SOCIAL PROTECTION & JOBS DISCUSSION PAPER No. 2304 | JUNE 2023 Social Protection in a World of Crisis: Learning from the Response to the COVID-19 Pandemic in Eastern Europe and the South Caucasus Sarah Coll-Black, Cornelius von Lenthe, Stefanie Brodmann, William Shaw, Judith Sandford, Alejandro Gonzalez, and Jamele Rigolini This project is funded by the European Union. © 2023 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: +1 (202) 473 1000 Internet: www.worldbank.org 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 necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: +1 (202) 522 2625; e-mail: pubrights@worldbank.org. Abstract retro geometric background: © iStock.com/marigold_88 Project 41595 Social protection in a world of crisis: Learning from the response to the COVID-19 pandemic in Eastern Europe and the South Caucasus Sarah Coll-Black, Cornelius von Lenthe, Stefanie Brodmann, William Shaw, Judith Sandford, Alejandro Gonzalez, and Jamele Rigolini Abstract This paper explores the social protection response to the COVID-19 pandemic in Armenia, Azerbaijan, Georgia, Moldova and Ukraine to learn lessons on how to build the resilience of their social protection system. These countries made substantial efforts to address the most serious consequences of the pandemic, pragmatically harnessing existing programs to reach vulnerable groups, while also introducing innovations to fill gaps in the existing social protection system. Rigidities in administrative systems, complex eligibility criteria, as well as weaknesses in information systems, limited governments’ ability to quickly identify and reach those households that were most vulnerable to the impact of the pandemic with adequate support. These challenges strengthen the case for investment in crisis preparedness – most immediately by improving the functioning of social protection systems and setting out the design features and delivery systems to support a response to future covariate shocks. JEL Codes: H53, H55, H84, I3, O12, O52, O57 Key Words: Social Safety Nets, Resilience, Adaptive Social Protection, COVID-19, Social Protection, Social Assistance, Active Labor Market Programs, Social Insurance, Unemployment Benefits, Shocks Acknowledgements This paper has been prepared by Sarah Coll-Black, Cornelius von Lenthe, Stefanie Brodmann, William Shaw, Judith Sandford, Alejandro Gonzalez, and Jamele Rigolini. Much of the analysis relies on background notes for each country prepared by Susanna Karapetyan (Armenia), Isa Aliyev (Azerbaijan), David Okropiridze (Georgia), Sinchetru Alexandru, Olga Kupets (Moldova) and Katya Maynzyuk (Ukraine). Judith Sandford and Alejandro Gonzalez authored a set of notes on the response to the COVID-19 pandemic in Australia, Chile, Mexico and the United Kingdom that inform this chapter. The team is grateful for additional inputs and comments provided by Anastasiia Akulenko, Lucia Solbes Castro, Efsan Nas Ozden, Gozde Meseli Teague, Lucian Bucur Pop, Maddalena Honorati, Mirey Ovadiya, Renata Mayer Gukovas, Roberto Claudio Sormani, Roman Zhukovskyi, and Sirma Seker. This report was prepared under the guidance of Fadia Saadah (Regional Director, Human Development, Europe and Central Asia Region), Arup Banerji (Country Director, Eastern Europe), Sebastian Molineus (Country Director, South Caucasus), and Cem Mete (Practice Manager, Social Protection and Jobs Global Practice, Europe and Central Asia Region). It also benefitted from valuable feedback from Caryn Bredenkamp, Carolin Geginat, Mattia Makovec, Sarah Michael, Harry Patrinos, Mersedeh Tariverdi, William Wiseman and Levent Yener. The team would also like to thank representatives of the European Commission for useful feedback and inputs, in particular Hoa Binh Adjemian, Thibault Charlet and Corinne Deleu. The note was made possible by generous funding from DG NEAR’s Europe 2020 Trust Fund. Acronyms ALMPs Active Labor Market Programs AMD Armenian Dram ARM Armenia ASP Adaptive social protection AZE Azerbaijan AZN Azerbaijani Manat DOST Sustainable and Operative Social Protection Agency DRM Disaster risk management ECA Europe and Central Asia FB Family Benefit FLSEB Family Living Standards Enhancement Benefit GEL Georgian Lari GEO Georgia GMI Guaranteed minimum income HUS Housing Utility Subsidy IFE Emergency Family Income LRIS Last Resort Income Support MDA Moldova MDL Moldovan Lei MIS Management information system NEA National Employment Agency SARS Severe acute respiratory syndrome SESA State Employment Support Agency SMEs Small- and medium enterprises Social Protection Expenditure and Evaluation SPEED Database TSA Targeted Social Assistance UAH Ukrainian Hryvnia UI Unemployment Insurance UKR Ukraine USC Unified Social Contribution Unified Digital Application and Appointment sub- VEMTAS System WS Wage subsidy CONTENTS Abstract ........................................................................................................................................................... i Acknowledgements......................................................................................................................................... i Acronyms ........................................................................................................................................................ i Introduction ................................................................................................................................................... 1 Setting the stage: boosting resilience through shock-responsive social protection ..................................... 4 A brief review of foundational social protection programs ......................................................................... 10 The crisis and the emergency social protection response ........................................................................... 18 Adapting program design: responding to emerging needs and newly vulnerable populations .................. 27 Adapting the delivery chain: identification, enrollment, and provision to people in need ......................... 39 Better preparing for the next crisis: Building the resilience of the social protection system...................... 46 References ................................................................................................................................................... 52 Annex 1: Emergency social protection programs adopted during the pandemic ....................................... 59 List of Figures Figure 1: Four pillars of effective shock and crises responses................................................................................ 7 Figure 2: Modifications to social protection programs .......................................................................................... 8 Figure 3: The social protection delivery chain ........................................................................................................ 9 Figure 4: Social protection spending (% of GDP), countries in ECA, most recent year available ......................... 11 Figure 5: Employment by professional status, 15+ years ..................................................................................... 14 Figure 6: Coverage of social protection and social assistance, total and poorest quintile (Q1)........................... 14 Figure 7: Coverage and spending of poverty-targeted programs, total and poorest quintile (Q1) ..................... 16 Figure 8: Adequacy of social assistance and poverty-targeted programs for the poorest quintile ..................... 17 Figure 9: Percentage change in GDP between second quarter of 2019 and the second quarter of 2020 ........... 19 Figure 10: Estimated lost working hours in upper-middle-income ECA countries due to COVID-19 ................... 19 Figure 11: Additional financing allocated in 2020 to emergency social protection response.............................. 22 Figure 12: Approximate allocations of emergency resources in the study countries .......................................... 24 Figure 13: Simulation of proportional loss shock ................................................................................................. 38 Figure 14: Simulation of random loss shock ......................................................................................................... 38 Figure 15: Access to the internet, 2010 and 2020 ................................................................................................ 43 Figure 16: Access to computers, 2010 and 2020 .................................................................................................. 43 List of Tables Table 1: Select notable natural disasters in the study countries ................................................................... 2 Table 2: Summary description of main social protection schemes in the study countries ......................... 12 Table 3: Social protection response by type of worker and their employment status................................ 28 Table 4: Number of beneficiaries receiving poverty-targeted social assistance in 2020............................. 37 List of Boxes Box 1: Disaster Risk Profiles of the study countries ....................................................................................... 2 Box 2: Social protection systems ................................................................................................................... 5 Box 3: Disaster risk management (DRM) systems in the study countries ................................................... 13 Box 4: In Moldova, households driven into poverty by the pandemic differed from the pre-crisis poor ... 20 Box 5: Social care services during the COVID-19 pandemic......................................................................... 26 Box 6: The importance of design: simulating different response options in Moldova ................................ 38 Box 7: Ukraine’s Diya app facilitated applications for unemployment and sickness benefits..................... 42 Box 8: Facilitating access to an on-demand social assistance program: the case of Universal Credit ........ 45 Box 9: Rapidly identifying affected households through a social registry ................................................... 49 Box 10: Financing the social protection response to COVID-19: the case of the United Kingdom ............. 50 Introduction Social protection programs have played a central role in the response to the COVID-19 pandemic. Across the globe, countries harnessed social protection systems to provide much- needed financial support to households as they contended with job and income losses arising from public-health mandated lockdowns, a slowdown in global economic growth, and restrictions on travel. In many countries, the provision of social protection support to households was central to the overall strategy of ensuring that people stayed home to mitigate the spread of COVID-19 (New York Times 2020; Aminjonov and Bernard 2021). Social protection systems were also used to direct extraordinary support to households that were particularly affected by the pandemic, such as older people who faced enhanced risk of infection and thus needed basic services to be brought to their homes, or children whose schooling was switched from in-person to remote, with a commensurate increase in the time burden on families, often mothers. This paper explores the response to the COVID-19 pandemic and the associated economic impacts through the social protection systems in Eastern Europe and the South Caucasus: Armenia, Azerbaijan, Georgia, Moldova and Ukraine (henceforward “the study countries”). By looking at the social protection systems before the COVID-19 pandemic and considering what countries did in response to this large-scale crisis, we aim to glean lessons on how to improve the preparation of these systems for future crises. The paper is based on background papers from local experts, focus group discussions with social protection program recipients and social workers, a quantitative survey in one country, and some additional data and analysis (on expenditure data and simulations). We also draw on four case studies of the pandemic experience of upper-middle- and high-income countries Australia, Chile, Mexico and the United Kingdom, as well as regional and global reviews of the social protection response to the COVID-19 pandemic. Overall, the COVID-19 pandemic was a wakeup call for the study countries – as it was more broadly across the globe – on the urgent need to prepare for future shocks and disasters. The pandemic induced a massive economic shock that affected a large share of the global population. Despite the historical impact of influenza (such as the Spanish flu) and recent occurrences of other serious outbreaks of disease (including severe acute respiratory syndrome – SARS – and Ebola), countries all over the world were generally unprepared for the economic consequences of widespread illnesses that could only be contained, at least initially, by shutting down a major share of economic activity. The consequences for trade, output and government finances have underlined the vulnerability to disruptions of today’s technologically advanced, integrated global economy. Importantly, the COVID-19 pandemic 1 can be seen as one in a sequence of large-scale shocks, both global (such as the 2008 global financial crisis) and more localized. As seen in Box 1, the study countries are exposed to a range of natural disasters, which it is anticipated will become more frequent and severe with climate change. It reinforces the pressing need to prepare for future crises affecting large segments of national populations, alongside more localized shocks. This pressing need is vividly illustrated by the fact that even as economies and households recover from the COVID- 19 pandemic, the global economy is already facing the fall-out from Russia’s invasion of Ukraine. This conflict is not only affecting Ukraine; it is being felt regionally and globally because of disruption to supply chains, and its inflationary effect on food and energy prices, as well as the major refugee crisis overwhelming neighboring countries. Box 1: Disaster Risk Profiles of the study countries Countries in Eastern Europe and the South Caucasus regularly face the impact of a variety of natural disasters. Between 1990 and 2022, The countries experienced diverse natural disasters that affected more than 9.7 million people, causing economic damages in excess of US$ 7.7 billion. Table 1 shows some of the most notable natural disasters in the countries, and their damages: Table 1: Select notable natural disasters in the study countries Economic Number of Affected Date Type of disaster Loss deaths population ($ million) Ivano-Frankivsk region flood, 26 July 2008 38 224,725 1258 Ukraine January 2006 Extreme Winter, Ukraine 801 50,000 Ismayilli–Gobustan region flood, 16 April 2003 31,500 55 Azerbaijan Magnitude 4.8 Tbilisi earthquake, 25 April 2002 6 19,156 350 Georgia June 2000 Caucasus sub-region drought 993,000 400 24 August North region flood, Moldova 47 25,000 548 1994 Magnitude 7.0 Racha-Imereti 29 April 1991 100 100,000 10 earthquake, Georgia 10 March Adzharia region landslide, 98 2,500 423 1989 Georgia 12 December Magnitude 6.9 Spitak 25,000 1,642,000 14,200 1988 earthquake, Armenia 14 February Tbilisi region flood, Georgia 110 36,000 546 1987 Magnitude 4.2 Noyemberyan 18 July 1977 15,000 33 city earthquake, Armenia 2 In Armenia, Azerbaijan and Georgia, floods tend to be the most common natural disasters, while earthquakes have historically been the most serious. Since 1990, 29 floods, most of them in Georgia, have affected more than 20 million individuals. Earthquakes, however, have been the costliest form of natural disasters, causing damages in excess of $600 million since 1990. In 1988, the Armenian earthquake alone cost the lives of 25,000 Armenians. While Ukraine and Moldova are less affected by earthquakes, they are equally vulnerable to floods. Climate change is expected to lead to more severe, intense and frequent weather- related disasters, with important implications for the countries. Azerbaijan is a flood- prone area. In Georgia, the largest natural disaster occurred in 2015, when heavy rains triggered landslides. In Ukraine, droughts now occur, on average, once every three years. More than $2.6 billion in total damages has been caused by floods in Ukraine and Moldova and almost 2.5 million people have been affected. During the summer months, Ukraine was recently impacted by large wildfires, while Moldova has historically struggled with droughts, such as in 2000, 2007 and 2012. In the winter months, extreme temperatures regularly lead to deaths. Since 2000, more than 1,100 people lost their lives to extreme cold. Source: Guha-Sapir et al. 2022; World Bank 2009; USAID 2016. The growing frequency and severity of shocks highlights the importance of strengthening the resilience of social protection systems. A growing body of evidence shows that the poor are most affected by disasters, as long-lasting impacts undermine gains in human capital and poverty reduction (Hallegatte et al. 2017). In Europe and Central Asia (ECA) for instance, including the study countries, emerging evidence suggests that socio-economic status can determine how households cope with and recover from shocks, revealing a need for innovative policy responses, such as scalable or adaptive social protection (World Bank 2021b). Simulations of the distributional impacts of natural disasters show the detrimental impact on households with lower consumption levels. Around the world, countries are increasingly turning to their social protection systems to protect poor households and those vulnerable to falling into poverty from the negative effects of large-scale shocks (See, for example, Bowen et al. (2020)). Social protection has proven to be an effective means of providing direct support to poor and vulnerable households to help ensure their basic consumption and promote their human capital (World Bank 2018; Bastagli and Lowe 2021; Gentilini, Almenfi, and Dale 2022). Increasingly, these programs are being used to protect poor and vulnerable households from sudden losses of income and the rising costs of essential goods and services in the aftermath of crises, thereby helping them to withstand, manage and recover from shocks (Bowen et al. 2020). In many countries – particularly in 3 Africa and increasingly in South Asia and Latin America – social protection programs are becoming a pillar of the response to disasters and climate change (World Bank 2020e, 2022h). The response to the COVID-19 pandemic revealed shortcomings in the ability of social protection systems in the study countries to respond flexibly to shocks. The countries did make substantial efforts to address the most serious consequences of the pandemic, pragmatically harnessing existing programs to reach vulnerable groups, while also introducing innovations to fill gaps in the existing social protection system. However, rigidities in administrative systems, complex eligibility criteria, as well as weaknesses in information systems, limited governments’ ability to quickly identify and reach those households that were most vulnerable to the impact of the pandemic with adequate support.1 These challenges strengthen the case for investment in crisis preparedness – most immediately by improving the functioning of social protection systems and setting out the design features, delivery systems, information sources and institutional arrangements to support a response to future covariate shocks, be these climate-related events, earthquakes or pandemics. The experience of the study countries during the pandemic provides valuable lessons for preparing for future shocks and building the resilience of the social protection system. All the countries initiated emergency programs to reach those who lost incomes during the pandemic and many directed resources to groups who were particularly affected or vulnerable, such as older people and children. These responses involved – depending on the country – increasing the value of social assistance benefits; easing eligibility requirements to enter or remain in social assistance; extending assistance to new categories of beneficiaries; channeling public funds to the unemployed; subsidizing wages to retain jobs; and increasing benefits under social insurance programs, specifically pensions. Despite the magnitude of the response, in all countries groups were left uncovered, with the assistance funded through emergency social protection programs often being too little and arriving late. Reflecting on which populations were reached, the modifications to program design and the effectiveness of the delivery system offer important insights as countries prepare for the next crisis. Setting the stage: boosting resilience through shock-responsive social protection Social protection systems are designed to directly reduce poverty, while providing insurance against risks and promoting opportunities throughout the life cycle. By design, social protection systems help individuals and societies manage risks and volatility, protect against poverty and destitution, and facilitate access to economic opportunities. A range of 1 It is likely that available budgetary resources also limited the scale of the response. However, this is beyond the scope of this paper. 4 instruments across social assistance, social insurance, social care services and labor programs achieve these objectives, with some instruments contributing to more than one goal (see Box 2)(World Bank 2022c; Gentilini et al. 2012). Rooted in a robust and growing evidence base on the effectiveness of social protection programs for individuals, communities and societies (Bastagli et al. 2016; World Bank 2018; Moffitt 2014), countries across the globe have established and are expanding the coverage of social protection programs, including in the study countries (World Bank 2022e, 2022f, 2022g). Box 2: Social protection systems Social protection systems rest on four main pillars: social assistance, social care services, social insurance, and employment/labor market programs, which help families and individuals build resilience against events and shocks across the life cycle and build human capital.2 Social assistance is intended to protect people from falling into poverty and provide support at certain points in the life cycle or in response to particular vulnerabilities. It encompasses non-contributory (government-funded) programs, including non-contributory pensions (often called social pensions), and family and child cash benefits. Social care services support individuals and their families to improve their living conditions throughout the life cycle. Social insurance is intended to smooth income across the life cycle and protect people from shocks; it typically comprises benefits based on the length and level of individual contributions (old-age, disability and survivors’ pensions). Finally, employment and labor market programs are intended to improve the functioning of the labor market (through employment services), enhance labor supply (through training) and increase demand for labor (through subsidies or public works); the programs also seek to smooth income during unemployment (through unemployment insurance) or protect employment in the context of childbirth (through parental benefits). The COVID-19 pandemic threw into stark relief how rigidities within social protection systems can undermine these objectives – leaving many poor and vulnerable households exposed to shocks. Globally, despite decades of progress, social assistance programs often reach only a fraction of the people in the poorest quintile, though coverage rates tend to be higher in higher-income countries (World Bank 2022b). This limited coverage among the poorest leaves them vulnerable to covariate shocks, particularly when access to savings and insurance is low. Even as coverage increases, design features can cement the boundaries of these programs. These design choices may be driven by limited budgetary allocations, which prevent programs from expanding coverage to include additional beneficiaries or weaknesses in front-line implementation arrangements. 3 These rigidities also reflect limited information 2 Health insurance is not considered here, although it forms part of social insurance mechanisms. 3 For a discuss of such rigidities with respect to targeting, see: (World Bank 2022a) 5 about which households require support when a shock occurs, which can happen when early warning systems are incomplete or underutilized, and not integrated with or linked to social assistance databases (World Bank 2022a, 2022c). Together, these factors prevent programs from reaching households in need of support before, during or after a shock. These weaknesses in coverage, design and delivery increasingly appear to hold true for other types of social protection programs, such as unemployment insurance or active labor market programs. Across programs, low coverage, rigid delivery systems and limited information lead to a “missed middle”. This missed middle are those individuals in a country who tend to be workers without standard employment contracts, such as non-formal workers and increasingly gig economy workers, who are slightly better off than the poorest households but are neither covered by social assistance targeting the poor nor by social insurance schemes that are based on the existence of formalized employer-employee relationships and require regular monthly contributions. In the case of employment- or health-related shocks, this missed middle is often not protected by established social protection instruments (Guven, Jain, and Joubert 2021). In response, there is a growing focus on boosting resilience by strengthening the ability of social protection systems to respond to covariate shocks. This approach, called adaptive social protection (ASP) is an agenda within the broader field of social protection that focuses on preparing social protection programs and systems to better respond to covariate shocks, with the aim of building the resilience of poor and vulnerable households – before, during and after a shock. According to the World Bank, “adaptive social protection helps to build the resilience of poor and vulnerable households by investing in their capacity to prepare for, cope with, and adapt to shocks: protecting their wellbeing and ensuring that they do not fall into poverty or become trapped in poverty as a result of the impacts” (Bowen et al. 2020). While the risk profile of each country – as well as the contours of the existing social protection system – will determine the exact form ASP will take, investments are often concentrated in four key areas or building blocks (Figure 1): (1) design and governance; (2) delivery; (3) data and information; and (4) financing. Each of these four areas may be further separated into foundational and adaptive investments. Foundational investments are investments in social protection, disaster risk management or governance institutions that are not made with the express purpose of responding to shocks through the social protection system, but provide the context within which ASP can develop. These may range from a targeting system for a social assistance program that aims to reach the chronically poor, to a country’s early warning system, to a national ID system or the presence of local government offices across the country. These are then complemented by adaptive investments, which are conceived of and put into place with the express purpose of responding to covariate shocks through the social protection system (Bowen et al. 2020). 6 Figure 1: Four pillars of effective shock and crises responses Source: Rigolini et al. 2023, adapted from Bowen et al. 2020. This paper focuses on the design and delivery of social protection programs that enabled or hindered the response to the COVID-19 pandemic, in order to identify lessons for preparing for future shocks in the countries in Eastern Europe and the South Caucasus. To this end, in addition to the framework introduced in the synthesis paper (Figure 1), we also draw on a framework for shock-responsive social protection that sets out five ways in which social protection programs may be used to respond to large scale shocks: (i) design tweaks: small modifications to routine social protection programs; (ii) vertical expansion: temporary modification of program design to increase the value or duration of the benefit provided to existing beneficiaries; (iii) horizontal expansion: temporary modification of program design to expand the program to new beneficiaries; (iv) piggybacking: launching a new program in response to the crises using part of an existing social protection program, such as a national database or program staff; 4 and (v) alignment: when a new initiative is designed to resemble programs already in place, such as objectives, targeting method, or payment systems.5 Figure 2 illustrates vertical and horizontal expansions. 4 Importantly, in the shock-responsive social protection framework, piggybacking is when an emergency program – funded and most often delivered by humanitarian actors – uses the systems of a national program. In this chapter, we define “piggybacking”, as a government directing a new, temporary benefit with a specific objective for the beneficiaries of an existing program. 5 Again, in the shock-responsive framework, “alignment” generally applies to a situation in which a humanitarian program is delivered alongside an existing national program, in a manner that mirrors the national program in terms of objective, targeting populations, and benefit types and amounts. As a result, this aspect of the framework does not feature in the discussion below. 7 Figure 2: Modifications to social protection programs Source: Bowen et al. 2020. “Adaptive” modifications to programs – those that build the resilience of the social protection system – are often conditioned by existing delivery systems, as is the effectiveness of programs for achieving their objectives. As discussed above, the form ASP takes in a country is shaped by its social protection system, in terms of the existing programs, their objectives, coverage and design parameters, but also the delivery systems that they rely on. That is, the foundational investments. Drawing on the framework for delivery chains, which is set out in the Sourcebook on the Foundations of Social Protection Delivery Systems (Lindert et al. 2020), social protection delivery systems are the “operating environment for implementing social protection benefits and services.” As detailed in Figure 3, this framework is built around the core steps in the delivery chain. People (applicants and beneficiaries) and institutions (service providers, local governments, and central ministries) interact along the delivery chain, in ways that are enabled by communications, information systems, technology and other factors. As has been discussed extensively elsewhere, weak implementation – that is, poor quality delivery systems at any point in the delivery chain – can lead to reduced impact, errors of inclusion or exclusion, and leakage (Lindert et al. 2020). These weaknesses can also negatively affect applicants or potential beneficiaries, by increasing the financial and opportunity costs of applying, or exposure to harassment or fraud. When a crisis occurs, not only can the systems themselves be affected (for example if communication systems fail or roads are impassable), but their design and execution can also determine how quickly and effectively they can be deployed to identify and reach additional people negatively affected by the shocks (Smith and Bowen 2020). 8 Figure 3: The social protection delivery chain Source: Lindert et al. 2020. This focus on program design and delivery is extended across social protection instruments, in recognition of the level of maturity of the foundational systems in the study countries. While adaptive social protection is framed around the social protection system’s responsiveness to shocks, its application has, by and large, been limited to a focus on social assistance. 6 In contrast, the response to the COVID-19 pandemic among the countries has been through all the pillars of the social protection system, as countries harnessed a range of programs to meet the needs of diverse population groups. In recognition of this, the sections that follow apply the framework for the shock-responsive social protection system to selected programs across the pillars of social assistance, social insurance and labor. The rest of the paper is structured as follows: the first section describes the foundational social protection systems in the study countries before the COVID-19 pandemic. This includes the broad parameters of the social protection system, including overall levels of financing, coverage and mix of programs. We then briefly discuss the COVID-19 pandemic and its effects across the study countries in the early months of the pandemic, before considering the initial phase of the pandemic response through social protection systems, and specifically the emergency social protection programs that each country adopted to finance additional measures through social protection systems, in terms of level of financing, range of programs used, and populations reached. The next section reviews program design features across social assistance and labor programs within the foundational social protection system and the emergency program. This is followed by consideration of the delivery systems, focusing specifically on poverty-targeted social assistance (called Last Resort Income Support 6 See Bowen et al. 2015 or Oxford Policy Management 2015. This reflects the roots of adaptive social protection within certain low-income countries. 9 Programs). The final section concludes by returning to the framework set out in Figure 1 to draw out lessons under each of the pillars for how to strengthen the ability of the system to respond to future shocks, with the aim of boosting the resilience of households. A brief review of foundational social protection programs While the study countries all have established social protection systems, the mix of programs differs greatly, driven by varying levels of spending across pillars of the system. According to the definition of social protection set out Box 2, the study countries had established social protection systems, with varying levels of spending before the pandemic (Figure 4). As described in Box 3, these countries also had disaster risk management (DRM) systems in place; these are not considered in this chapter given its focus on the response to the COVID- 19 pandemic, despite their contribution to ASP. Before COVID-19, social protection spending was largely dominated by spending on social insurance, with all the countries, except Georgia, spending more on social insurance than on social assistance. 7 Among countries with social insurance spending, this ranged from 3.5 percent in Armenia to 10.7 percent of GDP in Ukraine, and tended to largely comprise spending on old-age pensions. In contrast, spending on social assistance tended to be lower than social insurance, although levels varied across countries. The average country in Europe and Central Asia (ECA) spent 1.8 percent of GDP on social assistance, with two study countries – Georgia and Ukraine – topping the list, spending the equivalent of 5.7 percent and 4.7 percent of GDP, respectively. Spending on social care services and labor programs across ECA countries is very low. Table 2 describes the main programs comprising the social protection systems in each country. 7 In Georgia, relatively low spending levels on social protection are mainly driven by modest spending on pensions relative to other countries, while social assistance spending is in line with ECA averages, as discussed further in the paragraph below. The pension system relies in particular on a social pension which, as it is non- contributory, is counted as social assistance. A contributory pension was rolled out in 2019; however its impact will likely only be seen in future decades. 10 Figure 4: Social protection spending (% of GDP), countries in ECA, most recent year available 18.0% Social Protection Spending (% of GDP) 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% GEO - 2018 MDA - 2017 EST - 2017 ALB - 2019 CZE - 2017 SVK - 2017 KSV - 2017 BGR - 2017 UKR - 2017 ECA Average** HRV - 2014 POL - 2012 ARM - 2019 AZE - 2019 LTU - 2016 MNE - 2019 BLR - 2017 SRB - 2019 RUS - 2016 UZB - 2017 TUR - 2016 MKD - 2019 TJK - 2017 KGZ - 2017 KAZ - 2017 LVA - 2017 BIH - 2018 HUNG - 2016 ROU - 2017 Labor market programs (%) Social assistance (%) Social insurance (%) Social care services (%) Total Source: Social Protection Evaluation and Expenditures Database (SPEED) (2022). Note: Georgia’s social protection expenditures are classified to consist solely of social assistance, as the only existing pension is a social pension. 11 Table 2: Summary description of main social protection schemes in the study countries Social protection Armenia Azerbaijan Georgia Moldova Ukraine category Description Exp. Description Exp. Description Exp. Description Exp. Description Exp. The Family Living Standards The Guaranteed Minimum Enhancement Benefits (FLSEB) The Ajutor Social provides a Income (GMI) is the main provide three cash benefits: i) Targeted Social Assistance guaranteed minimum income program to support a minimum Family Benefit (FB) is provided (TSA) is the main poverty- to households below a set level of income. It is Targeted Social Assistance (TSA) to families with children aged 0- targeted program. The income and proxy means test implemented as a cash payment is a monthly allowance targeted 18 registered in the FB system government reformed TSA to score. The benefit amount is to households, and the benefit is to low-income households, and assigned a vulnerability make the eligibility criteria the difference between the equal to the difference between defined by those whose average score above the threshold (57% more stringent and objective, guaranteed minimum income the guaranteed level of formal monthly income is below of MW); ii) Social Benefit (SB) is introduced a scheme of and the actual household subsistence minimum (21% of the total requirement for each Last-resort provided to families with no differentiated levels of benefits income. The current threshold the subsistence minimum for the household member. The children aged 0-18 registered in for the TSA, and introduced a is at about half the subsistence able-bodied, 85% for children social the FB system and assigned a 0.51% amount of social assistance is 0.11% Child Benefit Program, 0.56% minimum. Jobless individuals in 0.45% and 100% for disabled) and the 0.40% assistance then determined as the vulnerability score above a applicable for children aged 15 the household must be household’s income. For difference between the average defined threshold 32.7% of or less, and delivered alongside registered as unemployed with individuals of working age monthly income of the MW) and iii) Quarterly TSA transfers. Eligibility is some exceptions. Households without disabilities and not SA household and the sum of the emergency assistance targets designed for households that must apply for the benefit and caring for anyone, an assumed requirements for each member, persons/families registered in are predicted to live on less re-certify every 12 months. income is imputed to account for and is paid every month to a the system and assigned a than 65% of the subsistence Households with no members informal income. The benefit is designated bank account. vulnerability score below the minimum is targeted using a of working age capable of work also topped up for families with threshold, and finding proxy-means test. are granted the benefit for 24 children. The maximum benefit is themselves in an emergency months. set at 75% of the subsistence situation (32.7% of MW). minimum for the family. Other social assistance largely Other social assistance Other social assistance consists of a universal old-age Other social assistance Most social assistance spending Other social programs include, among programs include disability pension, which provides programs include a disability is absorbed by two categorical others, a disability benefit, a benefits, war-veteran benefits, relatively small benefits, assistance child-birth lump sum benefit, a 0.58% family and child allowance, 0.39% currently equal to 17% of the 5.16% benefit, war-time benefits, as 0.57% programs, the Housing Utility 3.74% programs well as birth grants and child Subsidy (HUS) and the childcare benefit and a social presidential pensions as well as average wage, for all women benefits for uninsured persons. untargeted childbirth grant. pension. small supplementary benefits. over the age of 60 and men over the age of 65. The largest proportion of Three types of contributory In 2019, a new contributory pensions is paid out through The elderly are well covered Following a 2017 reform of its pensions are offered: old-age pension scheme was the regular old-age pension, with pension benefits, even pension system, pensions are pension, disability pension and introduced for formal and self- followed by a general disability though the benefits are estimated to be sustained at survivors’ pension. Coverage of employed workers. Pensions pension. The pension system is 3.56% old-age pension is relatively 4.23% Contributory pensions are still 0% relatively low. Recent cuts to 5.95% close to 27% of the average 10.15% supplemented by a survivors’ the social security contribution wage. Significant transfers are low, but reforms in recent years at their onset and will show pension, an occupational injury rate will likely lead to required to cover the deficit of SI have improved the adequacy of their impact in at least one or pension and pensions for significant deficits. the fund. payments. two decades. service personnel. Contributory Currently, the cumulative Contributory paid maternity parental Paid maternity and sick leave A contributory maternity leave is 100% of average monthly duration of paid maternity and are available. Paid paternity 0.20% benefit provides women with <0.1% N/A 0% 0.86% income, available for 126 0.20% benefits and leave was introduced in 2020. 126 days of leave. parental leave available calendar days paid for by the sick leave exceeds three years. State Social Security Fund. Most expenditure goes toward ALMPs are not provided at Active Labor wage subsidies, training and a ALMPs include vocational large scale. The responsible Most of the small ALMP financial Market training, self-employment resources are allocated to provision to support initiating agency, SESA, established in ALMPs are not provided at <0.1% programs, and employment <0.1% <0.1% <0.1% vocational training and employer <0.1% Programs livestock breeding. Only 3% of services delivered through the 2020, is at its early stages of large scale. compensation (USC), followed by (ALMPs) registered jobseekers benefit development and coverage of State Employment Service (SES). public and temporary works. from ALMPs. services is limited. L Unemployment insurance was Unemployment benefits can be M Individuals who contributed to introduced in 2018. Individuals paid out to anyone registered as the social insurance system for are eligible if they have paid unemployed. A so-called partial Unemployment The contributory a minimum of 12 months are insurance for at least 3 years There is no contributory unemployment benefit pays unemployment benefit was 0% <0.1% 0% eligible for benefits between <0.1% 0.23% benefits abolished in 2013. prior to the termination of their unemployment benefit. 40% and 50% of their previous parts of wages if employees are employment contract and are forced to reduce working time average income, capped at the registered with the SES as due to the suspension average monthly salary. unemployed. (reduction) of production. Most recent total spending 2019 4.6% 2019* 4.75% 2018 5.72% 2017 7.86% 2017 14.74% Source: World Bank Social Protection Background Notes. Note: Exp. Shows expenditure as % of GDP, are simulated for Azerbaijan and do not include social services, unlike in Figure 4. SA = Social Assistance, SI = Social Insurance, LM = Labor Market Programs, MW = Minimum Wage 12 Box 3: Disaster risk management (DRM) systems in the study countries Adaptive social protection is inherently multi-sectoral, drawing together the policies, programs and systems that underpin the social protection and DRM sectors, in addition to broader investments in governance, foundational ID systems and other areas. In this paper, we focus exclusively on social protection systems. And yet, any move towards adaptive social protection in the study countries will require complementary investments in the DRM sector that may be foundational, with the aim of ensuring the basic functionality of the DRM system, or adaptive: that is, made with the explicit objective of establishing an adaptive social protection system. Recent DRM and climate resilience analysis and adaptive social protection country case studies suggest that the various DRM systems were primarily geared towards climate and natural disasters, rather than infectious diseases. Furthermore, very little attention was paid to the potential role of the social protection sector (except in Ukraine), and therefore there was not enough planning for how the social protection sector could be integrated into any disaster response. The study countries were not unique in this regard. According to the 2019 Global Assessment Report on Disaster Risk Reduction, most of the DRM systems around the world had a “traditional” set up before the pandemic. This traditional set up can be characterized as (i) being narrowly focused on climate and natural disasters, leaving aside other types of covariate shocks (including infectious diseases); (ii) on paper having an integrated approach to DRM, but in practice being geared towards emergency preparedness and response and with minimum provisions for disaster prevention and long-term resilience; (iii) suffering from limited integration and articulation mechanisms with other sectors (such as health and social protection); (iv) being heavily centralized with limited devolution of decision-making capacities and funds to local governments; and (v) being poorly funded with limited use of disaster risk financial instruments. Against this backdrop, improving the resilience of social protection delivery systems will necessitate investments to modernize the DRM systems as well. Source: World Bank 2020d, 2020f; UNDRR 2019. Social insurance is characterized by broad pension coverage and high social security contributions, with limited access to other contributory-based programs. Old-age pensions and disability pensions together cover more than 40 percent of the entire population in all the countries, except for Georgia, where a universal publicly financed pension provides 13 support in old age.8 Given high levels of informality, historically low statutory retirement ages, emigration and demographic characteristics, support ratios for old-age pensions have been low (Huitfeldt 2020). Until reforms in recent years, this meant that high social security contributions were needed to finance the broad coverage of pensions. All countries had weak links between the level of contributions and the size of pension benefits, which in turn acted as a disincentive to formalization (Figure 5)(Huitfeldt 2020). Other programs that tend to be contribution-based, such as parental leave or sick leave, are either non-existent or have very low coverage, with few exceptions such as Ukraine’s unemployment benefit and Moldova’s maternity leave. Figure 5: Employment by professional status, Figure 6: Coverage of social protection and 15+ years social assistance, total and poorest quintile (Q1) 120 Azerbaijan 99 98 97 100 79 Georgia 80 72777176 77 65 65 60 65 58 58 56 Armenia 60 48 Moldova 40 29 27 22 Ukraine 20 0 EU GEO-2019 UKR-2018 ARM-2019 MDA-2018 AZE-2015 0% 20% 40% 60% 80% 100% Social Protection Total Social Protection Q1 Self-employed and family workers Employees Social Assistance Total Social Assistance Q1 Source: Eurostat 2022. Note: In Armenia for persons Source: Social Protection Evaluation and Expenditures aged 15 – 75 years. Database (SPEED) (2022). Note: Coverage is defined as the number of individuals in the group who live in a household where at least one member receives the transfer divided by the number of individuals in the group. In contrast, coverage of social assistance varies greatly, reflecting a wide mix of programs and levels of funding. Coverage of social assistance ranges from a high of 71 percent of the population in Ukraine to a low of 22 percent in Armenia, based on the most recently available data, as seen in Figure 6. In all countries, less than half of social assistance expenditure is channeled to programs targeting the poor. While all countries have poverty-targeted programs, including last-resort income support programs, most spending tends to be allocated to programs that are categorically targeted, such as disability benefits, birth grants 8 That is, a social pension. 14 or social pensions, and to utility subsidies. Georgia, for example, has the second highest social assistance coverage study countries at 65 percent through its universal social pension, on which almost two-thirds of its social assistance budget is spent. In Ukraine, two-thirds of social assistance spending in 2017 went to two categorical programs: the Housing and Utility Subsidy (HUS) and an untargeted Birth Grant, which directly benefited an estimated 50 percent and 4 percent of the population, respectively. Similarly, in Azerbaijan and Moldova categorical programs, such as war-veteran benefits or birth grants, account for high shares of social assistance spending, which are, however, small in absolute terms relative to those in Ukraine and Georgia (World Bank 2022i). Spending on poverty-targeted social assistance programs is low and only covers a marginal proportion of the poor population. While levels of spending on programs that explicitly target the poor differ, all the countries spend less than half of their social assistance expenditure on them. In 2019 Azerbaijan spent the equivalent of 0.11 percent of GDP on its poverty-targeted program (TSA) and Ukraine – despite spending more than the equivalent of 4 percent of GDP on social assistance – spent 0.4 percent of GDP on its Guaranteed Minimum Income (GMI). As seen in Figure 7, coverage of those who are poor is strongly correlated with spending, but none of the programs cover more than half the population in the poorest quintile. Nevertheless, coverage of the poorest quintile tends to be higher than in comparable countries with available data, such as Albania (20 percent), Romania (16.7 percent) or Serbia (10.2 percent). This relatively higher coverage coupled with low spending leads to low benefit levels and adequacy, as elaborated on further below. 15 Figure 7: Coverage and spending of poverty-targeted programs, total and poorest quintile (Q1) 45.0% 0.60% 40.0% 0.50% 35.0% 30.0% 0.40% Coverage % of GDP 25.0% 0.30% 20.0% 15.0% 0.20% 10.0% 0.10% 5.0% 0.0% 0.00% AZE - 2015 UKR - 2018 MDV - 2018 ARM - 2019 GEO - 2018 Coverage total population (LH) Coverage Q1 (LH) Expenditure (RH) Source: Social Protection Evaluation and Expenditures Database (SPEED) (2022). Note: Coverage is derived from each country’s Household Budget Survey (HBS) for the year stated. Expenditure is derived from administrative data and may thus be more up to date than the indicated survey year. The poverty-targeted programs in the shown countries are: AZE - TSA; UKR - GMI; MDV - Ajutor Social; GEO - TSA; ARM - FLSEB. The adequacy of social assistance benefits is uniformly low. Social assistance coverage is highest in Ukraine (71 percent), but per capita benefits remain very low – equivalent to only 7.4 percent of household consumption. International evidence suggests that low per capita benefit levels limit the poverty reduction impacts of social assistance (See, for example, Bastagli et al. 2016). Social assistance benefit levels are higher in the other countries, but only in Georgia do they equate to more than 50 percent of the average consumption of the poorest quintile, though the adequacy of the poverty-targeted program TSA is also low (Figure 8). 16 Figure 8: Adequacy of social assistance and poverty-targeted programs for the poorest quintile 60% 54% 49% 50% (% of total consumption) 42% 40% 37% 36% Adequacy 30% 28% 23% 23% 19% 20% 10% 7% 6% 0% ECA Average GEO-2018 AZE-2015 ARM-2019 MDA-2018 UKR-2018 Social Assistance Poverty-Targeted Program Source: World Bank 2022i. Note: Latest year available. Adequacy is the mean transfer amount to a group as a share of the total household consumption aggregate of beneficiaries in that group. The coverage of labor programs in the study countries is also very low, with unemployment insurance systems present in only three of the five countries. Active and passive labor market programs can provide protection to people losing their jobs, or facilitate the transition into new or more productive jobs. Spending on active labor market programs (ALMPs) is low in all the countries, and the share of registered jobseekers who participate in ALMPs is equally low. Only Ukraine provides a large proportion of its registered jobseekers with ALMPs, mostly vocational training. In Armenia, only 3 percent of registered jobseekers are covered by ALMPs. Prior to the pandemic, Armenia, Georgia, Moldova, and Ukraine had functional wage subsidies, though most at very small scale. In addition, both Georgia and Armenia did not have established unemployment insurance. The coverage and generosity of benefits, and, ultimately, the flexibility of the social protection system was the foundation for the response to COVID-19 among the study countries. When COVID-19 hit, the study countries had well-established social protection systems, which enjoyed, in some cases, significant funding. However, these were mostly dominated by contribution-based social insurance programs to secure income in old age, which tend to be characterized by broad coverage, low adequacy and few contributors. The coverage of social assistance programs varied greatly across the countries, reflecting the differing mix of programs, their size, adequacy and objectives. In many instances, these programs were categorically targeted, with relatively more limited reach to those who were 17 poor. These rates of coverage – coupled with continued high rates of informality, as discussed in the sections below – limited the ability of the systems to provide immediate protection to poor and vulnerable households badly affected by the COVID-19 pandemic. In addition, the generally low value of the benefit paid to households through the social assistance systems left many poor and vulnerable households with little capacity to cope with the direct impact of the pandemic. These features of the social protection system will shape the response to future crises and, for this reason, the interaction of these features and the emergency social protection response to COVID-19 are described in more detail in the sections below. The crisis and the emergency social protection response The COVID-19 pandemic affected the national economies and household incomes of the study countries in several ways. Public health measures (such as lockdowns), combined with households being urged to reduce social contact, resulted in fewer people working and significant reductions in demand in certain sectors. Lockdowns in other countries and restrictions to global travel reduced opportunities for migrant labor and resulting remittances, and devastated the tourist industry (World Bank 2020a). COVID-19 infections increased absenteeism both by those who were off sick and by family members in some countries who were also required to self-isolate. 9 As a result, the economic impact of the COVID-19 pandemic crisis was initially severe. 10 GDP fell in all the study countries in the second quarter of 2020, and for the year as a whole it declined by 4.7 percent on average across the region (Figure 9). Unemployment rates increased by an average of 1 percentage point. While all countries saw simultaneous reductions in labor force participation and hours worked, the effects were greater and lengthier for women than for men (Figure 10). Evidence from other countries suggests that non-essential and informal sectors were more likely to become unemployed than those in the formal sector (Viollaz et al. 2022). Similarly, evidence suggests that formal firms in Georgia were able to return to increased sales and re-hire workers quicker than informal firms (Hatayama et al. 2021). 9 Working hours in 2020Q2 upper-middle income countries in Europe and Central Asia are estimated to have reduced by almost 21 percent, with that drop having been almost 3 percentage points higher for women than for men. 10 This section focuses on the initial effects of the COVID-19 pandemic in 2020, given the focus on the social protection response in the first part of the pandemic. It does not consider more recent evolutions in economic growth or poverty reduction. 18 Figure 9: Percentage change in GDP Figure 10: Estimated lost working hours in between second quarter of 2019 and the upper-middle-income ECA countries due to second quarter of 2020 COVID-19 25 20 0% 15 -2% 10 -4% -3% -6% 5 -8% 0 -10% -12% -11% -11.2% -14% -14% -13% Total Male Female -16% -15% Source: National Statistical Offices and Eurostat Source: ILO 2023. Note: Percentages are relative to (Indicator: TEINA011) baseline. The ILO notes that these estimates are subject to substantial uncertainty. There are 17 upper middle-income countries in ECA: Albania, Armenia, Azerbaijan, Bulgaria, Bosnia and Herzegovina, Belarus, Georgia, Kazakhstan, Moldova, North Macedonia, Montenegro, Romania, Russian Federation, Serbia, Turkmenistan, Turkey and Kosovo Increased return migration and reduced remittances aggravated the negative economic impact on households. The study countries are a significant source of economic migrants, who mainly travel to the European Union and Russia for work. For example, one in four Armenians live abroad, while around 17 percent of the working-age population of Moldova work abroad (International Labour Organization 2017). As a consequence of the travel restrictions imposed in response to the COVID-19 pandemic, 50 percent of Armenian migrants are estimated to have been unable to go to Russia for work (Honorati, Yi, and Choi 2020). In addition, as part of increased return migration and disruptions to the global economy, remittances fell by an average of 18 percent in the first quarter of 2020 and 16 percent in the second, compared to the 2019 annual average.11 The pandemic increased poverty headcount rates, although data are scarce in some countries. In Georgia, the proportion of people living on less than US$6.85 PPP per day12 increased by four percentage points to 58.3 percent of the population. Similar poverty increases were observed in other countries, with the international poverty rate increasing by 2.4 percentage points in Armenia and national poverty rates increasing by 1.6 percentage 11 According to balance of payments data from national central banks. 12 The international poverty line for upper-middle income countries. 19 points in Moldova, respectively (World Bank 2023). Only pre-war Ukraine was estimated to have seen its trend of declining poverty rates continuing, although this has since been disrupted by the ongoing conflict. There is some indication that those falling into poverty because of the pandemic differed significantly from those who were already poor. In the study countries, as in many developing countries, rural areas, and the agricultural sector account for a large proportion of households living under the poverty line. However, most of the households where incomes fell under the poverty line because of the pandemic were likely in urban areas. Simulations suggest that the urban share of newly impoverished households as a result of COVID-19 was 64 percent in Azerbaijan, 64 percent in Ukraine (although the urban share of the pre-crisis poor was only slightly lower) and 70 percent in Georgia.13 Consistent with the predicted significant impact on urban areas, the limits to mobility and collapse in trade as a result of the pandemic tended to affect industry and services more than they did agriculture. It is likely that few of the newly poor received social assistance before the pandemic. The differences between the newly poor and the existing poor in Moldova are illustrated in Box 4. In addition to increases in the poverty rate, those already impoverished before COVID-19 likely fell deeper into poverty because of the crisis. The effects of the COVID-19 pandemic were highly regressive, with lower-income households facing higher welfare shocks as a share of their consumption. For example, consumption among Georgian households in the lowest consumption quintile was estimated to have fallen by between 20 percent and 60 percent, whereas the percentage losses gradually decreased in the upper quintiles (World Bank 2021c). Similarly, the average rate of consumption decline in Azerbaijan among the bottom 20 percent of the population was estimated at over 20 percent, almost twice the mean estimated percentage change of around 10 percent (World Bank 2020c). 13 These simulations are based on survey data and use various macroeconomic forecasts for economic growth, labor market outcomes and remittances to estimate a counterfactual scenario, and use microsimulations with estimated income losses by subsector employment losses, remittance losses and assumed reduction in agricultural sales to estimate multiple shock scenarios. For more details see World Bank 2021c, 2020c, 2020b. 20 Box 4: In Moldova, households driven into poverty by the pandemic differed from the pre-crisis poor Simulations of the effects of the COVID-19 pandemic in Moldova show that employed individuals comprise a higher proportion of the new poor than the existing poor (82 percent versus 75 percent) and they tended to be concentrated in the services sector (47 percent), the agriculture sector (34 percent) and the industry sector (18 percent). By contrast, close to three quarters of the existing poor are concentrated in agriculture. The concentration of the new poor in the industry and services sector is largely consistent with these sectors experiencing the greater declines in employment during the crisis (5.2 and 4.7 percentage points, respectively). Close to a third of the new poor are concentrated in the agriculture sector. However, this concentration may also be due to the confounding effects of the drought that coincided with the harvesting season in quarter three of 2020. The new poor are also more likely to be employees (69 percent) and less likely to be self- employed (30 percent) than the existing poor, who are more likely to be self-employed (62 percent) and less likely to be employees (35 percent). The difference in employment types between the new and existing poor largely stems from the tendency for self- employment in agriculture, while work in industries and services is mostly employment- based. Close to 20 percent of the new poor were concentrated in the Chisinau region, compared with less than 3 percent of the preexisting poor. The difference in the geographic concentration of the new and existing poor in the capital region stems from geographic segregation of industries, whereby 35 percent of industrial jobs and 43 percent of service sector jobs are concentrated in Chisinau region. Source: World Bank 2020e Countries quickly adopted emergency programs that channeled support through the full range of social protection programs All countries launched dedicated social protection responses shortly after the pandemic hit, although these were relatively modest. By March 2020, it was evident that the COVID-19 pandemic would depress household incomes due to the drop in global demand, supply interruptions, and restrictions on mobility required to limit the spread of the virus. Measures authorizing emergency spending were passed in Armenia (March 26), Azerbaijan (April 4), 21 Georgia (May 4), 14 Moldova (March 13) and Ukraine (March 12). In most of the countries, assistance was provided to firms and individuals shortly after these authorizations were adopted. While this emergency financing was authorized quickly, in general, the amount allocated to social protection was modest (see Figure 11 below) when compared with routine social protection spending (see Figure 4 above)(World Bank 2020b, 2020g). Most countries financed these emergency packages through budgetary reallocations and increased public debt, partly generated through the issuance of large bonds or increased financing from international financial institutions. 15 Only Azerbaijan was largely able to finance its fiscal packages by drawing down assets from its State Oil Fund in addition to reallocating spending. Public central government debt as a share of GDP increased by almost 10 percentage points or more between 2019 and 2021 in all the countries except for Azerbaijan (IMF 2020, 2021a, 2021b, 2021c, 2022a, 2022b). Figure 11: Additional financing allocated in 2020 to emergency social protection response 2.00% Social Proteciton Spending (% of GDP) 1.80% 1.60% 1.40% 1.20% 1.00% 0.80% 0.60% 0.40% 0.20% 0.00% Armenia Azerbaijan Georgia Ukraine Moldova Sources: Gentilini, Almenfi, and Dale 2022, IMF 2022, IMF Policy Tracker, Background Notes and Annual Report of the Ministry of Finance Azerbaijan. Note: Spending includes 2020 spending on social insurance, social assistance, labor market programs or social services as part of emergency measures enacted by governments in response to the COVID-19 pandemic as a proportion of 2020 GDP. Funding from the regular government budgets also contributed to the COVID-19 response in some cases; however, reallocation of existing spending is not considered here. Moldova’s expenditure is an estimate based on IMF’s definition of social protection spending, which may not align with the other spending data. 14 On March 23, the government implemented a decree simplifying administration and disbursement of transfers under the Targeted Social Assistance Program. 15 These emergency packages tended to encompass spending not just on social protection but on multiple sectors, such as health, education, and other areas. 22 Countries responded through social assistance, social insurance, and labor programs; however, irrespective of the instrument, the additional support tended to be short-term. Social protection responses included expanding existing programs vertically and horizontally, design tweaks, and “piggybacking” the introduction of new programs on existing systems. More specifically, higher payments were made to beneficiaries of existing programs (vertical expansion); eligibility criteria were modified (particularly with respect to the requirement to have paid contributions for unemployment insurance; horizontal expansion); and new programs were developed and rolled out through existing programs (new payments to vulnerable groups) or using existing information or delivery systems (payments to unemployed workers based on tax authority data). Instruments included social assistance (payments to the poor and vulnerable, child allowances, food assistance, and payments to cover expenses such as utilities and tuition), social insurance 16 (pensions and contributory leave), labor market programs (unemployment benefits, wage subsidies or other ALMPs), fuel subsidies and social care services. Annex 1 provides an overview of the response in each country in 2020. However, this initial response to the COVID-19 pandemic was largely time- bound, and only lasted for the opening months of the pandemic.17 The allocation of emergency spending across pillars of the social protection system differed significantly across countries. Wage subsidies featured importantly in most of the countries, accounting for almost 60 percent of emergency spending in Armenia, and 17 percent in Ukraine (Figure 12). This is perhaps unsurprising given the widespread concern about job losses in the early days and weeks of the pandemic. 18 Among others, Ukraine allocated a relatively small proportion of the emergency spending to social assistance, while Georgia devoted the bulk of emergency spending to this purpose. Payments through the pension systems were only an important part of the emergency packages in Azerbaijan and Ukraine. Unemployment benefits varied from 14 percent to 26 percent of emergency spending in Azerbaijan, Georgia and Ukraine. Figure 12 presents the allocations of emergency spending across these countries. 16 Although social insurance instruments were used, these were financed or heavily subsidized by additional government funding. 17 Countries introduced new initiatives or extended initiatives into 2022. These are not considered here in this note, which focuses on the initial response to the COVID-19 pandemic. 18 For a more detailed look at the use and rationale behind wage subsidies and job retention schemes in response to COVID-19, see Ando et al. 2022 for a case study of EU countries’ responses. 23 Figure 12: Approximate allocations of emergency resources in the study countries 100% Allocation of emergency resources 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Armenia Azerbaijan Georgia Ukraine Social Assistance (other than Utility) Pensions Unemployment Benefits Wage Subsidies (WS) ALMPs (other than WS) + Tax Rebates Utility Subsidies Sources: Gentilini et al. 2020, IMF Policy Tracker, Background Notes and Annual Report of the Ministry of Finance of Azerbaijan. Notes: Social assistance includes any direct payment which is not based on the level of previous social security contributions, including payments to unemployed workers. These are approximations based on the description given in the emergency packages, and are broad categorizations. In Armenia, figures are based on actual spending using administrative data. Armenia did provide some assistance to unemployed workers, but these are grouped with social assistance because they were not based on contributions. Moldova is excluded due to a lack of disaggregated or detailed data. These emergency provisions complemented the protection afforded to households through existing social protection programs. The emergency response to the COVID-19 pandemic was implemented through or alongside the existing social protection systems in the five countries. Despite limitations in the coverage or generosity of benefits, all existing beneficiaries of social protection programs continued to receive support throughout the pandemic to help them meet their basic needs. As a result, the social protection system mitigated the impact of the pandemic on households through the combined effect of existing programs and the emergency measures. In Georgia, for example, the share of households in the bottom quintile receiving Targeted Social Assistance (TSA) or emergency-related transfers is estimated to have increased from 39.5 percent in 2018 to 62.6 percent in the last quarter of 2020.19 Additionally, in some countries, ongoing reforms to social protection systems had additional effects. For example, the onset of the COVID-19 pandemic in Georgia coincided with the rolling out of a reform of its social protection schemes. As a result, spending on social 19 World Bank staff calculations. 24 protection in Georgia almost doubled to 9 percent of GDP in 2020, while expenditure on the emergency schemes was equivalent to only 1.8 percent of GDP (World Bank 2022g). The emergency response prioritized workers, particularly those in the formal sector, and vulnerable groups The allocation of spending under the emergency response meant that payments to formal sector workers – whether employed, unemployed or retired – were predominant in most countries. That is, most emergency funding was channeled to formal sector firms to retain employees, workers, including those participating in the unemployment insurance system, or retirees receiving pensions. In Armenia, for example, most emergency funds were spent on wage subsidies paid to firms, while in Ukraine just over 60 percent of funds went to pensions and unemployment benefits. Armenia also allocated a significant amount of its social assistance response to unemployed workers. Much of this support likely benefited formal sector workers. In many ways, this allocation of emergency funding mirrors the existing social protection systems in each of these countries, which are dominated by social insurance spending. However, important differences also emerge given the focus of foundational social insurance on providing old-age pensions and the modest size of most labor programs in these countries, which required the introduction of new programs which piggybacked on existing systems and significant design tweaks, as discussed in the sections below. Beyond the focus on workers, all countries channeled resources to support vulnerable groups, often dwarfing the support to the poor. The elderly featured in the emergency response, with additional payments being made to pension beneficiaries in Moldova and Ukraine. Children were prioritized in Armenia, where support for workers who lost their jobs was conditioned on the presence of children in the households and disbursed in the form of child benefits and, to a lesser extent, in Moldova, which allocated a relatively larger increase in social assistance payments to poor households with children. Persons with disabilities received additional support in Georgia and in Ukraine, where one-time payments were made to disability beneficiaries. Modifications were also made to residential facilities in several countries to protect residents from COVID-19, and home-based services were modified, such as the provision of social care services to elderly persons living alone in Azerbaijan (Box 5). Support to vulnerable groups often far outweighs the amount allocated to support the poorest households. In Ukraine, expenditure on one-off grants to those considered newly vulnerable persons was almost 15 times the size of the increase in spending on GMI. 25 Box 5: Social care services during the COVID-19 pandemic Social care services are often provided to vulnerable populations, who may require tailored support and protection during a crisis. Demand for social care services may also rise during a crisis, as affected populations require psychosocial support or child protection services, for example. In some countries, the need for social services, such as shelter or tailored support for the elderly or children, is recognized in their DRM strategies or response plans. All the study countries offer social care services, although the levels of funding are very low, which suggests that the level and range of services provided is very limited. Social care services did not feature in the emergency response to the COVID-19 pandemic, although many countries instituted measures to protect people residing in residential facilities, such as homes for elderly people, and there are a few examples of more innovative responses. Ukraine and Azerbaijan asked social workers to be equipped with protective equipment, and distributed food among elderly homes. Armenia provided lump sum payments to persons providing residential social care services. Georgia’s Ministry of Labor, Health and Social Affairs provided recommendations on how to modify provision of social care services for existing beneficiaries to allow for remote delivery. Additionally, social care services were expanded to include provision of online services such as therapy classes and psychological assistance, as well as courses on developing capacity to cope with stress or helping with school assignments. Furthermore, existing beneficiaries and their families received monthly food stamp vouchers worth GEL 80 (400 percent of the local statutory monthly minimum wage). Sources: World Bank 2021a; ILO 2022 Finally, countries exploited programs within their social protection systems designed to protect households from economic shocks, such as energy price increases. Programs to protect vulnerable consumers are common within ECA, where access to energy, and specifically heating in the winter, is an important basic need. Utility subsidies or targeted support for vulnerable consumers often feature in social protection systems. These programs, which are often designed in manner that allows them to expand as prices rise or needs increase were used to respond to the COVID-19 pandemic. Ukraine simplified the administrative requirements to enroll in the Housing and Utilities Subsidy (HUS) program, which is the country’s largest social assistance program and reached close to half the country’s households in 2017 (6.5 million). The government also temporarily increased HUS benefits to mitigate the effect of increased utility costs during the lockdown. Armenia 26 included subsidies for natural gas and electricity bills (with maximum limits set on the level of usage) in the emergency measures, and Georgia provided a utility bill discount to cover payments for up to three months to low consumption households 20 for gas, electricity and other utilities in 2020. More recently, given rising global energy prices arising from the war in Ukraine, energy subsidies are once again looked toward to protect households, with the ability to scale them up quickly depending on administrative capacities such as the ability to identify and verify households, scale up the distribution of payments, and control fraud (For more information, see World Bank 2022d). Adapting program design: responding to emerging needs and newly vulnerable populations Countries implemented a suite of programs intended to mitigate the impacts of the pandemic on workers – through design tweaks, program expansion and the launching of new initiatives A core focus of the emergency response in all countries was to mitigate the impact of the COVID-19 pandemic on workers by protecting jobs or providing income support to the unemployed. This was seen in significant allocations to wage subsidies, ALMPs and unemployment benefits within the emergency packages, which were complemented by the natural responses to rising unemployment rates through unemployment insurance systems. While this support offered protection to many workers, some countries took additional steps to extend protection to formal workers who were not eligible for unemployment insurance, employed informally, or returning from abroad. In many cases, this required countries to adopt innovative approaches to setting the eligibility criteria for unemployment benefits, or to introduce new programs that creatively drew on available information, most often from the tax authority. In some cases, social assistance programs were also used to reach unemployed workers, as detailed in Table 3 below. In the sections that follow, we consider each of these programs in turn. 20 Households consuming less than 200 kWh of electricity and 200 m3 of natural gas per month. 27 Table 3: Social protection response by type of worker and their employment status Formal sector, Job retention Formal sector, not eligible Informal sector schemes eligible for UI for UI Employees in vulnerable sectors such as tourism or hospitality received Informal workers became one-off transfers A new program was launched to provide eligible for child allowances proportionate to their payments through the Social Security Agency, targeting families where one previous income in with the list of potential beneficiaries provided parent had lost a job, and January and February by the State Revenue Committee. Potential which required both parents Armenia 2020, of a minimum of beneficiaries did not have to apply and instead to not have a registered job. AMD 68,000 and a were provided the benefit based on the date of The assistance was equal to maximum of AMD dismissal. The assistance provided was equal to AMD 26,5000 per child, 136,000. Individual the minimum wage and disbursed through bank around 38% of the minimum entrepreneurs branches. wage. received 10% of turnover from Q4 but not more than twice the minimum wage. The government Rate of contributions continued to provide for self-employed benefits to those who The obligation to pay persons in became ineligible interest on unpaid Dismissed informal construction and trade during the special unemployment workers had to rely on Azerbaijan were reduced. The insurance premiums quarantine period, but social assistance salaries of 300,000 assessment of new was lifted until the programs. employees in affected beneficiaries was beginning of 2021. areas were partially conducted in the covered. normal way during the pandemic. The government introduced a temporary Any natural person unemployment allowance in two phases. The declaring to be first phase provided benefits to all those engaged in economic In April 2020, a wage formally employed for any of three consecutive activity in Georgia subsidy was months between July and December 2019, or for and/or who had introduced for all at least one month during the first quarter of income in the first retained jobs. Salaries 2020, and who were no longer recorded by the quarter of 2020 who Georgia of up to GEL 1,500 tax authority as receiving salaries. Beneficiaries registered and applied received a tax were granted GEL 200 per month for up to six for compensation at exemption of up to months. The benefit was withdrawn as soon as SESA before August 1, GEL 750 for a six- the beneficiary found a new formal job and re- 2020, regardless of month period. appeared in the tax registry. The second phase submitting evidence of provided the same benefit but to individuals economic activity, was who had been employed between January to eligible to get a one- November 2020 and no longer received a salary time transfer of GEL 28 for one or more months in the period from 300. Subsequently, December 2020 until February 2021. another one-off compensation payment of GEL 300 was provided to individual entrepreneurs and persons employed in facilities whose operations were suspended as a result of restrictions in December 2020 and January 2021. Minimum unemployment benefit increased to MDL Eligibility criteria were modified to allow all 2,775 per month (US$157). unemployed persons, including those who The mechanism for had returned from abroad, to apply for a administering Affected business minimum unemployment benefit until the unemployment insurance were reimbursed end of the state of emergency. Persons who was changed from in- 100% of their payroll had not lost their jobs prior to the person registration to taxes if they were declaration of the state of emergency and distance-based. Scanned Moldova forced to close down or photographed copies of did not meet the regular eligibility criteria and 60% if they were were obliged to enroll in the compulsory applications, declarations not forced to close health insurance system before applying for and documents could be down but became benefits. This was financed from the state sent remotely by mail or e- non-operational. budget. Ineligible formal workers and mail to the address of the informal workers also made use of the new territorial subdivision administrative procedures, which allowed corresponding to the for remote applications. applicant’s place of residence or registration. A partial unemployment benefit was introduced while The regular unemployment benefit is provided independent of length or level quarantine measures of social insurance. The minimum unemployment benefit was increased from were in place for UAH 650 to UAH 1,000 (from US$24 to US$37). For the period of the workers who were lockdown, the parliament relaxed the rule for the unemployed to renew their temporarily registration every month by personal attendance of the regional SES office. unemployed because Ukraine employers reduced or Rules for registration of the new unemployed were also simplified (applicants received unemployed status from the first day of registration and without the stopped activities. usual need to check whether an alternative employment option was Entrepreneurs, available). The SES, in partnership with the Ministry of Digital Transformation farmers and self- (MODT), also introduced an online registration option for the new employed were unemployed through the Diya digital application. exempted from paying Social Security Contributions until end of April 2020. 29 Promoting job retention was central to the response, with extensive use of wage subsides to maintain formal employment contracts, even in countries with high rates of informality The goals of wage subsidy schemes are to enable workers to keep their contracts with their employers even as work is suspended or reduced, and for firms to keep talent and expertise within the company so that operations can increase quickly once economic activity picks up. Globally, during the pandemic such schemes were favored by countries with low rates of informality. Eighty-eight percent of countries classified in the low informality tercile had such schemes, compared with only 32 percent among high informality countries (Gentilini et al. 2020). In the study countries, the converse was true: job retention programs tended to be more important in the high informality countries. Moldova and Ukraine, which are in the low informality tercile, in fact provided no or minimal funds for job retention through the emergency programs. Moldova did not implement any active labor market measure. Ukraine announced support to small and medium enterprises (SMEs) of one minimum wage per employee, and suspended the requirement to pay social security contributions. Armenia and Georgia, in the medium informality tercile, invested significantly more. Azerbaijan introduced significant financial support for public and private companies to maintain the salaries of contract-based employees, including coverage of social security payments for 1.7 million people. Job retention schemes were channeled directly to workers or through firms, with varying generosity and duration. In many countries wage subsidies were provided directly to employees. In Azerbaijan, the average beneficiary received AZN 454 during the months of April and May, or around two-thirds of the average monthly wage. In Moldova and Georgia, where wage subsidies were channeled through businesses in the form of tax exemptions, payments were substantially larger and remained in place for extended periods. Subsidies in Moldova were paid proportionate to payroll taxes, as paid prior to the pandemic (up to 100 percent). With benefit size limited only by the level of previously paid taxes, the subsidies’ generosity increased with the firms’ wage levels. 21 While the original period only covered two months (from mid-March to mid-May), it was further extended four times with payments for some firms continuing until the end of 2020. In Georgia, wage subsidies were provided for six months in the form of tax exemptions of up to GEL 750 or 67 percent of the average monthly wage. Armenian employees active in affected sectors received one-off cash transfers proportionate to their wage, up to AMD 136,000 or around 72 percent of the average monthly wage in 2020. 21 Sixty percent of the disbursed funds went to large companies with more than 250 employees each. Source: https://mf.gov.md/en/content/reports-1 30 These programs focused on employees in the formal sector, irrespective of how the funds were channeled. As the institutions eligible for applying for such schemes were registered workers, firms or SMEs, the schemes were limited to the formal sector. Gig economy workers and other non-standard contract or daily labor workers were typically excluded. As many countries had existing wage subsidy schemes either for the purpose of activation or established in response to the 2008 global financial crisis, these offered a relatively quick way of responding to what was believed to be a rapid but temporary surge in unemployment. While wage subsidies played an essential role in many countries in mitigating spikes in unemployment, they likely also increased existing labor market inequalities (Ando et al. 2022). Young people, those with low educational attainment and part-time workers tend to be most likely to be in non-standard forms of work, while: (i) the share of nonstandard employment was high in those sectors most affected by the pandemic; and (ii) nonstandard workers were most often not protected by employment-related benefits such as wage subsidies (Ando et al. 2022). The cost-effectiveness of the job retention schemes and trade-offs surrounding their design during the pandemic are not well evaluated. Despite the sizeable investments made in wage subsidies, to date no rigorous evaluations have been made of their cost-effectiveness or the impact of design decisions such as the type of workers covered, length of coverage, time period used as income reference period or the payment channels. Some evidence from Moldova suggests that the generous tax exemptions were effective at keeping beneficiary enterprises operational during the pandemic and retaining jobs. Of 399 supported enterprises, 96 percent continued to operate until the end of 2020 and maintained levels of employment effectively at the same level as at the beginning of the pandemic. At the firm level, however, job protection programs may also hinder labor reallocation from less productive to more productive firms.22 Emerging evidence suggests that job retention schemes played a crucial role protecting workers in some countries. A recent analysis of countries in ECA suggests that higher expenditure on job protection during the pandemic was associated with higher employment, less inactivity and lower poverty in countries with weak pre-pandemic social insurance systems (Demirgüç-Kunt, Lokshin, and Torre 2022). Notably, both Georgia and Armenia – which had no functioning unemployment insurance systems at the beginning of the pandemic – each spent a relatively high equivalent of 0.25 percent of GDP on job retention schemes alone. 22 The extent to which this may have been the case, and if the companies who received the support would not have continued operating in absence of the benefit, would require more in-depth evaluations. 31 Modifications to unemployment insurance expanded access – but benefits were generally low and of short duration, which undermined the effectiveness of the support Unemployment benefits featured in all the countries’ responses to the pandemic, though the range of adaptations and resulting increases in coverage varied greatly. Globally, unemployment insurance schemes have expanded during periods of economic crisis, when countries have also often modified eligibility criteria to facilitate access (Grosh, Bussolo, and Freije 2014; Dikmelik 2012; Robalino, Newhouse, and Rother 2013). By design, unemployment insurance schemes expand during periods of economic downturn, when employees who have paid into the system lose their jobs and become eligible for support. However, the COVID-19 pandemic revealed rigidities within the unemployment insurance programs in many countries that limited access, while also demonstrating the challenges of responding to such a devastating covariate shock through a contribution-based system. Globally, as of February 2022, roughly 45 percent of countries modified their unemployment benefits to respond to the pandemic. 23 Among upper-middle income countries, most modifications entailed temporary increases in benefit levels (such as Albania, Argentina, and Russia), or modified application procedures (such as in Bulgaria, Ecuador and Serbia). A response through unemployment benefits featured more strongly in the study countries, where all countries with unemployment insurance schemes (Azerbaijan, Moldova, and Ukraine 24), expanded these in response to the pandemic, while Georgia and Armenia introduced new unemployment benefit schemes in the absence of an existing unemployment insurance scheme. While countries introduced “design tweaks” to try to facilitate the access of eligible workers and provide more meaningful insurance payouts, the increase in coverage remained modest. Moldova and Ukraine simplified the application processes for unemployment insurance by allowing the online submission of documents. Azerbaijan’s system was already online, and no steps were taken to simplify the process. Moldova and Azerbaijan automatically extended the duration of payments to beneficiaries of unemployment insurance by, for example, waiving the need to recertify after a specified period. Moldova and Ukraine increased the amount paid to recipients of unemployment insurance. Despite the changes, in most cases coverage of formal unemployment insurance started from a low base, and the increase in beneficiaries because of these modifications was modest. Countries also harnessed their unemployment insurance schemes to reach workers who were not normally eligible for such support. This was done in two ways. First, countries modified the eligibility criteria to include workers who had contributed to unemployment insurance but were not yet eligible to access benefits (such as formal sector workers who did not meet 23 A total of 88 out of 194 countries modified unemployment benefits as part of their emergency responses (Gentilini et al. 2022). 24 Armenia had an unemployment insurance scheme that was abolished in 2013 and replaced with cash support that is provided, on a case-by-case basis, to unemployed job seekers who are assessed as being uncompetitive. 32 the minimum requirements in terms of duration of contributions). This was possible as information on these employees was already available to the unemployment insurance system. Secondly, countries extended benefits to all workers, including those working in the informal sector or who had returned from abroad, by allowing these groups of workers to apply for support. Moldova, for example, modified the eligibility criteria for unemployment benefits to allow all unemployed persons, including those who returned from abroad, to apply for a minimum benefit until the end of the State of Emergency. This ‘modified’ unemployment scheme covered about 27,500 workers, a much larger increase in coverage than that through the regular unemployment insurance system. 25 Armenia and Georgia introduced new schemes for people who had lost their jobs, drawing on data from the Tax Authority. During the pandemic Armenia provided assistance in two phases to workers who lost their jobs, the first covering 7,400 and the second 5,000 workers. 26 For the first phase, payments were made based on a list of beneficiaries generated by the Revenue Authority; in the second phase, the list of potential beneficiaries from the Revenue Authority was compared with online applications from individuals. These initiatives were complemented with support for families with children in which at least one parent had lost a job: this was initially paid according to the beneficiary list generated by the Revenue Authority, but was later amended to allow households in which both parents did not have registered jobs to apply for support through the official website of the Social Security Agency. 27 Georgia introduced two schemes that were each implemented in two phases: the first scheme was for formal sector workers who had lost their jobs based on information from the Tax Authority and applications by employers (the list and application had to match). The second scheme was for both registered and unregistered self-employed workers. 28 In both countries, the schemes were introduced and subsequently expanded to accommodate different contract types and informal workers, and, in Georgia, to include returning migrant workers. Despite significant efforts to reach unemployed workers, unemployment benefits were generally low and of short duration, which undermined the effectiveness of this support. For example, in Georgia the temporary unemployment allowance for the self-employed instituted during the crisis was a flat GEL 200 (US$64) a month, or 16.8 percent of the average wage in 2020 and about 64 percent of the subsistence minimum for a typical household. In Ukraine, the temporary increase in unemployment benefit was limited to the period of the lockdown. 25 They still had to pay for the compulsory health insurance program, which served as a barrier to participation for some otherwise eligible workers. Data are from NSIH; slightly different numbers are given by NEA. 26 There may have been some overlap between the two groups. The program was managed by the Ministry of Labor and Social Affairs and payments were made through the Social Security Agency. 27 Armenia rolled out a similar program to reach pregnant women who lost their jobs due to the pandemic. 28 This unemployment assistance was for people who had lost their jobs and, by design, required no contribution history. While the program may be classified as social assistance, given its focus on people who had lost their employment it is considered in this section. 33 Poverty-targeted social assistance programs were rarely used to reach poor workers who lost their jobs While social assistance programs offer an avenue to identify and reach poor households with members who lost income or employment, this option was rarely used in the study countries. While all countries financed the provision of benefits to mitigate the impact of the pandemic on workers, there are almost no examples of social assistance programs being modified to extend coverage to reach very poor people who had lost their jobs. This strategy could have mitigated gaps in coverage of job retention schemes and unemployment benefits by directing resources to poor households, with members who lost jobs, which were most likely to be in the informal sector and often the hardest to reach. The primary example of such an approach is Ukraine, which modified the eligibility for the GMI to enable people who lost their jobs during quarantine to become eligible for support. All the other countries that purposefully sought to reach informal workers did so through their unemployment or employment programs. Support to poor and vulnerable groups was largely channeled through existing programs, with somewhat limited expansion in coverage Countries channeled resources through established social assistance programs to reach vulnerable groups affected by the pandemic, exploiting existing flexibility within the system. As discussed above, countries channeled resources through a range of social assistance programs to reach vulnerable and poor populations. Often, this support was provided as a “top up” to beneficiaries of existing programs (that is, a vertical expansion). In this way, countries recognized the vulnerability of these populations to the pandemic, while also pragmatically exploiting existing program infrastructure, information sources and delivery systems. For example, additional payments were made to pension beneficiaries in Moldova and Ukraine, as that was the most robust system countries had in place. Interestingly, there are also examples of countries “piggybacking” on existing programs to provide beneficiaries with support that was intended to achieve new objectives. For example, Armenia provided an energy subsidy as a “top up” to existing beneficiaries of the poverty-targeted Family Living Standards Enhancement Benefit (FLSEB), while Georgia provided an educational grant to students in poor households under Targeted Social Assistance. This approach is consistent with the design of social assistance systems in some countries in ECA, which provide for automatic access to additional benefits once a household is enrolled in the poverty-targeted programs, while also drawing on readily available information on the groups that were likely affected by the pandemic. In a similar fashion, Ukraine provided one-off assistance to those considered newly vulnerable (9.7 million vulnerable pensioners, 424,000 disability program beneficiaries, and 176,000 social pensioners). In some cases, efforts were made to expand the coverage of existing programs by modifying eligibility criteria to reach new beneficiaries or introducing new programs for uncovered groups. While the study countries devoted emergency resources to reaching poor and vulnerable groups, few countries launched any new programs for groups that remained 34 outside the social assistance system 29 or modified the eligibility criteria of existing programs to expand them horizontally. This is in direct contrast to new programs that were introduced to reach workers through wage subsidies or unemployment benefits. Notable exceptions are found in Ukraine, which simplified administrative requirements for enrolling in the Housing and Utilities Subsidy (HUS) program – the country’s largest social assistance program, which in 2017 reached close to half of the country’s households (6.5 million) 30 – and modifications to the eligibility criteria for the country’s GMI. Moldova similarly increased eligibility for the Ajutor Social (the Last-Resource Income Support (LRIS)) with the aim of reaching more poor households to protect them from the pandemic; an aim mirrored in Georgia’s decision to provide temporary social assistance transfers to poor and vulnerable households, who were enrolled in the database and assessed with a Proxy-Means Test (PMT) score below an established threshold. 31 Below we explore these issues further, with a specific focus on the LRIS in each of the countries. Countries took steps to protect the beneficiaries of poverty-targeted social assistance programs Countries took steps to offer additional protection to existing recipients of poverty-targeted social assistance programs by modifying recertification rules or increasing the value of payments. All the study countries have at least one program that aims to reach the poorest members of society (that is, an LRIS program). All the countries, except Armenia, waived requirements that social assistance beneficiaries re-apply or be recertified, including any requirement to submit documents certifying their income or employment status. This change in program rules enabled existing beneficiaries to remain in these programs, while also minimizing demands for in-person interactions with staff. While not a formal horizontal expansion to new beneficiaries, in many cases the decision to waive these requirements resulted in a temporary increase in beneficiary numbers. Armenia, Georgia and Moldova also increased the size of benefits for existing participants. Putting these design changes into practice did not require any new information to be obtained on beneficiaries and, as a result, could be done quickly. These enabled many people already found to be eligible for the LRIS to quickly receive continuing support during the crisis – likely a lifeline for many very poor people. 29 An exception is the unemployment assistance provided in Georgia and Moldova, which was considered under unemployment benefits above. 30 The government also temporarily increased HUS benefits to mitigate the impact of increased expenditure on utilities during the quarantine 31 In Georgia, the government provided temporary social assistance payments to all households with a PMT score below 100,000, the cut-off point for eligibility for child benefits. This extended the TSA – which is paid to all households with a PMT score below 65,000 – to additional households, some of which were receiving the child benefit. Armenia introduced a new program to subsidize tuition fees for students in graduate, post- graduate and academic programs, with funds transferred to the institutions based on lists provided by the Ministry of Education. 35 While this design modification provided certainty to beneficiaries, the level of support was generally inadequate, reflecting an overall feature of social assistance in countries in Eastern Europe and the South Caucasus. As discussed above, social assistance programs in the study countries tend to pay benefits of low value. Even when these values were increased, focus group participants appeared to have mixed views on the value of the benefits, with Armenian participants viewing them as “quite irrelevant” to the needs of recipients, Georgian participants differing among themselves, and Moldovans asserting that although social aid was not enough to meet all basic needs, it was nevertheless very helpful. The additional support was also time-bound, generally limited to the state of emergency. In Georgia, for example, support was provided for six months. In Moldova, the increase in the income threshold used to determine eligibility and the higher benefit level for Ajutor Social, the main targeted social assistance program, were only effective in April and May. Changes to the eligibility criteria for some LRIS schemes enabled these programs to rapidly expand coverage of the poor. Among the study countries, all LRIS schemes are “on-demand” programs, that is, people may regularly apply for support and are accepted into the program if they meet the eligibility criteria. In principle, this creates flexibility in the LRIS schemes to expand (and contract) in response to changing needs among the poorest parts of society. However, eligibility criteria can slow the responsiveness of the programs when they are intended to identify the chronically poor or are allocated to poor households with specific characteristics, such as including children under a given age. The targeting methods used to identify eligible households may introduce further rigidities, as discussed in the section on delivery systems below. In recognition of these issues, some countries took specific steps to modify the eligibility criteria to accommodate additional chronically poor households or households driven into poverty because of the pandemic: that is, those who are transitory poor. Moldova and Georgia took steps to reach additional very poor households by raising the eligibility threshold within the existing targeting system. Moldova raised the threshold for the Ajutor Social from MDL 1,107 to MDL 1,300 during the emergency period (April and May 2020). Georgia provided temporary social assistance to poor households that, prior to the pandemic, were only eligible for the child benefit, in effect, raising the eligibility threshold for the Targeted Social Assistance. In contrast, Ukraine modified the eligibility criteria for the GMI to include people who had just lost their jobs by excluding their previous salaries from the assessment of income. 32 Overall, coverage of LRIS among the study countries increased, although this change in coverage ranged from 0.5 percent in Armenia to 32.6 percent in Ukraine. This modification was introduced because eligibility is assessed based on average income over the last six 32 months. 36 Table 4: Number of beneficiaries receiving poverty-targeted social assistance in 2020 2019 2020 year-on-year % Country Program Indicator (thousands) (thousands) change Armenia FLSEB Households 61.4 61.7 0.5% Azerbaijan TSA Individuals 296 322.1 8.8% Georgia TSA Individuals 427 525 23.0% Moldova Ajutor Social Households 46.5 53.1 12.5% Ukraine GMI Households 258 342 32.6% Source: World Bank 2021a. Notes: Armenia and Azerbaijan are at the end of the period, Moldova a yearly average and all other show total beneficiaries during year. The emphasis on serving existing beneficiaries likely failed to reach those people who were driven into poverty by the crisis. As mentioned above, across the study countries, those people falling into poverty as a result of the pandemic likely differed from those who were already poor. They were more likely to be found in urban areas and were more likely to be employed in industry and services than in agriculture. Few were already recipients of social assistance programs before the crisis hit. Therefore, the emphasis on increasing benefit levels to existing clients of LRIS programs, providing them with support through new programs, and allowing them to remain in the program through the emergency period likely had a very limited impact on the newly poor. Expanding the coverage of LRIS (and unemployment benefits, as discussed in the sections above) offers the potential to reach additional poor households, as was seen in Georgia, Moldova and Ukraine, although the effectiveness of this approach depends on existing program design, as discussed in Box 6 below. 37 Box 6: The importance of design: simulating different response options in Moldova Often, efforts to strengthen the responsiveness of social protection to shocks focus on delivery issues. And yet, a lack of attention to program design, and trade-off between design options – in terms of the population groups reached and the severity of their vulnerabilities and need – can undermine the effectiveness of the response. Simulations from Moldova illustrate how the type of shock and the population it affects requires differences in program design to achieve an optimal response, as shown in Figure 13 and Figure 14. More specifically, two types of shocks are simulated: one that leads to uniform income losses across the population (Figure 13) and the other that results in random losses in income to zero across the population (Figure 14). These are then compared with different response designs, with fixed budgets. When a shock is proportional across the population, providing additional support to current beneficiaries of social assistance can be an effective response, as is selecting beneficiaries based on their poverty status before or after the shock. When, however, a shock imposes random losses to individuals across the population, as is simulated in Figure 14, the optimality of program design changes and programs to support those individuals who suffered losses from the shock gain importance. That is, simulated reductions in selected poverty rates through cash transfers are simulated to be lower when selecting beneficiaries based on prior poverty status, rather than attempting to assess losses imposed by the shock. Figure 13: Simulation of proportional Figure 14: Simulation of random loss loss shock shock 0% 0% 0% 0% -10% -1% -1% -20% -20% -1% -1% -40% -30% -2% -2% -40% -2% -60% -2% -50% -3% -80% -3% -60% -3% -100% -3% -70% -4% -120% -4% -80% -4% -90% -5% -140% -4% Squared Poverty Gap (LH) Headcount Rate (RH) Squared Poverty Gap (LH) Headcount Rate (RH) Source: Hernandez et al. 2022 38 Adapting the delivery chain: identification, enrollment, and provision to people in need The choices countries made about who to reach and how to do so were often driven by the parameters of the existing social protection programs, their scope and reach. While these choices influenced the shape of the response, the delivery systems in place across the study countries enabled or hindered the achievement of these objectives, often with variability across different pillars of the social protection systems within the same countries. These differences proved especially pronounced in the case of the systems’ ability to reach new beneficiaries. In the section that follows, we explore modifications that countries made to their delivery systems, focusing specifically on those of the LRIS, guided by the framework for the delivery chain set out in Figure 3. This analysis is complemented by examples of the steps in the delivery systems for employment programs or social insurance when these offer examples of innovations or modifications that provide effective. Outreach Efforts to inform potential beneficiaries of changes to poverty-targeted programs were limited, reinforcing a general neglect of outreach, which undermines an on-demand system. Outreach is the first step in the delivery chain, through which potential beneficiaries learn of the opportunity to apply for a program. According to Lindert et al. (2020): “outreach involves interactions to inform people about social protection programs and delivery processes and to create adaptations to encourage them to engage.” Globally, this first step in the delivery chain is often the weakest (Lindert et al. 2020). For social assistance programs, specifically LRIS programs, this seems to hold true in the study countries. In Ukraine, for instance, information on simplified application procedures was available on the government website, but this was not supported by a dedicated communication campaign. Moldova provides insights into how the lack of a coherent outreach strategy or information campaign leads to limited awareness. Focus group participants were almost entirely unaware of the availability of assistance prior to the pandemic, and participants in Moldova’s targeted assistance program, Ajutor Social, were unaware of other government programs that they might benefit from. Information on the program was spread through informal channels (friends, relatives, neighbors, doctors and mailmen) and the media, although Moldovan social workers noted that information disseminated by the media could be inaccurate and misleading (Manea and Dumitru 2020). In contrast, countries adopted innovative communication strategies to inform other populations about opportunities to apply for support. Responding to the COVID-19 pandemic 39 required countries across the globe to adopt innovative – and more intensive – communication campaigns to their populations, asking for compliance with public health measures while also providing information on sources of health and financial support. In contrast to the experience of outreach for LRIS programs, innovations in communication were witnessed among labor programs in the study countries. In Ukraine, unemployment and sickness benefits were widely promoted and made extremely accessible through the Diya app. 33 In Georgia, workers in the State Employment Support Agency felt that citizens were reasonably well informed on the programs, owing to extensive advertisement and the various media used for dissemination. Further research is required to understand why similar innovations were not applied to the poverty-targeted programs, be this limited access to mobile phones and the internet about the poorest segments of the populations (see section below) or perceptions among policymakers and implementers on how best to reach this population. Application Simplified procedures facilitated – to some degree – applying to poverty-targeted programs in some countries, mostly for beneficiaries already in the programs. LRIS programs in the study countries often involve detailed applications and supporting evidence, particularly on incomes and/or assets, including for reapplication or recertification (World Bank 2014). Generally, people have to apply for support in person at a municipal or ministerial office to fill out application forms and provide supporting evidence, although efforts are underway to modernize these processes. These application processes, when complex and time consuming, can undermine the “on-demand” nature of the programs. For example, focus group participants viewed applications for the Family Living Benefit Program in Armenia as “complicated” and “incomprehensible”. By contrast, beneficiaries who participated in focus groups in Moldova found the application process simple, particularly with social workers helping to fill out forms, although problems obtaining some required documents and the need for repeated applications were noted (oddly, social workers found the forms too complicated and hard to understand). Some of the requirements were modified in response to the COVID-19 pandemic, as discussed in the section above concerning changes to program design. Most frequently, countries modified the recertification requirements, most effectively by automatic extension of the eligibility of households during the period of emergency (such as in Azerbaijan and Moldova). While this decision risked the inclusion of 33 Diya is a smartphone application available in the Apple App Store and Google Play which provides users with services such as a digital driver’s license or vehicle registration, and can also be used to apply for one-off allowances and receive sick leave certification digitally. By the end of 2021, 12 million individuals or around a quarter of Ukraine’s population used Diya. The mobile application was also used to process applications and payments of UAH 1,000 for individuals who obtained their vaccination certificate in the app. 40 (poor) households who might no longer be eligible for support, it provided immediate certainty to all beneficiaries that they would continue to receive support through the emergency period. In contrast, few countries adopted procedures that facilitated the application of new beneficiaries to the LRIS; this undermined changes in program design that were intended to expand programs horizontally. While Ukraine took steps to expand the GMI to poor people affected by the pandemic, application rules were not eased, which still necessitated in-person visits for first time applicants to the GMI to register. The experience in Moldova offers insights into how the delivery systems can enable or undermine program design features. The database of applicants to the Ajutor Social covers 20-23 percent of the total population (it is used to pay a heating allowance), and validation or eligibility checks can be done automatically. When the government expanded the program horizontally by increasing the threshold for eligibility, the program was automatically extended to those households already in the management information system (MIS) that were found to be eligible. In contrast, the mechanism for receiving and processing new applications failed to adapt to the lockdown period, and, therefore, households that fell into poverty or, for some reason, were not yet already enrolled in the database, were unable to apply due to closed welfare offices and inconsistent procedures for applying by phone. This failure to consider the application procedure, unfortunately, undermined the effectiveness of the policy decision to expand the Ajutor Social horizontally. Investments in online systems positioned some programs to adapt quickly to the COVID-19 pandemic. Azerbaijan recently improved the application process for Targeted Social Assistance. Families can now apply to the program online through the Unified Digital Application and Appointment sub-System (VEMTAS). The information in this initial application is verified by integration with other government authorities. The outcome of all applications is transferred to the central database through VEMTAS. Georgia harnesses investments in a social registry to expand the TSA horizontally to additional households. Households enrolled in the social registry with rating scores less than 1,000 received temporary social assistance payments.34 Across programs, online applications were commonly used for employment programs, including for newly launched programs such as in Ukraine (see box below). 34 Under the TSA, social assistance is paid to poor households with scores of less than 65,000. Households with children with scores between 65,001 and 100,000 receive a child benefit. 41 Box 7: Ukraine’s Diya app facilitated applications for unemployment and sickness benefits The Diya platform, launched by the Ministry of Digital Transformation in 2020, includes website and mobile applications. The Diya mobile app enables citizens to have digital national IDs, taxpayer identification documents, driving licenses, biometric international passports, vaccination certificates, and other documents. The Diya currently uses information available from state registries and databases. All data are transmitted and stored in encrypted form; for critical data Diya uses the blockchain technology of distributed data storage. The Diya connects users to services using their Bank IDs. By the end of 2021, 12 million individuals, or around a quarter of Ukraine’s population, were using Diya. Source: World Bank 2022j However, reliance on digital processes could also have served as a barrier to participation for anyone with a low level of digital skills and internet access. Digital procedures can create difficulties for the elderly, those lacking access to computers or the internet, and those unable to pay for internet services. While both internet and computer access have expanded sharply over the past 10 years, significant proportions of the populations of the five countries still lack one or the other, or both. According to focus groups in Georgia, only a small share of those potentially eligible had the wherewithal to access services online. However, countries continued to provide in-person support by counselors or social workers on applications to persons who faced barriers to using online applications. In Ukraine, one-off payments were exclusively accessible by digital means; 15.1 percent of the population lack any digital skills, and Ukraine ranks lowest in household access to personal computers. 35 35 Forty-seven percent of the Georgian population lacked the basic knowledge to use computers, in Armenia 34 percent of individuals lacked basic digital skills and a survey in Ukraine showed that 15.1 percent did not have any digital skills and 53 percent had lower than basic digital skills. 42 Figure 15: Access to the internet, 2010 and Figure 16: Access to computers, 2010 and 2020 2020 100% 90% Households with access to personal 90% 80% Internet Access at Home (%) 80% 70% 70% Households with 60% computers (%) 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% 2010 2020 2010 2020 Source: Eurostat. 2022. European Neighbourhood Policy - East - statistics on science, technology and digital society. Note: Data for Armenia are for 2018 instead of 2020. EU data on personal computers are for 2017 instead of 2020. Assessment Steps were taken to simplify the assessment process for new or existing beneficiaries, with greater use of data verification through interoperability with government databases. Frequently, countries waived the requirement that social workers conduct a home visit to verify the information provided in the application and instead relayed on self-declared information (as was the case in Moldova) or information verified through data cross-checks with other government databases (in Georgia). Steps were also taken to simplify the documents that households were required to submit, such as relying on self-declared income, assets and family structure in Moldova. In Armenia, verification of household eligibility through other government databases did not always lead to more efficient assessment. More specifically, while the FLSEB is designed to be “on-demand”, the potential for the program to expand horizontally to new beneficiaries was undermined because: (i) the method for assessing eligibility through the Vulnerability Assessment System was complex; and (ii) registration and enrolment procedures remained challenging due to bureaucracy and low administrative capacity. More specifically, registration remains paper-based, and several verification checks must be carried out manually with other civil registers, while for those that are automated the data repositories are not integrated. All this slowed eligibility determinations, increased the burden on applicants, and increased reliance on in-person submission of new applications to expand programs, which was constrained during the 43 lockdown. Armenian participants complained about lengthy delays in eligibility determination, in some cases of up to 5-6 months. Social workers noted that limits on private visits imposed by the pandemic and the lack of documents did lead to lengthy delays in some cases. The integration of databases was important for reducing the burden on government and recipients involved in the assessment process. The existing MIS in Georgia enabled the employment agency (SESA) to routinely access the beneficiary list of targeted social assistance (TSA) to identify and reach out to them for various SESA-administered ALMPs. In Ukraine, verification of application information was automated, so that verification could be conducted in a matter of minutes. Armenia and Georgia harnessed the data that existed in their tax authorities, as discussed above. However, the simple existence of a database on the poor was not necessarily sufficient to improve programs. For example, in Armenia, the Quarterly Emergency Assistance targets families registered in the VAS but not determined to be eligible for support. This provides the government with something akin to a social registry. Nevertheless, because of paper-based registration and non-automated verification checks, Armenia’s social assistance system did not make use of this database to expand vertically. In some cases, attempts to increase the integration of databases during the pandemic initially created challenges. Making substantive changes to administrative processes in the middle of an emergency, however necessary to achieve the needed expansion of coverage, can overburden administration. In Armenia, problems during the integration of databases led to social workers being provided with inconsistent information from different ministries, leading to an increase in the administrative burden and stress for the workers, exacerbated by a lack of written guidance on some procedures. On the other hand, social workers participating in focus group discussions felt that despite the difficulties, the experience had taught them new skills and that most difficulties were eventually overcome. In Georgia the lack of modularity in the existing MIS required a new one to be built when previously non-existent unemployment benefits were introduced, leading to delays in implementation. However, focus group participants also reported that these problems were sorted out and administrative processes improved over time. Finally, social workers were overburdened, which affected all initial stages of the delivery chain. Social workers in Moldova and Georgia focus groups complained about the increased workload during the pandemic, due to the increase in beneficiaries from the pandemic, rising domestic violence cases, colleagues’ absence due to illness, and inadequate equipment. A lot of unpaid overtime work was required, and stress increased, particularly given the low pay. There also were concerns about increased health risks in dealing in person with clients during the pandemic. Some vowed to leave the profession due to this experience. Box 8 describes 44 the response to the COVID-19 pandemic in the United Kingdom through Universal Credit and the steps taken to facilitate access to this program, including the redeployment of thousands of staff. Box 8: Facilitating access to an on-demand social assistance program: the case of Universal Credit In the United Kingdom, Universal Credit is targeted to those who have low incomes or are out of work and not in receipt of a pension. From its introduction, Universal Credit has been managed and accessed almost entirely digitally, with claimants applying online. This was an advantage during the pandemic, as applications could proceed without face-to-face contact. To support the increase in claims all local job centers were repurposed as claims processing teams to support the processing of new claims; and more than 8,500 staff were redeployed within the Department of Work and Pensions (DWP) and 1,000 staff transferred from other government departments to assist with claims processing. Despite the huge increase in claims during the early months of the COVID-19 crisis, the timeliness of claims processing improved, with 93-95 percent of claimants receiving their first payments on time rather than the pre-pandemic figure of 85 percent. However, it should be noted that Universal Credit has a built in five-week wait between when a claimant applies and receipt of the first benefit. Additionally, no specific outreach was undertaken to identify and support new claimants, and there were reports that new claimants struggled to navigate the system. Source: Mackley 2022; Ross and Clarke 2021; Sandford 2021. Payments The reach of the banking system in the study countries, coupled with the established use of these systems for social protection payments, enabled the rapid distribution of payments. Access to commercial banks differs substantially across the region, in turn limiting the extent to which social benefits can be paid out electronically. While in Georgia and Ukraine almost two-thirds of the population report having accounts at banks or other financial institutions, less than half do so in the other countries (Demirgüç-Kunt et al. 2018). 36 All countries were already paying some beneficiaries through bank accounts or bank branches, which enabled quick disbursement of payments once people were enrolled in the program, and facilitated adherence to social distancing protocols. In Ukraine, for example, 65 percent of payments to 20 million beneficiaries are made through bank accounts. Some countries used SMS notifications of deposits in beneficiaries’ accounts. In Armenia, however, most social 36 Data are from 2017, however the most recent available data from Georgia and Ukraine do not show an increase in commercial bank accounts since. Only Moldova has seen an increase in commercial bank accounts. 45 assistance benefits and pensions are paid in cash. Beneficiaries who received bank transfers mentioned the need to travel to banks to check whether transfers had arrived due to the lack of SMS notifications, although it was not clear if this reflected the lack of a phone service or failure to establish a system. In Armenia, the provision of information and some individuals’ familiarity with banking infrastructure was so low that, under government policy, some benefits were returned after having been left untouched for 12 months in bank accounts. In Moldova, payments are made through a government service for electronic payments (MPay) that enabled payment services using multiple payment methods: credit cards, payment terminals, e-banking and cash payments. Better preparing for the next crisis: Building the resilience of the social protection system The COVID-19 pandemic was not the first large-scale crisis and will not be the last. Already, the economic ripple effects of Russia’s invasion of Ukraine are being felt far beyond Ukraine’s borders. At the same time, climate change is leading to more frequent and intensive weather events and climate-induced shocks. While the timing, intensity and nature of future shocks is unpredictable, the fact that shocks will continue is as close to a certainty as is typically afforded in making predictions. All of this emphasizes the importance of preparing for future shocks. The experience of responding to the COVID-19 pandemic through social protection systems in the study countries suggest avenues for building the resilience of the social protection system, as follows: Strengthening the overall performance of social protection systems, especially poverty- targeted programs and unemployment benefits, would contribute to a more effective response to future crises, provided that these investments result in improved access to benefits for poor households and workers with non-standard employment contracts – populations that were often missed in the response. This would require: (i) reconsidering the financing and reach of poverty-targeted social assistance programs so that they provide broader coverage to a greater proportion of the poor, irrespective of household characteristics, and more adequate benefits; and (ii) reviewing the parameters of unemployment insurance (and in some cases establishing programs, such as is currently under consideration in Georgia) to facilitate uptake by eligible workers, while also considering options to extend coverage to those currently ineligible for support, including workers in the informal sector and those in the gig economy, 37 and given the experience of the pandemic in many countries, return migrant workers. In multiple countries, increases in coverage and adequacy can be achieved through better use of existing resources, such as by reducing the 37 These issues are considered in Truman et al. 2019. 46 fragmentation of programs, reallocating financing to more effective programs or improving methods to activate social assistance beneficiaries. Continuing and extending reforms in delivery systems, as discussed in the paragraph below, is also important for improving the performance of foundational social protection systems. 38 Ongoing reforms to delivery systems should continue and be furthered, drawing on lessons from across social assistance, social insurance and labor programs within and across countries. To provide effective social protection support, delivery systems must enable the identification of people in need of support, assess their eligibility, enroll them and deliver support to them at the right time, in a manner that is transparent, accountable and as costless as possible for the applicant. These features of an effective social protection delivery system come to the fore during a crisis, when quickly identifying people and providing them with support is of paramount importance. Ongoing reforms to delivery systems in the study countries – such as investments in online applications, enrolment, and inter-operability with government databases to assess eligibility – are improving the efficiency and effectiveness of these systems. Several of the study countries had switched to digital payment procedures through the banking system before the pandemic. These systems largely worked well and were of particular value in a context where social distancing was required. And yet innovations in delivery systems, particularly in terms of outreach and application, differed across social assistance and labor programs, suggesting scope for learning among programs within countries and additional scope for innovation, particularly among social assistance programs. Finally, the delivery chain of a program is only as strong as its weakest link, which suggests that investments should be made at all steps from outreach to manage, including the surge capacity required to function effectively during a crisis. To enable an effective response to future crises, options for program design should be set out in advance to support vulnerable populations and meet their needs. A core challenge in responding to crises is to quickly identify those people in need of support because of the shock. Across Eastern Europe and the South Caucasus, countries acted pragmatically, rapidly expanding programs or introducing new programs to reach populations that were deemed to be vulnerable to the pandemic. Program designs were chosen in response to early understandings of the effects of the pandemic, with a strong focus on protecting the elderly and children, and workers who lost jobs. In some countries, such as Armenia and Georgia, modifications were introduced to program design as gaps in coverage emerged. This real- time decision making, based on available information, enabled a rapid response, although more thorough reviews point to important gaps in coverage across the population, such as 38 For a further discussion of reforms to strengthen the foundational social protection systems in the study countries, please see World Bank 2022c, World Bank 2022d, World Bank 2022e. 47 among poor and informal workers. This suggest that all countries could benefit from planning ahead for the next crisis by seeking to identify populations that may potentially be at risk from different types of crises and setting out, in advance, options for response in terms of target populations, objectives and design parameters. This is illustrated through the simulated response options to shocks in Moldova. Investments in data management, information systems, including social registries, can support speedier and more effective responses. Beyond the question of which populations or groups will require support is the mechanics of how these people will be identified, assessed, and enrolled when a crisis occurs. This issue seems to have featured strongly in the response to the COVID-19 pandemic in the study countries. A common feature across the countries and instruments is the use of existing data and information on beneficiaries or potential beneficiaries as the basis for the response: increases in social assistance were largely provided to existing beneficiaries of various programs; and new programs were introduced based on available information housed within tax authorities or other ministries. Rapidly collecting new information proved to be a challenge, even for those programs designed to be on-demand, namely the LRISs. Focusing investment on data and information systems and procedures to overcome this gap will be key to improving the effectiveness of the social protection response to future crises: examples of such investments are found in the COVID-19 response. Georgia and Moldova’s use of their social registries or integrated beneficiary registries played a pivotal role in quickly increasing the coverage of poverty- targeted social assistance programs to people whose data was already within these registries; expanding further to new people, that is enrolling them into the databases in real time, provided challenging. Box 9 presents the case of Chile, which used its social registry to reach 14 million people with emergency COVID-19 support. Ukraine’s widespread communication and digital applications for unemployment and sickness benefits offer insights into innovations that would be required to strengthen the on-demand nature of LRIS programs in the region and/or facilitate the enrolment of new people into registries that support these programs. Armenia’s automatic payments to workers registered with the tax authority, pregnant women recorded by the Ministry of Health, and schools based on student enrolment records held by the Ministry of Education point to how information collected by various ministries may be used to inform a response. 48 Box 9: Rapidly identifying affected households through a social registry Chile has one of the most advanced social protection systems in Latin America and the Caribbean. However, to address the effects of the COVID-19 pandemic, the government was forced to create several new schemes, as existing schemes were not equipped to cater for rapid declines in income and were ill-suited to incorporate new target groups. The government was able to harness its pre-existing Household Social Registry, make use of its interoperability with other government administrative databases, and thus rapidly identify, target, enroll, and provide new benefits to 14 million Chileans (73 percent of the population) in only a few months. The first payment of Emergency Family Income (IFE) to 1.3 million households was made on May 23, 2020, just two weeks after the first mandatory lockdown in the capital city Santiago. Source: Gonzalez 2021 More broadly, preparing for the next crisis requires a focus on investing in the adaptive elements of the social protection system that will increase the resilience of the population before, during and after a shock. The framework set out in the introduction to this report (Figure 1) draws attention to investments required in program design and delivery, but also in governance, data and information and financing. In addition to the recommendations emerging around the need to consider program design options in advance and invest in delivery systems to identify people for support, social protection systems can also be made more adaptive by setting out in advance the rules that would guide any such changes, such as modifications to eligibility criteria, system procedures, and benefit amounts. Support for staffing – including strengthening skills and knowledge, providing equipment and resources, and putting in place procedures to reallocate staff or hire temporary staff to support surges in capacity – are also required to ensure that these systems can “surge” when a crisis occurs. Sources of financing should be identified in advance and, if needed, be pre-positioned, including through disaster risk financing (Cubas, Gunasekera, and Humbert 2020). Most countries financed their responses to the COVID-19 pandemic, including those through social protection systems, using ex-post budgetary reallocations and additional public debt from international financial institutions or international financial markets. Box 10 describes the process the United Kingdom followed to finance its social protection response to the pandemic, which built on an annual budgeting process that flexibly financed social assistance programs based on actual demand. The use of ex-post budgetary reallocations can lead to an inadequate response, in terms of timing and the size of financing, with high opportunity costs. Additionally, international experience suggests that adopting a risk layering approach for disaster risk financing, which combines risk retention (such as budgetary allocations or contingent credit lines) and risk transfer (such as insurance), can ensure that funds are 49 available for rapid response to a crisis (Cubas, Gunasekera, and Humbert 2020). Finally, complementary investments are likely required in early warning systems and disaster preparedness and response, although further analysis is required to identify the entry points for a shift towards adaptive social protection. Box 10: Financing the social protection response to COVID-19: the case of the United Kingdom The natural expansion of existing programs was already covered by existing legislation, which allowed all those people who meet the eligibility requirements of a scheme to be enrolled and start benefiting. Spending on most benefits is not subject to departmental expenditure limits, but instead make up the ‘Annually Managed Expenditure’ portion of the United Kingdom Government’s budget, which largely relates to functions considered demand-led, such as welfare budgets. Adjustment to existing schemes did require secondary legislation in the form of new regulations. For example, the Employment and Support Allowance and Universal Credit (Coronavirus Disease) Regulations and the Social Security (Coronavirus) (Further Measures) Regulations allowed for increases to benefit amounts, and adjusted some of the key conditions for those claiming Universal Credit and its legacy predecessor benefits. While the regular budgetary process has some built-in flexibility for natural expansion of existing welfare schemes, the huge increase in funding requirements and the introduction of policy changes and new schemes meant that additional processes were necessary. The March 2020 Contingencies Fund Act approved the use of up to GBP 260 billion in emergency funding (an increase from the previously approved level of contingency funding from 2 percent to 50 percent), but approval for the use of the Contingency Fund is temporary and actual spending still required approval by parliament through the normal estimates process. The Supply and Appropriation Act 2020 (May 2020) and the February 2021 Supplementary Estimates were the mechanisms for parliamentary approval of spending and the resulting deficits in the Financial Year 2020/2021. Source: Sandford 2021 Finally, it is essential to continuously monitor the effectiveness of the response, across social assistance, social insurance and labour programs, to support ongoing decision making on the magnitude and duration of emergency assistance. Evidence-based policy making is central for strengthening the effectiveness of the overall social protection system and its response to future crises. For this, monitoring and evaluation systems are required to assess the performance of programs, their coverage, adequacy and impacts. These same systems 50 should be used to assess the speed and quality of a response to a crisis, to enable a critical review of program design and well as the performance of delivery systems. A first step towards such evidence-based policymaking would be a review of the effectiveness and efficiency of the wage subsidy programs that were used extensively in some of the study countries in response to the COVID-19 pandemic, given that the current level of information available on the functioning of these programs is thin. Well-functioning grievance and redress mechanisms can support this by creating feedback loops between applicants, beneficiaries and institutions. 39 While the social protection response offered immediate relief, programs could also be harnessed to help mitigate the long-term scars that are emerging in human capital. The initial emergency response in all countries was modest and time-bound, often only providing support during the state of emergency in the early months of the pandemic. And yet the pandemic continued over multiple years, with the long-term effects on human capital only beginning to emerge. The emergency support provided through the social protection systems offered some immediate relief to households and workers. Situating this short-term emergency support within a broader, longer-term response could help to mitigate the impacts on human capital, particularly among poor and vulnerable households. Such a strategy would, for example, provide dedicated support to poor households to consider their eligibility for the LRIS for longer-term support (possibly with modifications to eligibility to enable more people to enter for a given period of time) or a transition strategy from unemployment benefits to active labour market programs as job seekers re-enter the labour market. Social protection programs could also support longer-term recovery efforts led by health and education stakeholders, for example, by providing additional support or incentives for children to attend school or participate in remedial sessions. Bastagli and Lowe 2021 provide a more thorough discussion of the role of grievance and redress 39 mechanisms. 51 References Aminjonov, Ulugbek, and Tanguy Bernard. 2021. “Gimme Shelter: Social Distancing and Income Support in Times of Pandemic.” Ando, Sakai, Ravi Balakrishnan, Bertrand Gruss, Jean-Jacques Hallaert, La-Bhus Fah Jirasavetakul, Koralai Kirabaeva, Nir Klein, et al. 2022. “European Labor Markets and the Pandemic.” IMF Departmental Paper. Bastagli, Francesca, Jessica Hagen-Zanker, Luke Harman, Valentina Barca, Georgina Sturge, Tanja Schmidt, and Luca Pellerano. 2016. “Cash Transfers: What Does the Evidence Say? A Rigorous Review of Programme Impact and of the Role of Design and Implementation Features.” ODI Report, no. July: 1–300. https://odi.org/en/publications/cash-transfers-what-does-the-evidence-say-a- rigorous-review-of-impacts-and-the-role-of-design-and-implementation-features/. Bastagli, Francesca, and Christy Lowe. 2021. “Social Protection Response to Covid-19 and beyond: Emerging Evidence and Learning for Future Crises.” ODI Working Paper 614. Bowen, Thomas, Carlo del Ninno, Colin Andrews, Sarah Coll-Black, Ugo Gentilini, Kelly Johnson, Yasuhiro Kawasoe, Adea Kryeziu, Barry Maher, and Asha Williams. 2020. Adaptive Social Protection: Building Resilience to Shocks. Adaptive Social Protection: Building Resilience to Shocks. https://doi.org/10.1596/978-1-4648-1575-1. Cubas, Diana, Rashmin Gunasekera, and Thibaut Humbert. 2020. “Disaster Risk Finance for Adaptive Social Protection.” Disaster Risk Finance for Adaptive Social Protection. https://doi.org/10.1596/34133. Demirgüç-Kunt, Asli, Leora Klapper, Dorothe Singer, Saniya Ansar, and Jake Hess. 2018. “The Global Findex Database 2017: Measuring Financial Inclusion and the Fintech Revolution.” https://datacatalog.worldbank.org/int/search/dataset/0039935. Demirgüç-Kunt, Asli, Michael M. Lokshin, and Iván Torre. 2022. “Protect Incomes or Protect Jobs? The Role of Social Policies in Post-Pandemic Recovery.” World Bank Policy Research Working Paper 10166. Dikmelik, Aylin Isik-. 2012. “Do Social Benefits Respond to Crises ? Evidence from Europe & Central Asia During the Global Crisis,” no. 1219. 52 Eurostat. 2022. “European Neighborhood Policy - East - Labor Market Statistics.” 2022. https://ec.europa.eu/eurostat/statistics- explained/index.php?title=European_Neighbourhood_Policy_-_East_- _labour_market_statistics#Employment_by_economic_activity. Gentilini, Ugo, Mohamed Almenfi, and Pamela Dale. 2022. “Social Protection and Jobs Responses to COVID-19: A Real-Time Review of Country Measures. ‘Living Paper’ Version 16. Washington, D.C.: World Bank.” Social Protection and Jobs Global Practice. Gentilini, Ugo, Mohamed Almenfi, Pamela Dale, Robert Palacios, Harish Natarajan, Gustavo Demarco, Indhira Santos, et al. 2020. “Social Protection and Jobs Responses to COVID- 19 : A Real-Time Review of Country Measures. Version 16.” 13. Gentilini, Ugo, Maddalena Honorati, Ruslan Yemtsov, Session Brief, Session Summary, Session Framing, Kathy Lindert, et al. 2012. “Managing Risk, Promoting Growth: Developing Systems for Social Protection in Africa. The World Bank’s Africa Social Protection Strategy 2012-2022.” The World Bank’s Africa Social Protection Strategy 2012-2022, no. May: 135. http://siteresources.worldbank.org/INTAFRICA/Resources/social-protection-full- report-EN-2012.pdf. Gonzalez, Alejandro. 2021. “Social Protection Response to the COVID-19 Pandemic in Chile,” Unpublished World Bank high-income case study. Grosh, Margaret E., Maurizio Bussolo, and Samuel Freije. 2014. Understanding the Poverty Impact of the Global Financial Crisis in Latin America and the Caribbean. Guha-Sapir, D., R. Below, and Ph. Hoyois. 2022. “EM-DAT: The CRED/OFDA International Disaster Database - Www.Emdat.Be.” Université Catholique de Louvain – Brussels – Belgium. Guven, Melis, Himanshi Jain, and Clement Joubert. 2021. “Social Protection for the Informal Economy. Operational Lessons for Developing Countries in Africa and Beyond.” World Bank Group. Hallegatte, Stephane, Adrien Vogt-Schilb, Mook Bangalore, and Julie Rozenberg. 2017. Unbreakable: Building the Resilience of the Poor in the Face of Disasters. Hatayama, Mayo, Yiruo Li, and Theresa Osborne. 2021. “Understanding and Predicting Job Losses Due to COVID-19 : Empirical Evidence from Middle Income Countries.” 53 Hernandez, Carlos, Jamele Rigolini, Ana Maria Oviedo, and Sarah Coll-Black. 2022. “Protecting Who? Social Protection Responses to Shocks with Limited Information.” Draft, Forthcoming. Honorati, Maddalena, Soonhwa Yi, and Thelma Choi. 2020. “Assessing the Vulnerability of Armenian Temporary Labor Migrants during the COVID-19 Pandemic.” Assessing the Vulnerability of Armenian Temporary Labor Migrants during the COVID-19 Pandemic, no. 2003. https://doi.org/10.1596/34359. Huitfeldt, Henrik. 2020. Towards Better Adequacy & Sustainability: A Review of Pension Systems & Pension Reforms in Eastern Partnership Countries. Vol. 8022. https://doi.org/10.2765/972935. ILO. 2022. “COVID-19 and the World of Work: Country Policy Responses.” 2022. https://www.ilo.org/global/topics/coronavirus/regional-country/country-responses. ———. 2023. “Working Hours Lost Due to the COVID-19 Crisis, by Sex -- ILO Modelled Estimates (%) -- Annual” HOW_2LSS_S. IMF. 2020. “Republic of Armenia: Third Review under the Stand-By Arrangement and Modification of Performance Criteria.” ———. 2021a. “Georgia: 2021 Article IV Consultation Staff Report.” ———. 2021b. “Republic of Armenia: 2021 Article IV Consultation, Fourth and Fifth Review under the Stand-By Arrangement.” ———. 2021c. “Republic of Azerbaijan: 2021 Article IV Consultation Staff Report.” ———. 2022a. “Montenegro: 2021 Article IV Consultation-Press Release; Staff Report; and Statement by the Executive Director for Montenegro.” IMF Staff Country Reports 2022 (060): 1. https://doi.org/10.5089/9798400201332.002. ———. 2022b. “Republic of Moldova: 2021 Article IV Consultation.” IMF Staff Country Reports. International Labour Organization. 2017. Migrant Workers: The Case of Moldova. Lindert, Kathy, Tina George Karippacheril, Ines Rodriguez Caillava, and Kenichi Nishikawa Chavez. 2020. Sourcebook on the Foundations of Social Protection Delivery Systems. 54 Mackley, Andrew. 2022. “Coronavirus: Universal Credit during the Crisis.” House of Commons Library Brief Paper No 8999. Manea, Madalina Elena, and Razvan Dumitru. 2020. “Review of the Ajutor Social Benefits Delivery to Improve the Uptake of the Poor.” https://imagebank2.worldbank.org/search/32124145. Moffitt, Richard. 2014. “Unemployment Benefits and Unemployment.” IZA World of Labor, no. May: 1–10. https://doi.org/10.15185/izawol.13. New York Times. 2020. “Paid to Stay Home: Europe’s Safety Net Could Ease Toll of Coronavirus,” 2020. https://www.nytimes.com/2020/03/06/business/europe- coronavirus-labor-help.html. Oxford Policy Management. 2015. “Shock-Responsive Social Protection Systems Research: Conceptualising Shock-Responsive Social Protection.” Rigolini, Jamele, Sarah Coll-Black, Rafael De Hoyos, and Ha Thi Hong Nguyen. 2023. “Protecting Human Capital Through Shocks and Crises: How Lessons Learned from the COVID-19 Response across Eastern Europe and the South Caucasus Can Be Used to Build Better and More Resilient Human Development Systems.” Washington, D.C. https://openknowledge.worldbank.org/handle/10986/39716. Robalino, David A., David Newhouse, and Friederike Rother. 2013. Labor and Social Protection Policies during the Crisis and the Recovery. Working through the Crisis: Jobs and Policies in Developing Countries during the Great Recession. https://doi.org/10.1596/978-0-8213-8967-6_ch4. Ross, Antrim, and Anna Clarke. 2021. “Credit Where It’s Due? Claiming Universal Credit during the Covid-19 Pandemic.” Peabody. Sandford, Judith. 2021. “Social Protection Response to the COVID-19 Pandemic in the UK,” Unpublished World Bank high-income case study. Smith, Gabrielle, and Thomas Bowen. 2020. “Adaptive Social Protection: The Delivery Chain and Shock Response.” Truman, Packard, Gentilini Ugo, Grosh Margaret, O’Keefe Philip, Palacios Robert, Robalino David, and Santos Indhira. 2019. Protecting All: Risk Sharing for a Diverse and Diversifying World of Work. 55 UNDRR. 2019. “Global Assessment Report on Disaster Risk Reduction 2019.” https://www.undrr.org/publication/global-assessment-report-disaster-risk-reduction- 2019. USAID. 2016. “Climate Factsheets: Georgia, Armenia, Azerbaijan, Ukraine,” no. Washington DC: USAID. Viollaz, Mariana, Mauricio Salazar-Saenz, Luca Flabbi, Monserrat Bustelo, and Mariano Bosch. 2022. “The COVID-19 Pandemic in Latin American and Caribbean Countries: The Labor Supply Impact by Gender.” IZA Discussion Paper No. 15091. World Bank. 2009. “Central Asia and Caucasus Disaster Risk Management Initiative (CAC DRMI): Risk Assessment for Central Asia and Caucasus Desk Study Review.” ———. 2014. “The Role of Last-Resort Income Support in the Eastern Europe and Central Asia Region.” ———. 2018. The State of Social Safety Nets 2018. Washington, DC: The World Bank. http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-1254-5. ———. 2020a. Europe and Central Asia Economic Update, Spring 2020: Fighting COVID-19. Europe and Central Asia Economic Update, Spring 2020: Fighting COVID-19. https://doi.org/10.1596/978-1-4648-1564-5. ———. 2020b. “Poverty and Welfare Impacts of COVID-19 and Mitigation Policies in Armenia.” ———. 2020c. “Poverty Impacts of COVID-19 and Mitigation Policies in Azerbaijan.” ———. 2020d. “Strengthening Moldova’s Disaster Risk Management and Climate Resilience: Facing Current Issues and Future Challenges.” https://openknowledge.worldbank.org/handle/10986/35318. ———. 2020e. “Towards Adaptive Social Protection Systems in Latin America and the Carribean.” ———. 2020f. “Urban and Disaster Risk Management Responses to COVID-19.” http://pubdocs.worldbank.org/en/575581589235414090/World-Bank-Urban- DRM- COVID-19-Responses.pdf. 56 ———. 2020g. “Welfare Impacts of the COVID-19 Pandemic in Moldova: Main Channels and Impact of Mitigation Measures.,” no. 4. ———. 2021a. “Country Background Notes,” Internal distribution only. ———. 2021b. “Overlooked: Examining the Impact of Disasters and Climate Shocks on Poverty in the Europe and Central Asia Region.” ———. 2021c. “Poverty and Welfare Impacts of COVID-19 and Mitigation Policies in Georgia.” ———. 2022a. A New Look at Old Dilemmas: Revisiting Targeting in Social Assistance Contents. ———. 2022b. “ASPIRE: The Atlas of Social Protection Indicators of Resilience and Equity.” Internal Database. https://www.worldbank.org/en/data/datatopics/aspire. ———. 2022c. “Charting a Course Towards Universal Social Protection: Equity, Resilience, and Opportunity for All.” ———. 2022d. “Energy Crisis: Protecting Economies and Enhancing Energy Security in Europe and Central Asia.” Washington, DC. © World Bank. https://openknowledge.worldbank.org/handle/10986/38101. ———. 2022e. “Human Capital Review Armenia.” Washington, D.C. http://documents.worldbank.org/curated/en/099082001272340595/P1735300d500e8 0b20840902a6e742162a6. ———. 2022f. “Human Capital Review Azerbaijan.” Washington, D.C. http://documents.worldbank.org/curated/en/099082001272341155/P1735301c30e9a 0b1efa3147da1823d147928a86e81a. ———. 2022g. “Human Capital Review Georgia.” Washington, D.C. http://documents.worldbank.org/curated/en/099435008172221325/P1735300c417d2 026096d50dd8d8218cd90. ———. 2022h. “Responsive by Design: Building Adaptive Social Protection Systems in South Asia,” Draft. Forthcom. ———. 2022i. “Social Protection Evaluation and Expenditures Database (SPEED).” 57 ———. 2022j. “Ukraine Crisis Social Protection Briefing Note (Internal),” Internal distribution only. ———. 2023. “Poverty and Inequality Platform (Version 20220909_2017_01_02_PROD).” Pip.Worldbank.Org. 2023. ———. “Poverty and Inequality Platform (version 20220909_2017_01_02_PROD) [data set],” pip.worldbank.org. 2023. Accessed on 2023-03-02 58 Annex 1: Emergency social protection programs adopted during the pandemic Increased resources were devoted to raising the level of payments from poverty-targeted programs to beneficiaries, increasing the coverage of existing programs, and initiating new programs. All countries expanded last-resort income support programs, by increasing the level of benefits (Armenia, Georgia, Moldova and Ukraine), allowing automatic extensions for existing participants (all the countries except Armenia), increasing the threshold of household income used in eligibility determinations (Georgia, Moldova and Ukraine)40 or reducing the documentation required to participate (Georgia and Moldova). Ukraine made those who lost jobs in the pandemic eligible for GMI and persons under lockdown eligible for disability assistance, while Armenia provided additional payments to vulnerable families participating in other social assistance programs. New programs provided one-time payments to low- income individuals (Azerbaijan and Moldova), disability beneficiaries (Ukraine), pregnant women and students (Armenia) and all children (Georgia). Other approaches to supporting incomes included efforts to control the prices of key foods (Ukraine and Georgia); and subsidies to cover utility bills (Armenia, Azerbaijan, Georgia and Ukraine), food purchases (Georgia) and tuition (Azerbaijan and Georgia). Most countries expanded support to unemployed workers, and some to pensioners. Armenia, Georgia, and Ukraine provided income support to workers (in some cases also entrepreneurs) who lost work during the pandemic. Azerbaijan extended the eligibility period for unemployment insurance for current beneficiaries. Moldova and Ukraine increased unemployment benefits under existing schemes and extended benefits to workers who were not formerly eligible (for example, in Moldova returning migrants and entrepreneurs without income). Pensions were increased or one-time payments were made to pensioners in Georgia, Moldova and Ukraine. And pension rules were modified to index benefits to inflation (Georgia, Moldova and Ukraine). Table A1 provides a summary of unemployment systems before the crisis. 40 Moldova also changed the formula used to calculate the income threshold. 59 Table A1: Unemployment systems before the COVID-19 crisis Azerbaijan Moldova Ukraine All residents of Azerbaijan whose employment contracts Employed persons, including had been terminated because casual workers, the self- Coverage of the liquidation of a state Employees. employed and military agency or legal entity or as a personnel. result of shedding workforce or staff reduction. Registered with the state Registered with the state employment services, employment services, actively seeking work, willing Registered with the state Eligibility actively seeking work, willing to work and aged between 16 employment service as a requirements to work and aged between 15 and pensionable age, not jobseeker. and pensionable age. studying in a form of full-time education Employed for 12 months in 26 calendar weeks of full or the 24 months preceding Employed for 12 months in part-time employment in the Qualifying period unemployment and with at the 24 months preceding 12 months preceding least 3 years of social unemployment. unemployment. insurance records. Minimum is the minimum subsistence level. Maximum Minimum is minimum Minimum is the legal monthly Minimum and benefit is the preceding monthly wage. Maximum is minimum wage. Maximum is maximum benefit month's national average the national average monthly the national average monthly amount earnings of the sector in salary. salary. which the recipient was previously employed. 1-10 years of contributions: 5 Duration of unemployment First application: six months; months; 10-15 years of benefits may not exceed 360 Pay-out period Repeated applications: three contributions: 7 months; 15+ calendar days during a two- months. years of contributions: 9 year period. months Between 50-70% of previous average earnings depending Between 50-60% of the on length of contribution. Between 40-50% depending average monthly salary Benefits reduce as time goes Replacement rate on who decided to terminate depending on length of on. 100% of the benefit for the working relationship. contribution. the first 90 days, 80% for the subsequent 90 calendar days and 70% thereafter. Source: MISSCEO comparative tables (Accessible at www.missceo.coe.int) and Social Security Administration - Social Security Programs Throughout the World (Accessible at www.ssa.gov/policy/docs/progrdecs/ssptw). Note: Armenia and Georgia had no functional unemployment insurance prior to the COVID-19 crisis. 60 Subsidies were used to retain jobs in all the study countries except Moldova. Subsidies took the form of direct payments to firms, tax exemptions and coverage of social security payments in salary subsidies. Efforts were also made to create jobs in agriculture and through expanding public works programs. There was little emphasis on social services programs during the crisis, likely reflecting pandemic restrictions, difficulties in scaling up service provision and funding shifts to social assistance and unemployment compensation. Azerbaijan provided free delivery of food for the elderly. Georgia established the State Employment Support Agency (SESA) in 2019 to provide job intermediation services, and switched to remote delivery of social services but stopped taking in new beneficiaries temporarily. 61 Table A2: Overview of main social protection policy responses to COVID-19 Armenia Azerbaijan Georgia Moldova Ukraine Social HE: One-time child HE: Lump sum HE: Assistance to VE: Increased Guaranteed HE: Relaxed eligibility rules for GMI assistance allowance. Payments for payments for low- individuals who lost Minimum Income (GMI) and and program extended to those measures employees who lost jobs. income individuals. jobs or were on unpaid indexed benefits to inflation. who lost jobs during quarantine. Categorical payments for Food assistance. leave. Allowances for Raised child benefit adult pregnant women, low- For vulnerable all children under age equivalency coefficient. Payments to pensioners, and income families, households, 18. Subsidies to cover workers and entrepreneurs who students, and payments for utility bills and food. HE: Payments to pensioners lost jobs. Simplified and eased workers/entrepreneurs utility bills and Price stabilization for and low-income individuals. restrictions on enrollment for in affected sectors. tuition. selected foods. Payments to some dismissed Housing and Utilities Subsidy Subsidies for utility bills. workers, returning migrants, (HUS). Price controls on critical HE and VE: Households and informal workers. products. Extended eligibility for VE: Support programs for eligible for targeted temporary disability assistance to vulnerable families social assistance those under quarantine. One-time participating in SA expanded and allowance for families of health programs. additional benefits workers who died from COVID-19. provided to some participating DT: Increased reliance on digital households. processes for applications and payment of benefits for SA. DT: Simplified rules for administration and VE: Increased duration of GMI disbursement of payments. Automatic re- targeted social registration of GMI and HUS assistance and the participants. social package Temporarily increased HUS benefits and eased compliance rules. Payments for disability beneficiaries. 62 Social Expanded Pensions increased and Unemployment benefit Minimum unemployment benefit insurance coverage of indexed to inflation, increased. All unemployed and coverage increased; measures unemployment and rules for persons, including returning administration improved. insurance and administration and migrants, made eligible. Registration requirements for increased period disbursement Workers who lost job during unemployment insurance eased. of eligibility. simplified. state of emergency not Pension benefits increased and required to participate in indexed to inflation. compulsory health insurance. Registration for unemployment shifted to remote system. Unemployment benefit provided to entrepreneurs without incomes. Pension benefits indexed to inflation. Job One-time salary subsidy Payments to retain Tax exemptions for Firms allowed to adopt more retention for firms/entrepreneurs jobs and maintain employers who flexible working hours. Subsidies / wage engaged in most affected wages, and to retained jobs and to SMEs who suspended support sectors. Package to cover social payments to self- operations, based on salaries. schemes create jobs in agriculture. security payments. employed. Tax Job creation in concessions and public sector. financial support for micro, small and medium-sized firms in affected sectors. Source: World Bank 2021c. Notes: HE = horizontal expansion; VE = vertical expansion; DT = design tweak 63 Social Protection & Jobs Discussion Paper Series Titles 2021-2023 No. Title June 2023 2304 Social protection in a world of crisis: Learning from the response to the COVID-19 pandemic in Eastern Europe and the South Caucasus April 2023 2303 Social Protection Program Spending and Household Welfare in Ghana by Dhushyanth Raju, Stephen D. Younger, and Christabel Dadzie 2302 Digital-First Approach to Emergency Cash Transfers: Step-Kin in the Democratic Republic of Congo by Anit Mukherjee, Laura Bermeo, Yuko Okamura, Jimmy Vulembera, and Paul Bance March 2023 2301 Productive Inclusion Programs in Urban Africa by Jorge Avalos, Thomas Bossuroy, Timothy Clay, and Puja Vasudeva Dutta December 2022 2215 Tracking Global Social Protection Responses to Inflation. Living paper v.4 by Ugo Gentilini, Mohamed Almenfi, Hrishikesh TMM Iyengar, Giorgia Valleriani, Yuko Okamura, Emilio Raul Urteaga, and Sheraz Aziz 2214 Childcare Arrangements for Low-Income Families: Evidence from Low- and Middle-Income Countries by Alisa Currimjee, Jem Heinzel-Nelson Alvarenga Lima, Sara Troiano November 2022 2213 Affordable childcare: A needs assessment of low-income mothers and childcare providers in urban Bangladesh by Fahmina Rahman, Anmol Kamra October 2022 2212 Guiding Social Protection Targeting Through Satellite Data in São Tomé and Príncipe by Peter Fisker, Jordi Gallego-Ayala, David Malmgren-Hansen, Thomas Pave Sohnesen, and Edmundo Murrugarra September 2022 2211 Tracking Global Social Protection Responses to Price Shocks (Living paper v.3) by Ugo Gentilini, Mohamed Almenfi, Hrishikesh TMM Iyengar, Yuko Okamura, Emilio Raul Urteaga, Giorgia Valleriani, and Sheraz Aziz July 2022 2210 Tracking Global Social Protection Responses to Price Shocks (Living paper v.2) by Ugo Gentilini, Mohamed Almenfi, Hrishikesh TMM Iyengar, Yuko Okamura, Emilio Raul Urteaga, Giorgia Valleriani, Jimmy Vulembera Muhindo, and Sheraz Aziz June 2022 2209 Tracking Social Protection Responses to Displacement in Ukraine and Other Countries by Ugo Gentilini, Mohamed Almenfi, Hrishikesh TMM Iyengar, Yuko Okamura, Emilio Raul Urteaga, Giorgia Valleriani, Jimmy Vulembera Muhindo, and Sheraz Aziz April 2022 2208 Tracking Global Social Protection Responses to Price Shocks (Version 1) by Ugo Gentilini, Mohamed Almenfi, Hrishikesh TMM Iyengar, Yuko Okamura, Emilio Raul Urteaga, Giorgia Valleriani, Jimmy Vulembera Muhindo and Sheraz Aziz 2207 Adapting Social Protection to FCV and Insecurity - The Case of the Democratic Republic of Congo by Silvia Fuselli, Mira Saidi, Afrah Alawi Al-Ahmadi 2206 Voluntary Savings Schemes to Protect Informal Workers in Jordan by Friederike Rother, Carole Chartouni, Javier Sanchez-Reaza, Ernesto Brodersohn, Montserrat Pallares-Miralles 2205 Enhancing Workers’ Protection in Jordan by Friederike Rother, Carole Chartouni, Javier Sanchez-Reaza, Gustavo Paez Salamanca, Belal Fallah 2204 Humanitarian and Social Protection Linkages with Examples from South Asia by Steen Lau Jorgensen and Maria Virginia Ceretti March 2022 2203 Cash in the City: The Case of Port-au-Prince by Olivia D’Aoust, Julius Gunneman, Karishma V. Patel and Caroline Tassot January 2022 2202 Tracing Labor Market Outcomes of Technical and Vocational Training Graduates in Saudi Arabia: A study on graduates from the Technical and Vocational Training Corporation (TVTC) by Nayib Rivera, Mehtabul Azam and Mohamed Ihsan Ajwad 2201 From Protracted Humanitarian Relief to State-led Social Safety Net System: Somalia Baxnaano Program by Afrah Al-Ahmadi and Giuseppe Zampaglione December 2021 2110 Early Lessons from Social Protection and Jobs Response to COVID-19 In Middle East and North Africa (MENA) Countries by Alex Kamurase and Emma Willenborg 2109 Migration in Bulgaria: Current Challenges and Opportunities by Daniel Garrote-Sanchez, Janis Kreuder, and Mauro Testaverde 2108 Social Protection and Labor: A Key Enabler for Climate Change Adaptation and Mitigation by Jamele Rigolini June 2021 2107 Intent to Implementation: Tracking India’s Social Protection Response to COVID-19 by Shrayana Bhattacharya and Sutirtha Sinha Roy 2106 Social Assistance Programs and Household Welfare in Eswatini by Dhushyanth Raju and Stephen D. Younger May 2021 2105 The Coal Transition: Mitigating Social and Labor Impacts by Wendy Cunningham and Achim Schmillen April 2021 2104 Social Protection at the Humanitarian-Development Nexus: Insights from Yemen by Yashodhan Ghorpade and Ali Ammar January 2021 2103 Review of the Evidence on Short-Term Education and Skills Training Programs for Out-of-School Youth with a Focus on the Use of Incentives by Marguerite Clarke, Meghna Sharma, and Pradyumna Bhattacharjee 2102 Welfare, Shocks, and Government Spending on Social Protection Programs in Lesotho by Joachim Boko, Dhushyanth Raju, and Stephen D. Younger 2101 Cash in the City: Emerging Lessons from Implementing Cash Transfers in Urban Africa by Ugo Gentilini, Saksham Khosla, and Mohamed Almenfi To view Social Protection & Jobs Discussion Papers published prior to 2021, please visit www.worldbank.org/sp. ABSTRACT This paper explores the social protection response to the COVID-19 pandemic in Armenia, Azerbaijan, Georgia, Moldova and Ukraine to learn lessons on how to build the resilience of their social protection system. These countries made substantial efforts to address the most serious consequences of the pandemic, pragmatically harnessing existing programs to reach vulnerable groups, while also introducing innovations to fill gaps in the existing social protection system. Rigidities in administrative systems, complex eligibility criteria, as well as weaknesses in information systems, limited governments’ ability to quickly identify and reach those households that were most vulnerable to the impact of the pandemic with adequate support. These challenges strengthen the case for investment in crisis preparedness – most immediately by improving the functioning of social protection systems and setting out the design features and delivery systems to support a response to future covariate shocks. ABOUT THIS SERIES Social Protection & Jobs Discussion Papers are published to communicate the results of The World Bank’s work to the development community with the least possible delay. This paper therefore has not been prepared in accordance with the procedures appropriate for formally edited texts. For more information, please contact the Social Protection Advisory Service via e-mail: socialprotection@ worldbank.org or visit us on-line at www.worldbank.org/sp