THE GRADUAL RISE and RAPID DECLINE of the Middle Class in Latin America and the Caribbean THE GRADUAL RISE and RAPID DECLINE of the Middle Class in Latin America and the Caribbean LAC Team for Statistical Development Regional Poverty and Inequality Report Poverty & Equity Global Practice May 20, 2021 © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not nec- essarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encour- ages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be ad- dressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522- 2625; e-mail: pubrights@worldbank.org. CONTENTS FOREWORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 SECTION 1. COVID-19 CONTEXT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 SECTION 2. LATIN AMERICA AND THE CARIBBEAN’S (SLOW) SOCIAL GAINS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 SECTION 3. THE VULNERABILITY OF LATIN AMERICA AND THE CARIBBEAN TO THE IMPACTS OF COVID-19​ . . . . . . . . . . . . . . . . . . . 33 SECTION 4. EXPECTED IMPACT OF COVID-19 ON POVERTY AND THE MIDDLE CLASS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 SECTION 5. FINAL REMARKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Annex 1. Reported COVID-19 Cases and Government Response Index, by Country . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Annex 2. Household Surveys from SEDLAC and LABLAC Harmonization . . . . . . . . . . . 57 Annex 3. SEDLAC Survey Availability and Comparability . . . . . . . . . . . . . . . . . . . . . . . . . 58 Annex 4. Methodological Changes in the Surveys and Projections . . . . . . . . . . . . . . . . 59 Annex 5. The Macro-Microsimulation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Annex 6. Growth and Distribution Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Annex 7. Shapley Decomposition by Components of a Welfare Measure . . . . . . . . . . 65 Annex 8. Poverty and Middle Class Estimates, with and without Mitigation Measures, by Country (2020) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Annex 9. Population Covered by Mitigation Measures, by Percentile . . . . . . . . . . . . . . 69 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 About the Poverty Reports in LAC The Poverty and Labor Brief (PLB) and Poverty and PLBs and PIMs are designed to inform fact-based Inequality Monitoring (PIM) series present the latest decision-making and discussion by providing readers trends in poverty, inequality, and shared prosperity with detailed and comparable statistics related to the in Latin America and the Caribbean (LAC) using com- World Bank’s twin goals of eradicating extreme pover- parable regional household and labor force surveys ty and boosting shared prosperity. PLBs offer a deeper (SEDLAC and LABLAC, respectively). The reports are explanation of the labor market as well as other issues produced by the Latin America and Caribbean Team for related to poverty dynamics. PIMs, on the other hand, Statistical Development (LAC TSD) in the Poverty and tend to be shorter and more specific. Along with the Equity Global Practice. previous reports, many of the indicators reported in the series are available at the country level in the LAC Equity Lab website at www.worldbank.org/equitylab. Recent Poverty Briefs: June 2018 Stagnant Poverty Reduction in Latin America (PIM) May 2017 Social Gains Show Signs of Stagnation in Latin America (PIM) May 2016 Inequity in Access to Opportunities April 2016 A Slowdown in Social Gains (PIM) June 2015 Working to End Poverty in Latin America and the Caribbean: Workers, Jobs, and Wages February 2014 Social Gains in the Balance: A Fiscal Policy Challenge for Latin America and the Caribbean June 2013 Shifting Gears to Accelerate Shared Prosperity in Latin America and the Caribbean August 2012 The Effect of Women’s Economic Power in Latin America and the Caribbean On the Edge of Uncertainty: Poverty Reduction in Latin America and December 2011 the Caribbean during the Great Recession and Beyond April 2011 A Break with History: Fifteen Years of Inequality Reduction in Latin America October 2010 Did Latin America Learn to Shield its Poor from Economic Shocks? ​​ 5 FOREWORD The Latin America and Caribbean (LAC) region is at a All subregions saw their middle class grow, with crossroads. The region has been disproportionately Brazil and the southern cone leading the charge. But affected by the COVID-19 pandemic, with many coun- much like with poverty alleviation and reductions in tries still struggling to contain the virus. With 8 per- inequality, the rise of the middle-class was faster in the cent of the global population, the region has borne 20 first decade of the new century (2002-2014), and much percent of infections and 32 percent of accumulated slower in the five years before the COVID-19 pandemic global deaths. The depth of the health crisis has been year (2020). matched by the largest economic recession ever on re- Latin America was a region that was already strug- cord. LAC’s economy contracted by 6.5 percent in 2020, gling with slow growth in the years before the pan- the sharpest regional economic contraction seen since demic, particularly since 2014. The per capita growth we have reliable data. And the rebound in 2021 will be rate for the region was below 1 percent over the past insufficient to return to 2019 GDP levels. decade, a reflection of poor productivity growth and This report documents important trends, facts, and faltering reform efforts. The pandemic intensified in- figures about poverty, inequality, vulnerability, and the equality, with progress reversed back several years. In middle class in LAC. It illustrates that despite twenty addition, most countries in the region saw a deepening years of reductions in overall poverty and inequality, of poverty levels and an erosion of the middle class. and a gradual increase in the size of the middle class, Many countries put in place mitigation mecha- the devastating effects of the COVID-19 pandemic have nisms that helped supplement income losses with led to significant reversals in these gains. emergency transfers. Targeted social transfers were The impact of the pandemic is particularly con- scaled up to include new beneficiaries and prevent cerning given that the region had managed to reduce them from falling into vulnerability and poverty. How- poverty by half between the early 2000s and 2014. Two ever, despite these mitigation measures, millions of decades of gradual improvements in earnings helped people shifted out of middle-class status; some fell lift people out of poverty and vulnerability and into into vulnerability and some into poverty as the ‘new’ the middle class. In fact, in 2018, for the first time ever poor. Brazil is a notable exception, where income loss- in the region, the number of households classified as es were fully mitigated in 2020 thanks through a ramp middle class was larger than the number of house- up in temporary support measures that are unlikely to holds living in either poverty or vulnerability. be sustained for long. 6 Despite the shock waves caused by the pandemic, growing new employment opportunities in these new there are rays of hope in the region. This has also been areas of the economy. The region’s future growth and a time of unprecedented structural transformation prosperity will very much depend on the way Latin in the region. The COVID-19 crisis brought a surge of Americans and their leaders manage to turn these op- high-productivity sectors, including ICT, finance, and portunities into a better future for all. logistics. The use of digital technologies has acceler- ated and promises to change the way Latin Americans Carlos Felipe Jaramillo work, the way they get education, health, finance, and World Bank Vice President, the way they access financial resources. There are Latin America and Caribbean 7 ACKNOWLEDGMENTS This report was produced by the Poverty and Equity World Bank’s views, its Board of Executive Directors, or Global Practice in the Latin America and Caribbean the governments they represent. The World Bank does Region of the World Bank. The core team consisted of not guarantee the accuracy of the data included in this Carolina Diaz-Bonilla, Laura Moreno Herrera, Giselle work. Nothing herein shall constitute or be considered Del Carmen, Diana Sanchez Castro, Gustavo Canavire to be a limitation upon or waiver of the privileges and Bacarreza, Karen Barreto Herrera, Hernan Winkler, immunities of the World Bank, all of which are specifi- Flavia Sacco Capurro, and Pamela Gunio. The team cally reserved. worked under the guidance of Ximena Del Carpio. The numbers presented in this brief are based on The team received valuable country-specific inputs two regional data harmonization efforts known as the from Maria Davalos, Alejandro De la Fuente, Gabriela Socio-Economic Database for Latin America and the Inchauste, Jacobus Joost De Hoop, Roy Katayama, Caribbean (SEDLAC) and the Labor Database for Latin Gabriel Lara Ibarra, Sergio Olivieri, Monica Robayo, America and the Caribbean (LABLAC), joint efforts of Lourdes Rodriguez Chamussy, and Javier Romero; as the World Bank and the Center of Distributive, Labor well as Agustin Arakaki, Juan Pablo Baquero, Kiyomi and Social Studies (CEDLAS) at the National University Cadena, Oscar Roberto Castillo Anazco, Ronald Cueva of La Plata in Argentina. They increase cross-country Chavez, Jia Gao, Christian Gomez Canon, Julieta La- comparability of selected findings from official house- dronis, Rafael Macedo Rubião, Juan Manuel Monroy, hold and labor surveys. For that reason, the numbers Eder Oliva, Julieth Pico Mejia, Ana Rivadeneira, Trini- discussed here may be different from official statistics dad Saavedra Facusse, Fabio Saia Cereda, Diego Tuz- reported by governments and national offices of statis- man, and James Sampi Bravo. Additional support was tics. Such differences should not be interpreted in any provided by Cristina Cifuentes and Desiree Gonzalez. way as a claim of methodological superiority, because The document was edited by Daniel McNaughton. both sets of numbers serve the same important ob- The findings, interpretations, and conclusions jectives: regional comparability and the best possible expressed in this work do not necessarily reflect the representation of the facts of individual countries. The welfare aggregate used in this study is income based. 8 EXECUTIVE SUMMARY THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 9 Latin America and the Caribbean (LAC) reported pled with the dramatic fall in activity caused by the over 30 million Coronavirus (COVID-19) cases and COVID-19 crisis will negatively impact living standards around 960,000 deaths as of May 2021. Official track- and well-being across the region. Poverty projections ing data shows that Brazil, Colombia, and Argentina for 2020 suggest that the number of the poor increased have the highest number of reported cases throughout in most LAC countries. Brazil, however, implemented a LAC, which in turn is the region with among the high- generous emergency transfer program that benefited est numbers across all developing regions. Moreover, 1 almost 67 million people and lifted millions out of pov- Brazil is the third-worst affected country worldwide, erty. As a result, poverty in the LAC region is expected after the United States and India, with approximately to decline marginally from 22 percent in 2019 to 21.8 15.4 million infections. Dramatic declines in economic percent in 2020. Had no mitigation measures been im- activity are expected throughout the LAC region due to plemented, the region may instead have seen 28 mil- the global pandemic. Unfortunately, many LAC coun- lion new poor in 2020. tries entered the crisis with low potential economic The current global crisis is expected to result growth and high levels of inequality, following the re- in a rapid decline in the size of the middle class in gion’s recent period of stagnant growth. most countries, setting LAC back as a majority-mid- The 2020 COVID-19 crisis will likely reverse in a dle-class region. By 2019, Latin America was predom- short time frame many of the social gains that took inantly a middle-class region, with 38 percent of its decades to materialize in Latin America and the Ca- population, approximately 230 million people, reach- ribbean. In the past two decades, the region has seen ing middle-class status. However, this socioeconomic a reduction in the number of people living in poverty group is projected to have declined to 37.3 percent of by nearly half and an increase in the size of its middle the population in 2020, resulting in a net loss of 4.7 class. Income inequality also decreased, as income 2 million people from the middle class. Without mitiga- growth has been primarily pro-poor in recent years. tion measures, in particular without Brazil’s emergen- Despite variations across countries, most have expe- cy transfers, projections suggest the global pandemic rienced positive welfare gains since the early 2000s. 3 could have resulted in more than 20 million people los- However, the growth deceleration of 2014–2019 cou- ing middle-class status. 1 Data from the Center for Systems Science and Engineering (CSSE), Johns Hopkins University, at https://systems.jhu.edu/. 2 The LAC aggregate used for poverty, inequality, and the middle class is based on 18 countries in the region for which mi- crodata are available at the national level (i.e., “LAC-18”) over a long period of time. In an effort to increase cross-country comparability, microdata are harmonized as part of the project called the Socio-Economic Database for Latin America and the Caribbean (SEDLAC), a joint effort between CEDLAS and the World Bank). In cases where data are unavailable for a given country in a given year, values have been interpolated using WDI data only for the purpose of calculating regional measures. 3 Welfare monitoring and poverty estimation requires reliable and frequently collected household-level data. In the case of Venezuela, such data have not been shared with the World Bank since the mid-2000s, barring the World Bank from calcu- lating reliable poverty and inequality estimates for the country and a Venezuela-inclusive regional aggregate for the LAC region. The most recently collected data are not available for public use, and in addition, due to the ongoing crisis and economic instability in the country, we believe that any estimation of poverty from such a survey will inherently suffer from many shortcomings. 10 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Persistent inequalities throughout the LAC re- the high degree of uncertainty as to the impact and du- gion add to the challenge of overcoming the ongo- ration of the COVID-19 crisis, especially if another wave ing crisis and will likely result in unequal pandemic hits the region, LAC countries should prioritize equita- impacts. Latin American countries face high levels of ble access to essential services such as water, sanita- informality and self-employment, particularly among tion, and electricity. the poor, resulting in lower-quality and more-vulner- The recovery from 2021 onward may also de- able jobs. The crisis has put governments and health pend on the vaccine rollouts. Latin American coun- systems under immense stress, highlighting the re- tries face important challenges in terms of vaccine gion’s limited access to and quality of affordable health rollout, and to date only Chile has reported significant care. In addition, lockdown measures implemented to progress. Chile has administered more than 49 doses contain the virus’s spread underline inequities in ac- per 100 people (above the United States—40—and the cess to basic services, such as electricity, water and United Kingdom—47). In contrast, other LAC countries sanitation, and even the internet. Households who lag significantly, with only 1 to 13 doses having been were already poor, and have now lost further human administered for every 100 people.4 Across the region, or physical capital accumulation, will have the hardest governments are having problems securing enough time recovering from this crisis, and inequality across doses of vaccines to cover their populations, as well as multiple dimensions is likely to get worse. efficient and effective systems to distribute and admin- The LAC region must continue to target policies ister them. to prevent contagion and support the most vulner- The accelerated digital transformation in the able populations, yet careful to protect livelihoods. region has been an unexpected positive outcome Moreover, as lockdown measures are phased out, of the COVID-19 pandemic. Stay-at-home orders and governments should address preexisting inequities. social distancing have highlighted the need for alter- Though general lockdowns are the most effective way native methods of purchasing goods and services. A to prevent mass contagion, they come at the expense significant boost in e-commerce and e-services has of an increase in unemployment, general loss of in- been evident throughout the region as several super- come, and an increase in poverty. Well-targeted tem- markets and restaurants have shifted to online delivery porary income transfers do provide vulnerable groups services, in some cases through social media. More- with at least some income security during containment over, several governments have switched to online periods. Although these policies adequately support platforms to continue operating. It is unlikely firms will low-income households, they are temporary, and in return to business as usual once the pandemic ends. addition, they may not be enough to prevent the sharp Thus, countries should continue to invest in digital in- decline of the middle class. On the other hand, social frastructure to accelerate these changes and enact leg- protection programs can be re-assessed to adjust their islation for the expansion of the digital economy. reach and thus incorporate new beneficiaries. Given 4 Data from Our World in Data, https://ourworldindata.org/covid-vaccinations. Accessed on 05/13/2021 11 SECTION 1. COVID-19 CONTEXT 12 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN As of May 2021, the Latin America and the Carib- with more COVID-19 cases identified in urban centers bean (LAC) region has reported over 30 million (Maps 1.1 and 1.2). Coronavirus (COVID-19) cases and approximately Measures to slow the spread of the virus and to 960,000 deaths. The region’s two most populous na- mitigate the resulting economic and poverty im- tions, Brazil and Mexico, have seen the highest number pacts have varied across the region. Most countries of deaths, with more than 428,000 and 219,000, respec- have adopted nationwide lockdowns—in some cas- tively. Brazil is the third-worst affected country world- es voluntary and in others compulsory—to slow the wide, after only the United States and India, with ap- spread of the virus. Some of the most common mitiga- proximately 15.4 million infections. However, Panama tion measures include the use of face masks and cover- has the highest number of cases per 1 million people ings in public spaces, health screenings, quarantines, (approximately 87,000), followed by Brazil (73,000) and school closures, and travel restrictions. In response Argentina (71,000). Within countries there are substan- to lockdowns, countries have had to apply fiscal and tial differences in terms of the impact of the disease, monetary measures to protect the most vulnerable Map 1.1 Map 1.2 Confirmed cases of COVID-19 in Latin Confirmed deaths from COVID-19 in Latin America and the Caribbean America and the Caribbean Esri, HERE, Garmin, USGS Esri, HERE, Garmin, USGS Source : Public health ministries. Note : Confirmed cases of COVID-19 in these maps are as of March 25, 2021. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 13 and contain the fall of economic activity. Along these the crisis with low potential for growth. Over the lines, most LAC countries have introduced, adapted, or last decade, there are two distinctive periods in terms expanded their social protection programs in response of growth and poverty in the region: one of sustained to COVID-19. In Brazil, the government expanded the 5 growth and strong poverty reduction (the “Golden Bolsa Familia program and implemented emergency Decade” up to 2013), and a subsequent one of poverty measures that reached almost 67 million people. Sim- stagnation and growth deceleration (2014 to 2019).8,9 ilarly, most nations have approved additional borrow- Just when the region was experiencing a glimpse of ing to support the public health systems and provide recovery from the stagnation period, the COVID-19 out- support to the vulnerable. The amount ranges from 1 6 break forced economies to shut down. The LAC econo- percent of GDP in Uruguay and Mexico to 11 percent in my is expected to contract by 6.7 percent in 2020, mak- Brazil. Moreover, several countries have extended tax ing it the most profound recession in the region,10 with filing deadlines and temporarily suspended payment significant differences across countries. Brazil’s GDP de- by households for some public services (water, elec- clined by 4.1 percent, whereas Central America’s GDP is tricity, telephone, internet). 7 expected to contract by 6.1 percent. Similarly, per cap- Dramatic declines in economic activity are ex- ita, household income is estimated to have decreased pected throughout the LAC region due to the glob- by 3.2 percent in 2020. By 2021, the LAC economy is ex- al pandemic; additionally, many countries entered pected to recover, growing by 4.4 percent (Figure 1.2). 5 (a) Economic Commission for Latin America and the Caribbean (ECLAC) 2020 and (b) Gentilini et al. 2020. 6 Many countries in LAC have requested emergency financing from the International Monetary Fund (IMF), the World Bank, the Inter-American Development Bank (IADB), and the Development Bank of Latin America (CAF). 7 For more details on government responses, see IMF, Policy Responses to COVID-19, at https://www.imf.org/en/Topics/imf- and-covid19/Policy-Responses-to-COVID-19#H. 8 The first period (2003–2013) saw the LAC-18 region registering GDP per capita growth at a higher rate than the world aver- age, with the exception of the global Great Recession of 2009. This period of mostly continuous growth was characterized by a sustained reduction in poverty. Similarly, growth in Per Capita Household Income (PCHI) (from the household surveys) was positive during the whole period and in line with macro trends. In the second period (2014–2019), the region experi- enced growth deceleration and a contraction, growing far less than the world. Poverty decreased slightly or stalled com- pletely, depending on the poverty threshold, and PCHI decreased in 2015 for the first time in a decade and remained largely stagnant from 2016 to 2019. 9 Brazil’s importance is evident in both its share of the total LAC population (almost 38 percent) and its share in the popula- tion living in poverty (almost half of the total population of LAC under the International Poverty Line of $1.90 per day, 2011 PPP). Hence changes in poverty in the LAC aggregate are mainly driven by changes in Brazil. During the 2014–2018 stagna- tion period, some countries like El Salvador and Panama actually managed to lift people out of poverty. However, because of the smaller populations of these countries, these instances of poverty reduction do not visibly impact the regional aggre- gate. 10 During Latin America’s Debt Crisis of the 1980s and the 2009 global financial crisis, the LAC economy contracted by 2.5 and 1.9 percent, respectively (World Bank 2020b). The estimates for 2020 are from World Bank 2021b. 14 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Box 1.1 Government Response to the COVID-19 Outbreak in LAC The Oxford COVID-19 Government Response Tracker (OxCGRT) provides a systematic cross-coun- try, cross-temporal measure to understand how government responses have evolved during the COVID-19 outbreak. Daily information on government interventions is monitored across a set of stan- dardized indicators to generate four composite indices: (1) containment and health, (2) stringency, (3) economic support, and (4) an overall government response index. These cover data on school closures, stay-at-home orders, restrictions on domestic and international travel, fiscal measures, and emergency investment in health care, among others. Indices range from 0 to 100 and are intended to reflect govern- ment action level—not their effectiveness—in certain dimensions. Governments throughout LAC have varied in the measures they have taken to mitigate the COVID-19 outbreak and how quickly they have adopted them (Annex 1). Except for Nicaragua, all countries expanded their policy responses as the number of confirmed COVID-19 cases increased throughout the region. In the case of Brazil and Mexico, though these countries have some of the highest infection rates, the governments have enacted less-severe lockdown measures than their peers. Others, like El Salvador and Costa Rica, implemented mitigation measures before the first reported case. As of May 12, 2021, Nicaragua shows an overall lower policy-action level than the average LAC country. Con- versely, El Salvador has a high level of response and a low number of cases (Table 1.1 and Figure 1.1). Table 1.1 Reported COVID-19 cases and maximum government response level, as of May 12, 2021 Number of Maximum government Country COVID-19 cases response level Brazil 15,400,000 78.4 Argentina 3,215,572 84.9 Colombia 3,048,719 83.3 Mexico 2,371,483 66.9 Peru 1,865,639 81.8 Chile 1,260,448 87.6 Ecuador 404,632 78.1 Panama 368,930 79.2 Bolivia 322,578 71.4 Paraguay 304,889 71.9 Costa Rica 276,887 60.4 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 15 Number of Maximum government Country COVID-19 cases response level Dominican Republic 273,497 77.6 Guatemala 237,682 76.6 Uruguay 228,102 81.6 Honduras 222,992 83.3 Venezuela, RB 210,948 78.9 El Salvador 70,915 85.9 Haiti 13,227 61.5 Nicaragua 7,086 20.0 LAC 30,104,226 74.2 Figure 1.1 Response-to-risk ratio 90 SLV HND CHL ARG URY PAN PER COL 80 Maximum Government Response Level VEN ECU DOM GTM 70 PRY BOL MEX 60 HTI CRI 50 40 30 20 NIC 10 0 0 500,000 1.000,000 1.500,000 2.000,000 2.500,000 3.000,000 3.500,000 Number of COVID 19 - Cases Source: Thomas Hale, Sam Webster, Anna Petherick, Toby Phillips, and Beatriz Kira 2020. Oxford COVID-19 Government Response Tracker. Blavatnik School of Government. Available at www.bsg.ox.ac.uk/covidtracker. Note: Brazil was excluded from the scatter plot due to the high volume of cases reported. Data as of May 12, 2021. Countries above the trend line can be interpreted as having stricter measures than the average LAC country per their number of confirmed cases. Conversely, countries below the line show a lower level of policy action. 16 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Two discernible periods of growth in Latin America in the last decade Figure 1.2 Growth of GDP in Latin America and the world, and growth of mean Per-Capita Household Income (PCHI) Sustained Growth Stagnation COVID-19 8 6 4 2 0 % -2 -4 -6 -8 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020e 2021f GDP - World GDP - LAC PCHI- LAC17 Source : SEDLAC (CEDLAS and World Bank); WDI 2020; World Bank 2021b; and original calculations for this publication. Note : Per capita household income (PCHI) is calculated using pooled data as described in Annexes 2–4 for LAC-18. PCHI projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata projected using National Accounts data (private consumption per-capita), job losses, and remittances from the World Bank Macroeconomics, Trade and Investment Glob- al Practice (MTI GP) and the Poverty and Equity Global Practice (POV GP), based on a macro-microsimulation model that assumes 12 months of labor income loss and mitigation measures. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming) and Annex 5. The weak growth in GDP per capita during 2 percentage points (Figure 1.3).11 This highlights that the stagnation period was worse for the poorest economic growth in the most recent years was not households. Between 2015 and 2019, GDP per capi- enough to restart social gains, particularly among the ta increased by 0.1 percent and poverty rates under poorest households who are also among the most vul- the international poverty line (IPL) increased from 4.2 nerable to the effects of the COVID-19 crisis. percent to 4.4 percent, while poverty under the $5.50 The region’s high levels of inequality and infor- per day poverty line (2011 PPPP) decreased by nearly mality pose a significant challenge in relation to 11 The International Poverty Line (IPL) is USD 1.90 per day per person in 2011 Purchasing Power Parity (PPP). All dollar amounts are US dollars unless otherwise indicated. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 17 The growth elasticity of poverty was higher for poorer households in the past, but the current weak economic growth is not enough to help the poorest. Figure 1.3 Changes in poverty, GDP, and elasticity Elasticity of Poverty to GDP Per Capita Growth 2% 20 1% 15 Change in Poverty (annualized) 0% 10 - 1% 5 - 2% 0 - 3% -5 - 4% - 10 - 5% - 15 - 6% - 7% - 20 - 8% - 25 - 9% - 30 $1.90 $5.50 $1.90 $5.50 2008 - 2014 2015 - 2019 (1.8% GDP pc growth) (0.1% GDP pc growth) Poverty Change Elasticity Source : Author’s estimations using SEDLAC (CEDLAS and World Bank) and World Development Indicators 2020 . Annual- ized growth rates. Note : For the years of early stagnation, we have an important break in the survey series for Brazil and Mexico, as ex- plained in Annex 4. GDP refers to GDP at factor cost (NY.GDP.FC.ST.KD). GDP per capita growth is a change of GDP over total population between two periods. overcoming the pandemic. Many countries continue The 2020 crisis will likely reverse, in a short time to present high income-inequality levels as mirrored by frame, many of the social gains that took decades the Gini Coefficient (see Figure 2.9) and unequal access to materialize in Latin America and the Caribbe- to basic services. Moreover, Latin American countries an. During the last 20 years, the region has managed face high levels of informality and high self-employ- to reduce poverty by nearly half, decrease income in- ment levels, resulting in lower-quality and more-vul- equality, and simultaneously increase the size of the nerable jobs. High levels of remittances in some coun- middle class, making LAC in 2019 a predominantly tries are also seeing dramatic drop-offs, affecting poor, middle-class region (Figure 1.4). Similarly, access to near-poor, and even middle-class households (see Sec- basic services has improved throughout Latin America tion 4). The crisis has put governments and health sys- (see Section 3). However, the 2020 crisis is expected to tems, as well as the economy and employment, under have led to poverty increases in almost all countries, immense stress. with the important exception of Brazil, thanks to the 18 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 1.4 Poverty trends and projections (2000–2020) 50 45 40 38 38.5 Vulnerable $5.5 - $13 37.3 35 37 Headcount (%) 30 Middle Class $13 - $70 25 Poor $5.5 21.8 20 22 15 10 5 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020e Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consumption per capita, job losses, and remittances from MTI and POV GPs. The current projections shown are based on a macro-microsimulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming) and Annex 5. Note: (1) The Latin America and Caribbean regional aggregate is estimated based on a sample of 18 countries (LAC-18); see Annex 2 and Annex 3. In cases where data are unavailable for a given country in a given year, values have been inter- polated or extrapolated using World Development Indicator (WDI) data and then pooled to create the regional estimate. (2) Due to important methodological changes in Mexico’s official household survey in 2016 that created a break in the poverty series, we have created a break in the LAC-18 aggregate. More details are available in Annexes 3 and 4. government’s generous, but temporary, emergency 20 million people across the region fell into poverty transfer program that benefited almost 67 million Bra- (below the $5.50 poverty line), with another 1.4 mil- zilians. Poverty in the LAC region is expected to de- 12 lion increase due to population growth. On the other cline from 22 percent in 2019 to 21.8 percent in 2020 hand, social transfers across the region lifted 22 million due to Brazil’s transfers (see Section 4), for a net de- people out of poverty, of whom more than 77 percent cline of almost 400 thousand poor. This net decline is were in Brazil. Poverty in the LAC region without Brazil, a combination of transitions into and out of poverty for however, is projected to have increased by 3 percent- different households. Projections suggest more than 12 World Bank Macro Poverty Outlook, available at https://pubdocs.worldbank.org/en/114751582655277329/mpo-bra.pdf (April 2021). Brazil’s Auxilio Emergencial (AE), conceived as a temporary program, ended in December 2020. However, the government of Brazil launched a new wave of AE in April 2021 with lower benefits that targeted about 44 million individuals. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 19 Figure 1.5 LAC poverty and middle-class trends and projections, with and without Brazil (2019–2020) 38.4 37.3 35.0 31.6 26.7 23.5 22.1 21.8 2019 2020 2019 2020 LAC with Brazil LAC without Brazil Poor $ 5.50 Middle Class ($13 - $70) 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-microsimulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forth- coming) and Annex 5. age points between 2019 and 2020 (Figure 1.5), for a ed to have declined to 37.3 percent of the population in net increase of almost 13.7 million people in poverty. 2020, resulting in a net loss of 4.7 million people from After decades of a gradual rise in incomes, LAC of the middle class, while the vulnerable class has in- had finally become a predominantly middle-class creased to 38.5 percent. Without mitigation measures, region, until the global pandemic caused millions in particular without Brazil’s emergency transfers, pro- to lose income and fall out of middle class status. jections suggest the global pandemic could have re- LAC’s middle class, defined as those households whose sulted in 20 million people losing middle-class status income is between $13 and $70 per day (2011 PPP), (see Section 4). became the largest income class in the region for the This report presents an overview of LAC’s social first time in 2018. By 2019, 38 percent of the popula- gains in the past two decades, followed by an analysis tion (around 230 million people) were considered mid- of the region’s vulnerability to the pandemic. The last dle class, while 37 percent were considered vulnerable two sections include poverty and inequality projec- class (those living on $5.50 to $13 per day; around 220 tions in light of the COVID-19 outbreak and a series of million people), and 22 percent poor. Despite region- policy recommendations going forward. wide mitigation measures, the middle class is project- 13 13 Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. 20 SECTION 2. LATIN AMERICA AND THE CARIBBEAN’S (SLOW) SOCIAL GAINS THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 21 Over the last decade, LAC saw a 22 percent gain in terval (between 2014 and 2019), these trends slowed average per capita household income. In 2008, the markedly, to -0.9 percentage points for poverty and average PCHI was about $13.9 a day in 2011 PPP pric- +1.5 percentage points in the middle class. The final es. By 2018, the region’s average PCHI was $17.1 a day year shows that about 2 in 5 people in LAC lived on (2011 PPP), or almost 3 PPP dollars more per day on more than $13, classifying them as either middle class average. Similarly, the median per capita household in- or higher, a number slightly higher than in 2014. come increased from $8 per day (2011 PPP) in 2008 to All regions experienced an increase in the size of $10.6 a day in 2018. Although the region’s median per their middle class over the decade, but at different capita income grew by 25 percent in the 10-year period, paces. As with median and mean income, the LAC re- the growth at the subregional level was not homoge- gion has clear differences by subregions across the neous, as Mexico and Central America became a larger income distribution. As the region became wealthier, share of the population below LAC’s median PCHI. Bra- poverty declined, and both the size and share of the zil, the Andean Region, and the Southern Cone became middle class became bigger. However, using the same relatively wealthier than Mexico and Central America in picture of the distribution as in Figure 2.1 but over- this period. lapped with the different income thresholds, one can In 2018, for the first time in nearly two de- see differences in the trends in middle-class growth cades, LAC’s middle class became the largest across subregions (Figure 2.2). The Andean subregion’s socioeconomic group. The middle class, defined as middle class saw a large increase, from 23.8 percent in people with incomes between $13 and $70 a day in 2009 to 32 percent in 2014, and then a much smaller 2011 PPP, increased from an estimated 28.9 percent one, reaching 33.4 percent in 2019. Similarly, Brazil’s of the population in 2008 to 37.4 percent in 2018 and middle class grew strongly from 34.6 percent to 44.5 38 percent in 2019. The size of the vulnerable class percent in 2009–2014, and then more slowly to 44.6 (or aspiring middle class), defined as people with in- percent by 2019. Central America’s middle class, on comes between $5.50 and $13 a day (2011 PPP), hardly the other hand, experienced a very modest increase changed throughout the decade. This class made up 36 between 2009 (22.6 percent) and 2014 (24.6 percent), percent of the population in 2008 and 36.9 percent in but then grew strongly in the next six years to reach 2019 (Figure 2.1). 29.3 percent of the region’s population. The South- The movement towards a larger middle class ern Cone’s middle class grew almost continuously was stronger between 2002 and 2014 than during throughout the decade, from 44.3 percent in 2009 to 55 the most recent five-year period (2014–2019). The percent in 2019. The size of Mexico’s middle class, on regional distribution barely changed between 2014 the other hand, barely changed in the past decade. and 2019, affecting the “almost middle-class” trend At 42 percent, Brazil accounts for the largest in 2002–2014 (Figure 2.1). Between 2002 and 2014, as share of the Latin American middle class. Although poverty fell from 43 percent to 25 percent, the middle its share is much smaller, the Andean subregion’s mid- class increased from 22 percent to 35 percent of the dle class increased from 13.3 to 16.5 percent between regional population. This means that poverty declined 2002 and 2019. On the other hand, Mexico’s share of the by 18 percentage points and the middle class grew by Latin American (LA) middle class declined from 19.3 13 percentage points. In the most recent six-year in- percent to 17 percent during the same period. Similar- 22 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Changes in distribution were small during 2014–2019 compared with 2002–2014 Figure 2.1 Distribution of income in Latin America and the Caribbean 2002 2014 .6 .6 .6 .6 .4 .4 .4 .4 Density Density Density Density .2 .2 .2 .2 43% 43% 33% 33% 22% 22% 25% 25% 37% 37% 35% 35% Poor Poor Vulnerable Middle Vulnerable Middle Class Class Poor Vulnerable Middle Poor Vulnerable Middle Class Class 0 0 0 0 $5.5 $5.5 $13 $13 $70 $70 $5.5 $5.5 $13 $13 $70 $70 Per Capita Per Capita Income Income USD USD 2011 2011 PPP PPP Per Per Capita Capita Income USD 2011 Income USD 2011 PPP PPP 2002–2019 .6 .6 .4 .4 Density Density .2 .2 1% 1% 37% 37% 38% 38% Vulnerable Middle Poor Vulnerable Poor Class Middle Class 0 0 $5.5 $5.5 $13 $13 $70 $70 Per Capita Per Income USD Capita Income 2011 PPP USD 2011 PPP Source : SEDLAC (CEDLAS and World Bank). Note: Pooled data using logarithmic scale and values lower than 99% of the distribution. Solid lines indicate the differ- ent poverty and income thresholds; dotted line in the 2002–2019 graph refers to the income distribution in 2002. ly, Central America’s share of the LA middle class de- to 11.5 percent during the same period. While pover- clined at first and then remained relatively stable in the ty reduction in Central America was stagnant during past decade, with 7.7 percent by 2019 (Table 2.1). most of LAC’s ‘Golden Decade,’ the Andean region and Likewise, LAC has reduced poverty by nearly half Southern Cone experienced steep declines. Howev- since 2000, with differences across countries. With er, from 2013 to 2019 the trends reversed, as Central a $5.50 (USD 2011 PPP) a day poverty line, the region’s American countries led poverty reduction in LAC with population living in poverty fell from 44.5 to 26.5 per- an 8.5 percentage point decrease under the $5.50 (USD cent between 2000 and 2013. Similarly, with a $3.20 2011 PPP) line (Figure 2.3). per day (USD 2011 PPP) poverty line, it fell from 24.8 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 23 Box 2.1 Defining the vulnerable and middle classes Evidence-based policy making requires an analysis of transitions in and out of poverty. It is important to not only lift the poor above the minimum income threshold (poverty line), but to protect the vulnerable (those close to the poverty line) from falling into poverty. Moreover, as countries grow and move toward middle-class income status, which characterizes most LAC countries today, it becomes imperative to analyze the transitions into the middle class over time. Thus, at any point in time, an individual can be classified as poor (based on the IPL, lower middle-income [LMIC], and/or upper middle-income [UMIC] thresholds), as vulnerable, or as being in the middle class. These classifications are dependent on eco- nomic stability, i.e., low transition probabilities of falling in and out of poverty. For example, an individual is defined as vulnerable if the probability of falling back into poverty over a five-year interval is greater than 10 percent, which is approximately the average probability of falling into poverty in countries like Argentina, Colombia, and Costa Rica (Ferreira et al. 2012). This, in turn, yielded an upper bound of $10 per person per day in 2005 PPPs to be classified as vulnerable. This upper bound also served as the lower bound for the classification of individuals in the middle class. The upper bound for the middle class was set at $50 per person per day using self-perceptions data, based on analysis from Brazil, Chile, Colombia, Mexico, and Peru. The key consideration here is to pick an income threshold that is robust to changes in the distribution of income right around the threshold, so that changing the upper bound slightly should not move a significant proportion of people in and out of the middle class. In contrast, moving the lower bound should significantly affect the percentage of the excluded/included population. More recently, the thresholds were updated to be expressed in terms of 2011 PPPs. First, using the 2005 PPP conversion factor of each country, the vulnerable- and middle-class lines were converted to local cur- rency units at 2005 prices. Second, these values were deflated to 2011 prices using each country’s CPI and converted back to US dollars using their corresponding 2011 PPP conversion factors. Finally, a simple aver- age of the resulting lines was taken to obtain a regional value. By rounding to the closest unit, the vulnera- ble- and middle-class lines in 2011 PPP for LAC were then set at US$5.5–13 and $13–70 a day, respectively. This implies that an individual/household can be (1) earning below the IPL ($1.90), (2) earning be- tween the IPL and the LMIC line ($1.90 to $3.20), (3) earning between the LMIC and UMIC lines ($3.20 to $5.50), (4) vulnerable, i.e., earning between $5.50 and $13; (5) be in the middle class (earning $13–$70), or (6) earning more than $70.14 14 14 Previously, the income thresholds used to identify the vulnerable and the middle classes were set at $4–10 and $10–50 vis-à- vis the $1.25, $2.50, and $4 as the IPL, LMIC, and UMIC thresholds, respectively. All these were expressed in terms of 2005 PPPs. 24 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Fewer people in poverty, more moving in the middle class, but at different paces. Figure 2.2 Distribution of income in Latin America and the Caribbean by subregion 2009 2019 2009 2019 50,000 Subregion 50,000 Subregion Brasil Brasil Population (in thousands) Population (in thousands) Mexico Mexico 40,000 40,000 Andean Region Andean Region Central America Central America 30,000 30,000 Southern Cone Southern Cone 20,000 20,000 10,000 10,000 0 0 $1.9 $3.2 $5.5 $13 $70 $1.9 $3.2 $5.5 $13 $70 Per Capita House Income Per Capita House Income Source: SEDLAC (CEDLAS and World Bank) pooled data. Note : Data representation uses the logarithmic scale and trims values higher than 99% of the distribution. Lines indicate the different poverty and income thresholds. In cases where data are unavailable for a given country in a given year, values have been interpolated or extrapolated using WDI data (World Bank 2020b) and then pooled to create the region- al estimate. More details are available in Annex 3 and Annex 4. See Annex 2 for complete information on the availability of surveys. Table 2.1 Share of the Latin American middle class by subregion and by year (%) REGION 2002 2008 2014 2019* Andean 13.3 14.7 16.8 16.5 Brazil 42.7 41.7 45.3 41.6 Central America 9.1 7.6 6.9 7.7 Mexico 19.3 18.7 13.6 17.0 Southern Cone 16.4 18.7 18.8 18.1 Source: SEDLAC (CEDLAS and World Bank) pooled data. Note: The Andean subregion is the aggregate of Bolivia, Colombia, Ecuador, and Peru; the Central American subregion is the aggregate of Costa Rica, Guatemala, Honduras, Nicaragua, Panama, El Salvador, and the Dominican Republic; and the Southern Cone subregion is the aggregate of Argentina, Chile, Paraguay, and Uruguay. *Latest data for Mexico is 2018. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 25 Figure 2.3 Poverty trends by subregion (2000–2019) 70 58.1 60 50 38.9 40 44.5 % 29.0 30 27.8 30 25 26.5 20 22 10.1 10 10 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Andean Region Central America LAC Southern Cone Source: SEDLAC (CEDLAS and World Bank) and World Development Indicators 2020. Note : (1) The Latin America and Caribbean regional aggregate is estimated based on a sample of 18 countries (LAC-18); see Annexes 2 and 3. In cases where data are unavailable for a given country in a given year, values have been inter- polated or extrapolated using WDI data (World Bank 2020b) and then pooled to create the regional estimate; (2) due to important methodological changes in Mexico’s official household survey in 2016 that created a break in the poverty series, we have created a break in the LAC-18 aggregate. More details are available in Annex 3; see Annex 4 for more information. Despite accounting for a lower share of the over- tral American countries having deeper differences all population, rural areas continue to host a larger than the rest of the region. Colombia, Bolivia, and number of the poor in LAC. In 2019, rural areas hosted Peru have rural-urban $5.50-based poverty differenc- 21 percent of the total population while accounting for es of between 27 and 45 percentage points; in Central over 54 percent of the poor population under the $1.90 America, Honduras, Nicaragua, and Guatemala face line and 41 percent of the poor under the $5.50 2011 rural-urban poverty differences of about 30 percent- PPP line. Moreover, despite a reduction in overall pov- age points. Countries with relatively low poverty levels erty over time, the rural-urban poverty gap persists. tend to have very small gaps (mostly per definition), Based on the $5.50 poverty line, the poverty headcount with the notable exception being Panama, where the was 60.1 percent in rural areas and 14.8 percent in ur- poverty rate is less than 12.1 percent, but there is a ban areas in 2019 (Figure 2.4). sizeable gap between urban and rural areas (23.3 per- This contrast between urban and rural areas is centage points) (Figure 2.5). not shared by all countries, with Andean and Cen- 26 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 2.4 Figure 2.5 Urban vs. rural poverty headcounts under Urban vs. rural poverty under the $5.50 the $1.90 a day line (IPL) and $5.50 a day a day line, last year of data for each (UMIC) (circa 2019) country (circa 2019) 90 70 80 60 70 50 % 60.1 60 40 50 30 % 40 20 30 10 21.0 20 0 14.8 10 Dominican Rep. Uruguay Chile Costa Rica El Salvador Paraguay Ecuador Brazil Mexico Panama Bolivia Nicaragua Colombia Guatemala Honduras LAC Peru 2.6 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 IPL Urban IPL Rural UMIC Urban UMIC Rural Gap Rural Urban Source: LAC Equity Lab using SEDLAC (CEDLAS and World Bank). Note : Data are from 2019 or the closest available dataset for each country. The Latin America and Caribbean regional aggregate is estimated based on a sample of 18 countries (LAC-18); see Annexes 2 and 3. In cases where data are un- available for a given country in a given year, values have been interpolated or extrapolated using WDI data (World Bank 2020b) and then pooled to create the regional estimate. Due to important methodological changes in Mexico’s official household survey in 2016 that created a break in the poverty series, we have created a break in the LAC-18 aggregate. More details are available in Annex 3; see also Annex 4. In line with the slow poverty reduction in recent with total population growth rates of about 5 percent, years, progress on shared prosperity indicators has registered annualized increases of 7 or more percent also been limited. Prior to the economic slowdown, 15 for the B40. This, however, changed significantly after LAC shared prosperity was relatively high, with the re- 2014. Between that year and 2019, although the region gion being second among the World Bank regions in reported a positive shared prosperity premium, at 0.45 terms of average shared prosperity. For instance, be- 16 percentage points it was much lower than the previous tween the years 2009 and 2014, income growth for the five-year period. Moreover, the top five performers (El bottom 40 percent (B40) (4.1 percent) was nearly one Salvador, Dominican Republic, Panama, Chile, and Bo- percentage point higher than the income growth (3.5 livia) saw between 3 percent and 5.5 percent growth in percent) for the total population. The top five perform- the incomes of the B40 (Figure 2.6). ers (Chile, Ecuador, Paraguay, Uruguay, and Brazil), 15 Shared Prosperity is defined over a five-year period. For each country, only the surveys that are comparable can be used for the five-year period of shared prosperity. This means that if a country has only three years of comparable data during that period, Shared Prosperity is calculated for those three years only. See Annex 2 for more information on the surveys used. 16 According to data from circa 2009-2014. World Bank 2018. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 27 A slowing in the gains in shared prosperity Figure 2.6 Shared prosperity (circa 2009–2014 and 2014–2019). Five best and five worst performers by B40 growth. a. 2009–2014 b. 2014–2019 10 10 8 8 6 6 4 4 2 2 0 0 -2 -2 -4 -4 -6 -6 CHL ECU BRA BOL COL NIC PER PAN LAC DOM ARG CRI MEX HND DOM PAN CHL BOL MEX PER HND LAC CRI COL ECU BRA ARG PRY URY SLV SLV PRY URY Growth Total Growth b40 Growth Total Growth b40 Source : LAC Equity Lab using SEDLAC (CEDLAS and World Bank). Note : Growth rates are annualized. Although the shared prosperity indicator is purely national in focus (World Bank 2018), previous editions of the Poverty and Labor Briefs have included LAC averages. For these averages, we used pooled data for the given years using extra- or interpolations. Panel a shows annualized growth rates for 2009–2014 for all countries except Chile (2009–2013), Costa Rica (2010–2014), Mexico (2010–2014), and Honduras (2009–2013). Panel b shows annualized growth rates for 2014–2019 for all countries except Mexico (2016–2018), the Dominican Republic (2017–2019), Chile (2013–2017), and LAC (2015-2019). Income growth and its redistribution both re- Despite low overall growth and corresponding- duced poverty during the ‘Golden Decade.’ Howev- ly lower shared prosperity premiums since 2012, er, during the period of stagnation, distributional income growth in most LAC countries continues to changes offset the negative changes in income. be pro-poor. Only 4 out of 16 LAC countries for which From 2002 to 2014, changes in the distribution and we have the harmonized data had a negative shared the growth of income were both important drivers of prosperity premium during 2014–2019. Most countries the reduction in poverty. Income growth accounted moved around in terms of their ranking based on B40 for 63 percent of the reduction in poverty at the $1.90 growth. Chile, for instance, had comparatively average line and for 76 percent of the reduction at the $5.50 growth (7 percent as a total) during the first period line (Figure 2.7). The period of early stagnation saw (2009–2014) but managed to maintain a relatively high a change in the usual pattern of the effects of growth growth rate in the latter years, both overall and for the and distribution: (negative) changes in growth in in- B40. El Salvador and the Dominican Republic jumped come were responsible for an increase in poverty but from the worst to the best performers in terms of over- were partially offset by distributional changes. In fact, all and pro-poor growth. Brazil became one of the for the $3.20 and $5.50 lines, distributional changes ef- worst performers in terms of overall growth, despite fectively countered the effects of the negative income being the fifth-best performer in the preceding five growth, leading to an almost stagnant poverty rate. years. Mexico continued to register one of the lowest 28 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Growth was the main driver during the sustained growth period, while distributional changes contributed to lower poverty during the period of stagnation. Figure 2.7 Distribution and growth decomposition (annual changes) 5 0 Percentage Points -5 -10 -15 -20 $1.9 $3.2 $5.5 $1.9 $3.2 $5.5 2002 - 2014 2015 - 2019 Growth Redistribution Total Source: SEDLAC (CEDLAS and World Bank). Note : See Annex 6 for the Datt-Ravallion decomposition. overall growth rates in the region: between 2009 and erty lines on the one hand and the increase in pover- 2014, both overall and B40 growth hardly registered ty based on the $1.90 on the other. The headcount of any significant change. After 2014, both overall and B40 $1.90 poor is related to the decrease in income per cap- income growth were positive, but were still small. ita in the bottom decile. Because the income per capita Given that most countries experienced pro-poor in the lowest decile decreased by 0.1 percent, poverty growth during the sustained growth period, the re- headcounts under the $1.90 line were higher in 2019. gion as a whole was also marked by higher income However, since the second and the third deciles had growth in the bottom deciles during that period growth rates around 1 percent, the overall effect on the but mixed results during the stagnation period. $5.50-based poverty rate was positive. The $3.20 head- The growth incidence curves (GIC), which plot growth count, which almost overlaps with the first decile, did rates at each quantile of per capita income, show that not increase, most likely because the largest income between the years 2008 and 2014, income growth in declines must be among the poorest households in the bottom decile (5.1 percent) was almost twice the that income decile. growth of income in the top decile (2.7 percent), and These changes also help explain the reduction 50 percent higher than the mean percentile growth in overall income inequality as mirrored by the Gini rate (3.4 percent). During the stagnation period (2015– coefficient​. During most of the ‘Golden Decade,’ the 2019), there was a general reduction in incomes at all Gini coefficient declined considerably in LAC, from 56.4 levels of the income distribution (Figure 2.8). in 2000 to 51.9 in 2011. From that point on, the reduc- These distributional changes can help explain tion rate slowed, with the Gini coefficient remaining the reduction in poverty at the $5.50 and $3.20 pov- flat through 2019 at around 51 (Figure 2.9). Mexico and THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 29 Growth was pro-poor during the sustained growth period; during the stagnation period, the poorest and richest households saw the deepest decreases. Figure 2.8 Growth incidence curves, 2008–2014 and 2015–2019 6 5 Growth Rate (Annualized) 4 Growth rate 2008–2014 2015–2019 2008 -2014 3 Mean 2.9 0.9 2 2015 -2019 Median 3.1 1.0 1 0 Mean percentile 3.4 0.9 -1 1 2 3 4 5 6 7 8 9 10 Deciles of Per Capita Household Income Source: SEDLAC (CEDLAS and World Bank). several Central American countries have driven this in- Poverty reduction was associated with increas- equality stagnation, with some countries even experi- es in earnings within sectors; nonetheless, the poor encing increases in inequality between 2014 and 2019. continue to work in low-paying sectors with high Even though the Andean subregion and Southern Cone levels of informality. In Brazil and the Andean and have contributed to the decline in inequality since Southern Cone subregions, poverty reduction during 2000, inequality reduction has slowed since 2013. the ‘Golden Decade’ among unskilled workers was Labor income has been the main driver of poverty associated with an increase in earnings in the ser- reduction in Latin America. Increases in labor income vices (construction, commerce, and hospitality) and have been invaluable in decreasing poverty during the agricultural sectors. In Mexico and Central America, sustained growth period (2009–2014). In fact, labor in- construction was associated with some reduction in come alone drove about half of the poverty reduction poverty among unskilled workers.18 As of 2019, LAC’s for the $1.90 line, two-thirds for the $3.20 line, and poor are concentrated in low-paying sectors such as about three-fourths for the $5.50.17 While labor income agriculture, commerce, and construction (Figure 2.11). contributed the most to poverty reduction throughout Moreover, 54.5 percent of workers in the LAC region are the ’Golden Decade’ for most countries, in Costa Rica, in the informal economy, particularly in the agriculture Argentina, Panama, and the Dominican Republic nonla- and services sectors (Figure 2.12). bor income was the main driver (Figure 2.10). 17 For $1.90 and $3.20 Shapley decompositions, visit the LAC Equity Lab: https://www.worldbank.org/en/topic/poverty/lac-equity-lab1. 18 World Bank (2015). 30 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 2.9 Gini coefficient by subregion (2000–2019) 0,6 0.56 0,6 0.55 0,5 0.54 Gini Coe icient 0,5 0.51 0,5 0.49 0.50 0,5 0.47 0,5 0,4 0.44 0,4 0,4 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Andean Region Central America LAC Southern Cone Source : SEDLAC (CEDLAS and World Bank). Note : (1) The Latin America and Caribbean regional aggregate is estimated based on a sample of 18 countries (LAC-18); see Annexes 2 and 3. In cases where data are unavailable for a given country in a given year, values have been interpolated or extrapolated using WDI data (World Bank 2020b) and then pooled to create the regional estimate; (2) Due to important meth- odological changes in Mexico’s official household survey in 2016 that created a break in the poverty series, we have created a break in the LAC-18 aggregate. More details are available in Annex 3; see Annex 4 for more information. Figure 2.10 Shapley decomposition of poverty by income source and country ($5.50 a day 2011 PPP) (2009–2014) 15 10 5 Percentage Points 0 -5 - 10 - 15 - 20 Labor earnings Other non - labor income Public transfers Remittances Retirement and pensions Share of individuals 15 - 64 years of age Share who are employed Total 10 Labor earnings Other non - labor income Public transfers Remittances Retirement and pensions Share of individuals 15 - 64 years of age Share who are employed Total THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 31 (2014–2019) 10 5 Percentage Points 0 -5 - 10 - 15 Labor earnings Other non - labor income Public transfers Remittances Retirement and pensions Share of individuals 15 - 64 years of age Share who are employed Total Source: SEDLAC (CEDLAS and World Bank). Note : See Annex 7 for Shapley decomposition. Panel a shows annualized growth rates for 2009–2014 for all countries except Chile (2009–2013), Costa Rica (2010–2014), Mexico (2010–2014), and Honduras (2009–2013). Panel b shows an- nualized growth rates for 2014–2019 for all countries except Mexico (2016–2018), the Dominican Republic (2017–2019), Chile (2013–2017), and LAC (2015-2019). 32 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 2.11 LAC earnings and employment by sector and number of poor ($5.50 a day 2011 PPP), (circa 2019) 1,800 1,600 Public Administration & 1,400 Defense Mean Labor Income (USD 2011PPP) Financial Intermediation 1,200 Education, Health & Personal 1,000 Electricity, Gas & Services Transportations Manufacturing 800 Construction Commerce 600 Industry Agriculture 400 Domestic Service 200 - 0 10,000,000 20,000,000 30,000,000 40,000,000 50,000,000 60,000,000 70,000,000 80,000,000 Number of Workers Source: SEDLAC (CEDLAS and World Bank). Note : Size of circles represents the number of poor in each sector. The number of workers is limited to individuals ages 15–64. Figure 2.12 Workers by sector and informality (circa 2019) Agriculture Industry Informal, 12% Informal, 11% Services Formal, 34% Services Informal, 32% Industry Formal, 9% Agriculture Formal, 2% 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 work- ers 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 ter- tiary education are considered formal workers. 33 SECTION 3. THE VULNERABILITY OF LATIN AMERICA AND THE CARIBBEAN TO THE IMPACTS OF COVID-19​ 34 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Persistent inequalities throughout the region will works in the informal sector and at low-paying jobs, likely result in the pandemic having unequal im- making them particularly vulnerable to income shocks. pacts. As was presented in the previous section, pov- Some countries in the region have mobilized signifi- erty reduction and income growth have not been ho- cant amounts of resources to strengthen their health mogenous across and within countries. Rural areas systems in order to confront the global pandemic; oth- continue to host many of the poor, and one in four indi- ers have limited capacity to provide quality affordable viduals continues to live on less than $5.50 a day (2011 health services. In addition, lockdown measures put in PPP) (Map 3.1). Moreover, a larger share of the poor place to contain the virus’s spread have highlighted in- Map 3.1 Subnational poverty rate ($5.50 2011 PPP) and COVID-19 confirmed cases Esri, HERE, Garmin, USGS Sources: Tabulation of SEDLAC (CEDLAS and World Bank) for poverty data; public health ministries for COVID-19 con- firmed cases (circa March 25, 2021). Note : Groups denote the population with income per capita lower than $5.50 a day (2011PPP). Subnational representative data at the administrative level are currently not available for Argentina, Costa Rica, Honduras, and Uruguay; therefore, the map currently shows the same poverty rate/indicator across states/departments/provinces within these countries. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 35 Figure 3.1 Informality rate (Middle class and poor, $5.50 2011 PPP), circa 2019 100 80 Percentage (%) 60 40 20 0 PRY GTM HND BOL PAN PER HTI ECU MEX COL NIC SLV LAC ARG URY BRA DOM CRI CHL Middle Class Poor ($5.50) 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 work- ers 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 ter- tiary education are considered formal workers. equities in access to basic services such as electricity, proximately four out of five workers are in the informal water, and sanitation, and even the internet. sector, whereas in Uruguay, Chile, and Costa Rica less Latin American and Caribbean countries face than a third are informal. Nearly 90 percent of workers high levels of informality 19 and self-employment, living on less than $5.50 a day (2011 PPP) are informal resulting in lower-quality and more-vulnerable relative to over a third among middle-class workers.20 jobs. Over half (54.4 percent) of the region’s workers Similarly, whereas only one of every five middle-class are in the informal sector, though there are signifi- workers is self-employed, over 30 percent are among cant variations within countries and socioeconomic the poor (Table 3.1). These numbers account for a large groups (Figure 3.1). In Guatemala and Honduras, ap- segment of the population who do not have labor con- 19 Informality refers to workers ages 15–64 who do not receive a pension. For Panama, estimates are limited to workers receiv- ing 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. 20 Although informality is linked most closely to poverty, it is still the case that one-third of middle-class workers are consid- ered informal. Hence, policies involving payrolls and unemployment insurance, for example, that could be used to mitigate shocks like COVID-19 would not reach this group. In addition, these households could be more subject to shocks, even though for the moment their incomes are high enough for them to be considered middle class. 36 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN tracts or access to unemployment insurance and rely pandemic with health systems that had recently dealt on day-to-day work that cannot be carried out from with or were facing seasonal diseases such as dengue, home. This may force many to keep working during chikungunya, and yellow fever. While this may have the confinement period to acquire basic necessities, prepared some of them, it also implied an overburden- ultimately exposing them to infection. Inequalities ing of the systems. World Bank High-Frequency Mon- between skilled (often salaried workers) and unskilled itoring Surveys (HFS) have found that nearly half of labor (commonly self-employed/informal workers) household members in Ecuador and 41.5 percent of in- throughout the LAC region will likely be heightened, dividuals in Peru needed but could not access medical because those who can work from home will not expe- treatment during quarantine. In Guatemala and Hon- rience such drastic drops in income. Even though sal- duras, approximately one in five household members aried workers may also experience income loss in the could not access medical treatment, compared with short run due to furloughs or wage cuts, informal or less than 10 percent in Costa Rica. self-employed workers do not count with benefits such Access to basic services such as water and sanita- as unemployment insurance. tion is marked by a high level of inequality. Frequent Health systems across Latin America and the Ca- handwashing has been widely promoted to reduce ribbean have limited resources for dealing with the the risk of contagion of COVID-19. Inadequate access COVID-19 crisis. On average, LAC countries invest only to water (both quantity and quality) poses additional 8 percent of GDP in health care, with some like Peru challenges to the maintaining of a clean environment spending below 5 percent. This contrasts with 10 and and the sanitizing of physical surfaces. Moreover, not 12.5 percent of GDP health expenditure globally and having an adequate water source inside the dwelling21 among OECD countries, respectively. Similarly, the LAC can limit households’ ability to socially distance and region has on average 2 doctors per 1,000 people and follow other recommended guidelines. While there has only 2.1 hospital beds per 1,000 people, well below the been an expansion in water and sanitation provision OECD average of 3.5 and 4.7, respectively. Countries in Latin American and Caribbean countries, there are like Guatemala, Honduras, and Haiti have just 1 doctor still significant gaps, particularly among socioeconom- per 1,000 people. As of early 2020, there were on av- ic groups. As of 2019, nearly 38 percent of poor house- erage just 9.1 intensive care units (ICU) beds per 1,000 holds did not have adequate sanitation relative to only people in Latin America and the Caribbean. The low- 12 percent among the middle class (Table 3.1). Similar- est ratios were observed throughout Central America, ly, nearly 1 in 10 poor households did not have access whereas Brazil and the Southern Cone countries were to improved water relative to 0.9 percent among the above the regional average. High out-of-pocket health middle class. Inadequate access to improved sanita- expenditures perpetuate inequities in access to health tion remains an issue especially across Central Amer- services and reflect a worse baseline at the onset of the ican and Caribbean countries like Guatemala and pandemic. Out-of-pocket health expenditures range Haiti, where more than 60 percent of poor households from 54 percent of total health expenditure in Guate- experience deficient sanitation levels. Costa Rica, by mala to only 15 percent in Argentina. Moreover, many contrast, performs significantly better, with indicators Latin American and Caribbean countries entered the 21 Access to water from outside sources includes public standpipes, wells, or surface water bodies. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 37 Figure 3.2 Figure 3.3 Population without improved sanitation Population without improved water (middle class and poor $5.50 2011 PPP), supply (middle class and poor $5.50 2011 circa 2019 PPP), circa 2019 80 35 70 30 60 Percentage (%) Percentage (%) 25 50 20 40 15 30 10 20 10 5 0 0 HTI GTM NIC BRA LAC BOL PER MEX COL DOM ECU HND CRI ARG HTI NIC BOL PER GTM ECU LAC MEX HND BRA COL DOM CRI ARG CHL PRY SLV URY SLV PRY URY Poor ($5.50) Middle Class Poor ($5.50) Middle Class Source : SEDLAC (CEDLAS and World Bank). closer to those of more-developed countries (Figures more than half of poor households in urban centers. 3.2 and 3.3). Conversely, Chile and Mexico have less than 2 and 4 Overcrowded living situations combined with in- percent overcrowding, respectively, in their metropoli- adequate access to water and sanitation exacerbate tan areas (Figure 3.4). the epidemic’s risks. A large part of the Latin American The COVID-19 pandemic has caused a rise in food and Caribbean population faces a higher risk of illness insecurity due to financial hardship and a lack of within their households, especially under confinement storage. Access to durable goods such as a refrigerator measures. Physical distancing can be virtually impos- may be considered invaluable under lockdown mea- sible in crowded places, increasing the risk of diseases, sures. Stay-at-home orders across the LAC region have especially in urban areas where the cost of living tends limited daily visits to supermarkets or local stores to to be higher and several family members are forced purchase food. As of 2019, over a third of poor house- to share a single room. Overcrowding may also pose holds did not have a refrigerator, compared with only 5 threats to household members’ mental health during percent among middle-class households. Under lock- lockdowns, increasing the probability of conflict within down measures, adequate food storage is particularly the household. One in three women is affected by gen- important, because households cannot replenish food der-based violence in Latin America and the Caribbean items properly. Results from a set of World Bank HFSs and in some countries from which data are available, suggest an alarming increase in food insecurity. In domestic violence has doubled or even tripled with Honduras and Ecuador, over 40 percent of households stay-at-home orders (World Bank 2020a). In El Salva- report having adults who had to skip a meal in the last dor, Guatemala, and Nicaragua, overcrowding affects 30 days because of lack of money or other resources. 38 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 3.4 Household overcrowding (middle class and poor $5.50 2011 PPP) 100 80 Percentage (%) 60 40 20 0 NIC GTM SLV HTI BOL HND ECU MEX URY ARG PER PRY LAC DOM COL BRA Middle Class Poor ($5.50) Source: SEDLAC (CEDLAS and World Bank). Note: Overcrowding is defined by the number of household members divided by the total number of bedrooms in the household. If this ratio is 3 or more, or if the household does not have a bedroom, it is considered “overcrowding”. Also, nearly half of households in Honduras and Ecua- dle-class families. Across the region, over 75 percent of dor report running out of food in the last 30 days be- the Southern Cone countries’ population uses the in- cause of a lack of money. ternet, approximately 57 percent do so in the Andean Limited access to electricity and the internet is subregion, and less than half in Central America do so yet another source of vulnerability. As countries were (Figure 3.5). Access to a mobile phone can be an alter- forced to shut down economic activity, beginning in native means by which households can connect either mid-March 2020, many businesses and schools opted for work or school. In the LAC region, there are on aver- for telework and remote learning. However, individuals age over 100 mobile cellular subscriptions per 100 peo- without access to electricity and the internet were not ple: the number of subscriptions ranges from 57 in Haiti able to adjust to these new regimes. This will likely limit to 160 in Costa Rica (Figure 3.6). Results from the World productivity in the region and, in some cases, increase Bank HFSs indicate a large share of students could not unemployment. As of 2019, 9 percent of individuals access the platforms designed for e-learning and are re- living on less than $5.50 a day (2011 PPP) did not have lying on other social networks as alternative contacts. access to electricity, compared with less than 1 percent Many countries in the region are highly depen- of the middle class (Table 3.1). Similarly, only half of the dent on remittances, making them particularly poor report using the internet, whereas over 70 percent vulnerable to a global economic slowdown. Interna- of the middle class report doing so. Internet usage at tional remittances, mainly originating from the United home is even rarer among poor households, with less States, account for 23 percent of Haiti’s GDP and 21 than a third using it, compared with 64 percent of mid- percent of the GDPs of El Salvador and Honduras (Fig- THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 39 Figure 3.5 Figure 3.6 Households with access to Mobile cellular subscriptions the internet (%), circa 2019 (per 100 people) 90 90 80 80 180 180 70 70 160 160 60 60 140 140 120 120 50 50 100 100 % % % % 40 40 80 80 30 30 60 60 20 20 40 40 10 10 20 20 0 0 0 0 Chile Costa Rica Uruguay Dominican Rep. Argentina Mexico Paraguay Brazil Chile LAC Rica Colombia Uruguay Guatemala Rep. Panama Argentina Peru Mexico Paraguay Bolivia Brazil LAC Haiti Colombia Honduras Guatemala Nicaragua Panama Peru Rica Bolivia Uruguay Haiti Chile Honduras Panama Nicaragua Colombia Argentina Peru Guatemala Costa Rica ElParaguay Uruguay LAC Bolivia Chile Brazil Panama Mexico Colombia Argentina Peru Nicaragua Dominican Rep. Guatemala Honduras Paraguay Haiti LAC Bolivia Brazil Mexico Nicaragua Dominican Rep. Honduras Haiti Ecuador El Salvador Ecuador Salvador El Salvador Salvador Ecuador Ecuador Costa Costa Dominican El Source : World Bank 2019b. Figure 3.7 Figure 3.8 Remittances as a share of GDP (2019) Remittances from abroad as a share of household income (circa 2019) Chile 0.0 Chile 0.1 Argentina 0.1 Brazil 0.2 Peru 0.1 Uruguay 0.2 Colombia 0.1 Panama 0.9 Costa Rica 0.2 Costa Rica 0.9 Peru 1.5 Ecuador 0.2 Paraguay 1.7 Paraguay 0.3 LCN 1.9 Colombia 2.1 Mexico 0.4 Ecuador 3.0 Bolivia 0.4 Mexico 3.1 Bolivia 3.3 Guatemala 0.6 Dominican Rep. 8.3 Nicaragua 0.8 Nicaragua 13.5 Dominican Rep. 1.0 Guatemala 13.9 El Salvador 20.9 Honduras 1.1 Honduras 21.5 El Salvador 2.2 Haiti 23.2 0 5 10 15 20 25 0.0 0.5 1.0 1.5 2.0 2.5 Remittances as a % of GDP Remittances as % of household income Source : World Bank 2019b. Source : SEDLAC (CEDLAS and World Bank). ure 3.7). However, remittances represent a small share large share of international remittances tends to go to of total household income, ranging from 0.1 percent in the non-poor, a sharp fall in remittances can increase Chile to 2.2 percent in El Salvador (Figure 3.8). While a the likelihood of families falling into poverty, and in 40 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN some cases, reduce investments in human capital that tances are often sent by urban informal-sector workers, are often financed by remittances. Moreover, domestic including seasonal migrants, to their families in rural ar- remittances are an important income source for rural eas, a substantial shock to the urban informal sector is households, particularly in countries with a large seg- likely to directly reduce nonlabor income in rural areas. ment of agricultural workers. Since domestic remit- Table 3.1 Profile of the poor ($5.50 2011PPP) and middle class, LAC (2019) Middle Class Poor ($5.50 a day 2011 PPP) Total Female Male Total Female Male Access to Services Electricity 99.6 99.8 99.4 91.0 92.8 89.9 Internet (at home) 63.6 65.5 62.3 29.2 40.4 21.7 Internet usage 72.9 70 74.7 50.2 55.9 47.2 Mobile Phone 93.6 92.5 94.3 83.4 86.1 81.5 Mobile Phone (individual) 87.6 87.3 87.9 52.7 52.8 52.6 No Sanitation 11.8 11.3 12.3 37.6 36.9 38.4 No Water 0.9 0.7 1 10.3 9.9 10.8 Refrigerator 94.2 96.1 92.9 66 74.1 60.6 Education Average Years of Education 9.4 9.5 9.3 5.2 5.4 5 Never attended 7.1 7.2 6.9 17.3 17.4 17.1 Incomplete Primary 19.9 19.7 20.1 38.4 36.5 40.4 Complete Primary 8 7.9 8 10.8 10.7 11 Incomplete Secondary 13.8 12.8 14.9 20.1 20.3 19.8 Complete Secondary 21.6 20.7 22.5 10.3 11.4 9.1 Incomplete Tertiary 11.6 11.8 11.4 2.2 2.5 1.8 Complete Tertiary 18.1 20 16.1 1 1.1 0.9 Sector Agriculture 6.3 3.2 8.8 40.4 28.2 48.0 Industry 26.5 13.8 36.7 21.6 12.3 27.4 Services 67.2 83.0 54.5 38.0 59.5 24.6 Type of employment Employer 5 3.6 6.1 4.9 3.5 5.9 Not salaried 1.9 2.9 1.2 11.2 16.1 7.9 Salaried worker 68.2 69.9 66.8 36.9 30.6 41.2 Self-employed 20.3 18.3 21.9 31 30.1 31.7 Unemployed 4.6 5.3 3.9 15.9 19.7 13.3 Informality Informal Workers 38.0 37.0 38.8 89.0 91.7 87.2 Source: SEDLAC (World Bank and CEDLAS). Note: Type of employment, sector, and informality limited to working individuals ages 15–64. 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. 41 SECTION 4. EXPECTED IMPACT OF COVID-19 ON POVERTY AND THE MIDDLE CLASS ​ 42 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN The growth deceleration of 2014–2019 coupled with out of poverty. More than 20 million people across the the dramatic drop in economic activity caused by region are projected to have fallen into poverty (below the COVID-19 crisis has negatively impacted living the $5.50 poverty line) in 2020, with an increase of 1.4 standards and well-being across the region. Among million more poor due to population growth. On the other emergency measures undertaken to protect other hand, emergency social transfers across the re- households, LAC governments introduced, adapted, gion in 2020 are projected to have lifted 22 million peo- or expanded their social protection programs in re- ple out of poverty, of whom more than 77 percent were sponse to the pandemic. Nevertheless, the 2020 crisis from Brazil. The combination resulted in a net decline is expected to have led to poverty increases in almost of almost 400 thousand poor in LAC. Had no mitigation all countries, with millions of people falling into pover- measures been implemented, the region may instead ty. Brazil, however, is an important exception. The 22 23 have added 28 million new poor in 2020. To understand government of Brazil implemented a generous emer- the substantial impact of Brazil, poverty rates for the gency transfer program benefitting almost 67 million LAC region excluding Brazil were also projected. These Brazilians that not only protected families from falling projections suggest that poverty in the rest of LAC has into poverty but also lifted many people out of pover- increased even with mitigation measures (resulting in ty in 2020. Poverty is therefore projected to decline 24 13.7 million more people in poverty), but less than if no sharply in Brazil in 2020. As a result, poverty in the measures had been implemented at all (see Figure 4.1, LAC region is expected to decline marginally from 22 panel b). In summary, mitigation measures, especially percent in 2019 to 21.8 percent in 2020 (see Figure 4.1, in Brazil, helped limit the negative impacts in the short panel a). Without the emergency measures taken by 25 term. However, without a fast and inclusive econom- governments across LAC, poverty could have instead ic recovery and similar levels of mitigation measures, increased to 26.5 percent in 2020. poverty may rise again in 2021.26 LAC is projected to have almost 400 thousand Even though most countries adopted emergen- less poor in 2020 than in 2019, as social transfers, cy measures to counteract the negative impact of primarily from Brazil, helped lift millions of people the COVID-19 crisis, such policies’ generosity were 22 Based on poverty projections from Diaz-Bonilla, Moreno, and Sanchez (forthcoming); see also Annex 5. World Bank fore- casts are frequently updated based on new information and changing (global) circumstances. Consequently, macro- and microprojections presented here may differ from those contained in other World Bank documents, even if basic assess- ments of countries’ prospects do not significantly differ at any given moment in time. Due to lack of reliable data of ade- quate quality, the World Bank is currently not publishing economic output, income, or growth data for Venezuela, and that country is excluded from the cross-country macroeconomic aggregate (World Bank 2021a). 23 Another exception is Chile, where poverty is projected to remain constant under the $5.50 poverty line. Chile’s social protec- tion measures are expected to have helped offset the worst effects of the crisis, maintaining poverty at prepandemic levels. In all other countries, the poverty-mitigation measures were not enough to avoid poverty increases. 24 Brazil Macro Poverty Outlook (April 2021) https://pubdocs.worldbank.org/en/114751582655277329/mpo-bra.pdf 25 Brazil’s Auxilio Emergencial (AE) was conceived as a temporary program and ended in December 2020. However, the gov- ernment of Brazil launched a new wave of AE in April 2021 with lower benefits that targeted about 44 million individuals. 26 See Annex 8 for country-specific poverty projections for 2020. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 43 Figure 4.1 LAC poverty projections for 2020 (with and without mitigation measures) a. With Brazil b. Without Brazil 40.4 45 45 39.4 38.5 40 37.3 36.5 38 40 40 34.7 37 35 31.6 30.8 35 35 28.4 26.7 26.5 30 30 21.8 24 25 25 22 % % 20 20 15 15 10 10 5 5 0 0 2019 2020 e (COVID) 2020e (COVID, 2019 2020 e (COVID) 2020e (COVID, Mitigation Measures) Mitigation Measures) Poor $ 5.50 Vulnerable ($5.5 - $13) Middle Class ($13 - $70) 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 Diaz-Bonilla, Moreno, and Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. quite low across the region.27 With the exception of substantial benefits to their entire populations (Figure Brazil, the benefit incidence as a share of pretransfer 4.3 and Annex 9). income was on average a mere 15 percent. It ranged The current global crisis is expected to result from 3 percent in Ecuador to 33 percent in Argentina in a sharp decline in the size of the middle class in (Figure 4.2).28 Likewise, countries varied in the way most countries, setting LAC back as a majority-mid- they reached their populations and in their ability to dle-class region. After decades of gradual rise, LAC’s target benefits. Colombia, Brazil, Uruguay, and Chile middle class (per capita income between $13 and $70 showcase a clear effort to ensure that support reached per day in 2011 PPP) finally became the region’s largest those who needed it most. While Costa Rica, Mexico, income class. By 2019, the middle class accounted for and Argentina provided minimal support across the 38 percent of LAC’s population, or around 230 million income distribution, Bolivia and Guatemala provided 27 Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. 28 Coverage of emergency social transfers is simulated, based on potential recipients’ eligibility criteria. In the case of Argentina, the coverage includes Emergency Family Income (Ingreso Familiar de Emergencia—IFE), additional payments to Universal Child Allowance (Asignación Universal por Hijo—AUH and Asignación Universal por Embarazo—AUE), and Tarjeta Alimentar. Projections for Argentina are based on the population covered by the household survey in the first quarter of 2020, which represents around 62 percent of the total population. Coverage for all countries may be underestimated in the simulation results given the assumptions and data restrictions. 44 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 4.2 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 12 16 8 8 10 6 5 4 5 4 3 Bolivia Guatemala Peru Panama Brazil Chile Colombia Dominican Rep. Honduras Costa Rica Argentina Ecuador Mexico El Salvador Paraguay Uruguay 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-microsimula- tion model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. Figure 4.3 Population covered by mitigation measures, by percentile (%) Brazil Costa Rica 100 100 Population (%) Population (%) 80 80 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Bolivia Mexico 100 100 Population (%) Population (%) 80 80 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles 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-microsimula- tion model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, and Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. See Annex 9 for all country-specific distributions. Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. See Annex 5 for more details. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 45 Table 4.1 Transition matrix (%) 2020 (COVID) with mitigation measures 2019 $3.20 or less $3.20-$5.50 Vulnerable Middle Class $3.20 or less 7.0% 2.1% 0.4% 0.0% $3.20-$5.50 1.1% 8.4% 3.0% 0.1% Vulnerable 0.7% 2.3% 31.8% 2.0% Middle Class 0.2% 0.2% 3.2% 34.8% Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private con- sumption 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 Diaz-Bonilla, Moreno, Sanchez (forthcoming). Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. Estimates reported with mitigation measures. people, while the vulnerable class,29 who are near poor with the vulnerable class representing once again the but not yet middle class, accounted for another 220 largest socioeconomic group (Figure 4.4). million people. However, the 2020 global pandemic Estimates suggest income growth for the bottom is expected to reduce the middle class to 37.3 percent deciles would have been the most negatively affect- of the population, for a net loss of 4.7 million people. ed by the global crisis, but mitigation measures pro- Table 4.1 shows the transition of the middle class (by vided support. Income growth is expected to be even 3.5 percentage points) into either the vulnerable class lower throughout the income distribution relative to or poverty in 2020 due to the negative impact on in- the region’s stagnation period (2015–2019). However, comes and employment. The projected net loss is less taking into account the mitigation measures adopted negative than originally expected, due primarily to the by various countries, income growth is projected to be generous transfer program implemented in Brazil.30 positive for the bottom two deciles: growth estimates While 21.6 million people in LAC are projected to lose range from 4.5 percent for the bottom decile to -4.7 for middle-class status due to the crisis, around 17 million the top decile. In the absence of emergency measures, are projected to be added to LAC’s middle class (in- all income deciles would have experienced negative cluding through population growth), thanks primarily income growth, with the bottom decile experiencing to the emergency transfers, with Brazilians making up over three times what the top decile would have expe- more than 70 percent of this gain. Without Brazil, the rienced (-20.1 percent and -6.1 percent, respectively) rest of the region will likely show a sharp decline in the (Figure 4.5). size of the middle class, with a projected net loss of 12 Because lost labor income was supplemented million people. With or without Brazil, the final result by the emergency transfers, income inequality is is a leftward shift in the region’s income distribution, projected to decline in the region in 2020. As noted 29 The vulnerable class is defined as persons whose per capita income is between $5.50 and $13 per day in 2011 PPP. 30 Table 4.1 also shows the positive transition of the poor and vulnerable class (by 2.1 percentage points) into the middle class in 2020 due to cash transfers, netting out the negative 3.5 percentage point fall out of the middle class. 46 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 4.4. Distribution of income in Latin America and the Caribbean (2019–2020) a. With Brazil b. Without Brazil 2019-2020 2019-2020 .6 .6 .4 .4 Density Density .2 .2 22% 38% 37% 25% 37% 35% Poor Vulnerable Middle Class Poor Vulnerable Middle Class 0 0 $5.5 $13 $70 $5.5 $13 $70 Per Capita Income USD 2011 PPP Per Capita Income USD 2011 PPP Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private consump- tion per capita, job losses, and remittances from the MTI and POV GPs. The current projections shown are based on a macro-micro- simulation model that assumes 12 months of unemployment. See Diaz-Bonilla, Moreno, Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. Estimates reported with mitigation measures. Figure 4.5 Projected growth incidence curves, Latin America and the Caribbean (2019–2020) 10 5 2019-2020 2019-2020 2015 -2019 Growth Rate (Annualized) Growth (with (without 2015-2019 mitigation 0 Rate mitigation measures) measures) 2019-2020 (with -5 mitigation measures) Mean  0.9 -3.6 -7.9 - 10 2019-2020 (without - 15 mitigation measures) Median 1.0 -1.8 -9.8 - 20 Mean 0.9 -10.6 -1.5 - 25 percentile 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 Diaz-Bonilla, Moreno, Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 47 earlier, Brazil’s generous emergency transfer program As discussed in Section 2, labor income has been the provided additional income to almost 67 million indi- main driver of poverty reduction in the LAC region; viduals, which contributed to a strong decline in pov- thus, this component’s negative impact is expected erty and inequality in Brazil in 2020. Although not as to affect overall household income significantly. Clo- generous, emergency transfers in other countries also sure policies and mobility restrictions had an imme- helped either to decrease inequality or minimize the diate effect on the service industry, as it forced many increase in inequality. Overall, inequality in the LAC re- businesses to close, including restaurants, shops, and gion, as measured by the Gini coefficient, is expected other tourism-related firms. As mentioned in Section to have declined from 51 to 49.8 in 2020. On the other 2, many workers in the service sector are informal, so hand, the Gini coefficient is projected to be higher in they are especially vulnerable to income fluctuations. 2020 if we exclude Brazil from the regional estimates, Moreover, not all companies were prepared to operate increasing from almost 48 in 2019 to 48.2 in 2020 even under the conditions of a pandemic and remain open with mitigation measures (Figure 4.6). for months without generating income. Thus, the larg- Household welfare will primarily be affected by est job losses in most countries are expected to be in the reduction of labor income through lockdown-in- the service sectors. COVID-19-related job losses in ser- duced job losses, particularly in the service sector. vices to date range from 2.1 percent in El Salvador to 30.5 percent in Peru (Figure 4.7). Figure 4.6 LAC Gini coefficient trends and projections for 2000–2020 59 57 55 Gini coe icent Gini (with Brazil) 53 51 51 49.8 49 Gini (without Brazil) 48.2 48 47 45 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020e 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 Diaz-Bonilla, Moreno, Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. The Gini coefficient is a mea- sure between 0 and 1; the Gini index is equal to the Gini coefficient scale between 0 and 100. 48 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Figure 4.7 Job loss by sector (2020) 20% 10% 0% % Job losses -10% -20% -30% -40% -50% Haiti Peru Panama Costa Rica Honduras Chile Domican Republic Colombia Bolivia Brazil Guatemala Argentina Nicaragua Mexico Uruguay Paraguay Ecuador El Salvador Agriculture Industry Services Source : World Bank MTI and POV GPs. Workers in the industry and agriculture sectors High levels of remittances in some countries also will also be affected, though to a lesser extent. In saw dramatic drop-offs, affecting poor, near-poor, Honduras, job losses in industry are projected to be and even middle-class households. The expected de- higher than losses in services (19.2 and 10.2 percent, crease in remittances implies a reduction in household respectively), as the country was recently struck by two nonlabor income and therefore an increase in pover- hurricanes that damaged many factories. In El Salva- ty. As covered in Section 3, remittances as a share of dor and Paraguay, job losses in the industrial sector household income range from almost zero in Uruguay will be as high as the loss of employment in services. to 2.2 percent in El Salvador, even though they can Agriculture—the sector with the most informal workers represent almost 21 percent of GDP in a country like and poor in LAC—is the least affected by the crisis. Ag- El Salvador. Similarly, in both El Salvador and Hon- riculture job losses in most countries are expected to duras, around 6 percent of households receive remit- account for between 1 and 9 percent of total job losses tances (the largest in the region, represented by the (with Chile facing an even larger decline), yet other LAC size of the bubble in Figure 4.8), but remittances can countries, in particular Peru and Paraguay, are actual- represent almost 30 percent of income for that small ly expected to increase employment in this sector (8.2 share of households who receive them. The 2020 cri- and 8.3 percent, respectively) (Figure 4.7). ses resulted in large declines in remittances in most THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 49 Figure 4.8 Projected declines in remittances across LAC countries 3.0% Average remittances over total household income 2.5% El Salvador 2.0% 1.5% Domican Republic Nicaragua Honduras 1.0% Mexico Guatemala 0.5% Ecuador Costa Rica Bolivia Peru Colombia Uruguay 0.0% Paraguay 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. countries, with some exceptions (Figure 4.8).31 Colom- experienced between March and May, but shows the bia and Ecuador faced the highest declines (20 and 19 resilience of migrants from these countries when it percent, respectively), but this likely had small effects, comes to helping their families back home. because only 1 percent of households in these coun- The COVID-19 crisis is characterized by a high tries receive remittances (although for that 1 percent, degree of uncertainty as to its impact and duration. the impact was quite negative). The negative impacts As discussed in Section 1, governments throughout on the Dominican Republic, Nicaragua, and Guatemala the LAC region have implemented various mitigation were more pronounced: remittances declined between measures to protect their most vulnerable popula- 10 and 15 percent, represent from 0.6 to 1 percent of tions. Cash transfers and unemployment insurance total household income, and are received by 2.2 to 4 are projected to offset some of the short-term negative percent of households. Interestingly, in Honduras and welfare impacts of the global economic slowdown. El Salvador, where remittances are received by the Nonetheless, the recovery in 2021 onward will depend highest percentage of households, remittances expe- on vaccine rollouts, the continuance of 2020 COVID-19 rienced positive annualized growth in 2020 of 3.8 and government policies, and the overall global econom- 4.8 percent, respectively. This growth was lower than ic recovery. The region entered the pandemic already previous years, being affected by the sharp drop-off 31 The remittance data reflects projections as of the end of 2020. More recent data is emerging that suggests remittances de- clined 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. 50 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN struggling through a period of stagnant growth and ment losses, the crisis pushed down many households poverty reduction, and will now have to face the nega- who had been near poor or middle class. In the me- tive impacts from the loss of schooling and work expe- dium term, as the region begins to grow again, those rience, as well as high levels of debt, all of which could with higher education levels are best placed to bene- slow its recovery. fit from any future recovery in jobs and opportunities However, the region’s ‘new’ poor are better suit- (Table 4.2). Households who were already poor, and ed to recover from the crisis. Individuals whose in- have now lost further human or physical capital accu- come declined to less than $5.50 a day (2011 PPP) in mulation, will have the hardest time recovering from 2020 due to the COVID-19 pandemic have on average this crisis, and inequality across multiple dimensions higher education levels and more access to basic ser- is likely to get worse. As the region recovers, the most vices relative to the ‘old’ poor. Via income and employ- excluded will need more of a focus than ever before. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 51 Table 4.2 Profile of the ‘new poor’ ($5.50 2011 PPP) and ‘lost middle class’ (2020 and 2019) 2020 2019 New Poor Lost Middle Class Poor Middle Class Access to Services Electricity 95.2 99.4 91.0 99.6 Internet (at home) 29.4 50.1 29.2 63.6 Internet usage 65.2 71.6 50.2 72.9 Mobile Phone 93.8 94.7 83.4 93.6 Mobile Phone (individual) 63.1 83.3 52.7 87.6 No Sanitation 19.9 10.2 37.6 11.8 No Water 4.1 1.3 10.3 0.9 Refrigerator 68.6 89.9 66 94.2 Education Average Years of Education 6.8 8.3 5.2 9.4 Never Attended 13.5 8.4 17.3 7.1 Incomplete Primary 27.1 21.1 38.4 19.9 Complete Primary 11.3 10.4 10.8 8 Incomplete Secondary 23.5 20.5 20.1 13.8 Complete Secondary 13.5 20.6 10.3 21.6 Incomplete Tertiary 5.1 9.8 2.2 11.6 Complete Tertiary 6.1 9.3 1 18.1 Informality Informal Workers 45.9 43.7 89.0 38.0 Sector Agriculture 4.3 7.6 40.4 6.3 Industry 26.3 30.5 21.6 26.5 Services 69.4 61.9 38.0 67.2 Type of employment Employer 5.5 3.8 4.9 5.0 Not salaried 2.3 2.1 11.2 1.9 Salaried worker 71.4 69.7 36.9 68.2 Self-employed 18.3 20.4 31 20.3 Unemployed 2.4 4.0 15.9 4.6 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 Diaz-Bonilla, Moreno, Sanchez (forthcoming) and Annex 5. Note: The LAC aggregate includes projections for Haiti based on its 2012 household survey. Estimates for 2020 are re- ported with mitigation measures. 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 with complete tertiary education are considered to in the formal sector. 52 SECTION 5. FINAL REMARKS THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 53 The pandemic has hit Latin America and the Carib- jections suggest the COVID-19 crisis could have result- bean hard. Throughout the LAC region, governments ed in more than 20 million people losing middle-class have implemented a variety of stringent policies to status. Even though mitigation measures helped limit confront the COVID-19 crisis, including restrictions on the negative impacts in the short term, without a fast public gatherings, public transport, and school and and inclusive economic recovery and similar levels of workplace closures. Nonetheless, daily reported cas- mitigation measures, poverty may rise again. es and deaths continue to rise, even as they fall across Inequalities throughout the region have been a developed economies. Steep declines in economic ac- challenge to confront the crisis. Access to basic ser- tivity are expected in the LAC region, with an estimated vices such as electricity, adequate water, sanitation, drop of 6.7 percent in GDP and 3.2 in per capita house- and even the internet has become more essential un- hold income for 2020, making it the most profound re- der lockdown measures. However, less than 1 in 4 poor cession in the region’s history. households have adequate sanitation, 9 percent do The COVID-19 crisis is expected to reverse many not have access to electricity, and only 25 percent use of the social gains that took decades to materialize the internet at home. Furthermore, 9 out of 10 work- in Latin America, particularly the shift to becoming ers living on less than $5.50 a day (2011 PPP) are in the a majority middle-class region. In 2018, for the first informal sector, and nearly a third are self-employed. time in the region’s history, millions of Latin Americans These vulnerabilities add to the difficulty of overcom- reached middle-class status, making the middle class ing the income and health shocks tied to the COVID-19 the region’s largest socioeconomic group. Poverty and pandemic. Unfortunately, those who were worst off to income inequality also declined considerably, though begin with will likely be the most affected. with differences across countries. However, the dras- Governments must continue to target policies to tic fall in economic activity caused by the 2020 global prevent contagion and support the most vulnerable pandemic will negatively impact living standards and populations, while striving to protect livelihoods. well-being across the region. Poverty projections sug- As health systems are essential to confront the crisis, gest the number of poor in LAC increased in all coun- resources should continue to be earmarked to increase tries with the exception of Brazil, where the govern- access to and improve the quality of affordable health ment’s generous emergency transfer program resulted care. Even though general lockdowns are the most ef- instead in a large poverty decline. This led to a margin- fective way to prevent mass contagion, they come at al decline in poverty in the LAC region as a whole, from the expense of increases in unemployment, the gen- 22 percent in 2019 to 21.8 percent in 2020, represent- eral loss of income, and increases in poverty. Thus, ing almost 400 thousand fewer poor. Had no mitigation well-targeted temporary income transfers provide measures been implemented, the region may have in- vulnerable groups with some income security during stead reached 28 million new poor in 2020. Alternative- containment periods. Although informality is linked ly, simply excluding Brazil would result in an increase most closely to poverty, it is still the case that one- of 13.7 million new poor in the rest of LAC. Likewise, the third of middle-class workers are considered informal. pandemic resulted in a net loss of 4.7 million people Hence, policies that could be used to mitigate shocks from the middle class. Without mitigation measures, like COVID-19 involving payrolls and unemployment in particular without Brazil’s emergency transfers, pro- insurance, for example, would not reach this group. In 54 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN addition, these households could be more subject to ress. Chile has administered more than 42 doses per shocks, even though, for the moment, their incomes 100 people (more than the United States—40 per 100— are high enough to reach middle-class status. In addi- and the United Kingdom—47 per 100). In contrast, oth- tion, social protection programs may be re-assessed to er LAC countries lag significantly, with only 1 to 6 doses incorporate new beneficiaries or adjust their reach. For administered per 100 people.32 Across the region, gov- instance, poor households often rely on school feeding ernments are having problems securing vaccines to programs for their kids; thus school closures may lead cover all their population and implementing efficient to a decline in food intake among these children. and effective systems to distribute and apply them. As lockdown measures phase out, governments Nonetheless, stay-at-home orders and social dis- should address preexisting inequities. Given the high tancing have accelerated the region’s digital trans- degree of uncertainty as to the impact and duration of formation, a silver lining from the crisis. As the econ- the COVID-19 crisis, especially if a second wave hits the omy shut down, businesses were forced to reinvent region, LAC countries must broaden access to essen- their services to continue operating. A significant boost tial services among vulnerable populations. Access to in ecommerce and eservices has been evident through- electricity and the internet have marked the difference out the region. Several supermarkets and restaurants between privileged individuals who can telework and have shifted to delivery services either through their homeschool and those who cannot continue to work online platforms or WhatsApp and Instagram. Similar- or attend school due to a lack of connectivity. More- ly, the finance sector has opted to increase their online over, adequate water and sanitation are necessary to services to minimize traffic in their physical branches. reduce the risk of infection of COVID-19. Even governments have been forced to switch to digital The recovery in 2021 onward may also depend channels to continue working. It is unlikely businesses on the vaccine rollouts. Latin American and Caribbe- will abandon these measures once the pandemic ends. an countries face important challenges in this regard Thus, countries should continue to invest in digital in- and to date only Chile has reported significant prog- frastructure to boost these changes further and enact legislation to expand the digital economy. 32 Data from Our World in Data, https://ourworldindata.org/covid-vaccinations. Accessed on 05/13/2021 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 55 Annex 1. Reported COVID-19 Cases and Government Response Index, by Country Figure A.1.1 Reported COVID-19 cases and Government Response Index Costa Rica Dominican Republic 260,000 100 300,000 100 240,000 280,000 Government Response Index Government Response Index Reported COVID -19 Cases Reported COVID -19 Cases 220,000 260,000 200,000 80 240,000 80 220,000 180,000 200,000 160,000 60 180,000 60 140,000 160,000 120,000 140,000 100,000 40 120,000 40 80,000 100,000 80,000 60,000 20 60,000 20 40,000 40,000 20,000 20,000 0 0 0 0 29 50 71 92 111 132 153 174 195 216 237 258 279 300 321 342 363 384 405 -79 -58 -37 -16 113 134 155 176 197 218 239 260 281 302 323 344 365 386 407 6 -77 -56 -35 -15 8 27 48 69 90 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Guatemala Honduras 260,000 100 260,000 100 240,000 240,000 Government Response Index Government Response Index Reported COVID -19 Cases Reported COVID -19 Cases 220,000 220,000 200,000 80 200,000 80 180,000 180,000 160,000 60 160,000 60 140,000 140,000 120,000 120,000 100,000 40 100,000 40 80,000 80,000 60,000 20 60,000 20 40,000 40,000 20,000 20,000 0 0 0 0 114 135 156 177 198 219 240 261 282 303 324 345 366 387 408 104 125 146 167 188 209 230 251 272 293 314 335 356 377 398 -76 -55 -34 -13 -86 -65 -44 -23 9 30 51 72 93 20 41 62 83 -3 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Nicaragua Panama 100,000 100 400,000 100 Government Response Index Government Response Index Reported COVID -19 Cases Reported COVID -19 Cases 350,000 80,000 80 80 300,000 60,000 60 250,000 60 200,000 40,000 40 150,000 40 100,000 20,000 20 20 50,000 0 0 0 0 103 124 145 166 187 208 229 250 271 292 313 334 355 376 397 119 140 161 182 203 224 245 266 287 308 329 350 371 392 413 -87 -66 -45 -24 -71 -50 -29 19 40 61 82 14 35 56 77 98 -4 -8 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index El Salvador Haiti 100 100,000 100 250,000 Government Response Index Government Response Index Reported COVID -19 Cases Reported COVID -19 Cases 200,000 80 80,000 80 150,000 60 60,000 60 100,000 40 40,000 40 50,000 20 20,000 20 0 0 0 0 120 141 162 183 204 225 246 267 288 309 330 351 372 393 113 134 155 176 197 218 239 260 281 302 323 344 365 386 -91 -70 -49 -28 -98 -77 -56 -35 -15 8 15 36 57 78 99 29 50 71 92 -8 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Paraguay Uruguay 350,000 100 250,000 100 Government Response Index Government Response Index Reported COVID -19 Cases Reported COVID -19 Cases 300,000 80 200,000 80 250,000 200,000 60 150,000 60 150,000 40 100,000 40 100,000 20 50,000 20 50,000 0 0 0 0 109 130 151 172 193 214 235 256 277 298 319 340 361 382 403 102 123 144 165 186 207 228 249 270 291 312 333 354 375 396 -81 -60 -39 -18 -88 -67 -46 -25 4 25 46 67 88 18 39 60 81 -5 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Argentina Bolivia 3,500,000 100 350,000 100 Govern Govern Cases Cases 3,000,000 300,000 80 80 El Salvador Haiti El Salvador Haiti 100 100,000 100 250,000 100 100,000 100 Government Government 250,000 Cases Cases Government Government Cases Cases 80 80,000 80 56 200,000 THE GRADUAL RISE AND RAPID DECLINE OF 200,000 80 80,000 80 -19 -19 CLASS IN LATIN AMERICA AND THE CARIBBEAN THE MIDDLE 150,000 -19 -19 60 60,000 60 COVID COVID 150,000 60 60,000 60 Response Response COVID COVID Response Response 100,000 40 40,000 40 100,000 40 40,000 40 Reported Reported Reported Reported 50,000 20 20,000 20 Index Index 50,000 20 20,000 20 Index Index 0 0 0 0 0 0 0 0 120 141 162 183 204 225 246 267 288 309 330 351 372 393 113 134 155 176 197 218 239 260 281 302 323 344 365 386 -91 -70 -49 -28 -98 -77 -56 -35 -15 8 8 1515 3636 5757 7878 9999 2929 5050 7171 9292 -8 -8 120 141 162 183 204 225 246 267 288 309 330 351 372 393 113 134 155 176 197 218 239 260 281 302 323 344 365 386 -91 -70 -49 -28 -98 -77 -56 -35 -15 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Paraguay Uruguay Paraguay Uruguay 350,000 100 250,000 100 350,000 100 250,000 100 Government Government Cases Cases 300,000 Government Government Cases Cases 300,000 80 200,000 80 250,000 80 200,000 80 -19 -19 250,000 -19 -19 200,000 60 150,000 60 COVID COVID 200,000 60 150,000 60 Response Response COVID COVID 150,000 Response Response 40 100,000 40 150,000 40 100,000 40 Reported Reported 100,000 Reported Reported 100,000 20 50,000 20 50,000 Index Index 20 50,000 20 50,000 Index Index 0 0 0 0 0 0 0 0 109 130 151 172 193 214 235 256 277 298 319 340 361 382 403 102 123 144 165 186 207 228 249 270 291 312 333 354 375 396 -81 -60 -39 -18 -88 -67 -46 -25 4 4 2525 4646 6767 8888 1818 3939 6060 8181 -5 -5 109 130 151 172 193 214 235 256 277 298 319 340 361 382 403 102 123 144 165 186 207 228 249 270 291 312 333 354 375 396 -81 -60 -39 -18 -88 -67 -46 -25 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Argentina Bolivia Argentina Bolivia 3,500,000 100 350,000 100 3,500,000 100 350,000 100 Government Government Cases Cases 3,000,000 300,000 Government Government Cases Cases 3,000,000 80 300,000 80 2,500,000 80 250,000 80 -19 -19 2,500,000 250,000 -19 -19 2,000,000 60 200,000 60 COVID COVID 2,000,000 60 200,000 60 Response Response COVID COVID 1,500,000 Response 150,000 Response 40 40 1,500,000 40 150,000 40 Reported Reported 1,000,000 100,000 Reported Reported 1,000,000 20 100,000 20 500,000 50,000 Index Index 20 20 500,000 50,000 Index Index 0 0 0 0 0 0 0 0 109 131 153 175 197 219 241 263 285 307 329 351 373 395 417 101 122 143 164 185 206 227 248 269 290 311 332 353 374 395 -68 -46 -24 -89 -68 -47 -26 2121 4343 6565 8787 1717 3838 5959 8080 -3 -3 -6 -6 109 131 153 175 197 219 241 263 285 307 329 351 373 395 417 101 122 143 164 185 206 227 248 269 290 311 332 353 374 395 -68 -46 -24 -89 -68 -47 -26 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Chile Colombia Chile Colombia 1,400,000 100 3.500.000 100 1,400,000 100 3.500.000 100 Government Government Cases Cases 1,200,000 3.000.000 Government Government Cases Cases 1,200,000 80 3.000.000 80 1,000,000 80 2.500.000 80 -19 -19 1,000,000 2.500.000 -19 -19 800,000 60 2.000.000 60 COVID COVID 800,000 60 2.000.000 60 Response Response COVID COVID 600,000 1.500.000 Response Response 40 40 600,000 40 1.500.000 40 Reported Reported 400,000 1.000.000 Reported Reported 400,000 20 1.000.000 20 200,000 500.000 Index Index 20 20 200,000 500.000 Index Index 0 0 0 0 0 0 0 0 117 139 161 183 205 227 249 271 293 315 337 359 381 403 117 139 161 183 205 227 249 271 293 315 337 359 381 403 -82 -60 -38 -17 -82 -60 -38 -17 7 7 7 7 2929 5151 7373 9595 2929 5151 7373 9595 117 139 161 183 205 227 249 271 293 315 337 359 381 403 117 139 161 183 205 227 249 271 293 315 337 359 381 403 -82 -60 -38 -17 -82 -60 -38 -17 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Ecuador Peru Ecuador Peru 450,000 100 2,000,000 100 450,000 100 2,000,000 100 Government Government 1,800,000 Cases Cases 400,000 Government Government 1,800,000 Cases Cases 400,000 80 1,600,000 80 350,000 350,000 80 1,600,000 1,400,000 80 -19 -19 300,000 1,400,000 -19 -19 300,000 60 1,200,000 60 250,000 COVID COVID 60 1,200,000 1,000,000 60 250,000 Response Response COVID COVID 200,000 1,000,000 Response Response 200,000 40 800,000 40 150,000 40 800,000 40 Reported Reported 150,000 600,000 Reported Reported 100,000 20 600,000 400,000 20 100,000 Index Index 50,000 20 400,000 200,000 20 Index Index 50,000 200,000 0 0 0 0 0 0 0 0 116 137 158 179 200 221 242 263 284 305 326 347 368 389 410 119 141 163 185 207 229 251 273 295 317 339 361 383 405 -74 -53 -32 -12 -80 -58 -36 -15 9 9 1111 3232 5353 7474 9595 3131 5353 7575 9797 116 137 158 179 200 221 242 263 284 305 326 347 368 389 410 119 141 163 185 207 229 251 273 295 317 339 361 383 405 -74 -53 -32 -12 -80 -58 -36 -15 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Mexico Brazil Mexico Brazil 2,500,000 100 18,000,000 100 2,500,000 100 18,000,000 100 Government Government Cases Cases 16,000,000 Government Government Cases Cases 2,000,000 80 16,000,000 80 14,000,000 2,000,000 80 14,000,000 80 -19 -19 12,000,000 -19 -19 1,500,000 60 12,000,000 60 10,000,000 COVID COVID 1,500,000 60 60 10,000,000 Response Response COVID COVID 8,000,000 Response Response 1,000,000 40 8,000,000 40 1,000,000 40 6,000,000 40 Reported Reported 6,000,000 Reported Reported 500,000 20 4,000,000 20 4,000,000 Index Index 500,000 20 2,000,000 20 Index Index 2,000,000 0 0 0 0 0 0 0 0 120 142 164 186 208 230 252 274 296 318 340 362 384 406 100 122 144 166 188 210 232 254 276 298 320 342 364 386 408 -79 -57 -35 -14 -77 -55 -33 -12 1010 3232 5454 7676 9898 1212 3434 5656 7878 120 142 164 186 208 230 252 274 296 318 340 362 384 406 100 122 144 166 188 210 232 254 276 298 320 342 364 386 408 -79 -57 -35 -14 -77 -55 -33 -12 Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Reported COVID -19 Cases Government Response Index Source: Hale et al. 2020. Note : Data as of May 13, 2021. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 57 Annex 2. Household Surveys from SEDLAC and LABLAC Harmonization Sources included in the Socio-Economic Database for Latin America and the Caribbean (SEDLAC) Country Name of survey Acronym Year (2000s) Coverage Argentina Encuesta Permanente de Hogares—2nd Semester EPHC 2003–2019 Urban Bolivia Encuesta de Hogares EH 2005–2019 National Pesquisa Nacional por Amostra de Domicilios PNAD 2001–2011 National Brazil Pesquisa Nacional por Amostra de Domicilios—Contínua PNAD-C 2012–2019 National Chile Encuesta de Caracterización Socioeconómica Nacional CASEN 2006–2017 National Colombia Gran Encuesta Integrada de Hogares GEIH 2008–2019 National Costa Rica Encuesta Nacional de Hogares ENAHO 2010–2019 National Encuesta Nacional de Fuerza de Trabajo ENFT 2005–2016 National Dominican Republic Encuesta Continua Nacional de Fuerza de Trabajo Q03 ECNFT 2017–2019 National Ecuador Encuesta de Empleo, Desempleo y Subempleo ENEMDU 2007–2019 National El Salvador Encuesta de Hogares de Propósitos Múltiples EHPM 2000–2019 National Guatemala Encuesta Nacional de Condiciones de Vida ENCOVI 2000–2014 National Haiti Enquête sur les Conditions de Vie des Ménages Après le Séisme ECVMAS 2012 National Honduras Encuesta Permanente de Hogares de Propósitos Múltiples EPHPM 2014–2019 National Encuesta Nacional de Ingresos y Gastos de los Hogares ENIGH 2000–2014 National Mexico Encuesta Nacional de Ingresos y Gastos de los Hogares – Nueva Serie ENIGH NS 2016, 2018 Nacional Nicaragua Encuesta Nacional de Hogares sobre Medición de Niveles de Vida EMNV 2005–2014 National Panama Encuesta de Hogares EH 2008–2019 National Paraguay Encuesta Permanente de Hogares EPH 2002–2019 National Peru Encuesta Nacional de Hogares ENAHO 2004–2019 National Uruguay Encuesta Continua de Hogares ECH 2006–2019 National 58 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Annex 3. SEDLAC Survey Availability and Comparability The SEDLAC (CEDLAS and the World Bank) project is a in time, generating breaks in comparability between harmonized database that seeks to mitigate differenc- series. The graph below shows the availability and es in country-specific survey design and thus formulate comparability of the SEDLAC surveys that comprise the comparable indicators between LAC countries. Howev- LAC-18. Within each country, only series marked with er, some countries have made significant methodolog- the same color may be compared over time. ical changes to their survey designs at certain points 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Country Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Republic Ecuador Guatemala Honduras Haiti Mexico Nicaragua Panama Peru Paraguay Salvador Uruguay THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 59 Annex 4. Methodological Changes in the Surveys and Projections National statistics offices sometimes introduce im- survey used to estimate income and poverty rates. provements to their household surveys. The aim could Since 2012, Brazil’s IBGE has also been conducting a be to better capture income, expand the representa- new survey called the PNAD-Continua (PNAD-C). It in- tivity of the survey, or provide other important inputs. corporates improvements in methodology and survey When the methodological changes are significant, they collection, including improved income questions and can result in breaks in the comparability of a country’s larger samples. From 2012 to 2015, IBGE conducted poverty series over time (see Annex 3). In the case of the two surveys in parallel, and beginning in 2016 the Brazil and Mexico, given the size of their populations, traditional PNAD was discontinued. The PNAD-C cur- their methodological changes can also affect the over- rently covers the period 2012–2019. The PNAD-C also all estimates of poverty and inequality at the LAC ag- replaced the PME, which provided regular information gregate level. The LAC aggregate used for poverty, in- on the labor market. The main changes between PNAD equality, and the middle class is based on 18 countries and PNAD-C include in the region for which microdata are available at the 1. Timing and representativity: PNAD was conduct- national level (i.e., “LAC-18”) in recent years. This annex ed once per year using September as a reference provides a brief overview of the recent methodological month. PNAD-C is conducted throughout the year changes undertaken in Mexico and Brazil and how the and is representative at the quarterly and monthly LAC-18 aggregate used in this report has been created. level for a subset of indicators, particularly those In particular, the important methodological changes in related to employment. Mexico’s official 2016 household survey have created a 2. Sample design: Overall, PNAD-C expanded the break in the poverty series large enough that a decision sample size to include more areas beyond major was made to also break the LAC-18 aggregate series metropolitan areas and increased coverage in rural into two series: 2000–2015 and 2015–2019 (see Figure areas, yielding a larger number of observations. 1.4). The overlap for 2015 is created so that the reader 3. Labor income and employment: The new survey can get a sense of the difference in the two LAC series. incorporates some changes in employment and The alternative would have been to maintain a compa- labor income measurement following recent ILO rable series for 2000–2019 using projections for Mexico recommendations. The differences from PNAD in- based on its 2014 household survey. However, the im- clude (1) only individuals 14 or older are included portance of projecting the poverty impact of COVID-19 in the labor questions (the age of inclusion in the in the region requires using the latest available and PNAD was 10); (2) production for own consump- most accurate microdata for the region, which was the tion is no longer considered employment; (3) la- updated poverty series for LAC-18 for 2015–2019. bor income is excluded if an individual worked less Brazil than 1 hour during the reference week, excluding temporary leave from work (as they are no longer The Instituto Brasileiro de Geografia e Estadística considered employed); (4) for employers and the (IBGE) has traditionally administered the Pesquisa Na- self-employed, the PNAD-C asks about the “retira- cional por Amostra de Domicílios (PNAD), a household da,” which is the take-home income from the busi- 60 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN ness, while the PNAD had the same phrasing for The revised methodology that was applied in the wage workers and employers/self-employed. 2016 and 2018 ENIGHs is an improvement to the house- For more information, see Notas Metodologia PNAD hold surveys that aims to better capture income. How- e PNAD-C (IBGE 2014). ever, the values for income produced since 2016 are not directly comparable with the historical series (2014 Another important issue that arose was related to and earlier), and in fact result in quite a large difference household implicit rent. The 2012–2015 PNAD-C did not in Mexico’s poverty series. The relatively large impor- include the variables necessary for rent imputation. tance of Mexico in the LAC aggregate, combined with While the PNAD-C is an improvement over the PNAD in the large differences in poverty between the series, terms of survey methodology, the 2012–2015 surveys led the team to decide to break the LAC series to indi- did not collect data on dwelling characteristics, home cate that results are not comparable until a correction ownership status, or housing rent amount. These are method is implemented. the variables that are used for the SEDLAC rent impu- tation model used to increase comparability in LAC’s We implemented different robustness exercises to household surveys. Therefore, in order to harmonize evaluate the sensitivity of the regional aggregate to the data, the LAC Stats team and the World Bank Bra- each decision. After careful consideration, the team zil Poverty team developed and tested a model that decided to use actual data rather than extrapolations imputes expected rent throughout the income distri- for Mexico for 2016 and 2018, given the less realistic bution. This was done separately for the PNAD-C for picture of poverty in Mexico with the extrapolations. 2012–2015. The result was a more comparable harmo- This is in contrast with the situation in Brazil, where nized PNAD-C series for 2012–2019. the new series (2012–2019) is not so different from the 2011 poverty rate as to have warranted a break in the Mexico LAC series in that year. In Mexico, official poverty estimates are produced Haiti and released to the public every two years by CONE- VAL (Consejo Nacional de Evaluación de la Política de Haiti has traditionally been excluded from the LAC re- Desarrollo Social) based on data from the household gional aggregate due to the lack of up-to-date micro- survey MCS-ENIGH (Módulo de Condiciones Socioeco- data for poverty measurement. Currently, the latest nómicas Encuesta Nacional de Ingreso y Gasto de los available living conditions survey is the 2012 Enquête Hogares), which is generated by INEGI (Instituto Nacio- sur les Conditions de Vie des Ménages après le Séisme nal de Estadística y Geografía). The main differences of (ECVMAS—the Post-Earthquake Household Living Con- the 2016 survey compared to previous years are: ditions Survey). However, Haiti is an important country due to the size of its poor population and the fact that 1. Bigger Sample: a larger sample size to allow urban/ it is one of the few Caribbean countries with available rural disaggregation at the state level; harmonized income data. The LAC-18 aggregate there- 2. Expansion of the survey: both income and expendi- fore includes the 2012 household survey, projected tures were collected for all households; and to 2019 using a distributionally neutral projection of 3. Field operation improvements: in particular, better household per-capita income. Including Haiti thus pro- training for interviewers and an automated system vides a more precise number of poor people in LAC, be- to monitor data collection. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 61 cause Haiti is the poorest country in the region and has not available. Table A.4.1 shows where microdata have 10 million inhabitants. been interpolated or extrapolated across the 18 coun- tries in LAC for the 2000–2019 period. Unavailable or nonexistent microdata The final LAC-18 series is split into two. The first se- Survey microdata to measure poverty are not available ries covers the period 2000–2015 and since there are across all the years for all countries. Therefore, any LAC no microdata for Mexico in 2015 it is estimated as an aggregate that uses only available microdata will have extrapolation of the 2014 ENIGH. The second LAC-18 a compositional problem that will affect the compara- series is for the 2015–2019 period, where data for Mexi- bility of survey estimates. This problem arises when co in 2015 is instead extrapolated backwards using the changes in LAC poverty numbers do not result from new 2016 ENIGH. For all other countries the actual mi- changes in the welfare of any country in the region, but crodata are used, except for those cases where no mi- from the fact that the set of countries with data avail- crodata exist and thus interpolations or extrapolations able is not the same between one year and another. To using existing microdata must be applied as shown in circumvent that problem, we interpolate and extrap- Table A.4.1. olate country-specific estimates when microdata are Table A.4.1. Microdata inputs for the LAC aggregate Country 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 Arg 04 14&16 Bol 02&05 09&11 Bra 01 09&11 Chl 2006 06&09 09&11 11&13 13&15 15&17 17 Col 01 05&08 Cri Dom Ecu 03 Gtm 00&06 06&14 14 Hnd 01 Hti 2012 2012 Mex 00&02 02&04 06&08 08&10 10&12 12&14 16 18* Nic 01 01&05 05&09 09&14 14 Pan Per Pry 01 Slv Ury Note: White areas are country-year pairs with available survey microdata that was included in the regional aggregate. Gray areas represent country-year pairs for which we created microdata using an interpolation method of the years indicated in the cell. Blue cells are country-pair years where we created microdata by applying an extrapolation of the household income dis- tribution using a neutral distribution algorithm that applies the real growth rate in per-capita consumption. * The algorithm to produce Mexico’s 2019 extrapolation is nondistributionally neutral and implemented by the World Bank’s Poverty Group Practice Poverty Economist for the country. 62 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Annex 5. The Macro-Microsimulation Model To be able to model both the poverty and distributional 5.1 MACRO-MICROSIMULATION impacts of the 2020 global crisis caused by the COVID-19 MODEL: INPUTS pandemic, the team chose to implement a macro-mi- This model requires five main inputs. Country-specif- crosimulation model. To capture the impact of such a ic harmonized household-survey microdata for each large negative shock on inequality, poverty, and on the country in the Latin America and Caribbean region are size of the middle class required going beyond neutral based on the SEDLAC database. The model is applied distribution methodologies. This methodology ap- to the 18 countries in the Latin America and Caribbe- plies country-specific macroeconomic projections to an region with recently available household surveys. country-specific behavioral models built on household Projected annual growth rates in private consumption survey microdata that were harmonized as part of the per capita from national expenditure accounts are pro- Socio-Economic Database for Latin America and the vided by the World Bank Macro, Trade, and Institutions Caribbean (SEDLAC) project. Although many inputs are Global Practice (MTI GP) for each country. Where pos- country specific, the model applies the same method- sible, the model uses sectoral job losses and job hires ology across all countries in order to estimate poverty provided by the World Bank Poverty and Equity Global and distributional impacts in a consistent manner for Practice (POV GP), based on country estimates under- LAC as a whole. In addition, this class of models aims taken by the corresponding Poverty Economist. When to maintain consistency between the macro- and mi- these data are not available, sectoral GDP growth pro- cro-projections and therefore tends to focus on annual jections in agriculture, industry, and services provided impacts rather than short-run impacts. The microsim- by the MTI GP are used to estimate projected sectoral ulations are based on a household income generation job losses using a GDP to employment elasticity. Pro- model (Bourguignon and Ferreira 2005). For the macro jected changes in remittances by country are also pro- side, rather than use a computable general equilibrium vided by the MTI GP. The final input is a set of popula- (CGE) model for each country per the macro-microsim- tion projections for 2019 and 2020 based on the World ulation models of Bourguignon, Bussolo, and Pereira Development Indicators projections database. When da Silva (2008) and Ferreira et al. (2008), the macro projections incorporate mitigation measures, this ad- model here uses a variety of macroeconomic projec- ditional input is applied, based on information provid- tions as determined by the World Bank POV GP’s Pov- ed by the POV GP Poverty Economists. erty Economist for each country. Some countries apply a CGE model, others use simpler sectoral macroeco- 5.2 MACRO-MICROSIMULATION nomic projections to estimate job losses, while others MODEL: METHODOLOGY are able to apply actual job-loss data from household This approach takes the 2019 microdata for the 18 SED- surveys conducted in the field. This annex presents the LAC countries mentioned above as a starting point. The macro-microsimulation model. methodology then incorporates three main channels THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 63 of transmission of the 2020 shock: through job loss- job gains, a new labor income is estimated via a tradi- es, labor income changes, and nonlabor (remittance) tional Mincer equation, in which the logarithmic of in- income changes. For the simulations that incorporate come is regressed on sex, age, education level, urban/ mitigation measures, a fourth channel of transmission rural, dependency rate, and a dummy if the household is applied that incorporates cash transfers to eligible has income from remittances. Workers who are em- households. ployed in public administration, utilities, health, or Job losses/gains per sector provided by the POV GP extraterritorial agencies are protected from job losses or estimated using sectoral GDP growth projections or income changes. For the remaining workers (not in and a sectoral GDP to employment elasticity are im- protected sectors and who have not lost their jobs), posed on the household survey data. A probit model, their labor incomes are adjusted up or down by the applied to each of the three sectors and by formal and overall change in private consumption per capita of informal workers, provides a probability of employ- their specific country. ment in each sector for each person in each household Projected changes in remittances by country are survey, based on a set of characteristics (sex, age, edu- also included in the microsimulations. The percentage cation level, urban/rural, dependency rate, dummy for change in remittances at the national level is applied member working in the public sector, and a dummy for as a direct percentage change in the remittance in- remittances). Workers are ranked by the probability of come of any household who receives this income in the employment, and those with the lowest employment household survey. probability are simulated to lose their jobs until total Lastly, coverage of emergency social transfers is job loss matches the macroprojections for job loss by simulated based on potential recipients’ eligibility cri- sector. For job gains, new workers are chosen among teria, as determined by each country. Estimates are the unemployed according to the probability of being limited to cash-transfer mitigation measures that were employed in that sector until total job gains match the measurable in household surveys. These are provid- macroprojections by sector. ed by the World Bank POV GP’s Poverty Economists. For job losses, the chosen worker (by probability of In-kind transfers were not included. Coverage may be employment based on the probit model) is simulated underestimated in the simulation results, given the as- to lose 100% of his or her labor income. In the case of sumptions and data restrictions. 64 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Annex 6. Growth and Distribution Decomposition The decomposition of poverty changes on growth and curve while keeping the mean income constant at the income distribution was proposed by Datt and Raval- reference level μr ” ( 277). lion (1992). The idea is to separate the effects on pov- The residual component R ( r = 1) exists because erty according to the changes that occurred in income the poverty index is not additively separable between between two periods: (1) the income growth compo- mean income growth and income distribution. In other nent is the change in poverty due to a change in the words, the mean income growth is endogenous to the mean income in the absence of changes in income dis- Lorenz curve. A way to eliminate the residual compo- tribution, and (2) the distribution component is the nent of the poverty change is to decompose the change change in poverty due to changes in the Lorenz curve in poverty by changing the point of reference and av- while keeping the mean income constant. eraging its components.33 In this case, the second de- Mathematically, let Pt (μt, Lt) be the poverty rate in composition will be anchored to period 2 ( r = 2): time t = {1,2} that depends on the mean income μt and P2 - P1 = ΔP = [ P ( μ2, L2 ) - P ( μ1, L2 ) ] the Lorenz curve Lt . By taking t = 1 as the period of ref- + [ P ( μ2, L2 ) - P ( μ2, L1 )] + R (r = 2) erence, the decomposition of the change in the poverty Thus, by construction rate from period 2 to period 1 in its growth and redistri- R ( r = 1) = [ P ( μ2, L1 ) - P ( μ1, L1 ) ] bution components is - [ P ( μ2, L2 ) - P ( μ1, L2 ) ] P2 - P1 = ΔP = [ P ( μ2, L1 ) - P ( μ1, L1 ) ] = [ P ( μ1, L2 ) - P ( μ1, L1 ) ] + [ P ( μ1, L2 ) - P ( μ1, L1 )] + R (r = 1) - [ P ( μ2, L2 ) - P ( μ2, L1 ) ] According to Datt and Ravallion (1992), “The growth component [ P ( μ2, L1 ) - P ( μ1, L1 ) ] of a = - R ( r = 2) change in the poverty rate is defined as the change in and the whole equation becomes: poverty due to a change in the mean while holding the [ P ( μ2, L2 ) - P ( μ1, L2 ) ] + [ P ( μ2, L1 ) - P ( μ1, L1 )] P 2 - P 1 = ΔP = Lorenz curve constant at some reference level Lr . The 2 redistribution component [ P ( μ1, L2 ) - P ( μ1, L1 ) ] [ P ( μ2, L2 ) - P ( μ1, L2 ) ] + [ P ( μ2, L1 ) - P ( μ1, L1 )] + is the change in poverty due to a change in the Lorenz 2 33 This is known as the Shapley value of components, which is used to correct for path dependency. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 65 Annex 7. Shapley Decomposition by Components of a Welfare Measure The Shapley decomposition by components of a wel- Therefore, income per capita can be written as fare measure was developed by Azevedo, Nguyen, and nA noF yiL Ypc = 1 n Ʃin=1 yi = n nA Ʃinϵ F F o no F Sanfelice (2012) based on Barros et al. (2006). This noM yiL methodology estimates the relative effect of changes + nA Ʃinϵ M M o no M +1 n Ʃi gi n (1) in different income sources (i.e., labor income, nonla- Per-capita household income is a function δ of each bor income, and transfers) on poverty and inequality of the j components, in this case j = 4. changes during a specific period. nA no yil yinl Ypc = δ n ,n , no , n (2) Mathematically, in order to decompose the chang- A es in poverty and inequality by each of the income Note that any poverty or inequality measure I is a components, the per-capita income must be expressed function θ that depends on the income distribution as a function of its components. Barros et al. (2006) de- across households. Defining F (Ypc) as the cumulative fine income per capita as the sum of each individual’s distribution function of per-capita income and replac- income divided by the number of household members ing equation (2), it concluded that any poverty or in- n. The individual’s income yil can be split into labor in- equality measure is a function of the income compo- come yinl and nonlabor income. Nonlabor income in- nents: cludes per-capita levels of pensions, capital, transfers nA no yil yinl I= θ F δ n ,n , no , n (3) A (both public and private), and imputed rent, among Based on equation (3), the change in the indicator is other factors. expressed over a period t and t-1 as a result of changes Departing from the methodology used in previous in the value of its components ∆ It,t - 1 = It – It - 1 . Using PLBs, here we use this decomposition to adjust for de- the method of Barros et al., the distribution of income mographic transition while separating labor income is simulated by changing each of these i components, by returns and employment level among the men and one at time, to calculate their contribution to the ob- women of the household. Specifically, labor income served changes in poverty or inequality. is only earned by members of the household who are Using information from all components in each pe- employed, n0. These employed n0 are also a function riod, the participation of component j is estimated in of the number of members in the household who are the change of the analyzed indicator between t-1 and of working age nA . Based on these two conditions, the t. This can be done by constructing a counterfactual per-capita labor income of the household 1 n Ʃi =1 yi can n l distribution for period t by substituting each income be split into three components: per-worker labor in- yil component’s observed level in (t-1) for its value in (t). come ( Ʃinϵ o o no ) , the employment rate of the household no nA A counterfactual indicator for period t is then comput- ( nA ) , and working age rate ( n ) . In addition, labor in- ed, based on the previous counterfactual distribution. come can be split into female and male labor income: yil yil The difference between the counterfactual and the ob- ( Ʃ no F n F ) and ( Ʃ no M n M ) . Note that other income gi iϵF iϵM o o served value of the analyzed indicator is the effect of includes nonlabor income and labor income of non- component j on the indicator’s change. working-age adults. 66 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN In the absence of panel data, Azevedo, Nguyen, The sum of the marginal effects of each compo- and Sanfelice (2012) use the rank preservation princi- nent, however, does not give us the total change from ple to transpose the distribution from one period to (t-1) to (t), because the decomposition suffers from another. This means that the distribution in each pe- path dependence: the order in which each component riod is ranked using per-capita income. Thus, the first is changed matters. Azevedo, Nguyen, and Sanfelice observation period (t-1) will be linked with the first (2012) solve this problem using the Shapley value, observation in the period (t). The difference between which computes all possible j! ways to decompose the the observed indicator value and the counterfactual indicator. Then, the weighted average of these j effects indicator is the effect σ no of the occupied rate on the is computed, which is the total effect of component j nA change of the analyzed indicator. This is described in on the indicator’s observed change. the next equation (where the hat represents the coun- terfactual indicator in period t): nA no yil yinl σ no nA = Î – I = θ F δ n ,n , A no , n nA no yl yinl – θ F δ n ,n , i no , n (4) A THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 67 Annex 8. Poverty and Middle Class Estimates, with and without Mitigation Measures, by Country (2020) Poverty Estimates ($5.50 per day, 2011 PPP) 2020 2020 Country 2019 Without mitigation measures With mitigation 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 trans- fers were not included. 68 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Middle Class Estimates ($13 - $70 per day, 2011 PPP) 2020 2020 Country 2019 Without mitigation measures With mitigation 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 Republic 42.4 34.1-38.9 38.8-42.9 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 micro- data. (b) Estimates are limited to cash-transfer mitigation measures that were measurable in household surveys. In-kind transfers were not included. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 69 Annex 9. Population Covered by Mitigation Measures, by Percentile Brazil Chile 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Colombia Uruguay 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Bolivia Panama 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Paraguay Peru 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles 70 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Mexico Ecuador 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Argentina Honduras 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Dominican Republic Costa Rica 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 20 40 60 80 100 0 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Guatemala El Salvador 100 100 80 80 Population (%) Population (%) 60 60 40 40 20 20 0 0 20 40 60 80 100 0 0 20 40 60 80 100 Post-COVID Income Percentiles Post-COVID Income Percentiles Source: Projections based on 2019 SEDLAC (CEDLAS-World Bank) microdata and macroeconomic projections of private con- sumption 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. THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN 71 References Azevedo, Joao Pedro, Minh C. Nguyen, and Viviane Diaz-Bonilla, Carolina, Laura Moreno Herrera, and Di- Sanfelice. 2012. “ADECOMP: Stata module to es- ana Sanchez Castro. Forthcoming. “Projected timate Shapley Decomposition by Components 2020 Poverty Impacts of the COVID-19 Global Cri- of a Welfare Measure,” Statistical Software Com- sis in Latin America and the Caribbean.” Poverty ponents S457562, Boston College Department of and Inequality Monitoring Brief (PIM). Washing- Economics, revised 12 Jan 2019. ton, DC: World Bank. Barros, Ricardo Paes de., Mirela Carvalho, Samuel Fran- Economic Commission for Latin America and the Ca- co, and Rosane Mendoça (2006). “Uma Análise das ribbean (ECLAC). 2020. “Report on the Economic Principais Causas da Queda Recente na Desigual- Impact of Coronavirus Disease (COVID-19) on Lat- dade de Renda Brasileira.” In: Revista Econômica. in America and the Caribbean: Study Prepared by Volume 8, número 1, p.117-147. Univers deral Flu- the Economic Commission for Latin America and minense. Rio de Janeiro. Available in http://www. the Caribbean (ECLAC), at the request of the Gov- uff.br/revistaeconomica/V8N1/RICARDO.PDF ernment of Mexico in its Capacity as Pro Tempore Bourguignon, François, and Francisco Ferreira. 2005. Chair of the Community of Latin American and Ca- “Decomposing Changes in the Distribution of ribbean States (CELAC), at the Virtual Ministerial Household Incomes: Methodological Aspects” in Meeting on Health Matters for Response and Fol- The microeconomics of income distribution dy- low-up to the COVID-19 Pandemic in Latin Ameri- namics in East Asia and Latin America, eds. Bour- ca and the Caribbean, Held on 26 March 2020, (LC/ guignon, F., Ferreira, F., and Lustig, N. Washington, TS.2020/45),” Santiago. DC: Oxford University Press and The World Bank. Ferreira, Francisco H. G., Julian Messina, Jamele Rigo- Bourguignon, François, Maurizio Bussolo, and Luiz lini, Luis-Felipe López-Calva, Maria Ana Lugo, and Pereira da Silva. 2008. “The impact of macroeco- Renos Vakis. 2012. Economic Mobility and the Rise nomic policies on poverty and income distribu- of the Latin American Middle Class. Washington, tion: macro-micro evaluation techniques and DC: The World Bank. tools” in The Impact of Macroeconomic Policies Gentilini, Ugo, Mohamed Almenfi, Ian Orton, and Pa- on poverty and Income Distribution, eds. Bour- mela Dale. 2020. “Social Protection and Job Re- guinon, F., Bussolo, M., and Pereira da Silva, L. The sponses to COVID-19: A Real-Time Review of Coun- World Bank, Washington, DC. try Measures (Living Paper). Version 12 (July 10, Datt, Gaurav and Martin Ravallion. 1992. “Growth and 2020).” Washington, DC, World Bank. redistribution components of changes in poverty Hale, Thomas, Sam Webster, Anna Petherick, Toby measures: A decomposition with applications to Phillips, and Beatriz Kira. (2020). Oxford COVID-19 Brazil and India in the 1980s.” Journal of Develop- Government Response Tracker. Blavatnik School ment Economics, Volume 38, Issue 2, April, Pages of Government. Available at: www.bsg.ox.ac.uk/ 275-295. covidtracker. 72 THE GRADUAL RISE AND RAPID DECLINE OF THE MIDDLE CLASS IN LATIN AMERICA AND THE CARIBBEAN Instituto Brasileiro de Geografia e Estadística (IBGE). 2014. World Bank. 2018. Poverty and Shared Prosperity 2018: Pesquisa Nacional Por Amostra de Domicílios Con- Piecing Together the Poverty Puzzle. Washington, tínua Notas Metodológicas. Vol. 1. https://www.ibge. DC: World Bank. https://openknowledge.world- gov.br/estatisticas/sociais/trabalho/9173-pesqui- bank.org/handle/10986/30418. sa-nacional-por-amostra-de-domicilios-conti- World Bank. 2019a. “Global Economic Prospects: nua-trimestral.html?edicao=23841&t=downloads. Darkening Skies.” World Bank, January 8. Instituto Brasileiro de Geografia e Estadística (IBGE). http://documents.worldbank.org /curated/ 2018. Síntese de Indicadores Sociais : Uma Análise en/307751546982400534/Global-Economic-Pros- Das Condições de Vida Da População Brasileira : 2018. pects-Darkening-Skies. https://www.ibge.gov.br/estatisticas/sociais/popu- World Bank. 2019b. World Development Indicators 2019. lacao/9221-sintese-de-indicadores-sociais.html. Washington, DC: World Bank. Instituto Nacional De Estadística y Censos (INEGI). 2016. World Bank. 2020a. “Gender Dimensions of the “Mercado de Trabajo: Principales Indicadores. Se- COVID-19 Pandemic (English).” Washington, DC: gundo Trimestre de 2016 Consideraciones Sobre World Bank Group. http://documents.worldbank. La Revisión, Evaluación y Recuperación de La En- org/curated/en/618731587147227244/Gender-Di- cuesta Permanente de Hogares (EPH).” mensions-of-the-COVID-19-Pandemic. OECD/The World Bank. 2020. Health at a Glance: Latin World Bank. 2020b. World Development Indicators 2020. America and the Caribbean 2020. Paris: OECD Pub- Washington, DC: World Bank. lishing. https://doi.org/10.1787/6089164f-en. World Bank. 2021a. “Global Economic Prospects, Woetzel, Jonathan, Anu Madgavkar, Kweilin Ellingrud, January 2021.” Washington, DC: World Bank. Eric Labaye, Sandrine Devillard, Eric Kutcher, https://openknowledge.worldbank.org /han- James Manyika, Richard Dobbs, and Mekala dle/10986/34710. Krishnan. 2017. How Advancing Women’s Equality World Bank. 2021b. Renewing with Growth. LAC Semi- Can Add $12 Trillion to Global Growth. McKinsey & annual Report; April 2021. Washington, DC: World Company. Bank. https://openknowledge.worldbank.org/ World Bank. 2015. “Working to End Poverty in Latin handle/10986/35329. America and the Caribbean—Workers, Jobs, and Wages: LAC Poverty and Labor Brief, June 2015.” Washington, DC, World Bank.