Poverty and welfare impacts of COVID-19 and mitigation policies in Georgia1 This note summarizes the results of simulations of the poverty and welfare impacts of the COVID-19 pandemic in Georgia, undertaken by the Poverty and Equity team of the World Bank (WB). First, the poverty and distributional effects are estimated using a microsimulation model of household income shocks. Next, selected mitigation measures implemented by the Government of Georgia (GoG) are simulated to assess their effectiveness to improve household welfare and to prevent impoverishment from the economic impacts of COVID-19. Lastly, simulations based on short- and medium-term macroeconomic projections are reported.2 Key Takeaways • The short-term impact of COVID-19 on household incomes in Georgia could significantly increase poverty in 2020 in the absence of appropriate mitigation measures. The 2020 poverty rate could increase by 9 percentage points, from a projected 38% (counterfactual scenario) to 47% after the economic shocks derived from the pandemic (considering the upper middle-income class poverty line of US$5.5 a day, 2011 PPP). The incidence of extreme poverty in the population could more than double to 7.4%. The national poverty rate could be higher in 2020 than its 2013-level.3 • The policy measures implemented by the GoG to mitigate the economic shocks of COVID-19 on household incomes have progressive effects on the population. Combined together, the mitigation measures are significant, though insufficient, to fully revert the impoverishment effects of the pandemic. The combined effect of the analyzed mitigation policies would reduce or revert poverty by up to 4.7 percentage points (national poverty line). • Social protection and social assistance mechanisms (pensions and TSA) remain essential to reduce impoverishment. • The economic shock resulting from COVID-19 could impoverish 350 thousand people in Georgia,4 and force over 800 thousand people to suffer downward mobility, transitioning to a lower-income group. • The identified mitigation measures could lift approximately 120 thousand people out of poverty, considering the upper middle-income class poverty line (or up to 175 thousand people out of poverty defined by the national poverty line). • The analysis in this note is based on available information as of November 2020. Hence, results reflect trends in the impact of COVID-19 and mitigation policies observed between March and November 2020. As more information becomes available, the analysis will be updated and refined. • Lessons from the available simulations suggest that Georgia may be able to target vulnerable and impoverished groups by a combination of labor market actions and social safety nets. Interestingly, the set of policies implemented by the GoG to mitigate the effects of COVID-19 complement each other. They provide coverage to households located at different points of the income distribution and respond to different needs. The implemented programs have served as important mitigation vehicles. However, as the situation continues to evolve, a swift and sustainable recovery will fundamentally require more information and policy flexibility to reach and support affected sectors and populations. 1 Produced by the South Caucasus Poverty and Equity Team led by Alan Fuchs (afuchs@worldbank.org) and including Natsuko Kiso Nozaki (nkiso@world bank.org) and Maria F Gonzalez Icaza (mgonzalezicaza@worldbank.org), under guidance of Sebastian-A Molineus (Country Director, ECCSC) and Salman Zaidi (Practice Manager, EECPV). The team received useful comments from Evgenij Najdov, Mariam Dolidze and Arvind Nair. All errors are our own. 2 The aim of the note is to assess the impact of COVID-19 on welfare and share good experiences on timely and evidence-based policy responses to rapidly changing circumstances. Analysis can be updated once additional information becomes available to the team. 3 Georgia was classified as upper-middle income country (UMIC) for the first time in 2017, and again in 2020 (based on Gross National Income per capita). The upper middle-income class poverty line ($5.50, 2011 PPP) is the international poverty threshold corresponding to UMICs. However, Georgia was classified as lower-middle income (LMIC) in 2018-2019. Hence, the analysis also applies alternative poverty thresholds (lower-income class poverty line [$3.20, 2011 PPP] and national poverty line) for comparability. The international extreme poverty line is $1.90 per capita per day (2011 PPP). 4 As measured by the upper middle-income class poverty line of $5.50 per capita per day (2011 PPP). Page 1 Background • Georgia responded swiftly to the outbreak of COVID-19, declaring national state of emergency and curfews between March 21 and May 22, 2020. Strict containment measures, lockdowns of high-risk districts, businesses and school closures, and bans on border crossings contributed to milder health impacts, compared to regional peers (IMF 2020). • However, economic recovery after reopening has been slow, and the number of active cases accelerated and increased 10-fold in September to October 2020. • The pandemic has contributed to important economic slowdowns in key sectors of the economy, including tourism (which accounts for 8% of GDP). Remittances inflows in the region are expected to decrease by 27.5% in 2020 (World Bank 2020). The initial forecast of 5% growth of GDP in Georgia during 2020—held before the pandemic outbreak—has been updated to -6.0% (World Bank October 2020). Mitigation measures • The GoG announced an assistance package worth GEL 3.4 billion (close to 7% of GDP) to mitigate the socio-economic impacts of COVID-19 (IMF 2020 and MTI). Assistance includes direct government spending, state-sponsored loans and increased investments.5 The supplemental budget approved to cover these initiatives is expected to increase the fiscal deficit to 8.5% of GDP in 2020. As of October 2020, 686 thousand people had benefited from the initiatives (GoG). • A review of available sources was conducted to identify mitigation measures directly targeting Georgian households. The identified measures are summarized in Appendix 1. The combined fiscal budget of identified measures is over GEL 750 million, according to the latest information accessible. Other measures beyond the scope of this exercise include tax waivers; public spending for the medical response, and programs directed to the private sector (credit guarantees, interest subsidies, microgrants, special support for hospitality, construction and agriculture sectors, etc.). Similarly, some of the new mitigation measures to be introduced after November 2020 to support citizens and the private sector amid a second wave of the pandemic were not fully incorporated due to lack of detailed available information. Data and methodology • The analysis is based on data from the 2018 Household Incomes and Expenditure Survey (HIES), the national household budget survey collected by the National Statistics Office of Georgia (Geostat).6 Additional macroeconomic parameters are retrieved from forecasts produced by the Macroeconomics Trade and Investment (MTI) Global Practice, the Global Knowledge Partnership on Migration and Development (KNOMAD), and Geostat. Inputs on the eligibility of Target Social Assistance were provided by Social Protection and Jobs (SPJ) specialists. Additional monitoring documents published by the SPJ Global Practice and the International Monetary Fund (IMF) were used to identify mitigation measures. • The identification strategy of the impact of COVID-19 relies on the comparison of a counterfactual scenario—what would poverty and inequality be in 2020 in the absence of COVID-19?—to a “shock” scenario—what will poverty and inequality be in 2020 after the effects of COVID-19. • First, the counterfactual scenario is based on a nowcasting method to predict welfare in 2020, based on the macroeconomic forecasts held before the outbreak, as of February 2020 (hence, excluding the effects of COVID-19), • Second, two complementing simulation methodologies are implemented to estimate the COVID-19 shock scenarios. I. Microsimulation. Using a bottom-up approach to identify different impacts on households’ labor and non-labor incomes. The estimated impacts include unemployment and labor income losses by subsector of employment; remittances inflows, and agricultural sales. The microsimulations incorporate heterogenous and distributional effects of COVID-19, as well as the mitigation effects of policy measures. 5This package is above par when compared to EMDEs. The GoG announced new measures in late November 2020, coinciding with new restrictions. 6The HIES is representative of the national population. It is collected quarterly and used by the GoG and the World Bank to measure annual poverty and inequality indicators. In 2018, the sample included 11,056 households. Page 2 II. Macrosimulation. Based on macroeconomic projections for GDP growth for 2020 produced by MTI, as of October 2020 (hence, including the effects of COVID-19), and a growth-poverty elasticity. This method assumes a top- down neutral distribution of growth. Due to this neutral-distribution assumption, macrosimulations reflect poverty impacts but do not account for potential inequality or distributional effects of COVID-19. However, they can be informative of general-equilibrium effects in the short- to medium-term. • All methodological details and assumptions applied are summarized in Appendix 2. Microsimulation results • Results from microsimulations suggest that the income shocks derived from the pandemic could significantly increase the headcount poverty rates in Georgia by up to 7 to 10 percentage points (Table 1).7 The simulated effects under each poverty thresehold are illustrated in Figure 1 in the Appendix. The microsimulation model also forecasts a significant increase in inequality, increasing the Gini coefficient from 36.4 to 38.9, after COVID-19. Table 1. Main poverty and inequality results Counterfactual COVID-19 Microsimulation Poverty Headcount Poverty Gap Gini Poverty Headcount Poverty Gap Gini $1.90 (2011 PPP) 3.6 1.0 7.4 3.1 $3.20 (2011 PPP) 13.3 3.8 36.4 20.3 7.2 38.9 $5.50 (2011 PPP) 38.0 12.8 47.3 18.5 National poverty 17.0 4.8 32.1 27.1 10.3 35.7 Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from MTI, and inputs from the WDI and Geostat. Notes: International estimates using the harmonized household welfare aggregate per capita. National calculation based on the national consumption aggregate per adult equivalent. • In addition to impoversihment, COVID-19 would force thousands of Georgian households to suffer downward mobility.8 350 thousand people would become poor after the pandemic ($5.5, 2011 PPP poverty line), and close to 805 thousand people would face losses that would taken them to a lower welfare group (Table 2). • The share of the population living in poverty or vulnerability is expected to increase. While the share of those considered middle-class and higher-incomes9 is expected to shrink from 24% to 20% of the population (Figure 2). Table 2. Impoverishment and downward mobility effects of COVID-19 (Thousand people) After COVID-19 impact Extreme Poor Poor Moderate Poor Vulnerable Middle class Total (<$1.90) (<$3.20) ($3.20 to $5.50) ($5.50 to $10.0) (>$10.00) Extreme Poor (<$1.90) 135 0 0 0 0 135 Before COVID-19 Poor (<$3.20) 66 295 0 0 0 361 Moderate Poor ($3.20 to $5.50) 47 147 724 0 0 918 Vulnerable ($5.50 to $10.0) 25 34 264 1,006 0 1,330 Middle class (>$10.00) 3 3 20 194 760 980 Total 276 479 1,008 1,200 760 3,724 Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI and Geostat. Notes: Microsimulations before mitigation measures. Poverty lines in USD, 2011 PPP. Shaded cells along the diagonal represent people who did not suffer mobility after COVID-19. Cells highlighted in red represent impoverishment: people who were nonpoor before but became poor after the pandemic (relative to $5.5 poverty line, 2011 PPP). • Effects are heterogeneous across population groups (Figure 3). Given the shock to remittances inflows, households identified as receivers of remittances are expected to suffer large poverty increases. 7 Using the lower-middle income poverty line of $3.20 (2011 PPP) and the absolute national poverty line, respectively. 8 Downward mobility is the movement (or re-classification) of households from one welfare group to a lower welfare group, as a result of income losses after COVID-19. For example, a household moving from middle-class to vulnerable, or from poor to extreme poor. 9 Middle-class and higher incomes are defined as people living with >$10 per capita per day (2011 PPP). Page 3 • Geographically, Tbilisi and other urban centers will be most affected by the economic shock and impoverishment. • There are no clear gender disparities in the short-term monetary effects of COVID-19 captured in this exercise. Males observe a slightly higher increase in poverty rates than females. However, COVID-19 is likely to deepen existing gender inequalities in the medium- to long-term. For example, firm-level data collected in 2020 suggests that female workers are most affected by unemployment and the double burden of household responsabilities. The elderly population seems more shielded from the poverty increases. • Adjara—which remains particularly affected by COVID-19 outbreak—is also the region that observes the highest increase in poverty rates (including extreme poverty).10 Tbilisi ranks second in the risk of impoverishment. In absolute numbers, most of the impoverished population (close to 125 thousand people) concentrate in Tbilisi (Map 1). • One quarter of those impoverished by COVID-19 are children (Figure 4), assuming no mitigation measures were implemented. A large share of the new-poor population has high educational attainment. Although the effects of COVID-19 are not as apparent in rural households, 30% of the new-poor live in rural areas. Only 5% of the new impoverished population were eligible to receive TSA before COVID-19. Approximately 45% of the impoverished population live in households receiving pensions. • Among the simulated effects, wage losses (either due to employment or reduced earnings) are expected to be the main shock to household incomes. Remittances are the second highest shock, though their effect is more significant for medium to higher-income deciles (Figure 4). Overall, all deciles and welfare groups suffer negative income shocks derived from the pandemic (Figure 6). • The simulations and parameters assumed suggest that the largest effect on poverty could be driven by unemployment in wholesale and retail trade, followed by shocks to tourism, construction and manufacturing (Figure 7). Introducing COVID-19 mitigation measures • Several mitigation measures were evalutated, based on available public information on benefits, eligibility and program budgets. The identifying assumptions in the microdata are summarized in Appendix 1. In cases when elibility identified in the data surpassed the budgetary capacity of the program, beneficiaries were randomly selected to comply with the budget ceiling. It is assumed that all budgets are distributed as benefits (i.e. no administrative costs). • Each mitigation policy reduces the (post-COVID-19) poverty rates (Table 3). Jointly, the mitigation measures analized reduce the national poverty rate by 4.7 percentage points (this is the largest mitigation effect in absolute number of people avoiding impoverishment). However, in relative terms, mitigation measures are most effective to upset extreme poverty, the joint mitigation effect could upset up to two thirds of the impacts of COVID-19 on extreme poverty. Table 3. Summary of mitigation effects Marginal effect on headcount poverty rate (percentage points) Employees Self-employed Vulnerable Households Households One-time All facing facing households (PMT: with young facing assistance to all combined unemployment unemployment 65K to 100K) children disabilities children below 18 measures $1.90 (2011 PPP) -0.9 -0.2 -0.2 -0.1 -0.1 -1.2 -2.4 $3.20 (2011 PPP) -1.3 -0.2 -0.7 -0.2 -0.2 -1.9 -4.4 National poverty -1.9 -0.3 -0.7 -0.1 -0.3 -1.7 -4.7 $5.50 (2011 PPP) -0.9 -0.1 -0.6 -0.1 -0.2 -1.1 -3.2 Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI, Geostat, IMFand KNOMAD. • Figure 9 illustrates how the short-term impacts of COVID-19 shift the welfare distribution to the left, increasing poverty. The mitigation measures contribute to shifting back the distribution, but they are insufficient to fully compensate the poverty impacts of the pandemic. 10The microsimulations rely on employment and income microdata. They do not factor in health data or regional incidence of COVID-19 cases. Hence, this exercise constitutes only a low-bound estimation of the complex economic effects of COVID-19 in hard-hit regions, such as Adjara. Page 4 • The incidence of COVID-19 is regressive, with lower-income households facing higher welfare shocks (as share of their consumption) than higher-income households (Figure 10). The combined mitigation measures are somewhat progressive, ensuring that households in the bottom 20 percent of the population capture higher relative benefits than higher-income households. • While the mitigation measures aim to reduce the negative economic shock of COVID-19 on consumption across households, their impact on poverty reduction is also significant. The mitigation measures prevent close to 80 thousand people from impoversihment related to COVID-19 (upper-middle income poverty line of $5.50, 2011 PPP). As secondary effect, 40 thousand people who were poor before the pandemic overcome poverty after receiving mitigation benefits.11 In total, the mitigation measures reduce poverty by 120 thousand and 175 thousand people ($5.50 and national poverty lines, respectively). The matrix of impoverishment and social mobility after mitigation is presented in Table 5 (Appendix). Fiscal costs of mitigation • The simulated measures allocate a budget of GEL 421 million, below the level of fiscal spending originally announced (Table 4, Panel A).12 Unemployment benefits are the largest policy under consideration. Based on the parameters of the model, close to 200 thousand workers could receive unemployment benefits (considering both programs, for formal employees, and for informal or self-employed workers). • Since the government response was not necessarily designed to target the poor and impoverished, the mitigation measures could cost close to GEL 2,400 per case of poverty reduction (investment cost of GEL 421 million / 175 thousand people overcoming the national poverty thresehold). • The unemployment program for the formal sector and the one-time transfer to children are expected to reduce the largest number of poor. However, due to their coverage of a large share of non-poor population, neither measure would seem to be the most cost-effective per poverty case adverted (Table 4, Panels B and C). • The mitigation measures seem to complement each other, by targeting different socio-economic groups. While targetting formal workers reduces moderate poverty, protecting the self-employed and informal workers, along with providing cash transfers to children, could be more cost-effective to tackle extreme poverty. Table 4. Fiscal cost, poverty mitigation and cost/benefit ratios Employees Self-employed Vulnerable Households Households One-time All facing facing households (PMT: with young facing assistance to all combined unemployment unemployment 65K to 100K) children disabilities children below 18 measures Panel A. Mitigation budgets (million GEL) Announced budget 150 75 55 15 27 160 482 Executed budget in simulations 150 20 55 13 24 160 421 Panel B. Poverty mitigation (thousand people lifted out of poverty) * $1.90 (2011 PPP) 33 8 9 2 4 46 90 $3.20 (2011 PPP) 49 7 25 9 8 69 164 National poverty 72 9 26 5 12 63 175 $5.50 (2011 PPP) 34 4 21 2 7 43 120 Panel C. Cost-benefit of mitigating poverty (GEL executed / people lifted out of poverty) ** $1.90 (2011 PPP) 4,566 2,442 6,161 5,904 6,429 3,443 4,667 $3.20 (2011 PPP) 3,051 2,981 2,210 1,376 3,134 2,306 2,571 National poverty 2,095 2,109 2,078 2,580 1,903 2,559 2,409 $5.50 (2011 PPP) 4,443 5,245 2,670 6,295 3,321 3,759 3,507 Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI, Geostat, IMFand KNOMAD. Notes: *Bold numbers indicate highest poverty reduction (number of people), for each poverty line. **Bold numbers indicate cheapest cost per case of poverty reduction, in GEL. 11 Around 135 thousand people improve their welfare status after COVID-19, as their benefits from mitigation result in higher welfare than the counter-factual (pre-COVID-19). The majority were poor before COVID-19, and less than 10 percent were vulnerable. 12 This simulated figure is below the planned budget of GEL 482 million for the measures under consideration (or GEL 750 million for all identified measures targeted to households). The difference comes mainly from lower simulated public expenditures in unemployment benefits for the self- employed. In contrast, other cash transfer benefits were capped, as potential beneficiaries identified in the data surpassed fiscal capacity. Page 5 Effects of social protection and social assistance • Existing government transfers—pensions and social assistance—are relevant and effective policies to protect households against unforeseen shocks, including COVID-19. • Estimations for 2018 suggest that the absence of pension incomes could triple the population living in extreme poverty; while poverty below the lower middle-income poverty line ($3.20 2011 PPP) would double (Figure 11). Incomes received from Social Assistance programs also contributed to halve extreme poverty in 2018. Due to their efficiency to target lower-income households, pensions and social assistance are less impactful on higher poverty thresholds. However, in the absence of pension incomes, poverty under the upper middle-income threshold ($5.50 2011 PPP) could increase by one third (or 14 percentage points) in 2018. • While the design of these policies is not targeted to revert the economic shocks of COVID-19, pensions and TSA make up significant sources of household incomes, prevent impoverishment, and constitute relevant tools for public policy implementation in Georgia.13 Macrosimulations of medium-term trends • The microsimulations capture the potential poverty and distributional effects of COVID-19 across different sectors and population groups in the short-term. However, they are based on static assumptions, and they exclude behavioral responses and general-equilibrium effects. • Hence, while it is difficult to foresee longer-term scenarios, Figure 12 projects macro-poverty trends and potential recovery paths through 2022. The macrosimulations are based on the most recent macroeconomic forecasts (MTI, October 2020). Their main advantage is to allow for a general equilibrium-approach, including the effects of COVID-19, as well as market responses and government interventions. Their main limitation is the potential underestimation of poverty shocks, due to the absence of inequality effects in the model. • The macrosimulations suggest that the national poverty rate could face an increase of 3 percentage points in 2020. The incidence of extreme poverty would increase by 25%, or 1 percentage point. Only by 2022 Georgia would observe a lower poverty rate than the headcount poverty rate in 2019.14 Potential caveats and limitations • The analysis in this note is based on available information and assumptions as of November 2020. The results more accurately reflect trends in the impact of COVID-19 and mitigation policies observed from March to November 2020, before the surge in the number of positive COVID-19 cases in the country. • The simulation models do not incorporate other non-monetary or indirect effects that with longer-term economic impacts, including population health shocks, foregone human capital accumulation, or risks to gender equality in the context of lockdowns. • Only a subset of mitigation policies is identifiable in the microdata. It is possible that additional and new policies could mitigate the shock to household incomes, directly or indirectly, and foster recovery. • Poverty and inequality are calculated as annual indicators. However, facing volatile markets and lacking appropriate smoothing mechanisms and diversified income sources, many households could face poverty immediately, during a fraction of the year. The annualized measurement of poverty in 2020 would not capture those temporary but relevant cases of impoverishment. 13 Since February 2020, the number of TSA beneficiaries increased by 70 thousand (to a total of 510 thousand people). Similarly, pension coverage increased in 2020. Since this section is based on coverage reported in the HIES 2018, those expansions are not incorporated. The incidence of social protection and social assistance for reducing poverty and COVID-19 related impoverishment could be even more significant in 2020. 14 Under the upper-middle income international poverty line. Figure is nowcasted for 2019. Page 6 Appendix 1. Identified mitigation measures targeted to households Table A1.1 Assumed shock to labor incomes by subsector of employment Budget or Category Policy target Mechanism Eligibility Benefit estimated population Temporary Families with a PMT Flat benefit of Vulnerable Budget of 55 (a) (monthly) cash rating score of 65,000 - 100 GEL for up families million GEL. transfers 100,000. to 6 months. Expected to Families with a PMT Benefit of 100 benefit about Temporary rating score of 0 - GEL for Targeted Household cash Families with 43,000 (b) (monthly) cash 100,000 who have three Social Assistance transfer young children individuals. transfers and more children for up to 6 Budget of 15 under the age 16. months. million GEL. Temporary Persons with severe Direct transfer Families with Budget of 27 (c) (monthly) cash disabilities and children of 100 GEL for disabilities million GEL. transfers with disabilities. up to 6 months. People who lost their 1,200 GEL over Unemployment Monthly jobs because of the the course of six Budget of 150 (d) benefits payment coronavirus crisis or are months, 200 GEL million GEL. on unpaid leave. per month. Unemployment benefits People employed in the informal sector or the One-time One-off cash Budget of 75 (e) Informal workers self-employed with assistance of transfers million GEL. substantiated claim of 300 GEL. job loss. One-time One-time All children below 18 Budget of 160 (f) All children payment of 200 payment years old million GEL. GEL. Support for children and Assistance to Vulnerable Students from youth cover one students in Education vulnerable families Budget of 35 (g) semester of tertiary subsidy (social score < 150 million GEL. university education* thousand) tuition. Households which Subsidy for Subsidize utility consume less than 200 electricity bills, fees for three Budget of 270 Utility subsidies (h) Utility subsidy* kWh of electricity and sanitary service, months (March, million GEL. 200 cubic meters of gas and water April, May) natural gas per month. bills. Source: SPJ 2020 and IMF 2020. Notes: * Not simulated as part of this exercise, due to data constraints. Hence, data restrictions limit the microsimulation analysis to measures with a total combined budget of GEL 482 million in 2020. Page 7 Appendix 2. Methodological details of macro- and micro-simulation models Nowcasting methodology. The nowcasting methodology applies the real growth rate of GDP per capita between the survey year (2018) and the nowcasted year (2020) to household’s welfare aggregate. It assumes a neutral distribution of growth, and a 0.87 passthrough rate from GDP per capita growth to private consumption growth. This methodology is regularly used and published by the World Bank as part of the Macro-Poverty Outlook. Counterfactual scenario. The counterfactual scenario is based on nowcasted poverty and inequality indicators using macroeconomic forecasts of GDP per capita growth, held by the MTI team until February 2020. Because COVID-19 was an unforeseen event in macroeconomic projections at the beginning of 2020, it is assumed that this scenario excludes any influence from the pandemic. Macrosimulations. Macro-simulations are conducted to estimate actual poverty and inequality indictors in 2020, after the economic shocks derived from the pandemic. Computationally, the macro-simulations are also based on the nowcasting method, applying the most updated macroeconomic projections of GDP growth in 2020 (calculated by MTI as of early October 2020). Because the nowcasting method assumes a neutral distribution of growth, macro-simulations do not account for potential distributional effects or heterogeneous economic effects of COVID-19. To the extent that the most recent macroeconomic projections for 2020 already factor-in the effects of government responses, they already incorporate some mitigation effects. Box A.1. Timeline of counterfactual and macroeconomic forecasts, 2020 Mar 21. International Mar 11. WHO passenger traffic Feb 29. Feb 26. Georgian declares suspended. Georgia closes citizen returning from COVID-19 schools. Mar 23. First Iran diagnosed with global COVID-19. strict quarantine pandemic restrictions. 2020 FEB MAR APR-SEP OCT Feb 25. Estimated Oct 15. Estimated GDP GDP per capita growth per capita growth 2020: 2020: 4.6% (MTI). -5.8% (MTI). “Counterfactual” “Macrosimulation” (Accounting for COVID-19) (Assuming no COVID-19) Source: Based on information from WHO (2020) and the Government of Georgia. Page 8 Microsimulations. Micro-simulations are based on a bottom-up approach. They leverage microdata on household and individual characteristics to capture both poverty and distributional effects of the pandemic. Building on the nowcasted (counterfactual) welfare aggregate for each household represented in the national household budget survey, this method simulates the effects of COVID-19 on households’ labor and non-labor incomes. The estimated effects and assumptions are the following: Box A.2. Economic shocks of COVID-19 in micro-simulation model Shock to household welfare (1) (2) Remittances (3) Agricultural Sales Labor Incomes Unemployment Reduced Earnings Souce: World Bank. ▪ Workers (either hired or self-employed) face a probability of unemployment based on their subsector of employment (Table A2.1). The assignment of unemployment is realized randomly within each subsector of employment. All workers who become unemployed lose 100%. All workers who remain employed face a wage loss, depending on their subsector of employment, and defined as a share of their reported wage incomes. These labor income shocks are assumed to last for 6 months in 2020. ▪ Remittances incomes fall by 27.5% in 2020. ▪ Agricultural sales fall by 10%, during 6 months in 2020. Page 9 Table A2.1 Assumed shock to labor incomes by subsector of employment (Micro-simulations) Wage income Probability Sector Subsector of employment loss in Note unemployment, % employment, % Agriculture 1 Agriculture, forestry, fishing 0 10 2 Mining and Quarrying 20 20 3 Manufacturing 20 20 Industry 4 Electricity, gas, steam and conditioning 0 0 Water supply, sewerage, waste 5 0 0 management 6 Construction 20 20 Wholesale and retail trade; repair of motor 7 vehicles, motorcycles and personal and 50 50 household goods 8 Transportation and storage 20 20 Accommodation and Food Service Considered entirely 9 50 50 Activities as tourism 10 Information and communication 10 10 11 Financial and Insurance Activities 0 0 12 Real Estate Activities 20 20 13 Professional, scientific and technical 0 0 Services Administrative and Support Service 14 20 20 Activities 15 Public Administration and defense, comp 0 0 16 Education 0 0 17 Human Health and Social Work Activities 0 0 18 Arts, Entertainment and Recreation 50 50 19 Other service activities 20 20 20 Activities of households as employers; 50 50 21 Activities of extraterritorial organizations 0 0 99 Tourism 50 50 Source: Author’s. Notes: The tourism subsector is constructed based on the 4-digit codes of the Statistical classification of economic activities in the European Community, NACE REV.2. Workers in tourism are defined as those with economic activities in accommodation and food services; select transport activities; activities; activities of travel agencies; and select recreational, cultural and sporting activities. Simulation of mitigation impacts The methodological steps for calculating the impact of mitigation measures for COVID-19 during 2020 are the following: ▪ Eligibility to each mitigation measure is identified based on households’ and individuals’ characteristics, and the available information about each mitigation program (collected by the IMF and the World Bank). ▪ The maximum potential benefit is assumed for all eligible beneficiaries. Households and individuals are allowed to simultaneously receive several mitigation benefits, if eligible to each. While some mitigation measures are allocated at the individual level and others at the household level, all benefits are ultimately added for each household represented in the HIES. Page 10 ▪ The final monetary benefits for each mitigation measure are adjusted to comply with the budget constraints announced (if available). Among all eligible beneficiaries, a random process is used to select the final beneficiaries, subject to the corresponding budget constraint. It is assumed that the entire budget is allocated to transfers or benefits (i.e. there are no administrative costs paid by the budget). ▪ The value of each mitigation measure (in GEL at 2020 prices) for the household is added to the simulated post-COVID consumption aggregate of that household. Finally, poverty and inequality indicators are recalculated under the mitigation scenario. Other parameters and assumptions Other parameters and assumptions are described below or included in Table A2.2. ▪ All simulated income sources (wages, remittances and agricultural incomes) are assumed to grow at the same rate as consumption (i.e. real GDP per capita) between the year of the microdata (2018) and 2020. ▪ The value of government transfers (pensions and TSA) are assumed to remain constant in real term between the year of the microdata (2018) and 2020. Table A2.2 Other parameters and assumptions Assumptions/Parameters Value Period Note Shock on remittances incomes -27.5% Annual Source: KNOMAD Shock on agricultural sales -10% 6 months Month/per National poverty line 2018 GEL 152.66 Source: Geostat capita GDP per capita growth 2018 5.0% Annual Source: MTI. Date: 15 October 2020 GDP per capita growth 2019 5.3% Annual Source: MTI. Date: 15 October 2020 GDP per capita growth 2020 (COVID-19) -5.8% Annual Source: MTI. Date: 15 October 2020 GDP per capita growth 2020 (counterfactual) 4.6% Annual Source: MTI. Date: 25 February 2020 2011 PPP conversion rate 0.835 2011 Source: International Comparison Program Souce: World Bank. Page 11 Figure 1. Estimated effect of COVID-19 on poverty rates Panel A. Extreme poverty line, Panel B. Lower middle-income class $1.9 (2011 PPP) poverty line, $3.2 (2011 PPP) International poverty rate International poverty rate Counterfactual Counterfactual COVID-19 Macrosimulation COVID-19 Macrosimulation COVID-19 Microsimulation COVID-19 Microsimulation 14.0 12.0 12.0 40.0 % of the population 10.0 35.0 30.6 % of the population 30.0 8.0 7.4 25.0 20.8 20.3 6.4 17.8 6.0 20.0 13.9 4.9 15.7 4.0 4.5 15.0 3.9 3.6 10.0 2.0 13.3 5.0 - - 2017 2016 2010 2011 2012 2013 2014 2015 2016 2018 2019N 2020F 2010 2011 2012 2013 2014 2015 2017 2018 2019N 2020F Panel C. Upper middle-income class Panel D. National absolute poverty line poverty line, $5.5 (2011 PPP) Official poverty rates International poverty rate Counterfctual Counterfactual COVID-19 Macrosimulation COVID-19 Macrosimulation COVID-19 Microsimulation COVID-19 Microsimulation 70 40.0 37.3 60.8 60 35.0 49.5 30.0 47.3 % of the population 45.7 27.1 % of the population 50 30.0 26.2 40.0 23.5 42.8 25.0 40 19.5 20.3 20.0 30 38.0 15.0 17.0 20 10.0 10 5.0 0 - 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019N 2020F 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020F Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: N= Nowcasted. F = Forecasted. All international poverty lines expressed in USD PPP 2011. Page 12 Figure 2. Distribution of the population across welfare and vulnerability groups < $1.90 (2011 PPP) $1.90 to $3.20 (2011 PPP) $3.20 to $5.50 (2011 PPP) $5.50 to $10 (2011 PPP) ≥ $10 (2011 PPP) Middle-class 100 12.0 11.8 11.2 10.4 13.0 12.5 12.6 13.7 16.2 19.8 21.2 21.5 21.4 22.9 22.4 20.490 24.3 Vulnerable 80 25.1 26.9 26.6 26.6 29.1 27.0 27.6 27.2 29.0 70 30.8 32.2 33.2 33.7 35.3 34.1 35.2 60 35.7 Moderate poor 50 32.0 31.2 32.1 30.1 31.0 31.1 31.0 30.7 29.5 40 28.7 27.1 27.9 29.1 27.2 26.9 27.0 30 26.1 Poor 20.9 18.7 19.4 19.8 18.7 18.8 17.2 20 17.9 17.0 14.4 12.9 Extreme poor 12.8 12.3 11.1 11.0 10 12.0 10.0 10.7 10.3 11.6 10.2 10.1 12.0 11.1 10.0 8.5 6.4 7.4 4.9 3.7 3.8 5.0 4.5 3.9 0 2020 Microsim 2019N 2013 2004 2005 2006 2007 2008 2009 2010 2011 2012 2014 2015 2016 2017 2018 Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: Based on prefered Micro-simulation results. N= Nowcasted. 2020 Microsim = Forecasted for 2020. Page 13 Figure 3. Distribution of impoverishment effects of COVID-19, across population groups Panel A. Marginal poverty increase ($3.2, 2011 PPP) - 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Location All National population 7.0 Tbilisi 8.3 Other Urban 8.1 Rural 5.2 Female 6.8 Male 7.2 Individual characteristics Young children (0-6 years old) 8.3 Children and youth (7-25 years old) 7.3 Adults (26-64 years old) 7.7 Elderly (65+) 3.3 None or incomplete primary 2.3 Incomplete secondary 4.2 General Secondary 7.3 Specialized Secondary 6.6 Tertiary 6.3 Male 7.0 Household head Female 6.9 None or incomplete primary 1.5 Incomplete secondary 2.6 General Secondary 7.6 Specialized Secondary 7.0 Tertiary 6.9 Household does not recieve pension 8.7 vulnerability protection Social Household receives pension 5.6 Household does not receive SA 7.4 Household receives SA 3.4 Households does not receive remittances 6.6 Household Households receives remittances 12.3 TSA poor (PMT<65,001) 3.4 TSA vulnerable (PMT: 65,001 to 100,000) 6.9 TSA noneligible (PMT>100,000) 7.4 Panel B. Marginal poverty increase ($5.5, 2011 PPP) - 5.0 10.0 15.0 20.0 25.0 Location All National population 9.4 Tbilisi 10.8 Other Urban 12.7 Rural 6.1 Female 9.2 Male 9.6 Individual characteristics Young children (0-6 years old) 10.0 Children and youth (7-25 years old) 9.9 Adults (26-64 years old) 10.1 Elderly (65+) 5.8 None or incomplete primary 3.1 Incomplete secondary 6.7 General Secondary 8.3 Specialized Secondary 11.0 Tertiary 9.9 Male 9.9 Household head Female 8.2 None or incomplete primary 4.6 Incomplete secondary 6.3 General Secondary 8.7 Specialized Secondary 11.9 Tertiary 9.6 Household does not recieve pension 11.1 vulnerability protection Social Household receives pension 8.1 Household does not receive SA 10.4 Household receives SA 0.7 Households does not receive remittances 8.6 Household Households receives remittances 19.6 TSA poor (PMT<65,001) 0.7 TSA vulnerable (PMT: 65,001 to 100,000) 9.9 TSA noneligible (PMT>100,000) 7.4 Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI and Geostat. Notes: Prefered Micro-simulation results. Page 14 Figure 4. Profile of impoverished population due to COVID-19 Who will be impoverished in Georgia? Household and individual characteristics 100% Elderly (65+), 7.5% 90% Rural Tertiary, 27.8% Tertiary, 20.3% 30.4% 80% Male 48.9% No pension 54.4% % of Impoverished population 70% Adults (26-64) Male 57.3% 69.9% Specialized Secondary Specialized Secondary 60% 20.0% 12.4% Other Urban No TSA 50% 32.5% 94.8% 40% 30% General Secondary General Secondary Female Youth (7-25) 31.3% 49.2% Pension 51.1% 20% Tbilisi 23.8% 45.6% 37.1% Female 30.1% 10% Children (0-6) Incomplete secondary 11.4% Incomplete secondary TSA 2.0% 2.0% 5.2% 0% Location Sex Age Educational Household head Household head Household - Household - attainment* - Sex - Educational Social Pensions attainment Assistance** Tbilisi Other Urban Rural Female Male TSA No TSA Pension No pension Children (0-6) Youth (7-25) Adults (26-64) None or incomplete primary Incomplete secondary General Secondary Specialized Secondary Tertiary Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI and Geostat. Notes: Estimations exclude potential mitigation measures. Notes: Impoverishment relative to the $3.20 (2011 PPP) international poverty line. * Educational attainment restricted to individuals 25 years-old and older. ** Based on reported income from social assistence. Page 15 Map 1. Geographic distribution of poverty effects of COVID-19 Panel A. Increase in extreme poverty ($1.90, 2011 PPP), percentage Panel B. Increase in poverty ($3.20, 2011 PPP), percentage points points Panel C. Increase in poverty ($5.50, 2011 PPP), percentage points Panel D. Impoversihment ($3.20, 2011 PPP), thousand people Panel E. Impoversihment ($5.50, 2011 PPP), thousand people Panel F. Impoversihment (national poverty line), thousand people Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI and Geostat. Notes: Based on microsimulation model. Estimations exclude potential mitigation measures. Notes: Imereti, Racha-Lechkhumi and Kvemo Svaneti are estimated together. Page 16 Figure 5. Source of income losses due to COVID-19 (Microsimulation) Panel A. By Welfare group Panel B. By decile Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: No mitigation measures assumed. Page 17 Figure 6. Effects of COVID-19 on household income sources (Microsimulation) Panel A. By Welfare group Panel B. By Decile Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Note: No mitigation measures assumed. Page 18 Marginal effect on poverty ($3.2 2011 PPP) poverty line 0 1 2 0.5 1.5 2.5 Wholesale and retail trade… Tourism Construction Manufacturing Households as employers Arts, Entertainment and Recreation Other service activities Transportation and storage the WDI and Geostat. Note: No mitigation measures assumed. Mining and Quarrying Effect of COVID-19 on wage incomes Information and communication Admin support service activities Agriculture, forestry, fishing Real State Activities Subsector of employment Electricity, gas, steam and conditioning Figure 7. Sectoral analysis of impoverishment due to losses in wage incomes Water supply, sewage, waste mngment Financial and Insurance Activities Prof, scientific and technical activities Public Admin and defence Effect of COVID-19 on unemployment Education Human Health and Social Work Activities Activities of extraterritorial orgs Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from Page 19 Figure 8. Estimated effect of select mitigation measures Panel A. Marginal effect on poverty rates, by mitigation measure Persistent COVID-19 impact Mitigated COVID-19 impact Marginal poverty effect (percentage points) 10.0 8.0 6.0 4.0 2.0 - $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP $1.90 PPP $3.20 PPP $5.50 PPP National poverty line National poverty line National poverty line National poverty line National poverty line National poverty line National poverty line Employees facing Self-employed Vulnerable Households with Households facing One-time All combined unemployment facing households (PMT: young children dissabilities assistance to all measures unemployment 65,000 to children below 18 100,000) Panel B. Microsimulated poverty rates 50.0 47.3 44.1 45.0 40.0 38.0 35.0 % Population 30.0 27.1 25.0 22.4 20.3 20.0 15.9 17.0 15.0 13.3 10.0 7.4 3.6 5.0 5.0 - COVID-19 + Mitigation COVID-19 + Mitigation COVID-19 + Mitigation COVID-19 + Mitigation Counterfactual Counterfactual Counterfactual Counterfactual COVID-19 COVID-19 COVID-19 COVID-19 $1.90 USD ppp $3.20 USD ppp $5.50 USD ppp National poverty Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Page 20 Figure 9. Distribution of household welfare under counterfactual, COVID-9 and mitigation scenarios Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: The graph excludes all households above percentile 95th for illustration purposes. Vertical lines correspond to international poverty thresholds (2011 PPP). gallT is the main household consumption aggregate harmonized by the World Bank poverty team and used for international poverty calculations. Table 5. Impoverishment and downward mobility effects of COVID-19, after mitigation measures (Thousand people) After COVID-19 impact + Mitigation measures Extreme Poor Poor Moderate Poor Vulnerable Middle class Total (<$1.90) (<$3.20) ($3.20 to $5.50) ($5.50 to $10.0) (>$10.00) Extreme Poor (<$1.90) 113 22 0 0 0 135 Poor (<$3.20) 34 263 64 0 0 361 Pre-COVID Moderate Poor ($3.20 to $5.50) 28 92 758 40 0 918 Vulnerable ($5.50 to $10.0) 9 27 218 1,066 10 1,330 Middle class (>$10.00) 2 2 12 180 785 980 Total 186 405 1,052 1,286 794 3,724 Souce: World Bank calculations based on data from the HIES 2018, MTI, WDI and Geostat. Notes: Highlighted cells represent people who did not change welfare status before and after the pandemic (including mitigation). Bold numbers in red represent people who suffered downward mobility as a result of COVID-19 (inspite of mitigation). Bold numbers in green represent people who improved their welfare status in 2020 (as a result of mitigation benefits). Page 21 Figure 10. Distributional incidence of COVID-19 shock and mitigation measures Panel A. Relative incidence of COVID-19 on household consumption Panel B. Relative incidence of COVID-19 combined mitigation measures on household consumption Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: Percentiles in Panel A. correspond to the pre-COVID-19 household welfare aggregate. Percentiles in Panel B. correspond to the household welfare aggregate, post-COVID but before the introduction of mitigation measures. Page 22 Figure 11. Poverty mitigation of social protection (2018) Actual Poverty rate Additional poverty 60.0 50.0 13.49 1.64 % of the population 40.0 30.0 15.96 20.0 42.5 42.5 3.11 10.0 11.83 15.5 4.20 15.5 4.5 4.5 - $1.90 USD ppp $3.20 USD ppp $5.50 USD ppp $1.90 USD ppp $3.20 USD ppp $5.50 USD ppp No pensions No Social Assitence Souce: World Bank calculations based on data from the HIES 2018. Notes: Pensions and SA based on self-reported incomes in the HIES. Additional poverty rates calculated by substracting self-reported pensions and SA incomes from the household consumption aggregate. Figure 12. Macrosimulations, 2020-2022 $1.90 Counterfactual $1.90 COVID-19 Shock $3.20 Counterfactual $3.20 COVID-19 Shock $5.50 Counterfactual $5.50 COVID-19 Shock 45.0% 40.0% 35.0% Headcount poverty rate 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 2018 2019N 2020F 2021F 2022F Souce: World Bank calculations based on microdata from the HIES 2018, macroeconomic projections from the MTI, and inputs from the WDI and Geostat. Notes: Based on prefered macrosimulation results. N= Nowcasted. 2020 F = Forecasted for 2020 onwards. All international poverty lines expressed in USD PPP 2011. Page 23