The Economics of Winning Hearts and Minds Programming Recovery in Eastern Ukraine The Economics of Winning Hearts and Minds Programming Recovery in Eastern Ukraine © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 23 22 21 20 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not nec- essarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accu- racy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immuni- ties of The World Bank, all of which are specifically reserved. Rights and Permissions Attribution—Please cite the work as follows: World Bank. 2021. The Economics of Winning Hearts and Minds: Programming Recovery in Eastern Ukraine. Wash- ington, DC: World Bank. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each compo- nent of the content contained within the work. The World Bank therefore does not warrant that 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 such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, fig- ures, or images. All queries on rights and licenses should be addressed to World Bank Publi- cations, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Cover and interior art: Design and Development Minds. Contents Acknowledgements............................................................................................................10 Abbreviations....................................................................................................................... 12 Overview................................................................................................................................. 15 Chapter 1. The Long Shadow of History and Geography...................................... 43 The Legacy Growth Model and Its Demise..................................................................................46 Social Corollaries ...........................................................................................................................................56 The Role of Preferences: When People Vote with Their Feet .........................................65 Synopsis...............................................................................................................................................................68 Chapter 2. The Consequences of the Conflict.........................................................73 Conflict Dynamics..........................................................................................................................................75 Destruction.........................................................................................................................................................80 Displacement ...................................................................................................................................................87 Disorganization................................................................................................................................................94 Synopsis...............................................................................................................................................................110 Chapter 3. Current Conditions and Challenges......................................................115 Economic Activity.......................................................................................................................................... 117 Labor Markets and Demography........................................................................................................126 Access to Public Services........................................................................................................................ 131 Synopsis..............................................................................................................................................................143 Chapter 4. Future Development: 864 Scenarios...................................................147 An Integrated Approach to Future Development.................................................................. 149 The Impact of Policies ..............................................................................................................................157 Implications for Effective and Efficient Policies ..................................................................... 174 Synopsis..............................................................................................................................................................183 Chapter 5. Conclusions and Conjectures for a Development Strategy........ 187 Where Do We Stand? A Recap........................................................................................................... 189 3 What Is to Be Done?.....................................................................................................................................191 Conclusion.........................................................................................................................................................197 Appendixes.........................................................................................................................199 Appendix A Ukraine before Conflict: Trade, Labor, and Demographic Indicators, 2013 ...................................................................................................................................................................... 201 Appendix B Ukraine during Conflict: Characteristics of Internal Migration........209 Appendix C Recent Demographic and Labor Market Indicators.................................214 Appendix D Technical Aspects of the Simulation Model.................................................217 The Economics of Winning Hearts and Minds 4 List Of Boxes, Figures, Maps, And Tables Boxes O.1 A Simulation Approach to Future Policy Design.......................................................... 33 2.1 Chronology of the Conflict in Eastern Ukraine, 2013–present.............................81 2.2 The National Monitoring System (NMS) Survey: Definitions and Methodology....................................................................................................94 2.3 What Disorganization Looks Like: The Story of the Luhansk TPP................... 103 4.1 Characteristics of the Simulation Model .........................................................................152 Figures O.1 Population and Regional GDP Trends in Ukraine and Selected Regions, 2004–13...................................................................................................................................................20 O.2 Cargo Turnover at the Port of Mariupol in Donetsk Oblast, by Cargo Type, 2013–18................................................................................................................. 25 O.3 Differences in Sectoral Outputs and Productivity in Ukraine and Donetsk and Luhansk Oblasts, before and after Onset of Donbas Conflict................. 26 O.4 Decomposition of Job Vacancies and Unemployment, by Occupation Type, in Ukraine and the Donetsk and Luhansk GCAs, December 2019..... 28 O.5 Satisfaction with Public Services in Donbas GCAs and NGCAs, 2017–19...30 O.6 Simulated Changes in Regional GDP in Donetsk and Luhansk Oblasts, by Conflict Scenario........................................................................................................................30 O.7 A Decision Tree Approach to Designing an Economic Recovery Strategy for Eastern Ukraine............................................................................... 38 1.1 Implicit Energy Subsidies as a Share of GDP in Ukraine, 2003–13.................. 47 1.2 GDP Growth and External Demand in Ukraine, 2000–13....................................... 47 1.3 Global Commodity Price Trends, 2000–19 ....................................................................49 1.4 Energy Intensity of CEE and CIS Countries, 2017........................................................50 1.5 Potential GDP and Output Gap in Ukraine, 2001–20..................................................51 1.6 Sectoral Structure of Employment in Ukraine, by Regional Group, 2013..... 52 1.7 Merchandise Trade Balance in Ukraine, by Region, 2002–13.............................. 53 1.8 Energy Independence in Ukraine, by Region, 2018..................................................... 53 1.9 Contributions to Ukraine’s GDP Growth, by Selected Region, 2005–13 ....54 1.10 Changes in Manufacturing Shares of Employment and GDP, by Country Income Level, 1994–2011...............................................................................................................55 5 1.11 Regional GDP Convergence or Divergence in Ukraine, 2004–13......................55 1.12 Rank-Size GDP Distribution of Ukrainian Regions, 2004 versus 2013...........56 1.13 Average Monthly Wages in Ukraine, by Region, 1995–2013 ................................. 57 1.14 Integrated Index of Regional Human Development in Ukraine, by Dimension, 2013...........................................................................................................................58 1.15 Population Change in Donetsk and Luhansk Oblasts, 1989–2013.................. 60 1.16 Growth of Nominal Wages Relative to Regional GDP Per Capita in Donetsk and Luhansk Oblasts, 2004–13................................................................................................ 62 1.17 Sources of Household Incomes in Donetsk and Luhansk Oblasts Compared with Rest of Ukraine, 2013 ................................................................................63 1.18 Mother Tongues in Ukraine, by Region, 2001 .................................................................64 1.19 Net Migration within Ukraine, Donetsk and Luhansk Oblasts, 2002–13.......65 1.20 Net International Migration in Donetsk and Luhansk Oblasts, 2002–13...... 67 1.21 Gross Migration Flows in Donetsk and Luhansk Oblasts, by Location and Flow Type, 2013 .................................................................................................................................. 67 2.1 Assessment Bias from Changing Spatial Coverage of Data in Donbas: A Hypothetical Example................................................................................................................ 83 2.2 Challenges in Obtaining Employment Reported by Unemployed IDP Heads of Household in Ukraine, by Destination Region, 2018 ............................89 2.3 Educational Attainment of IDP and Returnee Household Heads in Donbas and Rest of Ukraine, by Current Place of Residence, 2018 ......... 92 2.4 Cargo Turnover at the Port of Mariupol, Donetsk Oblast, by Cargo Type, 2013–18..................................................................................................................................................... 97 2.5 Perceived Levels of Corruption in Donbas GCAs, by Behavior Type, 2019...............................................................................................................................................110 3.1 Estimated Changes in Regional GDP Using Spatially Variable GDP Statistics and Nighttime Light Emissions, Donetsk and Luhansk Oblasts, The Economics of Winning Hearts and Minds 2010–18 ...................................................................................................................................................118 3.2 Differences in Sectoral Output and Productivity in Ukraine and Donetsk and Luhansk Oblasts before and after Onset of Donbas Conflict ............... 120 3.3 Merchandise Trade Balance in Ukraine and Selected Regions, 2010–18....125 3.4 Differences in Enterprise Size and Employment in Ukraine and Selected Regions, 2013 (Whole Oblasts) versus 2018 (GCAs only) ....................................124 3.5 Volume and Interest Rates of Lending and Deposit Operations in Donetsk GCA, 2018.....................................................................................................................126 3.6 Differences between Donbas and Rest of Ukraine in Sectoral Structure of Employment, 2013 (Whole Oblasts) versus 2019 (GCAs Only)........................ 127 3.7 Decomposition of Job Vacancies and Unemployment, by Occupation Type, in Ukraine and Donetsk and Luhansk GCAs, December 2019 ............129 3.8 Satisfaction with Public Services in Donbas GCAs and NGCAs, 2017–19..135 6 3.9 Numbers of School Facilities and Students in Donetsk and Luhansk Oblasts, 2013–18 ...............................................................................................................................139 3.10 Perceptions of Educational Quality and the Capacity-Building Needs of Teachers in Donetsk and Luhansk GCAs, 2018 ..........................................................140 3.11 Water Supply Service Coverage of Settlements in Donetsk and Luhansk Oblasts, by Urban and Rural Location................................................................................142 4.1 Simulation Model to Assess Implications of New Policies in Eastern Ukraine.......................................................................................................................................................151 4.2 Conflict and Policy Scenarios................................................................................................. 156 4.3 Projected Regional Medium-Term GDP Gains, by Scenario Combination, in Donbas and Kyiv City ............................................................................................................. 159 4.4 Simulated Medium-Term Implications of Conflict Scenarios, by Selected Region .......................................................................................................................................................161 4.5 Simulated Effects of Medium-Term Mobility Cost Reductions (from 2020 Values) in Donbas and Kyiv City, by Conflict Scenario and Location .......... 173 4.6 Medium-Term Policy Outcomes under Uncertainty in Donbas and Ukraine, by Type ...............................................................................................................................182 5.1 A Decision Tree Approach to Designing an Economic Recovery Strategy for Eastern Ukraine...........................................................................................................................193 A.1 Ukraine’s Trade Patterns, 2013 ............................................................................................... 201 Maps O.1 Changes in Nighttime Light Emissions in Donetsk and Luhansk Oblasts, 2013–20.................................................................................................................................................... 22 O.2 Railroad Infrastructure and the Contact Line in Donbas, 2019.......................... 24 2.1 Casualties (Deaths) in Donetsk and Luhansk Oblasts, by Event, 2014–19.....................................................................................................................................................80 2.2 Changes in Nighttime Light Emissions, Donetsk and Luhansk Oblasts, 2013–20....................................................................................................................................................85 2.3 Railroad Infrastructure in Donbas in Relation to Conflict Zones, 2019 .........96 2.4 Water Distribution Coverage in Donetsk Oblast, by Settlement, 2015 .....100 2.5 Water Distribution Coverage in Luhansk Oblast, by Settlement, 2015 ..... 102 3.1 Levels of Satisfaction with Infrastructure in Donbas GCAs and NGCAs, 2017–19 ................................................................................................................................134 3.2 School-Age Population and School Attendance across Donbas Rayons, 2013 versus 2018 ..............................................................................................................................137 7 Tables O.1 Factors Driving the Uptrend and Decline of the Legacy Economic Model in Ukraine, before and after 2009 .......................................................................... 19 O.2 Main Labor Market Indicators in Ukraine and Donetsk and Luhansk GCAs, 2018 ....................................................................................................................... 29 O.3 Intentions of Working-Age Household Heads of IDPs to Return to Place of Origin in Donbas, by Region of Current Residence, 2018 .................. 29 O.4 Simulations for the GDP Impact of Policies: Policy Rank Switches between Two Scenarios in Eastern Ukraine ...................................................................36 2.1 Housing Stock of Donetsk and Luhansk Oblasts, 2013–18 ..................................84 2.2 Sociodemographic Characteristics of IDPs, Returnees, and Working- Age Population in Donetsk and Luhansk Oblasts Compared with Rest of Ukraine, 2018 ........................................................................................................................................88 2.3 Main Reasons for Returning to Donbas NGCAs, by Current Place of Residence and Age Group, 2018............................................................................................. 92 2.4 Indicators for Intergroup Contact and Openness to Dialogue with Specified Groups in Ukraine and Donbas GCAs, 2017–19 ..................................106 2.5. Indicators of Support for Reforms and Institutional Trust in Donbas GCAs................................................................................................................................108 2.6 Trust in Central and Local Institutions in Donbas GCAs, 2018–19 ................109 2.7 Trends in Perceived Corruption and Tolerance of Corruption in Ukraine and Donbas GCAs, 2017–19..........................................................................................................111 3.1 Main Labor Market Indicators in Ukraine and Donetsk and Luhansk GCAs, 2018 ......................................................................................................................128 3.2 Intentions of Working-Age Household Heads of IDPs to Return to Place of Origin in Donbas, by Region of Current Residence, 2018 ................ 130 3.3 Sociodemographic Characteristics of Working-Age Household The Economics of Winning Hearts and Minds Heads of IDPs, by Intention to Return to Place of Origin in Donbas, 2018 ...........................................................................................................................................................132 4.1 Scenario Approach to Considering Future Uncertainty and Policies ........ 164 4.2 Simulated Medium-Term Changes in Donbas and Ukraine, by Conflict, Investment Scale, and Allocation Scenario ...................................... 168 4.3 Simulated Medium-Term Changes in Donbas and Ukraine, by Conflict, Transfer Scale, and Target Scenario ....................................................... 177 4.4 Simulated Changes in Medium-Term GDP versus Inequality-Adjusted Personal Incomes in Donbas and Ukraine, by Policy Choice and Conflict Scenario .............................................................................................................................179 A.1 Selected Demographic and Labor Market Indicators in Ukraine, by Zone and Region, 2013..........................................................................................................203 A.2 Indicators of Employment in Ukraine, by Zone and Region, 2013..................206 8 B.1 Sociodemographic Characteristics of IDPs and Returnees, by Region, 2018.................................................................................................................................209 B.2 Selected Characteristics of IDPs (Heads of Household) in Ukraine, by Destination, 2018........................................................................................................................212 C.1 Selected Demographic and Labor Market Indicators in Ukraine, by Zone and Region, Recent Year...........................................................................................214 9 Acknowledgments This study was prepared by a multidisciplinary team led by Harun Onder (lead author) under the guidance of Kevin Carey (practice manager) and Jasmin Chakeri (practice manager). Contributing authors by field are as follows: •• Macroeconomy: Lubomir A. Mitov (macroeconomic expert) and Harun Onder. •• Labor markets and demography: Olga Kupets (labor markets and migration expert). •• Economic model: Erhan Artuç (senior economist), Harun Onder, and Nicolás Gómez Parra (economist). •• Social sector: Holly Welborn Benner (senior social development specialist), Ray Salvatore Jennings (social development expert), Oleksandra Shatyrko (social development specialist), and Nina Kolybashkina (senior social devel- opment specialist). •• Conflict analysis: Spyridon Demetriou (senior operations officer), Nadia Fernanda Piffaretti (senior economist), Georgia Christina Kosmidou (fragility, conflict, and violence (FCV)), and Allison N. Grossman (FCV expert). •• Infrastructure and public services: Emil Mihailov (metallurgy expert), James A. Gresham (education specialist), Gulcan Yayla (education expert), Ivaylo Hristov Kolev (WSS expert), Irina Zapatrina (water supply and sanitation ex- pert), Dmytro Glazkov (senior energy specialist), Fabrice Karl Bertholet (se- nior financial specialist), Volodymyr Ryabtsev (energy expert), Ilya Polyak (energy expert), and Yevhen Bulakh (transport specialist). The team would like to express sincere gratutude to authorities from Ministry for The Economics of Winning Hearts and Minds Reintegration of the Temporarily Occupied Territories of Ukraine for their gener- ous feedback, to local administrations of Donetsk and Luhansk Oblasts for their comments and hospitality, and to representatives from civil society organiza- tions and the business community in the Donbas region for several rounds of consultations. The team is also grateful for guidance and support from Anna Bjerde (Europe and Central Asia vice president), Arup Banerji (country director for Ukraine), Lal- ita Moorty (Equitable Growth, Finance and Institutions [EFI] regional director), Steven N. Schonberger (Sustainable Development Network [SDN] regional di- rector), Marcello Estevão (Macroeconomics, Trade and Investment [MTI] Global Practice director), Satu Kähkönen (previous country director for Ukraine), Gallina Vincelette (previous MTI practice manager for Ukraine), Baher El-Hifnawi (pro- gram leader), and Enrique Blanco Armas (lead economist). The team benefited from excellent comments from the peer reviewers Tito Cor- della (adviser, MTI), Xavier Devictor (FCV practice manager), Thomas Farole (lead 10 economist, SDN), Apurva Sanghi (lead economist, MTI), and Karlis Smits (pro- gram leader, EFI). The following colleagues kindly provided additional comments: Baher El-Hifnawi (SDN program leader), Faruk Khan (program leader), Vara Vem- uru (practice manager), Kevin A. Tomlinson (practice manager), Anthony Gaeta (country program coordinator), Tehmina Khan (senior country economist), Anas- tasia Golovach (senior country economist), Dmytro Derkach (senior EXT officer), Erik Caldwell Johnson (senior social development specialist), Klavdiya Maksy- menko (senior country officer, ECCUA), and Boris Ajeganov (EXT specialist). As- sistance and support from Thomas Djurhuus (lead partnership specialist), Patri- cia Miranda (senior counsel), Fabiola Altimari (senior counsel), Valery R. Ciancio (senior operations officer), Ekaterina G. Stefanova (senior program assistant), Nadia Kislova (team assistant), and Julia Samoslied (program assistant) are also gratefully acknowledged. This study was financed by generous financial contributions from the Ukraine Multi-Partner Trust Fund and the State and Peacebuilding Trust Fund. Technical assistance and data sharing by the International Organization for Migration (IOM) is gratefully acknowledged. The team would also like to acknowledge numerous government officials, international community members, and civil society or- ganization members who joined several rounds of consultations and provided invaluable comments and suggestions. Remote sensing-based assesments used in this report were prepared by Ipsos and theSocial Cohesion and Reconciliation (SCORE) survey analysis was provid- ed bythe Centre for Sustainable Peace and Democratic Development (SeeD). Mary A. Anderson conducted the editorial review, and Design and Development Minds performed the typesetting. 11 Abbreviations EU European Union GCA government-controlled area GDP gross domestic product GFC Global Financial Crisis GRP gross regional product IDPs internally displaced persons IPI inequality-adjusted personal income LFS Labor Force Survey MoSP Ministry of Social Policy of Ukraine NGCA non-government controlled area NMS National Monitoring System OSCE Organization for Security and Co-operation in Europe ROI return on investment SCORE Social Cohesion and Reconciliation Index SEZ special economic zone SMEs small and medium enterprises The Economics of Winning Hearts and Minds SSSU State Statistics Service of Ukraine TCG Trilateral Contact Group TFP total factor productivity TPP thermal power plant UN United Nations UXOs unexploded ordnance WSS water supply and sanitation 12 Key Messages 1. Since 2014, the armed conflict in Ukraine’s eastern provinces (oblasts) of Donetsk and Luhansk has dealt a heavy blow to people’s lives. As of 2020, more than 13,000 people had been killed, about 1.4 million Ukrainians were registered as internally displaced persons, and about 38 percent of Don- bas—the combined territories of Donetsk and Luhansk Oblasts—remained outside government control, separated from the rest of Ukraine by a 457 kilometer line of contact. 2. The conflict has magnified the long-standing problems and created new ones. Donbas was already suffering from a diminishing legacy growth model and rapid aging in industry, infrastructure, and demography before the con- flict. Between 2004 and 2013, Donetsk and Luhansk Oblasts lost 7.1 percent and 8.3 percent of their populations, respectively, as a result of out-migra- tion and demographic aging, and their shares of Ukraine’s gross domestic product (GDP) decreased by 2.2 and 0.7 percentage points, respective- ly. The conflict accelerated infrastructure depreciation by damaging stra- tegic assets (the destruction channel) and deepened demographic aging by displacing many people, especially the young and economically active (the displacement channel). It also created many transactional frictions in a previously integrated region (the disorganization channel), including con- flict-driven risks and uncertainty. Together, these three channels (the “3 Ds”) have further crippled economic prospects in Donbas. 3. This study shows that scaling up efforts in the government-controlled areas (GCAs) of Donbas is desirable despite the subdued productivity in the region. This is because the productivity gaps between Ukrainian regions are small relative to interregional mobility costs; thus, labor is not sufficient- ly reallocated toward higher-productivity areas. However, the optimal scale and composition of such efforts—such as through investments, transfers, or mobility policies—need to be calibrated carefully. Simulations in this report show that the best policy instrument for the job varies by conflict scenario (status quo versus reintegration) and that there are trade-offs between pol- icy objectives (for example, GDP in Donbas versus GDP in Ukraine). 4. Policies in Donbas should be coordinated around a unified development path. Given the structural and conflict-driven challenges, the region needs a comprehensive and integrated strategy (a complete and contingent plan) with the following characteristics: •• Balanced (no silver bullet): The strategy should explicitly weigh trade- offs between different policies. For instance, transfers are effective for boosting GDP in the GCAs but not in Ukraine overall. Thus, a “silver bullet” approach (for example, only infrastructure investments or special eco- nomic zones) will not work. A broad policy approach that balances such trade-offs is needed. 13 •• Nuanced (one size does not fit all): Policies need to be recalibrated be- tween the conflict (status quo) and peace (reintegration) scenarios. For instance, lowering mobility costs is highly effective in boosting Ukrainian GDP under the status quo, but its effect weakens under reintegration. •• Transformative (distortions do not fix distortions): The region’s struc- tural problems were sustained by long-standing, policy-driven distor- tions. Thus, fixing market fundamentals should be a priority. Without a better business climate, rule of law, and anticorruption efforts, selective incentive schemes for attracting investors are likely to be exhausted by rent-seeking actors. 5. This study recommends a decision tree approach to programming re- covery in Donbas. Given the looming uncertainties and scenario-sensitiv- ity of optimal policies, the recovery strategy should distinguish “contingent policies” from “no-regret policies.” Contingent policies change between the status quo and the reintegration scenarios, and they include interventions to mitigate conflict-related risks, risk-related transfers to address skill-short- ages in GCAs, and investments for a contingent infrastructure strategy. By comparison, no-regret policies are desirable regardless of the conflict dy- namics. They include the reforms to eliminate regulatory burdens and cor- ruption; policies to open up the housing market; investments to modernize education for jobs and target low-hanging fruits in infrastructure; and efforts to produce better data to address knowledge gaps. The Economics of Winning Hearts and Minds 14 Overview S ince 2014, the armed conflict in Ukraine’s eastern provinces (oblasts) of Donetsk and Luhansk has dealt a heavy blow to people’s lives. By United Nations (UN) estimates, about 13,000 people had been killed, a quarter of them civilians, and about 30,000 had been injured in the conflict as of 2020.1 The Ministry of Social Policy of Ukraine put the number of registered internally displaced persons (IDPs) at 1,420,523 in July 2020. The conflict has also torn the region’s economic and social fabric in complex ways. Some of these effects (like physical destruction) are measurable. Others, including diminished social trust in key institutions, are hard to quantify despite being equally important, if not more so. As of 2020, about 38 percent of Donbas—the combined territories of Donetsk and Luhansk Oblasts—remained outside government control (“non-government controlled areas,” or NGCAs) and are bounded by a 457 kilometer line of contact (“contact line”) with the “government-controlled areas” (GCAs). This division of previously integrated areas has introduced additional obstacles to economic activity by cutting off connectivity, public service delivery systems, and supply chains. Although recent negotiations have resulted in a significant decrease in hostilities and the partial disengagement of forces, a lasting settlement remains elusive. In response, the Ukrainian government created the Ministry of Temporarily Oc- cupied Territories and IDPs (MTOT) in 2016, which was renamed as the "Ministry for Reintegration of the Temporarily Occupied Territories of Ukraine" in 2020. The ministry embodies the public authority to shape and implement policies on recovery and peacebuilding in conflict-affected areas and on reintegration of of the NGCAs. Recently, MTOT embarked on producing a Strategy for the Eco- nomic Development of Donetsk and Luhansk Oblasts, with the approval of the concept note by the Cabinet of Ministers on December 23, 2020. The strategy is firmly grounded in President Volodymyr Zelenskyy’s “winning hearts and minds” approach, and it aims to implement large-scale economic initiatives and invest- ments “to ensure the sustainability of local communities, to create new jobs, and to fill local budgets” (Reznikov 2021). The final document, with associated bills, is expected to be prepared by summer 2021. This World Bank study aims to analyze the economic underpinnings of future recovery in eastern Ukraine. The strength of the analysis lies in its technical ap- proach to comparing how effectively various policy measures can fulfill different policy objectives under different conditions. This comparison can help policy makers to prioritize alternative policies and thus transform a long list of potential policy actions into a structured strategy. That is, it can help policies to have the most impact within a given resource envelope by balancing the trade-offs asso- ciated with different actions. To this end, the report first reviews the economic and social trends in Donbas before 2014. This discussion makes clear the limitations of traditional economic dynamics that will persist even after the conflict. Next, the report studies the mechanisms through which the conflict has changed these dynamics, focus- Overview ing on three main channels—destruction, displacement, and disorganization—to trace changes in economic activity, labor markets and demography, and access to public services. 17 Finally, to evaluate future policy options, the report employs a model-based approach. In this framework, three categories of policy actions (investments, transfers, and mobility cost reduction) are compared for outcomes regarding three different objectives: boosting gross domestic product (GDP) in the GCAs, boosting GDP in Ukraine, and boosting “inequality-adjusted average personal in- comes” in Ukraine. This comparison helps to identify trade-offs associated with each policy intervention quantitatively. Given the sensitivity of economic and social outcomes to conflict dynamics, and the uncertainty surrouding the future trajectories of those dynamics, the report performs such comparisons under three conflict scenarios: status quo, intermediate, and reintegration. The remainder of this overview presents some key results from the analysis in a nontechnical fashion. For more detailed descriptions of the methodologies be- hind the findings, refer to relevant sections in the main body of the report. Before the Conflict: A Legacy Economic Model on Life Support The economic problems of the Donbas region did not begin with the conflict. They have deep roots in how the economy operated long before the onset of the current conflict. Such operating principles should be considered carefully because they may continue to constrain the region’s future development in the absence of mitigating policies. Modern development of the Donbas region began in the late 19th century and continued under Soviet industrialization of the 1930s, during Nazi occupation in World War II, and throughout the postwar reconstruction years. In 1913, Donbas was producing 74 percent of the pig iron and 87 percent of the total coal output in the Russian Empire (Kohut, Nebesio, and Yurkevich 2005). The region would continue to be the largest producer of coal in the Soviet Union until the 1960s, which also gave rise to ancillary industries. Heavy engineering facilities emerged The Economics of Winning Hearts and Minds in Luhansk, Kramatorsk, and other industrial centers. Chemical industries for coking byproducts and rock salt were established in Artemivsk and Sloviansk. Over a few decades, Donbas became the most heavily settled region of Ukraine, housing a fifth of the nation’s urban settlements. It attracted people from else- where in Ukraine and from other parts of the Soviet Union, including the current Russian Federation. Through large exports in energy-intensive heavy industry (largely metallurgy), the region continued to play a major role in Ukraine’s eco- nomic output, employment, and exports after the country’s independence in 1991. After independence, there was little investment and modernization in the re- gion’s industrial base and infrastructure. Yet the legacy model still worked rel- atively well in the 2000s, thanks to favorable external conditions and policy- driven domestic distortions favoring the traditional industries (table O.1). 18 Table O.1 Factors Driving the Uptrend and Decline of the Legacy Economic Model in Ukraine, before and after 2009 Factor Uptrend around 2003–08 Downtrend since 2009 External demand Growth spells in trade partners (annual average Slowdown and import substitution: GDP growth, in order of country’s trade share): Cooling economies; installation of modern steel Russian Federation, 7 percent; China, 10.5 per- technologies in main export markets cent; ASEAN, 5 percent; Turkey, 6 percent Commodity prices Price boom: Price bust: Oil prices increase fourfold; iron ore, fivefold; Prices collapse after Global Financial Crisis steel, threefold (2008–09) without subsequent full recovery Input (energy) Implicit subsidies of oil and gas imports:a Expiration of implicit subsidies:a prices Annual average of around 8 percent of GDP; siz- Sharp price hike after Russian gas dispute in able domestic subsidies in coal and electricity 2009; most discounts eliminated by 2011 Source: Data from World Development Indicators database. Note: ASEAN = Association of Southeast Asian Nations. a. Implicit energy subsidies refer to the cost savings driven by discounted hydrocarbon imports from the Russian Federation as a share of Ukrainian GDP. On the external side, the booming economies of major trade partners boosted demand for Ukrainian exports, especially for steel, and soaring commodity pric- es increased Ukraine’s export revenues while raising input costs for the country’s competitors. In addition, significant subsidies and protection afforded to do- mestic coal and electricity production in Ukraine, as well as the largely conces- sional gas and oil imports from Russia (discounts amounting to 8 percent of GDP annually), shielded Ukrainian producers from rising input costs with commodity price hikes. Together, these factors reduced incentives for innovation and invest- ments, delaying an eventual structural transformation in the economy. However, these favorable conditions changed after the Global Financial Crisis of 2008–09. With anemic postcrisis recovery, demand from Ukraine’s major trade partners collapsed. This process also kindled a renewed import substitution drive, and old clients of Ukraine invested in modern steel technologies them- selves. On the supply side, a gas row with Russia brought subsidized imports to a halt, raising energy input prices. Amid this demand and a supply squeeze, Donbas—at the center of the legacy growth model—was hit disproportionately. As Ukraine unrolled one of the stark- est deindustrialization patterns in the world, the gravity center of domestic eco- nomic activity gradually shifted westward. Between 2004 and 2013, both Do- netsk and Luhansk Oblasts lost a sizable share of their population (7.1 percent and 8.3 percent, respectively), and their economies stalled (figure O.1). As a re- sult, Donetsk Oblast’s share of Ukrainian GDP decreased from 13 percent to 10.8 percent, and that of Luhansk Oblast decreased from 4.3 percent to 3.6 percent. Overview 19 Figure O.1 Population and Regional GDP Trends in Ukraine and Selected Regions, 2004–13 50 40 City of Kyiv, 2013 Change in regional GDP (%) 30 20 Ukraine, 2013 10 0 Positions Luhansk Donetsk in 2004 oblast, oblast, -10 (normalized) 2013 2013 -20 -10 -8 -6 -4 -2 0 2 4 6 8 Change in population (%) Donetsk Luhansk City of Kyiv Ukraine Source: World Bank calculations using State Statistics Service of Ukraine data. Note: Lines show the actual path followed by the two indicators: change in population and change in regional GDP. Therefore, the markers that identify 2013 values denote cumulative changes (the difference between 2013 and 2004 values). This downward trend in economic conditions was accompanied by a broader sociodemographic deterioration. Donetsk and Luhansk Oblasts, despite having previously favorable conditions compared with other regions, had relatively low scores by 2013 on the integrated index of human development produced by the State Statistics Service of Ukraine (SSSU 2014). This decline was driven by poor performance in three index components: demographic development (for exam- ple, lower birth rates); social environment (for example, high crime rates, alco- holism, and drug abuse); and comfort of living conditions (for example, pollution from mining and industry). As a result, Donbas—once a migration destination—has become a major source The Economics of Winning Hearts and Minds of migration, further worsening the region’s demographic aging. If not for certain obstacles that made migration more costly, like an outdated residence regis- tration system that tied people down and underdeveloped housing and credit markets that made housing unaffordable for migrants, the demographic outflow would be even more pronounced (Koettl et al. 2014). Conflict Onset: Magnifying Old Problems, Creating New Ones Before the conflict, Donbas was already suffering from diminishing prospects for its legacy growth model and from aging industry, infrastructure, and demogra- phy. With the onset of the conflict in 2014, these problems were magnified, and new problems were created. 20 Most directly, the conflict accelerated infrastructure depreciation by damag- ing strategic assets (destruction channel) and deepened demographic aging by displacing many people, especially the young and economically active (dis- placement channel). It also created many transactional frictions in a previously integrated region (disorganization channel), including conflict-driven risks and uncertainty. These frictions did not necessarily reduce the stock of productive assets directly, but they reduced the rate at which the economy uses any sur- viving assets. Unfortunately, given severe data constraints—also exacerbated by the conflict— it is difficult to produce a complete account of these effects. The official region- al statistics cover entire oblasts until 2014 but only the GCAs since 2014. This change in geographic scope prevents a direct comparison between the periods before and after 2014. Similarly, in many areas, on-the-ground assessments (in- cluding of damages and needs) are patchy and often outdated. Finally, official demographic statistics rely on residence-based registrations, which are likely to be misleading because they do not capture de facto movements that are not re- flected in residence records. Nevertheless, some useful insights can be derived from secondary sources, as summarized below by channel. Destruction The conflict, especially during its initial phase (2014–15), destroyed many stra- tegic assets, particularly in the energy, water and sanitation, and transportation sectors. In the absence of comprehensive data, we use nighttime light emis- sions to infer changes in infrastructure conditions. Between 2013 and 2019, light emissions decreased by 20.2 percent in the Donetsk GCA and by 28.1 percent in the NGCA (map O.1). In Luhansk Oblast, the GCA lost 7.2 percent of its light emissions, and the NGCA lost 42.9 percent. However, many intermittent damages that were fixed before 2019 are not cap- tured by these comparisons. For instance, during the initial phase of hostili- ties, several key power facilities were severely damaged. In spring 2014, about 20,000 shells put the Slovianska thermal power plant (TPP) out of service for nine months. Similarly, many essential water, sanitation, and hygiene (WASH) in- frastructure facilities near the contact line were subjected to frequent attacks, causing water service losses and shortages for both GCAs and NGCAs. One of the most prominent ones, the Donetsk Filter Station, has been damaged 71 times since 2016. Overall, the conflict has certainly significantly damaged the infrastructure. How- ever, our knowledge about the true extent of this damage needs to be updated with an on-the-ground assessment. Overview 21 Map O.1 Changes in Nighttime Light Emissions in Donetsk and Luhansk Oblasts, 2013–20 The Economics of Winning Hearts and Minds Net loss of light emissions (%) Donetsk GCA 20.2 Donetsk NGCA 28.1 Luhansk GCA 7.2 Luhansk NGCA 42.9 Source: World Bank estimations from US National Oceanic and Atmospheric Administration (NOAA) data. ©World Bank; further permission required for reuse. Note: Gray lines denote oblast boundaries. Within each oblast, the “contact line” (dark red) divides the government- controlled area (GCA) outside the line from the non-government controlled area (NGCA) inside the line. 22 Displacement The conflict also accelerated the region’s demographic aging problem by dis- placing a significant portion of the population, especially among the younger generations. At the peak of hostilities, in 2016, about 1.7 million people were reg- istered as IDPs, according to the Ministry of Social Policy of Ukraine. This number gradually fell to about 1.4 million people as of July 2020, about 45 percent of whom were residing in Donetsk and Luhansk Oblasts, according to the UN High Commissioner for Refugees (UNHCR). Besides the sheer scale of the displacement problem, the age and skill composi- tion of the displaced people is also consequential for regional economic dynam- ics. On average, IDPs in other regions are younger than in Donbas, according to National Monitoring System (NMS) surveys by the International Organization for Migration (IOM). Similarly, those who returned to their original locations (a rela- tively small group) have been older (average age 59.4 years) than the nonreturn- ees (37.5 years) (IOM 2019). A relatively lower share of the returnees have tertiary education, and a relatively higher share of them have vocational education. This suggests that individuals having region- or industry-specific education and skills could not find relevant jobs and sufficient pay for rented housing in their new locations. The NMS surveys find that IDPs face major challenges in finding employment because the skill composition of the population in industrialized Donbas differed from that of other locations that had relatively higher employ- ment concentrations in services and agriculture. Similarly, with rigid market con- ditions and strict residency regulations, IDPs report “lack of own housing” as another major challenge. Disorganization The conflict also triggered intangible effects that proved detrimental to the economy. The contact line dividing the region’s economy in two brought disrup- tions in connectivity, coordination problems, and a weakening social cohesion that posed challenges not seen before. Disruptions in connectivity. Before the conflict, Donetsk and Luhansk Cities were the primary urban hubs of Donbas as well as the transportation network’s centers of gravity. With the conflict, the contact line cut off this network, leading to connectivity islands and higher transportation costs, especially in railways (map O.2). The fragmentation in railroads also cut off some connections to Mar- iupol, the region’s primary seaport, reducing its cargo turnover by half (figure O.2). Similarly, major airports in Donetsk and Luhansk cities (Donetsk Sergei Prokofiev International Airport and Luhansk International Airport, respectively) are com- pletely destroyed. Other smaller airports in the GCAs could potentially be acti- Overview vated with significant rehabilitation. However, their proximity to the contact line restricts the viability of major projects as security threats continue, which would possibly increase if the airports were to be rebuilt. 23 Map O.2 Railroad Infrastructure and the Contact Line in Donbas, 2019 Rubizhne Lysychansk Sloviansk Schastya Kramatorsk Zolote Stanytsya-Luhanska Popasna Holubivka Bakhmut Luhanska Luhansk Pervomaisk Kadiivka Alchevsk Mayorske Myki vka Debaltseve Slovianoserbsk Herlivka Yenakiyeve Yasynovata Makiivka Vuglegirsk Khartsyzk Chys akove Donetsk Maryinka Novotroyitske Ilovaisk Railroad Railroad Station Hnutove Crossings Contact Line Mariupol Source: ©World Bank; further permission required for reuse. Note: The black outline denotes the area covered in the assessment. Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine. Coordination problems. The contact line also cuts through important public service provision systems, leading to a fragile interdependency. In both Donetsk and Luhansk Oblasts, for example, the contact line crosses important pipelines, pumping stations, and filtration stations, making the GCA and NGCA water sys- tems interdependent. Attacks on this critical infrastructure have taken place frequently, leading to service delivery disruptions. In the absence of better co- operation across the contact line, the entire system has been operating in a suboptimal mode. The Economics of Winning Hearts and Minds Furthermore, there are deep sectoral interdependencies (for example, a water-electricity-mining nexus), and government policies determine who gets the rent. When cheaper electricity is imported, local electricity generation is un- profitable, and without demand for coal from thermal power plants (TPPs), there is excess coal supply. When electricity imports are restricted, there is excess demand for coal, which needs to be imported at a high price (because importing directly from NGCAs is prohibited). These two cases—the water systems divi- sion and the water-electricity-mining nexus—provide textbook examples of how arbitrary barriers can disrupt economic systems. 24 Figure O.2 Cargo Turnover at the Port of Mariupol in Donetsk Oblast, by Cargo Type, 2013–18 16 14 Metric tons, millions 12 3.5 1.3 10 2.2 1.6 8 2.1 6 2.6 1.8 1.3 0.3 0.8 0.5 4 2 8.5 6.7 4.3 4.3 4.3 4.0 0 2013 2014 2015 2016 2017 2018 Metals Coal Clay Grains Other Source: Donetsk Regional State Administration 2019. Weakening social cohesion. The conflict has deepened mistrust and divisions in the country. The Social Cohesion and Reconciliation Index (SCORE) surveys conducted for this report by the Centre for Sustainable Peace and Democratic Development (SeeD) show a generally decreasing trend since 2015, both na- tionally and in the Donbas GCAs, in respondents’ exposure to other social and political perspectives. The presidential elections in 2019 provided an uptick in trust regarding central institutions in Donbas. However, deep suspicions about reforms and acute per- ceptions of corruption prevail—and have even increased in certain cases. In the Donbas GCAs, half to two-thirds of the respondents considered corruption to take place “sometimes” or “always” across a wide range of activities in 2019. The mistrust of and overall lack of support for reforms correlates with low aware- ness about the details of announced reforms. Therefore, any future recovery programs should pay close attention to building local ownership and pursuing a careful communication strategy. Economic Consequences of the “3 Ds” Together, these “3 Ds”—destruction, displacement, and disorganization—have crippled economic activity in Donbas since the onset of the conflict. Although it is not possible to pinpoint this effect exactly (before 2014, official regional GDP series cover entire oblasts, but after 2014, only the GCAs), we can approximate it by using nighttime light emissions, as noted earlier. Confirming the loss of activity indicated by the light emissions, SSSU sector- level data show that industrial output continued to shrink in both the Luhansk and Donetsk GCAs since 2015 (figure O.3). In the Donetsk GCA, this decline was accompanied by increasing productivity, but in the Luhansk GCA, productivity Overview decreased along with output. Agriculture provided the only bright spot, where both output and productivity increased in both GCAs. This sector, however, comprises only a small share of output and employment in these regions. 25 Figure O.3 Differences in Sectoral Outputs and Productivity in Ukraine and Donetsk and Luhansk Oblasts, before and after Onset of Donbas Conflict a. Industrial output, 2009–19 120 Whole oblast GCA only Output index (2010=100) 100 80 60 40 20 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Ukraine Donetsk Luhanks b. Industrial productivity, 2009–19 140 Productivity index (2010=100) Whole oblast GCA only 120 100 80 60 40 20 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Ukraine Donetsk Luhanks c. Agricultural output, 2011–19 150 140 Whole oblast GCA only Output index (2011=100) 130 120 110 100 90 80 70 60 2011 2012 2013 2014 2015 2016 2017 2018 2019 The Economics of Winning Hearts and Minds Ukraine Donetsk Luhanks d. Agricultural productivity, 2011–19 140 Productivity index (2011=100) Whole oblast GCA only 130 120 110 100 90 80 70 60 2011 2012 2013 2014 2015 2016 2017 2018 2019 Ukraine Donetsk Luhansk Sources: State Statistics Service of Ukraine database; World Bank calculations. Note: The graphs track sectoral output and productivity in Ukraine and in Donbas (the Donetsk and Luhansk Oblasts) before and after the armed conflict began in 2014. Pre-2014 statistics include all of Donetsk and Luhansk Oblasts (to left of dashed line), whereas 2014–19 statistics exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts (to right of dashed line). GCA = government-controlled area. 26 Structural and Conflict-Driven Obstacles to Future Growth Today, the Donbas region faces formidable obstacles against a swift recovery, and these obstacles are both structural and conflict-related. The structural problems have led to an aging industry and infrastructure that have eroded the region’s competitiveness internationally. For instance, in 2019, Ukraine used 444 kilograms of coke per metric ton of pig iron production. European countries used 20 percent less, on average: 359 kilograms per metric ton. This wedge is in part the outcome of long-standing policy distortions that re- duced incentives to modernize Ukraine’s industrial base. As subsidies and priv- ileges shielded firms from competition and systematic corruption reduced the entry of unconnected businesses, firms grew accustomed to seeking rents (which does not require technological upgrading) instead of profits (which does). Primary Bottlenecks The onset of the conflict, and the subsequent increase in risks and uncertainty, have only aggravated these problems. For instance, although the conflict has reduced capacity in steel production, without improvements in competitive- ness, additional capacity is deemed unnecessary. Together, the structural and conflict-driven factors have led to bottlenecks in three major areas (among oth- ers). Disorganization and risk. The conflict has increased the uncertainty and risk in an already unfavorable business environment suffering from chronic corruption and regulatory burdens. With further irregularities in supply, higher costs in key inputs (such as anthracite), bottlenecks in connectivity, and unfavorable exter- nal competition, profit prospects remain low and investments suppressed. The imposition of trade restrictions by Russia; logistical bottlenecks (with all major transportation routes to Russia’s southeast remaining in the NGCAs); and the Ukrainian government’s ban on large-volume trade with the NGCAs have all con- tributed to these trends. Access to finance. Enterprises in the GCAs were hit by deteriorating financial access. With sharply rising uncertainty and risks, most banks suspended opera- tions throughout Donbas. Credit fell from 34.6 percent of regional GDP in 2014 to just 10 percent of GDP in 2018 (less than a third of the national average), accord- ing to SSSU data. For corporations, the decline in lending was particularly steep, slumping from 23 percent of regional GDP in 2014 to just 4.2 percent in 2018, underscoring the scale of the drop in activity and demand for credit. Those who can borrow paid the equivalent of a 15 percent real interest rate as opposed to 5.6 percent nationwide, which effectively ceased credit to small Overview and medium enterprises (SMEs). This large and widening spread attests to the heightened risks that banks perceive in the region. 27 Figure O.4 Decomposition of Job Vacancies and Unemployment, by Occupation Type, in Ukraine and the Donetsk and Luhansk GCAs, December 2019 338,163 registered 59,018 10,803 registered 662 7,827 registered 599 unemployed people vacancies unemployed people vacancies unemployed people vacancies 100 14 12 11 11 11 15 13 14 16 18 75 21 34 Share of total (%) 12 20 20 9 23 2 1 1 5 14 50 1 9 15 14 23 14 7 4 3 5 11 4 11 4 11 25 10 4 11 11 10 9 7 17 12 12 6 14 16 6 6 9 8 0 Unemployed Vacancies Unemployed Vacancies Unemployed Vacancies Ukraine Donetsk Luhansk Without profession Skilled forestry, fisheries Specialists Operation or maintenance Trade and services workers Professionals of machines Skilled workers Technical officers Legislators, sr. civil servants, executives Source: World Bank estimates from State Employment Service of Ukraine data. Note: Regional and national data exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. GCAs = government-controlled areas. Skill mismatch. That the conflict hit industry more than other sectors has deep- ened labor market imbalances. A comparison of vacancies and unemployment profiles from the State Employment Service of Ukraine reveals an oversupply of basic machine operators and maintenance workers in the GCAs, especially in the Luhansk GCA (only 14 percent of the vacancies but 34 percent of the unem- ployed in this group) and an undersupply of skilled workers in both GCAs (figure O.4). This situation indicates a need for upskilling existing workers and also for The Economics of Winning Hearts and Minds providing more skills-intensive training in vocational education. Challenges in Labor Markets and Demography Currently, the GCAs in Luhansk and Donetsk Oblasts have the worst labor market conditions in Ukraine and the highest unemployment rates of all oblasts (table O.2). They also have the highest median age and old-age dependency ratios in Ukraine, according to SSSU data. This aging trend was already problematic be- fore the onset of the conflict, but it has been aggravated by a disproportionate displacement of younger generations toward other places in Ukraine. In the NMS surveys, the vast majority of IDPs report no intention to return in the near future despite facing difficulties in transferring skills and finding jobs— especially if they are young, male, live outside Donbas, and have been displaced for a long time (table O.3). This will further constrain economic growth opportu- nities in Donbas. 28 Table O.2 Main Labor Market Indicators in Ukraine and Donetsk and Luhansk GCAs, 2018 Old-age Labor force Unemployment rate dependency ratio participation rate Rank Value Rank Value Rank Value (ratio) Area (out of 25) (%) (out of 25) (%) (out of 25) Donetsk 0.325 24 58.9 24 13.6 24 Luhansk 0.331 25 68.1 1 13.7 25 Ukraine 0.254 n.a. 63.4 n.a. 8.2 n.a. Source: State Statistics Service of Ukraine data. Note: The 25 ranked regions are all oblasts. Ukraine also includes the City of Kyiv (not ranked). Both regional and national data exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. GCAs = government-controlled areas; n.a. = not applicable. Table O.3 Intentions of Working-Age Household Heads of IDPs to Return to Place of Origin in Donbas, by Region of Current Residence, 2018 Ukraine Intention to return Donetsk (GCA) Luhansk (GCA) Total (excl Donbas) Yes, in the near future 1.0 0.9 1.0 1.0 Yes, after end of conflict 33.2 31.4 18.1*** 25.9 Yes, maybe in the future 17.3 9.2*** 13.6** 14.1 No 29.3 25.8** 47.6*** 37.0 Difficult to answer, 19.1 32.7*** 19.8 22.0 or no response Source: World Bank calculations based on National Monitoring System (NMS) survey data of the International Organization of Migration. Note: Joint dataset is of IDPs from face-to-face and phone interviews (NMS rounds 9–12, weighted), based on their chosen answers to question 3.9: “Do you plan to return to the place of origin?” Information is provided only for the heads of house- hold, excluding IDPs from Crimea, individuals older than 70 years, and those who reported that they have already returned. Donbas comprises the Donetsk and Luhansk Oblasts. GCA = government-controlled area; IDPs = internally displaced persons. Significance level of difference in means between (a) Luhansk, and (b) Ukraine (excluding Donbas): *** = 1 percent, ** = 5 percent, * = 10 percent. Public Service Access Issues Conflict has affected both the quantity and quality of public services. Safety remains an important challenge, with military presence, shelling, and unexploded ordnance (UXOs) being common near the contact line. Infrastructure needs ma- jor rehabilitation. Both the providers and the receivers of public services need psychosocial support. Water and sanitation conditions remain problematic in rural areas (espe- Overview cially in Luhansk Oblast, where only 7 percent of rural settlements received water supply in 2018, according to SSSU data). The quality of water has his- torically been poor in the Donbas region, but it was worsened by the con- flict. Mine closures and flooding continue to contaminate water sources. 29 Figure O.5 Satisfaction with Public Services in Donbas GCAs and NGCAs, 2017–19 8 7.0 7 6.6 6.7 6.7 6.3 6.3 6.1 6.0 Score (0–10) 6 5.8 5.9 5.9 5.7 5.7 5.4 5.5 5.1 5.2 5.1 5.0 5.0 5 4.7 4.2 4.3 4 3.2 3 Roads Justice Health Public Administrative Utilities Social Basic services care transport services services (welfare) schooling Donbas GCAs 2018 Donbas GCAs 2019 NGCAs 2019 Source: Social Cohesion and Reconciliation Index (SCORE) surveys, Centre for Sustainable Peace and Democratic Develop- ment (SeeD). Note: Values are measured on a 0–10 scale, with 0 being no satisfaction and 10 being maximum satisfaction. Donbas com- prises the Donetsk and Luhansk Oblasts of eastern Ukraine, each of which is divided into a government-controlled area (GCA) and a non-government controlled area (NGCA). NCGAs were not surveyed in 2018. Figure O.6 Simulated Changes in Regional GDP in Donetsk and Luhansk Oblasts, by Conflict Scenario a. Status quo scenario 0 -32,2 -43.6 -5.9 -60.4 -2.7 -10 Change (percentage points) -20 -30 -3.5 The Economics of Winning Hearts and Minds -40 -5.1 -50 -60 -5.8 -70 GCA NGCA GCA NGCA Donetsk Luhansk Impact so far Future impact 30 b. Reintegration scenario 80 60 Change (percentage points) 40 20 27.6 39.6 0.9 54.8 0 -32.2 -43.6 -60.4 -2.7 -20 -40 -3.5 -5.1 -60 -5.8 -80 GCA NGCA GCA NGCA Donetsk Luhansk Impact so far Status quo effect Reintegration effect Source: World Bank estimations. Note: The “status quo” scenario assumes continuation of the Donbas conflict, and “reintegration,” a full postconflict recovery. GCA = government-controlled area; NGCA = non-government controlled area. Even the central water distribution system often does not meet the safety stan- dards: 20.6 percent of samples in the Donetsk GCA and 41.9 percent in the Lu- hansk GCA failed the test in 2018. Surveys detect a noticeable increase in satis- faction with publicly provided services in the Donbas GCAs between 2018 and 2019 (figure O.5); however, the absolute levels of satisfaction remain relatively low. Overall, the region faces multifaceted challenges. As in other decision-making processes, these issues will need to be prioritized for the highest policy impact. Furthermore, many of these challenges can change quickly, depending on the future trajectory of the conflict. Thus, for future policies to be effective, policy makers must incorporate this fluidity into ex ante policy design and adapt to materialized changes as needed ex post. Next, we turn attention to these future policies. The Case for Scaling Up Efforts in Donbas As the discussion has emphasized so far, the current challenges faced by Donbas stem from two distinct but related sources: structural problems and conflict-driven problems. The simulations in this report (described in box O.1) show that as the conflict continues—the status quo scenario—the Donbas re- gion will suffer more economic losses in the absence of policy interventions. Overview As the productivity shocks from the conflict linger, the GCAs in Donbas are ex- pected to lose another 3 percentage points of regional GDP in the medium term (about a decade) (figure O.6, panel a). 31 In contrast, a reintegration can bring about a “peace dividend”—an automat- ic partial economic recovery led by the removal of conflict-driven distortions (figure O.6, panel b). In the GCAs, these gains average about 17 percentage points in the medium term. Unsurprisingly, reintegration is projected to be particularly beneficial for the NGCAs, which are set to lose another 5.4 percentage points of regional GDP under the status quo scenario, on average, but can gain an average of 47 percentage points with reintegration. Note, however, that the peace dividend described here is only a partial rebound that takes place once the conditions improve under a reintegration scenario. This process, by itself, cannot alleviate the structural problems faced by the region before the conflict. In fact, it cannot even fully counter the conflict-driven effects in the medium term because some of those effects (for example, population displacement) are naturally persistent and will require active policies to mitigate. The Efficiency Argument for Targeting Donbas Given the region’s low economic productivity currently, is it efficient to use fi- nancial resources to boost growth in the Donbas GCAs? A standard “efficiency argument” from spatial economics literature would suggest that investments should target the areas with the greatest economic potential. In this case, wouldn’t Kyiv City, or other areas with promising economic patterns in recent years, be better alternatives to focus on? Then, once better returns to such in- vestments are gathered, the people in Donbas can be helped by means of other transfers (the “equity argument”). The simulations in this report suggest otherwise. Even without reintegration, scaling up policies in Donbas is better than investing elsewhere—for GDP in both the Donbas GCAs and in Ukraine as a whole. This is because the current produc- tivity gaps between the regions are small relative to the mobility costs between regions; thus, investments elsewhere cannot attract labor from low-productivity areas more effectively. It is important to note that, with different productivity gaps or mobility costs The Economics of Winning Hearts and Minds between regions, these results can change. If, for example, broader policy re- forms or better access to European markets increase productivity elsewhere in Ukraine, an “efficiency” argument that favors investing there can become rele- vant. In contrast, reintegration in Donbas can eliminate such concerns by closing the productivity gap between Donbas and the rest of Ukraine. 32 Box O.1 A Simulation Approach to Future Policy Design The forward-looking analysis in this report adopts a quantitative approach based on economic modeling. There are some good reasons for this choice: •• First, unlike descriptive case studies, the validity of this approach can be challenged by further evidence, logical or empirical. •• Second, this approach helps to ground the expectations for future trends by explicitly studying the relationship between policy actions, conflict conditions, and the behavioral responses of economic agents. •• Third, unlike simple empirical extrapolations, the model-based simu- lations can analyze new policies and dynamics under different shocks (for example, when past relationships change after a structural break such as conflict) by taking economic agents’ preferences into consid- eration. Another strength of the simulation model is its endogenous treatment of Ukraine’s migration patterns. That is, rational agents respond to changes in economic opportunities and nonmonetary factors like amenities by mov- ing to other regions when incentives are more aligned. This helps in the analysis of structural factors (for example, productivity gaps and mobility costs); conflict-driven effects (destruction, displacement, and disorgani- zation); and the effects of policy responses. The analysis focuses on three major groups of future policies: investments, transfers, and mobility cost reduction (figure BO.1.1). Investments (physical or human capital) increase labor productivity, transfers supplement wage income, and mobility cost reduction facilitates a better allocation of labor from low-productivity areas to high-productivity areas. Both investments and transfers can take three sizes: (a) status quo (no change from current levels); (b) a moderate increase (about 15 percent of the preconflict investments in Donbas, annually); and (c) a large increase (about 30 percent of the preconflict investments in Donbas, annually). They can also be geographically targeted: GCA only, GCA and NGCA, and Kyiv City (for comparison purposes). Finally, given major uncertainties about the future dynamics, we adopt a scenario approach to comparing alterna- tive future conflict trajectories: status quo, intermediate, and reintegration. Together, these amount to 864 different future paths that correspond to different combinations of these scenarios. Overview 33 Figure BO.1.1 Conflict and Policy Scenarios 1. CONFLICT SCENARIOS • Status quo • Intermediate • Reintegration 2. TRANSACTION COSTS • Mobility costs: Baseline or low • Jobs: Restricted or unrestricted public employment 3. INVESTMENTS • Scale: status quo, moderate, or large • Target: GCAs only or GCAs + NGCAs 4. TRANSFERS • Scale: Status quo, moderate, or large • Target: GCAs only or GCAs + NGCAs 864 alternative paths Source: ©World Bank. Further permission required for reuse. Note: In the conflict scenarios, the “status quo” assumes continuation of the Donbas conflict; “intermediate,” suffi- cient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “reintegra- tion,” a full postconflict recovery. GCAs = government-controlled areas; NGCAs = non-government controlled areas. This framework helps us to evaluate some important policy questions: The Economics of Winning Hearts and Minds •• What are the expected impacts of different policy instruments under current conditions? •• How would these effects change under different conflict conditions? •• Which policy is the most effective under each conflict scenario (con- sidering, for example, the impact on regional GDP)? •• What are the trade-offs, if any, between promoting growth in Donbas and promoting growth in Ukraine overall? •• Which policies are less prone to such trade-offs? Because future development policies will face budget constraints like all other policies, the answers to these questions can help prioritize policy actions to maximize development impact. 34 Principles of a Comprehensive Strategy Nuances like these highlight the importance of conducting a big push in Donbas, carefully and coherently. The region needs a comprehensive and integrated strat- egy (a complete and contingent plan) that can coordinate future interventions around a unified development path. Such a strategy would be, among other things, •• Balanced: no silver bullet. The strategy should explicitly weigh trade-offs be- tween different policies. Some policies like transfers are effective for boost- ing GDP in the GCAs but not in Ukraine overall (table O.4). Similar trade-offs also hold for investments and mobility cost reduction. Thus, a “silver bullet” approach (for example, only infrastructure investments) will not work. Al- though mixed policies do not deliver a first-best outcome for any objective, they still provide a second-best outcome for many objectives. Thus, they can help avoid undesirable trade-offs between objectives. Overall, the re- gion’s multifaceted problems call for a comprehensive approach. •• Nuanced: one size does not fit all. To be effective, policies need to differ between the conflict (status quo) and peace (reintegration) scenarios. For instance, simulations in this report show that lower mobility costs can largely boost Ukrainian GDP under the status quo by reallocating labor from low-productivity areas to high-productivity areas. However, with reintegra- tion, regional productivity gaps shrink and this effect is weakened. Therefore, interventions need to be calibrated accordingly: what is good under the sta- tus quo is not necessarily good in reintegration, and vice versa. A compre- hensive strategy should include an action plan for each case. •• Transformative: distortions do not fix distortions. The region’s structural prob- lems were sustained by long-standing, policy-driven distortions. Authorities should not introduce new distortions to offset the adverse effects of the old ones. Fixing market fundamentals and removing long-standing policies that hamper market contestability can work better. Without improving the business climate, rule of law, and anticorruption efforts, selective incentive schemes for attracting investors are likely to be exhausted by rent-seeking actors. Thus, removing such barriers should be considered a priority. Together, these principles can be used to benchmark development ideas for the region. For instance, the creation of special economic zones (SEZs) has recently been floated as a potential intervention. Although appealing theoretically, SEZs by themselves do not constitute a panacea for development—as lessons from over 300 SEZs worldwide show (Hyung-Gon 2016). International experience shows that, to be effective, SEZs must be embedded within a greater development framework (no silver bullet). This typically involves exerting maximum effort to (a) reduce regulatory burdens and corruption; (b) provide essential infrastructure within SEZs and in their hinterlands; (c) ensure the presence of a dynamic labor force with matching skills as facilitated by re- formed educational institutions; and (d) provide the broader elements of a com- Overview fortable living that can attract foreign investors. The exact configuration of these factors is determined by the broader socioeconomic factors, and the future dy- namics of conflict can shape them drastically in the case of Donbas (one size does not fit all). 35 Table O.4 Simulations for the GDP Impact of Policies: Policy Rank Switches between Two Scenarios in Eastern Ukraine a. Status quo scenario Donetsk GCA Luhansk GCA Ukraine Policy alternative Value Rank Value Rank Value Rank No policy -3.5 ** -2.7 ** 0.2 * Investments 23.6 ***** 10.9 *** 1.6 ***** Transfers 19.9 **** 25.9 ***** 0.5 ** Mobility -12.9 * -14.8 * 1.3 **** Policy mix 18.9 *** 18.4 **** 1.1 *** b. Reintegration scenario Donetsk GCA Luhansk GCA Ukraine Policy alternative Value Rank Value Rank Value Rank No policy 24.2 * -1.8 * 3.8 * Investments 58.7 *** 11.4 *** 5.6 ***** Transfers 78.7 ***** 42.2 ***** 4.7 ** Mobility 31.2 ** 7.5 ** 5.1 *** Policy mix 63.0 **** 23.2 **** 5.2 **** Source: World Bank calculations. Note: The table simulates, under each of two scenarios—status quo and reintegration—the percentage point change from current GDP under each policy alternative. The “status quo” scenario assumes continuation of the Donbas conflict; “Interme- diate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegra- The Economics of Winning Hearts and Minds tion,” a postconflict recovery. Stars denote effectiveness, from * (least effective) to ***** (most effective). GCA = government-controlled area. Finally, when established in isolation, and without the broader framework, the incentive structures put forth by the SEZs are unlikely to attract foreign investors but instead are likely to be captured by others, with no significant effect on the region’s economic trajectory (distortions do not fix distortions). 36 A Decision Tree Approach to Programming Recovery Given the abounding uncertainties and scenario-sensitivity of optimal policies, the recovery strategy should distinguish “contingent policies” from “no-regret policies.” For some public services, the efficiency of the system relies on its scale. This depends on population size (GCAs only versus GCAs and NGCAs) on the demand side, and on whether the infrastructure in the NGCAs can be used again, on the supply side. Reconfiguring the system only for the GCAs prema- turely can be costly if the NGCAs’ facilities open up, because two unlinked and redundant systems would be operating in parallel. However, delaying such re- configuration can prolong welfare losses if reintegration does not happen. In the end, policy decisions must weigh these two cases against each other by considering the likelihood of reintegration. If a policy is highly sensitive to such calculations, then it is a “contingent policy.” If it is not sensitive—that is, it is desirable under both the status quo and reintegration scenarios—then it is a “no-regret policy.” Separating the interventions in this way, and following a deci- sion tree approach (figure O.7) can help open areas for making progress. Contingent Policies: Front-Load the Low-Hanging Fruit Contingent policies change between the status quo and the reintegration sce- narios. But they can also change between different policy objectives. From a Ukrainian GDP perspective, investments in the Donbas GCAs play a more prom- inent role under both the status quo and reintegration scenarios. Reducing mo- bility costs between Donbas and the rest of Ukraine is effective in increasing Ukrainian GDP under the status quo because it helps people move to more- productive areas. But it is less important under reintegration as the productivity gaps shrink. In contrast, from a Donbas GCA GDP perspective, transfers are more effective under reintegration, and investments are more effective under the status quo (especially in the Donetsk GCA). Mobility cost reduction between Donbas and the rest of Ukraine reduces GDP in the Donbas GCAs, especially under the status quo. With these underlying mechanisms in mind, the analysis suggests several steps, described below. Transfers to address skill shortages in GCAs. In addition to no-regret policies (discussed below), the “status quo-contingent” transfers can include social as- sistance and insurance provisions, with the possibility of a carefully ring-fenced direct transfer scheme and supplemental incomes for workers, including hazard pay and a mobility premium in conflict-affected areas. This is similar to the “in- surance against conflict risks” for investors, as considered by the authorities. Overview The difference is that the hazard pay is paid to individuals up front, and the insurance is paid to businesses in the case of risk realization. Similarly, supple- mental assistance to IDPs can reflect ongoing hardships during active conflict, with additional support to gain skills for employability and access to housing. 37 Figure O.7 A Decision Tree Approach to Designing an Economic Recovery Strategy for Eastern Ukraine Recovery strategy p=0 p=1 Status quo Reintegration Investments Transfers Mobility Investments Transfers Mobility Contingent (status quo only) policies Contingent (reintegration only) policies No-regret policies Source: ©World Bank. Further permission required for reuse. Note: “Contingent” policies are specific to either the status quo (continued conflict) or reintegration (removal of all con- flict-driven distortions) scenarios. “No-regret” policies refer to those that would be desirable under both the status quo and reintegration scenarios (for example, improvements in business climate and housing market). Interventions to mitigate conflict-related risks. A safe school environment entails eliminating military presence, shelling, and UXOs at schools. Road safety and psychosocial support are also priorities. For businesses, public assistance in designing insurance mechanisms for conflict-driven risks can be considered; however, serious consideration should be given to moral hazard and monitor- ing problems, and alternative mechanisms (such as index insurance) should be evaluated. The authorities are currently evaluating these options and are advised to pursue them if those potential implementation issues can be addressed. In addition, public sector interventions to improve access to finance and simplify taxes (or breaks) for new SMEs are needed. (These fall into the “no-regret” cat- egory to some extent, but an active conflict may necessitate a greater, more The Economics of Winning Hearts and Minds targeted effort.) Investments for a contingent infrastructure strategy. All infrastructure invest- ments should be coordinated around a coherent infrastructure strategy aligned with economic objectives and conflict conditions for a “status quo only” reality. This process would include picking low-hanging fruit in (a) connectivity (railways extensions and roads) to eliminate current transportation bottlenecks; and (b) service delivery systems like water and sanitation services (while being mindful of possible redundancy in the case of reintegration). Attention should be paid to managing the quality of public investment projects in both the fiduciary sense and in terms of project selection rules based on economic returns and absorp- tive capacity constraints. Categorically, all policies for the NGCAs are contingent on reintegration. These include both major reconstruction projects (like the two main regional airports in Donetsk and Luhansk Cities) and other economic and social interventions aimed 38 at promoting growth. However, some reconstruction or rehabilitation projects for the GCAs will also need to be postponed until reintegration. These include major reconstruction projects (for example, Mariupol airport) that are deemed too costly in the absence of a broader service area or that risk attracting military attention. No-Regret Policies: Build Foundations for Economic Transformation Many of the potentially most beneficial and transformative policy interventions are desirable under both the status quo and reintegration scenarios. These are essential not only to mitigate the impact of the conflict but also to eliminate the long-standing policy-driven distortions that delayed structural transformation of the region’s economy before the conflict. These include, first and foremost, policies to improve the business climate. Reforms to eliminate regulatory burdens and corruption. The institutional framework in Ukraine, and particularly in Donbas, provides a weak foundation for building a robust economic recovery strategy. The World Bank’s current System- atic Country Diagnostic emphasizes governance, the rule of law, and anticorrup- tion reforms as critical elements of Ukraine’s economic future. This is particularly true in Donbas, where conflict has further weakened prevailing institutions. Although trust in institutions increased with the last election, residents in the east do not trust reforms. Perceptions of the accountability, responsiveness, and transparency of local authorities remain consistently low. Thus, any attempt to win hearts and minds should feature beneficiary feedback mechanisms, griev- ance redress processes, and transparency and accountability measures, with transgressions being addressed promptly. Policies to open up the housing market. The lack of sufficient housing, which is one of the main obstacles to more efficient interregional mobility in Ukraine, traps labor in low-productivity areas. Ukraine’s housing market problems are in- tertwined with land market problems, residence and registration regulations, and other institutional factors that overregulate a process that other countries sim- ply manage much more effectively. Addressing these institutional problems to promote a more dynamic housing market is crucial for both Donbas and Ukraine in general, with or without conflict. Investments to modernize education for jobs. Skill mismatches and de- mographic aging-driven labor supply problems limit growth at its foundation. Policies should align vocational schools with market conditions; boost adult (re)training; improve access to childcare facilities and family-friendly jobs (to activate women); and subsidize employment for disadvantaged groups (persons with disabilities, older people, and so on). Overview Involving employers in study program revisions would improve the employabil- ity of graduates, and increasing the focus on civic education and ethics in the curriculum could help foster opposition to corruption. In addition, increasing the effectiveness of the State Employment Service of Ukraine (including expansion 39 of its digital skills and jobs databases) could help increase labor market partic- ipation. Investments that target low-hanging fruits in infrastructure. Much of the in- frastructure in the GCAs needs rehabilitation (for example, schools, water and sanitation systems, and transportation systems). Rehabilitation of those, and improving environmental outcomes (such as eliminating pollution of water re- sources) are desirable under both the status quo and reintegration scenarios. New infrastructure should target the elimination of service delivery and connec- tivity islands and prioritize the projects with the most welfare impact and least risk of redundancy. Projects should be coordinated by a master plan that is firm- ly embedded in the broader recovery strategy, prioritized according to viability (subject to rigorous public investment management principles), and communi- cated well to the public. Better data to address knowledge gaps. In Ukraine, better data are need- ed urgently in three major areas: (a) demography, (b) physical damages and needs, and (c) regional statistics. First, official demographic statistics are res- idence-based and unlikely to reflect the situation on the ground. We recom- mend that a census be considered in the near future. Second, our knowledge about the conflict-driven damages, overall conditions of infrastructure systems, and service access needs are fragmented and in most cases outdated. A new round of damage and needs assessments is needed to inform the future recov- ery strategies (in addition possibly to a census of firms and UXOs). Finally, official statistics cover the entire oblasts of Donetsk and Luhansk before 2014 and only their GCAs since 2014. This often leads to misinterpretations. Backcasting the GCAs’ data series or forecasting NGCAs’ series can help to resolve this problem. A Final Word The Economics of Winning Hearts and Minds Finally, a word on implementation is in order. Reforms, to be successful, require local ownership. This is particularly important in Donbas, where trust in top- down reforms is very low, as discussed earlier. Therefore, extra care must be given to involve local authorities and beneficiaries in decision-making process- es. Commendably, the authorities have implemented consultation processes, but more is needed. Citizen-government engagement (facilitated as needed) and community consultation during the design and implementation of policies, reforms, and projects will also engender greater trust and ownership. It is also important that all Ukrainians feel they have a stake in the economic recovery of the east. The social cohesion data (as further discussed in chapter 2 of this work) suggest that the low levels of social proximity and high prevalence of negative stereotyping are key drivers of the marginalization, mistrust, neglect, and isolation expressed by majorities of survey respondents. Overcoming these challenges is not optional for winning hearts and minds. 40 Note 1 Estimates by the UN Office of the High Commissioner for Human Rights (OHCHR) in 2020, https://www.ohchr.org/Docu- ments/Countries/UA/29thReportUkraine_EN.pdf. References Donetsk Regional State Administration. 2019. “Development Strategy of Donetsk Region for the Period Up to 2027.” Krama- torsk, Ukraine. https://dn.gov.ua/ua/projects/strategiya-rozvitku-doneckoyi-oblasti-na-period-do-2027-roku. Hyung-Gon, Jeong. 2016. “Are Special Economic Zones a Panacea for Developing Countries? Lessons for Developing Coun- tries.” KIEP Opinions 88, July 19. Korea Institute for International Public Policy, Seoul. IOM (International Organization for Migration). 2019. “National Monitoring System Report on the Situation of Internally Dis- placed Persons, June 2019.” Report of the IOM Mission in Ukraine, Kyiv. Koettl, Johannes, Olga Kupets, Anna Olefir, and Indhira Santos. 2014. “In Search of Opportunities? The Barriers to More Efficient Internal Labor Mobility in Ukraine.” IZA Journal of Labor & Development 3 (1): 1–28. Kohut, Zenon E., Bohdan Y. Nebesio, and Myroslav Yurkevich. 2005. Historical Dictionary of Ukraine. Plymouth, UK: Scarecrow Press. Reznikov, Oleksii. 2021. “Russia Remains Unwilling to End Seven-Year Ukraine War.” Atlantic Council UkraineAlert (blog), January 9. https://www.atlanticcouncil.org/blogs/ukrainealert/russia-remains-unwilling-to-end-seven-year-ukraine-war/. SSSU (State Statistics Service of Ukraine). 2014. “Regional Human Development Index.” Annual statistical report, SSSU, Kyiv. Overview 41 Chapter 1 The Long Shadow of History and Geography M odern development of the Donbas region began in the late 19th century and continued under Soviet industrialization of the 1930s, during Nazi occupation in World War II, and throughout the postwar reconstruction years. In 1913, Donbas was producing 74 percent of the pig iron and 87 percent of the total coal output in the entire Russian Empire (Kohut, Nebesio, and Yurkevich 2005, 158). The region would continue to be the largest producer of coal in the Soviet Union until the 1960s and play a major role in Ukraine’s economic output, employment, and exports until the onset in 2014 of armed conflict in the eastern provinces (oblasts) of Donetsk and Luhansk. The intensive production of metals and coal also gave rise to ancillary industries. Heavy engineering facilities emerged in the cities of Luhansk and Kramatorsk and other industrial centers. Chemical industries, centered on coking byprod- ucts and rock salt, were established in Artemivsk and Sloviansk Cities. Manga- nese was mined and processed in Marhanets and Nikopol. Agricultural exports from Donbas were less significant than industrial output, particularly in Donetsk Oblast. Industrial specialization led to comparatively higher wages and attracted labor from other regions. Before the conflict, 15.7 percent of Ukraine’s working popu- lation (about 3.2 million Ukrainians) concentrated in Donbas, employed in the manufacturing and mining sectors.1 Steel, machinery, and coal exports as well as supply chain connections with the Russian Federation were well developed at the time of independence in 1991 and throughout the interceding years be- fore 2014. Donbas (comprising Donetsk and Luhansk Oblasts) became the most heavily settled region of Ukraine, hosting 20 percent of the nation’s urban set- tlements. Following independence, the region’s economy deteriorated significantly. Real wages declined by nearly 80 percent over periods of hyperinflation, and out- put fell to less than half the region’s preindependence gross domestic product (GDP). GDP then grew rapidly between 2000 and 2008, only to experience the shock of global recession that triggered the Ukrainian financial crisis between 2008 and 2010.² This volatility brought significant declines in GDP countrywide, but the impact in Donbas was particularly acute. With contractions in tradition- The Long Shadow of History and Geography al Eurasian markets—and as the region’s outdated production facilities limited its competitiveness in the wake of rapid changes in global markets—mines and other underperforming Soviet-era facilities were gradually forced to reduce their operations or to close.3 Despite these growing challenges, the region has managed to delay a more structural transformation thanks to geopolitical factors. After Ukraine’s indepen- dence, Donbas continued to enjoy certain geopolitical rents, including conces- sional pricing in energy imports from Russia and preferential export access to Eurasian markets. However, as Ukraine’s access to western markets improved gradually, the opportunity cost of this delayed transformation has increased for the entire country. Thus, even before the onset of the conflict, uncertainties about the region’s economic future loomed large. 45 The Legacy Growth Model and Its Demise After independence, the Ukrainian economy continued to use the capacities left from the former Soviet Union with relatively little upkeep, investment, or modernization. This economic model also relied on the abundance of local coal, whose price was kept artificially low, benefiting coal users and transferring value added to them. This was particularly true for the metallurgical sector, which re- corded its best postindependence performance in the years before the 2008– 09 Global Financial Crisis (GFC). Ukraine’s steel export volumes increased by 60 percent from 2000 through 2007—one of the largest increases among the major steel-making countries—and made up about 40 percent of all Ukrainian exports.4 By 2007, Ukraine became the world’s fourth largest exporter, behind only China, Germany, and Japan. Until 2008: The Model Holds This modus operandi worked well during the first decade of the millennium (until the 2008–09 GFC), thanks to the combination of several favorable factors:5 •• Strong external demand in Russia and the major emerging markets. Russia’s GDP grew by an annual average of 7 percent during 2000–08; China’s, by 10.5 percent; the Association of Southeast Asian Nations (ASEAN) countries’, by more than 5 percent; and Turkey’s, by close to 6 percent. Import demand in these countries rose even more, boosting Ukraine’s export markets by 13 percent over the same period. •• A global commodity boom that pushed demand and prices for key Ukrainian ex- The Economics of Winning Hearts and Minds ports (metals, iron ore, and fertilizers) to record highs. Oil prices quadrupled between 2000 and 2008, iron ore prices more than quintupled, and steel prices tripled. •• Access to concessionally priced Russian oil and gas (discounted by 35–40 percent), resulting in an implicit subsidy of 8 percent of GDP a year during 2000–08 (figure 1.1).6 This implicit subsidy not only cushioned the cost impact of the energy price jump7 but also boosted the competitiveness of key exports such as metallurgy and chemicals, both of which are energy intensive. •• Availability of subsidized domestic inputs—especially electricity and coal (partly via the budget, partly via arrears in wages)—as well as low domestic wages at that time. •• Numerous tax exemptions and the use of tax havens that maximized profits to own- ers of major industrial facilities. These profits, however, were not invested but rather shifted abroad. 46 Figure 1.1 Implicit Energy Subsidies as a Share of GDP in Ukraine, 2003–13 10 15 10 8 Subsidy as share of GDP (%) 5 6 GDP growth (%) 0 4 -5 2 -10 0 -15 -2 -20 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Crude oil Natural gas GDP growth (right scale) Sources: State Statistics Service of Ukraine, National Bank of Ukraine, and Haver Analytics databases; World Bank estimates. Notes: Implicit energy subsidies refer to the cost savings driven by discounted hydrocarbon imports from the Russian Feder- ation as a share of Ukrainian GDP. Figure 1.2 GDP Growth and External Demand in Ukraine, 2000–13 25 20 15 10 Change (%) 5 0 -5 -10 The Long Shadow of History and Geography -15 -20 -25 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Ukraine GDP Export markets Russia GDP GDP trend Sources: State Statistics Service of Ukraine, National Bank of Ukraine, and Haver Analytics databases; World Bank estimates. 47 2008–13: From Slump to Stagnation These conditions, however, changed with the advent of the GFC in 2008. With the reversal of all the growth drivers that had supported growth in the 2000s, Ukraine was hit hard. Real GDP slumped 15 percent in 2009 (figure 1.2), led by a 22 percent drop in export volumes. This was the deepest recession in Europe as well as in any major emerging market. The shock was partly cyclical (Russian im- ports plunging 30 percent that year), partly terms-of-trade related (commodity prices collapsing), partly political (the natural gas row with the Timoshenko gov- ernment erasing most of the implicit energy subsidies), but mostly structural. The cyclical rebound in global demand in 2010–11 returned the economic growth rate to positive, but the recovery was shallow and brief. In 2011, Ukraine entered a period of structural stagnation despite strong global synchronized growth, am- ple liquidity, strong risk appetite, and record-low global interest rates. The pre- crisis growth model had stopped working well before the conflict began in 2014, with the model’s past drivers no longer in place: •• Global growth never returned to the precrisis pace, and Russia entered a pe- riod of structural stagnation in 2012. Investment was curtailed by the need to repair balance sheets and by tightened financial markets regulation. •• The acceleration of trade that had accompanied global economic growth since the mid-1990s lost momentum in the post-GFC world. •• Financial flows sharply decreased, in line with the trade slowdown post- 2009. Private capital flows to emerging market and development economies (EMDEs) dropped from 6.5 percent of GDP in 2008 to 2.5 percent in 2018, and foreign direct investment (FDI) inflows from 3.2 percent to 1.5 percent, the lowest level since 1993. •• Commodity prices never recovered near their pre-2008 levels and since 2014 went into a sustained structural decline (figure 1.3). The Economics of Winning Hearts and Minds •• Import demand underwent major changes in emerging markets. Most of Ukraine’s previous main customers—such as China, the Arab Republic of Egypt, India, the Islamic Republic of Iran, Russia, Turkey, and the Gulf Coop- eration Council (GCC) countries—have stepped up import substitution and have sharply reduced their import demand. Many have built up significant export capacities that are more efficient and technologically advanced than those in Ukraine. •• Export opportunities were curtailed amid a combination of reduced import de- mand, stepped-up export capacities elsewhere, and ramped-up competi- tion. This was particularly evident in steel, Ukraine’s main export commodity. •• Access to Russian energy at concessional prices diminished following the 2009 natural gas dispute and was largely phased out after 2011. This significantly raised domestic production costs because of the prevalence of energy-in- tensive, obsolete technologies in Ukraine. 48 Figure 1.3 Global Commodity Price Trends, 2000–19 300 250 Price index (2005=100) 200 150 100 50 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Crude oil Natural gas Coal Steel Copper Iron ore Sources: Haver Analytics database; World Bank estimates. Structural Problems Revealed These exogenous factors have also revealed the structural problems in the Ukrainian economy. It is now well understood that the availability of cheap ener- gy reduced incentives to invest in more-efficient technologies and helped pre- serve the legacy Soviet equipment. Ukraine remains among the least efficient energy consumers in Europe and Central Asia, using about three times as much energy per unit of GDP as the regional median, even though the per capita con- sumption is only around the median (figure 1.4). That the country’s energy efficiency is much worse than its peers suggests an adverse economic structure and inefficient and obsolete facilities. Energy con- sumption in Ukraine improved only after the Russian implicit subsidies were The Long Shadow of History and Geography slashed (falling by 56 percent between 2009 and 2017 after remaining stagnant from 2000 to 2009), and energy intensity fell by 52 percent over the same peri- od, mostly after 2013 when Ukraine began paying world prices for energy. Although total domestic and foreign investment was never particularly high— approaching the Central and Eastern Europe (CEE) average of 23–25 percent of GDP in the second half of the 2000s—it has collapsed since 2009. Fixed investment averaged 16.4 percent of GDP during 2009–19, one-third less than in the rest of CEE, returning in 2019 to 21.6 percent of GDP (or roughly its 2004 level in real terms).8 The result has been a marked erosion of potential growth (figure 1.5).9 It has been accompanied by the conservation of inefficient and often obsolete technolog- ical processes, which has limited production to mostly semifinished products with low value added and made large swaths of Ukraine’s industry (particularly the metallurgy and chemical industries) unprofitable and less competitive than other major exporters like China, Russia, and Turkey. 49 Figure 1.4 Energy Intensity of CEE and CIS Countries, 2017 Per capita energy use (MMBTU per person) 250 200 CZE 150 SLV SLK POL BGR LIT 100 CEE median: 91.5 HUN SER LAT CRO EST TUR ROM BIH 50 CIS median: 60.4 ALB CEE median: 6.8 CIS median: 18.7 0 0 5 10 15 20 25 30 35 40 Energy use (BTU, thousands) per unit of GDP CEE CIS Sources: Energy Information Administration of the US Department of Energy; World Bank estimates. Note: Countries are labeled using ISO alpha-3 codes. BTU = British Thermal Units; CEE = Central and Eastern Europe; CIS = Commonwealth of Independent States; MMBTU = 1 million British Thermal Units. Economic policies sometimes aggravated the structural problems as well: •• Fiscal policy has been strongly procyclical—expansionary during good times and contractionary during bad times. Thus, instead of stabilizing growth, it has exacerbated its cyclicality, adding to demand pressures in good times and to the downturn in bad times. •• Incomes policy has been even more expansionary. Wage growth has remained strong irrespective of productivity. (The latter has been virtually flat since 2008, while real wages soared 50 percent.) The strong wage growth led to the demand pressures that contributed to the 2009 and 2014–16 financial crises as well as to the competitiveness loss, undermining Ukraine’s exports. The Economics of Winning Hearts and Minds •• Exchange rate policy also played a role. Competitiveness losses have been reinforced by a misplaced exchange rate policy, which sought to keep the Ukrainian hryvnia (Hrv) stable for long stretches followed by abrupt forced devaluations as financial pressures mounted. Relative unit labor costs (RULC) doubled as a result between 2000 and 2013, contributing to the decline in exports and the loss in market share.10 50 Figure 1.5 Potential GDP and Output Gap in Ukraine, 2001–20 800 15 Output gap as share of potential GDP (%) 700 10 GDP (2010 Hrv, billions) 600 5 500 0 400 -5 300 -10 200 -15 march 2001 july 2001 nov. 2001 march 2002 july 2002 nov. 2002 march 2003 july 2003 nov. 2003 march 2004 july 2004 nov. 2004 march 2005 july 2005 nov. 2005 march 2006 july 2006 nov. 2006 march 2007 july 2007 nov. 2007 march 2008 july 2008 nov. 2008 march 2009 july 2009 nov. 2009 march 2010 july 2010 nov. 2010 march 2011 july 2011 nov. 2011 march 2012 july 2012 nov. 2012 march 2013 july 2013 nov. 2013 march 2014 july 2014 nov. 2014 march 2015 july 2015 nov. 2015 march 2016 july 2016 nov. 2016 march 2017 july 2017 nov. 2017 march 2018 july 2018 nov. 2018 march 2019 july 2019 nov. 2019 march 2020 Gap (RHS) Output gap (right scale) GDP trend Sources: State Statistics Service of Ukraine; Haver Analytics databases; World Bank estimates. Note: Potential GDP growth is estimated based on Hodrick-Prescott (HP)-filtered quarterly series in constant 2010 prices. Hrv = Ukrainian hryvnias; SA = seasonally adjusted. The Preconflict Hit to Donbas Although the economic dynamics described above were relevant for the whole country, their magnitude and impact were the strongest in eastern Ukraine. The Donbas region made up a large share of the country’s mining, metallurgy, and coal extraction activities. Agriculture plays a much more subdued role in the Donbas economy, accounting for 6.8 percent of GDP in Luhansk Oblast in 2013 and 4.6 percent in Donetsk Oblast, compared with the Ukrainian average of 8.4 percent. Meanwhile, construction accounted for 4.5 percent of GDP in Donetsk Oblast but only 3.4 percent in Luhansk Oblast (averaging 3.7 percent country- The Long Shadow of History and Geography wide).11 The region’s employment composition also differed from the rest of Ukraine. Do- netsk Oblast, for example, had about 10 percentage points’ higher share of em- ployment in industry, and about 7 percentage points’ lower share in agriculture than the country as a whole in 2013 (figure 1.6). With limited domestic demand in Ukraine, industry in Donbas relied heavily on exports. Thus, both Donetsk and Luhansk Oblasts were the traditional export powerhouses of Ukraine, accounting for 35 percent of the country’s export earn- ings and carrying a net positive trade balance over the years, in contrast with other oblasts, including Kyiv City (figure 1.7). 51 Figure 1.6 Sectoral Structure of Employment in Ukraine, by Regional Group, 2013 Central 17.0 13.5 4.0 32.9 30.8 1.8 South 15.8 18.2 3.9 35.6 24.5 2.1 West 25.1 12.3 5.0 29.0 25.6 3.0 Kharkiv 13.5 19.4 4.3 35.4 26.0 1.5 East Luhansk 12.9 23.8 4.1 36.7 20.8 1.7 Donetsk 10.7 25.2 5.2 35.6 20.8 2.4 Overall Ukraine 17.5 16.0 4.4 33.5 26.3 2.2 0 10 20 30 40 50 60 70 80 90 100 Share of employed, aged 15–70 years (%) Agriculture Industry Construction Less-knowledge-intensive Knowledge-intensive Other types of economic services sectors activity Sources: State Statistics Service of Ukraine data; World Bank estimates. In addition, the traditional industries in Donbas were highly energy intensive, which was aggravated by their reliance on outdated technologies and equip- ment (with a depreciation rate of more than 80 percent). Therefore, in 2018, Do- netsk and Luhansk Oblasts were the most energy-dependent regions in Ukraine (figure 1.8). The energy efficiency in Luhansk Oblast, for example (at Hrv 9.8 per metric ton reference fuel), was about 15 times lower than in Odesa (Luhansk Regional Military and Civil Administration 2016). This dependence made Donbas particularly vulnerable to the turmoil in global demand and commodity prices that unfolded from 2008 onward. The Economics of Winning Hearts and Minds In 2004, Donetsk Oblast generated more than 13 percent of Ukraine’s GDP, and Luhansk Oblast another 4.3 percent. By 2013, however, their shares fell to 10.8 percent and 3.6 percent, respectively. In fact, Luhansk and Donetsk Oblasts made negative net contributions to Ukrainian GDP growth in 2013 (figure 1.9). Nevertheless, Donetsk Oblast remained the second largest contributor to the economy (after the City of Kyiv), Luhansk Oblast was the eighth largest, and to- gether these regions accounted for 15 percent of the country’s GDP and a similar share of employment. In terms of productivity, however, both oblasts fared worse. By 2013, Donetsk slipped to 4th place (behind Dnipropetrovsk, Poltava, and the City of Kyiv), and Luhansk to 10th. In both oblasts, productivity has been on the decline since 2007. This—along with the steady erosion of both oblasts’ shares in the econo- my—points both to long-term structural problems that deepened with the GFC and to the oblasts’ vulnerability to shifts in global commodity markets. 52 Figure 1.7 Merchandise Trade Balance in Ukraine, by Region, 2002–13 30 20 10 0 US$, billions -10 -20 -30 -40 -50 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Donetsk Luhansk Other Easternª City of Kyiv Other Total Sources: State Statistics Service of Ukraine and Haver Analytics databases; World Bank estimates. a. “Other Eastern” includes the oblasts of Kharkiv, Dnipropetrovsk, Zaporizhzhia, Poltava, Sumy, and Kherson. Figure 1.8 Energy Independence in Ukraine, by Region, 2018 Vinnytsya City of Kyiv 0.65 Volyn Chernihiv 0.60 Dnipropetrovsk Chernivtsi 0.55 Donetsk 0.50 Cherkasy Zhytomyr 0.45 0.40 Khmelnytskiy Zakarpattya 0.35 0.30 Kherson Zaporizhzhia 0.25 Kharkiv Ivano-Frankivsk The Long Shadow of History and Geography Ternopil Kyiv Sumy Kirovohrad Rivne Luhansk Poltava Lviv Odesa Mikolayiv Ukraine average Sources: State Statistics Service of Ukraine database; Donetsk and Luhansk regional strategies; World Bank estimates. Note: The radar diagram illustrates each region’s energy independence relative to others as well as the Ukraine country aver- age (dashed line). Index values are ranked from 0 (least energy independent) to 1 (most energy independent). 53 Figure 1.9 Contributions to Ukraine’s GDP Growth, by Selected Region, 2005–13 12 Contribution to GDP growth (percentage points) 10 8 6 4 2 0 -2 2005 2006 2007 2008 2009 2010 2011 2012 2013 Donetsk Luhansk City of Kyiv Average other Sources: State Statistics Service of Ukraine database; World Bank estimates. Broader Trends Overall, the preconflict dynamics in Donbas were part of a broader trend. An in- dependent Ukraine exhibited a gradual decoupling from its Soviet legacy, which at times was driven by external shocks like the GFC. In so doing, the country showed one of the starkest deindustrialization patterns in the world after 1994. By 2011, the manufacturing share of GDP regressed by 26 percentage points and the manufacturing share of total employment by 7 percentage points—a clear outlier to global trends (figure 1.10). This decoupling process was also accompanied by a shift in the gravitational center of economic activity subnationally. Between 2004 and 2013, both Lu- The Economics of Winning Hearts and Minds hansk and Donetsk Oblasts performed remarkably worse than other oblasts in average growth rates, which led other regions to catch up with Donetsk Oblast and Luhansk Oblast lagging further behind (figura. 2004 1.11). In contrast, Kyiv City reinforced its own favorable initial position by growing much faster than other regions. A comparison of rank-size (GDP) distribution of regions between 2004 and 2013 reveals interesting observations in this regard. In 2004, this distribution was too flat (that is, too many small regions or too few large regions) relative to the typ- ical alignment along the 45-degree line observed in market economies (figure 1.12, panel a), as indicated by Zipf’s Law-based analyses.12 This was likely reminis- cent of a planned economy in which place-based policies constrained market forces and agglomeration. By 2013, the rank-size distribution became steeper and converged with the 45-degree line, a process largely driven by the dispro- portionate growth in Kyiv City and slowdown in the east (figure 1.12, panel b). 54 Figure 1.10 Changes in Manufacturing Shares of Employment and GDP, by Country Income Level, 1994–2011 15 ETH Change in employment share (% points) 10 SEN KEN GMB CHN 5 OMN MWI NIC LKA TZA THA ZMB IND ALB ZAF NGA IDN MEX BRA SYR GHA 0 TTO BHS GIN MDA PHL VEN BWA ABW POL MNG JAM ITA MAR PAN BRB EGY SWE SLV MYS CAN BLR ECU CHEUSA KGZ HND AZE AUS CHL NLD –5 NZL CRIURY PRY ISL UKR BOL FRADNK ESP ARG ISR DEU CUB IRL KOR DOM CYP JPN RUS GBR PER MDV COL –10 SGP SVN MUS –15 HKG –20 –30 –25 –20 –15 –10 –5 0 5 10 15 Change in MVA share (% points) HIC UMC LMC LIC Source: Hallward-Driemeier and Nayyar 2018. Note: Countries are labeled using ISO alpha-3 codes. Manufacturing value added (MVA) data are from the United Nations Industrial Development Organization (UNIDO) MVA database 2017. Country income classifications are defined by the level of per capita income in 1994. HIC = high-income country; LIC = low-income country; LMC = lower-middle-income country; UMC = upper-middle-income country. Figure 1.11 Regional GDP Convergence or Divergence in Ukraine, 2004–13 10.5 Losing ground Fast ahead Mean growth= 2,6% Real GRP growth (CAGR, %) 10.0 9.5 Donetsk City of Kyiv 9.0 Averages Mean GRPpc= 8,8 (Hrv 6.400) 8.5 The Long Shadow of History and Geography Luhansk 8.0 Low and slow Catching up 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Real GRP growth (CAGR, %) Sources: State Statistics Service of Ukraine; World Bank estimates. Note: Bubble size corresponds to relative size of regional GDP. CAGR = compound annual growth rate; GRP = gross regional product (that is, gross regional GDP); pc = per capita. Another effect of the decoupling was in real wage dynamics. Figure 1.13 shows the evolution of real wage distribution by region (normalized in each year to re- move scale effects). Between 1995 and 2013, two main patterns are (a) the di- vergence of wages in Kyiv City from those in the rest of the country, and (b) a compression of wage dispersion among other regions. In fact, the standard de- viation among all regions remained roughly equal between 1995 and 2013 (0.20 and 0.19, respectively). However, when Kyiv City is excluded, it decreased from 0.18 to 0.11 in the same time frame. 55 Figure 1.12 Rank-Size GDP Distribution of Ukrainian Regions, 2004 versus 2013 a. 2004 5.0 City of Kyiv 4.5 Donetsk 4.0 3.5 Luhansk LN(GRP) 3.0 2.5 2.0 1.5 SMSE: 7.1 Slope: -0.85 1.0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 LN(rank) b. 2013 5.0 City of Kyiv 4.5 Donetsk 4.0 3.5 LN(GRP) Luhansk 3.0 2.5 2.0 1.5 SMSE: 3.5 Slope: -0.87 1.0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 LN(rank) Sources: State Statistics Service of Ukraine database; World Bank estimates. Note: SMSE refers to sum of mean squared errors around the 45-degree line. LN = natural logarithm; GRP = gross regional product (gross regional GDP). This shift away from the previous center of productive capacity and the wage The Economics of Winning Hearts and Minds gap also had social consequences, with status panic, resentment, and an in- creasing sense of persecution among principal actors, parties, unions, and oth- ers in the east becoming more common. We turn to this dimension next. Social Corollaries In a world where economic and social processes are interdependent, Ukraine is no exception. The country’s legacy growth model and the gradual but delayed dissolution of that model have been accompanied by corresponding social dy- namics. This section focuses on the preconflict patterns in human development and social context. 56 Figure 1.13 Average Monthly Wages in Ukraine, by Region, 1995–2013 2.0 Average wage (index, normalized annually) 1.8 1.6 1.4 1.2 1.0 0.8 0.6 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1 St.Dev exc. City of Kyiv 1 St.Dev Donetsk Luhansk City of Kyiv Ukraine average Sources: State Statistics Service of Ukraine database; World Bank estimates. Human Development The well-being of individuals and societies have both income and nonincome determinants. Although the former often receives much attention for provid- ing relatively more measurable and commonly understood statistics, the latter may be equally important if not more so. However, the problem with assessing the latter is not that there are too few indicators; it is that there are too many, and each indicator is incomplete in different ways.13 Nonetheless, it is particu- larly important to consider the available information for Donbas—in this case, the regional human development index produced annually by the State Statis- tics Service of Ukraine (SSSU)—because income and nonincome dimensions of well-being vary considerably in this region. The Long Shadow of History and Geography Before the conflict, in 2013, Donetsk and Luhansk Oblasts had relatively low scores on this integrated index despite their initially favorable economic con- ditions relative to most other regions (SSSU 2014b). This discrepancy resulted predominantly from poor performance in three index components: demograph- ic development, social environment, and comfort of living conditions (figure 1.14). Underlying the low scores in Donbas were multiple long-standing issues. For ex- ample, Donetsk and Luhansk Oblasts had serious environmental problems, with exceptionally high pollution from the mining, metallurgy, chemical, and other heavy industries concentrated in these oblasts. Social problems such as high crime rates and large incidences of tuberculosis, alcoholism, and drug abuse could be partly explained by lifestyle peculiarities of the residents of these re- gions. They can also be explained by ineffective regional management of so- cial and economic infrastructure development, especially in “company towns” where economic activity was driven by large industrial enterprises but whose enterprises were either closed or downsized during the 1990s (Kupets 2009). 57 Figure 1.14 Integrated Index of Regional Human Development in Ukraine, by Dimension, 2013. Demographic Social Comfort Decent Aggregate Development Environment of Living Education Work Index Volyn Lviv Crimean AR Donetsk Zaporizhzhia Kharkiv 1 0.86 0.74 0.72 0.79 0.65 4.21 Ternopil Ternopil Zakarpattia Kharkiv Kharkiv Chernivtsi 2 0.74 0.69 0.78 0.85 0.65 4.16 Kyiv Chernivtsi Odesa Dnipropetrovsk Chernivtsi Lviv 3 0.73 0.65 0.69 0.85 0.63 4.01 Rivne Kharkiv Kyiv Zaporizhzhia Mykolaiv Crimean AR 4 0.72 0.63 0.84 0.62 4.0 0.64 Poltava Ivano- Frankivsk Volyn Vinnytsia Luhansk Zakarpattia 5 0.72 0.61 0.62 0.84 0.61 3.98 Khmelnytsky Rivne Lviv Khmelnytsky Chernihiv Zaporizhzhia 6 0.62 0.84 0.61 3.91 0.72 0.57 Ivano-Frankivsk Zakarpattia Kherson Poltava Poltava Kyiv 7 0.72 0.54 0.62 0.83 0.61 3.87 Chernivtsi Volyn Chernivtsi Odesa Donetsk Poltava 8 0.62 0.83 0.6 3.83 0.72 0.53 Lviv Khmelnytsky Rivne Lviv Dnipropetrovsk Odesa 9 0.71 0.51 0.6 0.83 0.6 3.82 Crimean AR Vinnytsia Kharkiv Luhansk Crimean AR Ternopil 10 0.71 0.47 0.6 0.83 0.6 3.77 Vinnytsia Sumy Poltava Kyiv Cherkasy Cherkasy 11 0.46 0.59 0.83 0.58 3.77 0.7 Kharkiv Zhytomyr Mykolaiv Kirovohrad Kyiv Volyn 12 0.7 0.45 0.59 0.83 0.58 3.74 Rank Cherkasy Cherkasy Kirovohrad Chernihiv Khmelnytsky Mykolaiv 13 0.7 0.45 0.59 0.83 0.57 3.74 Zakarpattia Poltava Cherkasy Sumy Zakarpattia Khmelnytsky 14 0.69 0.44 0.59 0.82 0.56 3.73 Sumy Kyiv Ternopil Mykolaiv Odesa Dnipropetrovsk 15 0.69 0.57 0.82 0.56 3.72 0.42 Zaporizhzhia Donetsk Zhytomyr Kherson Kirovohrad Luhansk 16 0.41 0.56 0.82 0.55 3.7 0.68 Odesa Crimean AR Luhansk Crimean AR Vinnytsia Chernihiv 17 0.41 0.55 0.82 0.55 3.69 0.68 Dnipropetrovsk Chernihiv Dnipropetrovsk Cherkasy Sumy Vinnytsia 18 0.68 0.41 0.55 0.82 0.55 3.65 Zhytomyr Zaporizhzhia Zaporizhzhia Chernivtsi Lviv Rivne 19 0.4 0.53 0.81 0.55 3.65 0.67 The Economics of Winning Hearts and Minds Mykolaiv Odesa Vinnytsia Zhytomyr Volyn Donetsk 20 0.67 0.38 0.53 0.79 0.54 3.64 Luhansk Mykolaiv Chernihiv Volyn Zhytomyr Ivano-Frankivsk 21 0.67 0.36 0.53 0.79 0.54 3.63 Kherson Dnipropetrovsk Sumy Ternopil Kherson Sumy 22 0.66 0.52 0.78 0.54 3.6 0.36 Chernihiv Luhansk Khmelnytsky Rivne Ternopil Kirovohrad 23 0.66 0.35 0.51 0.76 0.54 3.56 Kirovohrad Kirovohrad Ivano-Frankivsk Ivano-Frankivsk Ivano- 0.52 Zhytomyr 24 0.64 0.35 0.51 0.76 Frankivsk 3.5 Donetsk Kherson Donetsk Zakarpattia Rivne Kherson 25 0.63 0.35 0.46 0.73 0.52 3.5 Source: SSSU 2014b. Note: The regional human development index is produced annually by the State Statistics Service of Ukraine (SSSU). Since 2012, the index has comprised 33 indicators clustered under six dimensions, as listed in notes a.–f. below. The aggregate index is an unweighted sum of scores (on a 0–1 scale) in these six dimensions. Kyiv denotes Kyiv Oblast. Crimean AR = Autonomous Republic of Crimea. a. “Demographic development” encompasses fertility, mortality, and life expectation estimates. b. “Social environment” considers statistics for crime; disease (tuberculosis); teenage fertility; and self-harm (suicide). c. “Comfort of living” considers the availability of housing; certain amenities (capacity of the outpatient clinics); and public services. d. “Education” considers indicators of educational attainment (enrollment ratios) and quality (common test scores). e. “Decent work” considers labor market statistics and other measures like the ratio of the average and minimum wages and the coverage of social insurance. f. “Well-being” considers multidimensional poverty indicators (not shown in the graph because of space constraints). 58 Demographic Development Demographically, Donetsk and Luhansk Oblasts performed badly, with some of the country’s highest rates of depopulation (figure 1.15) because of low birth rates, high morbidity and mortality rates, and negative net migration. Between 1992 and 2013, Luhansk and Donetsk Oblasts lost 22.2 percent and 18.8 per- cent of their populations, respectively, compared with 11.4 percent in the rest of Ukraine. Luhansk and Donetsk Oblasts also had among the oldest populations in Ukraine as of January 2013: the median ages of the resident population were 42.1 years and 41.9 years, respectively, compared with 39.7 years in Ukraine as a whole, according to SSSU data. Despite high urbanization and industrialization, which is usually associated with younger populations, the old-age dependency ratio also was higher in Donetsk and Luhansk Oblasts than in many other regions. Overall, in 2013, Donetsk Oblast had the worst demographic development score in Ukraine, and Luhansk Oblast ranked 21st out of 24 oblasts and Crimea (SSSU 2014b). Education In education, Donetsk was the leading region in 2013 (SSSU 2014). Given its vari- ety of higher educational institutions and jobs for high-skilled professionals and associate professionals, Donetsk was long one of the major educational hubs in the Soviet Union and subsequently in Ukraine. This status is reflected in the high educational attainment of the adult population of Donetsk Oblast relative to most other regions in Ukraine. With over half of its total and employed pop- ulations aged 25–70 years having tertiary education, Donetsk ranked third in Ukraine on that measure in 2013 (appendix A, table A.1). In Luhansk Oblast, the share of adult population with tertiary education—46.5 percent of total popula- tion and 49.5 percent of the employed population—was lower than in Donetsk and in Ukraine as a whole, but the oblast still outperformed many other Ukrainian regions. The Long Shadow of History and Geography Notably, vocational education has played an important role in Donbas. Among the ninth-grade graduates (lower secondary education) in May 2013, 22.3 per- cent in Luhansk Oblast and 17.3 percent in Donetsk Oblast were enrolled in tech- nical and vocational education and training (TVET) institutions, compared with 16.2 percent in Ukraine as a whole (SSSU 2014a). Moreover, these two oblasts were leaders in the number of TVET institutions (111 in Donetsk Oblast and 78 in Luhansk Oblast), accounting for nearly 20 percent of all TVET institutions in Ukraine (SSSU 2014a). In addition, Donetsk Oblast was Ukraine’s leading region in on-the-job vocational training and skills upgrading of workers in 2013, accounting for 18.2 percent and 14.8 percent, respectively, of all such trained workers in Ukraine (NASU 2015). Luhansk Oblast ranked third in on-the-job vocational training (6.8 percent) and fifth in skills upgrading of workers (5.8 percent). These good performances were mainly because of the internal training of workers by large metallurgical and ma- chine-building enterprises. 59 Figure 1.15 Population Change in Donetsk and Luhansk Oblasts, 1989–2013 a. Donetsk 40 20 Change in population, thousands 0 -20 -40 -60 -80 93 1 10 11 12 13 89 0 90 91 94 95 96 97 98 99 2 3 4 5 6 7 8 9 92 0 0 0 0 0 0 0 0 0 20 0 20 20 19 20 20 19 19 19 19 20 19 19 19 19 19 20 20 20 20 20 20 19 20 20 Net migration Births minus deaths Total b. Luhansk 40 20 Change in population, thousands 0 -20 -40 -60 -80 93 10 11 12 13 89 95 96 97 98 99 0 1 2 3 4 5 6 7 8 9 90 91 92 94 0 0 0 0 0 0 0 0 0 20 0 20 20 19 20 20 19 19 19 19 20 19 19 19 19 19 20 20 20 20 20 20 19 20 20 The Economics of Winning Hearts and Minds Net migration Births minus deaths Total Source: State Statistics Service of Ukraine database (in Ukrainian only), http://database.ukrcensus.gov.ua/MULT/Dialog/stat- file_c.asp. Labor Market Labor market conditions in Donbas were relatively favorable before 2008. Im- proving external conditions and growing investment within Ukraine kept unem- ployment in Donbas relatively low compared with many other regions of Ukraine. By 2013, the unemployment rates in Luhansk and Donetsk Oblasts (6.2 percent and 7.8 percent, respectively) were comparable to the Ukrainian average (7.2 percent), according to SSSU Labor Force Survey (LFS) data (appendix A, table A.1). Unemployment was mainly structural since local employers sought skilled craft and related trades workers (such as miners, electricians, mechanics, weld- ers, and turners), whereas job seekers were looking for nonmanual jobs in market services (Kupets 2009). 60 On the other hand, the negative demographic and social situation in the region decreased the number of people in the labor force and their share of the to- tal working-age population. Between 2000 and 2013, the number of individuals aged 15–70 years in the labor force shrank by 10.5 percent in Donetsk Oblast and by 5 percent in Luhansk Oblast, compared with 3.7 percent in Ukraine as a whole. Moreover, the labor participation rate in Luhansk Oblast (63.3 percent) was among the lowest in Ukraine in 2013, the LFS data show (appendix A, table A.1). On the SSSU’s human development index, Donetsk and Luhansk Oblasts scored close to the average in the “decent work” component. This may seem surpris- ing given their relatively low unemployment rates, fast wage growth relative to regional GDP per capita (figure 1.16), and widespread social insurance and gen- erous fringe benefits for a large share of workers. However, a major factor that counterbalanced these advantages concerned the harmful working conditions (occupational risk) that affected almost 50 percent of the workers in Luhansk Oblast and 44 percent in Donetsk Oblast, compared with 29.5 percent in Ukraine (NASU 2015). In addition, the region accounted for over half of the total wage arrears in Ukraine, owing to traditionally high wage arrears in coal mining (Kupets 2009). Income versus Quality of Life Overall, Donetsk and Luhansk Oblasts were among the leading regions in Ukraine in terms of material well-being measured by regional GDP per capita, household disposable income and purchasing power, and the share of households having the basic set of durable goods (NASU 2015), especially before the GFC. Yet these favorable conditions were at odds with the oblasts’ poor performance in overall human development, suggesting that their commodity-fueled economic wealth did not necessarily translate into better lives. As a result, in 2013, Donetsk and Luhansk Oblasts ranked in the bottom half of the SSSU’s aggregate human de- velopment index (SSSU 2014). The income profiles of households in Donbas also illustrate this observed trade- The Long Shadow of History and Geography off between income and quality of life. Like in other regions of Ukraine, wage in- come was the main source of household income in Donetsk and Luhansk Oblasts (figure 1.17). However, their shares of pension benefits in total household income were much higher than in other regions of Ukraine in 2013: 28.8 percent and 29.4 percent in Donetsk and Luhansk, respectively, compared with the Ukrainian av- erage of 23.4 percent. Two factors drove this disparity: First, given their higher average ages and old- age dependency ratios, Donetsk and Luhansk Oblasts had higher shares statis- tically of retirement-age people. Second, a relatively larger share of workers in those oblasts retired earlier because many had lost their ability to work owing to injuries and occupational diseases (NASU 2015, 5). At the same time, the shares of social assistance, income from subsistence farming, and private monetary transfers were much smaller in Donbas than in Ukraine as a whole. 61 Figure 1.16 Growth of Nominal Wages Relative to Regional GDP Per Capita in Donetsk and Luhansk Oblasts, 2004–13 a. Donetsk 600 500 400 Index (2004=100) 300 200 100 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Wage b. Luhansk 600 500 400 Index (2004=100) 300 200 100 0 The Economics of Winning Hearts and Minds 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Wage Sources: State Statistics Service of Ukraine database; World Bank calculations. The Role of Social Cohesion The last item in the list above relates to the concept of social cohesion—the existence of strong bonds in a society, including cooperation among individuals and groups and a shared perception of inclusion in economic, social, and politi- cal processes (Boarini et al. 2018). As such, social cohesion can be a determinant of both economic growth and demographic mobility. Under certain conditions, group identities can play more dominant roles (à la Akerlof and Kranton 2000), leading to fragmentations, introducing significant frictions in economic process- es, and reducing mobility. 62 Figure 1.17 Sources of Household Incomes in Donetsk and Luhansk Oblasts Compared with Rest of Ukraine, 2013 Donetsk 53.0 28.8 3.9 2.1 12.2 Luhansk 49.7 29.4 5.8 2.2 12.9 Ukraine excluding 50.8 23.4 2.6 3.2 20.0 Donbas 0 10 20 30 40 50 60 70 80 90 100 Share of total household income (%) Wages and Pension Use of savings Social assistance Other salaries or loans and subsidies Sources: State Statistics Service of Ukraine database; World Bank estimates. Note: “Other” income sources include entrepreneurial activities, sale of agricultural goods, and private monetary or nonmon- etary transfers. In Donbas, most of the population speaks Russian. According to the most re- cent census (in December 2001), nearly 75 percent of the population in Donetsk Oblast and 69 percent in Luhansk Oblast reported that their mother tongue was Russian (figure 1.18). These numbers have almost certainly changed since 2001, but this is not possible to verify. However, studies have not found language dif- ferences to be an important determinant of interregional migration patterns be- fore 2013 (Kupets 2014). Overall, with relatively small differentials of income (with cost-of-living adjust- ments and balanced by other amenities like environmental quality and living standards) and regardless of language differences, incentives to move away from extended families and neighborhoods were not substantial in preconflict The Long Shadow of History and Geography Donbas. A more recent study found that, in addition to these factors, people who do not live at their registered domicile are more likely than those who do to have difficulty accessing public services, mainly reflected in the need to pay more for such services (Slobodian and Fitisova 2018).14 Limitations of Metrics An important shortcoming of many conventional human development indicators is that they do not necessarily incorporate individuals’ subjective assessments. Preferences differ over various dimensions of well-being, including the nonin- come aspects of human development, and can be influenced by the identity problem under the social cohesion umbrella, as the next section discusses fur- ther. 63 Figure 1.18 Mother Tongues in Ukraine, by Region, 2001 Sevastopol Crimean AR Donetsk Luhansk Zaporizzhia Kharkiv Odesa Dnipropetrovsk Ukraine Mykolaiv Kyiv City Kherson Sumy Chernihiv Kirovohrad Poltava Kyiv (oblast) Cherkasy Zhytomyr Chernivtsi Vinnytsia Khmelnytsky Lviv Zakarpattia Rivne Volyn Ivano-Frankivsk Ternopil 0 10 20 30 40 50 60 70 80 90 100 Share of population (%) The Economics of Winning Hearts and Minds Ukrainian Russian Other Source: 2001 census data, State Statistics Service of Ukraine, http://database.ukrcensus.gov.ua/MULT/Database/Census/ databasetree_uk.asp. Note: Crimean AR = Autonomous Republic of Crimea. Therefore, it is not sufficient to measure only the objective indicators of well-be- ing; we also must consider subjective assessments. The chapter turns next to this issue. Absent a historical, systematic, subjective well-being assessment, we consider the next best option: the preferences revealed through migration de- cisions. The idea here is that we can infer how individuals weigh interregional differences in income and nonincome factors, including political identities, from their migration patterns. 64 The Role of Preferences: When People Vote with Their Feet The Donbas region had become a major source of internal migration even be- fore the conflict. According to administrative statistics on population registra- tion by place of permanent residence,15 Luhansk Oblast was the leading region in Ukraine for total population losses due to internal migration between 2002 and 2013. This negative trend worsened starting in 2009, and in 2013 the oblast’s net population losses equaled the previous peak of 2004 (figure 1.19). Donetsk Oblast also faced a serious out-migration of population to other regions of Ukraine, especially in 2012–13. These trends were partially offset by a net positive international immigration, especially in Donetsk Oblast (figure 1.20). Luhansk Oblast had a net negative international flow in the 2000s, which turned to strong positive inflows in 2012 and 2013. Like in the rest of Ukraine—where about 60 percent of migration flows took place within the same administrative region—migration was largely a localized phenomenon in Donbas.16 Interregional migration, according to SSSU data, con- stituted 20 percent and 24 percent of total inflows and 25 percent and 34 per- cent of total outflows in Donetsk and Luhansk Oblasts, respectively, between 2002 and 2013. However, there were also differences between Donbas and the rest of the coun- try. Elsewhere in Ukraine, migration within and between regions led to an increase in urbanization. In contrast, the urban bias in net migration to other regions re- duced the urban share of total population in Donbas (figure 1.21). Although intra- regional and international migration (figure 1.20) increased the region’s urbaniza- tion, these gains were not large enough to offset the negative urbanization effect of interregional migration. The Long Shadow of History and Geography The bulk of interregional migration flows in Donbas took place between neigh- boring regions, Kyiv City, and Kyiv Oblast—a common pattern in Ukraine. Specifi- cally, over 40 percent of inflows to and 35–40 percent of outflows from Donetsk Oblast were within neighboring oblasts to the west (Kharkiv, Dnipropetrovsk, and Zaporizhzhia), and 15–20 percent of the inflows and outflows were with the Luhansk Oblast. Migration flows to and from Luhansk Oblast were even more concentrated within Eastern Ukraine: 36 percent of all inflows to Luhansk were from Kharkiv, Dnipropetrovsk, and Zaporizhzhia, and over 32 percent were from Donetsk Oblast. The share of other oblasts in interregional inflows to Donetsk and Luhansk Oblasts declined with distance. 65 Figure 1.19 Net Migration within Ukraine, Donetsk and Luhansk Oblasts, 2002–13 0 -500 -1,000 -1,500 Net registered inflows -2,000 -2,500 -3,000 -3,500 -4,000 -4,500 -5,000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Donetsk Luhansk Source: State Statistics Service of Ukraine database. Note: Net migration within Ukraine is defined as total registered inflows from other regions of Ukraine to a given region minus total registered outflows from the region to the rest of Ukraine. National results for Ukraine (excluding Donbas) are omitted because its flows mirror that of Donbas (comprising Donetsk and Luhansk Oblasts), by definition. Kupets (2014) analyzed the drivers of interregional bilateral flows in Ukraine by using residence-based administrative data. Accordingly, geographic proximity and a common land border between regions were found to increase interregion- al migration. Higher incomes at the destination and a higher unemployment rate at the origin were also associated with higher interregional flows. The analysis found that air pollution and language differences between Ukrainian regions did not significantly affect interregional migration when other factors were consid- ered. A higher crime rate was a significant determinant of migration, but it in- creased both inflows and outflows. The Economics of Winning Hearts and Minds Previous World Bank assessments, however, found that internal migration within and between Ukrainian regions was only half of what would be expected rela- tive to other countries (Koettl et al. 2014; World Bank 2012). Accordingly, the top barriers that prevented workers from moving within Ukraine before the conflict included the following: •• Administrative procedures, reflected in a population registry system that binds people to their places of residence and increases the costs of internal migration •• Underdeveloped housing and credit markets that make housing in leading regions unaffordable to potential migrants •• Strong ties to families, friends, and home communities, including the im- portance of informal social networks to find employment and housing and gather social support. 66 Figure 1.20 Net International Migration in Donetsk and Luhansk Oblasts, 2002–13 6 60 4 40 Net registered inflows, thousands Net registered inflows, thousands 2 20 0 0 -2 -20 -4 -40 -6 -60 -8 -80 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Donetsk Luhansk Ukraine excluding Donbas (right axis) Source: State Statistics Service of Ukraine database. Note: Net international migration denotes total registered inflows from abroad to a given region minus total registered out- flows from the region abroad. Donbas consists of Donetsk and Luhansk Oblasts. Figure 1.21 Gross Migration Flows in Donetsk and Luhansk Oblasts, by Loca- tion and Flow Type, 2013 Donetsk Luhansk 35 Registered gross inflows, thousands 31.3 30.0 30 25 20 17.0 16.7 14.6 15 10.4 9.6 10 6.3 5.0 5.7 5 3.1 3.7 3.9 3.5 The Long Shadow of History and Geography 1.9 1.0 1.3 0.1 0.1 0.6 1.0 1.0 0.1 0.1 0 Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Intraregional Interregional International Intraregional Interregional International Inflows Outflows Source: State Statistics Service of Ukraine monthly bulletins (January 2014) on the economic and social situation in Luhansk and Donetsk Oblasts. Note: Data refer to registered migration flows based on registration and deregistration of population by the place of perma- nent residence. 67 Synopsis This chapter analyzed the economic and social trends before the onset of the conflict in eastern Ukraine. This is important because both the onset of the con- flict in 2014 and its economic consequences are best understood within the context of this economic and social continuum. Any future efforts to promote economic development in this part of the country must also carefully consider the constraints and opportunities indicated by such a context. Key observations emerged in several areas. Legacy growth model. The Ukrainian economy continued to use the industrial and mining capacities left from the former Soviet Union, with little moderniza- tion. This model, with Donetsk and Luhansk Oblasts at its center, remained viable in the 2000s thanks to three factors: •• Favorable external demand: Major trade partners of Ukraine boomed in the 2000s. •• Favorable commodity prices: Oil prices quadrupled in the 2000s, iron ore prices quintupled, and the price of steel (a major Ukrainian export) tripled. •• Subsidies and rents: Although commodity prices increased the costs for other steel producers, Ukraine enjoyed implicit subsidies of imported oil and gas (amounting to 8 percent of GDP annually) and sizable domestic subsi- dies in coal and electricity. Decoupling from the legacy growth model before the conflict. The legacy model stopped working years before the conflict. External demand and com- modity prices were hit by the GFC, leading Ukraine’s trade partners to invest in modern steel technologies. Input prices in Ukraine rose sharply after the Russian gas dispute in 2009, and discounts were eliminated by 2011. These develop- ments hit the economy of Donbas badly. As the country unrolled one of the The Economics of Winning Hearts and Minds starkest deindustrialization patterns in the world, its economic gravity center shifted from east to west. Social corollaries of the decoupling. Before the conflict, in 2013, Donetsk and Luhansk Oblasts had relatively low scores on the SSSU’s integrated regional in- dex of human development despite their initially favorable economic conditions. This discrepancy resulted predominantly from poor performance in three index components: demographic development, social environment, and comfort of living conditions. The region had serious environmental problems (pollution from mining and industry) and social problems (high crime rates, alcoholism, and drug abuse). Migration trends. Once a major destination of internal migration, Donbas be- came a major source of migration in the decade before 2013. Many Donbas resi- dents moved to neighboring regions and Kyiv (city and oblast), largely motivated by higher incomes and better living standards. Further out-migration was curbed by several obstacles, including an outdated residence registration system that 68 tied people down and underdeveloped housing and credit markets that made housing unaffordable for migrants. The next chapter will analyze the mechanisms through which the onset of the conflict and subsequent events have changed these trends since 2014. It will discuss the visible factors (such as physical destruction) and invisible factors (such as disorganization through disrupted connectivity, problems in coordinat- ing economic activity, and weakening social cohesion) that affect the economic and social determinants of well-being. The Long Shadow of History and Geography 69 Notes 1. Donetsk and Luhansk are similar yet distinct. Donetsk City and Oblast have traditionally been the economic engine of the region. Donetsk Oblast had over twice the population of Luhansk Oblast (4.5 million and 2.2 million, respectively), and the Donetsk economy was close to four times the size of Luhansk’s before the conflict. Donetsk has also been more industri- alized than Luhansk; however, a far larger share of Luhansk’s exports (43 percent) than Donetsk’s (22 percent) went to the Russian Federation in 2012 (UN Comtrade Database). 2. GDP data from the State Statistics Service of Ukraine (SSSU). 3. According to official statistics, there were 216 functioning mines in Donetsk and Luhansk Oblasts in 2000. This dropped to 123 mines in 2014 and to 50 by 2019 (Government of Ukraine, administrative data 2019). 4. Steel volume and export data from the UN Comtrade Database. 5. All GDP, trade, and commodity price data from the World Bank’s World Development Indicators database. 6. The implicit subsidy is calculated as the difference between (a) the oil (respectively, natural gas) price paid by Ukraine, and (b) the Russian export non-Commonwealth of Independent States (CIS) price, multiplied by the volume of imports. 7. Given the nature of the bilateral agreements, Ukraine benefited the most when oil and natural gas prices were the highest because the implicit subsidy versus the prices paid by the rest of the world was the highest. 8. All investment data from the World Development Indicators database. 9. Potential growth is estimated based on Hodrick-Prescott (HP)-filtered quarterly series in constant 2010 prices. 10. Relative unit labor costs (RULC) represent the labor costs for producing a unit of output adjusted for the changes in the labor costs in the trading partners’ countries and the nominal effective exchange rate. 11. All sectoral GDP data from the SSSU. 12. See Gabaix (2016) for a more detailed discussion of power laws in economics, including city-size distributions. 13. For a detailed discussion of these measurement and assessment issues, see Fleurbaey and Blanchet (2013). 14. Residential migration statistics underestimate the true magnitude of workers’ flows because employment-related migration is often not accompanied by corresponding changes in the place of residence through the registration system. To shed light on labor migration, it is suggested to use the Labor Force Survey (LFS) data—namely, information about employed individuals who reported working in a settlement other than where their households lived and who took part in the survey. According to such LFS-based analysis, workers of Donetsk and Luhansk Oblasts had low rates of commuting or temporary labor migration in 2013: 10.2 percent and 5 percent, respectively, compared with 12.1 percent in Ukraine as a whole. Moreover, as in other industrialized regions in eastern and southern Ukraine, residents of Donbas were much less likely than residents of western, central, and northern oblasts to work outside the boundary of their oblast. (Of all employed Donbas residents, 0.2–0.4 percent worked in some other oblast, and 0.1–0.2 percent worked abroad.) The most popular destinations for those few interregional commuters or labor migrants from Donbas were Donetsk Oblast for residents of Luhansk Oblast (and vice versa) and Kyiv City. 15. Based on a national survey, 12 percent of individuals do not live at their registered domiciles, equaling at least 3.3 million adult Ukrainians (Slobodian and Fitisova 2018). Hence, migration statistics based on registration and deregistration of population by the place of permanent residence underestimate true migration flows and should be interpreted with caution. 16. For a detailed analysis of migration trends in Ukraine before the conflict, see Kupets (2012). References Akerlof, George A., and Rachel E. Kranton. 2000. “Economics and Identity.” Quarterly Journal of Economics 115 (3): 715–53. Boarini, Romina, Orsetta Causa, Marc Fleurbaey, Gianluca Grimalda, and Ingrid Woolard. 2018. “Reducing Inequalities and Strengthening Social Cohesion through Inclusive Growth: A Roadmap for Action.” Economics E-Journal 12 (2018-63): 1–26. The Economics of Winning Hearts and Minds Fleurbaey, Marc, and Didier Blanchet. 2013. Beyond GDP: Measuring Welfare and Assessing Sustainability. New York: Oxford University Press. Gabaix, Xavier. 2016. “Power Laws in Economics: An Introduction.” Journal of Economic Perspectives 30 (1): 185–206. Hallward-Driemeier, Mary, and Gaurav Nayyar. 2018. Trouble in the Making? The Future of Manufacturing-Led Development. Washington, DC: World Bank. Koettl, Johannes, Olga Kupets, Anna Olefir, and Indhira Santos. 2014. “In Search of Opportunities? The Barriers to More Efficient Internal Labor Mobility in Ukraine.” IZA Journal of Labor & Development 3 (1): 1–28. Kohut, Zenon E., Bohdan Y. Nebesio, and Myroslav Yurkevich. 2005. Historical Dictionary of Ukraine. Plymouth, UK: Scarecrow Press. Kupets, Olga. 2009. “Preliminary Analysis of the Regional Human Development Impacts of the Crisis in Ukraine.” Background paper, World Bank, Washington, DC. Kupets, Olga. 2012. “Characteristics and Determinants of Internal Labor Mobility in Ukraine.” Background paper for Report No. 68824-ECA, World Bank, Washington, DC. Kupets, Olga. 2014. “Interregional Migration in Ukraine: Spacial, Economic and Social Factors.” Demography and Social Econo- my 2 (22): 164–75. Luhansk Regional Military and Civil Administration. 2016. “The Luhansk Region Development Strategy for the Period until 2020.” Updated strategy document, Luhansk Regional State Administration, Severodonetsk, Ukraine. NASU (National Academy of Sciences of Ukraine). 2015. “Donbas Recovery: Assessment of Social and Economic Losses and State Policy Priority Directions.” Report, NASU, Kyiv. https://reliefweb.int/sites/reliefweb.int/files/resources/Summary%20 23.09.2015_Eng.pdf. 70 Slobodian, Oleksandra, and Anastasia Fitisova. 2018. Registering Domiciles in Ukraine: How the System Affects Ukrainians. A Sociological Survey. Kyiv: CEDOS Think Tank. SSSU (State Statistics Service of Ukraine). 2014a. “Continuing Education and Obtaining a Profession: A Statistical Bulletin.” [In Urkrainian.] Electronic resource, SSSU, Kyiv. http://www.ukrstat.gov.ua/. SSSU (State Statistics Service of Ukraine). 2014b. “Regional Human Development Index.” Annual statistical report, SSSU, Kyiv. World Bank. 2012. “In Search of Opportunities: How a More Mobile Workforce Can Propel Ukraine’s Prosperity.” Vol. 1 and 2. Report No. 68824-ECA, World Bank, Washington, DC. The Long Shadow of History and Geography 71 The Economics of Winning Hearts and Minds 72 Chapter 2 The Consequences of the Conflict T he onset of the conflict and the subsequent military activity have torn the economic and social fabric of the Donbas region in complex ways. Some of these effects (like physical destruction, population displacement, and combat-related casualties) are measurable. Others (including institutional ero- sion and changes in social trust and culture) are hard to quantify despite being equally important, if not more so. This chapter summarizes the conflict dynamics and how they changed the eco- nomic conditions in Donbas. We are primarily interested in quantifying these effects and thus cannot capture the unmeasured components. Our main objec- tive is to identify the current challenges that may constitute stumbling blocks to future development efforts. Given the complex nature of the economic impacts of the conflict, we classify them according to three different channels of transmission, also referred to as the “3 Ds”: •• Destruction denotes damages to physical infrastructure and productive as- sets of the economy, which reduce the productivity of labor. •• Displacement covers the voluntary or involuntary migration of people away from conflict-affected areas, either for safety or to pursue better economic opportunities elsewhere because such conditions at the original place of residence may become unbearable during the conflict. •• Disorganization includes the impacts on factors that, unlike capital and labor, are not productive by themselves but are nonetheless essential for eco- nomic production—such as “business environment” elements (like market institutions), physical and digital connectivity, and factors affecting social cohesion such as social trust. Together, the impacts through these channels translate into the observed changes in outcomes like regional gross domestic product (GDP), wages, and productivity. Conflict Dynamics The Consequences of the Conflict The ongoing armed conflict in the eastern provinces (oblasts) of Donetsk and Luhansk broke out in 2014. As of 2020, about 38 percent of Donbas (the com- bined territories of Donetsk and Luhansk Oblasts) remained outside government control—the non-government controlled areas (NGCAs)—and are bounded by a 457 kilometer line of contact. Numerous cease-fires between 2014 and 2020 have failed to halt the conflict. Although the most recent attempts have resulted in a significant decrease in hostilities and the partial disengagement of forces, a lasting political settlement remains elusive. 75 The conflict was rooted in a context of widespread social unrest and has both domestic and geopolitical dimensions, which explain its complex dynamics and protracted nature. This section provides a brief overview of the main drivers and key turning points of the conflict. A better understanding of these factors is im- portant, both to explain how conflict has altered the socioeconomic conditions on the ground so far and to examine how the same factors may determine the success of future attempts to promote stability and prosperity in the region.1 Drivers of the Conflict From a technical perspective, identifying the causal mechanisms behind the conflict is a daunting task. Several factors—including data constraints and the strategic behavior of parties to the conflict (both internal and external)—make such identification challenging. Therefore, a broad technical consensus has not yet materialized on what caused the conflict. In practice, however, the genesis of the conflict is commonly attributed to nu- merous drivers—that is, certain factors are considered to have created the conditions for, facilitated, or directly caused the violent conflict even if such causality cannot be verified technically. These include internal drivers (including structural factors like political and social developments and prospects) and ex- ternal drivers (including geopolitical dynamics and foreign intervention). Internal Drivers Political, social, and economic structural conditions enabled the onset of the conflict in 2014. Although Ukraine was traditionally a peaceful multiethnic soci- ety, the period since its independence from the former Soviet Union in 1991 has nonetheless witnessed diverging regional political and economic interests and tensions. These tensions reflect deep historical, cultural, linguistic, and econom- ic differences between the western regions of Ukraine (historically part of the The Economics of Winning Hearts and Minds Austro-Hungarian Empire) and its southern and eastern regions (historically part of the Russian Empire). These tensions were managed successfully in the early post-Soviet period by an inclusive framing of statehood and system of governance, beginning after 1994, that balanced political and regional interests and encouraged the socioeco- nomic integration of the country. Subsequent changes in government and the impacts of the global economic recession in 2008, however, eroded this delicate political balance, leading to an increasingly narrower framing and a polarization of views on the country’s future trajectory and orientation. Combined with the economic downturn in the early 2010s, this polarization reinforced perceptions of socioeconomic marginalization and neglect—particularly in the eastern and southern regions, which were disproportionately affected. 76 External Drivers The conflict emerged within a broader geopolitical context and with the direct intervention of external actors. Geopolitically, Ukraine became a theater of in- creasingly tense competition between rival forces over political and economic influence in the past two decades. This competition undermined the delicate internal political status quo in place since the early 1990s, leading to increasingly unstable shifts in political governance between eastern and western influences. This pendulum of transition magnified economic and political polarization and reduced the country’s complex political, social, and economic issues to a stark ethnic and territorial calculus. As traditional spaces for compromise and commitment eroded, centrifugal dy- namics became viable, enabling the conditions for the conflict. However, direct external intervention likely has facilitated the conflict (in terms of longevity and intensity). The true extent of this intervention is not clear, but direct military support and financial assistance to the parties in conflict (equipment, training, and troops) have been observed. Against this backdrop, a series of events and actions in 2013 and 2014 precipi- tated a process of destabilization that facilitated the outbreak of armed conflict in eastern Ukraine. President Viktor Yanukovych’s decision not to sign an associ- ation agreement with the European Union (EU) in late 2013 triggered widespread antigovernment demonstrations in the country. The “Maidan movement,” as it became known, represented both a rejection of a system of governance widely perceived as corrupt and ineffective and the crystallization of regional divisions. The latter issue (regional divisions) came to the fore in early 2014 when, after a period of government violence against demonstrators, the Verkhovna Rada (par- liament) voted to remove President Yanukovych from office. In parallel, several anti-Maidan protests took place in certain eastern and southern regions, and confrontations became increasingly violent in the spring of 2014. Such social discontent in different areas was instrumentalized by political actors who used regional and identity-based grievances as a vehicle for mobilizing po- litical, and subsequently military, action. The Consequences of the Conflict Three Phases of the Conflict From its onset in 2014 until 2020, the conflict in eastern Ukraine exhibited three distinct phases: •• Phase 1 (2014–15): Outbreak and initial round of conflict until the Minsk agreements •• Phase 2 (2015–18): Low-level conflict and unsuccessful attempts to stop the conflict •• Phase 3 (2019–present): Renewed progress in disengagement and peace ne- gotiations. 77 Box 2.1 presents a timeline of key events marking the corresponding turning points. Phase 1 (2014–15): The Onset The 2014–15 period marked the beginning of violent confrontation in the Donbas region and the loss of government control over portions of Donetsk and Lu- hansk Oblasts. The ousting of President Yanukovych in February 2014 triggered an armed insurgency in the southern and eastern parts of Ukraine. As Ukrainian security forces were overwhelmed, portions of Donetsk and Luhansk Oblasts were declared “People’s Republics” by April. Clashes intensified in the summer of 2014, and a first cease-fire agreement (the Minsk Protocol) was signed on September 5. Nonetheless, fighting continued until February 2015, when agree- ment was reached on a package of 13 measures to implement the Minsk Protocol (Minsk II). This first phase also constituted the most violent period of the conflict to date. It accounted for about 80 percent of all civilian deaths to date (map 2.1, pan- els a and b); mass population displacement; and significant destruction to pri- vate property, infrastructure, and economic assets. According to the Recovery and Peacebuilding Assessment (RPBA) conducted by the EU, United Nations, and World Bank during this period, the conflict directly affected an estimated 3.9 million out of 5.2 million people in the Donbas region (EU, UN, and World Bank 2015). This includes an estimated 7,000 civilian and military deaths, more than 18,000 wounded, and more than 1.6 million displaced internally and outside Ukraine. Meanwhile, several international initiatives tried to mediate an end to the conflict, with mixed results: •• The Special Monitoring Mission (SMM) to Ukraine was deployed in March 2014 by the Organization for Security and Co-operation in Europe (OSCE). •• The Trilateral Contact Group (TCG) on Ukraine, created in June 2014, com- prises representatives of the OSCE, Ukraine, the Russian Federation, and the separatists. The Economics of Winning Hearts and Minds •• The Normandy Format—named for a meeting of country representatives during the June 2014 D-Day commemoration in Normandy—brought togeth- er the governments of Ukraine, Russia, France, and Germany to negotiate a resolution of the conflict, leading to the Minsk Protocol and Minsk II. Phase 2 (2015–18): Positional Warfare, Severed Ties The second phase of the conflict (2015–18) was characterized by positional war- fare along the front lines (the “contact line”) and a series of short-lived cease- fire attempts. The Minsk II agreement resulted in a cease-fire that was respected until April 2015, when armed conflict resumed. Numerous cease-fires were ne- gotiated since then (notably in August 2015, September 2016, December 2017, and July and September 2018), none of which stopped hostilities permanently. 78 Clashes during this period were largely “positional”—military skirmishes across the contact line without significant changes in territorial control. During this period, a step-by-step approach to cease-fire negotiations was initiated, with the TCG’s August 2016 “Framework Decision” on disengagement establishing a complex series of steps toward a permanent cease-fire that formed the basis of subsequent cease-fire negotiations. This second period was also marked, in March 2017, by Ukraine’s imposition of an economic and energy embargo against the NGCAs. This severed remaining economic ties, leading to significant deterioration on both sides of the contact line, especially in the NGCAs. Although casualties were lower than during the initial phase of the conflict, armed violence and bombardments of civilian areas continued to directly affect civilians on both sides of the contact line, creating additional waves of pop- ulation displacement and leading to the continued destruction of private and public property and infrastructure (map 2.1, panels b–e). Phase 3 (2019–present): Limited Progress The third phase of the conflict has been characterized by a steady progress in cease-fire negotiations since 2019, which facilitated partial disengagement of forces and a reduction in the intensity of the conflict (map 2.1, panel f). Prospects for a resolution to the conflict in eastern Ukraine improved with the election of President Volodymyr Zelenskyy in Ukraine in early 2019, who signaled increased willingness to seek a negotiated end to the conflict. Negotiations within the TCG resulted in a comprehensive cease-fire in July 2019, leading to an 80 percent reduction in cease-fire violations, which lasted until August (a mere 17 days). This was followed by an exchange of prisoners between Ukraine and Russia in September, and the October signing of the “Steinmeier Formula” (first circulated in 2016 by then German Foreign Minister and OSCE Chairperson-in-Office Frank-Walter Steinmeier) on implementation of key provisions of the Minsk agreements. The latter was followed by the continued disengagement of forces in specific zones. The Normandy Format participants agreed upon additional disengagement areas in December, but this process The Consequences of the Conflict subsequently stalled with disagreements over the precise sequencing of mea- sures foreseen in Minsk. As of February 2021, the conflict has remained “in stasis”—marked by signifi- cantly reduced violence (including the lowest civilian casualties recorded during the conflict); a cease-fire that has held, for the most part, since July 2020; and continued political deadlock in the peace negotiations. The remainder of this chapter will analyze the far-reaching consequences of these conflict dynamics. Specifically, it focuses on the mechanisms through which the conflict has manifested its economic and social effects: destruction, displacement, and disorganization. The next chapter will analyze how these mechanisms have shaped the current conditions. 79 Map 2.1 Casualties (Deaths) in Donetsk and Luhansk Oblasts, by Event, 2014–19 2014 2015 2016 2017 2018 2019 1 200 400 Number of casualties per event 100 300 500 Source: Uppsala Conflict Data Program (UCDP) database, https://ucdp.uu.se/. ©World Bank; further permission required for reuse. Note: The maps depict the two oblasts within the Donbas region: Luhansk Oblast to the right of the gray bisecting border and Donetsk Oblast to the left. The “contact line,” in red, separates the (Ukrainian) government-controlled area (GCA) of each oblast from its (separatist-held) non-government controlled area (NGCA) where most of the conflict-related fatalities have occurred. Destruction The Economics of Winning Hearts and Minds Destruction of physical assets is one of the most easily recognizable conse- quences of conflict, together with human casualties and displacement. It is vi- sually verifiable (for example, in newspaper images), and it symbolizes all other hardships associated with the conflict, some of which are not visible. Unlike the destruction wrought by natural disasters, the conflict-driven destruc- tion of infrastructure often unleashes a plethora of these invisible damages (for example, weakened social cohesion, as discussed later in the “Disorganization” section). It is this combination of visible and invisible factors that often makes wars more catastrophic than natural disasters. Unlike the latter, wars both de- stroy physical capital (such as infrastructure) and suppress returns to invest- ment by means of invisible complications (such as disruptions in connectivity and the social fabric of the country, leading to reduced social trust and erosion of institutions). 80 Box 2.1 Chronology of the Conflict in Eastern Ukraine, 2013–present 2013 •• November 21: Protests begin against President Yanukovych’s decision to annul Ukraine’s association agreement with the EU. Violence rapidly escalates. •• November 2013–February 2014: Antigovernment protests intensify throughout the country, become more violent, and are met with re- pressive force. 2014 •• February 22: Parliament removes President Yanukovych from office. •• February–March: Violent clashes spread in the eastern regions be- tween pro- and anti-Maidan protesters. •• March: The Russian Federation annexes Crimea. •• April: Separatists take control of parts of Donetsk and Luhansk Oblasts, declaring them to be separate “People’s Republics.” •• June: The Trilateral Contact Group (TCG)—representing the OSCE, Ukraine, Russia, and the separatists—is established to facilitate talks. •• September: Minsk Agreement (I) is signed. 2015 •• February: Minsk Agreement (II)—an implementation package of 13 mea- sures—is signed but fails to stop the fighting. •• August: A new cease-fire is agreed to at the TCG meeting in Minsk, but it fails in November. 2016 •• September: The TCG’s Framework Decision on disengagement of forc- es and materiel results in limited progress on disengagement until May 2017 but is considered to have failed by early 2018. The Consequences of the Conflict 2017 •• March: Ukraine imposes an economic and energy embargo on the non-government controlled areas (NGCAs). •• December: The TCG reaches a new cease-fire agreement, which fails in January. 2018 •• April–June: Conflict escalates significantly despite the TCG’s cease- fire attempts. •• July and September: The TCG reaches new cease-fire agreements, but both are short-lived. 81 2019 •• March: A new TCG-negotiated cease-fire goes into effect. •• July: The parties agree to a wide-ranging cease-fire, which reduces vi- olations by 80 percent and lasts until August. •• September: Ukraine and Russia exchange prisoners. •• October: The parties sign the Steinmeier Formula to implement parts of the Minsk Agreements. •• October–November: Both sides carry out force disengagement in three designated zones. •• December: The Normandy Four (Ukraine, Russia, France, and Germany) meet, resulting in a thaw in the frozen negotiations, but there is no fur- ther substantive progress on implementation of the Minsk provisions. 2020 •• July: The 29th cease-fire agreement holds until September. From an economic point of view, destruction of physical capital should increase investment (because the marginal productivity of capital is higher at low capi- tal levels). This may be true for some natural disasters. In active wars, however, this usually does not happen because physical destruction is accompanied by productivity shocks (or a decrease in labor supply). In the end, the marginal pro- ductivity of capital decreases despite the initial destruction. Accounting for Data Discontinuity, Pre- and Post-2014 In Donbas, assessing the degree of physical destruction and its economic con- The Economics of Winning Hearts and Minds sequences is hampered by data problems. For many publicly available indicators on Donbas, a before-2014 and after-2014 comparison is misleading because of a change in geographical scope. The regional statistics from the State Statistics Service of Ukraine (SSSU) cover the entire oblasts of Donetsk and Luhansk until 2013. In 2014, however, they began to exclude the NGCAs because reliable data were not available. This prohibits a comparison between pre-2014 and post- 2014 (inclusive) periods unless a correction is made. To elaborate on this, consider a hypothetical example (figure 2.1). Let’s say that, before the conflict onset, one entire oblast had an observable 100 units—com- prising 40 units from the part that later became a government-controlled area (GCA) and 60 units from the part that became an NGCA. With the onset of the conflict, both the GCA and the NGCA values decrease to 30. But now only the GCA data (30 units) is observed. If this accounting change were ignored, the ob- servable data would suggest a 70-point drop in the oblast instead of the actual 40-point drop. The mistake here is to conflate two factors: an actual decrease in the GCA and an accounting omission of the NGCA. 82 Figure 2.1 Assessment Bias from Changing Spatial Coverage of Data in Donbas: A Hypothetical Example 140 Observed decrease = 70 points Actual decrease in GCA (unobserved) = 10 points 120 Actual total decrease (unobserved) = 40 points Actual decrease in NGCA (unobserved) = 30 points 100 80 Units NGCA: 60 60 NGCA: 30 40 20 GCA: 40 GCA: 30 0 Before conflict onset After conflict onset GCA NGCA Observed data Source: World Bank calculations. Note: The figure illustrates the difficulty in assessing the actual changes in key socioeconomic indicators after the Donbas conflict began in 2014, when Ukrainian regional statistics began to cover only the government-controlled areas (GCAs) of Donbas (Donetsk and Luhansk Oblasts). The conflict-driven division of these oblasts into GCAs and non-government con- trolled areas (NGCAs), as well as the changes in the NGCAs, are not statistically observable. To see how real data reflect this complication, we can use the housing stock se- ries for the Luhansk and Donetsk GCAs (table 2.1). These series report the entire oblast stock for 2013 and only the GCAs’ stock from 2014 onward. Thus, although the available data indicate that the housing stock in Luhansk Oblast decreased from 55 million square meters in 2013 to 17.7 million square meters in 2014, this is not true. By using only these series, we cannot assess the housing stock trends in either the entire oblast or in the GCA. There are two imperfect remedies to this problem. First, estimate the GCA share of the total oblast before the conflict. Then use this along with other observ- able data (for example, conflict intensity) to infer the NGCA indicator after the conflict. Second, estimate the current NGCA level, and then use it to infer the preconflict GCA/NGCA composition. In both cases, heterogeneity between the GCA and NGCA economies (for example, population size and age, urbanization levels, and sectoral compositions) will need to be accounted for. In the remain- der of this chapter, we use these approaches as needed. The Consequences of the Conflict In the absence of a comprehensive census, we use nighttime light emissions as a proxy for changes in infrastructure conditions.² To do this, we use data from the Suomi National Polar Partnership (SNPP) satellite of the National Aeronautics and Space Administration (NASA) and the National Oceanic and Atmospheric Administration (NOAA). The satellite uses a Visible Infrared Imaging Radiometer Suite (VIIRS) instrument to collect low light in spectral bands covering emissions generated by electric lights, excluding stray light, lightning, lunar illumination, and cloud cover. For the purposes of this study, the nighttime light emissions in this scheme are interpreted narrowly as the availability of electricity (grid or genera- tor) or more generally as a proxy measure for the existence of utilities. 83 Table 2.1 Housing Stock of Donetsk and Luhansk Oblasts, 2013–18 Square meters, millions Oblast 2013 2014 2015 2016 2017 2018 Donetsk 103.5 50.6 51.2 51.4 51.5 51.6 Luhansk 55.0 17.7 17.8 18.8 18.8 18.8 Source: State Statistics Service of Ukraine database. Note: Starting in 2014, data were provided for only the government-controlled areas (GCAs) of these oblasts. Map 2.2 provides the estimation results for GCAs and NGCAs in Donetsk and Luhansk Oblasts, using the 2020 contact line boundaries. Temporal averaging is done on a monthly and annual basis between December 2013 and December 2019. The areas shaded red show where emissions were lost, and those shaded green show emerging emissions. The blue shading denotes no change. The estimates suggest that the Donetsk GCA exhibited a 20.2 percent net loss of light emissions between 2013 and 2019, and the NGCA exhibited a 28.1 per- cent net loss. In Luhansk Oblast, the GCA lost 7.2 percent of its light emissions and the NGCA lost 42.9 percent, in net terms. It is important to remember that these emission changes reflect a combination of several factors including both physical damages to infrastructure and service interruptions for other reasons (for example, human capital and input losses). Damage to Housing Stock and Infrastructure Housing. Since 2014, more than 50,000 residential buildings on both sides of the contact line have been damaged, according to international agencies (NRC The Economics of Winning Hearts and Minds 2019). Many of them have already been repaired, thanks to a concerted effort by local authorities, the Ukrainian government, and the international community. The latest on-the-ground damage assessments were conducted by the Nor- wegian Refugee Council in Donetsk (2019) and Luhansk (2018) Oblasts, focusing on shelter damage along the contact line. The assessments covered 3,945 pre- identified properties in Donetsk Oblast (based on a possible damage report by local officials) and around 3,100 in Luhansk Oblast (NRC 2018, 2019). In Donetsk Oblast, about 1,900 properties were identified as needing further repairs, of which about 90 percent needed light and medium repairs and 9 percent heavy repairs. About 1 percent was deemed fully destroyed. In Luhansk Oblast, 1,289 properties needed further repairs, of which about 90 percent required light and medium repairs and 6 percent heavy repairs. The remaining 4 percent were de- stroyed. 84 Map 2.2 Changes in Nighttime Light Emissions, Donetsk and Luhansk Oblasts, 2013–20 Net loss of light emissions (%) The Consequences of the Conflict Donetsk GCA 20.2 Donetsk NGCA 28.1 Luhansk GCA 7.2 Luhansk NGCA 42.9 Source: World Bank estimations from US National Oceanic and Atmospheric Administration (NOAA) data. ©World Bank; further permission required for reuse. Note: Gray lines denote oblast boundaries. Within each oblast, the “contact line” (dark red) divides the government-con- trolled area (GCA) to the north and/or west of the line from the non-government controlled area (NGCA) to the south and/ or east of the line. Temporal averaging of light emissions was done monthly and annually between December 2013 and December 2019. 85 To complement the available knowledge and fill the gaps, a remote-sens- ing-based shelter damage assessment for Luhansk and Donetsk Cities was conducted for this study. The assessment analyzed the damages to residen- tial structures (both single-family homes and apartment buildings) in 2015 and 2019. In Luhansk City, no significant damage was detected. In Donetsk City, an estimated 0.4 percent of about 81,000 structures were identified as damaged. It is, however, important to note that these remote-sensing-based estimates are likely to present a lower bound of damage since they only reflect structural damages that are vertically observable from space. They do not consider lateral and light damage, which is known to make up the bulk of the cases, as confirmed by on-the-ground assessments in the GCAs. Power and water infrastructure. There were frequent attacks on strategic in- frastructure throughout the conflict, but especially in the earlier active phases in 2014 and 2015. During this initial phase of hostilities, some thermal power fa- cilities (TPPs) were severely damaged. In particular, in the spring of 2014, about 20,000 shells exploded on the site of Slovianska TPP, putting it out of service for about nine months. Many of the most essential water, sanitation, and hygiene (WASH) infrastructure facilities in Donbas are located on or near the contact line and are subject to frequent attacks, causing water service losses and shortages to residents on both sides of the line. Among the most frequently impaired WASH facilities is the Donetsk Filter Station, which provides water to 380,000 residents in the region and was damaged 71 times between October 2016 and May 2020. In addition, the 1st Lift Pumping Station of the South Donbas Water Pipeline, a critical facility that provides water to 1.2 million people, had 73 damage incidents during the same period.³ Transportation infrastructure. Transportation infrastructure suffered some of the heaviest damages. The Donbas region’s two major airports—Donetsk Ser- gei Prokofiev International Airport and Luhansk International Airport—were both destroyed during the most active phase of the conflict. Similarly, among the re- gion’s 27 major railway stations, 21 have been damaged and have not returned to full functionality. The 2015 multiagency Recovery and Peacebuilding Assessment The Economics of Winning Hearts and Minds (RPBA) also reported major damages to bridges and roads from both explosives and heavy use of military equipment (EU, UN, and World Bank 2015). Overall, the conflict has imposed significant damages to the infrastructure. How- ever, our knowledge about the true extent of this damage needs to be updated with an on-the-ground assessment. Since 2015, the conflict has been limited to the area around the contact line, and there have been continuous efforts to re- pair and rehabilitate the strategic infrastructure and settlements alike. Accord- ingly, this study recommends a comprehensive damage and needs assessment to identify the current status of the infrastructure in the Donbas region and how that infrastructure maps onto recent demographic and economic trends. We turn to this point next. 86 Displacement The conflict has severely affected the demographic composition of Donbas. Many individuals continued living in areas that are not controlled by the Ukrainian government (NGCAs), whereas thousands of individuals were forced to move to either the government-controlled areas (GCAs) of Donetsk and Luhansk Oblasts, to other regions of Ukraine, or abroad. Unfortunately, a complete picture of these dynamics is obscured by the lack of reliable statistics. The SSSU provides information about the de jure and de facto populations using a demographic balance approach—that is, adjusting an initial estimate of popu- lation by using registered births, deaths, and net registered migration.⁴ However, these estimates are based on the official place of permanent residence (the for- mer propiska, or residence registration, system) and likely provide an incomplete view. Many internally displaced persons (IDPs) and migrants who moved be- fore the conflict could still be counted in the populations of Donetsk or Luhansk Oblasts. In fact, a 2018 survey5 suggests that the share of non-IDP residents who were not living at their registered domiciles reached 9 percent in Donbas and 16 percent in the central and southwest regions of Ukraine (Slobodian and Fitisova 2018).6 Overall Displacement Dynamics According to the Ministry of Social Policy of Ukraine (MoSP), there were 1,420,523 IDPs in July 2020, compared with a record high 1,709,083 IDPs in 2016.7 As of June 2020, 45.3 percent of the country’s IDPs were registered in Donetsk and Luhansk Oblasts, according to the UN High Commissioner for Refugees (UNHCR 2021). Popular destination regions outside of Donbas, according to MoSP data, are Kyiv City (11.0 percent of all registered IDPs) and the following oblasts: Kharkiv (9.3 percent), Dnipropetrovsk (4.9 percent), Kyiv (4.4 percent), Zaporizhzhia (3.9 percent), and Odesa (2.6 percent). These are nearly the same regions that had The Consequences of the Conflict attracted the most migrants from Donbas before the conflict. The number of pensioners among registered IDPs (726,022) was almost the same as the total of working-age individuals and children (739,530). To gain a better understanding of the displacement dynamics, we use the Na- tional Monitoring System (NMS) data collected by the International Organization for Migration (IOM).8 The NMS is a regular comprehensive survey of IDPs living in GCAs as well as IDPs who returned to NGCAs (box 2.2), and it has been con- ducted quarterly since 2016. More details about the survey, its methodology, and key findings for round 15 (July–September 2019) can be found in IOM (2019). Table 2.2 provides a simple comparison between the working-age (15–70 years) populations of Ukraine (excluding Donbas), IDPs, and returnees.9 For similar in- formation on all IDPs and returnees without age limits, see appendix B, table B.1. 87 Table 2.2 Sociodemographic Characteristics of IDPs, Returnees, and Work- ing-Age Population in Donetsk and Luhansk Oblasts Compared with Rest of Ukraine, 2018 Share of respondents, percent Working-age population IDPs Returnees Ukraine Ukraine Donetsk Luhansk Donetsk Luhansk Donetsk Luhansk Variable Category excl. excl. (GCA) (GCA) (GCA) (GCA) (NGCA) (NGCA) Donbas Donbas Urban 84.5 72.1*** 66.7*** 92.1 83.4*** 89.3*** 92.0 91.8 Type of Rural 15.5 27.9*** 33.3*** 7.2 16.0*** 9.5*** 2.6 2.6 settlement NR n.a. n.a. n.a. 0.8 0.6 1.2 5.4 5.6 15–24 years 10.9 10.6 13.7*** 12.9 13.2 13.8* 5.4 4.9 Age 25–59 years 70.1 70.1 69.4 68.6 67.9 70.5** 54.8 53.5 60–70 years 19.0 19.3 16.9*** 18.5 18.9 15.7*** 39.8 41.7 Female 52.7 52.5 52.5 60.4 59.4 59.4 56.0 57.8 Gender Male 47.3 47.5 47.5 39.6 40.6 40.6 44.0 42.2 Unemployed 50.0 56.9*** 57.5*** 47.3 50.8*** 51.2*** 36.1 35.9 Employed 8.1 10.1* 5.3*** 4.7 4.2 4.5 2.6 2.6 Pensionersb 24.2 21.3*** 20.2*** 25.3 25.8 21.7*** 49.7 50.3 Labor force Students 6.3 5.7 8.0*** 8.5 8.8 9.2 3.6 3.3 statusa Housework 8.2 5.0*** 7.9 11.6 8.7*** 10.8 5.4 5.3 Inactive 3.2 1.0*** 1.2*** 2.0 1.5 2.0 1.6 1.5 NR n.a. n.a. n.a. 0.6 0.3** 0.7 1.0 1.1 The Economics of Winning Hearts and Minds Unemploy- % of labor 14.0 15.1 8.4 9.0 7.6 8.0 6.8 6.8 ment rate force NMS-2018 NMS-2018 Source LFS-2018 (rounds 9–12) (rounds 9–12) N Sample 16,176 13,267 254,674 5,238 2,180 33,580 2,131 1,868 Sources: World Bank calculations from 2018 Labor Force Survey (LFS) and National Monitoring Survey (NMS) data. Note: The LFS and NMS datasets include all surveyed household members aged 15–70 years. Donbas comprises the Donetsk and Luhansk Oblasts. GCA = government-controlled area; IDPs = internally displaced persons; NGCA = non-government controlled area; NR = not reported, unknown; n.a. = not applicable. a. Labor force status for IDPs and returnees is based on their answers to question 1.56: “What is your current employment status or main activity?” b. Pensioners include nonworking individuals who receive old-age pension as well as disability pension or assistance. Significance level of difference in means between (a) Luhansk, and (b) Ukraine (excluding Donbas): *** = 1 percent, ** = 5 percent, * = 10 percent. 88 According to the 2018 NMS data, over 57 percent of the working-age heads of IDP households lived in the GCAs of Donetsk and Luhansk Oblasts. Six regions outside Donbas accounted for another one-third of IDPs: Kyiv City (10.5 per- cent), Kharkiv Oblast (8.3 percent), Dnipropetrovsk Oblast (8.3 percent), Kyiv Oblast (3.9 percent), Zaporizhzhia Oblast (3.7 percent), and Odesa Oblast (2.5 percent). IDP Destination Patterns As the official statistics for IDP registration show, the main destination regions of IDPs from Donbas resemble the economic migration before the conflict, and the choice of destination can be explained by distance, kinship or language, employ- ment, and livelihood opportunities. Having relatives and friends in a new location was an important factor for IDPs in rural areas distant from Donbas, and employ- ment opportunities were decisive for IDPs in urban areas, whereas respondents who settled in Donbas and other regions near the conflict were more influenced by proximity to home areas than by either family or employment (World Bank 2017). Urban and rural choices. The overwhelming majority of individuals in all sub- groups and regions live in towns and cities. However, the share of rural popula- tion is much smaller among returnees than among either IDPs or non-IDP Ukrai- nians who live in the Donetsk and Luhansk GCAs, according to the 2018 NMS data: less than 3 percent of returnees live in rural areas, compared with 7.2–16 percent of IDPs and 15.5–28 percent of other working-age individuals who live in the Donbas GCAs. As observed before, residents of Luhansk Oblast (both IDPs and those who live in the GCA) are more likely than residents of Donetsk Oblast to live in rural areas. Analysis of the previous and current places of residence for IDPs shows that most of the IDPs who were surveyed in rural areas had lived in cities or towns before the displacement. This points to substantial migration of displaced peo- ple to rural areas, probably because the cost of living is lower in villages than in large cities. Income and livelihood factors. IDPs living in the rest of Ukraine seem to rely The Consequences of the Conflict more on earnings and social assistance (most likely paid to children), whereas IDPs in Luhansk Oblast rely heavily on government support to IDPs. At the same time, a much larger share of IDPs in Donetsk Oblast than in Luhansk Oblast and the rest of Ukraine reported pension and disability benefits among the sources of household income. Although IDPs living in the rest of Ukraine seem to be better off than their coun- terparts in Donbas in terms of their financial and labor market circumstanc- es, they are significantly more worried about their living conditions, particularly about lacking their own housing, because the overwhelming majority of them live in rented apartments or houses (appendix B, table B.2). 89 Figure 2.2 Challenges in Obtaining Employment Reported by Unemployed IDP Heads of Household in Ukraine, by Destination Region, 2018 Lack of job opportunities Low pay for proposed vacancies Lack of vacancies corresponding to qualifications Unsuitable work schedule Difficult to combine work and family responsibilities Restrictions on health, disability It takes a long time to get to work Did not experience any challenges Discrimination by age Lack of knowledge and skills Discrimination by IDP status, registration 0 10 20 30 40 50 60 70 80 Donetsk (GCA) Luhansk (GCA) Ukraine excluding Donbas Source: World Bank calculations from 2018 National Monitoring Survey (NMS) data of the International Organization of Migra- tion. Note: The joint dataset is of internally displaced persons (IDPs) from face-to-face and phone interviews (NMS rounds 9–12, weighted). Information is provided only for heads of household, excluding IDPs from Crimea and individuals older than 70 years. “Unemployed” are individuals (respondents in face-to-face interviews) who reported that in the last seven days they were without a job, actively looking for one, and ready to start a job within two weeks. Answers are shown in descending order of their share of the total. Respondents could choose multiple answers. Donbas comprises the Donetsk and Luhansk Oblasts GCA = government-controlled area. IDPs in other regions of Ukraine are likely to face more significant challenges than those from Donbas in finding a job that would correspond to their education and skills (figure 2.2). Most of the employed IDPs living in the rest of Ukraine work in various service activities, trade, and construction, according to 2018 Labor Force Survey data. By comparison, about 54 percent of IDP household heads working in Luhansk Oblast were employed in public services (public administration, ed- The Economics of Winning Hearts and Minds ucation, and health care). This is probably because some public services (and their employees) relocated from the NGCA to the GCA. Another important difference between IDPs in Donbas and those in the rest of Ukraine is the share who reported having jobs that correspond with their edu- cation. In Luhansk Oblast, 89.4 percent of respondents said that was the case, compared with only 53 percent of IDPs living and working in the rest of Ukraine. Moreover, IDPs in the rest of Ukraine reported more often than those in Donbas that retraining, education, and consultation from the State Employment Service would be helpful in their (re)employment. At the same time, the overwhelming majority of unemployed IDPs, regardless of destination, are interested in receiv- ing direct support for job placement. 90 Characteristics of Returnees Returnees are significantly older than both IDPs and the general population in the Donbas GCAs. With the entire sample in mind, the age discrepancy between IDPs and returnees is even more pronounced: households of IDPs include more children, whereas most returnees are older individuals and those who get pen- sions. (However, the age composition of IDPs is similar to the overall working-age population in the GCAs and elsewhere in Ukraine.) Older individuals returned to the NGCAs mainly because they had private prop- erty there and would not need to pay rent as well as for family reasons (table 2.3). Some IDPs whom the IOM defined as “returnees” were in fact not displaced; they kept living in their original place of residence in the NGCA but registered as IDPs in the GCA to get a pension from the Ukrainian Pension Fund.10 Along with housing or property issues and family reasons, an important reason for the return of prime-age workers was a lack of employment opportunities where they were temporarily living during displacement. Notably, about 3 per- cent of the heads of household returning to NGCAs reported a failure to inte- grate into the local community at the previous place of (displaced) residence. Employment and Educational Distinctions There are also remarkable differences in labor force status between IDPs, re- turnees, and non-IDPs—although those differences vary by place of residence. Although IDPs have somewhat lower employment and labor force participation rates than non-IDPs in general, they also have lower unemployment rates than non-IDPs living in Donetsk and Luhansk Oblasts specifically (table 2.2). IDPs are more likely to be inactive than the total working-age population of Ukraine, not only because they are more likely to be retired and receiving a pension but also because women are more prevalent among IDPs than among non-IDPs and are more likely to focus on housework and childcare. Similarly, most returnees do not work after returning to the NGCAs as they rely on pension. Returnees to the NGCAs also differ from IDPs living in the GCAs in terms of the The Consequences of the Conflict education of household heads:11 a relatively lower share of the returnees have tertiary education, and a relatively higher share of them have vocational educa- tion (figure 2.3). The relatively large share of returnees with vocational education may suggest that (a) individuals having region- or industry-specific education and skills could not find a relevant job and pay for rented housing at the new place, or (b) they may have retired early and relied on pension benefits. Interestingly, IDPs living in Ukrainian territories other than Donbas are, on average, more educated than IDPs who stayed in Donbas. This can be a sign of a positive self-selection of IDPs to the regions farther from Donbas both geographically and economically. 91 Table 2.3 Main Reasons for Returning to Donbas NGCAs, by Current Place of Residence and Age Group, 2018 Share of respondents, percent 18–24 25–59 Resgponses Donetsk Luhansk 60+ years Total years years There is private property and we 78.5 77.5 58.6 77.0 79.3 78.0 do not have to pay rent Family reasons 34.0 35.7 31.0 37.0 33.1 34.8 Lack of employment opportunities 6.6 7.8 6.9 14.1 1.5 7.2 Failure to integrate to local 2.7 3.0 3.5 3.1 2.6 2.9 community at the place of refuge Limited access to social services 1.9 2.8 10.3 1.6 2.8 2.3 (health care, education, etc.) Place of residence in Donbas 0.6 0.9 0.0 0.9 0.6 0.7 became under government control Other 7.3 7.4 10.3 6.7 7.8 7.3 N 1,461 1,277 29 1,215 1,492 2,738 Source: World Bank calculations from 2018 National Monitoring Survey (NMS) data of the International Organization of Migra- tion. Note: Dataset is of returnees currently living in a non-government controlled area (NGCA) (in NMS rounds 9–12, unweighted) who answered question 3.2: “What was the main reason for your returning to Donbas?” Donbas comprises the Donetsk and Luhansk Oblasts. Column percentages total more than 100 because respondents could choose multiple answers. Information is provided only for heads of household. Figure 2.3 Educational Attainment of IDP and Returnee Household Heads in Donbas and Rest of Ukraine, by Current Place of Residence, 2018 100 90 Share of respondents (%) 80 70 60 39.7 41.6 51.5 43.3 50 56.2 The Economics of Winning Hearts and Minds 57.5 40 61.4 30 28.3 20 27.0 24.4 26.2 35.9 39.1 37.5 10 14.9 18.7 13.1 14.9 12.9 15.7 14.2 0 Donetsk Luhansk Ukraine Total Donetsk Luhansk Total (GCA) (GCA) exluding Donbas (NGCA) (NGCA) IDPs Returnees Secondary and lower Vocational Tertiary Not reported (includinq incomplete) Source: World Bank calculations from 2018 National Monitoring Survey (NMS) data of the International Organization of Migra- tion. Note: The joint dataset is of internally displaced persons (IDPs) from face-to-face and phone interviews (NMS rounds 9–12, weighted). Donbas comprises the Donetsk and Luhansk Oblasts. Information is provided only for heads of house- hold, excludes IDPs from Crimea, and excludes individuals older than 70 years. GCA = government-controlled area; NGCA = non-government controlled area. 92 Box 2.2 The National Monitoring System (NMS) Survey: Definitions and Methodology The NMS gathers data about IDPs via telephone and face-to-face sam- ple surveys using equivalent questionnaires. A similar questionnaire is also used in the telephone survey of returnees to the Donbas NGCAs. Geo- graphically, the survey covers 24 oblasts of Ukraine and the city of Kyiv, including NGCAs for returnees. Definitions “IDPs” are defined as individuals who are either registered as such with the MoSP or who consider themselves as displaced people because of the conflict. (For example, they left their places of residence in the GCAs for fear of the potential start of fighting in their settlements.) During the phone interview, the respondents verbally confirm that they are IDPs and confirm their place of residence in the GCA (oblast, locality, and so on). “Returnees” are IDPs registered by the MoSP who returned to an NGCA. A potential respondent verbally confirms their place of residence in the NGCA (oblast, locality, and so on) and that they returned to the NGCA. Face-to-Face Survey Sampling The sample design of the face-to-face survey consists of two stages: •• Stage 1: A sample of the territorial units (TUs) stratified by regions of Ukraine is formed. The TUs and respondents are distributed in propor- tion to the number of IDP families registered by the MoSP. On average, eight respondents are surveyed within one TU. In the regions with a low number of officially registered IDPs, enough TUs and respondents are selected to ensure further analysis. Overall, the TUs correspond to the administrative and territorial units of Ukraine. The selection of the TUs within the stratums is carried out based on a random selection with probabilities proportional to the number of registered IDPs. •• Stage 2: IDPs are surveyed within each randomly selected TU. The re- The Consequences of the Conflict spondents are recruited using the time-location sampling (TLS) and snowball sampling techniques. One respondent is surveyed per each IDP household. The sample uses a rotation procedure of approximately 50 percent of the TUs. Telephone Survey Sampling The sampling for the telephone survey is derived from the MoSP’s Unified Information Database of Internally Displaced Persons. The distribution of the interviews, by oblast, is calculated based on the distribution of IDP families registered by the MoSP. In oblasts with a low number of registered IDPs, the number of interviews is increased to ensure further analysis. 93 The sample size is calculated considering the possibility of surveying those who returned to NGCAs. Datasets and Weighting The data from the telephone interviews of the IDPs are combined with the data from face-to-face interviews with IDPs. Both datasets are weighted according to the regional distribution of registered IDPs’ families. Data from telephone interviews are weighted according to the sociodemographic characteristics of IDPs interviewed face-to-face. The combined dataset is analyzed using statistical weights for the merged dataset. The data gath- ered from the telephone survey of returnees are analyzed in a separate dataset without weights. Overall, we use four datasets in our analysis: •• Dataset with basic information about all household members of IDPs, from which the structure of IDPs by age, gender, and labor force status can be analyzed •• Main dataset with information provided by the head of household or any person aged 18 or older (only one person per household) about employment, housing, intention to return, integration, and so on •• Similar datasets for returnees to NGCAs. Source: IOM 2019. Disorganization The Economics of Winning Hearts and Minds The destruction of physical infrastructure and the displacement of people in Donbas have not been the only outcomes of the conflict since 2014. The conflict has also introduced many transactional frictions in a previously more integrated economic region. These frictions have not necessarily reduced the stock of pro- ductive assets directly, but they have reduced the rate at which the economy uses the surviving assets. This section summarizes such frictions in three categories: •• Disruptions in physical connectivity, which make the spatial movement of merchandise and people costly (sometimes prohibitively costly) •• Disruptions in coordination, which reduce efficiency of economic activity (especially in publicly provided services like electricity and water) •• Weakening social cohesion, which further reduces the efficient allocations of productive factors across different uses. 94 Disruptions in Connectivity Before the onset of the conflict, Donetsk and Luhansk Cities were the primary urban hubs of the Donbas region. As such, the transportation infrastructure that connected the region with the rest of Ukraine and neighboring Russia gravitated toward these centers. Therefore, smaller urban centers and other settlements were connected with the rest of Ukraine and Russia through these large urban centers. With the onset of the conflict, however, both urban centers remained in the NGCA territory. Together with conflict-driven damages to infrastructure, this led to major diversions of traffic, with significant increases in transaction costs, for all settlements in the NGCAs and some in the GCAs. Rail Transportation The Donetsk Railway system was one of the largest in the former Soviet Union. It was a part of Ukrzaliznytsia (Ukrainian Railways)—covering all of Donetsk and Luhansk Oblasts in addition to parts of Zaporizhzhia, Kharkiv, and Dnipropetro- vsk Oblasts—and it connected industrial Donbas to the rest of Ukraine as well as to Russia and the Caucasus. In the south, the Donetsk Railway connected the region to industrial Mariupol and its seaport (map 2.3). In the north, the railway connected the major cities of Donbas—such as Bakhmut (formerly Artemivsk), Horlivka, Kramatorsk, and Druzhkivka—with Kharkiv Oblast and central Ukraine, including Kyiv. Freight routes. Freight transport had been the main revenue generator for both Donetsk Railways and Ukrzaliznytsia. In 2013, the last year before the conflict, Donetsk Railway transported 139 million tons, including 97.9 million tons in Do- netsk Oblast and 42.4 million tons in Luhansk Oblast, according to the Centre for Transport Strategies. Together, the two oblasts accounted for 37 percent of freight traffic in the country. At the same time, coal accounted for a significant part of the transportation structure in both oblasts: 77 percent in Luhansk and 49 percent in Donetsk.12 The most active phase of the war, in 2014–15, significantly damaged the infra- structure of the Donetsk Railway. In many places, the GCAs have been cut off from the rest of Ukraine’s rail network, or the capacity of the government-con- The Consequences of the Conflict trolled rails has been significantly reduced. The Ukrainian government lost control of half the important railway junctions, such as Donetsk, Yasynuvata, Mykitivka, Debaltseve, Ilovaisk, Krasnaya Mogila, and Horlivka. By August 2014, Ukrzaliznyt- sia could not send freight cars from more than 50 Donetsk Railway stations and could transport only 40 percent of its prewar capacity. This was primarily be- cause of the damage done to the Donetsk–Pokrovsk–Chaplino railway branch (west of Donetsk) that carried freight out of Donbas to the rest of the country. In Donetsk Oblast alone, freight traffic decreased by 47 percent between 2013 and 2018, from 96.6 million tons to 45.4 million tons. 95 Map 2.3 Railroad Infrastructure in Donbas in Relation to Conflict Zones, 2019 a. Railroad stations in relation to conflict zones Statobilsk Rubizhne Lysychansk Sloviansk Schastya Kramatorsk Zolote Stanytsya-Luhanska Popasna Holubivka Bakhmut Luhanska Luhansk Pervomaisk Kadiivka Alchevsk Mayorske Mykit ivka Debaltseve Slovianoserbsk Herlivka Yenakiyeve Yasynovata Makiivka Vuglegirsk Khartsyzk Chyst iakove Donetsk Maryinka Novotroyitske Ilovaisk Hnutove Mariupol Railroad Conflict zone Source: ©World Bank; further permission required for reuse. Note: White (boxed) outline denotes the area covered in the assessment. Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine. Mariupol. Connections to the city of Mariupol were impeded at the very be- ginning of the war and continue to be a challenge today. Before 2014, the main railway to Mariupol passed through Donetsk and Yasynuvata. During the summer of 2014, both cities were taken over by the pro-Russian forces, and the main rail The Economics of Winning Hearts and Minds passageway to Mariupol was lost. Traffic to Mariupol is now possible through the Volnovakha–Komysh–Zorya–Polohy–Zaporizhzhia branch. However, this route has limited traffic capacity because at one point between the villages of Roziv- ka and Khlibodarivka, there is only one railway track. This route cannot provide enough traffic capacity to the city of Mariupol and its seaport. With disruptions in connectivity and economic activity, the cargo volume in Mariupol seaport—the region’s largest port and the country’s fifth largest—de- creased significantly (figure 2.4). As a result of widespread physical damages, losses in control of railway segments, and a limited de facto blockade by Rus- sia at the Kerch Strait, the Port of Mariupol lost important cargo clients in the north. These include metal giants like Yenakiyevskiy, Alchevsky, Makiyivsky, and Donetskstal, as well as clay and mining businesses that are now located in the NGCAs. Instead, the port started to handle more cargo for local businesses in 2015 and 2016. The list included two Mariupol giants—Ilyich Iron and Works and Azovstal Iron and Works—as well as the Satelit grain processing plant. 96 Figure 2.4 Cargo Turnover at the Port of Mariupol, Donetsk Oblast, by Cargo Type, 2013–18 16 14 Metric tons, millions 12 3.5 1.3 10 2.2 1.6 8 2.1 6 2.6 1.8 1.3 0.3 0.8 0.5 4 2 8.5 6.7 4.3 4.3 4.3 4.0 0 2013 2014 2015 2016 2017 2018 Metals Coal Clay Grains Other Source: Donetsk Regional State Administration 2019. Northern Luhansk Oblast. The northern part of Luhansk Oblast is another chal- lenging spot. One of the longest railway branches, the Lantrativka–Kondrashevs- ka–Nova (north of Luhansk), covers nine districts of Luhansk Oblast and has been cut off from the rest of Ukraine railways since 2014. The branch is parallel to the Lysychansk–Svatove railway branch, another major branch in the Luhansk GCA. The two branches meet near Luhansk City, which is not under Ukrainian government control, and connectivity between the two branches is not possible. This gap isolates businesses and residents of the northern part of the Luhansk GCA. The Ukrainian government has explored multiple options to connect the two branches of the Luhansk railways. Many government and rail officials in Ukraine favor the construction of new connections in the absence of reintegration pros- pects. Donetsk Railways estimated that it could cost up to Hrv 10 billion and take three to five years to connect the isolated Lantrativka–Kondrashevska–Nova branch to the Ukrainian railway. In February 2020, it was reported that France planned to finance the connection of this line for €100 million (Xinhua 2020). This project would greatly benefit Luhansk Oblast’s economy by overcoming its “island” status in connectivity. Air Transportation The Consequences of the Conflict The region’s air travel conditions have been hit harder than other transportation systems. Two large commercial airports in the Donetsk and Luhansk NGCAs (Do- netsk Sergei Prokofiev International Airport and Luhansk International Airport, respectively) were destroyed at the outset of the conflict and have not yet been repaired or rebuilt. The Donetsk airport, 10 kilometers (6.2 miles) northwest of Donetsk City, was built in the 1940s and 1950s and rebuilt in 1973. The last renovation, in 2011, pre- ceded the 2012 European soccer championship cohosted by Ukraine. A total of US$875 million was spent on the airport’s reconstruction, and before the conflict began in 2014, its passenger traffic grew by double digits annually, outpacing the overall growth rate of passenger traffic across Ukrainian airports. It suspend- 97 ed flights on May 26, 2014, after it was subject to shelling. The airport was lat- er destroyed during active fighting in the area between June 2014 and January 2015. Since then, it has not resumed operations, nor is there any open discussion about its rehabilitation. In the GCAs, three airports were built but not used for passenger traffic for sev- eral years prior to 2014. These airports (Severodonetsk, Kramatorsk, and Mariu- pol) currently have limited damage, but all would require significant rehabilita- tion or rebuilding before becoming fully compatible with passenger traffic. The Mariupol airport is partially functional, and of the three airports in the GCAs, can be restored faster and at less expense than the other two. The major issue for all three is proximity to the contact line. Security threats would continue, and possibly increase, if they are rebuilt. Thus, unlike the pro- posed project to build the missing segment of the railway connecting the two Luhansk branches described above, the economic case for investing in these airport facilities is not unambiguous. Coordination Problems Among the intangible effects of the conflict is the arbitrary split of previously coordinated activities. Because economic activities are segregated under dif- ferent military control zones, supply chains break, markets fragment, and scale economies in production disappear. These effects lead to artificially generated demand-supply mismatches and, overall, to inefficient organization of produc- tion and distribution. Water Supply and Distribution The public water system provides a prime example of the contact-line-driven disruption in coordination. In both Donetsk and Luhansk Oblasts, the contact line The Economics of Winning Hearts and Minds crosses important pipelines, pumping stations, and filtration stations, making the GCA and NGCA water systems interdependent. Currently, there seems to be a mutual understanding of this interdependency among the parties to the con- flict, which prevents crushing attacks on such facilities. However, limited attacks on this critical infrastructure have taken place frequently, leading to service de- livery disruptions. Additionally, in the absence of better cooperation across the contact line, the entire system has been in a suboptimal operation mode. Donetsk Oblast. In Donetsk Oblast, the water supply system (map 2.4) com- prises the Seversky Donets–Donbas channel (the main water supply source in Donbas since its construction in 1958), the South Donbas water supply network (WSN), and the Second Donetsk WSN. The Seversky Donets–Donbas channel originates from the Seversky Donets River (the main river in the region) and ends with the Verkhnekalmiussky reservoir in the suburbs of Donetsk City (131.6 kilo- meters in total). The channel provided potable water to 3.1 million people before the onset of the conflict and supplied major industrial centers through 14 outlets (Zapatrina 2020). Currently, several major installations in the NGCAs are sourced 98 by water coming from the GCAs and provide water for civilians on both sides of the contact line. Water of Donbas, a Ukrainian government-owned company, manages the water supply systems in Donetsk Oblast. As a result of the conflict, this utility was split in two. The GCA segment is based in Mariupol, and its 11,000 employees provide water services to 289 GCA settlements and sanitation services to 79 settlements. The NGCA segment in Donetsk City provides services to 19 NGCA settlements. Water of Donbas produced 454.5 million cubic meters of water in 2018, which is comparable to its 2013 production (470 million cubic meters). However, the company faces significant financial challenges because the NGCA consumes most of the water supply owing to the geography of the conflict and the nature of the water infrastructure, without a full cost recovery. Yet the company also cannot discontinue service, which would risk the lives of millions of civilians on both sides of the contact line. Luhansk Oblast. The water distribution systems of the GCA and NGCA in Lu- hansk Oblast are similarly interdependent and have similar coordination prob- lems. The oblast’s water system relies on the Siverskyi Donets River as its prima- ry water source, and the system is composed of pipelines that run north to south (map 2.5). Until 2014, the Luhanskvoda (Luhansk Water) utility supplied water in central and southern Luhansk. The contact line, which (in eastern Luhansk) runs along the Siverskyi Donets River, splits the Luhanskvoda systems, whose main office is in Luhansk City (within the NGCA) but whose other offices are in the GCA. At the moment, Luhanskvoda is the only water supply and sanitation (WSS) util- ity in the Luhansk NGCA. It provides WSS services to 95 percent of all residen- tial and 99 percent of all industrial customers in the NGCA. It also operates 97 percent of the water supply and 98 percent of the wastewater networks in the NGCA. However, most (about 70 percent) of the water sources—consisting of pumped water and boreholes—are in the GCA. These are operated by Ukrainian government-controlled utilities, and they also service the NGCA, without recov- ering costs, given the nature of the WSS infrastructure and the placement of the contact line. Overall, although nearly all the Luhanskvoda water sources are in the GCA, nearly all consumers (90 percent) are in the NGCA (Zapatrina 2020). The Consequences of the Conflict The Water-Electricity-Mining Nexus The conflict has also aggravated a different kind of coordination problem—be- tween the water supply, electricity and heat generation, and mining sectors. In the Donbas region, these sectors’ activities are tightly connected. Several mar- ket distortions (including large subsidies), combined with the region’s legacy economic system and infrastructure, have generated a greater interdependen- cy between these sectors than the average input-output connections through market mechanisms. Thus the conflict, with all its associated impacts—damage, displacement, and disorganization—has emerged as the proverbial “bull in the china shop” that upends this fragile interdependency. 99 Map 2.4 Water Distribution Coverage in Donetsk Oblast, by Settlement, 2015 The Economics of Winning Hearts and Minds Source: OCHA Humanitarian Response 2016. ©United Nations Office for the Coordination of Humanitarian Affairs (UNOCHA). Reprinted, with permission, from UNOCHA; further permission required for reuse. Note: WTP = Water Treatment Plant. 100 Electricity and water supply. The disruption of electricity supply is a major is- sue for the Donbas water systems. Electricity is often switched off as a result of unpaid electricity bills, leaky pipes, and damages to the power lines. In particular, the water supply system in Luhansk Oblast depends on the elec- tricity supply from the Luhansk TPP, a subject of frequent shelling and shortages of fuel for its functioning (box 2.3). The proximity of the infrastructure to the demarcation line causes damages to the power lines. For example, after several shelling instances in July and September 2017, the entire Luhansk GCA and some parts of the NGCA were left without electricity. Earlier, in the fall of 2016, several pumping stations on the Karbonit pipeline were switched off from the electricity grid as a result of unpaid electricity bills (Streltsova 2016). The pumping stations were not functioning for 10 days, and nearly 600,000 residents lost their water supply on both sides of the contact line. This incident is similar to many others that affect water supply in Luhansk Oblast. Because of Ukrainian legal restrictions on payments between GCA and NGCA entities, no mechanism was in place for payments to be issued in 2016 between Luhanskvoda and the regional water utility, Popasna Vodakanal (ACAPS 2019). As a result, debts accrued for water supplied to the NGCA, which meant that Popasna Vodakanal could not pay the government-controlled Luhansk Energy Association (LEO) for the electricity used by its pumping stations for distributing water. In this situation, the International Committee of the Red Cross in Ukraine called for both sides to find a way to pay for electricity. With the OSCE’s help, both sides agreed on a mechanism for transferring funds, whereby NGCA representa- tives carry cash in suitcases to the Ukrainian side at a specific time (DN 2020). This is not the only example. On at least two other occasions, LEO cut off elec- tricity to multiple water distribution stations at once, which subsequently af- fected water distribution to the NGCA. The electricity at the Petrivske pumping station was switched off six times between February 2017 and January 2018. Water supply and heat generation. In the winter, lack of water supply to these cities has serious consequences. Without water, the Soviet-style heating sys- tems stop working, which can damage the entire system. This, in turn, increases households’ reliance on electric appliances to heat their homes, which could overload the electricity networks. The Consequences of the Conflict At a September 2019 panel discussion sponsored by the Russian-German Ex- change (DRA), a Berlin-based nongovernmental organization, speakers discussed the huge connection between the water and other critical infrastructure in east- ern Ukraine. One panelist described the situation as follows (CivilM+ 2019): Ukraine has very severe winters with temperatures dropping to -30 degrees Celsius. Without heating, it is impossible to survive. Most of the settlements in Eastern Ukraine have centralized heating systems. Those heating systems are outdated and lose a large proportion of water through leaks. In the win- ter, a constant supply of fresh water is needed to keep the system running. If the supply stops, the system can freeze, and restarting it can become a se- rious challenge. Ukraine has had emergencies like this prior to the conflict. 101 Map 2.5 Water Distribution Coverage in Luhansk Oblast, by Settlement, 2015 The Economics of Winning Hearts and Minds Source: OCHA Humanitarian Response 2016. ©United Nations Office for the Coordination of Humanitarian Affairs (UNOCHA). Reprinted, with permission, from UNOCHA; further permission required for reuse. For a functioning government with substantial resources, it took eight days to re- store heating in Alchevsk after the heating system broke down in January 2006. Since then, infrastructure in Alchevsk and other cities around the front line have aged even more and need continuous maintenance to ensure that no emergen- cies like this occur. (Mark Buttle, WASH Cluster Coordinator for Ukraine, UNICEF). 102 Box 2.3 What Disorganization Looks Like: The Story of the Luhansk TPP The Luhansk thermal power plant (TPP), the only one in Luhansk Oblast, is located near the town of Shchastia, 26 kilometers north of Luhansk City. As a result of the war, it is completely isolated from the Ukrainian grid, and the conflict in Donbas constantly threatens its shutoff and a regional blackout. The power plant provides electricity to more than a million people and to dozens of mines and plants. The power station belongs to DTEK, Ukraine’s largest energy company. Since 2014, the town and the plant have been under the Ukrainian government’s control, but the station is within shooting range of the contact line and has been a place of constant tension since the start of the conflict. It was re- peatedly shelled, and four of its energy blocks were destroyed in 2014 and cannot be repaired (photo B2.3.1). As a result, about 1 million people lost their electricity in the summer of 2014. DTEK repaired two blocks, and the station resumed its operation in September 2014. Photo B2.3.1 Conflict-Damaged Luhansk TPP, 2014 Source: © NurPhoto/Corbis via Getty Images. Used with the permission of Nur Photo/Corbis. Further permission required for reuse. Note: TPP = thermal power plant. The Consequences of the Conflict However, that was not the end of troubles for the Luhansk TPP. Because all of Luhansk Oblast depends on it for energy, heavy fighting for the station lasted until almost the end of 2015. The fierce confrontation stopped only after the Ukrainian military mined the area around the plant and prom- ised to blow it up if the pro-Russian forces tried to attack. In response, the pro-Russian forces blew up a railway bridge across the Siversky Donets River, by which the coal to the plant had been coming from DTEK Energo’s anthracite holdings of the companies Sverdlovantratsyt and Rovenkyan- tratsyt. 103 To this day, the Luhansk TPP is in a difficult position. It is half-destroyed, with a minefield nearby and armed activities constantly taking place. Even the light artillery can reach the power plant from the NGCA. In addition, there is no railway track to resume the supply of coal from Ukraine. Until 2017, the coal made its way to the plant from the NGCA. However, in March 2017, all DTEK companies in the NGCAs were seized by the self-proclaimed Donetsk People’s Republic and Luhansk People’s Republic. As a result, all companies, including DTEK, lost control over their assets. Authorities did not have many options. Constructing a new railway branch to the Luhansk TPP is a dangerous and lengthy process, and converting to gas coal is too expensive. As a solution, DTEK came up with the alternative to supply coal from its as- sets on Russian territory, through the railway in the northern part of the re- gion. The Russian coal started reaching the Luhansk TPP in April 2017, com- ing from the Obukhovskaya mine (also a DTEK property) in Russia’s Rostov region. On October 10, 2018, however, the supply of coal by this branch was stopped as Russia implemented new limitations on trade with Ukraine. On November 6, 2018, the Luhansk TPP was forced to switch to gas. Since then, the power plant had to switch between gas and coal from Russia sev- eral times. As reported by the Ukrainian media, the Russian coal supply re- sumed again in the spring of 2020 (IEACCC 2021). But the unreliable supply from Russia makes the operation of the station unpredictable, and the use of gas doubled the cost of energy production. Because of the high gas costs, the TPP’s cost per kilowatt hour exceeded the selling price on Ukraine’s electricity market (“Energorynok”) in recent years. The TPP owes more than Hrv 1 billion to the Ukrainian government for the supplied gas, and in turn, the Energorynok owes the Luhanska TPP Hrv 1.9 billion for the provided electricity. The Economics of Winning Hearts and Minds Electricity and coal mining. A similarly excessive interdependency is also evi- dent between electricity generation and coal production. After the onset of the conflict, the coal production in Ukrainian-controlled territories became insuffi- cient to supply the government-controlled TPPs. This shortage was driven by (a) the loss of government control in anthracite coal production facilities (on which about half of all TPPs had relied), and (b) the Ukrainian government’s 2015 deci- sion to roll back subsidies to the coal industry. As a result, the cost of TPP-pro- duced electricity doubled by the end of 2015, and the country’s anthracite im- ports increased from about US$1.5 billion in 2016 to about US$3 billion in 2018. At the same time, the coal-dependent TPP units were gradually converted to gas-using ones (UNIAN 2018). However, this excess demand problem for local coal turned into an excess sup- ply problem in 2019. In September 2019, Ukrainian authorities resumed electric- ity imports from Russia for the first time since 2015. As cheaper electricity from Russia and Belarus flooded the local energy market (which Russia sold at a lower 104 price than in Russia itself), local energy production took a hit. With surplus elec- tricity generation capacity, the Ukrainian power transmission operator Ukren- ergo forced all TPPs to operate in the “minimum allowable regime” in November 2019 (only one production unit worked at every plant). Thousands of workers were furloughed without pay. The decrease in electricity generation at TPPs reduced demand for coal, which began accumulating in the TPP sites and in the coal mines’ storage areas (about 1.5 million tons of local coal were reported unclaimed), and these miners were also furloughed without pay.13 In the first months of 2020, the coal industry of Ukraine appeared in crisis. The authorities came under pressure to stop import- ing electricity, which they did by decree on April 8, 2020.14 Implications of Disrupted Coordination These two cases—the water systems division and the water-electricity-mining nexus—provide textbook examples of how arbitrary barriers can disrupt eco- nomic systems. In both cases, frictions between the two sides of the organiza- tional structure (GCAs and NGCAs) lead to inefficient economic processes. For water, this is driven by a division of productive assets (channels and pump- ing stations) between the Donbas GCAs and NGCAs. And for the water-elec- tricity-mining nexus, the inefficiency is driven by a spatial heterogeneity in eco- nomic activity, which in turn leads to demand-supply mismatches. These problems both point to sources of economic inefficiency and define chal- lenges for future recovery strategies depending on how the arbitrary frictions may evolve in the future. Weakening Social Cohesion Another important component of disorganization involves social cohesion. This corresponds to a large number of concepts and indicators depending on dis- ciplinary definition; however, for the purposes of this study, we consider it to be the extent of connectedness and solidarity among groups in a society. It in- The Consequences of the Conflict cludes two main components: “the sense of belonging of a community and the relationships among members within the community itself” (Manca 2014, 261). Each of these characteristics constitutes an end in itself, but they are also im- portant determinants of economic efficiency. Unlike land, labor, and capital, they do not contribute to economic processes directly. However, frictions in these dimensions could hamper economic activity to some degree. A systematic account of changes in social cohesion indicators between pre- conflict and postconflict periods in Ukraine is not available. However, the So- cial Cohesion and Reconciliation Index (SCORE) surveys—which the Centre for Sustainable Peace and Democratic Development (SeeD) conducted annually for this report between 2015 and 2019—provide a useful benchmark to assess cross-sectional differences and changes over years, albeit in a limited fashion.15 105 Table 2.4 Indicators for Intergroup Contact and Openness to Dialogue with Specified Groups in Ukraine and Donbas GCAs, 2017–19. Average respondent score (0–10) Intergroup contact Openness to dialogue Ukraine Ukraine GCAs GCAs GCAs Ukraine Ukraine GCAs GCAs GCAs Group 2017 2018 2017 2018 2019 2017 2018 2017 2018 2019 IDPs 3.4 2.7 4.3 3.8 3.5 4.9 4.6 7.1 7.6 7.5 People from 3.8 3.7 n.a. n.a. n.a. 4.7 4.7 n.a. n.a. n.a. eastern Ukraine People living in n.a. 1.2 3.1 3.0 2.8 n.a. 3.2 6.5 6.9 6.8 NGCAs People from western 4.8 4.0 2.5 2.1 2.0 5.8 5.3 6.6 6.5 6.3 Ukraine People with pro-EU 4.4 3.9 2.6 2.5 2.7 5.8 5.3 6.3 6.7 6.6 orientation ATO personnel n.a. n.a. 2.2 2.4 2.4 n.a. n.a. 6.1 6.5 6.0 and veterans People with pro-Russia 3.6 3.3 2.8 2.9 2.7 4.7 4.4 5.9 6.5 6.4 orientation People supporting 0.8 0.1 1.7 1.7 1.8 2.4 2.4 5.4 5.8 5.9 secession of Donbas NGCAs Ukrainian 0.2 n.a. 0.7 1.1 1.0 4.7 n.a. 4.7 5.0 4.8 nationalists Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). Note: Values are measured on a 0–10 scale, with 0 being no contact or openness to dialogue with a given group and 10 being maximum contact or openness to dialogue. “GCAs” refers collectively to the government-controlled areas of Donbas (in the Donetsk and Luhansk Oblasts). ATO = antiterrorist operation; EU = European Union; IDPs = internally displaced persons; NGCAs = non-government controlled areas; n.a. = not applicable. The Economics of Winning Hearts and Minds Social Proximity Among other things, the survey explores social proximity—that is, the extent to which one accepts members of various groups in society as neighbors, friends, or family in-laws. This measurement begins with the survey respondents’ views of personal identity. About 89 percent of respondents in the Donetsk and Lu- hansk GCAs still identify as Ukrainian, either ethnically or civically. In addition, locality identity (in this case, an attachment to the Donbas region) is a salient, primary identity for 94 percent of respondents in the Donbas GCAs. This dualis- tic (national and regional) spatial identity, along with strong support for linguistic and political diversity in Donbas, highlights the surprising pluralism in the region. Between the Donbas GCAs and NGCAs, however, divergent views of primary identity are more significant. In 2019, only 25 percent of NGCA residents main- tained that they identify first as Ukrainian, with another 22 percent claiming to be citizens of Ukraine, and 35 percent identifying first as Russian. These 47 per- 106 cent of NGCA respondents who identify first as either Ukrainian or as a Ukrainian citizen is little more than half of the 89 percent of the GCA respondents who identify in the same way. Intergroup Relations To assess “intergroup relations”—another factor affecting social proximity—re- spondents are asked (a) how often they communicate with people from a menu of various groups; (b) which members of these groups they would be willing to engage in dialogue; (c) whether they can envision having members of each group as close friends, coworkers, or family members; and (d) finally, to assess social threat, which groups they see as endangering their well-being or the unity of their community. Among the results (table 2.4), Donbas GCA respondents scored 2.0 (out of 10) on contact with “people from western Ukraine”—reflecting the survey result that 88 percent of them rarely or never interacted with western Ukrainians. General trends both nationally and in the Donbas GCAs show a decrease in exposure to other social and political perspectives over time.16 Predictably, a decrease in interactions with other groups correlates with de- creasing openness, over time, to dialogue with other groups. Although behavioral changes and the breaking down of stereotypes often depend as much on the quality of interaction with members of other groups as on the quantity of inter- actions, it is also typical that increased exposure leads to more willingness to engage in dialogue as familiarity increases. Generally, the trends both nationally and in the Donbas GCAs reveal less interaction and less openness toward engag- ing members of other groups in 2018 and 2019 than in previous years, although the average GCA responses indicate more openness there than in Ukraine as a whole to dialogue with other groups. Measures of Trust toward Reforms, Perceptions of Corrup- tion Trust toward reforms. Another important measure of social cohesion concerns citizens’ trust regarding policies and the public agents of such policies. To as- The Consequences of the Conflict sess this, respondents were asked about their level of awareness of various re- form initiatives and the extent to which they agreed or disagreed with various reform-oriented statements. This was followed by a series of questions assess- ing support for particular reforms, including health care and decentralization. Levels of support for reforms (table 2.5, panel a) are closely aligned with trust in local and state institutions (table 2.5, panel b); trust in media outlets as sources; perceived levels of corruption; and the degree to which respondents believe corruption is just a fact of everyday life (figure 2.5). General skepticism  toward reforms in the Donbas GCAs remains comparable to nationwide trends on this measure in 2019, with Luhansk GCA respondents registering higher-than-average overall mistrust of reform processes. 107 Table 2.5 Indicators of Support for Reforms and Institutional Trust in Don- bas GCAs a. Support for reforms, 2019 Donetsk Luhansk Type of reform Oblast Contact line Oblast (GCA) Contact line Decentralization reforms 4.3 3.9 4.1 3.6 Health reforms 4.1 3.4 3.7 3.8 b. Trust in institutions, 2017–19 Donetsk Luhansk Oblast Contact line Oblast Contact line Type of institution 2017 2018 2019 2018 2019 2017 2018 2019 2018 2019 Central institutions 1.6 1.8 1.4 1.4 3.8 2.0 1.7 4.4 1.5 3.6 Local institutions 4.6 4.1 4.5 3.5 4.1 4.6 4.0 4.1 4.4 3.5 Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). ©World Bank; further permission required for reuse. Note: Values are measured on a 0–10 scale, with 0 being no support for the reform (panel a) or trust in institutions (panel b) and 10 being maximum support or trust. Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine. The “contact line” separates each oblast’s government-controlled area (GCA) from its non-government controlled area (NGCA). Contact line areas were not surveyed in 2017. Strong mistrust toward reforms is closely linked with perceived levels of cor- ruption. In 2018, nearly 70 percent of nationwide respondents thought that re- forms would only benefit the elite and that reform-oriented initiatives are only for publicity. In 2019, 60 percent of respondents in the Donbas GCAs thought that reform rhetoric is for publicity, and 62 percent thought that reforms would not benefit ordinary people.  The Economics of Winning Hearts and Minds Significant mistrust toward and overall lack of support for reforms also correlates with low awareness of the details of announced reforms. Awareness of the provi- sions of health reform and of recent land reform legislation is lowest in Luhansk,17 even though each reform process could directly affect the lives of residents in the east. Trust in central and local institutions. This lack of informed opinion and poor regard for reform processes results partly from low trust in the news media and equally low trust in central and local institutions. Respondents nationwide, but especially in the Donbas GCAs, generally distrust the information they receive about reform processes—as much as they distrust those who are tasked with implementing them. They make an exception for anticorruption measures, however. Over 80 percent of respondents throughout Ukraine strongly support anticorruption initiatives in principle although they know little about specific anticorruption proposals. 108 Table 2.6 Trust in Central and Local Institutions in Donbas GCAs, 2018–19 Share of respondents expressing trust, percent Trust level Institution 2018 2019 President 8.5 61.3 CSOs and NGOs — 53.4 Armed forces 36.3 49.0 Village head or mayor — 47.1 Village administration 41.8 43.3 Verkhovna Radaa 6.1 34.5 Cabinet of Ministers 7.5 33.0 Police 29.9 32.2 Oblast administration 20.6 31.5 Courts 24.9 11.4 Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). Note: Donbas comprises the Donetsk and Luhansk Oblasts. CSO = civil society organization; GCA = government-controlled area; NGO = nongovernmental organization; — = not available (excluded from 2018 survey). a. Verkhovna Rada is Ukraine’s unicameral parliament. Table 2.6 lists the institutions included in the SCORE measures of trust for re- spondents in the Donbas GCAs in 2018 and 2019. The most striking figures in- clude the support for the president, which skyrocketed between 2018 and 2019, although it is important to note that the SCORE survey wave for 2019 took place shortly after Volodymyr Zelenskyy’s victory in an election that saw overwhelm- ing second-round support for his campaign in Donbas. This survey result is an outlier, to be contrasted with 2018 data showing low levels of support for the previous president and all central institutions. Overall, Donbas GCAs exhibited a remarkable increase in trust in central insti- tutions in 2019, consistent levels of trust toward local institutions in both years, The Consequences of the Conflict and enduring low levels of faith in both the courts and policing over the period. The dramatic increase in levels of trust for central institutions should be under- stood within the context of regionwide enthusiasm for Zelenskyy and his cabinet during and after the election cycle of 2019. Zelenskyy appears to have afforded other central institutions the benefit of the doubt in the 2019 panel survey wave. Perceptions and tolerance of corruption. Although the “trust” levels increased markedly in Donbas with the election of a president who is also favored by the region, some other aspects of social cohesion remained problematic. The SCORE surveys also include questions on “perceived levels of corruption” and “tolerance of corruption.” In the Donbas GCAs, one-half to two-thirds of panel respondents perceived corruption “sometimes” or “always” across a wide range of social behaviors in 2019 (figure 2.5). 109 Figure 2.5 Perceived Levels of Corruption in Donbas GCAs, by Behavior Type, 2019 Local authorities ask for additional payments to provide services 25 15 22 18 20 Teachers put higher marks if one pays 13 15 25 26 21 Police won’t register traffic 10 13 28 29 21 violations if paid Doctors are willing to provide 6 9 29 50 6 higher-quality assistance Police are corrupt and are covering up organized crime 5 11 33 38 13 Judges or prosecutors can be bought if offered enough money 4 7 30 44 14 Parliamentarians can be bribed 4 9 27 40 20 to propose 0 20 40 60 80 100 Share of respondents (%) Never Rarely Sometimes Always DK Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). Note: Donbas comprises Donetsk and Luhansk Oblasts. DK = don’t know; GCAs = government-controlled areas. Nationally, there was a small decrease in the perceived level of corruption be- tween 2017 and 2018, yet there was also an increase in tolerance of corruption overall (table 2.7). This increase was most evident in the south and west as well as Kyiv City. The higher the tolerance of corruption, the less likely respondents were to perceive bribery and other similar practices as corruption—or to believe anything can be done about it.  Perceived levels of corruption are lower in the Donbas GCAs than in nationwide measures, but the responses suggest that levels of corruption are increasing. Tolerance of corruption was also much higher in the Donbas GCAs in 2019 than in 2017. While respondents in the Donbas GCAs sense more corruption, they also evince greater tolerance for it, possibly signaling enduring dissatisfaction with The Economics of Winning Hearts and Minds service delivery but greater reliance on corrupt practices as a coping mecha- nism to improve access to those services. Synopsis This chapter focused on the nature of the conflict (that is, its drivers and emerg- ing patterns) and the mechanisms through which its economic and social fallout has taken place so far. The analysis employed a “3 Ds” framework (destruction, displacement, and disorganization) to classify these effects. Several key obser- vations emerged from this exercise. 110 Table 2.7 Trends in Perceived Corruption and Tolerance of Corruption in Ukraine and Donbas GCAs, 2017–19 Average respondent score (0–10) Ukraine Ukraine Donbas GCAs Donbas GCAs Donbas GCAs View measured 2017 2018 2017 2018 2019 Perceived corruption 7.5 7.2 6.4 6.0 6.5 Tolerance of corruption 1.5 2.2 2.3a n.a. 4.5 Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). Note: Values are measured on a 0–10 scale, with 0 being no perception or tolerance of corruption and 10 being maximum perception or tolerance of corruption. Donbas comprises the Donetsk and Luhansk Oblasts. GCA = government-controlled area; n.a. = not applicable. a. Score is derived from nationwide data for that year. Drivers of the conflict. The conflict in Donbas was driven by several internal and external factors that crated enabling conditions for the breakout of violence. Three phases of the conflict. The first phase of hostilities (2014–15) imposed the largest numbers of casualties, the greatest physical damage, and the most forced displacement. The following two phases (2015–18 and 2019–present) were characterized by low-intensity skirmishes (often near the contact line) and short-lived cease-fires, with an improving outlook since 2019. Destruction. The conflict inflicted considerable damages on strategic infra- structure—especially in the transportation, water and sanitation, and energy sectors—particularly in the first phase of the conflict. •• Potential for statistical misinterpretations: Many official statistics report oblast-level information before 2014 and only GCA-level data thereafter. Thus, the statistics are not conducive to trend analysis (such that a differ- ence between 2013 and 2014 would indicate a real change). •• Data needs: Most data on damages are outdated, necessitating an on-the- ground, comprehensive, and up-to-date damage and needs assessment. Displacement. Close to 1.5 million Ukrainians remained in IDP status by July The Consequences of the Conflict 2020, almost half of whom were in the GCAs of Donetsk and Luhansk Oblasts. •• Challenges faced by IDPs: IDPs who moved away from Donbas are likely to face more significant challenges in finding a job corresponding to their ed- ucation and skills, and thus they need retraining and job-search assistance. •• Returnees: The IDPs who have returned to their places of origin are signifi- cantly older than IDPs and GCA populations in general, less likely to have a tertiary education, and more likely to face housing and employment prob- lems at the place of displacement. •• Data needs: Demographic statistics in the region (registration-based) are not reliable. There is an urgent need for a comprehensive census. 111 Disorganization. Several “invisible” factors magnified the economic impact of the conflict. •• Disruptions in connectivity: The contact line has cut off the region’s trans- portation network, leading to hikes in transportation costs and creating transportation islands, especially in railways. The region’s two major airports (Donetsk and Luhansk airports) are destroyed, and the region’s major sea- port (Mariupol) is performing well below capacity. Actions to mitigate isola- tion include some low-hanging fruits; however, larger projects are impeded by the looming risks associated with the active conflict. •• Coordination problems: The contact line cuts through major public service provision systems, leading to a fragile interdependency, such that low coop- eration across the contact line leads to a suboptimal (and costly) provision of services. This is particularly evident in the region’s water system. With legacy distortions (such as subsidies and trade protection), certain sectors exhibit high interdependencies (a water-electricity-mining nexus). When electricity imports were allowed, domestic electricity production was unprofitable, and there was excess coal supply in the market; when it was not allowed, there was excess demand for anthracite, which needed to be imported. •• Weakening social cohesion: The conflict has deepened the country’s divi- sions, which are more prevalent in the nonconflict areas: survey respondents in the east are more open than those in the west to dialogue with other groups. At the same time, the 2019 presidential elections marked a hopeful outlook. For the first time since independence, significant majorities of vot- ers in both Donbas and nearly all western and central regions of Ukraine vot- ed in similar fashion, with the Donbas GCAs exhibiting a remarkable increase in trust in central institutions in 2019. Nevertheless, deep suspicions persist toward reforms, and there are incessant perceptions of corruption. Overall, we observe that although the conflict was not the only source of prob- lems in the Donbas region, it added to them. First, it worsened many structural problems that existed before, including the double-aging problem: a faster aging of infrastructure by means of physical destruction and a deeper aging of de- The Economics of Winning Hearts and Minds mography by means of selective displacement. Second, it created new prob- lems like disorganization (disruptions in connectivity, coordination problems, and weakening social cohesion). The next chapter analyzes how these factors have affected the region’s eco- nomic and social conditions and which specific challenges lie in the way to pro- moting economic prosperity in the region and in Ukraine as a whole. 112 Notes 1. For a detailed discussion on how grievances related to exclusion—from access to power and natural resources to security and justice, for example—can drive conflicts and how development processes can interact with such factors, see World Bank and UN (2018). 2. A large literature analyzes whether nighttime light emissions (and other remote-sensing-based assessments) can provide sufficient proxies for different indicators including economic activity. For examples, see Henderson, Storeygard, and Weil (2012) and Mellander et al. (2015). In conflict settings, such proxies are particularly useful (because data are scarcer) but less reliable. This is mainly because the conventional relationships between economic activity and the provision of public services (for example, electricity through a grid) are often broken by conflicts. Violent acts such as attacks on electric grids can lead to a near collapse of light emissions. While this, by itself, would lead to a significant decrease in economic activity, we would expect such a decrease to be more limited than the loss in light emissions. Therefore, caution is ad- vised in using remote-sensing-based methods in the context of conflicts. 3. For more details about these incidents, see the United Nations Children’s Fund (UNICEF) WASH Cluster Incident Reports for Ukraine at https://www.humanitarianresponse.info/en/operations/ukraine/water-sanitation-and-hygiene. 4. According to the Organisation for Economic Co-operation and Development (OECD) Glossary of Statistical Terms, “The de facto population is a concept under which individuals . . . are recorded . . . to the geographical area where they were present . . . at the specified time” (https://stats.oecd.org/glossary/detail.asp?ID=571). “The de jure population is a concept under which individuals . . . are recorded . . . to a geographical area on the basis of the place of residence” (https://stats. oecd.org/glossary/detail.asp?ID=580). The difference between the two concepts in Ukraine is the difference between the number of people recorded as temporarily living in a given region and those recorded as temporarily absent from that region during the last census in December 2001. This difference between de facto and de jure populations in Donetsk and Luhansk Oblasts has been fixed at 12,885 and 4,597 individuals, respectively, since January 2002. 5. The survey targeted Ukrainians aged 18–75 years, except those living on occupied Ukrainian territory, IDPs, foreigners, stateless persons, and Ukrainian citizens who do not have a registered domicile. 6. In the absence of a more complete set of statistics, one approach would be to rely on the SSSU’s regional monthly bulle- tins between January 1, 2007, and January 1, 2020. These provide detailed population statistics for each city of regional importance as well as rayons (second-level administrative divisions, below the oblast level) in both the GCAs and NGCAs of Donetsk and Luhansk Oblasts. Accordingly, between January 2014 and January 2020, the population in the GCAs and NGCAs decreased by 7.4 percent and 3.3 percent, respectively. However, when compared with the preconflict trends, these numbers are too small, indicating a possible underestimation. For instance, Ukraine’s total migration flows also seem to have declined over 2014–19 compared with the preconflict period, which is not consistent with the IDP statistics showing major flows. 7. The MoSP registers IDPs entitled to government assistance, pensions, and free housing for a period of up to six months, with the possibility of extension. However, these data are also not free of measurement issues. On the one hand, the MoSP registry might not cover all IDPs: some displaced people might not need support from the government, would like to avoid conscription, or lack necessary documents. This would lead to an undercount. On the other hand, many residents of NGCAs reportedly register as IDPs for access to pensions and other benefits without being displaced. This would lead to an overcount. It is not possible to quantify these factors and their net impact on IDP statistics. 8. We use the data for IDPs and returnees to Donbas from four rounds: March 2018 (round 9) to December 2018 (round 12). Thus, we can compare the 2018 information about IDPs and returnees with the profiles of the working-age populations of Donbas and the rest of Ukraine from the 2018 Labor Force Survey (LFS). Given the large differences in survey outcomes between Donetsk and Luhansk Oblasts described earlier in the chapter, we present information for these two regions separately. From the original NMS sample of IDPs, we excluded people from Crimea or Sevastopol to focus the analysis on IDPs from Donbas. One of our subsamples also excludes individuals older than 70 years from the samples of IDPs and returnees to enable comparison of sociodemographic and labor market characteristics with those of the same approxi- mate age group in the LFS. For the dataset of IDPs, we applied sample weights calculated on the basis of sample weights for the merged datasets from the face-to-face and phone interviews provided in each round (specifically, we divided original weights by 4). The data on returnees do not include sample weights because there is no information about the total population who returned to the NGCAs. 9. Potentially, IDPs who moved to the Donetsk or Luhansk GCAs or to other oblasts could also have been part of the LFS sample. But there is no information in the LFS data to distinguish between IDPs and other people. We believe that the number of IDPs in the LFS sample is negligible, and therefore we use the LFS-based population as a comparison group for IDPs. 10. This subgroup appears among the open-ended answer, “Other,” in table 2.3. In the sample of 2,738 returnees from four rounds in 2018, 24 individuals reported having registered as IDPs in the GCA to get a pension. 11. A comparison between (a) IDPs or returnees, and (b) other Ukrainians is not feasible because the former is reported for household heads by the NMS and the latter for all individuals by the LFS. 12. All freight traffic data from Transport Book 2020 by the Centre for Transport Strategies, Kyiv, available at https://en.cfts. The Consequences of the Conflict org.ua/transport-book. 13. Other factors also contributed to a decrease in demand for coal: an extremely warm winter, the general economic decline, and the coronavirus pandemic (quarantine was introduced in Ukraine on March 12, 2020). 14. National Energy and Utilities Regulatory Commission of Ukraine (NEURC), Resolution No. 766 of April 8, 2020, “On Actions of Electricity Market Participants During the Quarantine Period and Restrictive Measures Related to the Spread of Coro- navirus (COVID-19).” 15. The first wave of SCORE surveys in Ukraine took place between August 2015 and December 2016, with a nationwide sample size of 10,278 persons from among 24 oblasts and Kyiv City. A subset survey used phone sampling of 641 persons in NGCAs. In 2017, SCORE surveys were administered in the five easternmost oblasts of Ukraine (the Donetsk GCA, the Lu- hansk GCA, Kharkiv, Dnipropetrovsk, and Zaporizhzhia). The Donetsk and Luhansk GCAs were further divided into clusters of comparable rayons (six clusters in Donetsk and four in Luhansk). The sample sizes were 350 persons per cluster, with 600 respondents each in Zaporizhzhia, Dnipropetrovsk, and Kharkiv Oblasts for a total of 5,300 respondents. The 2017 sample also included surveys with (a) local experts in rayons throughout the region; (b) 1,500 NGCA residents, interviewed at crossing points into government territory; and (c) 3,300 adolescent students attending schools in the five oblasts. A survey replicating the 2017 sample was completed in 2018 for eastern Ukraine, with the addition of 700 respondents living along the contact line. Also in 2018, additional SCORE surveys were administered nationwide in 24 oblasts and Kyiv City for a total of 9,018 respondents. An added 1,042 respondents from the NGCAs participated, with 445 face-to-face interviews in the Donetsk NGCA and 597 phone interviews in the Luhansk NGCA. As with the successive rounds of SCORE surveying in 2017 and 2018 focused on eastern Ukraine, this 2018 wave reproduced the nationwide sampling undertak- en in 2016. Over 81 percent of the 2018 nationwide sample in the GCAs had participated in the 2016 wave, creating a panel cohort of 6,102 persons. In 2019, a randomized survey of eastern Ukraine was conducted between August and November with a total sample size of 9,055, including 3,325 respondents from the Donetsk and Luhansk GCAs; 1,811 from 113 contact-line areas; and specialized subset boosters for veterans and urban residents of 15 cities (or “hubs”) within the Donetsk and Luhansk GCAs. A separate booster of 619 residents using a snowball methodology and phone sampling was employed to reach residents in the NGCAs. 16. Reasons for less contact with IDPs may be explained by long-term IDPs beginning to integrate into host areas and hence declining to present or introduce themselves as IDPs. Decreased interaction with people from western Ukraine in Donbas may be because of the decrease in volunteers in the east as humanitarian efforts wind down. Decreased interaction with pro-EU persons at the national level may be explained by conflict fatigue and avoidance of political discourse. 17. Data on understanding of land reform were collected in focus group discussions in the Luhansk GCA from November 2019 to February 2020. References ACAPS (Assessment Capacities Project). 2019. “Ukraine: Conflict in Donetsk and Luhansk.” Briefing note, November 4, ACAPS, Geneva. CivilM+. 2019. “7.5L of Water per Day: ‘Pipelines and Frontlines—Cooperation and Conflict in Eastern Ukraine.” CivilM+ News, September 23. https://civilmplus.org/en/news/7-5l-of-water-per-day-pipelines-and-frontlines-cooperation-and-conflict-in- eastern-ukraine/. DN (Donetskie Novosti). 2020. “Bring 58 Tons of Small Change: ‘LPR’ Pays for Water in Coins.” Donetsk News, February 19. https://dnews.dn.ua/news/741785. Donetsk Regional State Administration. 2019. “Development Strategy of Donetsk Region for the Period Up to 2027.” Krama- torsk, Ukraine. https://dn.gov.ua/ua/projects/strategiya-rozvitku-doneckoyi-oblasti-na-period-do-2027-roku. EU, UN, and World Bank (European Union, United Nations, and World Bank). 2015. “Ukraine Recovery and Peacebuilding Assess- ment: Analysis of Crisis Impacts and Needs in Eastern Ukraine.” Volume I: Synthesis Report. EU, Brussels; UN, New York; World Bank, Washington, DC. Henderson, J. Vernon, Adam Storeygard, and David N. Weil. 2012. “Measuring Economic Growth from Outer Space.” American Economic Review 102 (2): 994–1028. IEACCC (IEA Clean Coal Centre). 2021. “Ukraine Cuts Spending in Foreign Currency on Coal Imports by 40%.” News (from UNIAN Information Agency), IEACCC, January 21. https://www.iea-coal.org/ukraine-cuts-spending-in-foreign-currency-on- coal-imports-by-40/. IOM (International Organization for Migration). 2019. “National Monitoring System Report on the Situation of Internally Dis- placed Persons, June 2019.” Report of the IOM Mission in Ukraine, Kyiv. Manca, Anna Rita. 2014. “Social Cohesion.” In Encyclopedia of Quality of Life and Well-Being Research, edited by Alex C. Michalos, 261. Dordrecht, Netherlands: Springer. Mellander, Charlotta, José Lobo, Kevin Stolarick, and Zara Matheson. 2015. “Night-Time Light Data: A Good Proxy Measure for Economic Activity?” PLoS ONE 10 (10): e0139779. NRC (Norwegian Refugee Council). 2018. “Humanitarian Shelter Damage Assessment: Contact Line Communities in GCA, Luhansk Oblast, Ukraine.” Report, NRC, Kyiv, Ukraine. https://www.nrc.no/globalassets/pdf/reports/ukraine/damage-assess- ment-report_nrc_luhansk_loc.pdf. NRC (Norwegian Refugee Council). 2019. “Humanitarian Shelter Damage Assessment: Contact Line Communities in GCA, The Economics of Winning Hearts and Minds Donetsk Region, Ukraine.” Report, NRC, Kyiv, Ukraine. https://www.nrc.no/globalassets/pdf/reports/ukraine-humanitarian-shel- ter-damage-assessment-2019/damage-assessment-report-eng.pdf. Slobodian, Oleksandra, and Anastasia Fitisova. 2018. Registering Domiciles in Ukraine: How the System Affects Ukrainians. A Sociological Survey. Kyiv: CEDOS Think Tank. Streltsova, Maria. 2016. “ICRC: About 600,000 Residents of Luhansk Region May Be Left without Water.” Ukrainian National News (UNN), November 21. https://www.unn.com.ua/ru/news/1620683-mkchkh-blizko-600-tisyach-zhiteliv-luganschi- ni-mozhut-zalishitis-bez-vodi. UNHCR (United Nations High Commissioner for Refugees). 2021. “Registration of Internal Displacement.” Data portal (figures as of March 5, 2021, from the Ministry of Social Policy of Ukraine), UNHCR, Geneva. http://bit.ly/IDPs_UA. UNIAN (UNIAN Information Agency). 2018. “Luhansk TPP Starts Using Gas due to Lack of Russian Anthracite.” UNIAN, November 7. https://www.unian.info/economics/10328685-ukraine-s-npps-receive-300-less-for-electricity-than-coal-fired-power- plants-media.html. World Bank. 2017. "Socio-Economic Impacts of Internal Displacement and Veteran Return. Summary Report." Working paper, Report No. 116489, World Bank, Washington, D.C.  World Bank and UN (United Nations). 2018. Pathways for Peace: Inclusive Approaches to Preventing Conflict. Washington, DC: World Bank. Xinhua (Xinhua News Agency). 2020. “Ukraine, France Discuss 100 mln Euro Investment in Railway Construction.” Xinhuanet. com, February 20. http://www.xinhuanet.com/english/2020-02/20/c_138799691.htm. Zapatrina, Irina. 2020. “Donetsk and Luhansk Regions’ Water Supply and Sanitation before and after the Conflict in the East.” Unpublished background paper, World Bank, Washington, DC. 114 Chapter 3 Current Conditions and Challenges T he economic consequences of the channels described in the previous chapter—that is, the 3 Ds (destruction, displacement, and disorganiza- tion)—have been profound and multidimensional in Donbas. Through these channels, the conflict has not only further hampered the region’s economic po- tential but also resulted in structural shifts, especially in foreign trade and labor markets. This chapter provides an overview of these outcomes, emphasizing the areas in which the current constraints are likely to challenge future economic recovery efforts. For the reasons also described in chapter 2 (such as the change in spatial cover- age of official data before and after the onset of the conflict), the analysis here is careful to refrain from always interpreting intertemporal differences in available statistics as “changes.” In addition, because not all changes are driven by the conflict—other factors being exogenous effects like commodity price dynamics and broader trends in Ukraine and the region—it is important to remember that this analysis does not aim to infer causation of impact. Nonetheless, the remain- der of this chapter will take stock of conflict-driven outcomes in three major categories: (a) consequences for economic activity, (b) trends in the region’s labor markets and demographic composition, and (c) conditions in access to publicly provided services. Economic Activity The overall economic activity in Donetsk and Luhansk Oblasts has drastically de- clined since the conflict began. The official statistics on regional gross domestic product (GDP) report a 61 percent decrease in Donetsk GDP and a 72 percent reduction in Luhansk GDP between 2013 and 2018. However, as discussed in chapter 2, a strict interpretation of these numbers would be misleading because they cover whole oblasts before 2014 and only the government-controlled areas (GCAs) thereafter. To gain a more complete understanding of the trends on the ground, we use the nighttime light emission data in two ways: •• Estimate economic activity levels in the GCAs and non-government con- Current Conditions and Challenges trolled areas (NGCAs) before the conflict by using 2013 nighttime light weights and aggregate (oblast-level) statistics. •• Estimate the economic activity levels (a) in the NGCAs by using 2019 night- time light weights, and (b) in the GCAs by using official GDP statistics. The results are presented in figure 3.1, where the red lines show the observed or official regional GDP data (corresponding to whole oblasts before 2014 and only the GCAs thereafter). The blue and yellow areas denote GDP in the GCAs and NGCAs, respectively. 117 Figure 3.1 Estimated Changes in Regional GDP Using Spatially Variable GDP Statistics and Nighttime Light Emissions, Donetsk and Luhansk Oblasts, 2010–18 a. Donetsk GDP 120 100 GDP, index (2013=100) 80 60 40 20 2013 night-light weights 2019 night-light weights 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 b. Luhansk GDP 120 100 GDP, index (2013=100) 80 60 40 20 2013 night-light weights 2019 night-light weights 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 GCA est. NGCA est. Observed data Sources: State Statistics Service of Ukraine and US National Oceanic and Atmospheric Administration (NOAA) databases; World Bank estimations. Note: The graphs compare economic activity in Donbas (Donetsk and Luhansk Oblasts) before and after the onset of the conflict in 2014. Red lines denote official, observed regional GDP data—corresponding to whole oblasts before 2014 and only to government-controlled areas (GCAs) thereafter. Blue and yellow areas denote GDP in the GCAs and non-government controlled area (NGCAs), respectively. Pre-2014 statistics are weighted by 2013 nighttime light emissions and the 2014–18 statistics by 2019 light emissions. The Economics of Winning Hearts and Minds In Donetsk Oblast, the GCA lost an estimated 29 percent of GDP between 2013 and 2018, and the NGCA an estimated 36 percent (figure 3.1, panel a). In com- parison, the GCA in Luhansk Oblast lost an estimated 12 percent of GDP and the NGCA an estimated 45 percent in the same time frame (figure 3.1, panel b). The Luhansk GCA performed slightly better than the Donetsk GCA, probably because the bulk of the coal mining in Luhansk (which had been in decline and loss-making even before the crisis) remained in the NGCA along with sever- al manufacturing facilities with limited markets. In addition, the Luhansk GCA’s economy depends relatively more on agriculture, which has done quite well in recent years. Donetsk Oblast, by contrast, has suffered a stronger indirect im- pact of the conflict (apart from the physical loss of production capacities) be- cause of the ruptured cooperation and value chain links with the companies in the NGCA and the significantly more complicated and burdensome logistics.1 118 Activity by Sector Although the overall economic activity in Donbas has been depressed unam- biguously, it is more difficult to pin down the sectoral composition of such a decline. Like other indicators, the official sector-level data cover whole oblasts before 2014 and only the GCAs thereafter. However, unlike the approach to as- sess aggregate economic activity, nighttime light emissions are not a reliable proxy for sector-level activity. Therefore, we cannot identify the exact trends for each sector. Nonetheless, it is useful to compare the whole-oblast trends before 2014 with the GCA trends after 2014. Industrial Sector The differences between the current level of economic activity in the GCAs and the preconflict levels in the whole oblasts are most apparent in industry, where output began decreasing even before the conflict (figure 3.2, panel a). Luhansk Oblast fared especially badly in this regard. Industrial output in Luhansk Oblast as a whole had already decreased by about 6 percent in the three years before the conflict. This trend continued in the GCA (the only part with available data) even after the initial shock of the conflict: from 2015 to 2019, the Luhansk GCA experienced a further 15 percentage point decrease in industrial output. By comparison, industrial performance in the Donetsk GCA was slightly better but grim nonetheless. Following the plunge immediately after the onset of the conflict, industrial output in the Donetsk GCA decreased by another 10 percent- age points between 2015 and 2019, amounting to only 35 percent of the entire oblast’s industrial production in 2010 (figure 3.2, panel a). In terms of industrial productivity (output per worker), the Donetsk GCA fared relatively well. Since 2015, its productivity increased by 26 percentage points (figure 3.2, panel b). As of 2019, industrial productivity there has remained about 15 percentage points above the 2013 preconflict level (whole oblast). The same positive trend did not hold true for the Luhansk GCA. Despite a small increase since 2015, the Luhansk GCA’s industrial productivity remains nearly 40 percentage points lower than the oblast’s as a whole in 2013. This decline prob- Current Conditions and Challenges ably reflects the loss of not only most of the oblast’s industrial capacity but also virtually all of its higher-value-added facilities, which are now within the NGCA. Construction Sector A similar picture could be observed in construction, where productivity in the Luhansk GCA remains at half the 2011 level, while the Donetsk GCA recovered most of the post-2011 losses by 2019 (figure 3.2, panel f). Similar trends are also observed in Ukraine as a whole (except that Ukrainian construction output and productivity recovered in recent years) but not in the Luhansk GCA (figure 3.2, panels e and f). 119 Figure 3.2 Differences in Sectoral Output and Productivity in Ukraine and Donetsk and Luhansk Oblasts before and after Onset of Donbas Conflict a. Industrial output, 2009–19 120 Whole oblast GCA only 100 Output index (2010=100) 80 60 40 20 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 b. Industrial productivity, 2009–19 140 Whole oblast GCA only Productivity index (2010=100) 120 100 80 60 40 20 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 c. Agricultural output, 2011–19 150 Whole oblast GCA only 140 Output index (2011=100) 130 120 110 100 90 80 The Economics of Winning Hearts and Minds 70 60 2011 2012 2013 2014 2015 2016 2017 2018 2019 d. Agricultural productivity, 2011–19 140 Whole oblast GCA only Productivity index (2011=100) 130 120 110 100 90 80 70 60 2011 2012 2013 2014 2015 2016 2017 2018 2019 Ukraine Donetsk Luhansk 120 e. Construction output, 2010–19 160 Whole oblast GCA only Output index (2010=100) 140 120 100 80 60 40 20 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 f. Construction productivity, 2010–19 200 Whole oblast GCA only Productivity index (2010=100) 180 160 140 120 100 80 60 40 20 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Ukraine Donetsk Luhansk Sources: State Statistics Service of Ukraine database; World Bank calculations. Note: The graphs track sectoral output and productivity in Ukraine and in Donbas (the Donetsk and Luhansk Oblasts) before and after the conflict began in 2014. Pre-2014 statistics include all of Donetsk and Luhansk Oblasts, whereas 2014–19 statis- tics exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. GCA = government-controlled area. Agriculture Sector Agriculture has done much better in both oblasts (figure 3.2, panels c and d), but it accounts for only a small share of the Donbas region’s output and em- ployment. In addition, this bright spot reflects a far superior performance in the large agrarian holdings in crop production (mostly grain, sunflower, and rape- seed) than in the individual farms. In fact, animal breeding remains loss-mak- Current Conditions and Challenges ing, while crop production has been highly profitable, especially in the Donetsk GCA, where it accounts for the bulk of agricultural output. This likely reflects a better-developed network of storage and processing facilities, as well as easier access to export markets. By contrast, agricultural producers in the Luhansk GCA, which tend to be smaller than in Donetsk, have lost access to storage and processing facilities, and the lack of railway connection to the rest of the country inhibits their access to both the domestic and the export markets. Still, economic composition in the Lu- hansk GCA has shifted away markedly from industry toward agriculture. 121 Trade Another major structural hit on Ukraine’s economy in general and on eastern Ukraine specifically was the large loss in export potential. As a result of the hos- tilities’ impact in the GCAs and the loss of control in the NGCAs, Ukraine’s export potential was slashed by about 22 percent, the bulk of it in Donetsk Oblast, which was Ukraine’s main industrial and export hub. In addition to the forgone exports, the loss of most of the country’s coal mining industry (located largely in the NCGAs of Donetsk and Luhansk Oblasts) led to major disruptions of coal supplies, which in turn necessitated massive coal imports. These developments have led to the emergence of a sizable structural current account deficit. In 2013, Ukraine’s export powerhouse in the east generated a trade surplus equal to US$25 billion, or 19 percent of GDP (roughly half of it from Donetsk Oblast alone). This was significant compared with a deficit of US$40 billion, or 29 per- cent of GDP, for the rest of the country (figure 3.3). By 2018, the eastern oblasts’ aggregate surplus shrank to 6 percent of GDP and that of Donetsk to 2 percent of GDP (not counting the NGCA). Even though the aggregate deficit of the rest of the country has fallen as well, to 15 percent of GDP, the overall deficit remained little changed, at 9 percent of GDP, despite the wrenching adjustment from 2014 to 2016—highlighting the sizable structural deficit that has emerged after the loss of the NGCAs in Donbas. Other factors have also contributed to the faster drop in exports in Donetsk and Luhansk Oblasts than elsewhere in Ukraine. These include, inter alia, (a) the loss of access to most of the coke and a large part of the energy coal deposits, which have resulted in stepped-up imports at significantly higher prices and occasional shortages; and (b) major infrastructure bottlenecks created by the hostilities, which made the shipping of exports more costly and more difficult. Key in these respects is the partial blockade of the Azov ports of Mariupol and Berdyansk following the launch of the Crimean Bridge by the Russian Federation and the stringent restrictions imposed on the naval passage beneath it. This, The Economics of Winning Hearts and Minds along with the need for exporters in the Luhansk and Donetsk GCAs, as well as in Dnipropetrovsk Oblast, to use a lengthy bypass to avoid the NGCAs (through which most main transportation arteries pass) has significantly increased the duration and costs of transportation. Another major impediment to the exporters in eastern Ukraine, and especially in the Luhansk GCA, was the collapse of trade with Russia. This mostly reflected not only Russia’s imposition of progressively tightening and broad-ranging trade re- strictions but also logistical bottlenecks given that all main transportation routes to Russia’s southeast remained in the NGCAs and not under Ukrainian customs control. The impact of these bottlenecks has intensified since early 2017, when the Ukrainian government banned all transportation links with the NGCAs. 122 Figure 3.3 Merchandise Trade Balance in Ukraine and Selected Regions, 2010–18 30 20 10 US$, billions 0 -10 -20 -30 -40 -50 2010 2011 2012 2013 2014 2015 2016 2017 2018 Donetsk Luhansk Other Easternª City of Kyiv Other Total Sources: State Statistics Service of Ukraine database; World Bank estimates. Note: Pre-2014 data for Donetsk and Luhansk Oblasts include entire oblasts, whereas 2014–18 data include only the govern- ment-controlled areas (GCAs). a. “Other Eastern” includes the oblasts of Kharkiv, Dnipropetrovsk, Zaporizhzhia, Poltava, Sumy, and Kherson. Enterprise Structure With conflict and subsequent loss of control in the NGCAs came a major change in the enterprise structure of both regions. Traditionally, the economies in both Donetsk and Luhansk were dominated by large industrial enterprises, mostly in mining, metallurgy, coke production, and chemical industries, many of them ver- tically integrated. However, as a result of the hostilities, most of the large indus- trial enterprises and mines remained in the NGCAs. As shown in figure 3.4, the number of large enterprises in Donetsk Oblast fell from 103 in 2013 (the whole oblast) to 27 in 2018 (GCA only). In Luhansk Oblast, the number fell from 28 (the whole oblast) to just 6 (GCA only). With the loss of most of their large enterprises, both oblasts lost the core of their industrial base, and with it most of their skilled employees, along with a Current Conditions and Challenges major portion of their tax base. There have also been major losses in the oblasts’ contributions to their own socioeconomic development, which traditionally in this part of the world is maintained and financed by the large enterprises. This has left the economy in a much weaker and vulnerable position than before the conflict. 123 Figure 3.4 Differences in Enterprise Size and Employment in Ukraine and Selected Regions, 2013 (Whole Oblasts) versus 2018 (GCAs only) a. Small enterprises 100 90 80 Enterprises, thousands 70 60 50 40 30 20 10 0 Dnipropetrovsk Donetsk Zaporizhzhia Luhansk Odesa Poltava Kharkiv City of Kyiv 2013 2018 b. Medium-size enterprises 4,0 3,5 Enterprises, thousands 3,0 2,5 2,0 1,5 1,0 0,5 0 Dnipropetrovsk Donetsk Zaporizhzhia Luhansk Lviv Odesa Poltava Kharkiv City of Kyiv 2013 2018 c. Large enterprises 180 160 140 120 Enterprises 100 80 60 40 20 The Economics of Winning Hearts and Minds 0 Dnipropetrovsk Donetsk Zaporizhzhia Luhansk Odesa Poltava Kharkiv City of Kyiv 2013 2018 d. Change in employment 2013–18, by enterprise size 40 20 0 Change (%) -20 -40 -60 -80 -100 Ukraine Dnipropetrovsk Donetsk Zaporizhzhia Luhansk Kharkiv City of Kyiv Large Medium Small Sources: State Statistics Service of Ukraine database; World Bank estimates. Note: Small enterprises are those with 1–49 employees; medium-size, 50–249 employees; and large, more than 250 employees. 124 Financial Markets Banking operations. The enterprises in the GCAs were also hit by deteriorating financial markets. The onset of hostilities sharply raised the risks for domes- tic banks, most of which have suspended operations in Donbas. By 2018, the combined GCAs had only 5 bank branches, down from 11 throughout Donbas in 2013 (compared with 395 nationwide). In practice, only the three state-owned banks—Oschadbank, PrivatBank, and Ukrgazbank—are maintaining a physical presence and providing banking services in Donbas. Deposits and lending. Although the parts of Donbas that are now GCAs were relatively underserved even before the conflict, the hostilities have led to a col- lapse in credit. In the Donetsk GCA, credit slumped from 34.6 percent of regional GDP in 2014 (slightly more than half the national average) to just 10 percent of GDP in 2018 (less than a third of the national average), as shown in figure 3.5, panel a. Deposits have fallen markedly as well but have held relatively better thanks to households, which showed more resilience than enterprises in the Donetsk GCA. This helped reduce the loan/deposit ratio of the region’s banks from 147 per- cent in 2013 to 75–80 percent since 2016. This reduction most likely reflects the lack of access to funding from abroad and increased risk aversion, forcing banks operating in the region to rely on their own funding. What is more concerning is that the decline in lending was particularly steep for corporations, slumping from 23 percent of regional GDP in 2014 to just 4.2 percent in 2018, underscoring the scale of the drop in activity and demand for credit (figure 3.5, panel b). Borrowing costs. Another constraint in the area is the very high cost of borrow- ing. Borrowers in the Donetsk GCA paid, on average, 60 percent more in interest than the national average in 2018—equivalent to a real interest rate of 15 percent (versus 5.6 percent on average nationwide), 2.5 times higher than in 2014. The spread on deposits was much smaller than the national rate (only 14 percent in 2018, little changed from 2014), and in real terms they were only slightly positive at 2.6 percent. These developments led to the emergence of a massive interest margin (as high as 14 percentage points in 2018, the highest since 2014) and ought to have Current Conditions and Challenges boosted banks’ profits. This large and widening spread, however, attests to banks’ perceptions of the heightened risks in Donbas. Borrowing costs also increased, partly because of regulatory costs due to the restrictions on economic activities imposed by the 2014 Presidential Decree, “On Temporary Measures for the Peri- od of the Anti-Terrorist Operation,”2 as well as diminished competition as many banks (including all foreign-owned) left the region. 125 Figure 3.5 Volume and Interest Rates of Lending and Deposit Operations in Donetsk GCA, 2014–18 a. All lending institutions 80 30 70 25 Share of regional GDP (%) 60 20 Interest rate (%) 50 40 15 30 10 20 5 10 0 0 2014 2015 2016 2017 2018 Loans National loans Deposits Deposit rate (rs) National lending rate (rs) Lending rate (rs) National deposit rate (rs) b. Nonfinancial companiesª 25 30 25 Share of regional GDP (%) 20 Interest rate (%) 20 15 15 10 10 5 5 0 0 2014 2015 2016 2017 2018 Loans Deposits Lending rate (rs) Deposit rate (rs) National lending rate (rs) National deposit rate (rs) The Economics of Winning Hearts and Minds Sources: State Statistics Service of Ukraine and National Bank of Ukraine databases; Donetsk Regional State Administration 2019. Note: GCA = government-controlled area; rs = right scale. a. Nonfinancial companies cover all sectors including wholesale and retail trade, manufacturing, and real estate, among others. Labor Markets and Demography Employment Trends Sectoral composition. The changes in the economic composition of Donetsk and Luhansk Oblasts were accompanied by similar changes in labor markets. In the Luhansk GCA, the 2019 share of workers engaged in industry (19.2 percent) was substantially lower than that of the whole oblast in 2013 (23.8 percent). By comparison, the shares of workers in services like public administration, trade, professional activities, and construction are higher (figure 3.6). 126 Figure 3.6 Differences between Donbas and Rest of Ukraine in Sectoral Structure of Employment, 2013 (Whole Oblasts) versus 2019 (GCAs Only) 2.6 Services 3.5 0.3 -1.1 Construction 0.8 -0.1 0.4 Industry -4.6 -0.3 -1.9 Agriculture 0.3 0.1 -6 -5 -4 -3 -2 -1 0 1 2 3 Percentage points Donetsk Luhansk Ukraine excluding Donbas Source: World Bank staff estimates based on State Statistics Service of Ukraine data. Note: Figure shows the percentage point change in employment of individuals aged 15–70 years, by region. Donbas comprises Donetsk and Luhansk Oblasts. In contrast, the Donetsk GCA exhibited lower concentrations of employment in agriculture and construction in 2019 than the whole oblast did in 2013—1.9 percentage points lower in agriculture and 1.1 points lower in construction (figure 3.6). The GCA showed higher concentrations in services and industry (by 2.6 and 0.4 percentage points, respectively), although the latter change was small. The differences between the oblasts in sectoral composition of employment reflect these oblasts’ occupational profiles. With lower industrial employment in the Luhansk GCA, the share of craftsmen and machine operators in all occupa- tions is also lower. At the same time, the shares of elementary occupations and service workers are higher.3 Formality versus informality. Such dynamics also affect the broad employment profiles of these oblasts. For instance, according to Labor Force Survey data, the Current Conditions and Challenges 2018 share of wage and salaried workers in the Luhansk GCA (71 percent) was lower than that of the whole oblast in 2013 (77 percent). In the Luhansk GCA, the incidence of informal employment was substantially higher in 2018 than before the conflict (29 percent in 2018 versus 22.4 percent in 2013). This likely reflects the relatively large share of agriculture and services employment in the Luhansk GCA, which often remains in the informal sector. The Donetsk GCA, by contrast, with its higher share of industrial employment and larger enterprises, had relatively lower informal employment, on the order of 13 percent in 2018, still a significant level. Overall, the high share of informal employment throughout the Donbas GCAs, while providing an informal “safety net” during the crisis, represents a major drag on local government revenues, for which the personal income tax is the key revenue source. 127 Table 3.1 Main Labor Market Indicators in Ukraine and Donetsk and Luhansk GCAs, 2018 Old-age Labor force Unemployment rate dependency ratio participation rate Value Rank Value Rank Value Rank Area (ratio) (out of 25) (%) (out of 25) (%) (out of 25) Donetsk 0.325 24 58.9 24 13.6 24 Luhansk 0.331 25 68.1 1 13.7 25 Ukraine 0.254 n.a. 63.4 n.a. 8.2 n.a. Source: State Statistics Service of Ukraine data. Note: The 25 ranked regions are all oblasts. Ukraine also includes the City of Kyiv (not ranked). Both regional and national data exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. GCAs = government-controlled areas; n.a. = not applicable. Unemployment and labor participation. Currently, the GCAs in Luhansk and Donetsk have the highest unemployment rates (13.6 percent and 13.7 percent, respectively) of all Ukrainian oblasts (averaging 8.2 percent). The Donetsk GCA’s labor force participation rate (58.9 percent) is the second lowest in Ukraine, but the Luhansk GCA’s is the highest (68.1 percent), as shown in table 3.1. Employment was hit in both oblasts in two waves: a first wave in 2014–15 (largely reflecting the loss of the NGCAs and averaging around 70 percent total em- ployment for both oblasts), and a second wave in 2017 after the government cut off trade and transportation with the NGCAs. The second wave hit industrial employment the most (declining between 2015 and 2017 by 19 percent in the Donetsk GCA and by 28 percent in the Luhansk GCA). Since the second wave, employment recovered somewhat through 2019—more so in the Luhansk GCA (increasing by 4 percent) than in the Donetsk GCA (1.8 percent)—led again by agriculture, construction, and trade as well as public ad- ministration. In contrast, industrial employment declined further, by 5 percent in the Donetsk GCA and 14 percent in the Luhansk GCA, reinforcing the deindus- The Economics of Winning Hearts and Minds trialization of the latter.⁴ Labor supply versus vacancies. These sectoral shifts are also evident in the occupational compositions of the excess labor demand-supply patterns in the Donbas GCAs (figure 3.7). Not surprisingly, there is a wider gap between the number of unemployed and the number of vacancies in the GCAs than in the rest of the country. In Ukraine as a whole, there are about 5.7 unemployed per- sons for every vacancy. In the Donetsk GCA, there are 16.3 unemployed persons per vacancy and in the Luhansk GCA, 13.1 unemployed per vacancy. Skilled workers are undersupplied everywhere, making up about a fifth of the vacancies and a tenth of the unemployed on average. This situation calls for upskilling of current workers and also for providing more skills-intensive training in vocational education. 128 Figure 3.7 Decomposition of Job Vacancies and Unemployment, by Occu- pation Type, in Ukraine and Donetsk and Luhansk GCAs, Decem- ber 2019 338,163 registered 59,018 10,803 registered 662 7,827 registered 599 unemployed people vacancies unemployed people vacancies unemployed people vacancies 100 14 12 11 11 11 15 13 14 16 18 75 21 34 Share of total (%) 12 20 20 9 23 2 1 1 5 14 50 1 9 15 14 23 14 7 4 3 5 11 4 11 4 11 25 10 4 11 11 10 9 7 17 12 12 6 14 16 6 6 9 8 0 Unemployed Vacancies Unemployed Vacancies Unemployed Vacancies Ukraine Donetsk Luhansk Without profession Skilled forestry, fisheries Specialists Operation or maintenance Trade and services workers Professionals of machines Skilled workers Technical officers Legislators, sr.civil servants, executives Source: World Bank estimates from State Employment Service of Ukraine data. Note: Regional and national data exclude the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. GCAs = government-controlled areas. Conversely, workers with more-basic skills (maintenance, operation and control, and assembly skills) are oversupplied across the board, but the gap is particu- larly wide in the Luhansk GCA (only 14 percent of the vacancies and 34 percent of the unemployed). In contrast, the same gap in the Donetsk GCA (13 percent of the vacancies and 18 percent of the unemployed) is more comparable to the Ukrainian average (16 percent of the vacancies and 21 percent of the unem- ployed). Demographic Trends Current Conditions and Challenges An Aging Population On the labor supply side, the conflict seems to have deepened the Donbas re- gion’s demographic aging problem. As of January 2020, the GCAs in Luhansk and Donetsk Oblasts had the highest median ages of population in Ukraine (46.5 and 45.4 years, respectively, compared with 41.4 years in Ukraine) and also the highest old-age dependency ratios (0.33 in Luhansk and 0.325 in Donetsk, com- pared with 0.254 in Ukraine). (See appendix C, table C.1, for these countrywide and oblast-specific data.) 129 Table 3.2 Intentions of Working-Age Household Heads of IDPs to Return to Place of Origin in Donbas, by Region of Current Residence, 2018 Share of respondents, percent Ukraine Intention to return Donetsk (GCA) Luhansk (GCA) Total (excl Donbas) Yes, in the near future 1.0 0.9 1.0 1.0 Yes, after end of conflict 33.2 31.4 18.1*** 25.9 Yes, maybe in the future 17.3 9.2*** 13.6** 14.1 No 29.3 25.8** 47.6*** 37.0 Difficult to answer, 19.1 32.7*** 19.8 22.0 or no response Source: World Bank calculations based on National Monitoring Survey (NMS) data of the International Organization of Migra- tion. Note: Joint dataset is of internally displaced persons (IDPs) from face-to-face and phone interviews (NMS rounds 9–12, weighted), based on their chosen answers to question 3.9: “Do you plan to return to the place of origin?” Information is pro- vided only for the heads of household, excluding IDPs from Crimea, individuals older than 70 years, and those who reported that they have already returned. Donbas comprises the Donetsk and Luhansk Oblasts. GCA = government-controlled area. Significance level of difference in means between (a) Luhansk, and (b) Ukraine (excluding Donbas): *** = 1 percent, ** = 5 percent, * = 10 percent. These facts suggest that population aging is a more serious problem in the GCAs now than it had been within Donbas as a whole before the conflict. It is driven by two possible factors: First, the NGCAs were more urbanized than the GCAs before the onset of the conflict, with younger populations. Thus, the subsequent exclusion of these more urbanized zones from the oblasts increased the average age in each GCA. Second, the demographic profiles of the GCAs themselves could have changed as a result of conflict-driven selective displacement of young and prime-age individuals to other parts of Ukraine or abroad, as further described in chapter 2. Displacement Dynamics The Economics of Winning Hearts and Minds The demographic underpinnings of these labor market problems are likely to persist, as is evident from the displacement dynamics. Synthesizing data from the 2018 National Monitoring System (NMS) survey by the International Organization for Migration (IOM)—which assessed the return intentions (table 3.2) and personal characteristics (table 3.3) of internally dis- placed persons (IDPs), among other factors—we observe two main types of IDPs: (a) younger, better-educated individuals who were more willing to work and in- tegrate into the local community, had moved elsewhere in Ukraine, and would not like to return; and (b) relatively older, less-educated individuals who moved from the NGCAs to the Donetsk and Luhansk GCAs (being relatively close), many of whom would like to return to their places of origin. Among the IDPs surveyed, the vast majority of respondents reported no inten- tion of returning in the near future, especially the IDPs who moved away from Donbas (table 3.2). Among those who reside in the Luhansk and Donetsk GCAs, 130 about a third indicated willingness to return to their original locations after the end of the conflict. The same answer was provided by only 18.1 percent of those residing elsewhere in Ukraine. Similarly, the IDPs elsewhere in Ukraine were twice as likely as those in the Donbas GCAs to say they will not return to Donbas. Therefore, the current place of residence is likely to be an important determi- nant of future return patterns. To better understand which other factors could influence the intention of IDPs to return, we compared the characteristics of those who reported that they would not like to return to the place of origin in Donbas (“No”) with those who claimed that they would likely return in the near future, after the end of conflict, or later in the future (“Yes”). The “No” respondents tend to be younger, male, and stay longer at the new place of residence than potential returnees (table 3.3). Most “No” respondents (almost 48 percent) lack close relatives in the NGCAs and therefore have no kinship ties to return to. Similarly, the “No” group are in a better financial situation (partly because most of them are employed), are less dissat- isfied with access to public services, and feel more locally integrated. The most problematic issue reported by both nonreturnees and potential re- turnees is the lack of housing. Similarly, both groups believe that the most im- portant condition for integration in the current community is housing (over 83 percent in each group), followed by regular income (over 66 percent), and only then employment (over 50 percent). Hence, as in previous studies of IDPs in Ukraine (for example, World Bank 2017), the major concern for all IDPs, whether they plan to return to their place of origin in Donbas or not, is housing. Employment, access to administrative and public services, local communi- ty support, and the possibility of voting in local elections appear to be much less influential than housing or regular income for integration of IDPs into a local community. However, these aspects are more important to nonreturnees than to returnees and should be improved to ease their after-displacement lives. Access to Public Services Current Conditions and Challenges The conflict has affected service delivery access through all of the 3 Ds: de- struction, displacement, and disorganization. The quantity and quality of service delivery has been jointly determined by the degree of physical damage to supply facilities (destruction); the scale and age composition of migration or displace- ment, affecting both the demand for services and the skills required for service provision (displacement); and other adverse effects of the conflict, including psychosocial trauma, insecurity, and less access to resources (disorganization). 131 Table 3.3 Sociodemographic Characteristics of Working-Age Household Heads of IDPs, by Intention to Return to Place of Origin in Donbas, 2018. Share of respondents, percent % % Variable Category or indicator answering answering Diff. No Yes Mean age Years 39.4% 44.9% *** Female 72.4% 76.4% *** Gender Male 27.6% 23.6% *** Mean duration of stay at Months 39.6% 38.3% *** destination Donetsk Oblast 27.6% 44.0% *** Current place of residence Luhansk Oblast 13.3% 19.3% *** Ukraine excluding Donbas 59.1% 36.7% *** Yes, I have returned there once or several times, Have you visited your place 37.1% 66.7% *** but I don’t live there permanently of residence in Donbas after first displacement? No 62.1% 32.7% *** Do you have close family Yes 49.5% 63.2% *** member residing in the temporary non-government controlled area? No 47.7% 34.1% *** Have to limit expenses even for food 12.3% 15.5% *** Enough funds only for food; lack for other needs 36.9% 40.8% *** Financial situation (self- Enough funds for food and basic needs but assessed) 43.2% 38.7% *** cannot save Enough funds for food, basic needs, and other needs; have savings necessary for expensive 6.1% 3.5% *** purchases Lack of own housing 29.9% 21.9% *** The Economics of Winning Hearts and Minds Impossible to return to the place of origin 3.4% 12.4% *** Payment for utilities 8.6% 11.5% *** The most problematic Living conditions 12.0% 9.6% *** issue (only aspects with significance difference shown here) Unemployment 4.3% 5.8% *** Suspension of social payments or pensions 1.1% 1.8% *** Financing own business 1.2% 0.5% *** Access to education 0.3% 0.1% ** Rented apartment 51.9% 51.2% Where does your household live now? Rented room in an apartment 4.0% 4.6% 132 Table 3.3 continued % % Variable Category or indicator answering answering Diff. No Yes Rented house 10.4% 11.0% Host family or relatives 10.2% 14.9% *** Own housing 10.3% 5.2% *** Where does your household Dormitory 6.8% 5.1% *** live now? Collective centers 3.5% 4.3% * Hotel 0.1% 0.2% Other 2.5% 3.2% Health care services 8.9% 11.3% *** Employment opportunities 12.5% 14.1% ** Share dissatisfied with Education (for children) 2.0% 2.2% access to public services Administrative services 4.0% 4.6% Social assistance services 6.1% 7.0% (receiving of pension, social assistance) Yes 55.7% 37.5% *** No 12.7% 21.0% *** Do you consider yourself locally integrated? Partially 29.8% 38.4% *** No response 1.8% 3.1% *** Employment 52.3% 50.7% Regular income 70.4% 66.4% *** Housing 84.5% 83.7% Which conditions are To have family and friends in the same important for you to be 42.6% 42.7% settlement Current Conditions and Challenges integrated in your current community? (multiple Support of local community 27.1% 23.6% *** responses) Easy access to administrative services 22.2% 16.9% *** Access to public services 32.7% 30.3% * (education, health care, and so on) Possibility of voting in local elections 16.4% 13.1% *** Source: Calculations from National Monitoring Survey (NMS) data of the International Organization of Migration. Note: Joint dataset is of internally displaced persons (IDPs) from face-to-face and phone interviews (rounds 9-12, weighted). Information is provided only for the heads of household, excluding IDPs from Crimea, individuals older than 70 years, and those who reported that they had already returned. Donbas comprises Donetsk and Luhansk Oblasts. Significance level of difference in means between nonreturnees and potential returnees: *** = 1 percent, ** = 5 percent, * = 10 percent. 133 Map 3.1 Levels of Satisfaction with Infrastructure in Donbas GCAs and NGCAs, 2017–19 2017 4.6 Regional 2018 4.8 average: 2019 5.6 4.0 4.3 Luhansk 5.1 5.2 Donetsk 5.1 Luhansk 3.5 6.0 Contact Line 4.7 Donetsk 4.6 Contact Line 5.3 0 5 10 Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). ©World Bank; further permission required for reuse. Note: Values are measured on a 0–10 scale, with 0 being no satisfaction and 10 being maximum satisfaction. “Infrastructure” refers to the quality of roads; the function of public transportation; and the quality of utilities such as waste management, water, and heating. Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine. The “contact line” separates each oblast’s government-controlled area (GCA) from its non-government controlled area (NGCA). Contact line areas were not surveyed in 2017. The Economics of Winning Hearts and Minds Infrastructure, Social, and Administrative Services The Social Cohesion and Reconciliation Index (SCORE) surveys by the Centre for Sustainable Peace and Democratic Development (SeeD) provide an opportunity to assess satisfaction with the current service provision. The survey questions determine the degree to which respondents can access these services as well as the frequency of such access. “Infrastructure” refers to the quality of roads; the functioning of public transportation; and the quality of utilities such as waste management, water, and heating for eastern Ukraine (map 3.1 and figure 3.8). The Donbas GCAs exhibit improving satisfaction among panel sample respon- dents, year on year. However, these levels of satisfaction remain at least a full point less than comparable national scores. In addition, infrastructure services score significantly higher in Donetsk than in Luhansk, possibly because of the 134 Figure 3.8 Satisfaction with Public Services in Donbas GCAs and NGCAs, 2017–19 8 7.0 7 6.6 6.7 6.7 6.3 6.3 6.1 6.0 Score (0–10) 6 5.8 5.9 5.9 5.7 5.7 5.4 5.5 5.1 5.2 5.1 5.0 5.0 5 4.7 4.2 4.3 4 3.2 3 Roads Justice Health Public Administrative Utilities Social Basic services care transport services services (welfare) schooling Donbas GCAs 2018 Donbas GCAs 2019 NGCAs 2019 Source: Social Cohesion and Reconciliation (SCORE) surveys, Centre for Sustainable Peace and Democratic Development (SeeD). Note: Values are measured on a 0–10 scale, with 0 being no satisfaction and 10 being maximum satisfaction. Donbas com- prises the Donetsk and Luhansk Oblasts of eastern Ukraine, each of which is divided into a government-controlled area (GCA) and a non-government controlled area (NGCA). NCGAs were not surveyed in 2018. particularly poor quality of the road network in Luhansk. Notable are the high average satisfaction scores in the NGCAs, relative to the GCAs, for the years in which data are available. However, the NGCA scores have been declining over time. Trends in satisfaction with social and administrative services across both ar- eas of control are mixed. “Social services” indicators measure satisfaction with access to education, health care, judicial, and social welfare services. “Adminis- trative services” refer to the provision of documentation, licenses, certificates, permits, and fee collection. Satisfaction with administrative and social services increased significantly overall from 2016 to 2019 in both the Luhansk and Do- netsk GCAs. The contact-line areas, however, registered significant declines. Generally, the increasing satisfaction overall resembles the trends of increased satisfaction with social and administrative services nationwide. Trend data are not available on satisfaction with administrative and social ser- Current Conditions and Challenges vices in the NGCAs, although 2019 data suggest that satisfaction there is higher than in the GCAs on nearly every indicator except administrative services (pos- sibly because the number of administrative service centers in the Donbas GCAs have increased in recent years). Higher satisfaction with access to justice in the NGCAs may be because of the recent restoration of civilian courts in those ar- eas. It is also important to note that these answers are filtered cognitively: that is, responses in both the GCAs and the NGCAs would reflect those respondents’ political views and security concerns. One reason for the improved satisfaction in 2019 with access to services in the Donbas GCAs may be the increase in development-oriented investments there starting in 2018. These included European Union (EU)-funded municipal improvements, World Bank engagements, and the contributions from the United 135 States Agency for International Development (USAID) and other donors for the completion of administrative service centers, road repair projects, and other in- frastructure improvements in Donetsk and nearby Zaporizhzhia Oblasts. Respondents in interviews and focus groups have cited such factors, which ap- pear to have been catalytic, with even small investments having positive im- pacts.5 However, a careful distinction should be made concerning where credit is given for such improvements. Persistently low trust in local and central institu- tions suggest that higher satisfaction with services is not necessarily increasing residents’ regard for authorities. Education A detailed account of the underlying factors driving such service access satis- faction can be observed in education, particularly in the enrollment and educa- tional quality trends. Enrollment and Attendance Trends By 2015, the conflict had damaged 36 preschools (35 in Donetsk and 1 in Lu- hansk). Around half of the preschools and their students remained in the NGCAs in both oblasts. Given accelerating migration of younger people and low fertil- ity rates, the share of individuals in the total population aged 0–14 years has been decreasing in Donbas since the conflict started, especially in the NGCAs (map 3.2, panel a). Overall, in the GCAs, the decline in preschool enrollment has been steeper than the decrease in the number of preschools, especially in the Donetsk GCA, where the preschool enrollment rate still has not returned to its preconflict level (map 3.2, panel b). There are several reasons behind declining enrollment. Access to preschool ed- ucation is challenging, especially on the contact line, because of damaged facili- ties, shelling risks, a shortage of kindergartens, parents’ fear of sending their chil- The Economics of Winning Hearts and Minds dren to the remaining facilities, and financial hurdles to making tuition payments. At the primary and secondary school levels, 57 schools had been damaged by 2015.6 Around half of the schools remained in the NGCAs in both Donetsk and Luhansk Oblasts (figure 3.9). In both oblasts, around 35 percent of all general secondary schools in the GCAs are in the rayons bordering the contact line. In contrast to the preschool trends, primary and secondary enrollment in the GCAs has been increasing since 2015, especially around the contact line (map 3.2, panel c). By June 2018, 17 state universities, 2 private universities, and 11 research insti- tutions relocated from the Donbas NGCAs to either Kyiv, the Donbas GCAs, or other cities. A Council of Rectors of Displaced Higher Education Institutions was established. However, most funding and equipment was left behind. Universities had to reregister at their new locations to receive the funding that had been allocated to their accounts in the NGCAs. Despite these relocations, more than half of institutions stayed within the NGCAs. 136 Map 3.2 School-Age Population and School Attendance across Donbas Rayons, 2013 versus 2018 a. Share of total population (%) aged 0–14 years a.1 Before conflict, 2013 a.2 After conflict, 2018 b. Number of students per preschool b.1 Before conflict, 2013 b.2 After conflict, 2018 Current Conditions and Challenges 137 c. Number of students per general secondary school c.1 Before conflict, 2013 c.2 After conflict, 2018ª Source: World Bank estimates using State Statistics Service of Ukraine data. ©World Bank; further permission required for reuse. Note: Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine, each of which is divided by a contact line (shown in red) into a government-controlled area (GCA) above (or outside) the line and a non-government controlled area (NGCA) below (or inside) the line. Rayons are second-level administrative divisions, below the oblast level. a. No breakdown of data by rayon was available for Luhansk Oblast in 2018. According to the Higher Education Portal of the Ministry of Internal Affairs,7 dis- placed universities continued operations in 2016 with around 40,000 students and 3,000 research and teaching staff. However, thousands of students and staff stayed in the NGCAs, reducing the size of the displaced universities. Educational Quality The Economics of Winning Hearts and Minds The conflict has also affected the quality of education in numerous ways. The Ukraine Education Cluster of the UN Office for the Coordination of Humanitarian Affairs conducted a comprehensive survey of schools in the Donbas GCAs in September–October 2018, covering around half of the schools and 65 percent of the students (UEC 2018). Especially around the contact line, the survey found, quality in education has suffered from reduced attendance (figure 3.10, panel a). As a result, an estimated 25 percent of children in Donbas require catch-up classes. The main reasons for reduced attendance were ill health, closed schools, and security concerns, especially in rural areas. Also, although the number of students per teacher in- creased in general secondary schools after the conflict, overcrowding was not considered as an important issue affecting educational quality. 138 Figure 3.9 Numbers of School Facilities and Students in Donetsk and Luhansk Oblasts, 2013–18 a.1 Donetsk preschools a.2 Luhansk preschools 1,500 1,500 1 ,1 58 Preschools (no.) 1,000 1,000 564 559 568 549 598 523 500 227 206 221 21 4 500 204 360 594 237 248 253 250 367 50 49 55 56 55 353 347 31 9 335 238 1 88 1 88 1 93 1 97 1 95 0 0 2013 2015 2016 2017 2018 2013 2015 2016 2017 2018 a.3 Donetsk preschool students a.4 Luhansk preschool students Students enrolled, thousands 150 1 39.7 150 100 77.3 100 54.3 55.7 52.6 52.2 58.1 50 25.8 21 .3 22.4 21 .0 20.8 50 62.4 42.3 1 5.0 1 5.5 1 5.8 1 4.9 36.6 33.0 33.3 31 .6 31 .4 15.8 2.2 1 .9 2.0 2.1 2.0 1 3.6 1 3.1 1 3.4 1 3.7 1 3.0 0 0 2013 2015 2016 2017 2018 2013 2015 2016 2017 2018 105 97 98 101 95 66 63 62 62 60 GCA GCA - Buffer zone NGCA Students per school b.1 Donetsk secondary schools b.2 Luhansk secondary schools 1,500 1,500 1 ,099 Schools (no.) 1,000 1,000 531 690 526 553 542 51 8 500 223 1 88 21 6 221 500 390 21 0 308 300 293 280 568 62 74 73 72 62 345 338 337 321 308 300 238 234 1 88 227 1 93 221 1 97 21 8 1 95 0 0 2013 2015 2016 2017 2018 2013 2015 2016 2017 2018 b.3 Donetsk secondary students b.4 Luhansk secondary studentsb Students enrolled, thousands 400 400 332.8 300 300 1 79.0 Current Conditions and Challenges 200 1 56.0 1 61 .7 200 1 67.9 1 51 .9 1 37.5 62.1 58.3 61 .7 64.4 55.6 1 1 7.9 100 100 51 .5 52.0 53.0 55.0 153.8 91 .7 81 .9 93.6 94.3 97.3 5.6 7.9 50 44.4 43.6 0 0 - - - 2013 2015 2016 2017 2018 2013 2015 2016 2017 2018 271 261 275 288 312 167 167 173 181 196 GCA GCA - Buffer zone NGCA Students per schoola Source: World Bank estimates from State Statistics Service of Ukraine data and the Statistical Yearbook of Ukraine. Note: Data for non-government controlled areas (NGCAs) are available only for 2013. The “GCA buffer zone” in Donetsk Oblast includes the following rayons (second-level administrative divisions): Avdiyivka, Toretsk (formerly Dzerzhinsk), Mariupol; the Bakhmut area; and the Volnovas’kyi, and Marins’kyi Yasynuvats’kyi districts. The “GCA buffer zone” in Luhansk Oblast includes the following rayons: the Novoaidars’kyi, Popasnyans’kyi, and Stanychno-Luhansk’kyi districts. a. Average numbers of students per school include those in both the GCA and the GCA buffer zone. b. Black columns for 2016–18 represent total number of students in the GCA and GCA buffer zone. No rayon-level breakdown was provided for those years. 139 Figure 3.10 Perceptions of Educational Quality and the Capacity-Building Needs of Teachers in Donetsk and Luhansk GCAs, 2018 a. Reasons for deteriorating quality of education a.1. Preschools a.2. General secondary schools Disruption of school 52 Disruption of school 56 7 18 due to conflict 15 due to conflict 8 52 72 Conflict-related trauma 21 Conflict-related trauma 40 and stress of children 20 and stress of children 25 Conflict-related trauma 40 Conflict-related trauma 72 16 33 and stress of teachers 18 and stress of teachers 19 Safety and security Safety and security concerns in and around 52 concerns in and around 44 10 15 school including 10 school including 6 UXOs and shelling UXOs and shelling 56 31 Reduced attendance 18 Reduced attendance 20 20 10 24 44 Lack of teachers 5 Lack of teachers 14 7 5 20 Lack of school books 3 Lack of school books 25 10 or learning materials 4 or learning materials 9 8 36 Reduced school hours 0 Reduced school hours 13 1 6 4 0 Overcrowding 3 Overcrowding 2 4 2 0 25 50 75 0 25 50 75 Share of schools choosing option (%) Share of schools choosing option (%) b. Teachers’ capacity-building needs b.1 Preschools b.2 General secondary schools 56 53 Psychosocial support 40% Psychosocial support 47 31 36 Conflict sensitive 28 Conflict sensitive 33 12 28 education education 18 9 The Economics of Winning Hearts and Minds 20 44 Life skills education 30 Life skills education 45 20 29 20 31 School safety 12 41 13 School safety 26 12 25 Peacebuilding 6 20 9 13 Peacebuilding 0 25 50 75 0 25 50 75 Share of schools expressing need (%) Share of schools expressing need (%) From contact line 0 - 5km From contact line 5 - 20km From contact line 20+km Source: World Bank analysis from UEC 2018. Note: An online questionnaire was disseminated to all education facilities in government-controlled areas (GCAs). Responses include 920 education facilities: 557 schools, 309 kindergartens, 25 vocational schools, and 29 after-school education facilities. Donbas comprises the Donetsk and Luhansk Oblasts of eastern Ukraine, each of which has been divided since the conflict began in 2014 into a GCA and a non-government controlled area (NGCA). UXOs = unexploded ordnance. 140 The most alarming safety concerns continue to be military presence, shelling, and unexploded ordnance (UXOs) at schools, especially those within 5 kilometers of the contact line. Although these safety concerns are shared at the preschool and general secondary school levels, lack of road safety is another overwhelmingly ex- pressed concern at general secondary schools. Several media reports have doc- umented military recruitment of children en route to or from school (GCPEA 2018). In response, the Ukrainian government endorsed the Safe Schools Declaration in November 2019 as a political commitment to prevent military use of schools and to better protect children, teachers, and schools during war (UNICEF 2019). In addition, the schools have major rehabilitation needs. The conflict has wors- ened an already aging infrastructure by further damaging windows, roofs, and facilities such as classrooms and gymnasiums. Such damages become detri- mental to education, especially during the winter. Therefore, fixing the windows and roofs is among the top-expressed needs across all schools. Also, half of the schools throughout the GCAs expressed need for water, sanitation, and hygiene (WASH) facilities. A need for gymnasiums came up among the biggest needs at the general secondary school level. These school rehabilitation needs exist in all regions throughout the GCAs, from within 5 kilometers of the contact line to 20 kilometers or more from the line. Finally, psychosocial factors have detrimental effects on the quality of educa- tion. Around 40 percent of students in the Donbas GCA and 13 percent in the Luhansk GCA study in rayons bordering the contact line. Children’s conflict-re- lated trauma and stress were expressed as among the factors in deteriorating educational quality by around half of the preschools and 72 percent of the gen- eral secondary schools within 5 kilometers of the contact line. An estimated 25 percent of the children in Donbas have required psychosocial support since the conflict began. Psychosocial problems have also affected teachers. On the needs of teachers, psychosocial support was mentioned by more than half the schools within 5 ki- lometers of the contact line, followed by the need to improve life skills education, especially at the general secondary school level (figure 3.10, panel b). To support the schoolchildren of Donbas and their families, a delegation to Ukraine from the EU, UN, and World Bank Group introduced a target for the education authorities to train 1,000 professionals from the education sector and beyond (school psy- chologists, teachers, and social and health workers). In addition, almost half of the teachers in Donbas needed capacity-building support. Current Conditions and Challenges Water Supply and Sanitation A critical service access area where the conflict has aggravated the preexisting conditions is water supply and sanitation (WSS). Before the onset of the con- flict, Ukraine had 755 settlements without WSS services. Donetsk and Luhansk Oblasts housed 30.5 percent and 17.7 percent of them, respectively, according to State Statistics Service of Ukraine (SSSU) data. About 89.3 percent of the urban settlements and 34.5 percent of the rural settlements in Donetsk Oblast were supplied with water services (figure 3.11, panel a). In Luhansk Oblast, about 80 percent of urban settlements and 30 percent of rural settlements were sup- plied (figure 3.11, panel b). 141 Figure 3.11 Water Supply Service Coverage of Settlements in Donetsk and Luhansk Oblasts, by Urban and Rural Location a. Donetsk, 2012–20ª 120 Share of settlements covered (%) 100 80 60 40 20 0 2012 2013 2014 2015 2016 2017 2018 2019 2020 b. Luhansk, 2014–18 90 Share of settlements covered (%) 80 70 60 50 40 30 20 10 0 2014 2015 2016 2017 2018 Urban Rural Source: State Statistics Service of Ukraine data. Note: Data after 2014 are for government-controlled areas (GCAs) only. Since the onset of the conflict, the conditions in rural areas of Luhansk Oblast The Economics of Winning Hearts and Minds have deteriorated significantly (figure 3.11, panel b). By 2018, only 7 percent of rural settlements were provided with water supply. Unlike the urban settlements, where a postconflict drop in supply was remedied after 2016, the rural supply problem persists. The quality of water, historically poor in the Donbas region, has been worsened by the conflict. With insufficient environmental regulation, industrial and mining activities polluted the region’s water resources for many years. However, cur- rently, there is a high risk that water sources will further deteriorate. Closure and conservation (flooding) of mines in these oblasts can lead to the removal of salts from mine waters. (Donbas has about 300 mine water reser- voirs and 1,185 landfills—more than 300 of which are permanently burning, which is a significant source of underground water pollution.) Rising mine water levels can poison the drinking water and the environment. In addition, many hazardous toxic substances of both industrial and municipal origin leak into the water, in- cluding direct wastewater discharge without necessary sanitary requirements. 142 In 2018, 41.5 percent of water samples taken from water sources in the Donetsk GCA did not meet the drinking water standards (24.4 percent of samples from underground sources and 63.5 percent from decentralized ones). Similarly, 20.6 percent of samples from centralized water supply systems also failed to meet the standards. In the Luhansk GCA, the situation was much worse: 85 percent of water samples failed the drinking water standards test (69.7 percent of samples from underground sources and 87.6 percent from decentralized sources), and 41.9 percent of water samples from centralized water supply systems failed it (WASH Cluster Ukraine 2019). Sanitation conditions are alarming in both Donetsk and Lugansk Oblasts, es- pecially in rural settlements. In 2018, 51.4 percent of urban settlements in the Donetsk GCA and 50 percent in the Luhansk GCA had access to sanitation ser- vices. In contrast, only 11.2 percent of rural settlements in the Donetsk GCA and less than 1 percent of those in the Luhansk GCA had access to sanitation ser- vices. In the Donetsk GCA, almost all collected wastewater was reported to go through wastewater treatment plants (WWTPs) in 2018, and approximately 77 percent of the wastewater received biological treatment before being discharged to riv- ers. In the Luhansk GCA, 62 percent of the collected wastewater went through WWTPs, only 35 percent of which received biological treatment. Overall, numerous challenges remain to WSS provision in Donbas. Access to WSS services remains poor, especially in rural areas, and the quality of drinking water used by residents of the Donetsk and Luhansk GCAs is well below standards. This trend is likely to further deteriorate with ongoing pollution of surface and groundwater sources by mines (dead or alive) and industrial facilities. Finally, the WSS utilities operating in the GCAs are financially strained because they effectively serve both GCAs and NGCAs but only get paid in the GCAs. In light of these challenges, there is an urgent need for a comprehensive WSS strat- egy that can help coordinate the future policy actions in a consistent framework. This need also goes beyond the WSS issue and extends to all public services affected by the disorganization brought about by the conflict. We will return to this point in the last chapter of this report. Current Conditions and Challenges Synopsis The analysis in this chapter focused on what the direct effects of the conflict (destruction, displacement, and disorganization) have meant for the economic and social outcomes in the GCAs of Donetsk and Luhansk Oblasts. The main objective was to elaborate on these consequences and identify obstacles to fu- ture development in the Donbas region. We analyzed these issues in three major groups: overall economic activity, labor markets and demography, and access to public services. Several key observations emerged. 143 Avoiding data misinterpretations. Official regional statistics report whole oblasts before 2014 and only the GCAs thereafter. Thus, differences over time reflect both changes in accounting and the impact of the conflict, which cannot be separated without further information. Estimating aggregate economic activity. Estimates from nighttime light emis- sions suggest that, between 2013 and 2018, GDP decreased by 29 percent in the Donetsk GCA, 12 percent in the Luhansk GCA, 36 percent in the Donetsk NGCA, and 45 percent in the Luhansk NGCA. Note that these are just point compar- isons and not impact assessments. (The latter would require a counterfactual GDP estimate.) Understanding the sectoral composition of the new reality. Industrial output has continued to shrink in both the Luhansk and Donetsk GCAs since 2015. In the Donetsk GCA, this decline was accompanied by increasing productivity, but in the Luhansk GCA productivity decreased too. Agriculture provided the only bright spot, where both output and productivity increased in both GCAs. This sector, however, still comprises a small share of output and employment in these oblasts. Identifying constraints to growth. The Donbas region’s growth challenges re- flect a combination of its structural challenges and conflict-driven factors. Among others, three factors come out strongly: •• Disorganization and risk. The overall uncertainty and escalated risk associated with the conflict suppresses economic activity directly. In addition, supply irregularities and higher costs associated with key inputs (such as anthra- cite) and connectivity bottlenecks, combined with unfavorable external competition, further reduce profitability. •• Access to finance. Enterprises in the GCAs were hit by deteriorating financial markets. With sharply rising risks, most banks have suspended operations in the region. Credit slumped from 34.6 percent of regional GDP in 2014 to just 10 percent of GDP in 2018 (less than a third of the national average). Those who can borrow paid the equivalent of a 15 percent real interest rate (versus The Economics of Winning Hearts and Minds 5.6 percent on average nationwide). •• Skill mismatch. A comparison of vacancies and unemployment profiles reveals an oversupply of basic machine operators and maintenance workers in the GCAs (especially in the Luhansk GCA) and an undersupply of skilled workers in both GCAs. Assessing labor market conditions. Currently, the Luhansk and Donetsk GCAs have the worst labor market conditions in Ukraine. Employment in the GCAs was hit in two waves: a first wave in 2014–15, largely reflecting the impact of the initial phase of the conflict, and a second one in 2017 following the cessation of trade and transportation with the NGCAs. Facing demographic challenges. The conflict has deepened the demographic aging problem in the Donbas region, which adds to the labor market problems and further constrains growth. The GCAs in Donbas currently have the highest median age and old-age dependency in Ukraine. The disproportionate displace- 144 ment of younger generations toward other places in Ukraine has only deepened the aging problem. Assisting displaced people. The vast majority of IDPs report no intention of returning to Donbas in the near future, especially if they are young, male, live out- side Donbas, and have been displaced for a long time. This situation will further constrain economic growth opportunities in Donbas. IDPs face major challenges in finding jobs (when unable to transfer skills) and housing. They need more as- sistance in reskilling and job placement. Addressing public service access. Conflict has affected both the quantity and quality of public services. Safety remains an important challenge (regarding mil- itary presence, shelling, and UXOs near the contact line); infrastructure needs major rehabilitation (for example, fixing school windows and roofs); and both the providers and receivers of public services need psychosocial support. Water and sanitation conditions remain problematic in rural areas (especially in Luhansk), and the conflict has accelerated the water supply pollution problem drastically (through mine closures and flooding, which contaminate water sources). Even the central water distribution system often does not meet safety standards. Calling for a comprehensive strategy. The Donbas region faces multifaceted challenges. Thus, one-dimensional (for example, infrastructure investment only) and one-size-fits-all approaches (for example, wish lists that consider neither nuances nor trade-offs between alternative actions) are not likely to be effec- tive. The complexity of the problem calls for a concerted effort that coordinates interventions around a comprehensive medium-term strategy. The next chapter considers policy options for economic development in the Donbas region in light of the challenges described here. As discussed above, any discussion about the region’s economic future must consider the conflict-driv- en factors. However, major uncertainties cast a veil over the future evolution of such factors. Thus, a contingent approach—for example, different policies for different conditions—will be needed. Current Conditions and Challenges 145 Notes 1. Note that these numbers show only the changes between 2013 and 2018, and thus they are not synonymous with the “impact” of the conflict. To calculate such an impact estimate, we would need to know the GDP patterns in the absence of the conflict (that is, a counterfactual). 2. Presidential Decree No. 1669-VII as of 09/02/2014 was issued, among other measures, to fulfill the assumption of interna- tional obligations concerning combating of terrorism. 3. Elementary occupations include unskilled laborers (Category 9), including laborers in agriculture, mining, and other sec- tors, according to the International Standard Classification of Occupations (ISCO). 4. For further social and labor market indicators for the GCAs in Donetsk and Luhansk Oblasts in 2019, see appendix C, table C.1, which covers the same indicators discussed in chapter 2 for 2013. Therefore, a comparison is possible between the preconflict profiles in Donbas as a whole and the postconflict profiles of the Donbas GCAs. However, one should resist the temptation to see any differences as “changes” for the reasons discussed in the beginning of this chapter. 5. World Bank key informant and focus group data, Luhansk Oblast, 2019–20. 6. According to the Ministry of Education and Science of Ukraine (MoES), 200–300 primary and secondary schools were damaged by 2020.  7. The Educational Portal is available on the Ministry of Internal Affairs of Ukraine (MIA) website: https://osvita.mvs.gov.ua/ en/educational-portal. Note, however, that access to the functional part of the portal is available only to registered users, who may include MIA employees, higher education institutions with specific training conditions of the MIA, or applicants for education in these institutions. References Donetsk Regional State Administration. 2019. “Development Strategy of Donetsk Region for the Period Up to 2027.” Krama- torsk, Ukraine. https://dn.gov.ua/ua/projects/strategiya-rozvitku-doneckoyi-oblasti-na-period-do-2027-roku. GCPEA (Global Coalition to Protect Education from Attack). 2018. “Education under Attack 2018 – Ukraine.” Report, GCPEA, New York. UEC (Ukraine Education Cluster). 2018. “Education Cluster Needs Survey in Government Controlled Areas of Donetska and Luhanska Oblasts, September 2018: Overview of the Methodology and Key Findings.” UEC survey report, Global Education Cluster, Humanitarian Response, United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA), New York and Geneva. UNICEF (United Nations Children’s Fund). 2019. “UNICEF Welcomes Ukraine’s Move to Endorse the Safe Schools Declaration.” Press release, November 26. https://www.unicef.org/ukraine/en/press-releases/unicef-welcomes-ukraines-move-endorse- safe-schools-declaration. WASH Cluster Ukraine. 2019. “WASH Cluster Study of Humanitarian Needs in Eastern Ukraine 2019.” United Nations Children’s Fund (UNICEF), WASH Cluster Ukraine, Kyiv. The Economics of Winning Hearts and Minds World Bank. 2017. “Conflict in Ukraine: Socio-Economic Impacts of Internal Displacement and Veteran Return; Summary Report.” Working Paper No. 116489, World Bank, Washington, D.C. 146 Chapter 4 Future Development: 864 Scenarios T o analyze the future growth path of Donbas, the alternative approaches in- clude case studies, statistical (regression) analysis, and quantitative analy- sis based on economic modeling. The case study approach often relies on expert opinion driven by specific experiences. Case studies sometimes provide a useful road map, especially when other approaches are not feasible. However, because the descriptive nature of this approach does not lead to quantitative metrics, evaluating different recommendations and prioritizing them becomes daunting. A statistical approach provides a more transparent way to use the past relationships between economic factors to explain what may happen. This can help to rank different policy options in a verifiable manner and control for country-specific factors. However, in certain cases (for example, onset of a ma- jor conflict), historical patterns may fail to capture future dynamics after struc- tural breaks in the data. This chapter follows a quantitative analysis based on economic modeling for several reasons. Unlike case studies, this approach enables the validity of its findings to be challenged by further evidence, logical or empirical. We aim to de- limit expectations regarding future trends by explicitly studying the relationship between policy actions, conflict conditions, and the incentive-driven behavioral response of economic agents. This is done by observing economic relationships that have taken place so far. When such past relationships fail to shed light on future actions, which limits purely statistical approaches, we use an economic model to produce policy conjectures—an approach still grounded in what has happened so far through the calibration of the model but one that can also sim- ulate what might happen under different conditions through new shocks. Finally, given major uncertainties about the future dynamics, we adopt a scenario ap- proach to compare alternative future trajectories. An Integrated Approach to Future Development Future Development: 864 Scenarios This section analyzes the future economic trends in Donbas, and in Ukraine, in light of an evolving environment and alternative policies. To assess the implica- tions of new policies while incorporating the behavioral responses of economic agents to these policies, the analysis employs a simulation model based on eco- nomic theory and calibrated by using economic data, following the approaches of World Bank (2017) and World Bank (2020). Figure 4.1 illustrates the model’s main operating principles, box 4.1 summarizes the model’s characteristics, and appendix D contains the formal definitions. In a nutshell, the model features forward-looking, risk-neutral, and rational eco- nomic agents who respond to economic incentives (for example, wage differ- entials) by changing sector or location. This endogenous treatment of mobility patterns helps to provide a more realistic analysis of economic shocks (negative 149 or positive), including the possibility of analyzing the temporal and spatial prop- agation of such shocks as well as the conflict-driven internal displacement of people. Objectives of the Approach Our first objective is to assess the economic implications of distinct policy ac- tions, individually. We consider three major groups of policy actions: •• Investments: Investments that boost labor productivity •• Transfers: Residence-based transfers that supplement income •• Mobility: Policies that reduce interregional mobility costs. The policy assessment incorporates different scales of the policy (for example, depending on financial resources) and targeting preferences (such as geograph- ic implementation). The analysis can estimate the results of these policy actions in terms of regional gross domestic product (GDP), wage, and population trends in all oblasts—treating the government-controlled areas (GCAs) and non-gov- ernment controlled areas (NGCAs) of Donetsk and Luhansk Oblasts separately. Because such trends also rely deeply on future conflict trends, we also evaluate these policies under different conflict trajectories. Theoretically, these policies themselves could affect the conflict trajectory; however, we abstain from such a speculative exercise. The second objective is to analyze the trade-offs between alternative policies. The decision-making problem faced by Ukrainian policy makers and the interna- tional community is in essence a resource allocation problem. Resources need to be raised and allocated across different policy tools (investments, transfers, and mobility reforms in this case) and across different locations (Donbas GCAs, The Economics of Winning Hearts and Minds NGCAs, and elsewhere in Ukraine in this case) to satisfy an objective criterion (an outcome). Even when all policies are desirable (akin to a “wish list”), budget constraints would enforce a specific allocation. It is difficult to precisely identify an optimal allocation under such complex con- ditions. However, the model’s structure enables us to elaborate on the trade-offs between alternative policies and to provide conjectures regarding the tendency toward such an optimum. 150 Figure 4.1 Simulation Model to Assess Implications of New Policies in Eastern Ukraine Future Development: 864 Scenarios Source: ©World Bank. Further permission required for reuse. 151 Box 4.1 Characteristics of the Simulation Model The simulation model is based on that of Artuç, Gómez Parra, and Onder (2019) with a number of extensions, including the adoption of the solution method in Artuç, Bastos, and Lee (2021) but with a fixed number of job opportunities. The Structure of the Model The model enables us to consider the economic implications of various shocks (natural changes, policy shifts, and conflict) by taking economic agents’ preferences into consideration. This is done by accounting for the country’s endogenous migration patterns, which respond to changes in economic opportunities and nonmonetary factors like amenities and con- flict. In particular, we account for several factors, described below. Spatial structure. The model comprises 108 industry-region combinations in every time period, corresponding to four main industries (agriculture and natural resources, manufacturing, services, and public sector) in 27 regions (all oblasts and special regions in Ukraine, with the GCAs and NGCAs of Donetsk and Luhansk Oblasts treated as different regions). Economic agents. Agents are risk-neutral. They decide to live in a region and work in an industry by considering their current and expected future utilities (a function of wage). For simplicity, these decisions are undertaken with perfect foresight—that is, expectations about future conditions affect all decisions now, except for the unexpected conflict shock. Because the model features both aggregate shocks (for example, conflict and policy scenarios) and individual-specific shocks (independently and identically distributed), interregional migration (and churning in the labor market) occurs even in the absence of big changes. We calibrate mobility The Economics of Winning Hearts and Minds costs using preconflict labor allocations, interregional migration data, and historical job-switching patterns. Conflict. We model the conflict as a shock to utility through a decline in productivity and an increase in moving cost between GCAs and NGCAs in Donbas (Donetsk and Luhansk Oblasts). We calibrate the decrease in pro- ductivity by matching the number of internally displaced persons (IDPs) in the simulation and the data. Accordingly, conflict decreases the produc- tivity in the Donetsk and Luhansk GCAs by 25 percent and 13 percent, re- spectively, and it doubles the mobility costs between GCAs and NGCAs. By comparison, it decreases productivity in the Donetsk and Luhansk NGCAs by 23 percent and 47 percent, respectively. 152 Because there are no data for GCAs and NGCAs separately before 2014, and because the official data for Donetsk and Luhansk Oblasts after 2014 cover only the GCAs, we used changes in nighttime light emissions to dis- tinguish GCAs and NGCAs before the conflict and to infer NGCA trends after the onset of the conflict. A Simple Numerical Example It is useful to consider a simple numerical example to build intuition about the main principles of the model. Suppose we have two identical regions (A and B), each endowed with 10 units of capital (K), labor (L), and total factor productivity (A). Consider two periods (1 and 2), where the second period can be of variable length (that is, the future). Technology is defined by a Cobb-Douglas production function (that is, , where denotes output, and is the income share of capital), and workers are paid their marginal productivities (equal to 7 initially). Assume capital is fixed, there are no savings, and workers are risk-neutral (with a simple utility equivalent to the wage) for the purposes of this illustrative exercise. At the end of the first period, a conflict in Region A destroys half of the capital stock and reduces productivity by half (figure B4.1.1, panel a). Thus, the marginal productivity of workers decreases substantially. If there were no out-migration, the equilibrium wage would decrease to 2.8 in Region A because there would be too many workers and too little capital (and too low a productivity level). With wages still at 10 in the other region, workers have motives to move from the conflict region to the nonconflict region. However, as more workers move from Region A to Region B, wages will in- crease in Region A and decrease in Region B. So, how many workers will move to the other region? The answer depends on migration, or mobility, costs. Figure B4.1.1, panel b, shows the relationship between mobility costs and the number of workers who migrate from Region A to Region B and the subsequent equilibrium wages in Region A. In this example, nine workers would move if there were no mobility costs. For the chosen parameters in this example, a mobility cost around 3.3 induces three workers to move from A to B, which brings Future Development: 864 Scenarios wages in Region A to 3.1 and wages in Region B to 6.4. In actual simulations, workers have an infinite horizon, they consider all 25 locations in each period, and mobility costs are estimated by using gross migration patterns and wage differentials between each pair of those 25 locations while also considering other factors (like different amenities that may make certain locations more attractive beyond wages). 153 Figure B4.1.1 A Simple Numerical Example of Mobility between Two Regions after Conflict a. A two-region, two-period scenario PERIOD 1 PERIOD 2 Region A Region A TFP (A) TFP (A) Capital (K) Capital (K) Labor (L) Labor (L) Wage (w)= 7 Wage (w)= 3.1 (without migration = 2.8) Region B Region B TFP (A) TFP (A) Capital (K) Capital (K) Labor (L) Labor (L) Wage (w)= 7 Wage (w)= 6.4 (without migration = 7) Conflict b. Relationship between mobility costs, wages, and migration 4.5 6.0 4.0 5.5 3.5 5.0 3.0 Mobility cost 4.5 2.5 Wages 4.0 2.0 3.5 1.5 1.0 3.0 0.5 2.5 0 2.0 0 2.0 4.0 6.0 8.0 10.0 The Economics of Winning Hearts and Minds Migration A -> B Migration Wages in A (rhs) Source: World Bank calculations. ©World Bank. Further permission required for reuse. Note: TFP = total factor productivity. Scenarios for Analysis To elaborate on these points, the analysis features seven groups of scenarios, with two to four options in each group (table 4.2). The scenarios can be imple- mented in any combination. Conflict scenarios. The presence of conflict affects both migration costs (with- in Donbas and between Donbas and other regions) and labor productivity in 154 Donbas (both GCAs and NGCAs). We consider three possible cases: •• Status quo: Both mobility costs and labor productivities in Donbas remain at their current values indefinitely. •• Intermediate case: Both mobility costs and labor productivities in Donbas recover to undo the conflict-driven shock by half (that is, they take the mid- point between the current values and the preconflict values). •• Reintegration case: Both mobility costs and labor productivities in Donbas fully recover from the conflict-driven shock (that is, they take the preconflict values). Migration cost scenarios. These include factors that affect interregional mi- gration within Ukraine (excluding the cost within Donbas because the conflict scenarios already determine these in Donbas). Broadly, migration costs reflect several factors like transportation, the housing market, and the business envi- ronment (for example, how easy it is to liquidate assets). We consider two op- tions: •• Status quo: Migration costs (specific to regions) remain at their current val- ues. •• Lower migration costs: With improvements in underlying factors, interregion- al migration costs are reduced by half between all regions (except between Donbas and the rest of Ukraine). Jobs scenarios. To differentiate between private and public drivers of economic recovery, we explicitly consider public sector employment in two cases: •• Restricted public sector: In all regions, public employment is fixed at its cur- rent levels. •• Unrestricted public sector: Public sector employment can grow and shrink in different regions based on migration patterns. Investment scale scenarios. Investments increase labor productivity in the re- gion where they are undertaken. We consider three scenarios, reflecting possible Future Development: 864 Scenarios finances that can be mobilized in the future (without any reference to where they will be invested): •• Status quo: Investments remain fixed at their current levels. •• Moderate increase: The investment envelope increases by an amount equiv- alent to 15 percent of preconflict investments in Donbas (Donetsk and Lu- hansk combined). •• Large increase: The investment envelope increases by an amount equivalent to 30 percent of preconflict investments in Donbas (Donetsk and Luhansk combined). 155 Figure 4.2 Conflict and Policy Scenarios 1. CONFLICT SCENARIOS • Status quo • Intermediate • Reintegration 2. TRANSACTION COSTS • Mobility costs: Baseline or low • Jobs: Restricted or unrestricted public employment 3. INVESTMENTS • Scale: status quo, moderate, or large • Target: GCAs only or GCAs + NGCAs 4. TRANSFERS • Scale: Status quo, moderate, or large • Target: GCAs only or GCAs + NGCAs 864 alternative paths Source: ©World Bank. Further permission required for reuse. Note: In the conflict scenarios, the “status quo” assumes continuation of the Donbas conflict; “intermediate,” sufficient recov- ery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “reintegration,” a full postconflict recovery. GCAs = government-controlled areas; NGCAs = non-government controlled areas. The Economics of Winning Hearts and Minds Investment target scenarios. Once the investment envelope is identified, the next problem is to consider the spatial allocation of this envelope. To elaborate on efficiency trade-offs across spatial alternatives, we consider four different options: •• Kyiv without agglomeration: The identified financial envelope is invested in Kyiv City, and these investments are not propagated by other factors (for example, a better business environment or market access conditions). •• Kyiv with agglomeration: The identified financial envelope is invested in Kyiv City, and these investments are reinforced by agglomeration effects. There- fore, the return on investment (ROI) is larger in Kyiv than in other regions (and its current level). •• Donbas GCAs only: Additional investments are undertaken only in the Don- bas GCAs. 156 •• Donbas GCAs and NGCAs: Additional investments are undertaken in both GCAs and NGCAs.1 Transfer scale scenarios. To attract workers, a wage subsidy (paid to workers directly, in addition to their salaries) is paid indefinitely. The total magnitude of this subsidy is calibrated to match the total investment scales in the early years of the transfers. The populations in the regions with transfers increase as a re- sult of incentive-driven migration; thus, the total transfer bill also changes. We consider three cases: •• Status quo: No additional transfers are provided. •• Moderate transfers: The transfer per worker is set at 25 percent of the av- erage wage in preconflict Donbas. (In the short term, the total transfer bill in this case is roughly equivalent to the “moderate increase” in the investment scale scenario.) •• Large increase: The transfer per worker is set at 50 percent of the average wage in preconflict Donbas. (In the short term, the total transfer bill in this case is roughly equivalent to the “large increase” in the investment scale scenario.) Transfer target scenarios. All transfers are considered only in the Donbas re- gion. However, we consider two spatial options: •• GCAs only: Only the workers in GCAs receive transfers (including those who migrate to these regions in the future). •• GCAs and NGCAs: All workers in GCAs and NGCAs receive transfers (includ- ing those who migrate to these regions in the future). Together, different combinations of these scenarios amount to 864 different fu- ture paths. The following sections first present an overview of the results, de- scribing the range of outcomes under this “scenario space.” Next, we analyze specific policy scenarios individually and discuss various outcomes under dif- ferent conditions. Finally, we compare the effectiveness of different policies and analyze the trade-offs between them. Future Development: 864 Scenarios The Impact of Policies To see the 864 scenarios together provides a sense of the overall range of out- comes. Figure 4.3 shows the GDP impact of all scenarios in Donbas (GCAs and NGCAs) and Kyiv City. The kernel density estimates (figure 4.3, panels a, c, and e) show the density of outcomes by different scenarios (conflict, investment scale, and transfers scale) with no geographic separation. The box plots (figure 4.3, panels b, d, and f) show the outcomes by specific region (Donetsk GCA, Donetsk NGCA, Kyiv City, Lu- 157 hansk GCA, and Luhansk NGCA), where the boxes show the range between 25th and 75th percentiles, and the lines inside the boxes show the median values. In this structure, if a scenario significantly affects outcomes, we should see strong differences between different density curves and box plots. Conversely, ineffective scenarios should not change the distributions significantly. GDP gains, by conflict scenario. Other things being equal, improvements in conflict conditions clearly shift the density of GDP gains in all regions rightward (figure 4.3, panel a). That is, as the conflict subsides, GDP gains increase mo- mentously, for small and large changes alike. This is particularly clear for Donetsk (both GCA and NGCA) and the Luhansk NGCA (figure 4.3, panel b). The effect has the same direction, but is relatively muted, in the Luhansk GCA. Kyiv City is largely unaffected, but there is a small negative effect on economic activity. GDP gains, by investment scale scenario. Compared with the conflict scenar- ios above, the impact of the investment scale scenarios on regional GDPs re- mains muted on average. The Donetsk GCA exhibits the largest variation in its median and Q1–Q3 range over different investment scale scenarios, followed by the Luhansk GCA (figure 4.3, panel d). These are likely to be driven by the fact that the broader scenario space includes diverse actions (like concentrating the investments in GCAs or Kyiv City) with strong implications for regional GDPs. For the same reason, Kyiv City is also relatively more responsive to the investment scale scenarios. GDP gains, by transfer scale scenario. The transfer scale scenarios have a more pronounced impact on GDP than the investment scale scenarios. The ef- fect comes from two factors: First, there is a decreasing density of GDP effects on the low end of the spectrum (figure 4.3, panel e), where, unlike the conflict scenarios, some negative results prevail. Second, there is a tail extension on the high end, which is common among all three scenarios displayed in figure 4.3. In the transfer scenarios by region, the GCAs in Donetsk and Luhansk are the most responsive (figure 4.3, panel f). These comparisons already provide the bold lines of the simulation results. The Economics of Winning Hearts and Minds However, to narrow down the exact impact of each policy action in various cir- cumstances, and to explain why such an impact takes place, we will need more structure in the analysis. The following subsections analyze the marginal effects of conflict, investments, transfers, and mobility scenarios in more detail. The Gravity of Peace Conflict scenarios affect the labor productivity in Donbas (both GCAs and NGCAs) as well as the mobility costs between GCAs and NGCAs. In the status quo case, the current conditions prevail in the medium term. In contrast, the re- integration scenario fully removes the productivity and mobility shocks brought about by the conflict, and the intermediate scenario removes half of them. 158 Figure 4.3 Projected Regional Medium-Term GDP Gains, by Scenario Com- bination, in Donbas and Kyiv City a. Conflict scenarios, all regions b. Conflict scenarios, by region 4 200 3 150 Change in GDP (%) 100 Density 2 50 1 0 0 0 0.5 1.0 1.5 2.0 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 Change in GDP (%) Donetsk- Donetsk- Kyiv City Luhansk- Luhansk- GCA NGCA GCA NGCA Status quo Intermediate Reintegration c. Investment scale scenarios, all regions d. Investment scale scenarios, by region 200 2.5 2.0 150 Change in GDP (%) 1.5 100 Density 1.0 50 0.5 0 0 0 0.5 1.0 1.5 2.0 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 Change in GDP (%) Donetsk- Donetsk- Kyiv City Luhansk- Luhansk- GCA NGCA GCA NGCA Status quo Moderate Large e. Transfer scale scenarios, all regions f. Transfer scale scenarios, by region 200 3 150 Change in GDP (%) 2 100 Future Development: 864 Scenarios Density 50 1 0 0 0 0.5 1.0 1.5 2.0 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 Change in GDP (%) Donetsk- Donetsk- Kyiv City Luhansk- Luhansk- Status quo Moderate Large GCA NGCA GCA NGCA Source: World Bank estimates. Note: Kernel density estimates (panels a, c, and e) show GDP outcomes including Donetsk GCA, Donetsk NGCA, Kyiv City, Luhansk GCA, and Luhansk NGCA. In the scenarios by region (panel b), the conflict scenarios are as follows: 1 = Status quo; 2 = Intermediate; 3 = Reintegration. Similarly, the regional investment scale and transfer scale scenarios (panels d and f) are 1 = Status quo; 2 = Moderate; 3 = Large. Results include all combinations from nonfeatured scenarios (840 total cases). Boxes (panels b, d, and f) show the range between the 25th and 75th percentiles, with the horizontal lines designating median val- ues; the vertical lines, the margin of error; and the dots, outliers. Medium-term = 10 years; GCA = government-controlled area; NGCA = non-government controlled area. 159 At the outset, we expect wages in both GCAs and NGCAs to increase with the increase in productivity. This should, in return, increase migration from the rest of Ukraine to the Donbas region, which puts a limit on wage growth. (The post- conflict productivity boost is balanced with higher numbers of workers, which, ceteris paribus, reduces labor productivity.) Overall, wages, population, and re- gional GDP in Donbas should all increase. The effects for the rest of Ukraine are a mirror image: as labor moves toward Donbas, real wages should increase in the rest of Ukraine along with decreases in population and regional GDP. However, because these effects are dissipated across Ukraine, they should be relatively small. In net terms, a better conflict scenario should have a net positive effect on everyone’s well-being in Ukraine. The reduction in mobility costs, on the other hand, can facilitate a faster spatial convergence by removing the cost barriers that help to sustain wage differen- tials across regions. Figure 4.4 shows the simulation results, displaying the medium-term (10-year) implications of conflict scenarios for regional GDP (panel a), population (panel b), and wages (panel c) in Donbas (GCAs and NGCAs) and Kyiv City. The values are normalized to fix their 2013 levels at 1. Status quo results. Under the conflict status quo, the Donbas region will contin- ue to lose population as the ongoing hostilities depress economic activity. This depopulation is more prominent in the NGCAs (by about 16 percentage points in the Luhansk NGCA and 13 percentage points in the Donetsk NGCA) than in the GCAs (about 7 percentage points in the Luhansk GCA and 8 percentage points in the Donetsk GCA). As the population continues to shrink, the marginal prod- uct of labor increases, and real wages increase by 6–10 percentage points in the NGCAs and by 2–4 percentage points in the GCAs. Overall, regional GDP will contract in the medium term by an additional 6 per- centage points in the Luhansk NGCA, by 5 percentage points in the Donetsk NGCA, and by 3 percentage points in both the Luhansk and Donetsk GCAs. The impact on Kyiv City is as expected: a marginal increase in population and region- al GDP and a marginal decrease in wages. However, these changes are all less The Economics of Winning Hearts and Minds than half a percentage point. Reintegration results. Under reintegration, where productivity and mobility costs revert to their preconflict levels, economic activity rebounds in three out of four regions in Donbas. In the Donetsk GCA, the depopulation trend halts and, with higher productivity, regional GDP increases by about 24 percentage points. The NGCAs in Donetsk and Luhansk Oblasts gain population (by 9 and 19 per- centage points, respectively) along with a sizable recovery in GDP (by 34 and 48 percentage points, respectively). These improvements, however, are not suffi- cient to bring the regional GDPs to their preconflict levels within the 10-year time frame covered here. 160 Figure 4.4 Simulated Medium-Term Implications of Conflict Scenarios, by Selected Region a. Gross regional product 1.1 1.0 0.9 0.8 Index (2013=1) 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 -0.1 o te n o te n o te n o te n o te n qu io qu io qu io qu io qu io ia ia ia ia ia at at at at at ed ed ed ed ed us us us us us r r r r r eg eg eg eg eg m m m m m at at at at at nt nt nt nt nt r r r r r St St St St St te te te te te i i i i i Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA b. Population 1.2 1.0 0.8 Index (2013=1) 0.6 0.4 0.2 0 -0.2 -0.4 o te n o te n o te n o te n o te n qu io qu io qu io qu io qu io ia ia ia ia ia at at at at at ed ed ed ed ed s s s s s r r r r r u u u u u eg eg eg eg eg m m m m m at at at at at nt nt nt nt nt r r r r r St St St St St te te te te te i i i i i Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA c. Real wages 1.6 1.4 1.2 Index (2013=1) 1.0 0.8 0.6 Future Development: 864 Scenarios 0.4 0.2 0 -0.2 o te n o te n o te n o te n o te n qu qu qu io qu io qu io io io ia ia ia ia ia at at at at at ed ed ed ed ed s us s us us gr r gr r r u u eg eg eg m m m m m at at at at at te te nt nt nt r r r r r St St St St St te te te te te in in i i i Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA Current Medium-term Change Source: World Bank estimates. Note: Results are evaluated for baseline values for all other scenarios (current levels of migration costs, investments, transfers, and public sector employment). “Medium-term” = 10 years. The “Status quo” conflict scenario assumes continuation of the Donbas conflict; “Intermediate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of con- flict-driven shock; and “Reintegration,” full recovery, returning to preconflict values. GCA = government-controlled area; NGCA = non-government controlled area. 161 The Luhansk GCA exhibits a more distinct pattern, one resembling other parts of Ukraine but more intense. With conflict, the population in the Luhansk GCA increased disproportionately, which led to lower wages, but the population gain was sustained thanks to mobility barriers with the NGCAs. When these barriers are removed and economic activity in the NGCAs recovers, this trend is reversed. In net terms, the projected decrease in population (about 25 percentage points) roughly offsets the productivity gains in the medium term, and the regional GDP in Luhansk Oblast remains broadly stable in this scenario. It should be noted, however, that real wages increase everywhere in Donbas with reintegration; thus the welfare effect of a subsiding conflict is positive for both GCAs and NGCAs. Intermediate-case results. Finally, under the intermediate scenario, we observe similar effects in Donbas but at a more muted level than under the reintegration scenario. The only exception to this is regional GDP in the Luhansk GCA, where the increase in the intermediate case is slightly higher than in the reintegration scenario. This is driven by interaction effects between productivity and mobility cost trends. Overall, the three economic indicators observed in our simulations (wages, pop- ulation, and regional GDP) are sensitive to conflict scenarios. Effects are largely as anticipated, except in the Luhansk GCA, where a conflict-driven economic concentration is likely to dissipate once the arbitrary frictions (productivity ef- fects and mobility costs) disappear. As in other regions, this would be a wel- fare-improving outcome for Ukrainians. The Power and Limits of Investments Investments increase labor productivity where they are implemented. With that, real wages increase in the short term, which attracts workers from other regions at the margin. Migration continues until the wage differentials between regions become small enough. As a result, the population and GDP in the region where investments are undertaken are expected to increase permanently. The simula- The Economics of Winning Hearts and Minds tion results confirm these expectations (table 4.1). Investment Impacts, by Scale and Conflict Scenario For a given conflict scenario, investments in a specific region (Kyiv, GCAs, or both GCAs and NGCAs) lead to an increase in wages, population, and GDP in that re- gion. For example, under the status quo conflict scenario, a moderate increase in investments only in the GCAs leads to an 8.5 percentage point increase in wages in the Donetsk GCA instead of 4 percentage points in the baseline (in the absence of additional investments). A large investment would raise wages by 14.1 percentage points. When the investment budget is split between GCAs and NGCAs, these effects are still there but more modestly (a 10.8 percentage point wage increase from a large investment in the Donetsk GCA and NGCA, for example). 162 Investments in the GCAs also lead to reductions in population and GDP else- where, including in the NGCAs. This is because some workers migrate to the GCAs, where investments have boosted productivity and wages. However, for the GCAs, this additional boost in GDP, driven by such a migration from the NGCAs to the GCAs, is relatively modest. For example, in the Donetsk GCA, it amounts to a 0.7 percentage point increase in the case of a moderate increase in invest- ments (invested only in the GCAs) under the status quo conflict scenario. Do investments boost economic activity in Donbas more effectively in peace or in conflict? This may seem like a trivial issue at first sight, but it is not. As conflict subsides, labor productivity in both GCAs and NGCAs increase—the total factor productivity (TFP) effect—which in turn slows down out-migration. In parallel, mobility costs also decrease in peacetime, allocating labor across regions more effectively to eliminate wage gaps. Whereas the former effect boosts economic activity in Donbas, the latter can diminish it by increasing migration (if the wage differential is not in favor of Donbas). The net impact depends on the relative sizes of these effects. Simulations show that, in general, investments are most effective in the full rein- tegration scenario. This is because both factors described above—the TFP effect and the mobility cost effect—work in the same direction (or the former effect dominates). For instance, in the status quo conflict scenario, a large investment in only the GCAs would boost the regional GDP by 27.1 percentage points in the Donetsk GCA. By comparison, it boosts the same regional GDP by 34.5 percent- age points in the reintegration scenario, both relative to the baseline (no policy) cases. The Luhansk GCA is an exception, where the postconflict arrivals boosted local activity to a relatively large extent, and this effect disappears more rapidly with lower migration costs as conflict dissipates. Similarly, the intermediate conflict scenario has less of an impact than reintegra- tion on regional GDP. Overall, the effectiveness of investments in boosting GDP is greater in the status quo scenario than the intermediate scenario as well. This is likely because the mobility cost effects dominate the TFP effect when these costs are low. Spatial Trade-Offs: Investments in Donbas versus Kyiv Future Development: 864 Scenarios Another important point is the distinction between outcomes in Donbas and in Ukraine as a whole in the investment target scenarios. In other words, what is the potential trade-off between investing in Donbas and investing elsewhere? For the GCAs, the best outcomes are observed in the GCA-only investment scenarios. For instance, in the status quo conflict scenario, the largest invest- ment-driven GDP rise in the Donetsk GCA (23.6 percentage points) occurs when a large investment targets the GCAs only. This is also true for the NGCAs, which see the greatest GDP increases when investments target both GCAs and NGCAs. 163 Table 4.1 Simulated Medium-Term Changes in Donbas and Ukraine, by Conflict, Investment Scale, and Allocation Scenario Percentage point change in index value (2013=100) a. Regional GDP Investment scenarios SCALE -> Moderate increase Large increase Baseline Area GCA GCA COMPOSITION -> Kyiv-1 Kyiv-2 GCA Kyiv-1 Kyiv-2 GCA + NGCA + NGCA Status quo -3.5 -3.6 -3.7 8.3 1.8 -3.6 -4.0 23.6 8.2 Donetsk Intermediate 11.5 11.4 11.4 22.2 15.9 11.4 11.3 37.5 21.7 GCA Reintegration 24.2 24.1 23.9 38.5 30.5 24.0 23.8 58.7 39.1 Status quo -5.1 -5.2 -5.3 -5.8 -0.5 -5.2 -5.5 -7.1 4.9 Donetsk Intermediate 12.5 12.5 12.4 11.2 16.3 12.4 12.4 9.2 21.3 NGCA Reintegration 34.4 34.3 34.2 33.4 40.8 34.3 34.1 32.3 49.5 Status quo 0.4 4.1 10.0 0.2 0.1 6.1 18.1 -0.1 -0.1 Kyiv City Intermediate -0.3 3.3 8.9 -0.4 -0.4 5.1 16.6 -0.4 -0.4 Reintegration -0.9 2.7 8.3 -1.2 -1.2 4.6 16.1 -1.5 -1.6 Status quo -2.7 -2.9 -3.3 4.1 0.1 -3.0 -3.8 10.9 2.9 Luhansk Intermediate 2.6 2.5 2.3 8.2 4.7 2.5 2.0 12.9 6.2 GCA Reintegration -1.8 -2.0 -2.1 4.7 0.7 -2.0 -2.3 11.4 3.2 Status quo -5.8 -5.8 -5.9 -6.0 -3.9 -5.9 -6.1 -6.4 -2.0 Luhansk Intermediate 17.2 17.2 17.1 16.7 18.9 17.1 17.0 15.9 20.4 NGCA The Economics of Winning Hearts and Minds Reintegration 49.1 49.0 48.8 48.4 51.8 48.9 48.7 47.7 54.6 Status quo 0.2 0.52 1.07 0.57 0.80 0.70 1.84 1.09 1.64 Whole Intermediate 1.9 2.28 2.81 2.41 2.60 2.45 3.54 2.97 3.46 Ukraine Reintegration 3.8 4.14 4.67 4.32 4.56 4.31 5.40 4.97 5.55 164 Table 4.1 continued b. Population Investment scenarios SCALE -> Moderate increase Large increase Baseline Area GCA GCA COMPOSITION -> Kyiv-1 Kyiv-2 GCA Kyiv-1 Kyiv-2 GCA + NGCA + NGCA Status quo -8.3 -8.5 -8.9 1.1 -4.7 -8.6 -9.5 12.3 -1.2 Donetsk Intermediate -1.4 -1.5 -1.6 4.1 -0.4 -1.6 -1.9 13.3 1.3 GCA Reintegration -0.4 -0.5 -0.8 10.5 3.7 -0.6 -1.0 26.0 9.7 Status quo -12.3 -12.5 -12.8 -13.8 -9.4 -12.6 -13.2 -16.8 -6.8 Donetsk Intermediate -4.1 -4.1 -4.2 -6.6 -3.5 -4.1 -4.3 -10.2 -2.6 NGCA Reintegration 8.8 8.7 8.4 7.0 13.0 8.6 8.2 5.2 19.1 Status quo 0.8 3.8 8.4 0.3 0.3 5.3 15.0 -0.1 -0.1 Kyiv City Intermediate -0.6 2.1 6.2 -0.8 -0.7 3.5 12.2 -0.9 -0.8 Reintegration -1.9 0.9 5.2 -2.3 -2.4 2.3 11.0 -3.0 -3.1 Status quo -6.9 -7.4 -8.2 -1.4 -5.5 -7.6 -9.6 3.6 -4.4 Luhansk Intermediate -11.3 -11.6 -12.0 -9.1 -11.6 -11.7 -12.8 -8.8 -13.2 GCA Reintegration -33.7 -33.9 -34.2 -29.3 -32.8 -34.0 -34.5 -24.9 -31.8 Status quo -15.8 -16.0 -16.3 -16.5 -14.5 -16.0 -16.7 -17.4 -13.3 Luhansk Intermediate -2.6 -2.7 -2.8 -3.6 -2.3 -2.7 -2.9 -5.0 -2.4 NGCA Reintegration 19.1 19.0 18.7 17.9 20.5 18.9 18.5 16.6 21.9 Future Development: 864 Scenarios 165 Table 4.1 continued c. Wages Investment scenarios SCALE -> Moderate increase Large increase Baseline Area GCA GCA COMPOSITION -> Kyiv-1 Kyiv-2 GCA Kyiv-1 Kyiv-2 GCA + NGCA + NGCA Status quo 4.0 4.1 4.3 8.5 6.8 4.1 4.6 14.1 10.8 Donetsk Intermediate 15.0 15.0 15.1 21.1 19.0 15.0 15.2 27.2 23.7 GCA Reintegration 28.6 28.7 28.9 31.4 30.8 28.8 29.0 34.1 32.9 Status quo 6.7 6.8 7.0 7.7 10.1 6.9 7.3 9.6 14.7 Donetsk Intermediate 22.0 22.0 22.1 23.7 26.5 22.0 22.1 26.2 32.1 NGCA Reintegration 33.2 33.3 33.4 34.3 35.3 33.4 33.6 35.5 37.4 Status quo -0.4 0.4 1.5 -0.1 -0.1 0.8 2.8 0.1 0.1 Kyiv City Intermediate 0.3 1.2 2.5 0.4 0.4 1.6 4.1 0.4 0.4 Reintegration 0.9 1.7 3.0 1.1 1.1 2.2 4.5 1.4 1.5 Status quo 2.4 2.6 2.9 4.4 3.9 2.7 3.4 6.6 5.5 Luhansk Intermediate 10.7 10.8 11.0 13.9 12.9 10.8 11.3 17.9 15.7 GCA Reintegration 28.8 28.9 29.1 30.8 30.6 29.0 29.3 32.8 32.3 Status quo 9.8 10.0 10.2 10.4 11.8 10.0 10.6 11.1 13.8 Luhansk Intermediate 32.2 32.3 32.4 33.1 34.7 32.3 32.5 34.2 37.5 NGCA Reintegration 45.7 45.8 46.0 46.5 47.2 45.9 46.1 47.4 48.7 Status quo -1.0 -0.7 -0.2 -0.3 -0.3 -0.5 0.4 0.5 0.5 The Economics of Winning Hearts and Minds Whole Intermediate 1.9 2.2 2.7 2.5 2.4 2.4 3.3 3.3 3.1 Ukraine Reintegration 4.7 5.0 5.5 5.5 5.4 5.2 6.1 6.5 6.2 Source: World Bank estimates. Note: Kyiv (1) denotes the investments in Kyiv with no agglomeration effects, and Kyiv (2) denotes the investments in Kyiv with agglomeration effects. Results are evaluated for baseline values for all other scenarios (current levels of migration costs, transfers, and public sector employment). “Medium-term” = 10 years. The “Status quo” conflict scenario assumes continua- tion of the Donbas conflict; “Intermediate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegration,” full recovery, returning to preconflict values. GCA = government-controlled area; NGCA = non-government controlled area. However, the same is not true for Ukraine as a whole. In the status quo conflict case, the largest increase in Ukrainian GDP occurs when the investment is in Kyiv with agglomeration effects (1.8 percentage points in the large investment scenario). This is because agglomeration effects in Kyiv (possibly supported by better economic prospects in the European markets) boost productivity, and also because it is easier to move to Kyiv than to Donbas. 166 The spatial trade-off is clear under the status quo, but it weakens as the conflict dissipates. That is, as the conflict continues, investing in Donbas improves out- comes in Donbas itself, but there is an efficiency cost (forgone growth opportu- nities elsewhere) of doing so. However, because peace boosts productivity in Donbas, investing there under the reintegration scenario is also efficient from a national GDP perspective. For instance, a large investment targeting both GCAs and NGCAs in the reintegration scenario boosts Ukrainian GDP by 5.6 percentage points (the highest increase, also the best option for the NGCAs, obviously). Thus, although the policy prob- lem is more straightforward in peace, it is not so in conflict. We will discuss the elements of decision making under such trade-offs (and under uncertainty in future conflict scenarios) at the end of this chapter. The Indirect Role of Transfers Transfers—such as direct cash payments, social assistance, hazard pay, and other income supplements—do not affect labor productivity directly. However, where they are implemented, they provide incentives for workers not to migrate away or even attract workers from other regions. As a result, the total number of workers increases more in a location with transfers than in a place without them. The enlarged workforce leads to a reduction in marginal productivity and wages while increasing population and GDP in that location. In addition, as workers are attracted from other regions, this leads to a mir- ror-image effect in those places of origin: higher wages, lower population, and lower GDP. Therefore, the interregional migration performs as a mechanism that propagates the income effect of transfers to other regions. Our simulations con- firm these expectations (table 4.2). Effectiveness of transfers. The first observation from the simulation results concerns the effectiveness of transfers: where transfers are applied in Donbas, they slow the out-migration and, in certain cases, can even reverse the migra- tion trends. In the status quo conflict scenario, a moderate transfer scheme (25 percent of the average wage in preconflict Donbas) targeting both GCAs Future Development: 864 Scenarios and NGCAs increases the population by 3.1 percentage points in the Donetsk GCA (as opposed to an 8.3 point decrease without transfers), and it reduces the out-migration from the Donetsk NGCA from a baseline (without transfers) of 12.3 percentage points to 5.2 percentage points. These movements increase GDP in the Donetsk GCA by 1.2 percentage points instead of the baseline decrease of 3.5 percentage points. They also slow down the GDP decrease in the Donetsk NGCA, bringing a 2.2 percentage point de- crease instead of a 5.1 percentage point decrease. Results are similar in the Lu- hansk GCA and NGCA. 167 Table 4.2 Simulated Medium-Term Changes in Donbas and Ukraine, by Conflict, Transfer Scale, and Target Scenario Percentage point change in index value (2013=100) a. Gross domestic product TRANSFER SCALE -> Moderate increase Large increase Baseline Area TARGET -> GCA GCA + NGCA GCA GCA + NGCA Status quo -3.5 6.7 1.2 19.9 14.5 Do- netsk Intermediate 11.5 20.9 18.5 45.7 42.0 GCA Reintegration 24.2 44.2 41.2 78.7 69.3 Status quo -5.1 -8.2 -2.2 -14.7 -0.9 Do- netsk Intermediate 12.5 8.0 14.8 3.8 25.6 NGCA Reintegration 34.4 31.6 51.6 28.4 79.7 Status quo 0.4 -0.3 -0.4 -1.0 -1.8 Kyiv Intermediate -0.3 -0.6 -0.9 -2.3 -3.6 City Reintegration -0.9 -2.0 -3.5 -4.2 -7.6 Status quo -2.7 11.7 3.1 25.9 13.9 Lu- hansk Intermediate 2.6 11.1 6.1 31.0 24.1 GCA Reintegration -1.8 15.2 10.7 42.2 31.4 Status quo -5.8 -8.2 -3.3 -12.9 -2.5 Lu- hansk Intermediate 17.2 14.1 18.8 10.2 24.6 NGCA Reintegration 49.1 46.0 63.6 42.6 86.7 The Economics of Winning Hearts and Minds Status quo 0.2 0.4 0.0 0.5 0.0 Whole Intermediate 1.9 2.2 2.0 2.4 1.9 Ukraine Reintegration 3.8 4.3 3.8 4.7 3.2 168 Table 4.2 continued b. Population TRANSFER SCALE -> Moderate increase Large increase Baseline Area TARGET -> GCA GCA + NGCA GCA GCA + NGCA Status quo -8.3 17.9 3.1 56.6 39.4 Donetsk Intermediate -1.4 19.4 13.9 82.6 71.7 GCA Reintegration -0.4 38.8 32.4 121.3 96.7 Status quo -12.3 -19.2 -5.2 -32.4 -1.3 Donetsk Intermediate -4.1 -12.4 0.9 -19.7 24.9 NGCA Reintegration 8.8 4.0 41.6 -1.1 104.9 Status quo 0.8 -0.6 -0.9 -2.0 -3.4 Kyiv City Intermediate -0.6 -1.2 -1.9 -4.4 -6.9 Reintegration -1.9 -3.9 -6.7 -8.0 -14.2 Status quo -6.9 31.7 7.7 70.2 33.9 Luhansk Intermediate -11.3 7.8 -4.2 55.9 37.2 GCA Reintegration -33.7 -0.8 -10.3 60.3 34.0 Status quo -15.8 -21.7 -9.4 -32.2 -7.0 Luhansk Intermediate -2.6 -8.4 0.8 -15.3 13.8 NGCA Reintegration 19.1 13.8 45.8 8.3 95.1 Future Development: 864 Scenarios 169 Table 4.2 continued c. Wages TRANSFER SCALE -> Moderate increase Large increase Baseline Area TARGET -> GCA GCA + NGCA GCA GCA + NGCA Status quo 4.0 -7.2 -1.4 -17.5 -13.3 Donetsk Intermediate 15.0 5.3 7.6 -11.7 -9.4 GCA Reintegration 28.6 10.8 13.1 -8.3 -3.9 Status quo 6.7 11.4 2.5 22.9 0.1 Donetsk Intermediate 22.0 27.9 18.6 33.9 6.8 NGCA Reintegration 33.2 36.3 17.4 39.7 0.3 Status quo -0.4 0.3 0.4 0.9 1.6 Kyiv City Intermediate 0.3 0.6 0.9 2.0 3.3 Reintegration 0.9 1.8 3.2 3.8 7.1 Status quo 2.4 -8.8 -2.3 -15.5 -8.3 Luhansk Intermediate 10.7 3.9 8.2 -7.5 -3.3 GCA Reintegration 28.8 13.4 17.2 -2.9 3.0 Status quo 9.8 15.0 5.3 27.4 3.7 Luhansk Intermediate 32.2 37.1 29.5 44.0 21.0 NGCA Reintegration 45.7 49.5 31.8 53.7 16.1 Status quo -1.0 -1.4 -0.3 -3.8 -0.5 Whole Intermediate 1.9 1.7 2.4 -1.2 0.9 Ukraine The Economics of Winning Hearts and Minds Reintegration 4.7 4.9 4.8 -0.6 -4.4 Source: World Bank estimates. Note: Results are evaluated for baseline values for all other scenarios (current levels of migration costs, investments, and public sector employment). The “moderate increase” refers to a transfer per worker of 25 percent of the average preconflict wage in Donbas, and the “large increase” to 50 percent of the average preconflict wage. “Medium-term” = 10 years. The “Sta- tus quo” conflict scenario assumes continuation of the Donbas conflict; “Intermediate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegration,” full recovery, returning to preconflict values. GCA = government-controlled area; NGCA = non-government controlled area. Targeting matters. Another important observation is that targeting only the GCAs can indeed worsen the NGCA conditions, especially while the conflict is ongoing. GCA-only transfers create further incentives for migration from the NGCAs to the GCAs, and this effect is relatively strong when productivities re- main suppressed by the conflict. 170 In the status quo conflict scenario, for instance, with moderate transfers target- ing only the GCAs, the Donetsk NGCA loses an additional 7 percentage points in population (a 19.2 percentage point reduction as opposed to the 12.3 point reduction in the baseline). It also loses an additional 3.1 percentage points in GDP (an 8.2 percentage point decrease as opposed to a 5.1 point decrease in the baseline). In the reintegration case, these marginal effects narrow to 4.8 per- centage points for population and 2.8 percentage points for GDP. A Donbas peace dividend? Finally, in Donbas, transfers have the highest impact on economic activity (where they are applied) under the reintegration scenario. Both the GCA-only and the GCA and NGCA transfer schemes have greater im- pacts on the population and GDP numbers in corresponding target areas in the reintegration case. For instance, moderate, GCA-only transfers boost population in the Donetsk GCA by about 39.2 percentage points over the baseline in the re- integration case, while this effect is only at 26.2 percentage points in the status quo conflict case. This pattern also extends to the GDP numbers: the same transfer scheme boosts the Donetsk GCA by 20 percentage points under reintegration and by 10.2 percentage points under the status quo conflict case. Similar outcomes are observed from moderate, GCA-only transfers in the Luhansk GCA as well as in Donbas as a whole under transfer schemes that target GCAs and NGCAs to- gether. Mobility Reforms within Ukraine Interregional migration acts to limit wage differentials between regions. When conflict leads to a sudden drop in productivity and wages in one region, some people move to other regions. This movement raises wages in the conflict-af- fected region and decreases it in the destination regions. Thus, migration helps the original shock to dissipate through all regions, but this adjustment takes time. The speed and diffusion rate of such adjustment is determined by mobility costs. The higher the mobility costs, the slower the propagation. Such costs in- clude not only physical factors like the actual travel time and cost but also other Future Development: 864 Scenarios factors like the availability of certain amenities such as housing and publicly pro- vided services (including education, health, transportation, energy, and the like). Mobility costs can also include psychosocial factors like ethnoreligious identity, language, and social cohesion problems when such characteristics have spatial aspects. Previous chapters have analyzed these characteristics in detail. This section analyzes the effects of a hypothetical 50 percent decrease in mo- bility costs between all regions in Ukraine (except for the mobility costs between the Donbas NGCAs and other Ukrainian regions, which are treated separately in the conflict scenarios). Such a reduction can result from certain market reforms in housing policy, residence-registration requirements, transportation, and cit- izen access to finance, among other things. Simulation results for changes in regional GDP, population, and wages are presented in figure 4.5. 171 Population effects. For the Donbas GCAs, lower mobility costs in Ukraine lead to a greater out-migration in the case of a continued conflict and either a slower outflow or greater inflow in the case of reintegration (figure 4.5, panel b). Spe- cifically, in the status quo conflict scenario, the mobility cost reduction leads to an 18 percentage point increase in out-migration in the Donetsk GCA and a 25.5 point increase in the Luhansk GCA. The additional outflows from the NGCAs are similar. In contrast, under the reintegration scenario, the Donetsk GCA receives a 14.5 percentage point increase in population, and the outflow from the Luhansk GCA slows down by about 19 percentage points. As for the NGCAs, lower mobility costs add to net outflow even in the case of re- integration. The Donetsk NGCA turns from an 8.8 percentage point net inflow to a 6.8 point net outflow, a 15.6 point change. Similarly, in the Luhansk NGCA, a 19.1 percentage point inflow under reintegration is largely offset by the out-migration brought about by the lower mobility costs. GDP effects. Lower mobility costs affect GDP through population movements (figure 4.5, panel a). Given net out-migration, GDPs in the Donetsk and Luhansk GCAs decrease under the status quo scenario by 9.4 and 12.1 percentage points, respectively, relative to the baseline (without cost reduction). Under the rein- tegration scenario, however, lower migration costs add 7.0 and 9.4 percentage points to their respective GDPs. Similarly to the population effects, lower mobility costs decrease GDP in the NGCAs in both the status quo and reintegration scenarios—in status quo, by 12.1 percentage points in both the Donetsk and Luhansk NGCAs, and in reintegration, by 10.7 percentage points in the Donetsk NGCA and by 12.6 points in the Luhansk NGCA. Wage effects. As expected at the outset, migration patterns shape wages. As the lower mobility costs boost out-migration in the status quo and intermediate cases, wages increase in the GCAs by close to 10 percentage points (figure 4.5, panel c). In the reintegration case, lower mobility costs boost net immigration in the GCAs; thus wages in the Donetsk GCA decrease by 8.7 percentage points and in the Luhansk GCA by 10.7 points. In the NGCAs, however, lower mobility The Economics of Winning Hearts and Minds costs lead to a greater out-migration in all conflict scenarios, hence increasing wages. Finally, we should note that a sizable share of the emigrants from Donbas ar- rive in Kyiv City. However, Kyiv City is neither the only destination for Donbas emigrants, nor is Donbas the only source of migrants to Kyiv City. With lower migration costs in Ukraine, Kyiv’s population increases by more than 10 percent- age points, and the conflict situation in Donbas—and thus Kyiv-bound migration from Donbas—has a relatively limited effect on this. 172 Figure 4.5 Simulated Effects of Medium-Term Mobility Cost Reductions (from 2020 Values) in Donbas and Kyiv City, by Conflict Scenario and Location a. Gross Regional Product 0.6 0.5 0.4 Index (2013=1) 0.3 0.2 0.1 0 -0.1 -0.2 -0.3 o e n o e n o e n o e n o e n at at at at at qu io qu io qu io qu io qu io at at at at at i i i i i ed ed ed ed ed s s s s s gr gr gr gr gr u u u u u rm rm rm rm rm at at at at at te te te te te St St St St St te te te te te in in in in in Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA b. Population 0.3 0.2 0.1 Index (2013=1) 0 -0.1 -0.2 -0.3 -0.4 o te n o te n o te n o te n o te n qu io qu qu io io qu io qu io ia ia ia ia ia at at at at at ed ed ed ed ed s s s s s r r r r r u u u u u eg eg eg eg eg m m m m m at at at at at nt nt nt nt nt r r r r r St St St St St te te te te te i i i i i Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA c. Real wages 0.7 0.6 0.5 Index (2013=1) 0.4 0.3 0.2 0.1 Future Development: 864 Scenarios 0 -0.1 -0.2 o e n o te n o te n o te n o te n at qu io qu tio qu io qu io qu io ia ia ia ia at at at at i ed ed ed ed ed ra us s s us us r gr gr r u u eg eg eg rm rm rm rm m at at at at at te te nt nt nt r St St St St St te te te te te in in i i i Re Re Re Re Re In In In In In Donetsk GCA Donetsk NGCA Kyiv City Luhansk GCA Luhansk NGCA Baseline Lower cost Difference Source: World Bank estimates. Note: Results are evaluated using baseline values for all other scenarios (current levels of investments, transfers, and public sector employment). “Medium-term” = 10 years. “Lower cost” assumes a 50 percent decrease in mobility costs between all regions in Ukraine except to and from non-government controlled areas (NGCAs). The “Status quo” conflict scenario assumes continuation of the Donbas conflict; “Intermediate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegration,” full recovery, returning to preconflict values. GCA = government-con- trolled area. 173 Implications for Effective and Efficient Policies The analysis so far has focused on elaborating the outcomes of specific poli- cy actions under different circumstances. This has been done by focusing on two conditions: First, a given policy action is evaluated in isolation from others. Second, this evaluation uses a set of predetermined outcome indicators. This approach is a necessary first step of our diagnostics. However, designing effec- tive and efficient policies would benefit from an additional step in our thinking. Specifically, the answers to the following questions are important: •• Which metrics should be used for comparing alternative policy actions? •• Given such metrics, and for a given conflict scenario, how can we rank alter- native policy actions? •• How does uncertainty regarding conflict dynamics affect ex ante policy choice? The first question emphasizes the importance of choosing the right metric in as- sessing which policy works better. This is an important issue because these indi- cators eventually transform into policy objectives. The second question is driv- en by the observation that resources for policies are limited and that there are trade-offs between alternative policy actions. It is crucial to elaborate on these aspects to prioritize actions effectively and efficiently. Finally, the third question acknowledges that although conflict dynamics are unpredictable, policy makers may have prior beliefs regarding the likelihood of different paths ahead. Thus, if certain policies perform better in more-likely scenarios, then they are likely to be assigned greater weights in planning. The remainder of this section takes a detailed look at our simulation results to shed light on these issues. Given the multiplicity of scenarios and evaluation criteria, we will focus on key results. Comparisons will often abstract away from The Economics of Winning Hearts and Minds values and magnitudes, and focus on rankings, to make the discussion more tractable. Choosing Policy Objectives Policies are deemed successful when they induce the desired change in out- comes, but outcomes can be defined in different ways. In the case of economic recovery in Donbas, the region’s aggregate economic activity (for example, GDP in the Donbas GCAs) provides a pragmatic indicator for measuring policy effec- tiveness that is widely understood and feasibly calculated. However, this section shows that focusing exclusively on GDP in Donbas can hide some implicit trade- offs, including some concerning GDP in Ukraine as a whole. Moreover, in certain cases, the desirability of policies may depend on whether we use a GDP-based metric or a broader welfare-based metric. 174 To elaborate on these points, we compare the outcomes of different policy op- tions—moderate investments, moderate transfer increases, and reduced mobil- ity costs—by using two GDP-based indicators (local GDP in Donetsk [by GCA or NGCA] and nationwide GDP) and a welfare-based indicator (inequality-adjusted personal incomes, or IPIs). The latter uses personal incomes (wages + transfers) as the basis of analysis, and its aggregation for Ukraine as a whole reflects dif- ferent degrees of concern regarding inequality. Specifically, we use an Atkinson Index with three alternative inequality aversion parameter values (0, 2, and 5) to generate an inequality-adjusted indicator of personal income.2 Table 4.3 shows the results of these policy comparisons, presenting the changes in IPIs in percentage change terms3 and the changes in GDP in percentage points (using normalized 2013 values). This prohibits a cardinal comparison between different categories, but comparisons between the rankings of various policies within such categories are still possible. Trade-offs between Donbas and Ukraine GDPs. The first important observa- tion from table 4.3 is that there may be trade-offs between boosting GDP in Donbas and GDP in Ukraine as a whole. This is the clearest in the case of mobil- ity-cost reduction. In both the status quo and intermediate conflict scenarios, a 50 percent decrease in mobility costs within Ukraine increases Ukrainian GDP more than the Donbas-targeted investments or transfers would. However, this choice comes at the expense of GDP in Donbas. Intuitively, a reduction in mobility costs within Ukraine reallocates labor away from Donbas as a whole (also from the GCAs in particular), where labor produc- tivity suffers from conflict, faster than in other policy scenarios. This further re- duces GDP in Donbas. In fact, this is the least favorable policy for GDP in Donbas but the most favorable policy for Ukrainian GDP. Nevertheless, the net effect is positive because productivity is greater at the destination. This trade-off dis- appears under the reintegration scenario because the productivity in Donbas improves with dissipating conflict. A similar trade-off also exists between transfers and investments. With reinte- gration, increased transfers boost GDP in the Donetsk GCA more than invest- ments. In fact, a GCA-only transfers scenario, in this case, delivers the greatest increase in the Donetsk GCA’s GDP (44.2 percentage points) in the whole sim- ulation. Future Development: 864 Scenarios As far as Ukrainian GDP is concerned, however, investments are better than transfers. With reintegration, GCA-only investments boost Ukrainian GDP by 4.6 percentage points, but GCA-only transfers boost it by only 4.3 percentage points. This trade-off, however, does not hold in the case of status quo, where investments are better than transfers for the GDPs of both the Donbas GCA and Ukraine as a whole. Trade-offs between boosting GDP and IPIs. The second observation is that the GDP and IPI objectives can provide contradictory results. If the main objective is to maximize Ukrainian GDP, as noted earlier, then investments are preferred to transfers. For instance, moderate investments in the GCAs boost Ukrainian GDP by 0.8 percentage points annually in the medium term (10 years), whereas moderate GCA-targeted transfers boost it by only 0.4 percentage points in the same time frame. 175 When we switch to IPIs as the metric of comparison, the ranking switches, and transfers are preferred to investments. GCA-targeted transfers boost IPIs by 2 percent, and investments by 0.5 percent. This is mainly because investments have a more direct effect on GDP and an indirect effect on personal incomes, whereas transfers have a more direct effect on personal incomes and an indirect effect on GDP. These results hold in all conflict scenarios and with all degrees of inequality aversion. Trade-offs between supporting migration and wages. Finally, as far as the mo- bility-cost scenario is concerned, the IPI results take the same direction as the GDP results for the Donbas GCA and thus contrast with the Ukraine GDP results. Lower mobility costs reduce average wages outside Donbas because migration reduces marginal labor productivity in the recipient locations. Hence, the invest- ments and transfers scenarios fare better than the mobility-cost case in increas- ing incomes in Donbas, reducing migration, and preserving average wages else- where in Ukraine. Overall, increasing the productivity in Donbas through peace is more effective than reallocating agents through out-migration, because out-mi- gration leads to underutilization of fixed production factors in the Donbas region. Determining Functional and Spatial Allocation of Resources The analysis so far has compared investment and transfer policy scenarios in terms of their effects on GDPs or IPIs in isolation. In practice, both approaches can be pursued. The problem is about the scale. With budget constraints, allo- cating resources over multiple instruments will reduce the scale of each. How- ever, this may be a preferred situation if such a policy mix brings about better outcomes. This section considers the effectiveness of an investment-transfer policy mix compared with investment-only or transfer-only approaches. To take budget constraints into account, we compare large investments, large transfers, and a The Economics of Winning Hearts and Minds policy mix comprising moderate investments and moderate transfers. These would entail similar resource envelopes in the medium term (about 10 years in our simulations).4 The results are presented in table 4.4, where the first four rows in each conflict scenario (no policy, investments, transfers, and policy mix—all considered in Donbas with GCA and GCA + NGCA options) provide the informa- tion for this discussion. Alleviating trade-offs with a policy mix. The first important result from our functional allocation simulations is that investments and transfers satisfy dif- ferent policy objectives, and using a policy mix helps to alleviate some of the trade-offs associated with this dichotomous case. To see this, let us compare the effects of large, GCA-targeted investments and transfers on Ukrainian GDP and on IPIs with no inequality aversion. The investments boost Ukrainian GDP by 1.6 percentage points and the transfers by 0.5 percentage points. In comparison, the investments boost IPIs by only 1.2 percent, while the transfers boost them by 5.0 percent. Thus, whereas investments are preferred for the purposes of Ukrainian GDP, transfers are the choice for the purposes of IPIs. 176 Table 4.3 Simulated Changes in Medium-Term GDP versus Inequality- Adjusted Personal Incomes in Donbas and Ukraine, by Policy Choice and Conflict Scenario Percentage point change in index value (2013=100) Gross domestic product Inequality-adjusted personal incomesa (percentage point change) (percentage change, whole Ukraine) scenario Conflict Donetsk Donetsk Whole Moderate High No aversion GCA NGCA Ukraine aversion aversion Policy Policy tool target Value Rank Value Rank Value Rank Value Rank Value Rank Value Rank No policy -3.5 (17) -5.1 (15) 0.2 (17) 0.0 (16) 0.2 (16) 1.0 (16) GCA 8.3 (12) -5.8 (16) 0.8 (14) 0.5 (15) 0.8 (14) 1.5 (14) Investments (moderate) GCA + Status quo 1.8 (15) -0.5 (13) 0.6 (15) 0.6 (14) 0.8 (14) 1.5 (14) NGCA GCA 6.7 (14) -8.2 (17) 0.4 (16) 2.0 (13) 2.2 (12) 2.9 (12) Transfers (moderate) GCA + 1.2 (16) -2.2 (14) 0.0 (18) 3.4 (8) 3.5 (8) 4.0 (8) NGCA Mobility Ukraine -12.9 (18) -17.2 (18) 1.3 (13) -3.0 (18) -2.6 (18) -2.2 (18) No policy 11.5 (11) 12.5 (9) 1.9 (12) 2.4 (11) 2.7 (11) 3.4 (11) GCA 22.2 (7) 11.2 (10) 2.6 (8) 3.0 (9) 3.1 (10) 3.7 (9) Investments (moderate) Intermediate GCA + 15.9 (10) 16.3 (7) 2.4 (9) 3.0 (9) 3.2 (9) 3.7 (9) NGCA GCA 20.9 (8) 8.0 (11) 2.2 (10) 4.3 (7) 4.3 (7) 4.6 (7) Transfers (moderate) GCA + 18.5 (9) 14.8 (8) 2.0 (11) 6.2 (3) 5.7 (3) 5.5 (5) NGCA Mobility Ukraine 7.4 (13) 3.5 (12) 3.0 (7) -0.7 (17) -0.3 (17) 0.0 (17) No policy 24.2 (6) 34.4 (3) 3.8 (5) 5.1 (6) 5.0 (6) 5.2 (6) GCA 38.5 (3) 33.4 (4) 4.6 (1) 5.7 (5) 5.5 (5) 5.6 (3) Investments (moderate) Reintegration GCA + 30.5 (5) 40.8 (2) 4.3 (2) 5.8 (4) 5.6 (4) 5.6 (3) NGCA GCA 44.2 (1) 31.6 (5) 4.3 (3) 7.4 (2) 7.0 (2) 6.7 (2) Future Development: 864 Scenarios Transfers (moderate) GCA + 41.2 (2) 51.6 (1) 3.8 (6) 10.5 (1) 9.8 (1) 9.1 (1) NGCA Mobility Ukraine 31.2 (4) 23.8 (6) 4.3 (4) 2.1 (12) 2.1 (13) 2.0 (13) Source: World Bank estimates. Note: Results are for moderate investment and transfer scenarios. The “mobility” policy tool considers a 50 percent reduction in all bilateral mobility costs between Ukrainian oblasts. Scenarios that are not shown (for example, transaction costs in jobs and certain options in investment location scenarios) are held at baseline levels. The "Rank" columns indicate the order of benefit (1 being the most beneficial) to the jurisdiction indicated at the top of the column. “Medium-term” = 10 years. The “Status quo” conflict scenario assumes continuation of the Donbas conflict; “Intermediate,” sufficient recovery to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegration,” full recovery, returning to precon- flict values. GCA = government-controlled area; NGCA = non-government controlled area. a. Inequality-adjusted personal incomes (including wages + transfers) are generated using an Atkinson Index with three alter- native inequality aversion parameter values (0, 2, and 5). 177 Using a policy mix of both lands us between the two results—boosting Ukrainian GDP by 1.1 percentage points and the IPIs by 2.6 percent. Thus, the policy mix does not dominate either instrument in either objective, but it delivers a more balanced outcome across the two policy objectives. This conclusion holds for all conflict scenarios and degrees of inequality aversion. Balancing the policy mix by region. While using a policy mix, outcomes can theoretically be improved by taking into consideration the ROI in different plac- es. A standard efficiency argument from spatial economics literature suggests that investments should target areas with the greatest economic potential (for example, possibly with agglomeration effects).5 To consider this idea, we amend our policy mix scenario by undertaking these new investments in Kyiv instead of Donbas while keeping the transfers in Don- bas. The idea here is that productivity is higher in Kyiv because of the con- flict and legacy problems in Donbas (described in chapter 1) as well as potential agglomeration effects in Kyiv. Thus, by undertaking the additional investments there instead of in Donbas, the economic ROI would be increased. In addition, by employing transfers in Donbas, we would also be satisfying the equity argument. Thus, everyone would be better off. In table 4.4, the last two rows in each conflict scenario show the simulation re- sults that investigate the validity of these arguments in the case of Donbas. The Kyiv (1) scenario (without agglomeration effects) captures the actual productiv- ity differences between Donbas and Kyiv City as observed before the onset of the conflict. The Kyiv (2) scenario (with agglomeration effects) considers greater productivity in Kyiv City—potentially brought about by a better business environ- ment (for example, better governance, institutional quality, and market-friendly reforms)6 or greater benefits from further access to European markets. Simulations show that the “policy mix: Kyiv (1) + Donbas” scenario performs worse than the “policy mix: Donbas” scenario. That is, shifting investments to Kyiv City under the current business environment reduces their effectiveness regardless of the policy objective (the Donetsk GCA GDP, Ukraine GDP, or IPIs with all degrees of risk aversion). Specifically, the "policy mix with Kyiv (1) + Don- The Economics of Winning Hearts and Minds bas" scenario boosts Ukrainian GDP by only 0.7 percentage points and the IPIs by 2.3 percent in the status quo conflict case (as opposed to 1.1 percentage points and 2.6 percent, respectively, with the “policy mix: Donbas” scenario alone). Intuitively, with mobility costs (material and psychosocial) limiting a rapid demographic adjustment and the current business environment limiting better growth prospects, the productivity differential between Donbas and Kyiv City is not large enough to justify the efficiency argument of spatial economics, even for nationwide objectives. In contrast, the "policy mix with Kyiv (2) + Donbas" scenario performs slightly better than the “policy mix: Donbas” scenario in the status quo and intermediate conflict cases. With greater productivity in Kyiv, reallocating investments there and undertaking transfers in Donbas boosts Ukrainian GDP by 0.1 percentage points and IPI growth by 0.3 percent more than the "policy mix: Donbas" scenar- io (under status quo conflict and GCA-only transfers). However, this advantage also disappears in the reintegration case. 178 Table 4.4 Functional and Spatial Allocation Estimates in Donbas and Ukraine, by Conflict Scenario and Policy Objective Percentage point change in index value (2013=100) Gross domestic product Inequality-adjusted personal incomesa (percentage point change) (percentage change, whole Ukraine) scenario Conflict Donetsk Whole Moderate Donetsk GCA No aversion High aversion NGCA Ukraine aversion Policy Policy tool Value Rank Value Rank Value Rank Value Rank Value Rank Value Rank target No policy -3.5 (33) -5.1 (28) 0.2 (32) 0.0 (33) 0.2 (33) 1.0 (33) GCA 23.6 (17) -7.1 (29) 1.6 (23) 1.2 (31) 1.4 (32) 2.0 (32) Investments (large): Donbas GCA + 8.2 (27) 4.9 (22) 1.1 (25) 1.2 (31) 1.5 (31) 2.1 (31) NGCA GCA 19.9 (21) -14.7 (33) 0.5 (29) 5.0 (20) 4.8 (20) 5.0 (18) Transfers (large): Donbas GCA + 14.5 (25) -0.9 (25) 0.0 (33) 8.0 (9) 7.2 (10) 6.6 (11) NGCA Status quo GCA 18.9 (22) -10.2 (32) 1.1 (25) 2.6 (28) 2.8 (27) 3.4 (27) Policy mix: Donbas GCA + 6.0 (30) 1.6 (24) 0.5 (29) 4.0 (23) 4.0 (23) 4.3 (23) NGCA GCA 6.6 (28) -8.2 (30) 0.7 (28) 2.3 (30) 2.4 (30) 3.1 (30) Policy mix: Kyiv (1) + Donbas GCA + 1.1 (31) -2.2 (26) 0.4 (31) 3.8 (24) 3.8 (24) 4.2 (24) NGCA GCA 6.5 (29) -8.3 (31) 1.2 (24) 2.9 (27) 2.8 (27) 3.3 (29) Policy mix: Kyiv (2) + Donbas GCA + 1.1 (31) -2.2 (26) 0.9 (27) 4.4 (22) 4.1 (22) 4.4 (22) NGCA No policy 11.5 (26) 12.5 (17) 1.9 (21) 2.4 (29) 2.7 (29) 3.4 (27) GCA 37.5 (13) 9.2 (18) 3.5 (11) 3.6 (26) 3.7 (25) 4.1 (25) Investments (large): Donbas GCA + 21.7 (18) 21.3 (13) 3.0 (13) 3.7 (25) 3.7 (25) 4.1 (25) NGCA GCA 45.7 (6) 3.8 (23) 2.4 (19) 8.0 (9) 7.5 (9) 7.0 (9) Transfers (large): Donbas GCA + 42.0 (9) 25.6 (12) 1.9 (21) 12.6 (2) 11.2 (2) 9.7 (4) NGCA Intermediate Future Development: 864 Scenarios GCA 36.2 (14) 6.6 (21) 2.9 (15) 5.1 (18) 4.9 (18) 5.0 (18) Policy mix: Donbas GCA + 25.8 (15) 19.9 (14) 2.5 (17) 7.0 (13) 6.3 (12) 5.9 (14) NGCA GCA 20.8 (19) 8.0 (19) 2.5 (17) 4.7 (21) 4.5 (21) 4.8 (21) Policy mix: Kyiv (1) + Donbas GCA + 18.4 (23) 14.8 (15) 2.3 (20) 6.6 (15) 5.9 (16) 5.7 (16) NGCA GCA 20.7 (20) 7.9 (20) 3.0 (13) 5.3 (17) 4.9 (18) 5.0 (18) Policy mix: Kyiv (2) + Donbas GCA + 18.2 (24) 14.7 (16) 2.9 (15) 7.1 (12) 6.3 (12) 5.9 (14) NGCA 179 Table 4.5 continued Gross domestic product Inequality-adjusted personal incomesa (percentage point change) (percentage change, whole Ukraine) scenario Conflict Donetsk Whole Moderate Donetsk GCA No aversion High aversion NGCA Ukraine aversion Policy Policy tool Value Rank Value Rank Value Rank Value Rank Value Rank Value Rank target No policy 24.2 (16) 34.4 (6) 3.8 (10) 5.1 (18) 5.0 (17) 5.2 (17) GCA 58.7 (4) 32.3 (7) 5.6 (1) 6.5 (16) 6.2 (15) 6.1 (13) Investments (large): Donbas GCA + 39.1 (12) 49.5 (5) 5.0 (4) 6.7 (14) 6.3 (12) 6.2 (12) NGCA GCA 78.7 (1) 28.4 (11) 4.7 (6) 11.9 (3) 11.1 (3) 10.2 (2) Transfers (large): Donbas GCA + 69.3 (2) 79.7 (1) 3.2 (12) 20.1 (1) 19.1 (1) 17.4 (1) NGCA Reintegration GCA 63.0 (3) 30.7 (10) 5.2 (2) 8.3 (7) 7.8 (7) 7.4 (7) Policy mix: Donbas GCA + 49.8 (5) 60.3 (2) 4.4 (8) 11.5 (4) 10.7 (4) 9.9 (3) NGCA GCA 44.0 (7) 31.4 (8) 4.7 (6) 7.8 (11) 7.2 (10) 6.9 (10) Policy mix: Kyiv (1) + Donbas GCA + 41.0 (10) 51.4 (3) 4.2 (9) 10.8 (6) 10.0 (6) 9.2 (6) NGCA GCA 43.7 (8) 31.3 (9) 5.2 (2) 8.3 (7) 7.6 (8) 7.1 (8) Policy mix: Kyiv (2) + Donbas GCA + 40.7 (11) 51.2 (4) 4.8 (5) 11.4 (5) 10.4 (5) 9.5 (5) NGCA Source: World Bank estimates. Note: Policy mix scenarios consider moderate investments and transfers. Kyiv (1) and Kyiv (2) denote investments undertaken in Kyiv City (with baseline and higher productivities, respectively). Transfers are always undertaken in Donbas. Scenarios that are not shown (for example, transaction costs in jobs) are held at baseline levels. The "Rank" columns indicate the order of benefit (1 being the most beneficial) to the jurisdiction indicated at the top of the column. “Medium-term” = 10 years. The “Status quo” conflict scenario assumes that the Donbas conflict will continue indefinitely; “Intermediate” that Donbas will recover enough to undo, by half, the productivity and mobility cost effects of conflict-driven shock; and “Reintegration” that Donbas has full recovery, returning to preconflict values. GCA = government-controlled area; NGCA = non-government controlled area. a. Inequality-adjusted personal incomes (including wages + transfers) are generated using an Atkinson Index with three alter- native inequality aversion parameter values (0, 2, and 5). The Economics of Winning Hearts and Minds Intuitively, greater productivity in the rest of Ukraine creates a large enough pro- ductivity differential to justify the shift in investment. However, as the conflict dissipates and the productivity in Donbas rebounds, that gap shrinks, eliminat- ing the efficiency argument. In all cases, using a policy objective of boosting the Donetsk GCA’s GDP would automatically favor investments in Donbas regardless of the productivity dynamics in the rest of Ukraine. Overall trade-off conclusions. Overall, the discussion here makes two import- ant observations: first, a brick-and-mortar approach to economic recovery in Donbas (for example, blindly focusing on only infrastructure investments in a specific region) is not necessarily efficient or effective. This comes out strongly when trade-offs between different policy objectives (for example, GDP in Don- bas versus GDP in Ukraine) are considered explicitly as we did here. 180 Second, there is an economic rationale to scale up efforts in Donbas in the status quo scenario, which is good for both Donbas and Ukraine. If and when produc- tivity in other locations in Ukraine increases significantly (as a result of reforms or better market access), then an efficiency argument suggests against using resources for Donbas. But even that is argument is negated when productivity in Donbas rebounds under a reintegration scenario. Making Decisions under Uncertainty The discussion so far has focused on the ex post outcomes—that is, analyzing what would happen to various outcome indicators in a specific conflict sce- nario if a certain policy instrument were chosen. This approach abstained from assessing whether such a scenario is feasible going forward. However, in reality, policy decisions are made in an ex ante world in a stochastic manner. Policy makers often have prior beliefs about the future trajectories of certain events (for example, the likelihood of conflict scenarios), and they tailor policies to these expectations. To consider such an ex ante decision-making problem, we now incorporate a probabilistic approach to our simulation results. Specifically, we compare the effectiveness of three policy instruments: investment-only (large in the Donetsk GCA); transfer-only (large in the Donetsk GCA); and policy mix (moderate in- vestments and moderate transfers in the Donetsk GCA). To do this, we employ three policy objectives (outcome indicators): GDP in the Donetsk GCA, GDP in Ukraine, and IPIs with moderate inequality aversion. Finally, because this analysis does not assume any prior probability of conflict scenarios, we consider the whole range of probabilities. That is, we compare policy outcomes when the probability of reintegration can take any value between 0 (impossible) to 1 (with certainty).7 We drop the intermediate conflict scenario for expositional clarity. Results are presented in figure 4.6. Choices under uncertainty for Ukrainian GDP and IPIs. For both Ukraine GDP and IPIs, the likelihood of reintegration does not alter the ordering of policies. Figure 4.6, panels b and c, show that the gain lines representing different policy Future Development: 864 Scenarios instruments (red for investments, green for transfers, and yellow for the policy mix) do not cross. This is mainly because the same instrument is dominant in both the status quo and reintegration conflict scenarios (investment-only for Ukrainian GDP and transfer-only for IPIs). In both cases, the policy mix remains consistently in between the other options over all probabilities of reintegration. However, there is a major difference between the GDP (panel b) and IPI (panel c) figures. In the former case, the gap between the gains from the investment and transfer scenarios (indicated by the blue-shaded bars) decreases with greater likelihood of reintegration. In contrast, this gap widens in the case of IPI gains. This is mainly because the impact of transfers on both Ukraine GDP and IPIs is more responsive than the impact of investments to the probability of reintegra- tion. This works to close the initial investment-transfer gap in the Ukraine GDP case and to widen it in the case of IPIs. 181 Figure 4.6 Medium-Term Policy Outcomes under Uncertainty in Donbas and Ukraine, by Type a. GDP gains, Donetsk GCA 80 5 70 0 60 -5 Percentage points Percentage points 50 40 -10 30 -15 20 -20 10 0 -25 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Probability of reintegration b. GDP gains, Ukraine 6.0 1.2 5.0 1.1 4.0 Percentage points Percentage points 3.0 1.0 2.0 4 0.9 1.0 0 0.8 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Probability of reintegration c. IPI gains, Ukraineª 12 -2.5 10 -3.0 8 -3.5 Percentage points Percentage points 6 -4.0 The Economics of Winning Hearts and Minds 4 -4.5 2 -5.0 0 -5.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Probability of reintegration Investment-transfer gap (right scale) Policy mix, Donbas Investment only Transfer only Source: World Bank estimates. Notes: Results are evaluated for baseline values for all other scenarios (current levels of investments, transfers, and public sector employment). “Reintegration” refers to a full recovery, returning to preconflict values in productivity and mobility costs. Probability of reintegration values range from 0 (impossible) to 1 (with certainty). GCA = government-controlled area. a. Inequality-adjusted personal incomes (IPIs), including wages and transfers, are generated using an Atkinson Index (here, with moderate inequality aversion). 182 Choices under uncertainty for GDP in the Donetsk GCA. Perhaps the most in- teresting result is about the policy rankings concerning GDP in the Donetsk GCA (figure 4.6, panel a). With a low probability of reintegration, investments in the GCA stand out as the most preferred option. However, with about 20 percent or more likelihood of reintegration, the impact of transfers exceeds that of invest- ments. A higher probability of rreintegration reinforces the wedge of transfers over investments. To see the mechanism driving this result, note that transfers perform worse than investments in the status quo conflict scenario. Specifically, transfers increase GDP in Donbas only by attracting people into the region, which remains limited under the conflict conditions. In addition, everyone in the region, including the new arrivals, are still subject to the low-productivity conditions driven by con- flict. However, investments improve the productivity of both the incumbent res- idents and others who may be attracted by marginally higher productivity (and thus higher wages). Thus, the GDP effect of investments is significant in Donbas even when workers from other regions are not attracted. Therefore, in the status quo conflict scenario, investments dominate transfers when we use GDP in Don- bas as a metric. However, in the reintegration scenario, these differences disap- pear with rebounding productivity, which then makes transfers more effective in attracting people from other regions. Synopsis The chapter analyzed how different policy interventions can affect a variety of economic outcomes in Donbas and in Ukraine. To consider the trade-offs be- tween alternative actions, and to overcome the challenges of nonverifiability (associated with descriptive-only methods) and structural breaks brought by the conflict (associated with empirical extrapolations), we used a general equi- librium model with strong microfundamentals that are conducive to studying economic incentives for migration. These aspects are particularly important in Donbas, where a significant out-migration of younger generations poses a formidable challenge to future economic dynamics. The key results from these Future Development: 864 Scenarios analyses are summarized below. The case for peace. Although the conflict is not the only source of structur- al problems in Donbas, it prevents the effective implementation of policies to address those problems. The analysis in this chapter showed that there is a “peace dividend”—that is, a partial economic recovery led by the removal of conflict-driven distortions in the economy. By the same token, common policy instruments like investments are more effective in boosting growth under the re- integration scenario than under the status quo. Nonetheless, the end of conflict neither restores the economy to its preconflict levels nor resolves the structural problems dating from before the conflict’s onset. The case for scaling up development in Donbas. The analysis shows that there is an economic rationale for scaling up development policies in Donbas even in 183 the absence of reintegration. (In other words, scaling up policies in Donbas is better for Ukrainian GDP than scaling them up elsewhere.) However, if produc- tivity in the rest of Ukraine increases significantly (as a result of policy reforms or better access to European markets), then an “efficiency” argument that favors investing in the rest of Ukraine becomes relevant. Nevertheless, a reintegration can dismiss this caveat by closing the productivity gap between Donbas and the rest of Ukraine. The case for a nuanced development strategy. In defining the future policy agenda, authorities should avoid off-the-shelf, one-size-fits-all approaches that resemble wish lists. As discussed in previous chapters, the region needs a com- prehensive strategy (a complete and contingent plan) that is adaptive to chang- ing conditions on the ground, has clear policy objectives, and reflects a balance between various trade-offs (explicitly calculated). Policy rankings under the status quo (current conditions prevail). Both in- vestments and transfers are effective in improving economic outcomes in Don- bas, but there are trade-offs between different policy objectives: •• Donbas GCA GDPs as policy objective: Both investments (I) and transfers (T) are significantly effective (I>T for Donetsk and T>I for Luhansk) in boost- ing GDP in the GCAs. Lowering mobility costs in Ukraine is the worst option because it increases out-migration from the GCAs. •• Ukrainian GDP as policy objective: Investments are the first-best option. Lowering mobility costs is also effective because it reallocates labor from lower-productivity areas in Donbas to higher-productivity areas in the rest of Ukraine. •• Reconciling the two objectives: Using a mixed policy approach does not de- liver a first-best outcome for either objective, but it provides a second-best outcome in most cases, lowering the trade-offs (a) between investments and transfers when GCA GDPs are considered, and (b) between GCA GDPs and Ukrainian GDP when the mobility cost scenario is considered. Policy rankings under reintegration (productivity rebounds partially in Don- bas). With productivity bouncing back, the scope for different policy instru- The Economics of Winning Hearts and Minds ments changes under a reintegration scenario: •• Donbas GCA GDPs as policy objective: Transfers are more effective than investments in attracting people from other regions and boosting GDP in the GCAs (both the Donetsk GCA and the Luhansk GCA). •• Ukrainian GDP as policy objective: Investments are still the first-best option (which, unlike transfers, boost productivity). With smaller productivity gaps between regions after the conflict, lowering mobility costs is less effective than in the status quo scenario. •• Reconciling the two objectives: As in the status quo case, using a mixed policy approach can help balance Donbas-specific policy objectives with Ukraine-specific policy objectives. The next chapter will combine these points with the observations summarized in previous chapters to help characterize a number of policy recommendations. 184 Notes 1. Note that some combinations of different scenarios—for example, a status quo conflict scenario and investments in both GCAs and NGCAs—may be unrealistic or not feasible. These are excluded from discussion. 2. The Atkinson Index is a widely used income and wealth inequality index in the literature. The index allows the user to adjust a sensitivity parameter that can range from 0 (no aversion to inequality) to infinity (extreme aversion to inequality). 3. It was not meaningful to normalize IPI values as we did with other indicators (using the preconflict values as baselines) because, with different risk aversion preferences, the initial (preconflict) values also differ. 4. An exact match between investment and transfer program budgets is not feasible to establish in our simulation model. This is because the total transfer bill is endogenously determined: as transfers attract people to change locations, the total number of beneficiaries—and thus the total transfer bill—changes. In our simulations, the difference between the investment and transfer bills becomes significant in the long term (for example, the latter doubles), but in the short and medium terms, they are very similar. 5. For a discussion and critique of these approaches, see Rodríguez-Pose (2018). 6. For a detailed description of such market-friendly reforms, see World Bank (2020). 7. Because agents are risk-neutral and rational, it is possible to calculate expected welfare conditional on a specific prob- ability of reintegration by taking a linear combination of outcomes under reintegration and no-reintegration scenarios weighted by their probabilities. If the agents were risk-averse (for example, with a utility specification that is a nonlinear function of wages), then this simplification would not be possible. References Artuç, Erhan, Paulo S. R. Bastos, and Eunhee Lee. 2021. “Trade, Jobs, and Worker Welfare.” Policy Research Working Paper 9628, World Bank, Washington, DC. Artuç, Erhan, Nicolás Gómez Parra, and Harun Onder. 2019. “Return Simulations.” Background paper for The Mobility of Dis- placed Syrians: An Economic and Social Analysis. Washington, DC: World Bank. Rodríguez-Pose, Andrés. 2018. “The Revenge of the Places that Don’t Matter (and What to Do about It).” Cambridge Journal of Regions, Economy and Society 11 (1): 189–209. World Bank. 2017. The Toll of War: The Economic and Social Consequences of the Conflict in Syria. Washington, DC: World Bank. World Bank. 2020. The Mobility of Displaced Syrians: An Economic and Social Analysis. Washington, DC: World Bank. Future Development: 864 Scenarios 185 The Economics of Winning Hearts and Minds 186 Chapter 5 Conclusions and Conjectures for a Development Strategy T his report has focused on the current economic and social challenges faced by the Donbas region, analyzing the effectiveness of potential policy interventions. To put things into perspective, it first analyzed the economic and social trends in the run-up to the onset of the armed conflict. Next, it an- alyzed the mechanisms through which conflict has manifested its impact on these trends. These mechanisms constitute a “3 Ds” framework—destruction, displacement, and disorganization—for analyzing the drivers of current condi- tions and challenges, which were next discussed in three areas: economic activ- ity, labor market and demography, and access to public services. Against this backdrop, the report used a modeling approach to analyze the effec- tiveness of alternative policy actions (investments, transfers, and mobility cost reduction) in progressing toward different objectives: boosting gross domestic product (GDP) in the government-controlled areas (GCAs) of Donbas, boosting GDP in Ukraine, and boosting average personal incomes in Ukraine (with various degrees of inequality aversion). This approach helped to elaborate explicitly on the trade-offs associated with each policy intervention. Where Do We Stand? A Recap It is important to build the future trajectory of economic development in Donbas on a solid understanding of where the region’s economy stands today and how it arrived there. The following two subsections summarize our key findings in this Conclusions and Conjectures for a Development Strategy regard. Dynamics of Change The analysis so far has shown that the economic woes of the Donbas region did not start with the conflict. The region was at the center of a legacy growth model that relied on energy-intensive exports in heavy industry (largely metal- lurgy), a model that worked well during the 2000s, when the external conditions were favorable: (a) booming demand from major trade partners; (b) major price hikes in commodities, which increased export revenues and pushed up input costs for potential competitors; and (c) concessional access to energy inputs via subsidized coal, electricity, and gas and oil imports from the Russian Fed- eration. However, as these conditions changed after the Global Financial Crisis of 2008–09, this bubble burst, and the region faced increasingly unfavorable economic prospects. How did the conflict affect these trends? With the onset of the conflict in 2014, these problems were magnified along with the emergence of new challenges through several channels. 189 Destruction channel. The conflict has accelerated infrastructure depreciation by inflicting considerable damages on strategic assets, especially in the trans- portation, water and sanitation, and energy sectors, and particularly during the first phase of the conflict. Displacement channel. It also accelerated the region’s demographic aging problem by displacing people, especially the younger generations. Disorganization channel. The conflict triggered several “invisible factors” that proved detrimental for the economy. As a contact line divided the region’s econ- omy into two—into a government-controlled area (GCA) and a non-government controlled area (NGCA) in both Donetsk and Luhansk Oblasts—connectivity be- came a major new problem, with some areas effectively becoming islands, espe- cially concerning railways. The region’s two major airports (Donetsk and Luhansk airports) were destroyed, and the region’s major seaport (Mariupol) has been performing well below capacity. The contact line also cuts through major public service provision systems, lead- ing to a fragile interdependency. Specifically, a low level of cooperation across the contact line leads to a suboptimal (and costly) provision of services, especially in the water and energy sectors. This fragile interdependency also manifests through a water-electricity-coal nexus whereby divided markets have created artificial excess supply and excess demand, aggravated by market distortions and feeding into rent seeking. The conflict has also deepened social divisions in the country, with diminishing intergroup interactions and trust in policy reforms. Current Conditions and Challenges The conflict has suppressed economic activity significantly. However, it is often not possible to observe this directly from official data, which conflate the effects of conflict with an accounting omission: the data report indicators that cover the entire oblast before 2014 and only the oblast’s GCA since 2014. The Economics of Winning Hearts and Minds A comparison of nighttime light emissions shows that, between 2013 and 2018, economic activity decreased by 29 percent in the Donetsk GCA, 12 percent in the Luhansk GCA, 36 percent in the Donetsk NGCA, and 45 percent in the Lu- hansk NGCA. Much of this decrease was driven by a shrinking industrial output and, in the case of Luhansk, by shrinking industrial productivity too. Agricultural output and productivity increased in both GCAs, but it still comprises a small share of output and employment in these regions. The current constraints to growth include three prominent factors: •• Conflict-driven disorganization and risks suppress economic activity direct- ly as well as indirectly through supply irregularities, higher input costs, con- nectivity bottlenecks, and unfavorable external competition. •• Severe investment restrictions have resulted from diminished access to fi- nance and from very high risk premia paid by those who can borrow against all odds. 190 •• Changing patterns of economic opportunities have led to a significant skill mismatch problem in the labor market. For example, basic machine opera- tors and maintenance workers are oversupplied in the GCAs, especially in the Luhansk GCA, and skilled workers are undersupplied in both GCAs. Unfavorable demographic trends can also hinder dynamic future development in Donbas. Currently, the GCAs in Donbas have the worst labor market condi- tions in Ukraine. They also have the highest median age and the highest old-age dependency ratios in the country. This deepening aging problem has been ag- gravated by a disproportionate displacement of younger generations away from Donbas. The vast majority of those internally displaced persons (IDPs) have no intention of returning in the near future—especially those who are young, male, and have been displaced for a long time. This will further complicate economic growth opportunities in Donbas. Safety remains an important challenge given the military presence, shelling, and unexploded ordnance (UXOs) near the contact line. Infrastructure needs ma- jor rehabilitation; for example, fixing windows and roofs are among the top-ex- pressed needs across all schools. And both the providers and receivers of pub- lic services need psychosocial support. Water and sanitation conditions remain problematic in rural areas (especially in Luhansk), and the conflict has acceler- ated the water supply pollution problem drastically (through mine closures and flooding, which contaminate water sources). Even the central water distribution system often does not meet safety standards. In the absence of a change in conflict trajectory or a scaling-up of mitigation efforts, the economic and social fallout of the conflict will continue to unfold in the Donbas region. Economic activity is likely to deteriorate further, pushing more people away from the region. The analysis in this report shows that peace, by itself, will not be sufficient to bring the economy back to its preconflict level. Conclusions and Conjectures for a Development Strategy Moreover, it will certainly not fix the structural problems that constrained the economy even before the conflict. However, a reintegration can remove the con- flict-driven distortions discussed in this report, increase the effectiveness of mitigation policies, and drive a partial recovery (peace dividend). What Is to Be Done? Our analysis shows that, under current circumstances, there is an economic ra- tionale to scale up efforts in Donbas even without reintegration. Scaling-up would boost GDP in both Donbas and Ukraine as a whole. The conventional efficiency versus equity trade-off argument—that resources should be used elsewhere in Ukraine where productivity is higher—applies only if productivity in the rest of Ukraine would increase significantly as a result of policy reforms or better ac- cess to European markets. Even then, reintegration could render this argument null by reducing the productivity gap between Donbas and the rest of Ukraine. 191 The analysis also highlights the importance of scaling up in a coherent manner. In defining the future policy agenda, authorities should avoid off-the-shelf and one-size-fits-all approaches, as discussed next. The Case for a Comprehensive Strategy The region needs a comprehensive and integrated strategy (a complete and contingent plan) that can coordinate various interventions around the same de- velopment path in the years to come. Such a strategy would be, among other things: •• Balanced: Explicitly weighing trade-offs associated with different policy in- terventions and policy objectives (no silver bullet) •• Nuanced: Designing mechanisms that are responsive to changing conditions on the ground, especially the conflict scenarios (one size does not fit all) •• Transformative: Removing market distortions to unleash economic growth without introducing new distortions to mitigate the economic consequences of the old ones. The first characteristic in this framework—balanced—is driven by simulation re- sults, which show that some policy instruments may be effective in improving one indicator (such as GDP in the GCAs) but not others (such as GDP in Ukraine). Chapter 4 details how different policy instruments (investments, transfers, and mobility cost reduction) can be used to achieve different objectives (increased GDP in the GCAs, Ukrainian GDP, and inequality-adjusted personal incomes). A “silver bullet” approach (for example, only infrastructure investments or only Special Economic Zones) is not an effective approach. A mixed policy approach does not necessarily deliver a first-best outcome for either objective, but it does provide a second-best outcome in most cases, hence reducing trade-offs. The Economics of Winning Hearts and Minds The second characteristic—nuanced—emphasizes the finding that the compo- sition of effective, efficient policies will differ between a conflict scenario (status quo) and a peace scenario (reintegration). For instance, a mobility cost reduction approach is effective under the status quo for increasing Ukrainian GDP because it helps reallocate labor from low-productivity areas to high-productivity areas (which reduces GDP in the GCAs but increases wages, so both those who stay and those who migrate are better off). However, as reintegration closes produc- tivity gaps between Donbas and the rest of Ukraine, this effect is weakened. A complete and contingent plan would calibrate actions for each scenario closely. The third characteristic—transformative—suggests that fixing the market funda- mentals by removing distortions would promote development more sustainably than by introducing more distortions to offset the adverse effects of preexist- ing ones. For instance, any incentive mechanism designed to attract investors in the region is likely to be captured by rent-seeking actors in the absence of improvements in business climate, rule of law, and anticorruption policies. Thus, removing such barriers should be considered a priority. 192 Figure 5.1 A Decision Tree Approach to Designing an Economic Recovery Strategy for Eastern Ukraine Recovery strategy p=0 p=1 Status quo Reintegration Investments Transfers Mobility Investments Transfers Mobility Contingent (status quo only) policies Contingent (reintegration only) policies No-regret policies Source: ©World Bank. Further permission required for reuse. Note: “Contingent” policies are specific to either the status quo or reintegration scenarios. “No-regret” policies refer to those that would be desirable under both the status quo and reintegration scenarios (for example, improvements in business climate and housing market). Better Data for Better Policy Before discussing the next steps, we would like to highlight that an important element of any development strategy is information. In Ukraine, better data are needed urgently in three major areas: •• Demography: Official demographic statistics are residence-based and un- Conclusions and Conjectures for a Development Strategy likely to reflect the situation on the ground. We recommend that a census be considered in the near future. •• Damages and needs: Our knowledge about the conflict-driven damages, overall conditions of infrastructure systems, and service access needs are fragmented and in most cases outdated. A new round of damage and needs assessments is needed to inform the future recovery strategies (in addition possibly to a census of firms and UXOs). •• Regional statistics: Official statistics cover the entire oblasts of Donetsk and Luhansk before 2014 and only their GCAs since 2014. This often leads to misinterpretations: for example, differences between 2013 and the years af- ter are interpreted as actual changes, which is misleading. Efforts to either backcast the GCAs’ data series or to forecast the NGCAs’ series would help to resolve this problem. 193 Adapting Policies to Conditions The discussion now turns to how policies can adapt to different conflict scenar- ios as described above (“nuanced”). To this end, we propose a distinction be- tween (a) contingent policies, which have appeal only within particular conflict and peace contexts (the status quo or reintegration scenarios); and (b) no-re- gret policies, which would be desirable in both conflict and peace (reintegration) cases (figure 5.1). In the provision of many public services, the optimal scale and structure of the system depends on the area where the system operates and whether the pre- conflict infrastructure can be used again as a whole (for example, water canals, the electric grid, and railways). On one hand, recalibrating the system of service provision only for the GCAs may prove costly and ineffective if the previously more-integrated structure of the system becomes available (in which case a suboptimally disintegrated and two redundant systems—one in the GCAs and the other one in the NGCAs—would operate in parallel). On the other hand, delaying such recalibration may prolong suffering and cause welfare loss if reintegration does not take place. If a policy is highly sensitive to such calculations, then it is a “contingent policy.” If it is not sensitive—that is, it is desirable in both status quo and reintegration—then it is a “no-regret” policy. Separating the interventions in this way, and following a decision tree approach, can create opportunities for making progress. Contingent Policies Our simulations emphasized the need for a multipronged approach to design- ing development strategy. This was true for both the status quo and reintegra- tion scenarios. The question then becomes how contingent policies differ from no-regret policies. The difference sometimes lies in either the type of policy or its magnitude. The Economics of Winning Hearts and Minds Under the status quo scenario, both investments (I) and transfers (T) can effec- tively boost GDP in the GCAs, albeit to different degrees in different places (I>T for Donetsk and T>I for Luhansk). Reducing mobility costs, on the other hand, boosts Ukrainian GDP but lowers GDP in both of the Donbas GCAs. By compari- son, with productivity partially rebounding under reintegration in Donbas, trans- fers can boost GDP more effectively than investments in the GCAs (T>I for both GCAs). This is done primarily by attracting workers from other regions (including some IDP returnees). However, this is not ideal for Ukrainian GDP, for which a productivity-boosting investment works better even if small. Thus, from a Ukrainian GDP perspective, the policy mix calls for a greater weight for investments under both the status quo and reintegration scenarios. There is also a strong rationale for mobility cost reduction under the status quo, which weakens under reintegration. For GDP in the GCAs, however, the policy mix would give relatively more weight to transfers under reintegration and to investments under the status quo (especially in the Donetsk GCA). 194 What does this all mean from a practical perspective? The transfer component of the policy mix can help address the excess demand problem for skilled work- ers in the GCAs. In addition to the no-regret policies (discussed below), the “sta- tus quo-contingent” transfers would include a “conflict premium” in social assis- tance and insurance mechanisms (with the possibility of a carefully ring-fenced direct transfer scheme in conflict-affected areas) and higher supplemental in- come for workers, including hazard pay and a mobility premium. These mecha- nisms essentially provide a risk premium for working under escalated security risks and enduring hardship associated with lower access to public services, among other factors. Similarly, the assistance targeting IDPs should reflect ongo- ing hardships stemming from an active conflict and consider additional support for gaining job skills and access to housing. As for investments, some public services need to be recalibrated in terms of both magnitude and targeting. Safety and the reduction of conflict-related risks are priorities for investment. This includes provision of a safe school environment (eliminating military presence, shelling, and UXOs at schools); road safety for all; and psychosocial support (adapted to active conflict situations) for providers and recipients of social services. For businesses, public assistance in designing insurance mechanisms against conflict-driven risks can be considered; however, serious consideration should be given to moral hazard and monitoring problems, and alternative mechanisms (such as index insurance) should be evaluated. A public sector intervention to improve access to finance and simplifying taxes (or breaks) for new small and medium enterprises (SMEs) should also be consid- ered. (Again, such interventions also fall into the “no-regret” category to some extent, but an active conflict may necessitate a greater effort.) All infrastructure investments should be coordinated around a coherent infra- structure strategy aligned with economic objectives and conflict conditions for a “status quo only” reality. This process would include picking low-hanging fruit Conclusions and Conjectures for a Development Strategy in (a) connectivity (including railways extensions and roads) to eliminate cur- rent transportation bottlenecks; and (b) service delivery systems like water and sanitation services (while being mindful of possible redundancy in the case of reintegration). Special attention should be paid to managing the quality of public investment projects in both the fiduciary sense and in terms of project selection based on economic returns and absorptive capacity constraints. Categorically, all policies targeting the NGCAs are contingent on reintegration. These include both major reconstruction projects (like the two main regional airports in Donetsk and Luhansk Cities) and other economic and social interven- tions aimed at promoting growth. However, some reconstruction or rehabilitation projects for the GCAs will also need to be postponed until reintegration. These include major reconstruction projects (for example, Mariupol airport) that are either deemed too costly absent a broader service area or that risk attracting military attention. 195 No-Regret Policies The potential for “no-regret” policies lies in the desirability of many of the po- tentially most consequential policy interventions under both the status quo and reintegration scenarios. These policies, rather than following a unidimensional “brick and mortar” approach, would encompass multiple dimensions aimed at eliminating the legacy market distortions driven by past policies that ultimate- ly delayed Ukraine’s transition toward a more market-driven and contestable economy. Regulatory and anticorruption reforms. A first policy direction in this regard is the elimination of regulatory burdens and corruption. Perceptions of the ac- countability, responsiveness, and transparency of local authorities remain con- sistently low in the east. In dealing with this problem, economic recovery initia- tives should ensure that beneficiary feedback mechanisms, grievance redress processes, asset disclosures, and transparency and accountability methods are in place for investments and that transgressions are addressed promptly. This is important for the credibility of reforms because Ukrainian residents in the east generally do not trust that their institutions will effectively implement the chang- es they would like to see. Policy overtures, reforms, and investments should re- quire communication campaigns that properly frame the intent and nature of the engagement. Housing market opening. A second policy direction is the improvement of the housing market. The lack of sufficient housing is one of the main obstacles to efficient interregional mobility in Ukraine. The country has so far left a potential efficiency gain untapped as workers remain trapped in low-productivity areas. IDPs who moved away from Donbas also cite lack of adequate housing as one of the most important problems they face. Ukraine’s housing market problems are intertwined with land market problems, residence and registration regulations, and other institutional factors that over- regulate a process that other countries simply manage much more effectively. Addressing these institutional problems to promote a more dynamic housing The Economics of Winning Hearts and Minds market would be beneficial both in Donbas and in Ukraine. Labor market programs. A third no-regret policy direction concerns improve- ment of labor market conditions to offset skill mismatches as well as the labor supply problems driven by demographic aging. These policies would seek to align vocational schools with market conditions and boost adult (re)training. In addition, working-age individuals who are outside the labor force could be ac- tivated through improved access to childcare facilities and family-friendly jobs (to activate women); psychological support and career counseling of discour- aged individuals; subsidized employment for disadvantaged groups (persons with disabilities, older people, and so on); and part-time employment in the for- mal sector for youth and other categories. Increasing the effectiveness of the State Employment Service of Ukraine (including expanding digital skills and jobs databases) and the outreach of its active labor market programs (ALMPs) are important prerequisites to increasing labor market participation. Education system improvements. A fourth policy direction concerns the mod- 196 ernization of the Ukrainian education system. Adopting a modern, student-cen- tered, and problem- and project-based curriculum with thematic focus on entrepreneurship would be beneficial. Involving employers in skill-oriented pro- grams would improve the employability of graduates. A greater focus on civic education and ethics in the curriculum could also play an important role in fos- tering opposition to corruption. Targeted infrastructure investments. A fifth policy direction concerns improv- ing infrastructure as a means to catalyze private investment. There are several ways to improve upon the current conditions without a great risk of redundancy in the case of reintegration. Much of the infrastructure in the GCAs needs reha- bilitation (for example, schools, water and sanitation systems, and transportation systems). In general, repairs and rehabilitation in these areas, as well as mea- sures to improve environmental outcomes (such as reducing water pollution), are desirable regardless of future conflict scenarios. New infrastructure should target the elimination of service delivery and con- nectivity islands (physical and digital) and prioritize the projects with the most welfare impact and least risk of redundancy under a reintegration scenario. Pro- posed projects should be coordinated around a master plan, prioritized accord- ing to viability (subject to rigorous public investment management principles), and communicated well to local beneficiaries. Conclusion Conclusions and Conjectures for a Development Strategy Today, the Donetsk and Luhansk Oblasts of Ukraine are left in a difficult place by a nexus of demographic, social, infrastructural, and institutional problems. Most of these problems existed long before the conflict, and perhaps paved the way to it, but they have been left unaddressed and are now reinforced by the hostilities. To rekindle the economy, arrest the decline in potential output, and foster development, these two oblasts need a comprehensive reform and de- velopment program, strategically formulated, that sets realistic goals focused on key medium- and long-term priorities. Such a strategy cannot rely only on government action and discretion; it would need to attract private investment, both local and foreign, and stop or even re- verse the Donbas region’s adverse demographic trends (out-migration and de- mographic aging). The analysis in this report showed that there are simply no “silver bullets” that can deliver these outcomes with a simple touch. Instead, a multipronged strategy that is balanced, nuanced, and transformative is needed to be credible: The strategy needs to be balanced because policy decisions are subject to trade-offs, as the simulations in this report showed. It needs to be nuanced because changing conditions on the ground, like the trajectory of the conflict, call for different approaches. Finally, it also needs to be transformative because trying to re-create the past economic structures would fall short of delivering a brighter economic path for the region. 197 Thus, the strategy should focus on building a sound institutional foundation—in- cluding market principles and ample provision of public goods and services— upon which private agents can operate competitively. Given the complexity of current socioeconomic dynamics, the implementation of such a strategy must attend to a few concerns. To succeed, the proposed policies would require local ownership. This is particularly important in Donbas, where trust in top-down reforms is very low, as discussed earlier in this report. Therefore, extra care must be taken to involve local authorities and beneficiaries in decision-making processes. Commendably, the authorities have implement- ed consultation processes, but more is needed. Citizen-government engage- ment (facilitated as needed) and community consultation during the design and implementation of policies, reforms, and projects will also help to develop great- er trust and ownership. Finally, it is important to note that, although the conflict is not the only source of the region’s problems, in the absence of a permanent resolution and with associated uncertainties, the economic downtrend is likely to continue. The hostilities suppress the region’s economy both directly (by reducing productive capacity) and indirectly (by reducing the effectiveness of policies). Therefore, a permanent resolution to the conflict, and a swift removal of conflict-driven dis- tortions, are essential for a sustainable recovery in the future. In eastern Ukraine, peace and prosperity will likely go hand in hand. The Economics of Winning Hearts and Minds 198 Appendixes Appendix A Ukraine before Conflict: Trade, Labor, and Demographic Indicators, 2013 Figure A.1 Ukraine’s Trade Patterns, 2013 a. What did Ukraine export? Power Oil seeds Other generating Metalliferous transport machinery and ores and and equipment metal scrap oleaginous equipment fruit 1.85% Iron General industrial 2.25% machinery and equipment, nes, Crude fertilizer Cork and parts of, nes and crude ICT and 3.48% 1.72% minerals and wood Electric Machinery Telecommunicati… 0.90% machinery, specialized for particular 9.85% apparatus and industries 5.44% 1.00% 0.13% appliances, Fertilizers, Petroleum, steel nes, and 0.71% Inorganic chemicals manufactured petroleum Road Metal… products and parts, nes vehicles related Transport materials 0.20% 2.71% 1.08% 0.40% 0.04% 1.78% Feeding stuff Coal, coke Electric for animals current Cereals and (not including 1.50% 1.41% and unmilled Artificial Dyeing, Organic briquettes cereal cereals) resins and tanning and chemicals plastic colouring preparations materials,… materials 1.12% 0.40% 0.37% 0.27% Vegetables Dairy 0.69% and fruit products and birds' 1.36% 0.20% eggs Articles… Miscellane… Special transactions, Fixed commodity not classified 9.64% 18.32% according to 0.77% 0.70% vegetable class Coffee, tea, Meat… oils and Paper, Manufacture… Non-metallic Cork and cocoa, paperboard, mineral wood, cork spices, and Travel and and manufactures, nes manufactures manufactures thereof fats 0.81% articles of 0.79% 0.76% 0.06% 0.06% pulp, of 0.50% Beverages Furniture paper or of and parts 0.41% tourism paperboard 0.65% 0.68% thereof Non-ferrous Miscellan… Sugar,… Sanit… 0.47% metals 1.02% 3.70% 5.78% 1.29% 0.52% 7.52% 0.37% 0.37% 0.59% 0.20% 201 Appendixes The Economics of Winning Hearts and Minds 202 b. Where did Ukraine’s exports go? Poland Belarus Italy Turkey China Egypt Russia 4.19% 3.76% 3.05% 6.57% 4.15% Netherlands Azerbaijan 4.31% Moldova Czechia Kazakhstan Saudi Iraq Germany Arabia Morocco Nigeria 0.59% 0.50% Ethiop… South… Kenya 1.29% 1.27% 1.18% Tunisia Japan Lebanon Syria 1.56% 1.35% 1.34% Iran 0.25% 0.22% 0.21% France Romania United 0.46% 2.92% Kingdom 0.71% 0.67% 0.66% Uzbekista… Libya United Indonesia Jordan Hungary Arab 0.91% 0.83% Emirates 1.17% Austria Belgium 3.24% 1.13% 0.42% Israel British Slovakia 0.58% 0.54% United 0.63% 0.58%Armenia Virgin Bangla… Malays… States Singapore Turkmenist… 0.79% 0.69% India of Islands Lithuania Denmark Serbi… America 0.59% Greece Belize Mexico 2.44% 0.40% 0.28% 0.28% 0.27% 0.27% 0.19% Thailand Vietna… 1.16% 0.32% 0.91% Switzerla… 0.60% 0.27% 0.22% Spain Bulgaria 0.49% Georgia Cuba Portugal South 0.39% 0.27% 0.31% Korea 1.33% 23.57% Pakistan Cyprus 0.21% 0.13% 2.83% 1.99% 1.01% 0.42% 0.84% 0.59% 0.35% 0.22% Source: United Nations Comtrade database, https://comtrade.un.org/. Table A.1 Selected Demographic and Labor Market Indicators in Ukraine, by Zone and Region, 2013 Old-age dependency Share of urban Share with tertiary education (%)b Labor force Household disposable Median age Unemployment Average Zone and region ratio (per 1,000 persons population Total population Employed participation income per capita (years)a rate (%)c wage (Hrv)d aged 15–64 years)ª (%)ª (aged 25–70 years) (aged 25–70 years) rate (%)c (Hrv)e Zone 1 Donetsk 41.9 241 90.6 51.2 55.7 65.4 7.8 3,755 31,049 Luhansk 42.1 228 86.8 46.5 49.5 63.3 6.2 3,337 25,590 Zone 2 Dnipropetrovsk 40.3 223 83.5 47.2 51.4 66.5 6.5 3,336 30,301 Zaporizhzhia 41.2 223 77.1 48.5 52.5 65.6 6.6 3,142 28,388 Kharkiv 40.5 210 80.3 49.6 54.2 65.7 6.4 2,975 26,098 Zone 3 Kirovohrad 41.2 247 62.3 46.0 50.4 64.5 7.9 2,608 21,671 Mykolaiv 39.7 207 67.8 43.8 46.5 65.4 7.4 3,094 23,869 Poltava 41.4 241 61.4 46.6 49.6 64.7 8.2 2,988 25,371 Sumy 42 232 67.9 39.6 43.9 65.1 7.7 2,702 23,559 Kherson 39.5 205 61.2 43.7 43.2 65.2 8.5 2,464 21,724 Cherkasy 41.7 252 56.4 43.9 46.5 65.8 8.9 2,682 21,633 203 Appendixes The Economics of Winning Hearts and Minds 204 Table A.1 continued Old-age dependency Share of urban Share with tertiary education (%)b Labor force Household disposable Median age Unemployment Average Zone and region ratio (per 1,000 persons population participation income per capita (years)a Total population Employed rate (%)c wage (Hrv)d aged 15–64 years)ª (%)ª rate (%)c (Hrv)e (aged 25–70 years) (aged 25–70 years) Zone 4 Vinnytsia 40.4 248 50.2 44.6 46.4 65.1 8.4 2,651 23,001 Zhytomyr 39.4 236 58.4 39.8 42.0 66.0 9.3 2,561 21,652 Kyiv Oblast 39.7 218 61.8 38.5 41.8 63.4 6.1 3,351 27,391 Odesa 38.4 200 66.9 44.3 47.0 63.0 5.3 2,947 25,572 Chernihiv 42.8 274 63.5 45.7 48.1 66.8 9.3 2,504 23,600 Kyiv City 37.6 172 100 72.6 76.3 68.4 5.2 5,007 55,842 Zone 5 Volyn 35.7 187 52.1 38.1 41.6 64.8 7.8 2,580 19,805 Zakarpattia 35.1 158 37.2 30.5 36.5 63.5 7.8 2,553 17,929 Ivano-Frankivsk 37.3 198 43.4 40.6 42.1 59.8 7.2 2,679 20,988 Lviv 38 205 60.9 48.6 53.4 63.3 7.1 2,789 23,138 Rivne 35.2 178 47.8 39.1 41.1 65.7 9.4 2,844 21,165 Ternopil 38.6 221 44.1 44.3 45.9 62.0 9.4 2,359 18,994 Table A.1 continued Old-age dependency Share of urban Share with tertiary education (%)b Labor force Household disposable Median age Unemployment Average Zone and region ratio (per 1,000 persons population participation income per capita (years)a Total population Employed rate (%)c wage (Hrv)d aged 15–64 years)ª (%)ª rate (%)c (Hrv)e (aged 25–70 years) (aged 25–70 years) Khmelnytsky 40.4 239 55.5 44.5 47.6 64.9 8 2,641 22,789 Chernivtsi 36.9 195 42.5 34.2 31.8 63.4 7.4 2,484 19,438 Zone 6 Crimean AR 39.9 209 62.7 45.4 46.4 66.1 5.7 2,850 22,793 Sevastopol City 40.2 219 93.8 62.3 63.4 66.3 5.7 3,114 26,584 Ukraine 39.7 217 68.9 46.8 50.2 65.0 7.2 3,265 26,719 Sources: State Statistics Service of Ukraine (SSSU) and Labor Force Survey (LFS) data. Note: Following IOM (2019), the grouping of oblasts into zones is based on their distance from the present-day non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. Crimean AR = Autonomous Repub- lic of Crimea; GCA = government-controlled area; Hrv = hryvnias. a. SSSU statistics available at (in Ukrainian only) http://database.ukrcensus.gov.ua/MULT/Dialog/statfile_c.asp. b. Calculations based on the individual-level LFS data. Tertiary education includes incomplete, basic, and complete higher according to the Ukrainian classification. c. SSSU statistics based on LFS for population aged 15–70 years, available at http://ukrstat.gov.ua/. d. SSSU, average gross wage of employees employed in firms with at least 10 employees, available at http://ukrstat.gov.ua/. e. SSSU, household income per capita, available at http://ukrstat.gov.ua/. 205 Appendixes The Economics of Winning Hearts and Minds 206 Table A.2 Indicators of Employment in Ukraine, by Zone and Region, 2013 Sectoral composition of employment Occupational composition of employment Composition of employment by status Incidence of informal (%)a (%)b (%)b employment (%)b Zone and region Wage and Employers and own- Only wage Industry and Public Other High Skilled Skilled Total Agriculture Unskilled salaried account workers not in Other and salaried construction services services skilled nonmanual manual employment workers subsistence farming workers Zone 1 Donetsk 10.7 30.5 15.7 43.1 32.6 18.9 36.3 12.3 92.4 4.6 3.0 12.4 9.1 Luhansk 12.9 27.9 16.7 42.5 30.1 15.5 30.1 24.2 78.6 4.3 17.1 22.4 6.7 Zone 2 Dnipropetrovsk 6.9 30.1 17.2 45.8 33.5 21.4 31.1 14.0 90.6 3.1 6.3 12.1 6.0 Zaporizhzhia 16.4 26.9 17.3 39.4 33.4 19.2 28.1 19.3 83.0 4.4 12.6 20.1 7.6 Kharkiv 13.5 23.7 18.2 44.6 38.0 20.8 24.1 17.1 90.1 3.4 6.5 14.5 8.5 Zone 3 Kirovohrad 30.7 16.2 20.0 33.1 31.5 17.4 26.1 25.1 77.6 5.8 16.6 25.5 9.2 Mykolaiv 25.8 17.4 19.1 37.7 31.9 19.1 23.0 26.0 75.3 9.0 15.7 27.6 12.3 Poltava 19.8 22.3 19.8 38.1 29.0 19.0 29.2 22.8 82.8 3.4 13.7 20.2 6.5 Sum 21.6 20.8 20.3 37.2 24.8 16.2 23.1 35.8 72.1 4.0 23.9 34.2 13.4 Kherson 30.8 12.7 19.5 37.0 24.4 19.8 21.3 34.6 66.7 8.0 25.4 39.2 18.0 Cherkasy 27.8 19.5 19.7 33.1 32.7 15.7 21.0 30.6 74.7 5.8 19.5 30.5 13.2 Table A.2 continued Sectoral composition of employment Occupational composition of employment Composition of employment by status Incidence of informal (%)a (%)b (%)b employment (%)b Zone and region Wage and Employers and own- Only wage Industry and Public Other High Skilled Skilled Total Agriculture Unskilled salaried account workers not in Other and salaried construction services services skilled nonmanual manual employment workers subsistence farming workers Zone 4 Vinnytsia 33.0 13.4 21.2 32.4 26.9 16.2 17.5 39.5 64.8 5.4 29.9 37.3 10.6 Zhytomyr 15.4 19.0 22.4 43.3 27.6 18.7 23.6 30.2 75.1 5.8 19.1 27.2 9.1 Kyiv Oblast 7.0 22.1 22.5 48.4 30.9 25.4 28.6 15.1 88.4 2.9 8.8 13.3 4.6 Odesa 16.2 13.2 19.7 50.9 35.1 21.6 17.5 25.8 79.7 4.9 15.4 24.6 11.0 Chernihiv 26.5 15.1 22.1 36.2 31.2 17.4 21.4 30.1 72.1 4.8 23.1 28.3 6.0 Kyiv City 0.3 13.6 20.5 65.6 66.0 15.5 13.9 4.7 95.8 4.2 0.1 4.1 3.3 Zone 5 Volyn 26.7 15.3 22.5 35.5 30.4 16.8 16.3 36.5 75.8 3.2 21.0 32.3 14.4 Zakarpattia 25.4 16.8 19.1 38.7 29.8 16.6 14.4 39.2 68.5 4.9 26.6 44.3 23.9 Ivano-Frankivsk 28.0 17.4 20.6 33.9 25.3 17.3 24.1 33.3 74.9 5.9 19.3 44.0 29.7 Lviv 19.9 21.5 21.5 37.1 36.1 19.8 24.3 19.7 84.6 4.5 11.0 19.0 7.8 Rivne 18.9 18.0 20.9 42.3 26.4 14.0 18.6 41.0 58.9 6.0 35.1 45.5 13.2 Ternopil 33.5 12.9 22.0 31.6 29.7 12.6 15.7 41.9 65.2 3.8 31.0 42.6 17.2 207 Appendixes The Economics of Winning Hearts and Minds 208 Table A.2 continued Sectoral composition of employment Occupational composition of employment Composition of employment by status Incidence of informal (%)a (%)b (%)b employment (%)b Zone and region Wage and Employers and own- Only wage Industry and Public Other High Skilled Skilled Total Agriculture Unskilled salaried account workers not in Other and salaried construction services services skilled nonmanual manual employment workers subsistence farming workers Khmelnytsky 28.4 15.2 20.6 35.8 31.4 18.0 18.8 31.8 74.4 5.5 20.2 28.0 9.9 Chernivtsi 27.4 15.0 19.8 37.8 16.5 17.0 19.3 47.2 61.0 9.8 29.2 52.1 27.7 Zone 6 Crimean AR 20.2 12.7 20.3 46.8 33.5 20.1 22.6 23.8 80.8 4.7 14.5 26.3 12.8 Sevastopol Cityc 2.7 20.5 23.9 52.9 — — — —  — — — — — Ukraine 17.5 20.4 19.5 42.6 33.7 18.6 24.0 23.7 80.9 4.7 14.4 23.6 10.0 Sources: State Statistics Service of Ukraine (SSSU) and Labor Force Survey (LFS) data. Note: Statistics refer to employed population aged 15–70 years. Following IOM (2019), the grouping of oblasts into zones is based on their distance from the present-day non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. a. World Bank calculations based on SSSU statistics on employment by sector and region. “Public services” include public administration and defense, education, human health, and social work. b. World Bank calculations based on LFS data. “High skilled” includes International Standard Classification of Occupations (ISCO) groups 1–3: managers, professionals, and associate professionals. “Skilled nonmanual” includes ISCO groups 4 and 5: clerical support workers and service and sales workers. “Skilled manual” includes ISCO groups 6–8: skilled agricultural, forestry, and fishery workers; craft and related trades workers; and plant and machine operators. c. — = not available separately for Sevastopol City, whose occupational and employment data are included in those for the Autonomous Republic of Crimea (Crimean AR). Appendix B Ukraine during Conflict: Characteristics of Internal Migration Table B.1 Sociodemographic Characteristics of IDPs and Returnees, by Region, 2018 Share of respondents (%) SUBPOPULATION -> IDPs Returnees Variable Ukraine excluding REGION -> Donetsk (GCA) Luhansk (GCA) Total Donetsk (NGCA) Luhansk (NGCA) Total Donbas Urban 92.4 81.5 88.4 88.5 92.0 91.9 91.9 Type of settlement Rural 7.0 18.0 10.4 10.6 2.8 2.8 2.8 NR 0.6 0.5 1.1 0.9 5.2 5.3 5.3 0–14 years 26.6 20.6 24.8 24.7 10.6 10.4 10.5 15–24 years 8.7 9.6 9.8 9.4 4.0 3.8 3.9 Age 25–59 years 46.5 49.7 49.8 48.8 40.7 41.2 40.9 60–70 years 12.6 13.8 11.1 11.9 29.6 32.1 30.7 70+ years 5.7 6.3 4.4 5.0 15.0 12.5 13.9 Female 58.2 57.7 56.8 57.3 56.0 57.0 56.5 Gender Male 41.8 42.4 43.2 42.7 43.9 43.0 43.5 209 Appendixes The Economics of Winning Hearts and Minds 210 Table B.1 continued SUBPOPULATION -> IDPs Returnees Variable Ukraine excluding REGION -> Donetsk (GCA) Luhansk (GCA) Total Donetsk (NGCA) Luhansk (NGCA) Total Donbas Employed 32.1 37.1 36.3 35.3 27.3 27.8 27.5 Unemployed 3.2 3.1 3.2 3.2 2.0 2.1 2.0 Pensioners 22.8 25.1 19.6 21.3 51.5 51.0 51.3 Labor force statusa Students 20.1 17.6 20.3 19.8 9.1 7.9 8.5 Housework and 7.9 6.4 7.6 7.5 4.0 4.1 4.0 care Inactive, other 13.6 10.6 12.5 12.5 5.2 6.1 5.6 NR 0.5 0.2 0.6 0.5 1.1 1.0 1.0 N Sample 7,624 2,999 46,686 57,309 2,868 2,426 5,294 Source: World Bank calculations from National Monitoring Survey (NMS) data of the International Organization for Migration. Note: NMS datasets include all household members. Donbas comprises the Donetsk and Luhansk Oblasts. GCA = government-controlled area; IDPs = internally displaced persons; NGCA = non-government controlled area; NR = not reported, or unknown. a. Labor force status for IDPs and returnees is based on their answers to the question 1.6: “What is your current employment status or main activity?” Pensioners include nonworking individuals who receive old-age pension as well as disability pension or assistance. Table B.2 Selected Characteristics of IDPs (Heads of Household) in Ukraine, by Destination, 2018 Share of respondents (%) Ukraine excluding Survey question Response Donetsk (GCA) Luhansk (GCA) Donbas Salary 57.0 56.2 64.9 Government IDP support 60.3 70.9 47.9 Social assistance 25.8 22.8 32.9 Retirement pension 32.2 29.6 27.2 Sources of household income in the past Irregular earnings 7.4 9.7 9.4 12 months (multiple responses) Financial support from relatives residing in Ukraine 7.3 10.3 6.5 Disability assistance 7.6 5.2 5.7 Humanitarian assistance 7.9 3.4 5.3 Other pension (survivor’s benefit , social pension, and so on) 3.4 0.6 4.0 Have to limit expenses even for food 19.0 4.9 12.4 Enough funds only for food, lack for other needs 38.8 52.0 35.0 Financial situation (self-assessed) Enough funds for basic needs (food, clothing, footwear), unable to save 35.9 39.0 44.9 Enough funds for basic needs, also able to save for other expenses 4.3 2.6 5.9 The most problematic issue Lack of own housing 17.5 26.9 32.8 (only top 10 responses shown) 211 Appendixes The Economics of Winning Hearts and Minds 212 Table B.2 continued Ukraine excluding Survey question Response Donetsk (GCA) Luhansk (GCA) Donbas Living conditions 6.9 7.3 13.6 Lack of money 10.8 22.7 11.7 Payment for rent 9.9 8.1 8.4 Impossible to return to the place of origin 11.9 4.6 6.0 The most problematic issue Payment for utilities 17.2 6.0 5.1 (only top 10 responses shown) Unemployment 6.5 5.6 3.5 Access to medicines 2.1 1.1 1.8 Suspension of social payments or pensions 1.6 1.4 1.3 Access to health care 2.0 1.7 1.0 Rented apartment 53.6 43.0 53.1 Rented house 12.1 10.4 10.0 Host family or relatives 14.3 19.1 9.4 Where does your household live now? Rented room in an apartment 2.9 3.7 5.8 Dormitory 4.6 1.7 7.9 Own housing 6.0 10.3 7.1 Collective centers for IDPs 4.8 0.4 4.6 Table B.2 continued Ukraine excluding Survey question Response Donetsk (GCA) Luhansk (GCA) Donbas Agriculture 2.0 2.6 1.8 Industry 12.4 8.2 10.3 Construction 3.3 0.4 7.4 Trade 16.2 9.6 18.7 Transportation 6.3 0.9 6.6 Sector for employed in the last 7 days Public administration 13.7 26.5 5.9 (only face-to-face interviews) Education 14.7 20.6 6.5 Health care 5.6 6.4 5.4 Services and other activities 20.9 13.1 31.2 Own business (specify please) 0.9 2.0 1.2 Not reported 4.1 9.7 5.0 Source: World Bank calculations based on National Monitoring Survey (NMS) data of the International Organization for Migration. Note: Joint dataset is of internally displaced persons (IDPs) from face-to-face and phone interviews (rounds 9–12, weighted). Information is provided only for the heads of household, excluding IDPs from Crimea and individ- uals older than 70 years. Donbas comprises Donetsk and Luhansk Oblasts. GCA = government-controlled area. 213 Appendixes The Economics of Winning Hearts and Minds 214 Appendix C Recent Demographic and Labor Market Indicators Table C.1 Selected Demographic and Labor Market Indicators in Ukraine, by Zone and Region, Recent Year Old-age dependency Share of urban Share with tertiary education (%)b Labor force Median age Unemployment Average wage ratio (per 1,000 working- population participation rate (years)a Total population Employed rate (%)c (Hrv)d Zone and region age persons)a (%)a (%)c (aged 25–70 years) (aged 25–70 years) Jan 2020 Jan 2020 Jan 2020 2018 2018 2019 2019 2019 Zone 1 Donetske 45.4 325 90.9 48.8 55.7 58.9 13.6 11,716 Luhanske 46.5 331 87.1 50.2 53.1 68.1 13.7 8,731 Zone 2 Dnipropetrovsk 41.7 258 84.0 51.5 57.0 64.4 7.7 10,751 Zaporizhzhia 42.8 268 77.4 50.6 55.1 64.1 9.5 10,480 Kharkiv 41.9 250 81.2 55.5 59.0 65.4 5.0 9,081 Zone 3  Kirovohrad 42.6 277 63.4 49.2 55.1 62.5 11.0 8,360 Mykolaiv 41.3 248 68.6 49.1 52.9 65.1 9.3 9,976 Poltava 42.9 265 62.5 54.4 59.0 63.3 10.6 9,846 Table C.1 continued Old-age dependency Share of urban Share with tertiary education (%)b Labor force Median age Unemployment Average wage ratio (per 1,000 working- population participation rate (years)a Total population Employed rate (%)c (Hrv)d Zone and region age persons)a (%)a (%)c (aged 25–70 years) (aged 25–70 years) Jan 2020 Jan 2020 Jan 2020 2018 2018 2019 2019 2019 Sumy 43.8 272 69.4 45.1 48.3 64.8 7.7 8,579 Kherson 41.0 243 61.4 44.5 46.2 65.2 9.6 8,187 Cherkasy 43.5 283 56.9 49.0 53.6 64.7 8.3 8,838 Zone 4 Vinnytsia 41.7 266 51.7 48.5 52.2 64.0 9.4 9,299 Zhytomyr 40.5 249 59.3 44.8 50.0 64.7 9.6 8,528 Kyiv Oblast 39.9 228 62.1 46.7 53.2 63.1 5.9 11,003 Odesa 39.7 237 67.2 47.0 51.7 62.0 5.9 9,246 Chernihiv 43.9 299 65.5 50.9 57.7 65.6 10.2 8,206 Kyiv City 39.5 223 100.0 73.2 77.2 67.0 5.8 15,776 215 Appendixes The Economics of Winning Hearts and Minds 216 Table C.1 continued Old-age dependency Share of urban Share with tertiary education (%)b Labor force Median age Unemployment Average wage ratio (per 1,000 working- population participation rate (years)a Total population Employed rate (%)c (Hrv)d Zone and region age persons)a (%)a (%)c (aged 25–70 years) (aged 25–70 years) Jan 2020 Jan 2020 Jan 2020 2018 2018 2019 2019 2019 Zone 5 Volyn 37.2 196 52.3 41.4 49.4 56.9 10.6 8,663 Zakarpattia 36.7 177 37.2 27.9 36.7 60.9 9.1 9,202 Ivano-Frankivsk 38.8 209 44.4 42.7 46.9 61.0 7.2 8,817 Lviv 39.4 218 61.1 45.6 52.8 61.9 6.5 9,271 Rivne 36.5 189 47.5 40.6 42.5 63.7 8.3 8,967 Ternopil 40.2 227 45.6 48.7 58.4 59.8 10.0 8,275 Khmelnytsky 41.7 258 57.4 44.4 47.7 61.9 8.0 8,672 Chernivtsi 38.5 212 43.3 42.4 43.4 63.4 6.9 8,066 Ukraine 41.4 254 69.5 49.4 54.7 63.4 8.2 10,497 Sources: State Statistics Service of Ukraine (SSSU) and Labor Force Survey (LFS) data. Note: Following IOM (2019), the grouping of oblasts into zones is based on their distance from the non-government controlled areas (NGCAs) of Donetsk and Luhansk Oblasts. Hrv = hryvnias. a. SSSU statistics available at (in Ukrainian only) http://database.ukrcensus.gov.ua/MULT/Dialog/statfile_c.asp. b. World Bank calculations based on the individual-level LFS data. Tertiary education includes incomplete, basic, and complete higher according to the Ukrainian classification. c. SSSU statistics based on LFS for population aged 15-70 years, available at http://ukrstat.gov.ua/. d. SSSU, average gross wage of employees employed in firms with at least 10 employees, available at http://ukrstat.gov.ua/. e. Demographic statistics for Donetsk and Luhansk Oblasts include part of their NGCAs (to the extent possible); education, labor, employment, and income statistics cover only the government-controlled areas (GCAs). Appendix D Technical Aspects of the Simulation Model D.1 Variable and Parameter List Indexes: •• : Region-industry combination index •• : Alternative region-industry combination index •• : Time index Variables: •• : Probability of moving from to at time •• : Present discounted value in at time •• : Wage in at time •• : Population (working-age) in at time •• : Real output in at time •• : Supply of public goods in at time Parameters: •• : Moving cost from to •• : Discount factor •• : Productivity in at time •• and : Cobb-Douglas production function share parameters •• : Parameter determining elasticity of labor mobility Data: •• : Labor allocation at •• : Real wages at time Appendixes •• : Probability of moving from to at time . (We do not use the full matrix but rather the average probabilities of changing regions, changing sectors, and changing both. Thus, we need only three moments total for cal- ibration purposes.) 217 D.2 Main Equations Define the following operators for any variable : Flow equation (to use in steady state) The probability of moving from to at the initial steady state (time at ) is given as (D.1) where we define Flow equation (to use in transition) The probability of moving from to at time t is given as The Economics of Winning Hearts and Minds (D.2) Bellman equation The change in values is given as (D.3) Labor allocation equation 218 The population in at time is (D.4) Wage change equations Change of real output in nonpublic sectors is (D.5) where is the index of public sector associated with the region of . For exam- ple, if is Kyiv-manufacturing, then is Kyiv-Public. Change of real output in public sectors is (D.6) Changes of wages of the nonpublic and public sectors are, respectively, (D.7) and (D.8) D.3 Simulation Steps Calibration The moving cost function is parameterized as (D.9) where is the log of distance between regions of and . is set to zero if they are in the same region. If and are in the same oblast but in the GCA and the NGCA, respectively, we use half of the calculated log-distance, Appendixes assuming that moving inside an oblast was easier than moving between oblasts before the conflict. is 1 if and are associated with different sec- tors, such as manufacturing and services. is equal to 1 if and are associated with both different regions and different sectors, and is 0 otherwise. 219 We calibrate , , and to , and match the labor allocations, average probability of region change, average probability of industry change, and aver- age probability of changing both, in the data and in the simulations. We only use equations (D.1) and (D.4) for this process. The system is exactly identified since we calibrate 111 parameters (or variables) and match exactly 111 moments from the data. We set , , and . We find that , and . Note that we do not need to calibrate the produc- tivity parameter , since only the changes in productivity, that is, , matters. Characterizing the Conflict After the calibration process, we simulate conflict by changing the following variables: We reduce the productivity in government-controlled (GC) Donetsk and Luhansk by 25 percent and 13 percent, respectively. The productivity in non-government controlled (NGC) Donetsk and Luhansk decreases by 23 per- cent and 47 percent, respectively. We also double the moving cost from NGC to GC areas. These adjustments are implemented by changing and . During conflict, we do not allow growth of the public sector and limit the mobility from the public sectors of Donbas to other regions. We set the variables such that we roughly match the numbers of internally displaced persons (IDPs) in Ukraine through the conflict simulation. We have the wages and labor allocations— and , respectively—from the data. Consider a series of from to , and let us denote this series with . Using , we first calculate the flows, from to using equa- The Economics of Winning Hearts and Minds tion (D.2), the moving probabilities from the steady state solution and the change in moving cost . Then, with the initial labor allocation from data, we can calculate labor allocation from to , via equation (D.4). Then we can calculate wages for the same time horizon, via equations (D.7), (D.8), (D.5), and (D.6). Finally, using wages, , and steady state moving probability, , and the original itself, we can calculate the new updated . Therefore, we can consider the equations (D.2), (D.3), (D.4), (D.7), (D.8), (D.5), and (D.6) as a function, . We simply start with a guess of and update it using until we find a fixed point. 220 Reference IOM (International Organization for Migration). 2019. “National Monitoring System Report on the Situation of Internally Dis- placed Persons, June 2019.” Report of the IOM Mission in Ukraine, Kyiv. https://displacement.iom.int/system/tdf/reports/Ukraine_DTM_National%20Monitoring%20System%20Report_Round%20 14_June%202019.pdf?file=1&type=node&id=6707. Appendixes 221 S ince 2014, the armed conflict in Ukraine’s eastern provinces of Donetsk and Luhansk has led to more than 13,000 deaths and displaced about 1.5 million Ukrainians internally. It has also torn the region’s economic and social fabric in complex ways. Some of these effects (like physical destruction) are clearly identifiable. Others, includ- ing diminished social trust in key institutions, are harder to measure despite being equally important, if not more so. Recently, Ukrainian authorities embarked on producing a Strategy for the Economic Development of Donetsk and Luhansk Oblasts, following Presi- dent Volodymyr Zelenskyy’s “winning hearts and minds” approach. The strategy aims to implement large-scale economic initiatives and invest- ments “to ensure the sustainability of local communities, to create new jobs, and to fill local budgets”. This World Bank study considers the economic underpinnings of future economic recovery in eastern Ukraine. It analyzes the effectiveness of various policy instruments under different conflict scenarios to help prioritize interventions. The report first reviews the economic and social trends in Donbas before 2014 to identify structural challenges that may persist even after the end of the current conflict. Next, it studies the mechanisms through which the conflict has altered these challenges, focusing on three main channels—destruction, displacement, and disor- ganization. Finally, it employs a model-based approach with three cate- gories of policy actions (investments, transfers, and mobility cost reduc- tion) to evaluate future economic trends in Donbas and Ukraine as a whole. WITH CONTRIBUTIONS FROM