Poverty and Inequality Monitoring: Latin America and the Caribbean Projected 2020 Poverty Impacts of the COVID-19 Global Crisis in Latin America and the Caribbean Carolina Diaz-Bonilla, Laura Moreno Herrera, Diana Sanchez Castro • Poverty projections for 2020 in the LAC region suggest the crisis will likely reverse in a short time frame many of the social gains of the last decade. • However, emergency transfers, primarily from Brazil, lifted millions out of poverty, resulting in a marginal decline in the poverty rate and around 400 thousand less poor in 2020 relative to 2019. On the other hand, projections that exclude Brazil suggest an estimated net 13.7 million people fell into poverty in the rest of LAC. • Even though most countries adopted emergency measures to counteract the negative impact of the COVID-19 crisis, such policies’ generosity were quite low across the region. • LAC’s recently achieved majority-middle-class status is projected to be reversed, as the global crisis is expected to result in a loss of 4.7 million people from the Middle Class in 2020. Excluding Brazil, the negative impact is larger, with 12 million people falling out of the middle class. • Because lost labor income was supplemented by the emergency transfers, income inequality is projected to decline in the region in 2020, while it is projected to increase without Brazil. Dramatic declines in economic activity have swept throughout the Latin America and Caribbean (LAC) region due to the impact of a global slowdown and pandemic brought on by the Coronavirus (COVID-19) outbreak. High levels of uncertainty regarding the COVID-19 outbreak led to national quarantines and social distancing orders to try to preserve health care systems and save lives. Concerns continue to arise, however, regarding the negative impacts on livelihoods, especially for the poorest. Latin American countries face high levels of informality and high levels of self- employment, resulting in lower quality and more vulnerable jobs. High levels of remittances in some countries also saw dramatic drop-offs, at least for part of 2020, affecting poor, near-poor, and even middle-class households. The contraction in the region’s economic activity is expected be -6.7 percent in 2020, the largest economic shock in decades, despite unprecedented policy support.1 Poverty projections for 2020 in the LAC region suggest the COVID-19 crisis will likely reverse in a short time frame many of the social gains of the last decade. The LAC region made commendable achievements in the 2000s - "the golden decade" - in the fight against poverty, with the poorest benefitting most from sustained economic growth. This trend continued until 2014. Since then, poverty reduction has been mostly stagnant, yet the region continued to show some increases in its middle class. Poverty in the (LAC18) region, defined as living on less than $5.5 a day (PPP2011), was 22 percent in 2019 and was projected to continue at similar levels under pre-COVID 2020 expected 1 World Bank. 2021. “Renewing with Growth” LAC Semiannual Report (March), World Bank, Washington, DC. 1 JUNE 11, 2021 GDP growth (Figure 1). 2 However, the COVID-19 global crisis is expected to have led to poverty increases in almost all countries, with millions of people falling into poverty. Brazil, however, is an important exception. 3 The government of Brazil implemented a generous emergency transfer program benefitting almost 67 million Brazilians that not only protected families from falling into poverty but also lifted many people out of poverty in 2020. 4 Poverty is therefore projected to decline sharply in Brazil in 2020. 5 Poverty in the LAC region is expected to decline marginally, resulting in almost 400 thousand less poor in 2020, as social transfers, primarily from Brazil, helped lift millions of people out of poverty. Due to Brazil’s sharp poverty decline, poverty in the LAC region is expected to decline marginally from 22 percent in 2019 to 21.8 percent in 2020 (Figure 1). More than 20 million people across the region are projected to have fallen into poverty (below the $5.50 poverty line) in 2020, with an increase of 1.4 million more poor due to population growth. On the other hand, emergency social transfers across the region in 2020 are projected to have lifted 22 million people out of poverty, of whom more than 77 percent were from Brazil.6 The combination resulted in a net decline of almost 400 thousand poor in LAC. Without the emergency measures taken by governments across LAC, poverty could have instead increased to 26.5 percent in 2020 (Figure 1), and the region may instead have added 28 million net new poor in 2020. Excluding Brazil, however, the rest of the LAC region is projected to have experienced a sharp increase in poverty, even with mitigation measures. Projected poverty rates for the LAC region excluding Brazil increased from 23.5 percent in 2019 to 26.7 percent in 2020 (Figure 2). These projections that exclude Brazil suggest an estimated net 13.7 million people fell into poverty. Had no mitigation measures been implemented at all, poverty for LAC excluding Brazil would have increased further to 28.4 percent. In summary, mitigation measures, especially in Brazil, helped limit the negative impacts in the short term. However, without a fast and inclusive economic recovery and similar levels of mitigation measures, poverty may rise again in 2021. Mitigation measures helped poorer households relatively more, while the negative poverty impacts without mitigation measures would have been relatively worse. The 2020 projections show a larger percent decline in poverty for households living with income per capita below the $3.2 a day (2011 PPP) poverty line. Poverty in the region measured by the $3.2 a day poverty line was 9.5 percent in 2019 and was projected to decline to 8.9 percent in 2020 under a COVID scenario with mitigation measures (Figure 1). This implies that the mitigation measures were relatively better 2 The LAC18 aggregate is based on 18 countries in the region for which microdata is available at the national level: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay. 3 Chile is another exception, where poverty is projected to remain constant under the $5.50 poverty line. Chile’s social protection measures are expected to have helped offset the worst effects of the crisis, maintaining poverty at prepandemic levels. In all other countries, poverty-mitigation measures were not enough to avoid poverty increases. 4 Brazil’s Auxilio Emergencial (AE) was conceived as a temporary program and ended in December 2020. However, the government launched a new wave of AE in April 2021 with lower benefits that targeted about 44 million individuals. 5 Brazil Macro Poverty Outlook (April 2021) https://pubdocs.worldbank.org/en/114751582655277329/mpo-bra.pdf 6 For projections of the number of poor in the region, this note includes population figures for LAC24. This includes LAC18 (see earlier footnote) plus Dominica, Guyana, Jamaica, St. Lucia, St. Vincent and the Grenadines, and Suriname. All estimates continue to exclude Venezuela due to lack of available and representative data. 2 JUNE 11, 2021 targeted to the poorest. However, without mitigation meaures, poverty under the $3.20 poverty line would have increased relatively more than for the $5.50 poor, increasing to 12.2 percent in 2020. Similarly, excluding Brazil, poorer households would have fared worse even with mitigation measures in the rest of the region, as seen by the larger relative increase in poverty under the $3.20 (increasing from 9.7 percent in 2019 to 11.2 percent in 2020, Figure 2). Figure 1. LAC Poverty Projections for 2020: pre-COVID, COVID, and COVID+Mitigation 45 38.7 38.5 38.4 37.3 36.9 36.6 36.5 40 34.7 35 26.5 30 22.1 22.0 21.8 25 % 20 12.2 15 9.5 9.3 8.9 10 5 0 2019 2020 (Pre-Covid) 2020 (COVID) 2020 (COVID+Mitig) Poor $ 3.20 Poor $ 5.50 Vulnerable ($5.5 - $13) Middle Class ($13 - $70) Figure 2. LAC Poverty Projections for 2020 excluding Brazil 40.4 39.9 39.8 45 39.4 35.1 35.0 40 31.6 30.8 35 28.4 26.7 30 23.5 23.5 25 % 20 12.6 11.2 15 9.7 9.6 10 5 0 2019 2020 (Pre-Covid) 2020 (COVID) 2020 (COVID+Mitig) Poor $ 3.20 Poor $ 5.50 Vulnerable ($5.5 - $13) Middle Class ($13 - $70) Source: Macro-microsimulation model projections based on 2019 SEDLAC (CEDLAS-WB) microdata and projections of private consumption per-capita, job losses, and remittances from the MTI and POV Global Practices. “COVID” scenario assumes no mitigation measures. “COVID+Mitig” scenario incorporates mitigation measures. See Box 1 3 JUNE 11, 2021 BOX 1. Poverty Projection Methodologies The poverty projections are based on a macro-microsimulation model. Countries that do not have available 2019 microdata are first projected to 2019 using a neutral distribution methodology. The 2020 pre-COVID estimates are likewise projected using a neutral distribution method. On the other hand, the 2020 COVID projections use the microsimulation method where: [A] COVID estimates apply latest projections for GDP, job loss, and remittances; and [B] COVID+Mitigation adds projected mitigation measures. The Microsimulation approach provides an estimate of both poverty and distributional impacts from the COVID-19 global crisis across the LAC region (see Annex 2 for more detail). (a) Neutral Distribution This approach projects poverty based solely on projected annual growth in Private Consumption per capita from National Expenditure Accounts and corresponding population projections. This approach was applied to Chile, Guatemala, Haiti, and Nicaragua to estimate the 2019 baseline. The methodology assumes no changes in the distribution of income within a country; therefore inequality does not change, which is the largest disadvantage of this method. A neutral distribution approach can be acceptable for a few years’ forecast in cases when it is acceptable to expect minimal changes in the income distribution. For this approach, we consider the actual income distribution of the latest year of available household microdata for each country. We increase (or decrease) the household per capita income of each observation in a specific survey by the projected annualized growth of private consumption per capita from the World Development Indicators. In addition, we also adjust household weights using population projections for each country to allow for population growth. For the 2020 pre-COVID estimates, we use private consumption projections previously expected for 2020 in the 2019 October Macro Poverty Outlook briefs. (b) Microsimulation This approach takes the 2019 microdata for the 18 SEDLAC countries as a starting point. The methodology then allows three main channels of transmission of the 2020 shock: through job losses, labor income changes, and non-labor (remittance) income changes. Lastly, mitigation measures are also applied. Where possible, job losses/gains per sector by country are provided by the country Poverty Economist. When not available, a macro model converts changes in sectoral GDP growth projections (in Agriculture, Industry, and Services) into job losses or gains in the corresponding sector, by country. These job losses/gains per sector are then imposed onto the household survey data. A Probit model, applied to each of the three sectors and by formal and informal workers, provides each person in each household survey a probability of employment in each sector, based on a set of characteristics (sex, age, education level, urban/rural, dependency rate, dummy for member working in public sector, and a dummy for remittances). Workers with the lowest employment probability are simulated to lose their job, until total job loss matches the macro projections for job loss by sector. For job gains, new workers are chosen among the unemployed according to the probability of being employed in that sector until total job gains match the macro projections by sector. Workers who lose their job lose 100% of labor income. In the case of job gains, a new labor income is estimated via a traditional Mincer equation, in which the logarithmic of income is regressed on sex, age, education level, urban, dependency rate, and a dummy if the household has income from remittances. Next, projected changes in remittances by country are also included in the microsimulations. The percent change in remittances at the national level is applied as a direct percent change in the remittance income of any household that receives this income in the household survey. The model does not simulate changes in other nonlabor income. Lastly, mitigation measures are estimated and applied as direct cash transfers to eligible households. 4 JUNE 11, 2021 Even though most countries adopted emergency measures to counteract the negative impact of the COVID-19 crisis, such policies’ generosity were quite low across the region. With the exception of Brazil, the benefit incidence as a share of pretransfer income was on average a mere 15 percent. It ranged from 3 percent in Ecuador to 33 percent in Argentina (Figure 3). Likewise, countries varied in the way they reached their populations and in their ability to target benefits. Colombia, Brazil, Uruguay, and Chile showcase a clear effort to ensure that support reached those who needed it most. While Costa Rica, Mexico, and Argentina provided minimal support across the income distribution, Bolivia and Guatemala provided substantial benefits to their entire populations (Figure 4 and Annex 4). Figure 3. Mitigation-Measure Coverage and Benefit Incidence (%) 98 88 85 70 70 64 53 50 48 45 43 30 32 31 29 33 17 18 19 22 21 16 5 12 8 8 10 65 4 4 3 BOL SLV GTM PER PAN BRA PRY CHL COL DOM URY HND CRI ARG ECU MEX Coverage (as % of total population) Benefit incidence (as % of pre-transfer income) Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI and POV GPs. The current projections shown are based on a macro-microsimulation model that assumes 12 months of unemployment. See Annex 2. Figure 4. Population Covered by Mitigation Measures, by Percentile (%) Note: Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. The estimated mitigation measures are applied to microdata that has first been projected to include COVID-19 impacts for 2020. See Annex 2 for the macro-microsimulation model. 5 JUNE 11, 2021 In 2018, LAC’s middle class became for the first time the largest income group in the region, but this positive achievement is projected to be reversed by the current global crisis as 4.7 million people fall out of the middle class. After decades of gradual rise, LAC’s middle class (per capita income between $13 and $70 per day in 2011 PPP) finally became the region’s largest income class, larger than both the Vulnerable ($5.50-$13 per day) and the Poor. By 2019, the middle class accounted for 38 percent of LAC’s population, or around 230 million people, while the vulnerable class, who are near poor but not yet middle class, accounted for another 220 million people. However, the 2020 global pandemic is expected to reduce the middle class to 37.3 percent of the population, for a net loss of 4.7 million people (Figure 1). Meanwhile the vulnerable class is projected to remain the largest population group (at 38.5 percent of the population). The projected net loss in the middle class is less negative than originally expected, due primarily to the generous transfer program implemented in Brazil. While 21.6 million people in LAC are projected to lose middle-class status due to the crisis, around 17 million are projected to be added to LAC’s middle class (including through population growth), thanks primarily to the emergency transfers, with Brazilians making up more than 70 percent of this gain. Table 1 presents transition matrices of the share of the population moving between 2019 and 2020. Without Brazil, the rest of the region will likely show a sharp decline in the size of the middle class, with a projected net loss of 12 million people. With or without Brazil, the final result is a leftward shift in the region’s income distribution, with the vulnerable class representing once again the largest socioeconomic group. Table 1. Transition Matrix Source: Projections based on 2019 SEDLAC (CEDLAS-WB) microdata and macroeconomic projections of private consumption per-capita, job losses, and remittances from the MTI and POV Global Practices. 6 JUNE 11, 2021 BOX 2. LAC-18: From the Golden Decade to the COVID crisis The LAC region made commendable achievements during the 2000s - "the golden decade" - in the fight against poverty, with the poorest benefitting most from sustained economic growth. This trend continued until 2014; however, since then poverty reduction has been mostly stagnant. Worse still, the current crisis will result in an important setback in the region’s poverty reduction, and may undo part of the achievements from the golden decade. Figure B2.1 presents the LAC region’s poverty trend since 2008. The solid lines present the poverty trends that are published in the LAC Equity Lab, which includes 18 countries with recently available household survey data: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay. Due to a break in the poverty series in the Mexico data between 2014 and 2016, and given the importance of Mexico in the total LAC population, the LAC-18 aggregate published in the LAC Equity Lab now includes a break and overlap in the series in 2015. The earlier series is projecting the Mexican 2014 survey forward to 2015, The later series is projecting the Mexican 2016 microdata backward to 2015. Similarly, the Dominican Repulic also has a break in its series in 2016, therefore the 2015 data corresponds to extrapolating its 2016 data to 2015. The dashed lines below are projections, using the newest available series, under two assumptions: the impact of COVID with and without mitigation measures put in place to protect the most vulnerable. After accounting for differences in the levels due to the comparability of the series, the graph shows that without mitigation measures the region would have faced a setback in poverty reduction of around eight years. Figure B2.1. LAC Poverty Projections for 2020 40 Middle Class $13-$70 37.3 35 34.7 30 25 26.5 Poor $5.5 Percentage % 21.8 20 15 12.2 10 8.9 Poor $3.2 5 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020e Source: Macro-microsimulation projections based on 2019 SEDLAC (CEDLAS-WB) microdata and macroeconomic projections of private consumption per-capita, job losses, and remittances from the MTI and POV Global Practices. These scenarios represent projections with [ marker] and without [ marker] mitigation measures. See Box 1. Estimates suggest income growth for the bottom deciles would have been the most negatively affected by the global crisis, but mitigation measures provided support. Income growth is expected to be even lower throughout the income distribution relative to the region’s stagnation period of low poverty reduction (2015–2019). However, taking into account the mitigation measures adopted by various countries, income growth is projected to be positive for the bottom two deciles: growth 7 JUNE 11, 2021 estimates range from 4.5 percent for the bottom decile to -4.7 for the top decile. In the absence of emergency measures, all income deciles would have experienced negative income growth, with the bottom decile experiencing over three times what the top decile would have experienced (-20.1 percent and -6.1 percent, respectively) (Figure 5). Figure 5. Projected Growth Incidence Curves, Latin America and the Caribbean 10 Growth Rate (Annualized) 5 2015-2019 0 -5 2019-2020 (with mitigation measures) -10 -15 2019-2020 (without mitigation -20 measures) -25 1 2 3 4 5 6 7 8 9 10 Deciles of Per Capita Household Income Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI and POV GPs. The current projections shown are based on a macro-microsimulation model that assumes 12 months of unemployment. See Annex 2. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. Because lost labor income was supplemented by the emergency transfers, income inequality is projected to decline in the region in 2020. As noted earlier, Brazil’s generous emergency transfer program provided additional income to almost 67 million individuals, which contributed to a strong decline in poverty and inequality in Brazil in 2020. Although not as generous, emergency transfers in other countries also helped either to decrease inequality or minimize the increase in inequality. Overall, inequality in the LAC region, as measured by the Gini coefficient, is expected to have declined from 51 to 49.8 in 2020. On the other hand, the Gini coefficient is projected to be marginally higher in 2020 if we exclude Brazil from the regional estimates, increasing from almost 48 in 2019 to 48.2 in 2020 even with mitigation measures (Figure 6). The increase in inequality in many countries is driven mostly by labor income losses of households whose members work in the service and industry sectors. 8 JUNE 11, 2021 Figure 6. Gini coefficient projections for 2020 59 57 55 Gini coefficent Gini (with 53 Brazil) 51 51 49 49.8 Gini (without Brazil) 48.2 47 48 45 2020e 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI and POV GPs. The current projections shown are based on a macro-microsimulation model that assumes 12 months of unemployment. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. The Gini coefficient is a measure between 0 and 1; the Gini index is equal to the Gini coefficient scale between 0 and 100. The largest declines in private consumption per capita are projected for Panama, Argentina, Peru, and Mexico. Argentina, Nicaragua, Haiti, Ecuador, and Mexico faced declines in private consumption per capita in 2019, yet the consumption decline in 2020 due to the COVID-19 crisis is expected to be larger (excluding Nicaragua), and significantly larger in the case of Mexico. Although the rest of LAC experienced positive growth of private consumption per capita in 2019, all are expected to face significantly larger but negative growth in 2020 (Figure 7). Figure 7. Private Consumption Per Capita Growth (%) 15% 2019e 2020f 2021f 2022f 10% 5% 0% -5% -10% -15% -20% -25% Mexico Panama Paraguay Costa Rica Colombia Peru Guatemala Argentina Nicaragua Ecuador El Salvador Haiti Chile Brazil Honduras Uruguay Bolivia Dominican Republic Source: World Bank. 2019 estimated and 2020-2022 forecasted based on information and circumstances known in June 2020. Projections may differ from other vintages in other bank documents. Note: Private Consumption (Per Capita) Volume Millions 2015 real USD % change 9 JUNE 11, 2021 Household welfare will primarily be affected by the reduction of labor income through lockdown- induced job losses, particularly in the service sector. Lockdowns and border closures leading to internal and international mobility restrictions had an immediate effect on restaurants, shops, and tourism. Many businesses had to close on short notice. In addition, large numbers of workers in these services sectors are informal, increasing their vulnerability to falling into poverty or becoming even poorer. Not all companies could benefit from e-commerce solutions and services, which will depend on the type of activity and the profile of customers. For example, retail and restaurants may rely on delivery services and the penetration and literacy of financial products and the Internet in the local market. However, since not all companies could operate under COVID circumstances, nor survive closed for long periods, job losses in the service sectors are projected to be the largest in almost all countries (Figure 8). Workers in the industry and agriculture sectors will also be affected, though to a lesser extent. In Honduras, job losses in industry are projected to be higher than losses in services (19.2 and 10.2 percent, respectively), as the country was recently struck by two hurricanes that damaged many factories. In El Salvador and Paraguay, job losses in the industrial sector will be as high as the loss of employment in services. A majority of agricultural workers are informal (Figure 9) and a majority of the poor are in agriculture (Figure 10), yet agricultural employment is the least affected by the crisis. Agriculture job losses in most countries are expected to account for between 1 and 9 percent of total job losses (with Chile facing an even larger decline), yet other LAC countries, in particular Peru and Paraguay, are actually expected to increase employment in this sector (8.2 and 8.3 percent, respectively) (Figure 8). Figure 8. Job losses (%) per sector (COVID 2020 versus pre-COVID 2020 projections) 20% 10% 0% % Job losses -10% -20% -30% -40% -50% Costa Rica Domican Republic Honduras Ecuador Chile Colombia Bolivia Brazil Panama Haiti Mexico Nicaragua El Salvador Peru Guatemala Uruguay Argentina Paraguay Agro Industry Services Source: World Bank MTI and POV GPs. 10 JUNE 11, 2021 Figure 9. Workers by sector and informality in LAC, Figure 10. Workers in Poverty (%) per sector circa 2019 in LAC, circa 2018 ($5.50/day poverty line) Services, Agriculture, 38% 40% Industry , 22% Source: SEDLAC (CEDLAS and World Bank). Note: Informality refers to workers ages 15–64 who do not receive a pension. For Panama, estimates are limited to workers receiving an aguinaldo (salary bonus). In Argentina, Ecuador, Panama, and Mexico self-employed workers are not asked about pensions; therefore, in this report self-employed workers in these four countries who have completed tertiary education are considered formal workers. Figure 11. Informality across LAC countries, circa 2019 7% 18% 19% 22% 24% 24% 25% 30% 31% 31% 40% 44% 46% 50% 53% 64% 71% 73% 77% 93% 82% 81% 78% 76% 76% 75% 70% 69% 69% 60% 56% 54% 50% 47% 36% 29% 27% 23% Panama Mexico LAC Paraguay Colombia Guatemala Peru Costa Rica Nicaragua Ecuador El Salvador Honduras Argentina Haiti Brazil Chile Dominicana Bolivia Uruguay Informal Formal Source: SEDLAC (CEDLAS-World Bank) It is important to note that public administration, utilities, and health are among the sectors not affected by the lockdowns instituted across countries. All countries instituted some type of exception for essential services to respond to the current situation. This thus prevented the loss of jobs in these specific sectors. Workers in public services such as gas, electricity, telecommunications, public administration and defense, social work, and health are not expected to show job losses. For the microsimulations presented in this document, the model protects workers in these sectors from facing any employment or income losses. 11 JUNE 11, 2021 In addition to job losses and labor income changes, a third important channel impacting poverty are losses in (non-labor) remittance income. Remittances in some countries saw dramatic drop-offs, particularly mid-2020, affecting poor, near-poor, and even middle-class households. The expected decrease in remittances implies a reduction in household nonlabor income and therefore an increase in poverty. Remittances as a share of household income range from almost zero in Uruguay to 2.2 percent in El Salvador, even though they can represent almost 21 percent of GDP in a country like El Salvador. Similarly, in both El Salvador and Honduras, around 6 percent of households receive remittances (the largest in the region, represented by the size of the bubble in Figure 12), but remittances can represent almost 30 percent of income for that small share of households who receive them. The 2020 crisis resulted in large declines in remittances in most countries, with some exceptions (Figure 12). 7 Colombia and Ecuador faced the highest declines (20 and 19 percent, respectively), but this likely had small effects, because only 1 percent of households in these countries receive remittances (although for that 1 percent, the impact was quite negative). The negative impacts on the Dominican Republic, Nicaragua, and Guatemala were more pronounced: remittances declined between 10 and 15 percent, represent from 0.6 to 1 percent of total household income, and are received by 2.2 to 4 percent of households. Interestingly, in Honduras and El Salvador, where remittances are received by the highest percentage of households, remittances experienced positive annualized growth in 2020 of 3.8 and 4.8 percent, respectively. This growth was lower than previous years, being affected by the sharp drop-off experienced between March and May, but shows the resilience of migrants from these countries when it comes to helping their families back home. Figure 12. Projected declines in remittances across LAC countries. 3.0% El Salvador 2.5% Average remittances over total household income 2.0% 1.5% Domican Republic Nicaragua 1.0% Mexico Guatemala Honduras 0.5% Ecuador Costa Rica Uruguay Colombia Peru 0.0% Paraguay Bolivia Chile -0.5% -25% -20% -15% -10% -5% 0% 5% 10% Remittances Growth from 2019 to 2020 Source: Average remittances over total household income and share of households receiving remittances (size of the bubble) from SEDLAC (CEDLAS-World Bank) and remittance growth projection from the MTI GP. 7 The remittance data reflects projections as of the end of 2020. More recent data is emerging that suggests remittances declined less than projected and may have even increased in several countries beyond Honduras and El Salvador. This newer information will be incorporated into the next round of the poverty projections. 12 JUNE 11, 2021 The Latin America and Caribbean region is facing a global crisis without recent precedent, with persistent inequalities that result in unequal pandemic impacts, and some successful mitigation measures that protected the poor but yet are not sustainable. Governments throughout the LAC region have implemented various mitigation measures to protect their most vulnerable populations. Cash transfers and unemployment insurance are projected to offset some of the short-term negative welfare impacts of the global economic slowdown, but concerns are rising about the sustainability of these measures. Unfortunately, the region entered this crisis under difficult circumstances, having been in the midst of protests throughout 2019 and facing several years of relatively low growth and poverty stagnation. Worse, what could have been a source of celebration, that the LAC region became a majority-middle-class region for the first time, has been short-lived as millions of people likely fell out of the middle class under this global crisis. The region will now have to face the negative impacts from loss of schooling and work experience, as well as high levels of debt. The region’s “new poor” will be better suited to recover from the crisis as jobs returns and the region begins to grow again. However, households who were already poor, and have now lost further human or physical capital accumulation, will have the hardest time recovering. The region may therefore see inequality across multiple dimensions deteriorate, and may need to focus more on the excluded than before. As more data becomes available, this series will continue to be updated to share the most current understanding of the poverty situation in the LAC region. 13 JUNE 11, 2021 BOX 3. Microdata Data from SEDLAC-03 The numbers presented in the brief are based on the regional data harmonization effort known as the Socio-economic Database for Latin America and the Caribbean (SEDLAC) - a joint effort of the Poverty & Equity GP of the World Bank and CEDLAS from the Universidad Nacional de La Plata (Argentina). The LAC-18 aggregate used for a wide-range of WB analytics is based on the 18 countries that have comparable data available for recent years, in particular related to income. SEDLAC includes information from over 300 household surveys carried out in 19 LAC countries: Argentina, Bolivia, Brazil, Colombia, Costa Rica, Chile, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and St Lucia. For the current exercise, we pooled all the 2018 harmonized databases, or latest when 2018 is unavailable. Monetary variables are adjusted by consumption per capita growth and survey weights are adjusted by population growth. HOUSEHOLD SURVEYS USED FROM SEDLAC HARMONIZATION Circa 2019 – LAC Country Name of survey Coverage 18 Argentina Encuesta Permanente de Hogares- Continua 2019 Urban-31 Cities Bolivia Encuesta Continua de Hogares- MECOVI 2019 National Brazil Pesquisa Nacional por Amostra de Domicilios 2019 National Encuesta de Caracterización Socioeconómica Chile 2017 National Nacional Colombia Gran Encuesta Integrada de Hogares 2019 National Costa Rica Encuesta Nacional de Hogares 2019 National Dominican R. Encuesta Nacional de Fuerza de Trabajo 2019 National Ecuador Encuesta de Empleo, Desempleo, y Subempleo 2019 National El Salvador Encuesta de Hogares de Propósitos Múltiples 2019 National Guatemala Encuesta Nacional de Condiciones de Vida 2014 National Encuesta Permanente de Hogares de Propósitos Honduras 2019 National Múltiples Enquête sur les Conditions de Vie des Ménages Haiti 2012 National après le Séisme Encuesta Nacional de Ingresos y Gastos de los Mexico 2018 National Hogares Encuesta Nacional de Hogares Sobre Medición de Nicaragua 2014 National Niveles de Vida Panama Encuesta de Hogares 2019 National Paraguay Encuesta Permanente de Hogares 2019 National Peru Encuesta Nacional de Hogares 2019 National Uruguay Encuesta Continua de Hogares 2019 National 14 JUNE 11, 2021 ANNEX 1. 2.1 ABOUT THIS BRIEF This brief was produced by the Latin America and Caribbean Team for Statistical Development (LAC TSD) in the Poverty and Equity Global Practice of the World Bank (May 2021). The note was led by Carolina Díaz- Bonilla, Laura Moreno Herrera, and Diana Sanchez Castro. The team worked under the guidance of Ximena Del Carpio. The numbers presented in this brief are based on a regional data harmonization effort known as SEDLAC, a joint effort of the World Bank and CEDLAS at the National University of La Plata in Argentina (see Annex X for the list of surveys used in this brief). They increase cross-country comparability of selected findings from official household surveys. For that reason, the numbers discussed here may be different from official statistics reported by governments and national offices of statistics. Such differences should not be interpreted in any way as a claim of methodological superiority, as both sets of numbers serve the same important objectives: regional comparability and the best possible representation of the facts of individual countries. Indicators for LAC are calculated using data from Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay (LAC-18). Indicators for LAC-24 applied projections of LAC-18 to Dominica, Guyana, Jamaica, St. Lucia, St Vincent and Grenadines, and Suriname. ANNEX 2. THE MACRO-MICROSIMULATION MODEL To be able to model both the poverty and distributional impacts of the 2020 global crisis caused by the COVID-19 pandemic, the team chose to implement a macro-microsimulation model. To capture the impact of such a large negative shock on inequality, poverty, and on the size of the middle class required going beyond neutral distribution methodologies. This methodology applies country-specific macroeconomic projections to country-specific behavioral models built on household survey microdata that were harmonized as part of the Socio-Economic Database for Latin America and the Caribbean (SEDLAC) project. Although many inputs are country specific, the model applies the same methodology across all countries in order to estimate poverty and distributional impacts in a consistent manner for LAC as a whole. In addition, this class of models aims to maintain consistency between the macro- and micro- projections and therefore tends to focus on annual impacts rather than short-run impacts. The microsimulations are based on a household income generation model (Bourguignon and Ferreira 2005). For the macro side, rather than use a computable general equilibrium (CGE) model for each country per the macro-microsimulation models of Bourguignon, Bussolo, and Pereira da Silva (2008) and Ferreira et al. (2008), the macro model here uses a variety of macroeconomic projections as determined by the World Bank POV GP’s Poverty Economist for each country. Some countries apply a CGE model, others use simpler sectoral macroeconomic projections to estimate job losses, while others are able to apply actual job-loss data from household surveys conducted in the field. This annex presents the macro-microsimulation model. SEPTEMBER 2016 2.2 MACRO-MICROSIMULATION MODEL: INPUTS The model requires five main inputs. Country specific harmonized household survey microdata for each country in the Latin America and Caribbean region is based on the SEDLAC database. The model is applied to the 18 countries in the Latin America and Caribbean region with recently available household surveys. Projected annual growth rates in Private Consumption per capita from National Expenditure Accounts are provided by the Macro, Trade, and Institutions (MTI) Global Practice for each country. Sectoral GDP growth projections in Agriculture, Industry, and Services are also provided by the MTI GP and are used to estimate projected sectoral job losses using a GDP to employment elasticity. Wherever possible, the model uses sectoral job loss data estimated or acquired by the POV GP to have more accurate information. Projected changes in remittances by country are also provided by MTI. Another input is a set of population projections for 2019 and 2020 based on the World Development Indicators projections database. The three sets of MTI projections (private consumption growth, sectoral GDP growth, and remittance growth) are updated at least twice per year. A final input is a microdata base created by the country Poverty Economists that estimate the amount of public transfers provided in 2020 to each household. 2.3 MACRO-MICROSIMULATION MODEL: METHODOLOGY This approach takes the 2019 projected microdata for the 18 SEDLAC countries mentioned above as a starting point. The methodology then allows three main channels of transmission of the 2020 shock: through job losses, labor income changes, and non-labor (remittance) income changes. For the projections with mitigation measures, the model incorporates the amount of public transfers by household. A macro model converts changes in sectoral GDP growth projections (in Agriculture, Industry, and Services) into job losses or gains in the corresponding sector, by country. When these job losses or gains by sector are available directly from the Poverty Economist, these are used in place of the elasticity estimates. These job losses/gains per sector are then imposed onto the household survey data. A Probit model, applied to each of the three sectors and by formal and informal workers, provides each person in each household survey a probability of employment in each sector, based on a set of characteristics (sex, age, education level, urban/rural, dependency rate, dummy for member working in public sector, and a dummy for remittances). Workers are ranked by the probability of employment, and those with the lowest employment probability are simulated to lose their job, until total job loss matches the macro projections for job loss by sector. For job gains, new workers are chosen among the unemployed according to the probability of being employed in that sector until total job gains match the macro projections by sector. Workers who are simulated to lose their job experience year-long unemployment and 100% cuts in labor income. In the case of job gains, a new labor income is estimated via a traditional Mincer equation, in which the logarithmic of income is regressed on sex, age, education level, urban, dependency rate, and a dummy if the household has income from remittances. Workers who are employed in Public Administration, Utilities, Health, or Extraterritorial Agencies are protected from facing any job losses or income changes. For the remaining workers (not in protected sectors and who have not lost their jobs), their labor incomes are adjusted up or down by the overall change in private consumption per capita of their specific country. Next, projected changes in remittances by country are also included in the microsimulations. The percent change in remittances at the national level is applied as a direct percent change in the remittance income of any household that receives this income in the household survey. Lastly, coverage of emergency social transfers is simulated based on potential recipients’ eligibility criteria, as determined by each country. Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. These are provided by the World Bank POV GP’s Poverty Economists. In-kind transfers were not included. Coverage may be underestimated in the simulation results, given the assumptions and data restrictions. We use macroeconomic projections to derive the distributional impacts of COVID-19 exploiting heterogeneities in household characteristics in the microdata. The first source of heterogeneity is the difference in the composition of income per-capita by income source. Different shares from each source of income will imply a different effect of the macroeconomic projections for each household. For example, households that do not receive remittances are not affected by the economic shock on remittances. Similarly, household incomes that are derived from pensions, capital income, and public transfer are not affected by the economic shock. The second source of heterogeneity is the sociodemographic characteristics across households. Households differ by gender of the head of household, by education levels of members, by household composition, etc. Different characteristics will predict a different probability to suffer an unemployment shock. The third source of heterogeneity is the sector of employment, as job losses by sector vary according to macroeconomic projections. Each country faces different sectoral GDP projections. Workers in four sectors are protected from any shock: Public Administration, Utilities, Health, and Extraterritorial Agencies 2.4 MACRO-MICROSIMULATION MODEL: ADVANTAGES AND LIMITATIONS This model provides several advantages over other methodologies. First, it provides consistency between macro projections and the corresponding micro impacts, while also applying a consistent framework across all countries in the Latin America and Caribbean region. Second, the methodology can estimate distributional impacts and impact on middle class, thus not just poverty impacts. One can run the full gamut of statistical analysis, including decompositions, etc., on the resulting simulated microdata. Third, the availability of a varible for public transfers (that is part of the SEDLAC harmonization) makes it possibe to estimate the impact of mitigation measures based on income transfers. Lastly, the model can undertake many types of sensitivity analysis regarding the depth of the crisis, the pace of recovery, etc. The model also faces a set of limitations. First, it contains limited behavioral responses. Only the microsimulation model is based on a behavioral model. Instead, some of the job loss projections rely on GDP-to-employment elasticities. A second limitation is that these elasticities can vary considerably depending on the years chosen. However, it is important to note that as countries undertake their labor force surveys and share their employment indicators, the Poverty Economists update these job loss and gain inputs with actual data from official surveys. Third, the focus of the model is on average annual impacts, thus ensuring consistency between the macro and micro projections, so therefore it is difficult to estimate short-term quarterly or monthly impacts due to the lack of macro projections at this level. Lastly, the model does not currently take into account country-differentiated Probit and Mincer models. The country knowledge at a very disaggregated sector-level of the most affected industries is taken into account as the data becomes available. These are areas that could be improved within the model. ANNEX 3. PROJECTED POVERTY – BASELINE INTERNATIONAL POVERTY RATES BY COUNTRY Poverty Estimates ($5.50 per day, 2011 PPP) 2020 2020 Country 2019 Without mitigation With mitigation measures measures Argentina 14.4 18.5–22.9 16–20.2 Bolivia 19.9 27 25.5 Brazil 19.6 20.8–22.9 10.9–13 Chile 3.3–3.4 6.1–8.4 3.3–3.4 Colombia 29.4 34.6–39.6 30.9–34.7 Costa Rica 10.6 14.7–17.3 12.9–13.0 Dominican Republic 12.4 15.9–19.8 11.9–14.2 Ecuador 25.4 29.5–31.9 29.4–31.9 El Salvador 22.3 25.8–29.7 23.3–26.8 Guatemala 43.3 47.2–53.9 43.9–49.3 Haiti 77.6–83 79–87.5 79–87.5 Honduras 49 55.9 55.5 Mexico 20.7 24.9 24.8 Nicaragua 35.5 38.8 38.8 Panama 12.1 20.1–21.1 13.4–14.5 Paraguay 15.4 16.2–17.8 15–16.5 Peru 20.6 30.4–31.8 26.6–28.1 Uruguay 3.2 4.0–5.1 3.5–3.9 LAC 22 26.5 21.8 Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI GP and the POV GP. The current projections shown are based on a macro-micro simulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming). When two projections are shown, the second is based on the POV GP projections as published in the specific country’s Macro Poverty Outlook (April 2021 version). Note: Haiti’s estimates show both the consumption-based and income-based poverty projections using 2012 microdata. Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. Middle Class Estimates ($13 - $70 per day, 2011 PPP) 2020 2020 Country 2019 Without mitigation With mitigation measures measures Argentina 51.1 41.4-46.6 42.6- 47.8 Bolivia 36.5 30.1 30.9 Brazil 44.6 41.9-42.8 47.7 Chile 62.8-63.3 50.3-56.4 53.3-59.4 Colombia 30.5 23.8-26.9 24-27 Costa Rica 50.4 43.8-45.7 47.4-48 Dominican 34.1-38.9 38.8-42.9 Republic 42.4 Ecuador 33.3 30.4 30.4 El Salvador 29.0 23.8-25.6 25.3-27.3 Guatemala 17.5 13.9-16 15.1-17 Haiti 4.6 3.6 3.6 Honduras 17.8 14.2 14.2 Mexico 30.6 27.5 27.6 Nicaragua 20.8 19 19 Panama 56.9 44.6-45.2 50.5-51.3 Paraguay 43.8 40.1-41.7 40.6-42.3 Peru 36.7 25.8-26.6 27.2-28 Uruguay 68.3 63.1-66.4 64.6-67.1 LAC 38 34.7 37.3 Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI GP and the POV GP. The current projections shown are based on a macro-micro simulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming). When two projections are shown, the second is based on the POV GP projections as published in the specific country’s Macro Poverty Outlook (April 2021 version). Note: (a) Haiti’s estimates show both the consumption-based and income-based poverty projections using 2012 microdata. (b) Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. ANNEX 4. POPULATION COVERED BY MITIGATION MEASURES, BY PERCENTILE Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI GP. The current projections shown are based on a macro- micro simulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming). Note: Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. ANNEX 4 (CONT) POPULATION COVERED BY MITIGATION MEASURES, BY PERCENTILE Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from the MTI GP. The current projections shown are based on a macro- micro simulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming). Note: Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. ANNEX 5. BASELINE MACRO PROJECTION PRIVATE CONSUMPTION VOLUME MILLIONS 2015 REAL USD, PER CAPITA (IN MILLIONS) BY COUNTRY 2019e 2020f 2021f 2022f Argentina -7.3% -12.0% 6.7% 1.2% Bolivia 2.2% -9.4% 3.1% 2.3% Brazil 1.4% -7.3% 2.6% 2.1% Chile -0.1% -9.3% 5.7% 3.5% Colombia 1.6% -7.7% 4.1% 3.5% Costa Rica 0.7% -5.3% 1.6% 1.7% Dominican Republic 3.7% -2.4% 3.7% 4.0% Ecuador -1.4% -7.3% 1.6% 0.0% Guatemala 2.7% -4.0% 3.9% 3.0% Honduras 2.4% -9.3% 3.1% 2.7% Haiti -1.7% -3.5% -0.6% -1.6% Mexico -0.5% -11.1% 2.8% 2.6% Nicaragua -3.7% -3.3% 1.1% 1.0% Panama 1.7% -19.8% 10.3% 5.9% Peru 1.3% -11.1% 6.4% 3.0% Paraguay 0.0% -4.7% 3.0% 3.1% El Salvador 2.1% -8.5% 5.7% 2.3% Uruguay 0.1% -5.9% 2.1% 1.8% JOB LOSSES BY SECTOR AND COUNTRY Agro Industry Services Argentina -2.9% -9.7% -7.4% Bolivia -1.9% -14.7% -9.2% Brazil 0.0% -4.1% -8.5% Chile -21.4% -13.3% -10.0% Colombia 1.1% -5.9% -9.4% Costa Rica 1.2% -8.8% -17.6% Domican Republic -6.3% -5.7% -9.9% Ecuador -0.6% -15.3% -2.4% El Salvador -2.8% -2.8% -2.1% Guatemala -2.8% -3.9% -7.7% Haiti -1.5% -5.5% -42.6% Honduras -8.8% -19.2% -10.2% Mexico -3.3% -3.2% -4.8% Nicaragua -0.2% -6.0% -6.7% Panama -3.0% -2.2% -28.2% Paraguay 8.3% -3.7% -3.6% Peru 8.2% -26.0% -30.5% Uruguay 0.0% -6.9% -3.9% 3 REFERENCES Bourguignon, F. and F. Ferreira (2005). “Decomposing Changes in the Distribution of Household Incomes: Methodological Aspects” in The microeconomics of income distribution dynamics in East Asia and Latin America, eds. Bourguignon, F., Ferreira, F., and Lustig, N. Washington, DC: Oxford University Press and The World Bank. Bourguignon, F., M. Bussolo, and L. Pereira da Silva (2008). “The impact of macroeconomic policies on poverty and income distribution: macro-micro evaluation techniques and tools” in The Impact of Macroeconomic Policies on poverty and Income Distribution, eds. Bourguinon, F., Bussolo, M., and Pereira da Silva, L. The World Bank, Washington, DC. Diaz-Bonilla, Carolina, Laura Moreno Herrera, Giselle Del Carmen, Gustavo Canavire Bacarreza, and Diana Sanchez Castro (forthcoming). “The Gradual Rise and Precipitous Fall of Latin America’s Middle Class”. Poverty and Labor Brief. World Bank. July 2020 Ferreira, F., P. Leite, L. Pereira da Silva and P. Picchetti (2008). “Can the Distributional Impacts of Macroeconomic Shocks Be Predicted? A Comparison on Top-Down Macro-Micro Models with Historical Data for Brazil” in The impact of macroeconomic policies on poverty and income distribution: macro-micro evaluation techniques and tools, eds. Bourguinon, F., M. Bussolo, and Pereira da Silva, L. The World Bank, Washington, DC. LAC Equity Lab: https://www.worldbank.org/en/topic/poverty/lac-equity-lab1/overview World Bank. 2020. Global Economic Prospects, June 2020. Washington, DC: World Bank. © World Bank. World Bank (forthcoming), High Frequency Phone Surveys 1st Round, Washington, DC. Written by Paola Ballon, José Cuesta, Sergio Olivieri, and Ana Rivadeneira.