measures to Protect the portunities, and diminishing reliance on fossil fuels. Most Vulnerable Energy Poor? Examining energy expenditure patterns and affordabil- ity across various income levels and identifying vulnerable groups, as presented in the previous chapters, is essential for Overarching principles formulating effective policies that prioritize the socioeco- nomically disadvantaged during the shift toward environ- Effective social safety nets are vital in shielding individuals mentally sustainable practices. The transition toward greener from energy poverty, especially in the short term; simultane- and more-efficient energy sources stands as an important pillar ously, household energy-efficiency measures can significantly of the European Green Deal (European Commission 2023). enhance overall welfare, particularly in the medium term. Ef- However, a significant challenge lies in the fact that financial- fective policy solutions must focus on supporting enhancements ly constrained households, particularly those in lower-income in housing and heating appliance efficiency through subsidized brackets, encounter obstacles in transitioning to renewable and investments. Additionally, aiding households struggling to meet cleaner energy sources (González-Eguino 2015). This economic energy bill payments for heating is crucial. Energy efficiency barrier has substantial implications, leading to heightened en- initiatives can yield positive outcomes across various domains, ergy costs and potential negative health outcomes (Sagar 2005). including lowering energy costs, enhancing indoor comfort and Addressing the challenges faced by the more-vulnerable groups Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 95 is crucial to ensuring an equitable and inclusive transition to- adjustment for inflation. This erosion in their efficacy increases ward sustainable energy for all. the risk of households’ slipping into poverty. In the current inflationary environment and the con- Adhering to these principles, we explore potential strate- strained fiscal space, two aspects are critically important gies with a dual focus: (1) safeguarding the most vulnerable when designing effective policy packages: well-targeted mea- groups identified in our analysis, specifically poor households sures that are cost-efficient and have sufficient adequacy, to receiving means-tested and noncontributory cash transfers, the extent possible, so that the welfare losses of the rising en- as well as single-elderly and pensioner households; and (2) ergy prices are offset. enhancing the size or generosity of benefits. Considering the First, while augmenting the incomes of vulnerable seg- pinpointed inefficiencies in AROP rates and energy-poverty pat- ments can enhance overall societal well-being, policies to terns and recognizing the demonstrated effects on subgroups, support vulnerable populations necessitate careful balancing prioritizing poorer households and the elderly emerges as a with fiscal viability within resource constraints. Focusing on promising approach for short-term income mitigation strat- recipients based on their specific needs, rather than adopting a egies. These initiatives are not only practical in terms of im- universal approach, enables the allocation of limited resources plementation, but can also capitalize on existing infrastructure to those who require it the most, thus mitigating the ‘leakage’ within social protection measures designed for these identifiable of the poverty budget to non-needy individuals. This approach population segments. enables the more generous provision of benefits to a smaller yet deserving group and for the program to operate within a more Policy simulations constrained budget. In essence, strategic targeting enhances the effective allocation of resources (Coady et al. 2004; Skoufias In our simulations, we assess the potential mitigating impact and Coady 2007). of income-support initiatives on vulnerable populations fac- Therefore, countries should give precedence to protecting ing rising energy prices within a constrained fiscal space. This the susceptible population through targeted assistance while policy approach aims to offer temporary, targeted support to maintaining strict fiscal control to combat inflation. Across those affected by energy and income poverty, utilizing existing many EU nations, efforts to protect consumers, both individuals well-targeted programs to minimize administrative costs asso- and businesses, from escalating food and energy costs have pri- ciated with instituting new benefits. This strategy is built on the marily involved implementing price limits coupled with certain understanding that current social transfers are adequately tar- forms of financial aid. However, these general restrictions on geted but constrained in their level of assistance or generosity. energy costs and the provision of indirect utility subsidies can For each policy scenario, we consider an annual government also benefit higher-income groups, being less effective than pre- expenditure of 1.4 billion euros, equivalent to the yearly fiscal cisely targeted social assistance, and can strain fiscal resources budget allocated for electricity and natural gas price caps.51 We significantly. divide the total government expenditure across the beneficia- Second, it is crucial to thoroughly assess the adjustment ries of each potential program to estimate the size of the annual of benefit levels to match the increasing cost of living—an transfer per beneficiary. Then, we re-estimate welfare indicators often-overlooked aspect in policy formulation. Implement- (poverty headcount, poverty gap, and inequality). This entails ing mechanisms to index government benefits and tax credits disbursing cash transfers to the identified households and re-es- intended to alleviate financial strain is vital for maximizing their timating the poverty rate using simulated per adult equivalent impact on reducing poverty. Unfortunately, the value of numer- household income, which already includes the cash transfers. ous government benefits and tax credits designed to assist strug- We simulate two policy scenarios with alternative financ- gling families can easily erode due to the lack of proportional ing options. 51 Estimates based on data on yearly spending from April 2022 to April 2023 on electricity and gas from ANRE: https://anrero/suma-totala-verificata- de-anre-si-transmisa-spre-decontare-a-depasit-205-miliarde-ron/. 96 Romania Energy Poverty Assessment Policy Scenario 1: Income support The abovementioned groups have been chosen due to measures, financed with external funding their susceptibility to energy poverty, their ease of identifica- tion, and the existence of social protection measures in Ro- This policy scenario entails implementing income-support mania that already target them. Policy makers can, therefore, measures for households, with funding sourced from ad- leverage the existing infrastructure to implement cash transfer ditional channels, such as the SCF. Importantly, the existing programs for these groups without significant additional imple- structures for electricity and gas prices remain unchanged. Un- mentation and administrative costs. der this scenario, price caps are retained and income-support An effective strategy to counter the upsurge in poverty measures are financed through the SCF, constituting 23 percent resulting from rising energy costs involves extending finan- of the total SCF allocation. The fiscal envelope is assumed to cial assistance to pensioners at risk of poverty and single-el- be 1.4 billion euros. The income-support initiatives comprise derly households, two demographics at heightened risk. By a supplementary cash benefit integrated into means-tested and implementing cash transfer programs for these groups, one can noncontributory programs. Additionally, a top-up is introduced alleviate the decline in their income, thereby mitigating the ad- to target pensioners and single-elderly households identified as verse effects. This approach holds significant merit, considering highly vulnerable groups in our prior analysis. that single-elderly households can be readily identified and are In this scenario, we evaluate eight distinct policy mea- among the most adversely impacted by the energy price increas- sures aimed at addressing the heightened vulnerability of es, and a top-up transfer to pensioners at risk of poverty is easy various vulnerable groups, including single-elderly house- to implement. By tailoring support to these vulnerable demo- holds, pensioners, and AROP households receiving benefits graphics, we can ensure a targeted and efficient response to the through means-tested noncontributory programs. We chose challenges posed by rising energy prices and their subsequent these target groups based on their susceptibility to energy pov- impact on poverty levels. erty and existing targeting mechanisms in Romania. Given the Our findings indicate that more-targeted approaches crucial need to shield the poorest and most vulnerable segments are more cost-efficient in mitigating the impact of price in- of the population from the repercussions of rising energy pric- creases. According to the microsimulation results, adopting a es, it is important to determine the most suitable measures and universal approach where all Romanian households receive a ascertain when interventions would be deemed sufficient, as cash transfer would not sufficiently counteract the escalating highlighted by Guan et al. (2023). To identify the most effective poverty effect (policy measure 1, table 3). This insight is partic- and cost-efficient short-term strategies for mitigating the impact ularly relevant, given that various European governments have of energy price increases, we employ microsimulations to assess implemented general measures, such as price caps, energy bill the poverty-level effects of cash transfers. Our focus is on inter- discounts, subsidies, or tax reductions, to address rising energy ventions targeting the following groups: prices across the population (Ari et al. 2022). Alternatively, adopting more-targeted approaches, 1. Entire population (for illustrative purposes) specifically focusing on single-elderly households or pen- 2. Single-elderly sioners at risk of poverty, proves to be more cost-effective. 3. Single-elderly at risk of poverty Income-support programs directed at these groups provide a 4. Pensioners greater reduction in poverty and the poverty gap compared to 5. Pensioners at risk of poverty pre-price-increase levels within the same fiscal budget (policy 6. Recipients of means-tested noncontributory programs measures 2, 3, and 5, table 3). In these cases, the cash transfer 7. Bundle 1: Pensioners at risk of poverty and recipients of amounts to a substantial percentage, ranging between 23 and means-tested noncontributory programs 53 percent of their total household income. Introducing a top- 8. Bundle 2: Pensioners at risk of poverty and single-el- up transfer for recipients of means-tested noncontributory pro- derly at risk of poverty grams alone (policy measure 6, table 3) would reduce poverty Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 97 less than other measures targeting single-elderly households or Lastly, programs bundling pensioners at risk of poverty with pensioners at risk. It is essential to consider other dimensions cash-transfer recipients’ households (policy measure 7) or pen- of equity, and emergency measures should be implemented sioners at risk of poverty with single-elderly households at risk through inclusive consultations and active engagement with of poverty (policy measure 8) would also effectively reduce pov- specific communities and the Romanian population as a whole. erty and the poverty gap below the pre-price-increase level. Table 3 Si mulati on of Sh ort-Term M eas u re s Mit iga t ing t he D ire c t Ef fe c t s o f Ris ing Ene rg y Pr ice s under Curren t Prog ram Structure Simulated Simulated Simulated Simulated Monthly Simulated Simulated Poverty Reduction of Simulated Targeted Poverty Cash Simulated Policy measure Poverty gap Gini reduction poverty gap Adequacy population headcount transfer Beneficiaries (%) Index (percentage (percentage (% Income) (%) Size points) points) (Euros) 1 All 24.4 9.2 33.9 0.48 0.38 72 19,348,563 5.9 2 Single-Elderly 22.2 8.8 33.0 2.68 0.77 1,136 1,232,223 22.4 3 Single-Elderly AROP 22.2 8.7 32.8 2.68 0.82 2,696 519,197 53.0 4 Pensioner 24.0 9.1 33.4 0.93 0.46 332 4,211,665 6.2 5 Pensioner AROP 21.2 8.4 32.8 3.74 1.12 1,665 841,003 30.0 6 CT recipients 23.6 8.7 33.2 1.31 0.82 1,800 777,846 31.8 Bundle 1: Pool beneficiaries 7 Pensioner at-risk and CT 22.0 8.2 32.9 2.90 1.38 890 1,573,329 15.5 recipients Bundle 2: Pool beneficiaries 8 Pensioner at-risk and Single- 20.8 8.3 32.7 4.11 1.27 1,478 947,173 28.7 elderly at-risk Source: Own estimates based on the matched HBS and EU-SILC dataset. Note: This table shows eight potential short-term measures to mitigate energy price increases with a total fiscal budget of 1.4 billion euros for each measure, assuming that all other things stay equal. The pre-price-increase poverty level is 24.9 percent. Policy Scenario 2: Income support subsidies would increase an additional 20 percent, bringing the measures financed by phase-out of energy total increase to 60 percent based on IMF data52 (refer to annex price caps 7 for additional price-increase scenarios). As in previous scenar- ios, we maintain a total annual expenditure of 1.4 billion euros This policy scenario involves income-support measures for for each policy measure. households, financed with the resulting fiscal revenues gen- Although the elimination of the price caps implies an es- erated by the full elimination of the implicit residential price timated increase in energy prices, which may have adverse caps of natural gas and electricity. As price caps for electricity distributional consequences, these can be mitigated through and natural gas are eliminated, in this scenario energy prices the targeted measures. The same income support measures increase even further. We assume that energy prices without 52 Based on the “IMF Fossil Fuel Subsidies Data: 2023 Update”: www.imf.org/en/Topics/climate-change/energy-subsidies. 98 Romania Energy Poverty Assessment used in policy scenario 1 are assumed. The additional 20 percent with price caps to 34.1 percent without price caps. Income- sup- increase in energy prices due to the elimination of price caps port measures that target single-elderly and pensioners are ex- would increase poverty from 24.9 percent (in baseline scenar- pected to provide an effective cushion against the rising energy io—policy scenario 1) to 25.8 percent in policy scenario 2 (fig- prices, because more than half of the targeted populations are ure 64). Moreover, inequality would increase from 33.8 percent both energy poor (P10) and income poor (figure 65). Figure 64 Sh ort-Term Effects on Wel fare by Figu re 65 Cove rage o f Po te nt ial Mit iga t io n Polic y Scen ari o Meas u re s by Pove r ty G ro u ps 40 100% Percentage of the population 33.8 34.1 covered by the program 35 80% 30 60% 24.9 25.8 25 40% 20 20% 15 0% 9.6 9.9 All Single-elderly Single-elderly AROP Pensioner Pensioner AROP CT recipients Bundle: Pensioner AROP and CT recipients Bundle: Pensioner AROP and single-elderly AROP 10 5 0 Poverty rate Poverty gap Gini index (at-risk of poverty) (at-risk of poverty) With price-caps (policy scenario 1) Only energy poor (p10) Only at-risk of poverty Without price-caps (policy scenario 2) Both energy poor Neither energy poor and at-risk of poverty nor at-risk of poverty Source: Own estimates based on the matched 2019 HBS and 20202 EU-SILC dataset. Note: CT recipients refer only to those receiving benefits through means-tested and noncontributory programs. In this scenario, where residential price caps are entirely assisting the most vulnerable with managing higher energy prices. eliminated and the resulting fiscal revenues are allocated to Notably, providing a cash transfer of 1,665 euros per year to pen- income-support measures, targeted approaches become even sioners at risk (measure 5, table 4) not only mitigates the increase more important for mitigating adverse effects. Our simulations in poverty, it in fact reduces poverty levels much further than in show that these temporary income-support measures, while the baseline scenario. A similar effect is achieved by providing slightly more challenging to administer compared to the price a 1,478 euros cash transfer to pensioners at risk of poverty and caps, offer a cost-effective and likely more efficient approach to single elderly at risk of poverty (bundle 2, table 4). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 99 Table 4 Si mulati on of Sh ort-Term M eas u re s Mit iga t ing t he D ire c t Ef fe c t s o f Ris ing Ene rg y Pr ice s by Eliminat in Pri ce Cap s for Electri ci ty an d N a t u ral Gas Simulated Simulated Simulated Simulated Simulated Monthly Simulated Poverty Reduction of Simulated Policy Targeted Poverty Gini Cash Simulated Poverty gap reduction poverty gap Adequacy (% measure population headcount transfer Beneficiaries (%) Index (percentage (percentage Income) (%) points) points) Size (Euros) 1 All 25.1 9.5 34.1 0.69 0.39 72 19,348,563 4.7 2 Single-Elderly 23.1 9.1 33.2 2.67 0.84 1,136 1,232,223 22.7 Single-Elderly 3 23.1 9.1 33.0 2.68 0.88 2,696 519,197 53.6 AROP 4 Pensioner 24.9 9.4 33.6 0.86 0.49 332 4,211,665 6.2 5 Pensioner AROP 22.1 8.7 33.0 3.70 1.20 1,665 841,003 30.3 6 CT recipients 24.5 9.1 33.4 1.23 0.85 1,800 777,846 32.2 Bundle 1: Pool beneficiaries 7 23.2 8.5 33.1 2.55 1.44 890 1,573,329 15.6 Pensioner AROP and CT recipients Bundle 2: Pool beneficiaries 8 Pensioner AROP 21.7 8.6 32.9 4.07 1.36 1,478 947,173 29.1 and Single-elderly AROP Source: Own estimates based on the matched 2019 HBS and 2020 EU-SILC dataset. Note: This table shows eight potential short-term measures to cushion energy price increases with a total fiscal budget of 1.4 billion euros for each measure, assuming that all other things stay equal. The pre-price-increase poverty level is 25.8 percent. Currencies are in euros. Cash-transfer (CT) recipients refer only to those receiving benefits through means-tested and noncontributory programs. Adequacy measures the share of the CT with respect to the total household income. It is important to acknowledge that our microsimulations Romania using EU-SILC data from 2020. The available infor- of mitigation measures are based on several assumptions and mation in the EU-SILC dataset is often aggregated, which ham- are subject to some limitations. This microsimulation exer- pers the comprehensive reconstruction of the respective social cise that is subject to some limitations. First, we assume that protection programs. Additionally, crucial information neces- all other relevant parameters, such as prices, remain constant sary for a thorough analysis of these programs is frequently throughout the analysis. We also assume that there are no be- absent from the EU-SILC 2020 dataset. This observation aligns havioral changes among the population as a result of the cash with the findings of previous research conducted by Militaru et transfers’ being implemented. Second, we assume perfect take- al. (2022), who also emphasize the limited scope for simulating up of the programs among the targeted beneficiaries, meaning tax and social benefits in Romania. that all eligible individuals and households receive the cash A crucial aspect to contemplate when considering the im- transfers. Lastly, we operate under the assumption that there plementation of new measures is the level of government sup- is no take-up of the program by population groups who are port likely to be garnered when modifying existing protection not specifically targeted. Furthermore, we must highlight the schemes related to energy poverty. As per recent estimates from limited feasibility of replicating the social protection system in a rapid survey conducted by the World Bank in July 2023, over 100 Romania Energy Poverty Assessment half of the population perceived the price caps implemented in bills and maintaining daily living standards. However, these response to the energy crisis the previous summer as inadequate. are temporary solutions. To achieve lasting improvements, it is Only a minimal fraction of the population, specifically 4.3 per- necessary to also invest in energy efficiency measures that will cent, expressed the opinion that the response was too generous. have a medium-term impact, as these measures can strengthen These figures underscore the significance of a clear communica- the resilience of households against future energy price shocks. tion strategy and transparent dissemination of information re- garding the benefits and costs associated with policy interventions Strengthening energy-efficiency aimed at safeguarding the energy poor within the country. initiatives with insights from Regarding the modality, cash and in-kind transfers have dif- behavioral science ferent advantages that make each suitable for specific contexts. A careful analysis of the delivery systems for energy compensatory Households in Romania face energy inefficiencies (especial- transfers is important for Romania. Cash transfers typically have ly in the heating space) that could be addressed effectively lower administrative costs and give recipients greater freedom with more and better upgrade-support programs that focus over their consumption, providing flexibility in expenditures and on sustainable transitions. As highlighted above, most of the alleviating credit constraints. If properly indexed, they can help policy focus in Romania is on short-term price or income- mitigate the impact of overall increases in prices that may result support measures. Sustainable energy transitions require more due to higher energy prices. In-kind transfers ensure that aid is medium-term approaches that support shifts in technologies used for its intended purpose and can help mitigate the impact of (for example, heating technology and structural and insulation energy inflation when the real value of cash transfers diminish- investments in homes as well as energy efficient appliances), es (when not properly indexed). Some countries have opted for fuels, and use practices, especially for energy-poor households cash and others for in-kind electricity subsidies to protect energy who might not be income poor. Existing renovation programs poor households. France, for instance, has established a system of target mostly urban areas and multifamily buildings, meaning vouchers in place to help households pay their energy bills or to that a large concentration of energy-poor households in ru- cover energy renovations. Evidence of different modalities of cash ral areas and residing in single-family homes does not benefit transfers in the context of energy subsidy reform shows a wide from these programs. Medium-term energy-efficiency solu- range of policy responses. Ukraine initially increased the scope tions are aligned with the EC’s SCF, which aims to support and generosity of its financial assistance programs. Subsequently, it vulnerable households through an improvement in the envi- changed the way benefits were distributed, shifting from providing ronmental and energy-efficiency performance of buildings, budgetary transfers to utility companies to directly compensating including the replacement of fossil-fuel heating installations households. The Dominican Republic’s experience offers valuable with renewable ones.53 insights into coordinating and integrating compensatory cash As discussed in chapter 3, only a small share of the popu- transfers with existing programs, particularly conditional cash lation makes use of energy-efficient heating systems, and only transfers like those in the Programa Solidaridad. Brazil, Malaysia, half of this group can control the temperature at home, which Morocco, and Tunisia have implemented similar approaches with leaves substantial room for improvement in supporting of different levels of coverage. These delivery systems manage the sustainable heating transitions. Based on survey estimates entire process, from identifying beneficiaries to onboarding and from 2023, only 1 out of 10 Romanians indicated that they used making payments (Mukherjee et al, 2023). a high-efficiency stove or heating system (figure 66) and only 2 It is critical to accompany these income support measures, out of 10 Romanians could be classified as using modern heating which are more short-term, with some medium-term energy devices. Critically, half of Romanians rely on traditional/con- efficiency measures. Income support measures are designed ventional heating systems. Another sign of improvements need- to offer quick relief and address urgent needs, such as paying ed in energy efficiency in Romania is the fact that only half of 53 See https://www.consilium.europa.eu/en/infographics/fit-for-55-social-climate-fund/ and https://www.europarl.europa.eu/RegData/etudes/ ATAG/2023/753971/EPRS_ATA(2023)753971_EN.pdf Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 101 households could control the temperature in their home (figure willing to upgrade their heating system and 8 percent would be 67). Among the 80 percent of Romanians living in households willing to install or upgrade insulation in their home (and fewer with traditional heating technologies, only 10 percent would be than one-third of these people would do so within one year). Figu re 67 S ha re o f Po pu la t io n A ble to Co nt ro l Figure 6 6 M ai n Heati n g Source, 2 023 (%) Te m pe ra t u re in H o m e, 2 023 ( % ) 60.00 52.37 Traditional/convention 100% 50.00 al stove Traditional/convention 80% 40.00 al heat 60% High efficiency stove or 30.00 27.57 space 40% High efficiency heat distribut 20% 20.00 District or centralized 10.11 0% 10.00 7.09 heatin Percent 2.07 0.80 I am not sure about the 0.00 system Yes No Don't know/unsure Percent Source: Estimates based on World Bank household survey, July 2023. Energy efficiency initiatives should expand upon the ex- Energy and Business Environment offers support to micro, small isting grant scheme for energy efficiency and resilience in and medium-sized enterprises for the installation of photovoltaic multifamily residential buildings by focusing on energy effi- panels and energy storage systems. Similar programs should be cient investments in both multiapartment buildings and sin- developed with a focus on energy-poor and rural areas with a gle-family homes. Such initiatives should support investments high density of vulnerable communities. in both structural improvements (building envelope) among the Subsidy programs in Romania should learn from successful households with the lowest incomes as well as heating and cooling examples elsewhere in Europe, including by considering the systems, fuels (especially heat source), insulation, and A+, A++, distributional implications of support levels and willingness and A+++ rated appliances. These initiatives should focus not 54 to accept certain subsidy amounts. Three out of four Romanians only on the upgrading of appliances but on the diversification would find switching to a cleaner, more energy-efficient fuel or of energy sources, such as renewables rollout. Successful subsi- power source to be financially difficult, and less than half of re- dy programs have been designed in similar contexts, providing spondents agreed that they had enough saved up to cover large benefits that reflect the demands of different income groups. For unexpected expenses. With this in mind, it is likely that most example, the Clean Air Priority Program (CAPP) in Poland (a households will need part of the upfront costs of these upgrades program that has received financial and technical assistance from covered, though how much will depend upon a variety of factors. the World Bank) is a national subsidy scheme that supports prop- Similar to the income measures discussed above, subsidy levels erty renovations and heat source replacements (including thermal should vary by level of income, with income- and energy-poor retrofitting) in single-family homes in an effort to reduce energy households receiving the largest subsidy amount and non-in- consumption and reduce air pollution55. In Romania, the Elec- come-poor and non-energy-poor households the lowest. New tricUp program Initiated by the Romanian Ministry of Economy, fieldwork could be conducted to understand the willingness to 54 For details on the EU Energy ratings, see https://commission.europa.eu/energy-climate-change-environment/standards-tools-and-labels/ products-labelling-rules-and-requirements/energy-label-and-ecodesign/about_en. 55 https://commission.europa.eu/projects/update-clean-air-priority-programme_en. 102 Romania Energy Poverty Assessment accept certain subsidy levels by income and socioeconomic and resources for energy-poor households, hindering their ability to sociodemographic groups. A study conducted by the World Bank receive grants for renovations. This, together with the existing in four Western Balkan countries found that willingness to ac- 56 barriers faced by energy-poor individuals in accessing current cept hypothetical subsidy amounts to support sustainable heating renovation programs may further diminish their future capacity upgrades was significantly correlated with income and educa- to renovate their homes and transition out of energy poverty. tion: for example, wealthier and more-educated respondents To address this, better-targeted allocation of EU funds in these were significantly more likely than poorer and less-educated programs combined with the mobilization of private financing respondents to accept a subsidy of 30 percent of the total cost for non-energy-poor households is recommended. (30 percent subsidy was the baseline level, which increased in Behaviorally informed solutions to support 10 percentage-point increments to a maximum of 70 percent). sustainable energy and heating transitions Financial support alone, however, will not guarantee that these sustainable transitions happen through support pro- Subsidy programs supporting energy efficiency—and in par- grams; such programs must consider other factors that influ- ticular, sustainable heating upgrades—should address the ence decisions to invest in energy efficiency upgrades. While barriers and enablers uncovered in the diagnostic activity the financial barriers (in particular, real or perceived affordabil- in those programs’ design and implementation. While most ity concerns that influence intentions to change behaviors) are beliefs and attitudes about energy efficiency and sustainable important, these are complemented by information barriers and heating appear to suggest that a favorable environment exists barriers related to beliefs, attitudes, ideas, and perceptions that en- to some extent—for example, the share of the respondnts who compass the cognitive, psychological, and social factors that influ- saw the benefits of upgrades in terms of climate change and ence decision-making.57 The survey of Romanians identified low environmental impact (including local air quality), health, and awareness around support programs (nearly half of Romanians convenience was greater than 50 percent—there is still substan- were unaware of support programs) and beliefs and attitudes re- tial room for improvement. Subsidy programs need to raise garding energy efficiency upgrades that serve to dissuade people awareness about the programs as well as the benefits of these from making investments and participating in public support investments while dispelling the perception of barriers such as programs (including sentiments of exclusion, as captured in the inconvenience factors involved in the process (purchase, instal- qualitative fieldwork). Intentions to upgrade appeared to be most lation, correct use initially and over the long term) of adopting highly correlated with awareness of subsidy programs and fuel a new technology into the home. flexibility (for both heating technology and insulation upgrades) Subsidy program design and implementation should con- and expected increases in property value from heating technology sider the beneficiary journey to support the various decisions upgrades. Addressing these factors represents low hanging fruit and actions that take place at the various stages of program with regard to motivating sustainable heating transitions and any engagement. The decision to invest in energy efficient tech- support program should consider these. nologies and behaviors is a dynamic one, especially when the Finally, targeted allocation of EU grant funds specifically decision is dependent upon external support and influence. As for renovating the housing stock of energy-poor individuals is was identified in the context of the CAPP in Poland,58 beneficia- important. If EU (nonreimbursable) funds are extensively used ries must embark on a journey in participating and benefiting for renovating and investing in households of non-energy-poor from support programs. There is a decision stage (during which individuals, it may deplete the funds available to energy-poor they recognize the need or desire to upgrade, inform themselves households in the future. This could result in a limited pool of about upgrades and support programs, and ultimately decide to 56 “Behavioral Diagnostic of Sustainable Heating Transitions in the Western Balkans: Evidence from Bosnia & Herzegovina, Kosovo, North Macedonia, and Serbia” (World Bank 2022). Internal report. 57 For a discussion of how these factors were incorporated into the CAPP in Poland, see https://blogs.worldbank.org/climatechange/clean-air-and- heating-choices-how-change-homeowners-behavior-poland. 58 https://blogs.worldbank.org/climatechange/clean-air-and-heating-choices-how-change-homeowners-behavior-poland. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 103 move forward), an application stage (during which either subsi- uration and cognitive overload of the citizens is key for dy or loan applications or both are initiated, forms are complet- wide take-up of the message. ed, and financing is secured), and an installation stage (during 2. Use nontraditional messengers to raise awareness of and which beneficiaries wait for approval, commence a project, and motivate participation in subsidy programs. Low insti- learn how to operate new technology, which includes the sus- tutional trust can complicate engagement strategies tained use of the technology). Each of these stages must be com- highlighting the importance of energy efficient invest- pleted before a household benefits from upgrades (or does so in ments as well as participation in public programs. As the long term), and understanding the different pressure points discussed in chapter 4, the level of trust in information in each will help optimize program design and implementation. from government officials about sustainable heating up- Subsidy programs should have embedded solutions that grades was substantially lower than the level of trust in target identified bottlenecks at each stage of the beneficia- information from technicians and social networks, and a ry journey and be evaluated where possible to understand high share of those surveyed did not even trust the main those bottlenecks’ impacts on attitudes and behaviors. The messenger in this type of investment. Engagement strat- design and implementation of subsidy programs must be cen- egies should enlist trusted nontraditional messengers tered around solutions to alleviate identified bottlenecks in the (for example, independent technicians, installers, and process of behavioral change. Where possible, these solutions builders) to inform potential beneficiaries and change should include robust evaluation methodologies (for example, the narrative around sustainable energy use. Testing a experimental design) to identify the causal impact of specific variety of messengers of information about energy ef- strategies on household attitudes and behaviors. Such an ap- ficiency and subsidy programs will help identify those proach will identify both the most effective strategies to improve who are the most effective for particular audiences. take-up and for whom these strategies are most relevant (for 3. Harness social influence in investment decisions. While example, urban vs. rural households or older vs. younger audi- social influence seems to be less of an enabler in the ences). An evidence-based approach also helps programs be re- Romanian context (less than 40 percent of Romanians sponsive to changes in attitudes over time and improve features stated they would be more willing to upgrade if others of program design and implementation as programs are scaled did the same), those with intentions to upgrade (will- up. Below is a list of potential actions that should be considered ing to upgrade or had already upgraded) were signifi- to facilitate sustainable-energy transitions (particularly related cantly more likely to undertake upgrades when social to heating) by means of energy-efficiency subsidy programs. influence was present. As such, any programs seeking to motivate investments in energy efficient technolo- 1. Initiate holistic engagement strategies to elevate the im- gies should consider the importance of communicat- portance and urgency of sustainable energy transitions. ing about rising trends in energy use modernization as Holistic engagement strategies in the context of subsidy well as harness established role models to promote sus- programs serve to promote positive beliefs and attitudes tainable energy behaviors. Role models can be used as around sustainable transitions and thus target behav- messengers or early adopters can be recognized socially iors. Engagement can take the form of mass communi- to change the narrative around energy efficient invest- cation campaigns that raise awareness about the costs of ments. Especially when there is a lack of role models traditional energy-use practices (for example, heating or success stories, which can lead to the perception of practices) and the benefits of modern ones (for exam- participation as being risky, elevating the positive expe- ple, the consequences of heating practices on indoor rience of early adopters and early beneficiaries can have and local air quality) and dispel perceptions of the in- a meaningful impact of participation. convenience of sustainable upgrades. An important as- 4. Improve the framing of the costs and benefits of invest- pect of engagement strategies is to manage the amount ment. Engagement with potential beneficiaries through- of information being communicated to the population. out the decision-making journey should focus on Planning the timing of the campaigns and avoiding sat- improving the framing of the financial costs and ben- 104 Romania Energy Poverty Assessment efits of energy efficient investment to address financial formats for less-digitally-connected groups to reduce and affordability concerns. Perceptions of long pay- expected friction. Engagement should also promote back periods (uncovered in the qualitative fieldwork) inclusion by increasing transparency in the selection can be addressed by providing concrete information on of beneficiaries as well as the eligibility criteria to con- upfront costs, variable costs (in terms of maintenance front the sentiments of exclusion highlighted in the and fuel), and energy savings scenarios based on the focus group discussions. combination with insulation and other energy-saving 6. Consider the “sustained use” phase of the journey from practices (for example, solar panels and investment in the outset, especially with regard to communicating the A-rated appliances). Communications should under- costs and benefits of upgrading. Installation alone does score the rise in energy prices in the country and re- not guarantee that the new behaviors will continue in gion and how energy efficiency can mitigate this trend, the long run, particularly if there are unexpected costs as well as encourage potential beneficiaries to be more or maintenance responsibilities with new technologies future oriented in their approach to energy efficiency in (per the expression in the survey responses of poten- general and heating in particular. tial beneficiaries’ concerns about the reliability and 5. Remove pressure points in the application process itself. upkeep of modern heating systems and solar panels). Qualitative fieldwork uncovered challenges faced by Communication as to the costs and benefits of upgrades households in terms of the process of applying to pro- (highlighted above) should highlight warranties or grams offering support for energy efficiency upgrades. guarantees available with the new technology if these Application processes should be simplified where pos- are available, because these can serve to increase trust sible and offered in both online and in-person or paper in sustainable and clean energy technologies. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 105 Chapter 6 Conclusions and Policy Implications In Romania, a significant portion of households grapple with ment. Prioritizing energy poverty in policy agendas can en- energy poverty, lacking essential features for daily life, such hance well-being, environmental sustainability, and economic as adequate warmth, cooling, lighting and energy. Around 25 resilience. percent of the population experienced energy poverty in 2021, Examining energy spending patterns reveals that con- due to having to allocate 8.7 percent of their expenses to energy. nectivity and energy source availability significantly impact The urgency to address this crisis is heightened in light of the household expenses. Limited access to energy due to infrastruc- energy crisis in the wake of Russia’s invasion of Ukraine, which ture or high costs can lead to seemingly low energy spending is causing notable financial stress for households and impacting shares, highlighting another form of energy vulnerability—low well-being negatively overall. access. Analyzing household connectivity to different energy The pressing issue of energy poverty is a critical concern sources, we find nearly universal access to electricity or renew- for European policy makers, especially amid the current en- able energy in 2021, so that lack of connectivity can be dismissed ergy crisis. Steep increases in natural gas and electricity prices as a driving factor. However, variations emerge concerning nat- are further exacerbating the financial strain on households. This ural gas coverage, influenced by household preferences rather crisis is particularly acute, given its linkage to income poverty than accessibility. Additionally, wood proves a vital heating and and its adverse effects on health, education, and the environ- cooking source, particularly in rural areas, shedding light on the nuanced energy dynamics and confounding factors affecting income levels; single-elderly households are particularly vul- spending patterns. nerable. However, a more comprehensive study that considers Our estimates indicate that approximately a quarter of factors such as connectivity, preferences, technology efficiency, Romania’s population faced energy poverty in 2021, spending and associated costs is needed to grasp this relationship fully. an average of 8.7 percent of their household budget on energy. The focus must extend to the non-financial determinants of be- The overlap of energy and monetary poverty reveals that 7–15 haviors that sustain energy poverty, specifically the choice and percent of households were energy poor but not income poor, use of energy-intensive technologies. Non-financial determi- emphasizing the distinction between the two forms of poverty. nants include attitudinal and belief systems, social norms, in- This has significant policy implications, because it highlights tention to action gaps and cognitive biases that act as barriers to the need for targeted energy policies such as subsidies for ener- choosing and using energy efficient options. Integrating insights gy-efficient technologies, especially for low-income households. from behavioral science into the sustainable energy transition Moreover, enhancing energy infrastructure in low-income areas framework provides a holistic perspective on the barriers to and is crucial to ensure reliable and affordable energy access. The facilitators of behavior change, bridging the gap between energy data underscore the vulnerability of low-income groups to en- poverty conditions and the attitudes and behaviors perpetuat- ergy price hikes, underscoring the need to mitigate the dispro- ing them. This approach offers the nuanced insights crucial for portionate impact on their household budgets. policy development. We found that energy expenditure shares are influenced Drawing from the comprehensive empirical analysis pre- by the technologies used for heating and cooking and in- sented in this report, we have identified a range of valuable come levels; single-elderly households are particularly vul- policy insights. Building upon these findings, we propose a set nerable. However, a more comprehensive study that considers of thoughtful policy recommendations aimed at effectively ad- factors such as connectivity, preferences, technology efficiency, dressing and mitigating energy poverty within Romanian house- and associated costs is needed to grasp this relationship fully. holds. These recommendations are designed to offer pragmatic Certain household types, such as those of the single elderly, of solutions and guide policy makers in formulating strategies pri- pensioners, and those with unemployed members or a female oritizing the welfare and well-being of vulnerable households head, tend to have higher energy spending shares. Policy mak- grappling with energy poverty. ers should prioritize these households when it comes to energy The behavioral approach adds value to traditional diag- poverty interventions. nostics by providing a user-centric understanding of policy The impact of rising energy prices is particularly pro- challenges in residential sustainable energy transitions. Stan- nounced for households already exposed to energy poverty. dard development policies typically target financial resources, As energy prices increase, so does these vulnerable groups’ risk incentives, laws, or information provision, while a behavioral of falling into poverty. This situation has significant welfare and approach explores mindsets, decision-making frames, and so- equity implications, as it can exacerbate inequalities between cial environments. This holistic understanding sheds light on different socioeconomic groups and further marginalize the decision-making processes, crucial for effecting behavior change most vulnerable. Mitigating the effects of rising energy prices in specific policy contexts. Residential sustainable energy tran- in both the short and long terms is crucial to ensure that energy sitions involve upgrading behaviors and technologies for clean poverty is manageable for those struggling to make ends meet. and efficient energy use. Thus there is a pronounced need for Effective policy measures that address price increases can help policies and programs addresing both behaviors and attitudes. protect the well-being of energy-poor households and promote First, special attention should be directed toward sin- the more equitable distribution of resources. gle-elderly households, because they face increased vulner- Addressing energy poverty requires a thorough under- ability to rising energy costs and energy poverty. Given this standing of its diverse drivers, which encompass household demographic group’s higher rates of energy poverty and larger sociodemographics, energy technologies, and geographical shares of energy expenditure, policy makers must focus on them location. We found that energy expenditure shares are influ- when formulating measures to tackle energy poverty effectively. enced by the technologies used for heating and cooking and Previous studies have shown that single-elderly households are 108 Romania Energy Poverty Assessment more susceptible to health risks related to extreme tempera- tributing to the broader effort to combat energy poverty and tures, making implementing policies that address their specific promote environmental stewardship. needs and circumstances imperative. In Romania, single-elderly Even in a context of limited fiscal resources, efficient tar- households also report inadequate housing conditions and are geting of policy interventions aimed at mitigating price in- at a higher risk of poverty. Additionally, differences in the tech- creases through income-support measures can significantly nologies they use for cooking and heating further contribute to impact poverty reduction. Policy makers can enhance the ef- their vulnerability. By recognizing and prioritizing single-elderly fectiveness of their efforts to reduce energy poverty by tailoring households’ challenges, policy makers can implement targeted interventions to meet the needs of different vulnerable groups, solutions, such as financial assistance programs or tailored ener- thereby optimizing the allocation of resources and achieving gy-saving initiatives, to alleviate their energy-related hardships. more-significant outcomes. Among the targeted strategies, Second, the report’s findings also highlight a worrisome focusing on single-elderly households or pensioners at risk of aspect concerning pensioners. While coverage rates of social poverty emerges as notably practical and feasible, given their rel- programs among energy-poor target populations are similar or atively straightforward identification and vulnerability to ener- even higher compared to the coverage of the total population, gy poverty. However, policy makers must carefully balance cost they lag when it comes to pensions. This is alarming because efficiency with equity considerations, such as intergenerational pensioners are particularly susceptible to monetary and energy concerns and fiscal sustainability. While cost savings are essen- poverty. The impact of price increases can be particularly harsh tial, it is equally crucial to ensure that vulnerable populations for this group, leading to higher AROP rates. To tackle energy are adequately supported and that the long-term implications poverty effectively, it is essential to consider targeted strate- of policy measures are sustainable and equitable. gies focusing on pensioners, especially those in single-person Clear communication is vital to garnering support for households and at risk of poverty. By implementing measures new policy interventions targeting energy-poor households. that specifically cater to their needs, policy makers can create Providing the Romanian population with a transparent un- more-effective and efficient solutions to address energy poverty derstanding of the benefits and costs of such interventions is among this vulnerable segment of society. Finally, an easy way essential to secure public backing. When people comprehend to mitigate the rising energy costs faced by pensioners would be the potential positive impacts on their lives and the broader so- the implementation of a top-up transfer. ciety, they are more likely to embrace and actively participate in Third, it is evident that energy-inefficient housing dispro- these initiatives. Effective communication strategies can foster a portionately affects low-income households. To alleviate this sense of collective responsibility and engagement, encouraging burden, we recommend prioritizing renovation programs that a unified effort to combat energy poverty and promote a more target energy-efficient housing to benefit the poor. Such initia- sustainable future for all. tives align with the goals of the European Green Deal and can Finally, multisector approaches are essential for measur- significantly improve the living conditions of vulnerable commu- ing and addressing energy poverty due to its complex and nities by reducing energy leakages and the associated expenses. multidimensional nature. Tackling energy poverty requires the Addressing energy poverty requires a multifaceted ap- involvement of various sectors, including energy, transport, in- proach that considers the unique challenges of different so- frastructure, and social sectors such as social protection, health, cioeconomic groups. One key aspect to bear in mind is that and education. It also necessitates interconnected strategies at poorer households might need to be made more aware of en- multiple geographic levels, from EU-wide policies and compari- vironmental problems. Therefore, implementing interventions sons to local-level monitoring. Tackling energy poverty requires that involve the education of these households about pollution, a mix of public policies, including energy, social, and housing grime, and other environmental issues could be beneficial. Such policies. Consequently, a wide range of stakeholders must be in- education would not only raise environmental consciousness; it volved in measuring, defining, and monitoring energy poverty. also has the potential to spark energy-efficient behavioral pat- Implementing energy poverty measures should be a collabora- terns. By empowering these households with knowledge, they tive effort among different ministries, academic institutions, and can become active participants in sustainable practices, con- local communities. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 109 Some knowledge gaps are worth emphasizing for future renovations for low-income households, and targeting support studies, particularly on the usage of wood and its influence on for single-elderly households, policy makers can make signifi- energy expenditure and energy poverty measures. This study cant strides toward creating a more equitable and energy-con- could provide valuable insights into the complex interplay be- scious society. In addition, by educating poorer households tween energy usage, energy prices, and energy accessibility and about environmental issues, addressing the impact of rising their relationship to illegal logging and climate change. By un- energy prices on vulnerable populations, targeting pensioners derstanding these dynamics, policy makers can devise more-tar- with specific policy measures, and effectively communicating geted and practical strategies to combat energy poverty and the benefits of interventions, Romania can take significant promote sustainable energy practices. strides in alleviating energy poverty. These measures are vital to The policy recommendations outlined in this report pro- achieving long-term energy security, improving the well-being vide a foundation for addressing energy poverty and pro- of energy-poor households, and mitigating the adverse effects of moting sustainable energy practices. By conducting further energy poverty on vulnerable communities, ultimately contrib- research on wood usage, prioritizing energy-efficient housing uting to a more sustainable and equitable society. 110 Romania Energy Poverty Assessment Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 111 References ADĂSCĂLIŢEI, Dragoș, Cristina Rat, and Marcel Spătari. bericht.” Österreichische Zeitschrift für Politikwissenschaft 2020. Improving Social Protection in Romania. Friedrich 47 (2): 61–76. Ebert Stiftung. https://library.fes.de/pdf-files/bueros/bu- BACON, Robert, Soma Bhattacharya, and Masami Ko- karest/16834.pdf. jima. 2010. “Expenditure of Low-Income House- ALPMAN, A. 2016. “Implementing Rubin’s Alternative Multi- holds on Energy: Evidence from Africa and Asia.” ple-Imputation Method for Statistical Matching in Stata.” Extractive Industries and Development Series The Stata Journal 16 (3): 717–39. Working Paper 16. World Bank, Washington, DC. ARI, Anil, Nicolas Arregui, Simon Black, Oya Celasun, Dora https://openknowledge.worldbank.org/entities/ M. Iakova, Aiko Mineshima, Victor Mylonas, Ian W. H. publication/0c0a1d5b-4d85-5394-9603-fd77dc8ad9c8. Parry, Iulia Teodoru, and Karlygash Zhunussova. 2022. BADIANI-MAGNUSSON, Reena, and Eva Militaru. 2022. “Surging Energy Prices in Europe in the Aftermath of “The Distributional Impacts of Taxes and Social Spend- the War: How to Support the Vulnerable and Speed Up ing in Romania—2018 Model and Selected Simulations.” the Transition Away from Fossil Fuels.” International World Bank Group, Washington, DC. https://policy- Monetary Fund Working Paper. https://www.imf.org/en/ commons.net/artifacts/2680987/thedistributional-im- Publications/WP/Issues/2022/07/28/Surging-Energy-Pric- pacts-of-taxes-and-social-spending-in-romania/3704468/. es-in-Europe-in-the-Aftermath-of-the-War-How-to-Sup- BHATIA, M., and N. Angelou. 2015. “Beyond Connections: port-the-Vulnerable-521457 Energy Access Redefined.” ESMAP Technical Report. BACHER, J., and D. Prandner. 2018. “Datenfusion in der World Bank Group: Washington, DC. sozialwissenschaftlichen Wahlforschung–Begründeter BANERJEE, Rajabrata, Vinod Mishra, and Admasu Asfaw Verzicht oder ungenutzte Chance? Theoretische Vorüber- Maruta. 2021. “Energy Poverty, Health and Education legungen, Verfahrensüberblick und ein erster Erfahrungs- Outcomes: Evidence from the Developing World.” Energy Economics 101: 105447. BANKWATCH Network. 2022a. “The Energy Sector in 36–48. https://www.sciencedirect.com/science/article/abs/ Romania.” https://bankwatch.org/beyond-fossil-fuels/ pii/S0378778818319832. the-energy-sector-in-romania. CENTER for the Study of Democracy. 2018. The opportunity of BANKWATCH Network. 2022b. “The Solution to Romania’s natural gas in the residential sector in Romania. Energy Poverty Is Decentralised, Renewable Energy.” CLANCY, Joy, Fareeha Ummar, Indi- https://bankwatch.org/blog/the-solution-to-romania-s-en- ra Shakya, and Govind Kelkar. 2007. “Appropriate ergy-poverty-is-decentralised-renewable-energy. Gender-Analysis Tools for Unpacking the Gender-En- BATTISTINI, Niccolò, Virginia Di Nino, Maarten Dossche, ergy-Poverty Nexus.” Gender & Development 15 and Aleksandra Kolndrekaj. 2022. “Energy Prices and (2): 241–57. DOI: 10.1080/13552070701391102. Private Consumption: What Are the Channels?” ECB CLODNITCHI, Roxana, and Cristian Busu. 2017. “Ener- Economic Bulletin 3/2022. gy Poverty in Romania—Drivers, Effects and Possible BATTISTINI, Niccolò, Alina Bobasu, and Johannes Gareis. Measures to Reduce its Effects and Number of People 2023. “Who Foots the Bill? The Uneven Impact of the Re- Affected.” Proceedings of the International Conference on cent Energy Price Shock.” ECB Economic Bulletin 2/2023. Business Excellence 11(1): 138–45. https://doi.org/10.1515/ Who foots the bill? The uneven impact of the recent ener- picbe-2017-0015 gy price shock (europa.eu) CMS Law-Now. 2018. “Romania: Liberalisation of the Energy BERTELSMANNSTIFTUNG. 2022. Sustainable Governance Market.” https://cms-lawnow.com/en/ealerts/2018/01/ Indicators. https://www.sgi-network.org/2020/Romania/ romania-liberalisation-of-the-energy-market. Social_Policies. COADY, David, Margaret Grosh, and John Hoddinott. 2004. BOUZAROVSKI, S. 2014. “Energy Poverty in the European Targeting of Transfers in Developing Countries. World Union: Landscapes of Vulnerability.” WIREs Energy and Bank, Washington, DC. https://elibrary.worldbank.org/ Environment 3: 276–89. doi: 10.1002/wene.89. doi/abs/10.1596/0-8213-5769-7. BOUZAROVSKI, S., Harriet Thomson, and Marine Cornelis. CONG, S., D. Nock, Y. L. Qiu, and B. Xing. 2022. “Unveiling 2021. “Confronting Energy Poverty in Europe: A Research Hidden Energy Poverty Using the Energy Equity Gap.” and Policy Agenda.” Energies 14 (4): 858. https://doi. Nature Communications 13: 2456. https://doi.org/10.1038/ org/10.3390/en14040858. s41467-022-30146-5. BRIDGEN, P., and C. Robinson. 2023. “A Decade of Fuel CROON, T. M., J. S. C. M. Hoekstra, M. G. Elsinga, F. Dalla Poverty in England: A Spatio-temporal Analysis of Needs- Longa, and P. Mulder. 2023. “Beyond Headcount Statis- based Targeting of Domestic Energy Efficiency Obli- tics: Exploring the Utility of Energy Poverty Gap Indices gations.” Energy Research & Social Science 101: 103139. in Policy Design.” Energy Policy 177: 113579. https://doi. https://doi.org/10.1016/j.erss.2023.103139. org/10.1016/j.enpol.2023.113579. BMWI. 2022. “What Exactly Is ‘District Heating’? https:// CSEREKLYEI, Zsuzsanna. 2020. “Price and Income Elasticities www.bmwi-energiewende.de/EWD/Redaktion/EN/News- of Residential and Industrial Electricity Demand in the letter/2021/03/Meldung/direkt-account.html. European Union.” Energy Policy 137: 111079. https://doi. BTI. 2022. Romania Country Report 2022. https://bti-project. org/10.1016/j.enpol.2019.111079. org/en/reports/country-report/ROU. DONATIELLO, G., M. D’Orazio, D. Frattarola, A. Rizzi, M. BURKE, P. J., and A. Abayasekara. 2018. “The Price Elasticity Scanu, and M. Spaziani. 2016a. “The Statistical Matching of Electricity Demand in the United States: A Three-Di- of EU-SILC and HBS at ISTAT: Where Do We Stand for mensional Analysis.” The Energy Journal 39 (2): 123–45. the Production of Official Statistics.” In DGINS–Confer- CASTAÑO-ROSA, Raúl, Jaime Solís-Guzmán, Carlos Ru- ence of the Directors General of the National Statistical bio-Bellido, and Madelyn Marrero. 2019. “Towards a Institutes, 26–27. Multiple-Indicator Approach to Energy Poverty in the DONATIELLO, G., M. D’Orazio, M., D. Frattarola, A. Rizzi, European Union: A Review.” Energy and Buildings 193: M. Scanu, and M. Spaziani. 2016b. “The Role of the Con- ditional Independence Assumption in Statistically Match- 114 Romania Energy Poverty Assessment ing Income and Consumption.” Statistical Journal of the GROSSMANN, K., G. Jiglau, U. Dubois, A. Sinea, F. IAOS 32 (4): 667–75. Martín-Consuegra, M. Dereniowska, R. Franke, R. Guyet, D’ORAZIO, Marcell, Marco Di Zo, and Mauro Scanu. 2006. A. Horta, F. Katman, L. Papamikrouli, R. Castaño-Rosa, L. Statistical Matching. Theory and Practice. John Wiley, Sandmann, A. Stojilovska, and A. Varo. 2021. “The Criti- Chichester. http://dx.doi.org/10.1002/0470023554. cal Role of Trust in Experiencing and Coping with Energy DUMITRESCU, Radu. 2022. “Romania Committed 2,6% Poverty: Evidence from across Europe.” Energy Research of GDP to Fighting Energy Crisis.” Romania Insider, & Social Science, 76, 102064. August 16, 2022. https://www.romania-insider.com/ https://www.sciencedirect.com/science/article/pii/ romania-measures-against-energy-price-increase. S2214629621001572?casa_token=2ztikIYTdggAAAAA:Y- ENERGY Poverty Advisory Hub. 2021. Tackling Energy Pov- jADsKo_WTpJ8dJDKtoNcek2OWZcd1vOwUsfde2iOK- erty through Local Actions. Inspiring Cases from across 6MayPWs2t62k1e6C5yTsFFI3fKpVgY66U. Europe. https://energy-poverty.ec.europa.eu/system/ GUAN, Y., J. Yan, Y. Shan, Y. Zhou, Y. Hang, R. Li, Y. Liu, B. files/2022-02/EPAH_inspiring%20cases%20from%20 Liu, O. Nie, B. Bruckner, K. Feng, and K. Hubacek. 2023. across%20Europe_report_EN.pdf “Burden of the Global Energy Price Crisis on Households.” EUROPEAN Commission. 2023. “Energy and the Green Deal.” Nature Energy 8 (3): 304–16. https://commission.europa.eu/strategy-and-policy/prior- GUO, J., L. Du, and C. Wei. 2019. “Equity-Efficiency Trade- ities-2019-2024/european-green-deal/energy-and-green- Off in China’s Energy Capping Policy.” Energy Policy 126: deal_en#:~:text=The%20European%20Green%20Deal%20 57–65. focuses,and%20digitalised%20EU%20energy%20market. HARDING, LaToya. 2022. “Cost of Living: Older Peo- FODOR, Simona. 2022. Survey: Romanians Worry ple Face Biggest Income Squeeze from Surging En- about Inflation, Energy Crisis. Romania Insider, Au- ergy Bills.” Yahoo Finance UK. https://uk.finance. gust 30, 2022. https://www.romania-insider.com/ yahoo.com/news/cost-of-living-older-generations-in- survey-inflation-energy-crisis-ro-aug-2022. come-squeeze-surging-energy-bills-winter-000105252. FREUND, Caroline, and Christine Wallich. 1995. “Raising html?guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2x- Household Energy Prices in Poland: Who Gains? Who lLmNvbS8&guce_referrer_sig=AQAAAKsuZnW3MYp- Loses?” Policy Research Working Paper 1495. World G08NMENfJeX-osHDqgB-klhWugQbWR0xUwl6y- Bank, Washington, DC: CONxDlCLN62NUh-Dgp4D3VKRqfPtlLj_3uXuskSmp- FREUND, C., and C. Wallich. 1997. “Public-Sector Price Re- FzHEjWutQ9yVkMV28JPdcilpuTUDQe4fZbQBSgF0jQF- forms in Transition Economies: Who Gains? Who Loses? nBXdkdsrsEDoqxAEifRZovOwCdBshux4ka4-_56d&gu- The Case of Household Energy Prices in Poland.” Econom- ccounter=2. ic Development and Cultural Change 46 (1): 35–59. HARDY, A., D. Glew, and C. A. Gorse. 2019. “Assessing the FUKUI, S., T. Tanaka, and J. Guo. 2022. “Cross-Border Factors Equity and Effectiveness of the GB Energy Price Caps behind Social Unrest in Africa: An Analysis of Biofuel Pro- Using Smart Meter Data.” Energy Policy 127: 179–85. duction, Oil Price, Speculations, and the US Dollar Index.” https://www.sciencedirect.com/science/article/pii/ GAYE, Amie. “Access to Energy and Human Development.” S0301421518307882. Human Development Report 2007/2008. United Nations HARRIET, T., S. Bouzarovski, and C. Snell. 2017. “Rethinking Development Program Human Development Report Of- the Measurement of Energy Poverty in Europe: A Critical fice Occasional Paper. United Nations. Analysis of Indicators and Data.” Indoor + Built Environ- GESIS. 2022. European Union Statistics on Income and Liv- ment 26: 879–901. ing Conditions (EU-SILC). https://www.gesis.org/gml/ HILLS, John. 2012. Getting the Measure of Fuel Poverty: Final european-microdata/eu-silc. Report of the Fuel Poverty Review. STICERD, The London GONZÁLEZ-EGUINO, M. 2015. “Energy Poverty: An Over- School of Economics and Political Science. https://sticerd. view.” Renewable and Sustainable Energy Reviews 47: lse.ac.uk/dps/case/cr/casereport72.pdf 377–85. https://doi.org/10.1016/j.rser.2015.03.013. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 115 INCHAUSTE, G., and E. Militaru. 2018. “The Distributional MAHLER, Daniel G., Nishant Yonzan, Ruth Hill, Chris- Impact of Taxes and Social Spending in Romania.” World toph Lakner, Haoyu Wu, and Nobuo Yoshida. 2022. Bank Policy Research Working Paper 8565. World Bank, “Pandemic, Prices, and Poverty.” World Bank Blogs. Washington, DC. April 13, 2022. https://blogs.worldbank.org/opendata/ INOUE, N., S. Matsumoto, and K. Mayumi. 2022. “Household pandemic-prices-and-poverty. Energy Consumption Pattern Changes in an Aging Soci- MASUD, Jamil, Diwesh Sharan, and Bindu N. Lohani. 2007. ety: The Case of Japan between 1989 and 2014 in Retro- Energy for All: Addressing the Energy, Environment, and spect.” International Journal of Economic Policy Studies 16: Poverty Nexus in Asia. Asian Development Bank, Manila. 67–83. https://doi.org/10.1007/s42495-021-00069-y MAXIM, Alexandru, Costică Mihai, Constantin-Marius Apos- JOHNSON and Young. 2011. “Towards Best Practices in An- toaie, Cristian Popescu, Costel Istrate, and Ionel Bostan. alyzing Datasets with Missing Data: Comparisons and 2016. “Implications and Measurement of Energy Poverty Recommendations.” Journal of Marriage and Family 73 across the European Union.” Sustainability 8 (5): 483. (5): 926–45. https://doi.org/10.3390/su8050483. JIGLAU, G., A. Sinea, U. Dubois, and P. Biermann. eds. MEINSHAUSEN, N., and P. Bühlmann. 2006. “High-Dimen- 2020. Perspectives on Energy Poverty in Post-Com- sional Graphs and Variable Selection with the Lasso. The munist Europe (1st ed.). Routledge. https://doi. Annals of Statistics 34 (3): 1436–62. org/10.4324/9781003000976. MENYHÉRT, Bálint. 2022. The Effect of Rising Energy and JIGLAU, G., A. Sinea, and C. Murafa, 2018. The opportunity of Consumer Prices on Household Finances, Poverty and So- natural gas in the residential sector in Romania. Research cial Exclusion in the EU. Publications Office of the Europe- reports. Center for the Study of Democracy. an Union, Luxembourg. KIRIKKALELI, D., and I. Darbaz. 2021. “The Causal Linkage MILITARU, Eva, Madalina Popescu, Denisa Vasilescu, and between Energy Price and Food Price.” Energies 14 (14): Amalia Cristescu. 2022. Euromod Country Report Roma- 4182. nia. EUROMOD. KRISHNAMURTHY, C. K. B., and B. Kriström. 2015. “A MILLER, M., and A. Alberini. 2016. “Sensitivity of Price Cross-Country Analysis of Residential Electricity Demand Elasticity of Demand to Aggregation, Unobserved Het- in 11 OECD-Countries.” Resource and Energy Econom- erogeneity, Price Trends, and Price Endogeneity: Evidence ics 39: 68–88. from U.S. Data.” Energy Policy 97: 235–49. https://doi. LAMARCHE, P., F. Oehler, and I. Rioboo. 2020. “European org/10.1016/j.enpol.2016.07.031. Household’s Income, Consumption and Wealth.” Statisti- MIŠÍK, Matúš. 2022. “The EU Needs to Improve its Ex- cal Journal of the IAOS 36 (4): 1175–88. ternal Energy Security.” Energy Policy 165: 112930. LIDDELL, Christine, and Chris Morris. 2010. “Fuel Poverty https://www.sciencedirect.com/science/article/pii/ and Human Health: A Review of Recent Evidence.” Energy S0301421522001550?casa_token=g8MNtUX-Ot- Policy 38(6): 2987–97. wAAAAA:XMW2MJe2MzOVrnqlJT8MmjamAIQRwu- LIDDLE, B., R. Smyth, and X. Zhang. 2020. “Time-Varying WTLLAAmRiCrQhyTpYxPqzGjlN7Wow1qAaCuzG14ov- Income and Price Elasticities for Energy Demand: Evi- 54SU. dence from a Middle-Income Panel.” Energy Economics 86: MOHAMAD Taghvaee, V., M. Nodehi, A. Assari Arani, M. 104681. Rishehri, S. E. Nodehi, and J. Khodaparast Shirazi. 2022. LIHEAP Clearinghouse. 1995. Energy and the Low-Income “Fossil Fuel Price Policy and Sustainability: Energy, Envi- Elderly. https://liheapch.acf.hhs.gov/pubs/520.htm#:~:- ronment, Health and Economy.” International Journal of text=The%20program%20provides%20home%20 Energy Sector Management 17 (2). weatherization,any%20single%20agency%20can%20meet. MORIARITY, C., and F. Scheuren. 2003. “A Note on Rubin’s LUSTIG, Nora C., ed. 2018. Commitment to Equity Handbook: Es- Statistical Matching Using File Concatenation with Ad- timating the Impact of Fiscal Policy on Inequality and Pover- justed Weights and Multiple Imputations.” Journal of Busi- ty. Brookings Institutition Press, Washington, DC. ness & Economic Statistics 21 (1): 65–73. 116 Romania Energy Poverty Assessment MURAFA, C. 2022. Energy Poverty and the Vulnerable Con- World Bank Group, Washington, D.C. http://docu- sumer in Romania: A Curious Case of Policy Schizophre- ments.worldbank.org/curated/en/099051123175082267/ nia. Theoretical and Applied Economics XXIX (4[633]): P18028109bfab800b0a771047dfd6c90089. 57–68. http://store.ectap.ro/articole/1623.pdf. RAMOS, C., M. Alvargonzález, and B. Moreno. 2022. MUKHERJEE, A., Okamura, Y., Gentilini, U., Gencer, D., “Study of Energy Poverty in the European Union: The Almenfi, M., Kryeziu, A., Montenegro, M., & Umapathi, Effect of Distributed Generation.” Energy Sources, N. (2023). Cash transfers in the context of energy subsidy Part B: Economics, Planning, and Policy 17 (1). DOI: reform: Insights from recent experience. Energy Subsidy 10.1080/15567249.2021.2018525. Reform in Action Series. ESMAP Technical Report. World RÄSSLER, S. 2004. “Data Fusion: Identification Problems, Bank. http://hdl.handle.net/10986/39948 Validity, and Multiple Imputation.” Austrian Journal of MURAFA, C., A. Sinea, G. Jiglau, and G. Badescu. 2017. Ener- Statistics 33 (1&2): 153–71. gy Poverty and the Vulnerable Consumer: How Far Are We RÉFABERT, F., Mahapatra, K., Mainali, B., & Hourcade, J.-C. from Europe? European Commission, Brussels, Belgium. (2021). How to boost one-stop-shops for integrated home OLIVERAS, Laura, Andrés Peralta, Laia Palència, Mercè Gots- energy renovation in the EU? Energies Demain, France; ens, María José López, Lucía Artazcoz, Carme Borrell, and Linnaeus University, Sweden; Centre International de Marc Marí-Dell’Olmo. 2021. “Energy Poverty and Health: Recherche sur l’Environnement et le Développement Trends in the European Union Before and During the (CIRED), France. Economic Crisis, 2007–2016.” Health & Place 67: 102294. ROBAYO-ABRIL, Monica, and Rude, Britta. 2024. Conceptu- PAHLE, M., O. Tietjen, S. Osorio, F. Egli, B. Steffen, T. S. alizing and Measuring Energy Poverty in Bulgaria. Policy Schmidt, and O. Edenhofer. 2022. “Safeguarding the Research Working Paper; 10827. © Washington, DC: Energy Transition against Political Backlash to Carbon World Bank. http://hdl.handle.net/10986/41800 Markets.” Nature Energy 7 (3): 290–96. DOI: https://doi. ROBAYO-ABRIL, Monica, Ana Maria Tribin, Eva Military and org/10.1038/s41560-022-00984-0. Caren Grown “Unveiling Gendered Impacts: Analyzing PETROVA, Saska, and Neil Simcock. 2021. “Gen- Fiscal Incidence in Romania”, forthcoming. der and Energy: Domestic Inequities Re- ROSEMAN, Norfleet, Julia Renee, Ivailo Izvorski, Michael considered.” Social & Cultural Geography 22 M. Lokshin, Dorothe Singer, and Iván Torre. 2023. Weak (6): 849–67. DOI: 10.1080/14649365.2019.1645200 Growth, High Inflation, and a Cost-of-Living Crisis. Europe PHILIBERT, Cédric. 2009. “Assessing the Value of Price Caps and Central Asia Economic Update. Office of the Chief and Floors.” Climate Policy 9: 612–33. https://www.osti. Economist. Spring 2023. World Bank, Washington, DC. gov/etdeweb/servlets/purl/21589313. http://hdl.handle.net/10986/39617. POLIMENI, John M., Mihaela Simionescu, and Raluca I. Ior- REUTERS. 2022. Romania’s Government to Extend Energy gulescu. 2022. “Energy Poverty and Personal Health in the Support Bill, Tax Entire Energy Chain. Reuters, Septem- EU.” International Journal of Environmental Research and ber 1, 2022. https://www.reuters.com/business/energy/ Public Health 19 (18): 11459. romanias-government-extend-energy-support-bill-tax-en- POPKOSTOVA, Yana. 2022. “Europe’s Energy Crisis Conun- tire-energy-chain-2022-09-01/. drum.” European Union Institute for Security Studies. RIKSBANK. 2022. “High Energy Prices—How Will Other https://www.iss.europa.eu/sites/default/files/EUISSFiles/ Consumer Prices Be Affected?” https://www.riksbank. Brief_2_Energy%20Crisis.pdf. se/globalassets/media/rapporter/ppr/fordjupningar/ PRASAD, M., C. M. Wheeler, E. Timmis, A. S. Dospinescu, engelska/2022/220210/high-energy-prices--how-will-oth- E. A. S. Tawfik, M. R. Abril, N. T. V. Nguyen, R. C. Ba- er-consumer-prices-be-affected-article-in-monetary-poli- diani-Magnusson, M. Laco, F. U. Ruch, L. A. Delgado cy-report-february-2022.pdf. Prieto, and B. L. Rude. 2023. “EU Regular Economic ROMERO, José Carlos, Pedro Linares, and Xiral López. 2018. Report Nine: Energizing Europe - Part 1: Inclusive “The Policy Implications of Energy Poverty Indicators.” Growth - Inflation Chipping Away Income Gains.” Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 117 Energy Policy 115: 98–108. https://www.sciencedirect. SCHNELLER, Andreas, Jakob Hoffmann, and Anca Sinea. com/science/article/pii/S0301421517308789. 2021. Alleviating Energy Poverty in Romania and Beyond. ROTARU, D. V. 2013. “Romanian Energy Market Evolution: Strategy Proposals for the Local and National Level. Adel- Towards a Common European Market.” CES Working phi, Berlin. Papers 5 (4): 577–87. SCHULTE, I., and P. Heindl. 2017. “Price and Income Elastic- RUBIN, D. B. 1986. “Statistical Matching Using File Concate- ities of Residential Energy Demand in Germany.” Energy nation with Adjusted Weights and Multiple Imputations.” Policy 102: 512–28. Journal of Business & Economic Statistics 4 (1): 87–94. SGARAVATTI, G., S. Tagliapietra, C. Tra- RUDE, Britta, and Monica Robayo-Abril Forthcoming b. “Sta- si, and G. Zachmann. 2021. “National Poli- tistically Matching Income and Consumption Data: The cies to Shield Consumers from Rising Energy Case of Romania.” World Bank, Washington, DC. Prices.” Bruegel.org. https://www.bruegel.org/dataset/ SAGAR, A. D. 2005. “Alleviating Energy Poverty for the national-policies-shield-consumers-rising-energy-prices. World’s Poor.” Energy Policy 33 (11): 1367–72. SINEA, A., C. Murafa, and G. Jiglău. 2018. Energy Poverty and SANDU, D. D., Corad, B., Iamandi-Cioinaru, C., Man, T.- the Vulnerable Consumer in Romania and Europe. Presa C., Marin, M., Moldovan, C., Neculau, G., Tesliuc, E. Universitară Clujeană, Cluj-Napoca, Romania. D., Grigoras, V. A., & Stanculescu, M. S. (2016). Atlas SINEA, A., and G. Jiglău. 2019. “Dealing with Energy Poverty marginalized rural areas and local human develop- from the Bottom Up. Local Approaches to a Pan-Europe- ment in Romania (English). Washington, D.C.: World an Issue.” In Sustainable Development and Resilience of Bank Group. http://documents.worldbank.org/curated/ Local Communities and Public Sector Organizations, ed- en/847151467202306122/Atlas-marginalized-rural-ar- ited by C. Haruta, C. Hintea, and O. Moldovan, 584–602. eas-and-local-human-development-in-Romania Accent, Cluj-Napoca, Romania. SAREEN, Siddarth, Harriet Thomson, Sergio Tirado Herrero, SINEA, Anca. 2020. “Romania—Consumers’ Access to Afford- João Pedro Gouveia, Ingmar Lippert, and Aleksandra able Energy Sources.” EP-pedia, ENGAGER COST Action. Lis. 2020. “European Energy Poverty Metrics: Scales, SINEA, A., A. Vornicu, and G. Jiglau. 2021. Energy Poverty Prospects and Limits.” Global Transitions 2: 26–36. in Buildings in Romania. Center for the Study of De- https://www.sciencedirect.com/science/article/pii/ mocracy. https://www.democracycenter.ro/application/ S2589791820300037. files/7916/2686/2125/Energy_poverty_buildings_-_re- SAUNDERS, M. 2023. “How Indirect Effects of Energy Costs port.pdf. Are Adding to Inflation.” Research Briefing. Oxford SKOUFIAS, Emmanuel, and David P. Coady. 2007. “Are the Economics. https://resources.oxfordeconomics.com/ Welfare Losses from Imperfect Targeting Important?” hubfs/press-release/Energy-Prices-Effects-On-Inflation. Economica 74 (296): 756–76. https://onlinelibrary.wiley. pdf?utm_campaign=PR-Global&utm_medium=email&_ com/doi/full/10.1111/j.1468-0335.2006.00567.x hsmi=243182546&_hsenc=p2ANqtz--nj9HuUQ6wUpF- SOVACOOL, B. 2012. “The Political Economy of Energy Pov- 7zLCV-Phfw288Rgj_V_1AeKcXySNG9DzMV4odKEQY- erty: A Review of Key Challenges.” Energy for Sustainable SwfXZ3lkPoV1xspR-oDhKoff1w33tpNwBP3oOw&utm_ Development 16 (3): 272–82. https://doi.org/10.1016/j. content=243182546&utm_source=hs_email esd.2012.05.006. SCHALLER, J. 2021. “Datenfusion von EU-SILC und House- SURANOVIC, S. 2013. “Fossil Fuel Addiction and the Impli- hold Budget Survey–ein Vergleich zweier Fusionsmetho- cations for Climate Change Policy.” Global Environmen- den.” WISTA–Wirtschaft und Statistik 73 (4): 76–86. tal Change 23 (3): 598–608. https://doi.org/10.1016/j. SCHNEIDER, Gregory L. 2022. “The 1973 Oil Cri- gloenvcha.2013.02.006. sis and Its Economic Consequences.” Bill of Rights STATACORP. 2021. Stata: Release 17. Statistical Software. Institute. https://billofrightsinstitute.org/essays/ StataCorp LLC, College Station, TX. the-1973-oil-crisis-and-its-economic-consequences. TAGHVAEE, Vahid Mohamad, Abbas Assari Arani, Susanne Soretz, and Lotfali Agheli. 2022. “Comparing Energy Effi- 118 Romania Energy Poverty Assessment ciency and Price Policy from a Sustainable Development Emissions through energy poverty alleviation in Romania” Perspective: Using Fossil Fuel Demand Elasticities in (EnPoweR), implemented by Babeș-Bolyai University, in Iran.” MRS Energy & Sustainability 9: 480–93. https://link. partnership with adelphi GMBH (Berlin) and Cluj-Na- springer.com/article/10.1557/s43581-022-00024-0 poca municipality. Available at https://energy-poverty.eu/ TESCHNER, N., Sinea, A., Vornicu, A., Abu-Hamed, T., & system/files/documents/enpower_scientific_report.pdf. Negev, M. (2020). Extreme energy poverty in the urban UNITED Kingdom, Office for National Statistics (UKONS). peripheries of Romania and Israel: Policy, planning, and 2022. Energy Prices and Their Effect on households. Link: infrastructure. Energy Research & Social Science, 66, Energy prices and their effect on households - Office for 101502. https://doi.org/10.1016/j.erss.2020.101502 National Statistics (ons.gov.uk) THEMA, J., and F. Vondung. 2020. “Expenditure-Based In- VORNICU-CHIRA, A., Sinea, A., Lese, M., & Jiglău, G. (2024). dicators of Energy Poverty—An Analysis of Income and Country report: Identifying the vulnerabilities to the ETS Expenditure Elasticities.” Energies 14 (1): 8. 2 in Romania. Policy recommendations for the adoption THOMSON, H., N. Simcock, S. Bouzarovski, and S. Petrova. of the National Social Climate Plans. Center for the Study 2019. “Energy Poverty and Indoor Cooling: An Over- of Democracy. http://www.democracycenter.ro looked Issue in Europe.” Energy and Buildings 196: 21–29. WADUD, Z., R. Noland, and D. J. Graham. 2009. “A Cointe- https://doi.org/10.1016/j.enbuild.2019.05.014. gration Analysis of Gasoline Demand in the United THOMSON, H., S. Bouzarovski, and C. Snell. 2017. States.” Applied Economics 41 (26): 3327–36. https://doi. “Rethinking the Measurement of Energy Pover- org/10.1080/00036840701477306. ty in Europe: A Critical Analysis of Indicators and WORLD Bank. 2009. “Impact of Electricity Tariff Reform on Data.” Indoor and Built Environment 26 (7): 879–901. Households in the Western Balkans.” (mimeo). doi:10.1177/1420326X17699260 WORLD Bank. 2012. Europe and Central Asia Balancing Act: TODOSIJEVIĆ, B. 2012. “Transfer of Variables between Dif- Cutting Subsidies, Protecting Affordability, and Investing ferent Data Sets, or Taking ‘Previous Research’ Seriously.” in the Energy Sector in Eastern Europe and Central Asia. Bulletin of Sociological Methodology/Bulletin de méthodolo- World Bank Group, Washington, DC. gie sociologique 113 (1): 20–39. WORLD Bank. 2015. “Improvement of Power Tariffs and Ad- TYNKKYNEN, Oras, Tuuli Hietaniemi, Outi Haanperä, and dressing Social Impacts.” ESMAP. World Bank Group Hanna Hakko. 2022. “On the Brink of an Energy Crisis – WORLD Bank. 2019. Poverty and Distributional Analysis of What Can We Learn from the Past?” Working Paper. Sitra. Electricity Poverty and Protection of Vulnerable Customers https://www.sitra.fi/app/uploads/2022/04/sitra-on-the- in Kosovo. World Bank, Washington, DC. brink-of-an-energy-crisis.pdf. WORLD Bank. 2022. “Energy Crisis: Protecting Economies VELODY, Mark. 2003. A Regional Review of Social Safety and Enhancing Energy Security in Europe and Central Net Approaches in Support of Energy Sector Reform: A Asia.” Policy Note. World Bank,  Washington, DC. http:// Synthesis Report. United States Agency for International hdl.handle.net/10986/38101  Development. WORLD Bank. 2023a. Romania—Systematic Country Diagnos- UCLA: Statistical Consulting Group. 2022. Introduc- tic Update. World Bank Group, Washington, DC. https:// tion to SAS. https://stats.oarc.ucla.edu/sas/modules/ documentsinternal.worldbank.org/search/34020204. introduction-to-the-features-of-sas/ WORLD Bank. 2023b. Europe and Central Asia: Toward ULLAH, S., M. Khan, and S. M. Yoon. 2021. “Measuring Ener- a Framework for the Sustainable Heating Transi- gy Poverty and Its Impact on Economic Growth in Paki- tion. World Bank Group, Washington, DC. https:// stan.” Sustainability 13 (19): 10969. documents.worldbank.org/en/publication/docu- BABEȘ-BOLYAI University (2021). Research report on the ments-reports/documentdetail/099092023140527206/ extent, drivers, and symptoms of energy poverty in p1777440fed3230ce089060ff8ce59c9f5e. Cluj-Napoca. As part of the project “Mitigating GHG Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 119 Annex 1 Data Household Budget Survey the concept of the so-called household’s reference person, who is the person most representative of the respective household. For this study, we rely on the Household Budget Surveys Additionally, in the case of Romania, households are asked to (HBS). The household budget survey is a national survey mea- maintain detailed diaries on expenditure for a reference period suring households’ expenditure on goods and services. The of one month.2 The income period is also gathered for the ref- survey also provides information on income and some demo- erence month. graphic and socioeconomic characteristics. EU Member States Importantly, variables from the EU-SILC and the HBS are implement these surveys independently, which has resulted in only partly comparable. The lack of comparability mainly lies limited comparability across countries in the European Union. in the different purposes of the surveys. The EU-SILC is the The survey is conducted based on a gentlemen’s agreement, main source for income and poverty measurement in the EU, with frequency, timing, content, or structure varying by coun- while the HBS focuses on expenditure. Consequently, income try. We use scientific use files, which consist of variables about categories in the HBS might be less complete and refer to differ- the household as well as about the household members. We use ent reference periods. In particular, income, which is subject to information published on the 2016 HBS by the International seasonality, might not be captured well in the HBS. Household Survey Network (IHSN) to label the data. 1 We rely on information gathered under the housing cat- The HBS follows concepts that differ from traditional egory of the survey to measure expenditure for energy. The household surveys. The unit of analysis is the household, which housing category includes information on water, electricity, gas, in this case is a social unit living in the same dwelling and shar- and other fuels. Below, we depict the part of the questionnaire ing expenditure with each other. Moreover, the surveys follow that gathers information on these indicators. 1 https://catalog.ihsn.org/catalog/7357/data-dictionary/F109?file_name=sve 2 https://ec.europa.eu/eurostat/documents/54431/1966394/HBS_EU_QualityReport_2015.pdf/72d7e310-c415-7806-93cc-e3bc7a49b596. 122 Romania Energy Poverty Assessment Figure A .1 .1 Hous eh old B udget Survey Q u e s t io nna ire fo r Ro m a nia Source: 2022 HBS Questionnaire for Romania. Information on the payment of services forms part of Section 6 of the questionnaire within the category of “rent and housing maintenance.” The HBS collects data on several types of energy expenditures: Figure A .1 .2 Hous eh old B udget Survey Q u e s t io nna ire fo r Ro m a nia Source: 2021 HBS Questionnaire for Romania. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 123 European Union Statistics on Income • Unemployment benefits (contributory and and Living Conditions (EU-SILC) means-tested) • Unemployment benefits (contributory and non-means-tested) We rely on data gathered as part of the EU-SILC survey, which • Unemployment benefits (noncontributory and collects information on income and living conditions of the means-tested) full population (above the age of 16) of all Member States • Unemployment benefits (noncontributory and in the EU. The information gathered can be divided into four non-means-tested) broad areas: information, income, social exclusion, and housing. • Old-age benefits (contributory and means-tested) The survey collects information at the individual level, such as • Old-age benefits (contributory and non-means-tested) each individual’s demographic information, education, labor • Old-age benefits (noncontributory and means-tested) information, health, and income. The survey is conducted an- • Old-age benefits (noncontributory and nually and was launched in 2004 in 13 Member States and was non-means-tested) extended to all Member States from 2005 onward (GESIS 2022). • Survivor benefits (contributory and means-tested) The data are collected by national statistical institutes, and has a • Survivor benefits (contributory and cross-sectional and a longitudinal component. Some countries non-means-tested) combine administrative and register data. There are two differ- • Survivor benefits (noncontributory and means-tested) ent registers: a household register and a personal register. • Survivor benefits (noncontributory and Importantly, the data on income refer to the year pre- non-means-tested) ceding the date of the survey. The survey’s income measure • Sickness benefits (contributory and means-tested) follows international standards. The income reference year is • Sickness benefits (contributory and a 12-month period, which, depending on the country, might non-means-tested) be a fixed 12-month period (such as the previous calendar or • Sickness benefits (noncontributory and means-tested) tax year) or a moving 12-month period (such as the 12 months • Sickness benefits (noncontributory and preceding the interview).3 non-means-tested) The EU-SILC also contains detailed information on social • Disability benefits (contributory and means-tested) protection benefits in Romania. The following social protec- • Disability benefits (contributory and tion schemes form part of the EU-SILC: non-means-tested) • Disability benefits (noncontributory and • Employers social insurance contribution means-tested) • Optional employer social insurance contributions • Disability benefits (noncontributory and • Contributions to individual private pension plans non-means-tested) • Unemployment benefits (noncontributory and • Education-related allowances (contributory and non-means-tested) means-tested) • Pension from individual private plans (gross) • Education-related allowances (contributory and • Unemployment benefits (gross) non-means-tested) • Old-age benefits (gross) • Education-related allowances (noncontributory and • Survivor benefits (gross) means-tested) • Sickness benefits (gross) • Education-related allowances (noncontributory and • Disability benefits (gross) non-means-tested) • Education-related allowances (gross) 3 https://ec.europa.eu/eurostat/documents/203647/203704/DOC65.pdf/434b2180-33b3-0d8c-ed1e-2da912d6a685?t=1655461990699. 124 Romania Energy Poverty Assessment World Bank Rapid Household Surveys were selected primarily based their use of traditional heating technologies (for example, inefficient coal or wood boilers or stoves) or being connected to district heating in poorly insulated The World Bank developed a high-frequency monitoring sys- buildings. The FGDs and IDIs centered around five key themes: tem to monitor the impact of COVID-19 on people’s well-being attitudes toward sustainable heating practices (and knowledge in four EU countries (Romania, Bulgaria, Poland, and Croatia) of energy efficiency measures), the current status of heating in collaboration with survey firms in the region. Nine survey and energy efficiency in the home, knowledge of subsidy and rounds were designed and implemented in Romania between support programs to improve sustainable heating practices and 2020 and 2022. In 2023, the World Bank conducted another sur- vey round to assess the impact of Russia’s invasion of Ukraine energy efficiency, preferred channels of information about gov- and supply chain disruptions on households’ energy affordabil- ernment support programs, and intentions to upgrade heating ity and vulnerability. This quantitative household survey was technologies or practices. conducted in June and July 2023 using high-frequency com- puter assisted telephone interviews (CATI) and covered 1,500 Qualitative discussion guide households, aiming to monitor the effects of the crisis on energy affordability and vulnerability. The questionnaire was designed GROUP DISCUSSIONS GUIDE – to be comparable with the other rapid surveys conducted in the HOMEOWNERS IN ROMANIA region to allow for cross-country comparisons. The sample used Qualitative fieldwork is representative at the national, urban, and rural levels and the survey weights were based on the 2021 Population Census. Focus group discussions Qualitative Data Collection A. Introduction and warmup 10 min Qualitative data collection taking the form of focus group dis- cussions (FGDs) and select in-depth interviews (IDIs) with fo- • Hello and welcome . . . My name is [name of facilitator], cus group discussion participants was conducted in July 2023 and today I will spend some time with you. Today I and included homeowners living in both single-family homes want to talk to you about how you heat your homes and and multiapartment buildings. A total of 64 individuals partic- use different energy sources more generally, how much ipated in the qualitative fieldwork (7 focus group discussions money you spend on heating/electricity, how satisfied were conducted in total, 1 with a Roma community) across 7 you are with the warmth of your home and what you locations in the counties of Mehedinți (the historical region of plan to do about it in the future. Oltenia) and Vrancea (the historical region of Moldova). • We are here to hear your attitudes, so I invite you to The objective of the qualitative fieldwork was to complement openly and honestly exchange thoughts and ideas, re- the quantitative survey by exploring themes surrounding sus- specting different views. There are no right or wrong tainable energy transitions more deeply as well as capturing the answers; everything you have to tell us is equally im- views of harder-to-reach populations. The approach, as a result, portant to us, I would ask you to speak one by one so was to focus qualitative fieldwork mainly in rural areas and on that we can listen to each other and possibly comment those with lower socioeconomic status (lower levels of income or add something . . . and higher social marginalization). A separate FGD was carried • The information obtained in this way is anonymous out with Roma communities in one of the selected municipali- and confidential. I will ask you to allow me to record ties in light of the increased marginalization of this group. our conversation, because that is the only way for me FGDs comprised 10–15 participants per discussion and to remember everything we talk about, which is very aimed to balance gender (male/female) and age (under 45, 45, important to me. We will use the recording exclusively and older) dimensions of participants. Participants in the FGDs internally, for our analysis and drawing conclusions. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 125 • Please fill in the questionnaire with general questions 5. Retired with disability before we start the conversation, to save time for the 6. Retired due to age discussion (5 minutes, we will help if anyone needs it). Do you live in . . . ? B. Basic information on participants and their heating practices 1. Separate house 2. Flat or apartment Gender [Potentially to be provided directly by surveying 3. Other company] Is it just your family or more families living in your house/ 1. Male apartment? 2. Female 3. Other/prefer not to say 1. Just my family 2. More families Home location [Potentially to be provided directly by sur- veying company] Are you the sole owner of the house/apartment or co-owner? 1. Urban area 1. I am the sole owner of the house/apartment 2. Rural area 2. I am a co-owner of the house/apartment 3. I am not the owner of the house/apartment Age What is the field area of the house/apartment where you live? 1. Under 25 years It is ______ m2 2. From 25 to 35 years What are the main construction materials used for building 3. From 36 to 45 years your house/apartment? 4. from 46 to 55 years 5. from 56 to 60 years 1. Brick and block 6. above 60 years 2. Concrete 3. Wood Education 4. Prefabricated materials 5. Plywood 1. Completed primary school 6. Other. What?________________________________ 2. Completed secondary school 3. Completed college When was the house/apartment built? Enter the year or age 4. Completed university degree of the house/apartment _______? Are you? When (if at all) to your knowledge was your home last ren- ovated (e.g., major improvements to structure of home 1. Employed in a public company like walls, roofs, etc.)? Enter the number of years since last 2. Employed in a private company—small, medium or renovation_______? large How many rooms does the house/apartment have? Enter 3. Self-employed number of rooms______? 4. Unemployed 126 Romania Energy Poverty Assessment Do you heat your home with . . . ? What about decisions regarding investing / buying heating items? 1. Room stove 2. Centralized boiler in your home (heat from an individ- 1. You personally ually controlled boiler connected to radiators through- 2. Your spouse out home) 3. Both of you decide together 3. Centralized boiler in your building (heat from a build- 4. Other relatives in the household ing controlled boiler connected to radiators in every 5. Common agreement unit of the building) 4. Electric heaters (oil radiators, heaters, etc.) How much money do you spend on heating per year? (Open 5. District heating (city/public heating system) answer) 6. Some other way. What?________________________ How much of this is maintenance of the heating appliance and how much is the fuel? (Open answer) What type of fuel do you primarily use to heat your home? (mark all that apply) How much money do you think you would spend if you in- sulated your house or apartment or changed the carpentry? 1. Wood (Open answer) 2. Coal Can you control the heat generated by this stove or boiler 3. Pellet automatically or just manually by adding more fuel? 4. Electricity 5. Natural or LPG gas 1. It can be controlled automatically (e.g., with a 6. Oil and oil derivatives thermostat) 7. Other solid fuels (e.g., trash, mixed paper, old tires, etc.) 2. It can be controlled only manually (e.g., by adding or 8. District heating (city/public heating) removing fuel) Do you ever burn anything else to supplement this fuel (for Where in the house is the boiler or stove located? example, paper, waste, clothing, etc.)? 1. In one room 1. Yes 2. In the hallway 2. No 3. In the boiler room 4. In the garage Who in your families take care of heating during the winter? 5. In the kitchen Is it just you, other family members, both? Or homeowner? 6. Somewhere else. Where? _______________________ (prompt: who turns on heating, adds fuel and extinguishes) Are you heating the entire home or just some parts? 1. You personally 2. Your spouse 1. I heat entire home 3. Both of you decide together 2. I heat most parts of the home 4. Other relatives in the household 3. I heat only some parts of the home 5. Common agreement Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 127 Do you own any other heating appliances for your house/ that focus on using less energy and preserving the value of en- apartment? ergy used). Thinking about what you know about modern heating sys- 1. Yes tems (for example, the introduction of heat pumps, gas boilers 2. No or eco-design boilers or stoves and insulation), what are your opinions and concerns regarding these technologies? Are they Do you have access at home to gas for heating? worth the investment? How would you say a modern heating technology would 1. Yes influence the time spent on heating up your house? And how 2. No would it affect you financially? For those of you who have improved some aspect of heating C. DISCUSSION your home (e.g. with modern boilers or stoves, or insulation), what changes have you noticed? Current winter-time heating practice? To your knowledge, how many people in your community How many of you changed the heating technology that was al- have upgraded some aspects of their heating technology to more ready in place in your house (when you moved in or once you modern processes? Do you perceive a trend in your community began to make decisions about heating)? What did you install towards this type of improvements, and why? instead? Why? Based on what you know, what changes did you notice in the What are the main aspects or measures you take into ac- homes or lives of these people in your community? count regarding heating and insulation in your home. [Note for facilitator: ask their opinions in relation to costs, quality, tech- Upgrade decision nology and comfort if not mentioned] During the next 12 months, what do you plan on doing regard- What are your main concerns with respect to access to dif- ing the heating and/or insulation in your house / apartment, If ferent fuel types/heating materials? nothing/anything—what are the main reasons for this? Based on the info you’ve offered, you heat your home with. . [If the answer is yes]: To which heat source or fuel type . . (In case this is different from gas or electricity) Where do you would you like to upgrade if you had the chance to do so? get this (wood/coal, other) from? What would you use if this What kind of thermal modernization are you thinking source would become unavailable? about? What specifically would you improve regarding the carpen- What motivated you to think about these changes? Were try in your home if you’d have the chance? these local smog problems, climate change? To improve thermal What would you consider to be a favorable temperature in comfort in your home? Easier and cleaner operation of a mod- the house, and why? What are the thinking and priorities influ- ern heating source compared to a traditional solid fuel boiler? encing how you decide to set the temperature in your house, if Recommendations from friends and family who did it (what you can? types of recommendations)? Want to modernize your home? If you were to compare different priorities in your home (e.g. Increasing the value of the house? Reducing heating costs? food, appliances, health and education of householders, vehicles, From whom do you get your information or ask for advice etc.), how does heating your home in winter and cooling it on when you need to make a decision about the heating system of summer compare to others? your home (Remind with examples if the answers are minimal: Attitudes about modern heating systems, energy efficiency, friends, family, contractors, local authorities?)? and air quality What other sources of information have you investigated? What would motivate you to move forward with the heat- We would like to hear from you what you think about modern ing upgrade next year? What factors would make you hesi- heating systems and energy efficiency in terms of maintaining tate about it (e.g., cost of heating appliance? Cost of fuel and the heat of your home (by energy efficiency we mean practices maintenance?)? 128 Romania Energy Poverty Assessment If you’re not thinking about upgrading your heating technol- What are the main aspects you consider when deciding to ogy or insulating your building, why not? heat your home? (prompt with quality of heating, cost of heating If given the chance to partially finance a heating or insu- appliances, cost of fuel and maintenance, air pollution in your lation upgrade through a government program, would you be city, potential health implications for you and your family, etc.) willing to accept such a program? Why, why not? What do you usually use to heat your home during the winter? Knowledge and interactions with support programs Do you use anything else besides that? (prompt for burning Are you aware of programs that support upgrading of heat sys- paper, rubber, garbage, clothes, etc.) tems to clean, energy efficient ones (modern boilers, stoves, and If you were to compare different priorities in your home (e.g. heat pumps) or retrofitting of homes with insulation to improve food, appliances, health and education of householders, vehicles, heat retention? etc.), where is the heating of your home positioned compared (If familiar with these initiatives) How did you hear about to the others? them? Was it through family/friends/neighbors? Installers, con- Attitudes about modern heating systems, energy efficiency, tractors or retailers? Chimney sweeps? Government officials? and air quality Did you see any communications on TV, in a newspaper/bill- board/leaflet/website or hear about the initiatives on the radio? I would like to hear from you what you think about modern When did you hear about this? Has the communication of these heating systems and energy efficiency in terms of maintaining initiatives changed in the past year? the heat of your home (by energy efficiency we mean practic- (If not familiar with these initiatives) Where would you like es that focus on using less energy and preserving the value of to hear about them/ Through which communication channels energy used) would you like to be informed? Thinking about what you know about modern heating sys- Have you looked into the details of these initiatives? Are tems (for example, the introduction of heat pumps, gas boilers you aware of how much financial support you could receive? or eco-design boilers or stoves and insulation), what are your Did you look into the application process? Are you considering opinions and concerns regarding these technologies? Are they applying to these initiatives? What are the main reasons for or worth the investment?How would you say a modern heating against it? What are the main aspects of the application process technology would influence the time spent on heating up your that could be improved? How? house? And how would it affect you financially? What effects How much would you say you trust the government agencies would it have on the health of the household in your opinion? responsible for providing this type of support? For example, the In case you’ve made any upgrades regarding the heating in Ministry of Work and Social Solidarity, the county council/local your home, what changes have you noticed? authorities? To your knowledge, how many people in your community Are you aware of any loans from commercial banks that can have upgraded some aspects of their heating technology to more be used to help you finance investments in your home’s heating modern processes? Do you perceive a trend in your community technologies? Could you offer a few examples? Would you trust towards this type of improvements, and why? commercial banks enough to take out a loan for this purpose? Based on what you know, what changes did you notice in the Would you be able to afford such loans from commercial banks? homes or lives of these people in your community? In-depth interviews Upgrade decision During the next 12 months, what do you plan on doing regard- Current winter-time heating practice? ing the heating and/or insulation in your house / apartment, If What are the main reasons for which you chose to heat your nothing/anything—what are the main reasons for this? [If the home the way you currently do? answer is yes]: To which heat source or fuel type would you like Did that way of heating come with the place or did you opt to upgrade if you had the chance to do so? What kind of thermal for it? modernization are you thinking about? Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 129 What motivated you to think about these changes? Were heat pumps) or retrofitting of homes with insulation to improve these local smog problems, climate change? To improve thermal heat retention? comfort in your home? Easier and cleaner operation of a mod- (If familiar with these initiatives) How did you hear about ern heating source compared to a traditional solid fuel boiler? them? Was it through family/friends/neighbors? Installers, con- Recommendations from friends and family who did it (what tractors or retailers? Chimney sweeps? Government officials? types of recommendations)? Want to modernize your home? Did you see any communications on TV, in a newspaper/bill- Increasing the value of the house? Reducing heating costs? board/leaflet/website or hear about the initiatives on the radio? From whom do you get your information or ask for advice (If not familiar with these initiatives) Where would you like when you need to make a decision about the heating system of to hear about them/ Through which communication channels your home (Remind with examples if the answers are minimal: would you like to be informed? friends, family, contractors, local authorities?)? Have you looked into the details of these initiatives? Are you What other sources of information have you investigated? aware of how much financial support you could receive? Did What would motivate you to move forward with the heat- you look into the application process? Are you considering ap- ing upgrade next year? What factors would make you hesi- plying to these initiatives? Why or why not? If not, is there some tate about it (e.g., cost of heating appliance? Cost of fuel and aspect of the application process that could be improved? How? maintenance?)? How much would you say you trust the government agencies If you’re not thinking about upgrading your heating technol- responsible for providing this type of support? For example, the ogy or insulating your building, why not? Ministry of Work and Social Solidarity, the county council/local If given the chance to partially finance a heating or insu- authorities? lation upgrade through a government program, would you be Are you aware of any loans from commercial banks that willing to accept such a program? Why/why not? can be used to help you finance investments in your home’s heating technologies? Could you offer a few examples? Would Knowledge and interactions with support programs you trust commercial banks enough to take out a loan for this Are you aware of programs that support upgrading of heat sys- purpose? Would you be able to afford such loans from com- tems to clean, energy efficient ones (modern boilers, stoves, and mercial banks? 130 Romania Energy Poverty Assessment Annex 2 The Method for Measuring Energy Expenditure and Energy Poverty in Romania Matters Main Findings Different Approaches to Measuring Energy Expenditure and Welfare Here we examine diverse methodologies for assessing energy expenditure shares and energy poverty in Romania, utilizing Analyses conducted on people’s vulnerability to energy prices data from the 2021 HBS. Notably, substantial variations are often rely on estimating households’ energy spending shares. observed in both energy expenditure shares and energy poverty These shares can be informative, because larger shares might rates based on the chosen methodology. These discrepancies are indicate that households are more vulnerable to rising energy attributed to considerations in the numerator (the method of prices. Researchers and policy makers typically rely on these measuring household energy expenditure) and the denomina- estimates to better understand their population’s need for miti- tor (the method of measuring household welfare). The inclu- gation measures or social protection mechanisms. sion of car-related energy expenditures in the numerator yields Researchers have developed a variety of approaches to results that are somewhat less straightforward. The choice of measure energy expenditure shares. To date, researchers have expressing energy expenditure shares as a ratio over household used various measures of energy expenditure shares (for a full expenditure or household income impacts the outcomes, with overview, see Robayo-Abril and Rude, 2024). While some mea- patterns along the welfare distribution differing between the two sures rely on nonmonetary approaches, others use income and approaches. We conclude that results are generally more coher- expenditure data. However, even within the literature that uses ent when expressing energy expenditure shares as a proportion monetary approaches, significant methodological differences of household income rather than consumption. Moreover, en- exist in the empirical approaches taken, based on two factors. ergy poverty rates exhibit significant variation depending on the First, different measures consider different components when methodology employed, with greater alignment observed when estimating household energy expenditure. For example, some using an income-based method. Consequently, in our analysis might include car-related energy expenditures (for example, gas- we opt to measure energy spending shares and energy poverty oline). In contrast, others might abstract from this component, using income-based measures while excluding car-related ener- because they relate it more to transport than the energy sector. gy spending due to increased stability and intuitive. Especially when analyzing households’ vulnerability to fluctu- Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 131 ations in energy prices, it is crucial to carefully consider which a full overview see Robayo-Abril and Rude, 2024). The first one components to include when measuring absolute energy expen- defines all households as energy poor those who have an ener- diture. Second, the literature to date has used several different gy spending share at least twice as large as the median energy methodologies to set households’ absolute energy expenditure spending share in the population. The second one relies on the in relation to households’ income. While some approaches rely absolute energy expenditure and defines all households who on household income, others rely on household expenditure report absolute energy expenditure below half the national me- (consumption). dian as energy poor. The last measure defines energy poverty These diverging methodologies might result in diverg- as the share of households with an energy expenditure share ing pictures of energy expenditure patterns and households’ above 10 percent. vulnerability to energy prices, which in turn might impact We show that critically reflecting on the underlying meth- policy recommendations. Given that energy expenditure shares odology used to estimate energy expenditure shares and ener- might vary depending on the underlying measure used, it is cru- gy poverty is crucial, given that results differ. We demonstrate cial to base the chosen measure on empirical considerations. that energy expenditure shares and energy poverty rates differ In addition, to better understand the sensitivity of the insights significantly by the underlying methodology used. These differ- generated based on resulting energy expenditure shares to the ences are driven both by the numerator (the method of measur- underlying methodology, it is recommended to compare differ- ing household energy expenditure) and the denominator (the ent empirical approaches. Given that policy recommendations method of measuring household welfare). Expressing energy might differ depending on the resulting estimates, it is crucial to expenditure shares as a ratio over household expenditure or critically reflect on the underlying methodology used. household income affects the results, and patterns along the In the following, we conduct empirical analyses to choose welfare distribution also differ between both approaches. Results the best suitable monetary measure in the context of Roma- are overall more intuitive when expressing energy expenditure nia. To this end, we rely on data from the 2021 HBS and explore shares as a share of household income (not consumption). Last- different approaches to measure energy expenditure as well as ly, energy poverty rates vary largely, depending on the under- household welfare. More concretely, we analyze how estimates lying method. differ across different methodological approaches in the case Based on our empirical analysis, we choose the in- of Romania. We do so by considering different components of come-based measure and abstract from car-related energy households’ overall energy expenditure. In addition, we explore expenditure. We show that estimates along the welfare distri- two different approaches to measure households’ welfare: in- bution are more intuitive when using income-based measures. come and consumption. In addition, energy poverty rates fluctuate less when abstract- As we are ultimately interested in the poorest households’ ing from car-related energy expenditure. Based on these re- vulnerability to energy prices, we investigate different expen- sults, we conclude that the income-based measure that does diture patterns and measures by income and consumption not consider car-related energy expenditure is the most ap- quintiles. We are ultimately interested in the vulnerability of propriate measure for energy expenditure shares and energy poor households to rising energy prices. These households might poverty in Romania. be more in need of support or mitigation mechanisms when en- ergy prices rise. Therefore, in the analysis we investigate different The Method of Measuring measures of energy expenditure shares and energy poverty by Households’ Energy Expenditure consumption and income quintiles. Both consumption and in- Matters come are approaches used to approximate households’ welfare. Moreover, we explore three different approaches to mea- There are several possibilities to approach households’ energy sure energy poverty. While there are many ways to quantify expenditure (consumption). Eurostat, for example, considers energy poverty, we apply three of the most used measures (for the following components when analyzing households’ en- 132 Romania Energy Poverty Assessment ergy consumption: space and water heating, space cooling, line), solid fuels, liquid fuels, natural gas, thermal energy, and cooking, lighting and electrical appliances, and other end-us- electricity and renewables. es, which mainly refer to energy usage outside the dwelling. 1 To better understand the importance of car-related energy Bacon et al. (2010) note that measurements vary by country be- expenditure, we investigate the share of households who own cause they depend on the information collected in household a car by income quintile. Figure A.2.1 shows that only a small budget surveys. Household budget surveys, on the other hand, share of households in the lowest income quintile own a car. At are often not standardized nor harmonized, and how they col- the same time, car-related fuels play, on average, a significant lect information on energy expenditure might vary significantly. role in the overall energy expenditure of the Romanian popula- Bacon et al. (2010) divide energy expenditure into the following tion (figure A.2.2). On average, nearly 30 percent of all energy categories: kerosene, liquified petroleum gas, gasoline, diesel, expenditure by Romanian households is due to car-related en- electricity, natural gas, coal, firewood, and other forms of bio- ergy consumption. Still, given the evidence on car ownership, mass. They also mention the importance of analyzing expendi- it is not surprising that these patterns of results do not hold ture on motorized passenger transport, given that oil price is a across income quintiles. Figure A.2.3 plots the different energy critical component of its cost structure. components by income quintiles. The graph reveals that those We start by screening the codebook of the Romanian households in the lowest income quintile only spend 9.8 percent HBS and search for all energy-relevant expenditures. We of their household income on car-related energy expenditure, identify six expenditure categories relevant to the energy sec- whereas this makes up more than 35 percent of total household tor. These are car-related energy expenditures (such as gaso- energy expenditure in the upper-income quintile. Figure A .2 .1 Sh ares of Hous eh olds wi th a Ca r Figu re A . 2 . 2 Ene rg y E xpe nd it u re by Ene rg y by Incom e Qui n ti le, 2 02 0 S o u rce, 2 021 0.9 100% 0.8 90% Car-related fuels 0.7 80% 70% Solid fuels 0.6 60% 0.5 Liquid fuels 50% 0.4 Natural gases 40% 0.3 30% Thermal energy 0.2 20% Elecricity and 0.1 renewable energy 10% 0 0% Q1 Q2 Q3 Q4 Q5 Source: Figure A.2.1—2020 EU-SILC; Figure A.2.2—2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest. 1 https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Energy_consumption_in_households#:~:text=Energy%20consumption%20 in%20households%20by%20type%20of%20end%2Duse,-In%20the%20EU&text=Main%20cooking%20devices%20require%206.1,final%20 energy%20consumed%20by%20households. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 133 Figure A .2 . 3 Hous eh olds’ En erg y Ex p en d it u re by it s Co m po ne nt s a c ro ss I nco m e Q u int ile s 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Q1 Q2 Q3 Q4 Q5 Electricity and renewable energy Thermal energy Natural gas Liquid fuels Solid fuels Car-related fuels Source: Own estimates based on the 2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest. We next calculate energy expenditure shares as a ratio We find counterintuitive results, with expenditure shares of household expenditure. To analyze energy expenditure being largest in the second-lowest income quintile; car-relat- shares, we next divide the absolute energy expenditure, using ed energy expenditures explain these counterintuitive results. all components depicted in figure A.2.1, by each household’s Contrary from our hypothesis, the poorest do not spend the total expenditure, also denoted as the consumption aggregate. highest share of their household expenditure on energy (figure We then plot the resulting energy expenditure share by in- A.2.4). These results are largely driven by spending patterns come and consumption quintiles. This analysis allows us to on car-related energy expenditure (figure A.2.5). On average, get a better understanding of energy spending patterns along households in the upper income quintiles spend a larger share of the income distribution. Given that the poorest mostly re- their overall household expenditure on this energy component port lower consumption aggregates and that energy is often than do those in the lowest income quintile. These results, which considered a nonsubstitutable good, meaning that house- seem counterintuitive at first, hold when using consumption holds have to use heating, electricity, and cooking devices, quintiles—our alternative welfare measure—instead of income we expect to find higher energy spending shares in the lowest quintiles (figures A.2.6–A.2.7). Bearing in mind the evidence income quintiles. on car ownership revealed in figure A.2.1 explains these results. 134 Romania Energy Poverty Assessment Figure A .2 .4 En erg y Ex p en di ture Sh ares Figu re A . 2 . 5 Ene rg y E xpe nd it u re S ha re s (Expenditure- B as ed) by I n come Qui n ti le s , ( E xpe nd it u re -B as e d ) by Inco m e Q u int ile s 2 021 a nd Co m po ne nt s , 2 021 0.185 0.2 0.18 0.18 0.16 0.14 0.175 0.12 0.1 0.17 0.08 0.06 0.165 0.04 0.02 0.16 0 Q1 Q2 Q3 Q4 Q5 0.155 Electricity and renewable Thermal energy Natural gas Liquid fuels (share) 0.15 Q1 Q2 Q3 Q4 Q5 Solid fuels (share) Car-related fuels Source: Figure A.2.1—Own estimates based on 2020 EU-SILC; Figure A.2.2—Own estimates based on 2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest. Figure A .2 .6 En erg y Ex p en di ture Sh ares Figu re A . 2 . 7 Ene rg y E xpe nd it u re S ha re s (Expenditure- B as ed) by Con s ump ti on ( E xpe nd it u re -B as e d ) by Co ns u m pt io n Quint iles , 2 021 Q u int ile s a nd Co m po ne nt s , 2 021 0.185 0.2 Car-related fuels 0.18 0.18 0.16 Solid fuels 0.14 Liquid fuels 0.175 0.12 0.1 Natural gases 0.17 0.08 (share) 0.06 Thermal energy (share) 0.165 0.04 0.02 Elecricity and renewable energy (share) 0.16 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Source: Own estimates based on 2021 HBS. Note: Energy expenditure shares include car-related expenditure and are in relation to overall household expenditure (consumption). Results are weighted by the respective survey weights. Q1 indicates the poorest income quintile and Q5 the richest. Given that measurements of energy poverty depend di- counterintuitive. Figures A.2.8 and A.2.9 plot energy poverty rectly on energy expenditure shares, it is no surprise that rates across consumption and income quintiles by applying the energy poverty patterns across welfare quintiles are also three different approaches to measure energy poverty outlined Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 135 above. The two relative measures reveal that energy poverty reveal that the share of households who spend less than half rates do not vary much across income (or consumption) quin- the median value of absolute energy expenditure decreases by tiles, which is related to the patterns in energy expenditure income (consumption) quintile. These estimates indicate that shares observed previously. When using the absolute measure richer households have more disposable income and can there- of energy poverty, the results are more intuitive. The figures fore spend more on energy. Figure A .2 .8 En erg y Poverty by I n come Figu re A . 2 .9 Ene rg y Pove r ty by Co ns u m pt io n Quint ile, 2 021 Q u int ile, 2 021 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Energy poverty incidence rate (share) Energy poverty incidence rate (share) Energy poverty incidence rate (absolute) Energy poverty incidence rate (absolute) Energy poverty (> 10 percent) Energy poverty (> 10 percent) Energy poverty (LIHC) Energy poverty (LIHC) Source: Own estimates based on 2021 HBS. Q1 indicates the poorest quintile and Q5 the richest We next explore energy expenditure shares across income Energy expenditure shares along the welfare distribution quintiles when abstracting from car-related energy spending are slightly more intuitive when abstracting from car-relat- based on two reasons. First, car-related expenditure seems less ed energy expenditures. Figures A.2.10 and A.2.11 plot ener- significant for households at the lower end of the welfare dis- gy expenditure shares when not including car-related energy tribution. Given that this is the group in which we are primar- expenditure by income and consumption quintiles. The graphs ily interested, these expenditures might be less relevant for our show that estimates of energy expenditure shares are slightly analysis. At the same time, they might create a distorted picture, more intuitive. Particularly when plotting the average expen- given that high-income households are more severely affected by diture share by consumption quintiles, the results are more rising gasoline prices because of the energy crisis, and we would aligned with findings in the literature of higher expenditure be neglecting this part in our analysis of distributional impacts. shares in the case of poorer households. Nevertheless, in the Second, abstracting from car-related energy spending makes sense case of income quintiles, the second-lowest income quintile when thinking about the political distribution of responsibilities. reports, on average, higher energy expenditure shares than the Car-related spending patterns might fall under the responsibility lowest income quintile (figure A.2.10). of political actors in the transport sector and not the energy sector. 136 Romania Energy Poverty Assessment Figure A .2 .1 0 Average En erg y Ex p en di tu re Figu re A . 2 .11 A ve rage Ene rg y E xpe nd it u re Shares by I n come Qui n ti les , 2 021 S ha re s by Co ns u m pt io n Q u int ile s , 2 021 0.18 0.18 0.16 0.16 0.14 0.14 0.12 0.12 0.1 0.1 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Source: Own estimates based on 2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest Results Using Income-Based Using this alternative measure results in more-intuitive Measures results, with poorer households spending larger shares on energy. Figures A.2.12 and A.2.13 reveal that energy spending We next explore an alternative measure of energy expenditure patterns are significantly different than when using the expen- shares that relies on income-based measures of energy expen- diture-based measure. In contrast to our results using expen- diture shares. Where we previously defined energy expenditure diture-based energy spending shares, there is a clear pattern shares as the ratio of energy expenditure over total household across income and consumption quintiles, with lower quintiles expenditure, we now define it as the ratio of household income. spending higher shares of their overall income on energy. These Household income is another way of approximating a house- findings are more in line with findings from the literature. The hold’s wealth. Using this measure, therefore, also makes sense, figures also reveal that spending patterns are less dispersed when especially for European settings where inequality and poverty using consumption instead of income quintiles to approximate measures often rely on income. the welfare distribution. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 137 Figu re A . 2 .1 3 Ene rg y S pe nd ing S ha re s Figure A .2 .1 2 En erg y Sp en di n g Sh ares ( I nco m e -B as e d ) by Co ns u m pt io n Q u int ile s , (Income-B as ed) by I n come Qui n ti les , 2021 2 021 0.18 0.14 0.16 0.12 0.14 0.1 0.12 0.1 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Source: Own estimates based on 2021 HBS. Note: Energy expenditure shares include car-related expenditure and are in relation to overall household income. Results are weighted by the respective survey weights. Q1 indicates the poorest income quintile and Q5 the richest. Similar to previous observations, the share of car-related quintile spends on average 35 percent of overall energy expendi- energy expenditure is higher in the upper welfare quintiles. ture. The lowest income (or consumption) quintile, on the other Figures A.2.14 and A.2.15 reveal that the higher income quin- hand, spends approximately 2 percent on this energy compo- tiles spend a larger share of their household income on car-re- nent. These patterns of results are in line with previous evidence lated energy expenditure. The highest income (or consumption) from the expenditure-based measure. Figure A .2 .1 4 En erg y Sp en di n g Sh ares Figu re A . 2 .1 5 Ene rg y S pe nd ing S ha re s (Income-B as ed) by I n come Qui n ti les a nd ( I nco m e -B as e d ) by Co ns u m pt io n Q u int ile s Compone n ts , 2 021 a nd Co m po ne nt s , 2 021 0.18 0.14 0.16 0.12 0.14 0.12 0.1 0.1 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Electricity and renewable energy (share) Thermal energy (share) Electricity and renewable energy Thermal energy Natural gas (share) Liquid fuels Natural gas Liquid fuels Solid fuels Car-related fuels Solid fuels Car-related fuels Source: Own estimates based on 2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest. 138 Romania Energy Poverty Assessment Energy poverty patterns across quintiles are more intuitive energy poverty explored previously. The figures demonstrate that in case of the income-based measure, with poorer households in the case of the income-based measure, energy poverty rates are being more affected by energy poverty than richer households. larger in the lower income (consumption) quintiles than in the Figures A.2.16 and A.2.17 plot the average energy poverty rate by upper income (consumption) quintiles. These patterns of results income (and consumption) quintiles, using the three measures of mostly hold for all three measures of energy poverty. Figure A .2 .1 6 En erg y Poverty Rates (I n co m e - Figu re A . 2 .1 7 Ene rg y Pove r ty Ra te s ( I nco m e - Based ) by M eas uremen t an d I n come B as e d ) by Meas u re m e nt a nd Co ns u m pt io n Quint ile, 2 021 Q u int ile, 2 021 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Energy poverty incidence rate (share) Energy poverty incidence rate (share) Energy poverty incidence rate (absolute) Energy poverty incidence rate (absolute) Energy poverty (> 10 percent) Energy poverty (> 10 percent) Source: Own estimates based on 2021 HBS. Note: Q1 indicates the poorest income quintile and Q5 the richest. In line with the analysis that uses expenditure-based car-related energy expenditures. Households at the lower end measures, we next abstract from car-related energy expendi- of the welfare distribution spend higher shares of their overall tures. Figures A.2.18 and A.2.19 show that the overall pattern household income on energy than those at the higher end of of results remains close to the one observed when including the distribution. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 139 Figure A .2 .1 8 En erg y Sp en di n g Sh ares Figu re A . 2 .1 9 Ene rg y S pe nd ing S ha re s (Income-B as ed, No Car) by I n come Qui nt ile s , ( Co ns u m pt io n-B as e d , N o Ca r ) by 2 021 Co ns u m pt io n Q u int ile s , 2 021 0.16 0.14 0.14 0.12 0.12 0.1 0.1 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Source: Own estimates based on 2021 HBS. Note: We abstract from car-related energy expenditures in these cases. Q1 indicates the poorest income quintile and Q5 the richest. Table A .2.1 Reg ress i on of En erg y Ex p en d it u re S ha re s a m o ng H o u s e ho ld s w it h H ea t ing Sys te m s , 2 021 (1) (1) VARIABLES VARIABLES Energy exp. share (income-based, no car) Energy exp. share (income-based, no car) Household with children -0.0104*** -0.00870** Disability benefit recipient (<15) (0.00137) (0.00425) 0.00282* 0.0506*** Household with pensioner Connectivity to electricity (0.00164) (0.00485) 0.0420*** -0.0144*** Single-elderly household Modern heating (0.0142) (0.00209) 0.00743*** 0.0869*** Female-headed household Subsidy (thermal energy) (0.00146) (0.0263) -0.0107*** 0.0955 Household with Roma Subsidy (natural gas) (0.00402) (0.0872) Household with 0.0107*** -0.0797*** Subsidy (elecricity) unemployed (0.00265) (0.0269) -0.00886*** 0.0233** Urban Cooking (electricity) (0.00173) (0.00957) Household with more than 5 -0.0102*** 0.0384*** Cooking (natural gas) members (0.00239) (0.00942) -6.95e-07*** 0.00453** Income per capita Cooking (wood/coal/oil) (2.55e-08) (0.00193) Social aid from municipality 0.00176 0.0286*** Cooking (cylinder) (recipient) (0.00537) (0.00532) 140 Romania Energy Poverty Assessment (1) (1) VARIABLES VARIABLES Energy exp. share (income-based, no car) Energy exp. share (income-based, no car) -0.0349 -0.00565 Cooking (other) Hot water—no hot water (0.0851) (0.0320) Cold water, indoors from -0.00771 Sewage type—public -0.00720 public supply (0.00606) system (0.00509) -0.00246 Sewage type—in-house -0.00241 Indoors, from in-house (0.00620) system (0.00474) Outside residence, but -0.0128 Natural gas (cooking or -0.0162** inside building (0.00831) heating) (0.00818) Outdoors, fountain,pump, 0.00357 0.0540 Constant well (0.00494) (0.0332) -0.0157 Hot water—public system Observations 15,016 (0.0318) Hot water—in-house -0.0110 R-squared 0.163 system (0.0318) Source: Own estimates based on HBS 2021. Note: The table reports correlation coefficients from a simple regression on energy expenditure shares. Energy expenditure shares are income based; abstract from energy spending on car-related energy, such as gasoline; and range from 0 to 1. Different Approaches to Measuring Estimates of energy poverty fluctuate less when using Energy Poverty income-based measures. The energy poverty incidence rate fluctuates less when using income-based measures, namely be- We next compare different measures of energy poverty to tween 19.0 and 35.4 percent across the different energy poverty each other and find that rates vary significantly. Figure A.2.20 measures. In contrast, the rate varies more significantly in the plots the average energy expenditure share and the related en- case of expenditure-based measures. These results could mean ergy poverty rates when using expenditure-based measures, that the income-based measure is more robust across the dif- whereas figure A.2.21 plots these estimates using income-based ferent methodological approaches toward energy poverty. Rates measures. We find that the incidence rate of energy poverty in are even more aligned when abstracting from car-related energy Romania is highly sensitive to the underlying measure. First, expenditure (figure A.2.22). the incidence rates fluctuate significantly depending on the un- derlying measure of energy poverty. Second, the rate also varies depending on the approach used to measure welfare (income or expenditure). Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 141 Figure A .2 .2 0 Average En erg y Ex p en di tu re Figu re A . 2 . 21 A ve rage Ene rg y E xpe nd it u re Shares an d En erg y Poverty Rates Us i n g S ha re s a nd Ene rg y Pove r ty Ra te s Us ing Expenditure- B as ed M eas ures , 2 021 Inco m e -B as e d Meas u re s , 2 021 0.9 0.80 0.8 0.400 0.354 0.7 0.350 0.6 0.300 0.5 0.250 0.190 0.4 0.34 0.200 0.3 0.24 0.150 0.123 0.106 0.17 0.2 0.100 0.064 0.1 0.06 0.050 0 0.000 Share of 2M M/2 P10 LIHC Energy 2M M/2 M/2 LIHC expenditure expenditure in energy share Source: Own estimates based on 2021 HBS. Note: These measures also consider car-related expenditure. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). Figure A .2 .2 2 Average En erg y Ex p en di tu re S ha re s a nd Ene rg y Pove r ty Ra te s Us ing I nco m e -B as e d Measures an d No Car- Related En erg y E xpe nd it u re s , 2 021 0.30 0.261 0.243 0.25 0.20 0.15 0.123 0.10 0.087 0.057 0.05 0.00 Energy spending M/2 2M P10 LIHC share Source: Own estimates based on 2021 HBS. Note: These measures do not consider car-related expenditure. We use four different measures of energy poverty: (1) the rate of households spending more than twice the median value of energy expenditure shares (2M), (2) the rate of households spending less than half the median value of absolute energy spending (M/2), (3) the rate of households spending more than 10 percent of their income on energy (P10), and (4) the rate of households who are pushed into energy poverty due to high fuel costs and low incomes (LIHC). Energy expenditure shares also vary depending on the (figure A.2.20). This share is lower when using income-based methodology used. When using expenditure-based mea- measures, namely 10.8 percent (figure A.2.21). When using sures, the average energy expenditure share is 17.3 percent income-based measures and abstracting from car-related en- 142 Romania Energy Poverty Assessment ergy expenditure, the resulting average expenditure share is latter, we investigate the impact of using income or expenditure 8.7 percent (figure A.2.22). Consequently, the estimate of Ro- as a welfare measure. mania’s average energy expenditure share fluctuates with the The analysis reveals that the underlying methodology methodology used. affects the results and findings are more intuitive when us- Our estimates align with previous estimates from other ing income-based measures. On average, energy expenditure countries, especially when using income-based measures. shares for Romanian households are lower when using in- Previous research finds that energy expenditure shares fluc- come-based measures. In addition, estimates along the welfare tuate between 4.1 and 12.0 percent (Bacon et al. 2010). While distribution seem more intuitive when using income-based our income-based measure is within this range, the expendi- measures of energy expenditure shares than expenditure-based ture-based measure results in an average share above this range. measures. While households at the lower end of the welfare dis- In addition, Bacon et al. (2010) find that the spending share on tribution spend higher shares of their overall welfare on energy transport is between 0.2 and 3.2 percent, which again aligns with in the case of the income-based measure, the same is not true our estimates. when using the expenditure-based measure. Estimates of energy poverty also fluctuate less across methodologies when using in- Discussion and Conclusion come-based measures. Based on these results, we conclude that income-based measures of energy expenditure shares are more We explore several approaches to measure energy expendi- suitable in the case of Romania. ture in Romania. The literature to date uses a variety of different We also provide evidence showing that including car-re- methodological approaches to measure energy spending pat- lated energy spendings results in less-intuitive and robust terns of households. We explore two of these approaches, relying estimates. We find that results are slightly more intuitive when on monetary measures of energy expenditure shares. We vary abstracting from car-related energy spending by households. both the numerator and the denominator of energy expenditure Moreover, energy poverty rates are more stable across meth- shares. In the former, we analyze measures with and without en- odologies when using this approach. Consequently, we abstract ergy expenditure on car-related energy, such as gasoline. In the from this energy spending component in our final analysis. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 143 Annex 3 Construction of Underlying Indicators Monetary Measures generated from these plantations for heating). Similar reasoning might apply to renewable energy sources, such as solar panels. In To estimate the household income, we rely on information addition, in the case of wood, households might be reluctant to from the income roster of the HBS (SVE) (S7). The section report wood bought on the black market or illegally. contains a variable (r44) that denotes total income, which we use for the analysis at hand. Nonmonetary Measures To estimate household spending on energy, we rely on data from the service expenditure roster (S6) and the infor- For the nonmonetary energy affordability measures, we rely mation on household spending on nondurable goods (S5). We on the living condition roster included in the HBS (S10a). summarize data on the expenditure on “electricity and renew- This part of the questionnaire is only filled out by the main able energy” (r511), “thermal energy” (r512), and “natural gas” household in a dwelling and its representative household mem- (r513). We do not include information on “Water, sewer, sanita- ber. We use the information in this part of the questionnaire to tion” (r510), given that we cannot distinguish between spending gather information on the type of lighting and heating used in on hot and cold water. We then summarize the resulting value Romania. We also generate information on the type of energy by household and month. We also add information gathered used for cooking as well as the connection of the household to on spending on nondurable goods (S5). Here we distinguish the sanitation system. Lastly, we analyze the question of sub- between liquid fuels (liquefied gas, oil, and liquid fuel for the sidies to cover expenses with public utilities (thermal energy, radiator [r340 to 342]) and solid fuels (firewood and coal [r343 natural gas, wood, coal, petroleum fuels, electricity, and so on). and r344]). Adding all these components together results in the overall household spending on energy. Consumption Aggregate Our approach to measuring energy spending has sever- al limitations. First of all, it does not reflect discounted bills. For this report, we construct the ECAPOV consumption Given that we only observe the amount ultimately paid by the aggregate. We follow the World Bank’s definition of the con- household, we do not capture potential ex-ante subsidies by sumption aggregate.1 The consumption aggregate is the nominal the government. Moreover, our measure does not account for household consumption expenditure divided by the household energy acquired by the household from proper resources (for size, the temporal consumer price index (CPI), and the spatial example, households owning forest plantations and using wood consumer price index (CPI): 1 For training material on how to follow the World Bank’s definition of the consumption aggregate, see this presentation: https://thedocs.worldbank. org/en/doc/20f02031de132cc3d76b91b5ed8737d0-0050012017/related/lecture-2-1.pdf. 144 Romania Energy Poverty Assessment Importantly, the definition of the consumption aggre- tion, all goods that are subject to economies of scale within a gate does not entail adding up all expenditures during the given household. In theory, it is possible to account for econo- reference period, but relies on the selective consideration mies of scale by taking the following approach: of all expenditures that best proxy the use (consumption) of resources. Following this definition, we only consider those expenditures that mirror the “typical” consumption during the reference year (in our case 2021). More specifically, we where xi is the individual welfare aggregate, xh the ex- implement the following definition of the consumption ag- penditure level per household, n the number of household gregate: nominal consumption aggregate = monetary expen- members, and a is a factor that scales the number of house- ditures on food and nonfood nondurable goods and services hold members. If a = 1, there is no adjustment for economies consumed + value of in-kind consumption + value of use (not of scale. If a = 0 all goods in the household are public goods. In purchase) of durables + value of use of owner-occupied hous- the case of a = 0.5 a household of four members, for example, ing. We abstract from expenditure for time and leisure as well would need twice as much as a household of one person. The as public goods. decision to account for economies of scale depends on the share For some indicators, we account for household size and of household expenditure directed to public goods, such as heat- composition. The consumption aggregate measures individ- ing, housing, or transportation. ual welfare, not household welfare, although expenditure data Alternatively, one can account for household consump- are typically collected at the household level. To compare in- tion through the concept of adult equivalents. This concept is dividuals’ welfare across households, it is crucial to adjust for based on the idea that certain household members consume less household size and composition. Notably, there are certain ex- than others. One example is a child, who might eat less than an penditures in a household that are public goods. This means that adult. Moreover, it is often assumed that women consume less consumption by one household member does not necessarily than men. The common practice is to assign women a value of reduce the amount available for consumption by another house- 0.8 and children below 15 years old a value of 0.5. The equivalent hold member. Examples are housing, heating, and transporta- scale would then look as follows: Following this approach, rescaled household consumption For the purpose of this report, we abstract from the adjust- is calculated as follows: ment for purchasing power, given that we do not consider com- parisons over time or space. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 145 Annex 4 Methodology: Measuring Direct and Indirect Welfare Impacts of Energy Inflation Impacts on Energy Poverty To do so, we relate P’ to the old price P, and Q’ to the old quantity of energy consumed, Q. According to Freund and Wallich (1995), We apply income-based measures to calculate energy spend- Q’ = Q *(1-e). Lastly, we also do not observe the new income I’ and ing shares and energy poverty. We divide the total energy ex- need to express it as a share of I. We declare that income I’ equals penditure by total household income as follows: the old income I minus the share of income lost due to inflation or economic downturns, expressed as i. To calculate energy poverty rates under the scenario of price increases, we use the new energy spending share ES’ and relate it to the 10-percent threshold used where Q is the average quantity of energy consumed by a in the 10-percent measure of energy poverty (P10). household, P is the average price for each unit of energy, and I is The equation above indicates that all three variables in- the total household income. Importantly, we do not observe P or volved in the energy spending share within the post scenario Q in the household budget survey, but only the total energy expen- are subject to change. Because we lack direct observations of diture E reported by each household. This limits the granularity of the new price P’ and new quantity of energy consumed Q’, it be- our analysis, because we cannot observe heterogeneity in energy comes necessary to express the equation in terms of the report- prices faced by households. ed energy expenditure I by households. To accomplish this, we To calculate the impact of energy price increases on ener- establish a relationship between the new price P’ and the initial gy poverty, we first calculate how energy prices impact energy price P, as well as between the new quantity of energy consumed expenditure shares. We embark from equation 1 and assume Q’ and the original quantity Q. According to Freund and Wallich that the energy share after an energy price increase is (1995), the new quantity of energy consumed Q’ is calculated as Q multiplied by 1-e, where e represents the price elasticity. Furthermore, the new income I’ is not directly observed, and thus, we express it as a proportion of the initial income I, adjust- The formula indicates that all three variables that comprise the ing for the portion of income lost due to inflation or economic energy spending share in the post scenario are subject to change. downturns, denoted as i. As we do not observe P’ or Q’, we need to express the formula in We then estimate the new energy poverty rate. To com- relation to E, the energy expenditure reported by the household. pute energy poverty rates in the context of price increases, we 146 Romania Energy Poverty Assessment utilize the new energy spending share ES’ and relate it to the gy-related portion PPP from the original income. This revised threshold of 10 percent employed in the P10 measure. 1 income serves as the basis for evaluating the AROP rates. The rate change reflects the welfare effect resulting from the price increas- Impacts on Income Poverty es. Furthermore, we also examine variations in the Gini coeffi- cient, which provides insights into changes in income inequality. To assess the household welfare impacts, we investigate the The PPP loss in our baseline scenario is as follows: effects of price hikes on the AROP rates. Building upon the work of Freund and Wallich (1995), we consider that energy price increases result in alterations in consumer surplus as a proportion of expenditure or income. The specific relationship We simulate the impact of energy price increases on mon- can be expressed as follows: etary welfare measures relying on data from the HBS. To this end, we rely on the welfare measures from the HBS, because information on household income is relatively reliable, and the resulting AROP rate, standing at 21.4 percent, is close to the of- In this context, the energy expenditure share ES represents ficial one reported by the EU-SILC. Table A.6.1 in Annex 6 pres- the proportion of household income allocated to energy ex- ents the full results for all scenarios. Depending on the scenario, penses, as explained earlier. To assess the impact of energy price a positive price increase results in an increase of the AROP rate increases on welfare, we introduce the variable t, which denotes over the one reported in 2021 of -0.17 (when households fully the percentage change in energy prices, and e, which represents substitute away from energy prices, a highly unlikely scenario) the energy price elasticity. By applying these values, we can es- to 3.0 percentage points (assuming a price increase of 50 percent timate the adjusted household income by subtracting the ener- and a price elasticity of 0). 1 We do not analyze the effect on the M/2 and 2M measures of energy poverty, because these are relative measures and barely change over time. These patterns of results are by construction, because the M/2 and 2M measures are relative to the distribution. So, if the entire distribution moves into a certain direction—as is the case here—then the relative measures of energy poverty might not change significantly. The P10 measure of energy poverty might be more appropriate for tracking changes in energy poverty over time, although it is subject to serious empirical limitations as well (see the overview of Robayo-Abril and Rude [2024]) for more information. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 147 Annex 5 Descriptive Statistics of Single-Elderly Households Table A . 5.1 Des cri p ti ve Stati s ti cs of Si n g le -Eld e r ly H o u s e ho ld s Co m pa re d to A l l Ot he r H o u s e ho ld s , 2 021 (1) (2) Single-elderly households All other households Mean S.D. Mean S.D. Energy spending share (income-based, no car) 0.15 0.13 0.08 0.09 Household with children (<15) 0.00 0.00 0.30 0.46 Household with pensioner 0.99 0.12 0.36 0.48 Single-elderly household 1.00 0.00 0.00 0.00 Female-headed household 0.73 0.44 0.23 0.42 Household with Roma 0.01 0.07 0.03 0.16 Household with unemployed members 0.04 0.20 0.06 0.24 Urban 0.51 0.50 0.53 0.50 Household with more than 5 members 0.00 0.00 0.06 0.25 Income per capita 24569.34 25312.09 31928.47 26369.42 Social aid from municipality (recipient) 0.00 0.05 0.02 0.13 Disability benefit recipient 0.00 0.04 0.03 0.18 At-risk-of-poverty 0.30 0.46 0.22 0.41 Connectivity to electricity 0.98 0.14 0.98 0.13 Has thermal power station 0.14 0.34 0.14 0.34 Has central heating 0.36 0.48 0.49 0.50 Central heating type: wood/pellets 0.02 0.14 0.08 0.27 Central heating type: natural gas 0.33 0.47 0.40 0.49 Central heating type: electricity 0.00 0.04 0.00 0.06 Central heating type: Other 0.00 0.01 0.00 0.01 148 Romania Energy Poverty Assessment (1) (2) Single-elderly households All other households Mean S.D. Mean S.D. Has natural gas stove 0.05 0.21 0.02 0.14 Has wood/coal/oil stoves 0.45 0.50 0.34 0.47 No heating system 0.00 0.04 0.00 0.03 Disconnected 0.00 0.05 0.00 0.04 Subsidy (thermal energy) 0.01 0.10 0.01 0.11 Subsidy (natural gas) 0.00 0.03 0.00 0.01 Subsidy (wood/coal/oil) 0.00 0.02 0.00 0.02 Subsidy (electricity) 0.01 0.09 0.01 0.10 Subsidy (other) 0.00 0.00 0.00 0.00 None 0.99 0.10 0.99 0.11 Cooking (electricity) 0.00 0.02 0.00 0.06 Cooking (natural gas) 0.51 0.50 0.55 0.50 Cooking (wood/coal/oil) 0.23 0.42 0.20 0.40 Cooking (cylinder) 0.47 0.50 0.42 0.49 Cooking (other) 0.00 0.02 0.00 0.01 Cold water, indoors from public supply 0.70 0.46 0.74 0.44 Indoors, from in-house 0.09 0.29 0.12 0.33 Outside residence, but inside building 0.02 0.13 0.01 0.10 Outdoors, fountain, pump, well 0.20 0.40 0.13 0.33 River 0.00 0.00 0.00 0.00 Other 0.00 0.00 0.00 0.00 Hot water—N/A 0.00 0.04 0.01 0.11 Hot water—public system 0.17 0.38 0.16 0.37 Hot water—in-house system 0.55 0.50 0.68 0.47 Hot water—no hot water 0.27 0.45 0.15 0.35 Hot water—disconnected 0.00 0.04 0.00 0.04 Sewage type—N/A 0.00 0.04 0.01 0.11 Sewage type—public system 0.57 0.49 0.62 0.49 Sewage type—in-house system 0.20 0.40 0.25 0.43 Sewage type—none 0.22 0.42 0.12 0.32 Electric lighting 1.00 0.05 0.99 0.11 Natural gas (cooking or heating) 0.51 0.50 0.56 0.50 Observations 7,672 22,854 Source: Own estimates based on the 2021 HBS. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 149 Annex 6 Simulated Poverty and Inequality Rates across Various Scenarios Table A .6.1 Si mulated Es ti mates of ARO P Ra te s u nd e r a Fu l l S e t o f S ce na r io s Energy price PPP loss (% of HH At-risk of poverty At-risk of poverty Scenario Elasticity Gini increase income) (%) (equ.) (%) 1 0 0 0.00 29.90 21.42 0.364536 2 0 -0.25 0.00 29.90 21.25 0.364536 3 0 -0.5 0.00 29.90 21.25 0.364536 4 0 -0.75 0.00 29.90 21.25 0.364536 5 0 -1 0.00 29.90 21.25 0.364536 6 0.1 0 0.01 30.10 21.81 0.366431 7 0.1 -0.25 0.01 30.10 21.80 0.366406 8 0.1 -0.5 0.01 30.10 21.79 0.366382 9 0.1 -0.75 0.01 30.09 21.79 0.366358 10 0.1 -1 0.01 30.09 21.77 0.366334 11 0.2 0 0.02 29.90 21.25 0.368396 12 0.2 -0.25 0.02 30.27 22.40 0.368296 13 0.2 -0.5 0.02 30.28 22.38 0.368196 14 0.2 -0.75 0.02 30.26 22.35 0.368097 15 0.2 -1 0.02 30.26 22.31 0.367997 16 0.3 0 0.03 30.52 23.07 0.370436 17 0.3 -0.25 0.03 30.49 22.98 0.370203 18 0.3 -0.5 0.02 30.47 22.93 0.369971 19 0.3 -0.75 0.02 30.45 22.89 0.369739 150 Romania Energy Poverty Assessment Energy price PPP loss (% of HH At-risk of poverty At-risk of poverty Scenario Elasticity Gini increase income) (%) (equ.) (%) 20 0.3 -1 0.02 30.42 22.81 0.369509 21 0.4 0 0.03 29.90 21.25 0.372552 22 0.4 -0.25 0.03 30.62 23.63 0.372123 23 0.4 -0.5 0.03 30.58 23.51 0.371696 24 0.4 -0.75 0.03 30.56 23.37 0.371273 25 0.4 -1 0.03 30.55 23.22 0.370853 26 0.5 0 0.04 30.81 24.37 0.374747 27 0.5 -0.25 0.04 30.76 24.15 0.374052 28 0.5 -0.5 0.04 30.70 23.99 0.373366 29 0.5 -0.75 0.04 30.63 23.81 0.372687 30 0.5 -1 0.03 30.61 23.61 0.372016 Source: Own estimates based on the 2021 HBS. Understanding and Addressing Energy Poverty in Romania: Exploring the Roles of Structural and Behavioral Constraints 151 Annex 7 Simulation of Potential Short-Term Measures Mitigating Direct Effects of Rising Energy Prices by Eliminating Price Caps for Electricity and Natural Gas (70% Increase in Total) Table A .7 .1 Si mulati on of Poten ti al Sh or t-Te r m Meas u re s Mit iga t ing D ire c t Ef fe c t s o f Ris ing Ene rg y Pr ices by El i mi n ati n g Pri ce Cap s for Ele c t r ic ity a nd N a t u ral Gas ( 70 % Inc reas e in To tal ) Policy Poverty Poverty Poverty Reduction Cash Targeted population Gini Beneficiaries Adequacy measure Headcount gap reduction poverty gap transfer 1 All 26.7 10.1 34.6 0.62 0.41 72 19,348,459 4.5 2 Single-elderly 24.6 9.6 33.6 2.70 0.95 1,136 1,232,223 23.0 Single-elderly at-risk of 3 24.6 9.5 33.5 2.68 0.98 2,696 519,197 54.4 poverty 4 Pensioner 26.3 10.0 34.1 0.98 0.54 332 4,211,665 6.3 Pensioner at-risk of 5 23.7 9.2 33.5 3.60 1.33 1,665 841,003 30.7 poverty 6 CT recipients 25.9 9.6 33.9 1.39 0.88 1,848 757,757 33.4 Bundle 2: Pool 7 beneficiaries Pensioner 25.1 9.0 33.6 2.20 1.52 890 1,573,329 15.8 at-risk and CT recipients Bundle 3: Pool beneficiaries Pensioner 8 23.3 9.0 33.4 3.94 1.51 1,478 947,173 29.5 at-risk and single-elderly at-risk Source: Own estimates based on the matched 2019 HBS and 2020 EU-SILC dataset. 152 Romania Energy Poverty Assessment