Employment in Crisis the Path to Better Jobs in a Post-COVID-19 Latin America

WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES

Employment in Crisis The Path to Better Jobs in a Post-COVID-19 Latin America

Joana Silva, Liliana D. Sousa, Truman G. Packard, and Raymond Robertson

Employment in Crisis

WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES

Employment in Crisis The Path to Better Jobs in a Post-COVID-19 Latin America

Joana Silva, Liliana D. Sousa, Truman G. Packard, and Raymond Robertson © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org

Some rights reserved 1 2 3 4 24 23 22 21

This work is a product of the staff of The World Bank with external contributions. The findings, inter- pretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be 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

This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions:

Attribution—Please cite the work as follows: Silva, Joana, Liliana D. Sousa, Truman G. Packard, and Raymond Robertson. 2021. Employment in Crisis: The Path to Better Jobs in a Post-COVID-19 Latin America. World Bank Latin American and Caribbean Studies. Washington, DC: World Bank. doi:10.1596/978-1-4648-1672-7. License: Creative Commons Attribution CC BY 3.0 IGO. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third- party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images.

All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: [email protected].

ISBN (paper): 978-1-4648-1672-7 ISBN (electronic): 978-1-4648-1691-8 DOI: 10.1596/978-1-4648-1672-7

Library of Congress Control Number: 2021938884 Contents

Foreword ix About the Authors xi Acknowledgments xiii Executive Summary xv Abbreviations xxi

1 Overview...... 1 Rationale for this report...... 1 Road map ...... 4 Key insights ...... 5 Three dimensions of the policy response...... 10 Implications for the COVID-19 crisis ...... 16 Notes...... 17 References ...... 18 Annex 1A: Background papers written for this report...... 20

2 The Dynamics of Labor Market Adjustment...... 23 Introduction...... 23 Labor market flows: Unemployment versus informality...... 26 Job destruction and job creation in times of crisis...... 33 A changing employment structure and the disappearance of good jobs...... 41 Conclusion...... 44 Notes...... 45 References...... 46 Annex 2A: Employment transitions by country, gender, and skill level...... 48

v vi Contents

3 The impact on workers, firms, and places...... 51 Introduction...... 51 Workers: A bigger toll on the unskilled...... 53 Firms: The cost of limited market competition...... 63 Places: The role of local opportunities and informality ...... 68 Conclusion ...... 70 Notes...... 71 References...... 72

4 Toward an integrated policy response...... 77 Introduction...... 77 Three key policy dimensions...... 80 Aggregate: Stronger macroeconomic stabilizers ...... 82 Social protection and labor systems: Cushioning the impact on workers and preparing for change...... 87 Structural: Greater competition and place-based policies...... 109 Conclusion...... 123 Notes...... 125 References...... 127

Boxes 4.1 Family allowances as de facto unemployment insurance...... 99 4.2 Brazil’s social protection response to the COVID-19 (coronavirus) pandemic ...... 101 4.3 Latin America and the Caribbean’s social protection and labor responses to the COVID-19 (coronavirus) contraction of 2020...... 104 4.4 Permanent, systemic shocks: Responses to job dislocation caused by structural changes...... 108 4.5 How well have regional policies performed at strengthening economic opportunities? ...... 120 4.6 Evidence on the effects of place-based policies on mobility and labor market outcomes...... 121

Figures 1.1 Persistent employment loss following crises: The myth of economic recovery...... 2 1.2 How adjustment works and the policies that can smooth it...... 10 1.3 Stabilizers and macroeconomic frameworks: Policy reforms ...... 12 1.4 Addressing crises’ impacts and preparing workers for change: Policy reforms ...... 15 1.5 Tackling structural issues that worsen the impacts of crises on workers...... 16 2.1 Quarterly fluctuations in unemployment and GDP growth, 2005–17...... 27 2.2 Quarterly net flows into formal and informal employment, 2005–17...... 30 2.3 Part-time work as a margin of adjustment in Argentina, 2005–15...... 33 2.4 Quarterly job loss, formal and informal sectors, 2005–17...... 35 2.5 Quarterly net job finding rates, formal and informal sectors, 2005–17...... 37 2.6 Gross job flows in Brazil and Ecuador, formal sector...... 38 2.7 Gross job flows and differential rates in large and small firms in the formal sector...... 39 Contents vii

2.8 Net job creation rates in Brazil and Ecuador’s formal sectors...... 40 2.9 Quarterly share of workers entering unemployment per wage decile, formal and informal sectors, 2005–17...... 42 2.10 Impulse response functions, by type of employment, during the 30 months after the beginning of the recession...... 43 2.11 Estimates of Okun’s Law for countries in the LAC region, 1991–2018...... 45 2A.1 Quarterly net flows into part-time work, formal and informal sectors, 2005–17...... 48 3.1 Effect on wages of displacement caused by plant closings in Mexico...... 55 3.2 Unemployment rates by cohort, Argentina and Colombia...... 56 3.3 Employment and wage effects of higher local unemployment at labor market entry in Mexico...... 57 3.4 Dynamic effects of the global financial crisis on workers...... 59 3.5 Heterogeneity in effects of the global financial crisis across workers...... 61 3.6 Dynamic effects of the global financial crisis on workers by skill...... 62 3.7 Dynamic effects of the global financial crisis on firms...... 65 3.8 Effects of the global financial crisis on Brazilian workers depending on sectoral concentration and state ownership...... 67 3.9 Effects of the global financial crisis on Brazilian workers depending on local labor market informality...... 70 4.1 How adjustment works and a triple entry of policies to smooth it...... 81 4.2 Wage and unemployment responses during crises in the 2000s versus crises in the 1990s, Brazil and Mexico...... 84 4.3 Sensitivity of unemployment and wages to output fluctuations ...... 85 4.4 Stabilizers and macroeconomic frameworks: Policy reforms ...... 87 4.5 Effective coverage of unemployment benefits, selected countries, latest available year...... 90 4.6 Economic cycle, unemployment, and spending on labor policies and programs...... 92 4.7 Level and composition of government spending on social assistance transfer programs, selected LAC countries...... 97 4.8 Insufficient support, with many left behind ...... 98 4.9 Coverage of social registries and support received through social assistance programs during the COVID-19 pandemic...... 100 4.10 Expansion of cash transfer programs in response to crises...... 100 B4.2.1 Brazil’s COVID-19 social protection strategy for two major vulnerable groups ...... 101 4.11 Positive effects of welfare transfers on local formal employment...... 102 B4.3.1 Stylized social protection and labor policy responses to the COVID-19 pandemic...... 104 4.12 Employment and reemployment policies, by the nature of the shock causing displacement...... 106 4.13 Addressing crises’ impacts and preparing for change: Policy reforms ...... 109 4.14 Employment protection legislation in OECD member countries and selected Latin American countries, 2014 or most recent data...... 111 4.15 Regulation of employment in the LAC countries, circa 2019...... 112 4.16 Flexibility of labor regulation and spending on human capital and labor programs in selected countries in LAC compared to other regions...... 114 4.17 Labor market regulation instruments and the duration of unemployment...... 115 4.18 Tackling structural issues that worsen the impacts of crises on workers...... 122 viii Contents

Map 4.1 Unemployment insurance throughout the world...... 89

Tables 1A.1 Background papers written for this report...... 20 2.1 Cyclical components of GDP growth, the unemployment rate, and net flows out of the labor force...... 28 2.2 Cyclicality of net flows across sectors and out of employment, 2005–17...... 31 2.3 Correlation of job loss across sectors...... 36 2A.1 Cyclicality of employment transitions, by gender and skill level...... 49 3.1 Presence of negative effects on employment and wage scarring, by gender and education...... 63 4.1 Landscape of formal unemployment income support in the LAC region...... 88 Foreword

In a region as volatile as Latin America and that only worsen as time passes. This effect the Caribbean (LAC), the recurrence of crises happens because the short-run impact of a should teach lessons that can be translated crisis on the labor market is felt more strongly into better public policy responses. Looking through unemployment than through shifts back and learning from past crises is partic- to informality. ularly relevant at a time when the devastat- At the same time, long-running struc- ing effects of the COVID-19 (coronavirus) tural changes are shifting the nature of pandemic shatter years of progress on many work across the LAC region and globally. fronts, including job creation and new oppor- Crises accelerate these changes and thus the tunities, the formalization of employment, reduction of opportunities in what are tradi- and poverty reduction. Latin America and tionally considered “good jobs”—the stable, the Caribbean must give a different answer to protected employment associated with the the usual problems. formal sector. In this way, a crisis does not That is precisely what this report is about. just temporarily shape worker flows—it also Drawing on decades of data on economic has significant, long-lasting effects on the shocks and labor market responses in the structure of employment. The result is that LAC region, it analyzes the effects of crises traditional formal sector opportunities are on workers and firms to help regional leaders slowly shrinking in the LAC region. direct policies and scarce resources in ways Second, crises in the LAC region have very that bolster long-term, inclusive economic different impacts across workers, sectors, and growth. locations. Low-skilled workers suffer from Employment in Crisis: The Path to Better longer-term scarring, while higher-skilled Jobs in a Post-COVID-19 Latin America has workers rebound fast, exacerbating the three core messages. First, large crises in the region’s high level of wage inequality. LAC region have led to persistent employ- The characteristics of employers and ment loss. The evidence shows that crises are places affect the severity and duration of often turning points in employment, marking crises’ scars on workers. Formal workers the beginning of strong negative deviations suffer smaller employment and wage losses

ix x Foreword

in localities with higher informality. And and laying the foundations for equitable reduced job flows can decrease individual development. The key initial step is to put welfare, but workers in localities with more strong, prudent macroeconomic frameworks job opportunities, including informal jobs, and automatic stabilizers in place to shield bounce back better. labor markets from potential crises. Sound Third, crises can have positive effects fiscal and monetary policies can preserve that increase an economy’s efficiency and macroeconomic stability and avert system- productivity. However, these effects are wide financial strain in the face of a shock. dampened in the LAC region because of its Fiscal reforms, including less distortive less competitive market structure. Rather taxation, more efficient public spending, than becoming more agile and productive financially sustainable pension programs, and during economic downturns—allowing the clear fiscal rules, are the first line of defense “creative-destruction” process—protected against crises. The report also proposes coun- sectors and firms gain more market share tercyclical income support programs, such as and crowd out others, trapping valuable unemployment insurance and other timely resources. transfers to households during downturns, as It doesn’t have to be like this. As the LAC well as social protection and labor policies. region faces the biggest synchronized down- Stronger macroeconomic stabilizers turn in its modern history as a consequence and reforms to social protection and labor of the COVID-19 pandemic, the region has systems are not enough, however. Jump- placed a strong emphasis on managing the starting job recovery by supporting vigorous short-run effects of the crisis and has suc- job creation is also needed. In this context, ceeded in cushioning some of its initial eco- competition policies, region-specific policies, nomic effects. However, this report’s results and labor regulations are a third key policy show that job losses caused by crises can dimension. Without addressing these funda- be particularly painful in the LAC region mental issues, recoveries in the LAC region because of the region’s sluggish recoveries. will remain characterized by sluggish job Thus, more is needed than what has been creation. done in the past. What can be done? Employment in Crisis Carlos Felipe Jaramillo proposes a mix of policies capable of contrib- World Bank Vice President, uting to improving labor market conditions Latin America and Caribbean About the Authors

Truman G. Packard is a lead economist supply decisions, saving behavior, and in the Social Protection and Jobs Global risk management. Truman holds a Ph.D. in Practice of the World Bank Group, cur- economics from the University of Oxford in rently serving as human development prac- the United Kingdom. tice leader for Mexico. Truman was the lead author of the Social Protection and Jobs Raymond Robertson is a professor and holder report Protecting All: Risk Sharing for a of the Helen and Roy Ryu Chair in Economics Diverse and Diversifying World of Work. He and Government in the Department of also led the World Bank team that delivered International Affairs at the Bush School of the regional report East Asia Pacific at Work: Government and Public Service as well as the Employment, Enterprise, and Well-Being in director of the Mosbacher Institute for Trade, 2014. He served on the teams that produced Economics, and Public Policy. He is a research Golden Growth: Restoring the Lustre of the fellow at the Institute for the Study of Labor European Economic Model, published in in Bonn, Germany, and a senior research 2012, and the World Development Report fellow at the Mission Foods Texas-Mexico for 2009, Reshaping Economic Geography. Center. He was named a 2018 Presidential Truman led the World Bank’s Human Impact Fellow by Texas A&M University. Development program in the Pacific Islands, Papua New Guinea, and Timor-Leste, and Joana Silva is a senior economist in the Office he has been part of teams delivering finan- of the Chief Economist for Latin America and cial and knowledge-transfer services to gov- the Caribbean of the World Bank. She has ernments in Europe and Central Asia, East expertise in labor economics, international Asia and the Pacific, and Latin America and trade, poverty and inequality, firm productiv- the Caribbean. Trained as a labor econo- ity, and policy evaluation. Her research has mist, Truman’s work has focused primarily been published in leading academic journals, on the impact of social insurance—includ- including the American Economic Review ing pensions, unemployment insurance, and the Journal of International Economics. and health coverage—on households’ labor She has authored four books, including two

xi xii About the Authors

World Bank regional flagship reports: Wage World Bank. Her research focuses on ques- Inequality in Latin America: Understanding tions related to the labor market, includ- the Past to Prepare for the Future and ing employment shocks, decisions around Inclusion and Resilience: The Way Forward immigration and remittances, and women’s for Social Safety Nets in the Middle East economic inclusion. She has led and contrib- and North Africa. She has led several World uted to several studies and books, and her Bank lending operations and has extensive research has been published in Labour and experience advising governments, in par- the Review of Development Economics. She ticular on the design and implementation is currently the World Bank’s poverty econo- of economic reforms, social programs, and mist for Angola and São Tomé and Príncipe monitoring and evaluation systems. She holds and was previously the poverty economist for a Ph.D. in economics from the University of Brazil and, before that, Honduras. Prior to Nottingham. joining the World Bank in 2013, she was an economist in the Center for Economic Studies Liliana D. Sousa is a senior economist in the at the US Census Bureau. She received her Poverty and Equity Global Practice of the Ph.D. in economics from Cornell University. Acknowledgments

This book would have been impossible with- December 10, 2018, and March 2, 2020. The out the support of a generous team of col- authors are very grateful for insightful com- leagues and collaborators. Special thanks ments from Kathleen Beegle (World Bank), are due to Martin Rama, chief economist Andres Cesar, Marcio Cruz (World Bank), for the Latin America and the Caribbean Augusto de la Torre (Columbia University), Region, World Bank. He provided valu- Ximena Del Carpio (World Bank), Guillermo able comments and suggestions at various Falcone, Maria Marta Ferreyra (World stages during the production of this report. Bank), Alvaro Gonzales (World Bank), Sebastian Melo provided outstanding Henry Hyatt (US Census Bureau), Tatjana research assistance. Karina Kleineberg (World Bank), Maurice This report greatly benefited from the Kugler (George Mason University), Daniel comments and suggestions of our peer Lederman (World Bank), Daniel Mateo reviewers at the World Bank Concept Note (University of Southern California), Samuel and Decision Meetings, including Rafael Pienknagura (International Monetary Dix-Carneiro, professor of economics, Duke Fund), Daniel Riehra, Bob Rijkers (World University; Roberta Gatti, chief economist Bank), Dena Ringold (World Bank), Sergio for the Middle East and North Africa region, Schmukler (World Bank), Erwin Tiongson World Bank; Denis Medvedev, practice (Georgetown University), Carlos Végh manager for Firms, Entrepreneurship, and (John Hopkins University), Lucila Venturi Innovation, World Bank; Julian Messina, (Harvard Kennedy School), and Guillermo lead economist, Inter-American Development Vuletin (World Bank). Bank; and the meetings’ chair, Carlos Felipe This report builds on a series of back- Jaramillo, regional vice president for the ground papers (further described in annex Latin America and the Caribbean Region, 1A to chapter 1). The team is grateful to all World Bank. The report was developed under the authors of those papers. the guidance of Martin Rama. It was also dis- Design, editing, and production were cussed in dedicated authors’ workshops held superbly coordinated by Amy Lynn Grossman,

xiii xiv Acknowledgments

who made the process simple and smooth. Sara Horcas translated it into Spanish. The Laura Handley, of Publications Professionals team would also like to thank Jacqueline LLC, and Sandra Gain skillfully edited Larrabure: this report would not have been the English version. Leonardo Padovani possible without her unfailing administrative translated the volume into Portuguese, and support. Executive Summary

A better policy framework for preventing, dynamics of labor adjustments in the LAC managing, and helping people recover from region in the wake of shocks, the long-term crises is crucial to lifting long-term growth implications of short-term shocks and their and improving livelihoods in Latin America driving mechanisms, and the effects of and the Caribbean (LAC). The need for this policies on labor market adjustments. It uses policy framework has never been more urgent household and labor force surveys (cross- as the region faces the monumental task of sectional and panel), longitudinal employ- recovery from the worldwide COVID-19 er-employee matched data, and analysis of (coronavirus) pandemic. Whether specific national accounts to assess labor market policy responses will deliver the expected dynamics. It also exploits firm shocks stem- growth dividends is an open question. The ming from crises in order to quantify their answer will depend on the underlying under- impacts on labor markets, disentangling standing of how labor markets adjust to cri- their effects from those of concomitant sec- ses and on the quality of the policies enacted. ular forces that also affect employment and This report studies how crises change productivity over long horizons. Finally, it labor market flows in the LAC region, develops structural models and explores rare assesses how these changes affect workers quasinatural experiments to assess the wel- and the economy, and identifies key policy fare implications of crises and to derive better responses to the needs of working people in policy responses to them. the wake of crises. It does so using two dis- The report’s key findings are threefold. tinct but complementary lenses: the macro- First, crises lead to significant losses of economic perspective, focused on the effects employment and earnings, but their impacts of cyclical fluctuations and crises on aggre- vary across countries, sectors, and workers. gate labor market flows, and the microeco- Different labor market dynamics hide behind nomic perspective, focused on crises’ uneven similar reductions in employment. Even with effects across different workers, firms, and the LAC countries’ large informal econo- regions. The report brings together the main mies, which serve as buffers by absorbing findings of a large research project involv- some excess labor, outright unemployment ing ten background papers focused on the remains a significant margin of labor market

xv xvi executive Summary

adjustment to economic shocks in the short traditionally considered “good jobs”—the run. In Mexico, for example, a decrease of standard, stable, protected employment asso- 1 percentage point in GDP growth is associ- ciated with the formal sector. ated with an increase in the unemployment Second, this report finds that some rate of 7.9 percent. These large gross flows to workers recover from displacement and unemployment lead to significant reductions other livelihood shocks, while others are left in household income, increasing vulnerability scarred by them. In the LAC region, scarring as well as expanding and deepening poverty. is more prominent for lower-skilled workers, Among households not living in poverty, a particularly those with no college education. job loss by the household’s main earner For example, in Brazil and Ecuador, work- would push 55 percent of them into poverty. ers with higher education suffer minimal The LAC region’s high informality and impacts from crises on their wages and only formal sector job protections suggest a hier- short-lived impacts on their employment, but archy in adjustment costs, wherein informal the effects on the average worker’s employ- workers, who have fewer protections, face ment and wages are still present nine years the highest likelihood of job loss regardless of after the beginning of a crisis. Those who skill. Indeed, workers in lower-income quin- enter the labor market during a crisis face a tiles are more likely in general to experience worse start to their careers, from which it is negative labor transitions than are workers difficult to recover. in higher-income quintiles. However, this Although the specific mechanisms for study’s results suggest that formal employ- scarring and its duration vary between men ment is overall more responsive to growth and women, the overall story across coun- shocks than informal employment. Although tries in the LAC region is similar—among job loss was part of the equation, the key both men and women, scarring is more likely driver of increases in unemployment during for those with lower levels of schooling and the 2008–09 global financial crisis was a unlikely for those who are college-educated. sharp drop in net job finding rates in the for- It is also far more likely to occur through mal economy. higher unemployment and informality rates Crises do not just temporarily shape years down the road than through lower worker flows—they also have significant, wages. long-lasting after-crisis effects on the struc- Importantly, the conditions in local labor ture of employment. These effects are such markets affect the severity of crisis-induced that traditional formal employment oppor- employment and earnings losses for workers. tunities are gradually shrinking in the econ- Persistent earnings losses may reflect a lack omies of the LAC region. It takes LAC of opportunities in the economic rebound, economies many years to recover from the not just scarring caused by the loss of human contraction in formal employment induced capital associated with a period of unemploy- by a crisis—20 months after the start of a ment or lower-quality employment. In the recession, overall employment remains lower, wake of a crisis, employment losses last and formal employment remains lower for longer for formal workers in localities with more than 30 months afterward. Significant larger primary sectors, smaller service sectors macroeconomic aftereffects of crises per- and fewer large firms. sist for multiple years, and formal employ- At the same time, the presence of a large ment indices remain substantially depressed. informal economy may protect some work- This effect happens throughout the region, ers from shocks. Employment and wage despite differences in labor markets across losses in response to a crisis are smaller for countries. Although longer-running struc- formal workers in the private sector who tural shifts are changing the nature of work live in localities with higher rates of infor- globally, crises further contribute to reduc- mality. This finding suggests that informal- ing employment opportunities in what were ity, including self-employment, can be an Executive Summary xvii

important employment buffer in the medium which exhibits little market contestability, to long run as workers transition from unem- high inequality, and low productivity growth. ployment to formality. These three findings have important This report shows that the reduction policy implications, which are all the more in job flows caused by a crisis decreases relevant in the context of the COVID-19 welfare but that workers in localities crisis, the biggest coincident downturn faced with more job opportunities, either for- by the LAC region in decades. Because crises mal or informal, bounce back better. have a negative effect on the region’s welfare Where employment is scarce, there is less in the short-run, as well as on its potential worker churning, which in turn results for economic growth in the medium run, in lower-quality job matches. These low- policy makers should aim not just to cush- quality matches reduce firms’ productivity ion the economic effect of crises but also to growth and workers’ lifetime earnings and mitigate their adverse impacts on workers, welfare. After a severe crisis, employment firms, and locations. Policies should pay just may not recover to what it had been before; as much attention to building efficiency and the crisis may push the labor market into a resilience—the ability to bounce back when new, depressed equilibrium. exposed to an adverse shock—as to manag- Third, this report argues that crises can ing the short-run effects of crises. have positive cleansing effects that increase This report proposes a three-pronged pol- efficiency and productivity. However, the icy response to crises. First, strong, prudent LAC region’s less competitive market struc- macroeconomic frameworks and automatic tures may dampen this effect, hampering stabilizers should be the first line of defense productivity growth. Job losses spurred by protecting labor markets against poten- an economic crisis can reduce productivity by tial crises, whether these crises are caused destroying employer-employee matches and by external or internal forces. Sound fiscal the job-specific human capital arising from and monetary policies can protect macro- them. That said, large economic disruptions economic stability and avert system-wide can also free workers and other inputs of pro- financial strain in the face of a shock.1 The duction from low-productivity firms, allow- LAC region’s monetary policy has improved ing them to move to more-productive firms significantly since the 1990s, reducing the as the economy recovers. Similarly, crises can incidence of downturns and, with a few spur reallocation of firms out of sectors with exceptions, averting large financial crises. very low productivity. However, fiscal policy is key to macroeco- However, this study shows that sectors nomic policy, and the region’s recent track and firms sheltered from market competition record in this area is weaker. LAC countries’ adjust less during a crisis, suggesting they are fiscal policies have tended to be procycli- less likely to experience this cleansing effect. cal rather than countercyclical. Economic In sectors where a few firms hold a large per- upturns often lead to rapid growth in pub- centage of the market share, a shock does not lic spending that proves difficult to sustain, lead to downward adjustments of real wages while downturns see dramatic drops in or employment. Although the workers in government revenue, forcing painful cuts in these sectors are better insulated from crises, government budgets and the vital services the costs of this protection are borne by the they finance. Fiscal reforms, including economy as a whole. Rather than becoming less-distortive taxation, more-efficient pub- more agile and productive during economic lic spending, financially sustainable pension downturns, protected firms gain more mar- programs, and clear fiscal rules, are crucial ket share and further crowd out competition, to protect against crises. trapping resources that could be used more Countercyclical income support programs, efficiently elsewhere. This phenomenon is such as unemployment insurance and other particularly concerning in the LAC region, transfers to households in bad times, limit the xviii executive Summary

damage from contractions and help economic if crisis-induced, worker-level human capital recovery. However, these instruments are decay were reduced. Strengthening recover- currently missing in most LAC countries and ies will also require going beyond short-term are only weakly responsive to shocks in the income support to protecting human capital countries where they are present. Two-thirds and promoting faster, higher-quality transi- of countries in the region do not offer nation- tions for displaced workers into new jobs. al-level, countercyclical income support plans To achieve this objective, a second key for displaced workers. Where these pro- reform is needed. It consists of increasing the grams do exist, as in Argentina, Brazil, and capacity of LAC countries’ social protection Uruguay, there is only a weak correlation and labor policies and coalescing them into between unemployment insurance claims and systems that provide income support as well economic activity. as preparation for new jobs through reskill- Moreover, one challenge faced by the LAC ing and reemployment assistance (known as region is that large segments of its workforce active labor market programs, or ALMPs). earn their living informally and cannot be More effort is needed to move away from reached through traditional unemployment fragmented and rigid social protection and insurance. Nonetheless, this report shows labor programs toward integrated, adapt- that the expansion of transfer programs tar- able programs that build on comprehensive geted on the basis of household need rather and dynamic social registries and operate as than whether a lost job was formal or infor- coherent, concerted systems. This change mal can perform a key complementary stabi- will entail providing policy packages that are lizing function by supporting local demand, able to cushion the short-term impacts of cri- thus generating aggregate benefits for the ses, prevent lasting human capital losses, and local economy in addition to individual-level facilitate the redeployment of working people benefits. with reskilling and reemployment support. To address their lack of automatic stabi- However, the need to increase the coher- lizers, LAC countries could consider creating ence of the LAC countries’ social protec- or reforming their unemployment insurance tion and labor systems and the coordination programs, making short-term compensation among their interventions is long-standing, programs permanent and enabling these and the region is now living through the con- programs to adapt to changing conditions sequences of not having all the tools it needs more swiftly. By building on the region’s to respond to the current COVID-19 crisis. already-extensive social safety nets and mak- Recent evidence suggests that the global pan- ing some of these programs state-contingent demic and governments’ quick actions to and automatically activated when, for exam- expand some assistance programs can lead ple, unemployment rises above a determined to progress in a crucial step: building better threshold, welfare losses and labor market social registries that are linked and interop- adjustment costs could be lowered in the erable with other administrative informa- wake of future shocks.2 This change should tion management databases. This step can be accompanied by clear rules about the be undertaken in the short run, and it can duration of benefits, strategies for scaling reduce poverty and inequality by inform- down benefits as crises resolve, and the pro- ing the targeting of resources to those most grams’ fiscal costs. affected by crises. However, some crises will still occur, Reemployment will remain crucial to even in the presence of a robust macroeco- avoiding scarring, but the evidence available nomic policy shield, and how to cushion on the effectiveness of the LAC countries’ their long-term impacts on workers is a key ALMPs to date can be discouraging. On the question. The scarring effects documented basis of that evidence, the region should in the LAC region imply that higher long- renew its policy emphasis on reemploy- term growth could be achieved in the region ment support services and incorporate four Executive Summary xix

elements rarely associated with traditional that place-based policies could address the ALMPs: (a) specificity to the particular needs LAC region’s lack of geographic mobil- of job-seekers; (b) coherence and coordina- ity and maximize the gains to workers tion with other parts of the social protection from relocation. Reducing pockets of labor and labor system (most obviously unem- rigidity— especially those caused by restric- ployment income support plans); (c) moni- tions on human resource decisions by firms toring of the programs’ implementation and and individuals—could speed up labor mar- evaluation of their impact; and (d) adequate ket adjustments and shorten transitions. resources from national budgets. Similarly, addressing protectionism and Importantly, although social protection unfair market conditions through better and labor policies are typically the first competition laws, lower subsidies, less state instruments deployed to mitigate the impact participation, and better public procurement of crises, they also play a crucial continuous practices could promote stronger recoveries. role in enabling access to opportunities by Policy responses to crises need to tackle those building human capital (such as education), issues, with varying weights depending on the which, as shown in this report, increases country, the period, and other circumstances. workers’ resilience and ability to rebound As the LAC region faces significant eco- from crises. nomic and social fallout from the COVID-19 However, stronger macroeconomic stabi- pandemic, integrated approaches will pave lizers and reforms to social protection and the way to lower vulnerability and greater labor systems are not enough to change the preparedness for crises. Over the past year, outcomes of crises in the LAC region. More the region has focused on its initial response vigorous job creation is needed in order to to the emergency. Drawing upon lessons from generate better recoveries, and this change past crises, the findings of this report offer will require removing structural constraints. new insights into policy responses focused on The sectoral and spatial dimensions behind workers, sectors, and places. Together, these the region’s poor labor market adjustments responses will support a faster and more must be addressed, or else recoveries in the inclusive recovery from the current crisis, LAC region will remain characterized by laying the foundation for future economic sluggish job creation. growth. In this context, competition policies, regional policies, and labor regulations are Notes a third key policy dimension. For example, this report highlights the dichotomy between 1. Monetary and fiscal stabilization policies firms in the LAC region: some are exposed are powerful tools to respond to crises; they to competition, and others are protected and include the management of a country’s capital hence less prone to restructuring, which is an account, exchange rate policy, fiscal rules, and important source of productivity gains. The sovereign welfare funds as well as interest rate adjustment. Although these policies are cru- report also highlights the low g eographical cial, they are not the main focus of this study. mobility of the region’s working people, 2. During the COVID-19 crisis, many LAC coun- which magnifies the welfare effects of crises, tries took some of these steps: they expanded and pockets of labor rigidity that hinder cash transfer programs and introduced necessary transitions and adjustments in the short-term compensation programs to miti- labor market. gate unnecessary job losses, including work- The findings in this report, together time banking, furloughs, and job retention with a growing related literature, suggest subsidies.

Abbreviations

ALMP active labor market program AUH Asignación Universal por Hijo CCT conditional cash transfer COVID-19 coronavirus EPL employment protection legislation EU European Union GDP gross domestic product hh household survey HR human resource IADB Inter-American Development Bank I2D2 International Income Distribution Database LABLAC Labor Database for Latin America and the Caribbean LAC Latin America and the Caribbean LMP labor market programs OECD Organisation for Economic Co-operation and Development OLS ordinary least squares PATH Program for Advancement through Health and Education PCA principal component analysis PREGRIPS Registro Integrado de Programas Sociales del Estado Plurinacional de Bolivia RS Registro Social RSH Registro Social de Hogar SA social assistance SEDLAC Socio-Economic Database for Latin America and the Caribbean SIFODE Sistema de Focalización de Desarollo SIMAST Information System of the Ministry of Social Affairs and Labour SIMS Database of Labor Markets and Social Security Information Systems SISBEN Sistema de Selección de Beneficiarios de Programas Sociales SIUBEN Sistema Único de Beneficiarios

xxi xxii Abbreviations

SP social protection STC short-time compensation TFP total factor productivity UCT unconditional cash transfer UI unemployment insurance

For a list of the 3-letter country codes used by the World Bank, please go to https://datahelpdesk .worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups. Overview 1

conomic crises result in significant in one or more countries in the region.1 The hardships for millions of people around LAC countries have rebounded from some of Ethe globe, especially the poorest, who, these crises, but others have altered their eco- with few assets and little savings, are more nomic trajectories. This phenomenon is illus- vulnerable to income shocks. A better policy trated in figure 1.1, which shows the severity framework to prevent, manage, and help peo- and persistence of employment losses follow- ple recover from crises is crucial for countries ing Brazil’s debt crisis in the early 1980s, the in Latin America and the Caribbean (LAC) to Asian financial crisis of the 1990s in Chile, accelerate long-term growth and improve the and the 2008–09 global financial crisis in livelihoods of their people. The need for this Mexico. In addition to weak rebounds from policy framework has never been more urgent crises, countries in the region have also expe- as the region faces the monumental task of rienced generalized economic stagnation recovering from the worldwide COVID-19 since 2013. (coronavirus) pandemic. However, whether Although much has been written about such a framework will deliver the expected the frequency and severity of economic crises growth dividends is an open question. The in the LAC region, less is known about how answer will depend on the underlying under- these episodes affect workers, in both the standing of how labor markets adjust to crises short run and the long run, and about how to and on the quality of the policies enacted. respond to these effects with policies. Focus- ing on workers is important, because the Rationale for this report long-run impacts of crises on labor markets may drive deeper losses in income than has The LAC countries experience macroeco- been previously understood. Moreover, if nomic fluctuations more frequently and often crises destroy human capital, they can have more severely than most other regions of the long-run effects on aggregate economic world. And crises, not a single crisis, charac- growth. terize the recent history of most countries in Several open questions are impeding the region. One-third of the fiscal quarters progress in this field. First, how large is the between 1980 and 2018 were periods of crisis impact of crises on workers? The effects of

1 2 employment in Crisis

FIGURE 1.1 Persistent employment loss following crises: The myth of economic recovery

a. Debt crisis b. Asian financial crisis c. Global financial crisis of 2008–09 Brazil Chile Mexico 11.0 10.5 9.2

9.0 10.9

8.8 10.0 10.8 8.6 10.7 8.4

9.5 10.6

Employment (log of number workers) 8.2 1985 1990 1995 2000 2005 2010 2015 1985 1990 1995 2000 2005 2010 2015 2003 2005 2007 2009 2011 2013 2015 2017

Source: Regis and Silva 2021. Note: The vertical bars indicate recessions. Series are seasonally adjusted. Data are from the first quarter of each year.

crises in the LAC region are superimposed region adjust in face of economic shocks, on a slowing trend of slowing employment assesses how the employment adjustments in and productivity (Fernald et al. 2017). Dis- response to these shocks affect workers in the entangling their impact on this slowdown short, medium, and long runs, and discusses from that of other factors is difficult because key policy responses to mitigate these shocks’ concomitant secular forces, such as techno- negative consequences. Its findings provide a logical progress and globalization, also affect new understanding of the medium- and long- employment and productivity over long hori- term implications of crises for labor markets zons (Ramey 2012). Second, research thus far and suggest policy responses to the 2020 has focused separately on crises’ short- and COVID-19 crisis. long-run effects, but how are these effects This study draws on existing data on linked? In the missing medium term, defined economic shocks in Latin America. These as the 8–12 years after the start of a crisis, shocks include permanent shocks, such as crises appear to result in microeconomic technological change and the trade liberaliza- transformation with persistent but poorly tions of the 1990s,2 and transient shocks, such understood effects. And third, how can pol- as exchange rate fluctuations. Some transient icy makers improve the outcomes of crises? shocks are sector- or location-specific or are Because of the emergency mode of crises, idiosyncratic to a particular set of house- robust evidence is lacking on the effective- holds, while others are systemic, affecting ness of the main policy instruments used to the region’s whole economy. Although each intervene in them. However, newly available, type of crisis has implications for productiv- granular data and advances in empirical ity and welfare, this study focuses on eco- approaches have opened the door to a much nomic crises: large, negative economic shocks deeper understanding of this issue. (rather than small fluctuations in gross The need for this analysis and policy domestic product [GDP]) that are systemic framework has never been more urgent as rather than idiosyncratic and transient rather the LAC region faces the monumental task than permanent. Through novel identifica- of recovering from the worldwide COVID-19 tion strategies and the use of new data, this pandemic. This flagship research project study aims to tease out the effects of shocks examines how labor market flows in the from those of concomitant secular forces Overview 3

that also affect employment and productiv- on workers. How do sectors and firms adjust ity. Finally, although crises also have import- employment and wages? What other margins ant noneconomic consequences, this study of adjustment are used, beyond shedding focuses only on their economic implications. jobs, and what are their medium- to long-run Although this study was prepared within effects on efficiency? How do the charac- the context of the COVID-19 crisis and teristics of localities (such as their economic includes some analysis of this crisis’s imme- structure) matter? This study will examine diate impacts, such as the evolution of the each of these questions. economy in 2020, the study’s goal is more The effects of a crisis also depend on general than this crisis. First, given the LAC a second crucial dimension, one being region’s pattern of frequent crises, the study discussed globally during the COVID-19 aims to understand the effects of crises in crisis: the policy response to the crisis and general, not just of the COVID-19 crisis, whether it successfully connects welfare on labor markets in the region. Second, in considerations with the country’s growth order to understand how labor markets in agenda. Few doubt that avoiding crises in the region adjust to crises and to unpack the first place is a priority. A more stable the underlying mechanisms driving these macroeconomic environment decreases the adjustments, the analysis necessarily draws incidence of growth shocks, while automatic upon the medium- and long-term effects of stabilizers such as countercyclical income earlier crises. Although the COVID-19 cri- support for job seekers serve to smooth the sis, projected to be the most severe labor impacts of shocks on the national economy. market recession in some countries’ histo- These measures are a crucial first shield ries, differs from any previous crisis and against economic crises, but they are largely comes along with supply disruptions and lacking in the LAC region. Correcting this prolonged uncertainty, it also shares some deficiency will require adjusting not only fis- features with previous crises.3 These fea- cal and monetary policies but also social pro- tures include a global recession, a sharp fall tection and labor policies. in demand for many months, and financial However, even in the presence of a robust stress or impending financial crises in some macroeconomic policy shield, some crises countries. Indeed, the available information will still occur, and what to do to cushion on unemployment in 2020 suggests that the their impact on workers is a key question. COVID-19 crisis induced a pattern of unem- This study argues that there is scope, when ployment similar to that observed in previous increasing the capacity of the LAC region’s crises, for example, with lower-skilled work- social protection and labor policies, to ers more directly affected than higher-skilled coalesce these policies into systems that workers. Although the source of the shock is provide income support as well as prepare different, the COVID-19 crisis’s similarities workers for new jobs through reskilling to previous crises suggest that these events and reemployment assistance. But will these can offer relevant lessons for the present and, measures be enough to spur job creation importantly, for future crises. and generate better recoveries? Considering This study proposes a new understand- the evidence presented in this report, there ing of how the LAC region’s labor markets is an urgent need for LAC countries to also adjust to crises, triangulating effects on tackle structural problems, including low workers, sectors and firms, and localities. product market competition in some sec- This approach follows from the study’s find- tors, contestability issues, and the spatial ing that job losses caused by crises are par- dimensions behind poor labor market adjust- ticularly painful in the LAC region because ment. Without addressing these fundamental of the region’s sluggish recovery processes. issues, economic recoveries in the LAC region The pace of job creation depends on demand- will continue to be characterized by sluggish side factors, like firms and locations, not just job creation. 4 employment in Crisis

Road map explanations and empirical evidence on how crises affect sectors and firms and This study brings together the macroeco- how they affect efficiency in the medium nomics of the effects of cyclical fluctuations to long run; and (c) explains why these and crises on GDP and the microeconomics microeconomic processes play out so dif- of these crises’ uneven effects across work- ferently for workers and firms in different ers, sectors, and regions. It builds on the localities. main findings of a large research project that • Chapter 4: Toward an Integrated Policy produced 10 background papers focused on Response assesses the suitability of the cur- the dynamics of labor adjustments to shocks rent policy framework to address the long- in the LAC region, the long-term implica- term nature of labor market adjustments tions of short-term shocks, the mechanisms and discusses potential reforms to macro- driving these shocks, and the effects of pol- economic, social-protection and labor, and icies on labor market adjustment (see annex competition and place-based policies that 1A for additional details). could help cushion the short- and long- This study is organized around term impacts of crises and address their three analytical themes that are key to sources at the sector-and-firm and loca- understanding how labor markets adjust to tion levels, bridging welfare and growth crises and the consequences of these adjust- agendas. ments on workers in the short and long terms: This study builds on the key findings • Chapter 2: The Dynamics of Labor of the background papers and includes Market Adjustment studies how crises a richer treatment of the literature and a have affected labor market flows in the summary of recent policy interventions in LAC region over the past 20 years and order to provide new insights into current assesses the extent to which they shape policy debates. The background papers its employment structure. Rather than and complementary analysis developed by focusing on one number (for example, this study use rich and varied data sources the unemployment rate or the elasticity and analytical approaches, including new of unemployment to output changes), it evidence from harmonized household and considers the mechanisms of labor mar- labor force surveys (cross sectional and ket adjustment, the cyclicality of job panel) for 17 LAC countries; longitudinal flows, the extent to which these flows are employer-employee matched data for Brazil heterogenous across firms and workers, and Ecuador (following all formal work- and whether employment adjustments ers and firms for more than 15 years); and lead to changes in employment structure. national accounts, which were analyzed to • Chapter 3: The Impact on Workers, assess labor market dynamics. To examine Firms, and Places evaluates the medium- the causal effects of crises, the study also to long-term effects of shocks on labor exploits firm shocks stemming from foreign market outcomes at the worker, firm, and demand shocks. These exogenous shocks locality levels, extending beyond welfare allow crisis effects to be disentangled from to include broader efficiency issues. It those of other forces that affect employ- (a) evaluates the magnitude and dura- ment and productivity over long horizons. tion of the medium- to long-term costs This evidence is complemented with find- of crises for affected workers in terms ings from a structural model developed of earnings and employment and identi- using detailed data from Brazil to assess fies the types of workers most at risk for the welfare implications of labor market long-term welfare losses; (b) offers causal adjustments. Overview 5

Key insights 60 percent of household income among the poorest 40 percent of households in LAC This study offers new insights into each of countries. Among households not living its three analytical themes: the dynamics of in poverty, a job loss by the main earner labor market adjustment; the impact of cri- would push 55 percent into poverty. And ses on workers, firms, and places; and policy job loss imposes costs on workers that go far responses to crises. beyond the immediate loss of income.4 Eco- nomic adjustment through unemployment is Crises have significant impacts on the also especially costly because jobs are more structure and dynamics of employment quickly lost than gained; in other words, in Latin America employment losses can persist for a substantial period following a crisis. How do crises change labor market flows? A macroeconomic shock results in microeconomic reallocation at the worker and firm Net loss of formal employment is driven by levels. At these defining moments, workers’ reduced job creation and firms’ fates are linked. Firms can adjust This report reveals that most of the reduction their number of employees, hours of work, in employment in response to a crisis occurs and wages paid, and workers can choose in the formal sector. This result is important, to accept these offers or search for other because the potential for good worker-job options. From these interactions, a new matches, firm-specific earnings premiums, short-run equilibrium is formed. and match-specific human capital is highest among people who are employed in the for- Negative shocks result in unemployment mal economy. Fluctuations in unemployment more than informality in the short run are determined by the rates of transitions According to new research for this report, into and out of unemployment—job loss and the short-term adjustment to crises occurs job finding rates, respectively. During eco- primarily through unemployment (Sousa nomic downturns, these rates are driven by 2021). Although exits from the labor force increased job destruction (existing positions and shifts to part-time work do not appear are eliminated), reduced job creation (new to be significant margins of adjustment, positions are not created), and lower levels Sousa (2021) finds a strong negative cor- of job reallocation, or churning, as fewer relation between formal and informal job workers voluntarily leave their positions to flows in five of the six countries analyzed look for better matches.5 Addressing each (reductions in formality are often accom- of these transitions will take different policy panied by increases in informality, and vice tools. The relative contribution to unemploy- versa), where the countries’ large informal ment of each type of transition varies across economies serve as de facto safety nets by labor markets; some studies of high-income absorbing excess labor. But even with this economies find that cyclical unemployment buffering function, outright unemployment is driven by reduced job finding rates, while remains a significant margin of labor mar- others find that it is driven by increased job ket adjustment to economic shocks in the separation rates.6 LAC region. Sousa (2021) finds low cyclicality of job The large gross labor flows to unem- losses across formal and informal employ- ployment, in turn, represent significant ment. Instead, in most countries analyzed, reductions in household income, increas- adjustment in employment during the global ing vulnerability as well as expanding and financial crisis of 2008–09 was driven by a deepening poverty. Labor income represents drop in net job finding rates that was larger 6 employment in Crisis

in the formal economy than in the informal in the LAC region have significant effects on economy. Zooming in on the formal economy, the structure of employment that last for sev- Silva and Sousa (2021), using worker-firm eral years (Regis and Silva 2021). In Brazil, linked administrative data from Brazil and Chile, Ecuador, and Mexico, the shrinkage Ecuador, find that a reduction in job cre- of standard formal employment has been ation, rather than an increase in job destruc- strong and long-lasting. It takes LAC econo- tion, drives lower formal employment during mies multiple years to recover from the con- crises—net formal job losses are driven by a traction in formal employment induced by a reduction in new formal employment. They crisis. For 20 months after the start of a reces- also find that although larger firms tend to be sion, overall employment tends to remain more productive and more resilient to crises, lower, and formal employment remains lower they also exhibit higher cyclical fluctuations for more than 30 months. Significant mac- in labor demand. That is, although the firms roeconomic aftereffects of crises persist for themselves are resilient, employment in them multiple years, leaving formal employment may not be the most resilient to economic indices substantially depressed. This effect shocks. Once the higher “death rates” of happens throughout the region despite differ- small firms are taken into account, the pat- ences in labor markets across countries. terns of employment fluctuation look very Although long-running structural shifts similar for small and large firms. are changing the nature of work, crises Are labor flows for lower-skilled or infor- further contribute to reducing employment mally employed workers more cyclical than opportunities in what were traditionally con- those for formally employed or higher-skilled sidered “good jobs”— the standard, stable, workers? The LAC region’s combination of protected employment associated with the large informal economies and workers of formal sector. Moreover, although informal widely varying skill levels suggests a hierar- work seems to be a long-term buffer of cri- chy in adjustment costs, wherein informal ses in some countries, including Brazil and workers, who have fewer job protections, face Chile, in others, such as Ecuador and Mex- the highest likelihood of job (and livelihood) ico, informal employment is stagnant or loss regardless of skill. Workers in the lower falls in response to crises. These results income quintiles are more likely in general suggest that a crisis has the potential to push to experience negative labor transitions than the labor market into a new equilibrium workers in the higher income quintiles—but between formal and informal employment, overall, this study’s results suggest that high- with long-term implications for welfare and skilled employment is more responsive to productivity. growth shocks than low-skilled employment. This finding is consistent with the afore- Crises scar workers, but the characteristics mentioned greater cyclicality of employment of firms and places affect the severity and in large firms and higher cyclical job losses duration of these scars among formally employed workers, because such workers are more likely to be skilled. Although the evidence presented thus far suggests that economic crises have detri- Stable and protected job opportunities are mental impacts at the aggregate level, how gradually shrinking in the LAC region severe are their impacts on individual work- Changes in labor market dynamics can ers and the economy? What do they mean lead to changes in the composition of for welfare and efficiency when the three an economy’s workforce. The potential main dimensions of labor market adjust- macroeconomic aftereffects of a crisis on the ment (workers, sectors, and localities) are structure of employment can shape the cri- considered? The impacts of a crisis leave sis’s medium- to long-term effects on employ- scars on workers and firms. Many workers ment and wages. As this study reveals, crises do not fully recover even in the long run; Overview 7

their earnings do not bounce back and their informally) rather than in long-term effects careers follow a different, worse track. The on earnings. And it is more prominent among biggest losers lose a lot. Firms adjust to crises lower-skilled workers (those with no college in ways that affect their efficiency and resil- education) than among higher-skilled work- ience down the road. Overall, crises scar ers. For example, in Brazil and Ecuador, workers, but the structure of product markets effects on workers’ employment and wages and the conditions within local labor markets persist for an average of 9 years after the affect the severity of these scars. beginning of a crisis. Workers with higher education tend to suffer small impacts from Low-skilled workers suffer most from scar- the crisis on their wages and very short- ring, while high-skilled workers suffer small lived impacts on their employment (Moreno and brief impacts and Sousa 2021). Similarly, Fernandes and How extensive is scarring from economic Silva (2021) find stronger scarring effects crises in Latin America, and what forms from the 2008–09 global financial crisis in does it take? This flagship research project employment and wage outcomes for low- analyzes scarring across three dimensions— er-skilled workers than for higher-skilled that caused by job loss, by initial conditions workers in the formal sector in Brazil and at entry, and by the effects of crises on firms. Ecuador. One explanation for this effect In the case of scarring caused by job loss, is that competition is lower for skilled jobs this study finds large and long-lasting wage because of the relative scarcity of college effects from displacement because of firm graduates in the LAC region. In other words, closure. For example, two years after the clo- scarring likely exacerbates the region’s high sure of a plant, wages were 11 percent lower level of wage inequality across skills. for displaced workers than for nondisplaced Switching gears to consider how economic workers. Four years later, the wage gap was shocks pass from firms to their employees, 6 percent. Wages did not fully recover until Fernandes and Silva 2021 find that among nine years after the closure (Arias-Vázquez, workers with similar initial observable char- Lederman, and Venturi 2019). acteristics, it is more difficult for workers of Next, this project examines scarring in the firms that are most affected by a crisis to the LAC region caused by the labor market recover from the shock. However, the effects conditions when new workers first enter the on workers vary according to their employ- labor force (Moreno and Sousa 2021). Are ers’ or firms’ characteristics; for example, there long-term employment and wage conse- the effects are smaller for workers in large quences of entering the labor market during a firms. Evidence in this report also shows downturn, generating what the popular press that workers who lose their jobs, even if they calls “a lost generation”? This question is regain employment later, experience a last- particularly salient for the LAC region, given ing decline in their earnings. Furthermore, its high rates of youth unemployment and unskilled workers suffer the most from this its investments in increasing and improving decline, which has implications from an educational outcomes at the secondary and equity and poverty-reduction perspective. tertiary levels. Are these investments in the region’s human capital stock undermined by Reduced job flows can decrease individual frequent crises? welfare, but workers in localities with more Using detailed data for four LAC coun- job opportunities, including informal jobs, tries, the project’s findings confirm that bounce back better entering the labor market during a crisis can Negative aggregate demand shocks and cri- have long-term consequences. However, this ses reduce welfare in part by reducing job scarring is found in employment outcomes flows (Artuc, Bastos, and Lee 2021). Because (lower participation rates, higher unemploy- of this effect, during slowdowns and cri- ment rates, and higher likelihood of working ses, job match quality falls. The estimated 8 employment in Crisis

utility from job match quality also decreases, labor market struggle to start their careers. because workers become less mobile during Efficiency changes permanently, and positive crises. A structural model for Brazil, devel- employment effects depend on the econo- oped in a background paper for this proj- my’s ability to create jobs. Because firms are ect, shows that an adverse external shock a key transmission channel of crisis effects to a local labor market lowers welfare sig- to individual workers, the speed of work- nificantly within the market and that low ers’ adjustment and the outcomes of the mobility across regions magnifies this impact new equilibrium also depend on the initial (Artuc, Bastos, and Lee 2021). structure of the economy’s product markets, This report finds larger and longer-lasting rents, and rent-sharing mechanisms. losses in employment (and sometimes in wages) in the wake of a crisis for formal Crises can increase productivity and workers in localities with larger primary sec- efficiency tors, smaller service sectors, and fewer large During a crisis, employer-employee matches firms (Fernandes and Silva 2021). In such and the job-specific human capital arising cases, these workers’ persistent losses in earn- from them, which often take a long time ings may reflect their lack of opportunities to build and would regain viability when in the rebounding economy, not just scarring the economy goes back to normal, may be in the traditional sense of a persistent loss of permanently dissolved only because of the human capital associated with a period of severity of the temporary shock. These job unemployment or lower-quality employment. losses may slow the ramping up of produc- On the other hand, the presence of a large tion later, and they imply losses in productiv- informal economy may protect some workers ity; however, they can also have an important against shocks. This study finds smaller cleansing effect and lead to increased produc- employment and wage losses in response to tivity at both the firm and market levels. crises for formal workers in the private sec- A crisis can also have persistent effects tor who live in localities with higher rates on technology, which can be a margin of of informality (Fernandes and Silva 2021). adjustment firms use to cope. Firms adjust This finding suggests that informality can to crises through changes in productivity, be an important employment buffer in the changes in demand for various skills, mark- medium- to long-run, when workers may ups, and changes in their products to make transition from unemployment to informal them more appealing to consumers (Mion, employment. Such an effect was shown by Proenca, and Silva 2020). Negative demand Dix-Carneiro and Kovak (2019) in the case shocks cause more-affected firms to reduce of adjustment to trade liberalization. Indeed, their capital-to-worker ratios in some coun- transitions from unemployment to informal tries, such as Ecuador, whereas in other employment were twice as common in the countries, such as Brazil, firms simply adjust Brazilian data as transitions from unemploy- their employment and wages (Fernandes and ment to formal employment. Silva 2021). Firms also, in the presence of a negative external demand shock, increase the skill content of their production (the share Crises can have positive cleansing of skilled labor in their total employment) effects that increase efficiency and in countries such as Argentina (Brambilla, productivity, but less competitive Lederman, and Porto 2012), Brazil (Mion, market structures mitigate these effects Proenca, and Silva 2020), and Colombia As an economy transitions to a new equi- (Fieler, Eslava, and Xu 2018). librium, many workers lose their jobs or Moreover, crises can affect the structure see their earnings fall, some firms go out of a country’s economy. They induce firm of business, and new entrants into the exit—not on impact, but around two years Overview 9

after the shock, as shown in Brazil and Ecua- Protected sectors and firms adjust less dor (Fernandes and Silva 2021). Debt over- during a crisis, suggesting less opportunity hang problems may also occur and have for cleansing effects in these sectors the potential to scar firms. Crises destroy There is a complex interplay between less-resilient firms and increase the market labor markets, product markets, and local share of more-resilient ones. In addition to conditions—and understanding it is essential their effects on existing firms, crises can have to crafting sound economic policies. Going persistent effects on firms that are created in beyond the firm and considering the role of bad times. Demand is a key driver of firm market structure, this project’s data from Bra- capabilities, and if firms start at a time when zil show that employment in more-protected demand is low, they will have more difficulty firms, defined as those facing less competi- developing their network of clients and learn- tion, is less affected by crises compared with ing how to work well with them. New evi- employment in less-protected firms (Fer- dence from the United States indicates that nandes and Silva 2021). In sectors in which a firms that are born in times of crisis end up few firms hold a large percentage of the mar- stunted—that is, they grow slowly through- ket share, a bad shock does not lead to any out their life cycles, even when times improve downward real wage adjustments. Instead, it (Moreira 2018). These crisis effects can have can lead to an increase in employment: the persistent implications for the economy, and opposite of what normal economic mecha- firms may find it difficult to revert them. nisms would bring about. Similarly, employ- By inducing firm exit, bad economic ment responds less to a negative export shock times can have a cleansing effect and if the firm in question is state-owned. increase productivity. Suppose that a labor Although workers in protected firms are market is subject to large friction, so that better insulated from crises than those in very-low-productivity firms can survive by other firms, the costs of this protection are hiring workers for very low wages. Given borne by the economy, which ends up with the labor market’s friction, workers who lower overall productivity. Importantly, this receive these low-wage offers take them— result suggests that the presence of protected the opportunity cost of continuing their job firms and sectors in the LAC region may search is high because job matching rates are contribute to its low level of productivity by low. Therefore, these low-productivity firms reducing the potential gains in efficiency and can essentially trap resources that could be productivity caused by crises. Rather than more efficiently used elsewhere. In this con- becoming more agile and more productive text, large economic disruptions can have a during economic downturns, protected firms cleansing effect by freeing workers from such can increase their market share and further firms and allowing those workers to reallo- crowd out competition. As noted above, they cate themselves to more-productive firms as may also trap resources that could be used the economy recovers. Similarly, crises could more efficiently elsewhere. allow reallocation out of sectors that have Although this study focuses on Latin been living in very low productivity at the America and the Caribbean, its findings have margin of survival. This effect is good as long implications for understanding the upgrading as the economy sees job creation after the cri- process in other regions as well. In particular, sis is over. However, the effects of crises on the study’s results reinforce the idea that cri- productivity have gone in the other direction ses affect employment and productivity in the in Brazil: crises there have led to persistent long run, not just in the short run. The study’s reductions (not increases) in firm produc- particular setting, the LAC region, has the tivity. In Ecuador, on the other hand, crises advantage of allowing the clean identification have caused a positive, but small, increase in of a causal relationship between crises and a productivity (Fernandes and Silva 2021). broad range of welfare and efficiency effects, 10 employment in Crisis

but the study’s basic findings seem likely to ability to bounce back when exposed to an apply more broadly. adverse shock, which can be aided by healthy economic growth. Three dimensions of the policy Figure 1.2 presents a framework for think- response ing about the relevant areas of policy. Strong, prudent macroeconomic frameworks and Considering the importance of demand to automatic stabilizers (the shield in figure 1.2) an economy’s well-being and the triangle are the first line of defense to shield labor of workers, firms, and locations, how can markets from crises. Prudent fiscal and mon- policies mitigate the impacts of crises on etary policies are also powerful instruments, workers and promote better recovery from preventing many types of crises and ensuring crises? This study shows that crises have the fiscal space needed to provide support a meaningful negative effect on welfare in and avert system-wide financial strain when the LAC region. The labor market scarring crises do occur.7 effects it documents are likely to hamper In addition to macroeconomic policies, the the region’s economic growth potential. To typical automatic stabilizer used in Organ- mitigate them, policies should try to cush- isation for Economic Co-operation and ion the effects on workers in the short run: Development (OECD) countries is unemploy- shock impacts spread in unequal ways across ment insurance, which many LAC countries workers and firms, and many of them will lack. This type of social protection and labor not regain their lost jobs, wages, or clients. program is key to cushioning the impact of But policies should pay just as much attention crises on formal workers. However, many to efficiency and resilience, promoting the workers in the LAC region are informal, and

FIGURE 1.2 How adjustment works and the policies that can smooth it

Income support s k Active labor market programs r o WORKERS w e

SHOCK m a r f

o c

e Competition policies o

r Local investment + place-based policies

a Labor regulations

t S

Source: World Bank. Overview 11 the best way to protect their incomes and Automatic stabilizers and consumption is through cash and noncash macroeconomic frameworks as shields transfers. Targeted on the basis of household The LAC region has significantly improved its need rather than whether a lost job was for- macroeconomic framework over the past few mal or informal, these programs soften the decades. Because of these efforts, the region extent to which labor market adjustments has fewer domestic crises than it used to. With translate into short-term and long-term few exceptions, large monetary crises have impacts on the poor and vulnerable. Because not occurred in the region since the 1990s; reemployment is crucial to avoid scarring, instead, factors exogenous to the region now reskilling and reemployment services (known drive most of its crises. One important devel- as active labor market programs) are a third opment is that most countries in the region crucial type of social protection and labor have lowered their inflation. In the 1980s and program. The key role of a country’s social 1990s, when the region had high inflation, cri- protection and labor system in determining sis adjustment through lower real wages was the size and persistence of a crisis’s impacts is often mechanical: as inflation spiked during a illustrated by the upper arrow in figure 1.2. crisis, real wages fell. Now, adjustment to cri- Although social protection and labor ses is mostly done by changing employment, systems can cushion the impacts of crises on which is associated with long-lasting scarring workers, they do not address the structural effects, as mentioned earlier in this chapter issues that help determine the magnitude (Robertson 2021). of these impacts. For example, this study Obviously, it is better to avoid crises when- highlights a dichotomy between protected ever possible. A prudent macroeconomic (because of their market power and lack of framework is key to reducing the frequency competition) and unprotected firms in the of crises. But some crises are unavoidable. LAC region and the region’s low geographic Monetary and fiscal stabilization policies mobility among workers, both of which mag- are a powerful tool to respond to such crises. nify the welfare effects of shocks. The study These policies include the management of also highlights pockets of labor rigidity that the country’s capital account, interest rates, are slowing transitions across jobs. Hence, exchange rate policy, sovereign welfare competition policies, regional policies, and funds, and fiscal rules. Importantly, hav- labor regulations (illustrated by the lower ing fiscal space to provide a demand stimu- arrow in figure 1.2) are a third key dimension lus may be key to resolving a crisis, but the in policy responses to crises. This dimension space available depends on current as well touches on important structural issues that as past decisions. Fiscal reforms with a long- may explain poor adjustment, that might term perspective are key; these reforms may require interventions at the sector and local- involve tackling difficult issues, including ity levels (in addition to worker-level and tax policy, energy subsidies, the efficiency of economywide interventions), and that inter- social spending, and the financial stability of act with the aforementioned social protection old-age pension systems as populations age. and labor needs and incentives (as illustrated To protect a country from external by the vertical arrows in figure 1.2). The pol- shocks, a strong set of automatic stabilizers icy response to a crisis must squarely tackle is key. Among these stabilizers are nation- the relevant structural issues, which will ally administered income protection arrange- have different importance in each country or ments, such as unemployment insurance, and setting. other forms of countercyclical income sup- Given the complexity of labor market port that expand in bad times to help affected adjustments to economic crises in the LAC individuals. These programs stimulate con- region, this report argues that countries can sumption, providing a demand stimulus that improve their responses by advancing on limits damage and helps speed recovery. three fronts. 12 employment in Crisis

Automatic stabilizers help households smooth short- and long-term harm of layoffs. Social their consumption, reducing the immediate assistance cash transfer programs have also impact of the shock on aggregate demand been expanded, and evidence in this project and employment and therefore mitigating the shows that their expansion has increased magnitude and composition of its effects on employment at the aggregated local econ- labor markets. In other words, these policies omy level, in addition to causing positive can reduce the severity of a crisis. effects on poverty and inequality (Gerard, The LAC region still needs stronger Naritomi, and Silva 2021). Installing some automatic stabilizers in order to ensure effec- of these instruments as permanent parts of tive fiscal responses to crises. Missing or poorly their respective countries’ automatic stabiliz- functioning aggregate stabilizers limit govern- ers could lower losses and adjustment costs ments’ ability to offer dynamic, countercyclical in the wake of future shocks. Some of these spending, which makes crises harder to man- programs could be made state-contingent age and amplifies the effects of shocks. and automatically activated when, for exam- Beyond large-scale unemployment ple, unemployment rises above a determined insurance programs, other policies can also threshold. These microeconomic policies serve as automatic stabilizers. During the have macroeconomic consequences. COVID-19 crisis, for example, strategies like A more complete characterization of the work-time banking, furloughs, job reten- policy areas that can be focused on in order tion subsidies, and short-term compensa- to achieve stronger macroeconomic frame- tion programs8 have made up an important works and create automatic stabilizers (policy share of the spending meant to help limit the dimension 1) is given in figure 1.3.

FIGURE 1.3 Stabilizers and macroeconomic frameworks: Policy reforms

s k r o Prudent macroeconomic frameworks to avoid crises w e • Normalized in ation implies labor market adjustment on quantitative SHOCK m a employment, with long-lasting scarring r f

c i Monetary and scal stabilization policies to manage crises

o • Create scal space with a broader, long-term perspective (tax policy,

o energy subsidies, social spending eciency, and nancial sustainability of c

e pensionsystems)

r c

a Automatic stabilizers to smooth crises

• Create or reform unemployment insurance (UI)

e • Make short-time compensation (STC) programs a permanent part of the

i economy’s automatic stabilizers b

t S • Give UI and STCs the ability to adapt to changing conditions more swiftly

Source: World Bank. Overview 13

Workers: A policy package for Governments around the world have cushioning crises’ impacts and learned the importance of strong social pro- preparing for change tection and labor systems to limit scarring and other human capital losses caused by The labor scarring documented in this crises. Despite advances in this area, formal study and its adverse impact on countries’ assistance in the case of labor income loss— productivity potential imply that greater or other losses of livelihood associated with long-term growth could be achieved in the economywide transitory shocks—is still LAC region if crisis-induced, worker-level beyond the reach of most people in the LAC human capital decay was reduced. This region. Two-thirds of LAC countries do not change would require cushioning the short- yet offer nationally administered income term impact of job loss through income sup- support plans for people who lose their jobs. port to protect welfare. However, displaced These countries rely instead on severance workers need more than just income support pay mandates, which perform poorly in the to recover from crises; they also need social context of systemic shocks. In terms of job protection and labor systems that help build search support, most LAC countries spend human capital and promote faster, high- very little on active labor measures, and er-quality transitions into new jobs. Social even those that spend more have poor track protection and labor systems should help records for their programs. people to renew and redeploy their human At the same time, countries’ social pro- capital. In this broader sense, reforms to tection and labor systems are oriented prin- the LAC region’s existing social protection cipally toward providing cash transfers to and labor policies and systems are needed. chronically poor households. Although these These reforms will, in turn, affect labor programs offer vital “last-resort” support market flows and provide a responsive sys- and in some countries are able to quickly tem that contributes to countries’ automatic scale up during crises, they still fall short of stabilizers. meeting the needs of most displaced work- Although some workers can benefit from ers. In the COVID-19 crisis, countries have expansionary macroeconomic policies, this relied heavily on cash transfer programs to study shows that others are scarred more get money into the hands of vulnerable peo- permanently by crises and are unlikely to ple quickly. Some of these programs have respond to such policies. Social protection been more effective than others: for exam- and labor systems would be the second line ple, the success of these efforts in the LAC of response to avoid or mitigate the aforemen- region is largely determined by the coverage tioned scarring effects. However, in general, of the population by social registries, which and despite tremendous advances in the past allow programs to be quickly expanded to thirty years, countries in the LAC region include previously uncovered and newly still lack reliable and robust income protec- vulnerable groups. Countries that entered tion paired with effective job-search support the COVID-19 crisis with low-coverage services. The need for such programs is inten- social registries and weaker social assistance sified by the fact that the margin of adjustment programs were less able to provide robust to crises has shifted toward the quantity of income protection. employment, resulting in more cuts in hours, How can LAC countries do better for more dismissals, and, as shown by research workers and communities in terms of social for this study, much slower creation of new protection and labor responses to crises? Pol- formal employment relationships. Most peo- icy actions to buffer the effects of crises on ple who lose their jobs or whose livelihoods workers can be organized into the following are otherwise adversely impacted in a down- categories: turn are largely unprotected. 14 employment in Crisis

1. Augmenting unemployment income sup- the emergence of assistance “ghettos” port through the creation or redesigning by structuring benefits to incentivize the of unemployment insurance. A history return to work (with support from aug- of frequent systemic shocks combined mented reemployment services). with the emergence of a significant middle class has created more demand for 3. Building robust and coordinated em- robust unemployment insurance mech- ployment services to get people back to anisms in LAC countries than exists in work quickly. Several lessons from inter- other regions (De Ferranti et al. 2000). national experience can be used to guide Past crises and the 2020 pandemic shock the reform of reemployment support have dramatically demonstrated the use- services. First, it is important to move fulness of unemployment income sup- away from single interventions and to- port systems with deep and extensive ward providing an integrated package of risk pools that provide a channel for ad- services (such as combinations of coach- ditional, extraordinary support measures ing, training, information, and intermedi- when needed. In Latin America, several ation, all informed by market demand). countries have implemented changes to Even individuals who are affected by the their social insurance plans that loosen same type of shock seldom face identical eligibility requirements and increase ben- constraints to accessing new jobs. Hence, efits. For example, Brazil and Chile, in the success of a reemployment program addition to paying benefits to displaced depends on its ability to adapt its services workers, used their unemployment in- to different profiles and needs. Second, to surance systems to implement subsidized make this change, public employment as- furlough measures and other employ- sistance services need strong registration ee-retention programs. These systems and statistical profiling systems. Finally, make all the difference in how well labor modern monitoring and evaluation prac- markets adjust to crises. tices are key to assessing the results of programs and to introducing course 2. Improving the capacity of social corrections when needed. Making larg- assistance transfer programs to be robust er, more effective programs fiscally sus- and responsive. There are three main pol- tainable will require diverse sources icy priorities when improving the dyna- of financing:when governments make mism of social assistance cash transfers. risk-pooling structures more widely avail- First is improving the “adaptability” of able to cover shocks with uncertain and these programs, that is, their capacity to catastrophic losses, it is reasonable to ex- respond to shocks (such as economic or pect resources contributed by people and natural disasters), including by establish- firms to meet the needs of more-foresee- ing comprehensive and dynamic social able and less-costly shocks. Today, most registries that are usable across social active labor measures are financed from programs. On the basis of a rare quasi- general budget expenditures, spreading natural experiment, Gerard, Naritomi, already small allocations too thin. and Silva (2021) show that expanding welfare programs has aggregate benefits 4. Supporting working people through for the entire local economy in addi- periods of change by enhancing their tion to individual-level benefits. Second skills. This effort involves strengthen- is moving from budgeted programs to ing technical education and vocational protection guarantees, thus transforming training, expanding short-cycle higher these programs into safety nets that can education programs to reach low-income expand to catch all those vulnerable to students, and conditioning the funding impoverishment before they become poor for such programs upon participants’ (Packard et al. 2019). Third is preventing employability. Overview 15

A more complete characterization of the nature—and the impact on people—of a sys- policy areas that can be focused on in order temic shock from transitory to long-term. to achieve stronger social protection and The policy implication of this study’s labor responses to crises (policy dimension 2) findings and the related literature is that is given in figure 1.4. Evidence from multi- even if macroeconomic and social protec- ple contexts shows that each of these prior- tion and labor policies are pristine and flaw- ity areas can make a real difference in labor lessly implemented, better outcomes could market adjustment. still be achieved for workers in crises by complementing these policies with sectoral and place-based policies to deal with any Sectors and places: Tackling structural structural issues impeding strong recoveries issues from crises. This study shows how factors beyond the labor Such policies would address inefficiencies market affect the magnitude of the impacts of in labor market adjustment resulting from crises on workers. Structural challenges in the labor market legislation, product markets’ LAC region act to slow and even to prevent structures, insufficient geographic mobility, necessary labor market adjustment, hence and localized economic depression. Address- weakening economic recoveries and causing these structural challenges will require ing lasting effects on efficiency as described changes in legal frameworks and regulations above. These structural issues can change the as well as targeted public investment. A more

FIGURE 1.4 Addressing crises’ impacts and preparing workers for change: Policy reforms

WORKERS

s k Income support r o Active labor market programs

w e

SHOCK m a r f

o Cushion the short-term impact of shocks on workers

o • Augment unemployment income through the creation or redesign of c

e unemployment insurance

o r

c • Improve the capacity to assist with social assistance programs

+ Beyond short-term income support

r • O er robust and coordinated employment services to get workers back to

i the workforce quickly l

b Support working people for change: Enhance their skills a •

t S

Source: World Bank. 16 employment in Crisis

FIGURE 1.5 Tackling structural issues that worsen the impacts of crises on workers

Regional policies • Local investment and infrastructure development to promote local employment opportunities • Place-based policies to address the lack of spatial mobility and maximize the relocation potential

Competition policies

s • Address protectionism and unfair market conditions k r through better competition laws, fewer subsidies, o less state participation, and stronger procurement SHOCK w e practices

f Labor regulations WORKERS

c i • Reduce the pockets of labor rigidity (fewer

o restrictions on HR decisions) to speed up

o adjustments and shorten transitions

s Competition policies

e Local investment + place-based policies

l Labor regulations i

t S

Source: World Bank. Note: HR = human resource.

complete characterization of the policy areas Brazil—greatly mitigated the impact and, in to focus on in order to tackle the structural fact, is believed to have marginally reduced issues that worsen the impacts of crises on overall levels of poverty and inequality. workers (policy dimension 3) is given in (Diaz-Bonilla, Moreno Hererra, and Sanchez figure 1.5. Castro 2020). In this crisis, projected to be the most severe labor market recession in Implications for the COVID-19 some countries’ history, millions of workers crisis in Latin America and the Caribbean have lost their jobs, and millions more have seen sig- The COVID-19 pandemic is a convulsive, nificant reductions in their earnings. And this catastrophic crisis that is exacting a savage loss is not expected to be evenly shared across toll on labor markets in the LAC region. the income distribution, with some workers The region is experiencing an extraordi- (such as those who cannot work remotely) nary rate of employment destruction and impacted more than others. (Diaz-Bonilla, massive negative income shocks. While the Moreno Hererra, and Sanchez Castro 2020). predictions for 2020 were dire due to the Although this crisis—which was triggered widespread job loss experienced across by the public health imperatives of mitigating the region, the significant increase in social a global pandemic—is exceptional, it is also spending across the region—and especially in one more in a long series of aggregate demand Overview 17 shocks that have hit the countries of the LAC Notes region. On one hand, the COVID-19 crisis has several distinctive features. First, the 1. Following Vegh and Vuletin (2014), this paper defines a crisis as beginning in the quarter in lockdown has been bad for many workers which real gross domestic product (GDP) falls and worse for those for whom home-based below the preceding four-quarter moving aver- work (or access to high-quality internet) is age and as ending in the quarter in which real not an option. Second, prolonged uncertainty GDP returns to the precrisis level. This statis- about the crisis’s duration and outcome, partic was calculated using quarterly GDP series ticularly with regard to how employment will from International Financial Statistics and rebound, has delayed investment. And third, Haver Analytics. Although the specific number some countries have made stronger policy of crises that took place over this time period responses to this crisis than ever before. is sensitive to the definition of crisis used, inde- On the other hand, this crisis is not so pendently of that definition, this result shows different from previous ones. A large part of that unlike in developed countries, where output evolution is characterized by smooth its effects on the LAC region derives from the cycles, Latin America experiences frequent and global recession it has caused, a sharp fall in pronounced crises. demand for many months, and impending 2. There is an extensive literature on the dynamics financial crises in some countries. The LAC of labor market adjustment to technological region has a history of frequent and often change and international trade. See, for exam- severe economic slowdowns. What happens ple, Acemoglu and Restrepo (2017); Autor to workers during these slowdowns is deter- et al. (2014); Autor, Dorn, and Hanson (2015); mined largely by aggregate demand fluc- Dauth, Findeisen, and Suedekum (2017); tuations (plus some self-inflicted domestic Dix-Carneiro and Kovak (2017, 2019); and crises or mismanagement). Utar (2018). This deep crisis arrived just as many 3. Unique features of the COVID-19 crisis include lockdowns and the health risks asso- governments in the LAC region were grap- ciated with personal contact, which had det- pling with known structural challenges. rimental effects on many jobs, particularly It has accelerated some long-run structural informal jobs, and led to worse job losses in shifts that have been changing the nature of industries for which home-based work is not work, magnifying their potential to reduce an option. Another key feature of this crisis employment opportunities in what were is the prolonged uncertainty it has generated, traditionally considered “good jobs”— which has delayed investment and hiring, cast- the standard, stable, protected employment ing additional doubt on how employment is associated with the formal sector (Beylis going to rebound. et al. 2020). 4. These statistics were estimated by the authors The aforementioned employment dynam- on the basis of Socio-Economic Database for Latin America and the Caribbean data ics observed in many LAC countries will lead (World Bank and Center for Distributive, to sizable labor scarring effects stemming Labor and Social Studies). from the COVID-19 crisis. Sector and loca- 5. Although it is not a focus of this study, tion characteristics are likely to further mag- migration is an additional margin of nify these effects for some workers. However, quantitative adjustment in employment. This the three-dimensional policy framework factor is relevant across the LAC region—for proposed in this study provides a roadmap example, in the large flows of migrants from that could lead to a more resilient recovery. the República Bolivariana de Venezuela to How public and business policies address the neighboring countries, mainly Colombia. The current challenges will shape the progress Caribbean in particular is known for high of the economies of the LAC region and the unemployment and high migrant outflows; displaced workers often leave Caribbean well-being of the region’s workers and citi- countries in response to employment loss zens for decades. The challenge is immense, when there is slack in those countries’ labor but now is the defining time to take it on. markets. 18 employment in Crisis

6. Such studies include Elsby, Hobijn, and Sahin Arias-Vázquez, A., D. Lederman, and L. Venturi. (2013) and Shimer (2005). 2019. “Transitions of Workers Displaced Due 7. Monetary and fiscal stabilization policies are to Firm Closure.” Unpublished paper. powerful tools to respond to crises. Although Artuc, E., P. Bastos, and E. Lee. 2021. “Trade these policies are crucial to crisis mitigation, Shocks, Labor Mobility, and Welfare: Evi- they are not the main focus of this study. dence from Brazil.” Background paper written 8. In contrast to the United States’ experience for this report. World Bank, Washington, DC. during the COVID-19 crisis, job retention (See also annex 1A for additional details on programs in several European countries have this background paper.) served millions of workers. The magnitude Autor, D. H., D. Dorn, and G. H. Hanson. 2015. of the human capital destruction (scarring “Untangling Trade and Technology: Evidence effects) averted by these programs depends from Local Labour Markets.” Economic Jour- on: (a) the estimated losses in human capital nal 125 (584): 621–46. that would have been caused by the period Autor, D. H., D. Dorn, G. H. Hanson, and J. Song. of unemployment or nonemployment; (b) the 2014. “Trade Adjustment: Worker-Level unemployment permanently averted—that Evidence.” Quarterly Journal of Economics is, the workers in these programs who would 129 (4): 1799–1860. otherwise have been fired (directly or indi- Beylis, G., R. Fattal-Jaef, R. Sinha, M. Morris, rectly via firm bankruptcy or closure for lack and A. Sebastian. 2020. Going Viral: COVID- of liquidity); and (c) the unemployment tem- 19 and the Accelerated Transformation of porarily averted—that is, the workers who Jobs in Latin America and the Caribbean. are supported now but will be fired after the World Bank Latin American and Caribbean end of the support period, or even before then, Studies. Washington, DC: World Bank. because of their firms’ bankruptcy. In terms of Brambilla, I., D. Lederman, and G. Porto. 2012. costs per worker, such programs make sense “Exports, Export Destinations, and Skills.” from the governments’ perspectives, because American Economic Review 102 (7): 3406–38. without these programs, unemployment Dauth, W., S. Findeisen, and J. Suedekum. 2017. insurance would have to be paid in full to “Trade and Manufacturing Jobs in Germany.” each laid-off worker. Three key choices when American Economic Review 107 (5): 337–42. implementing such programs are their size, De Ferranti, D., G. Perry, I. S. Gill, and L. their duration, and their coordination with Servén. 2000. Securing Our Future in a existing unemployment insurance and social Global Economy. World Bank Latin Ameri- assistance. They are adequate for temporary, can and Caribbean Studies. Washington, DC: short-lived shocks but not for extended cri- World Bank. ses. As a crisis continues, key trade-offs arise: Diaz-Bonilla, C., L. Moreno Herrera, D. Sanchez should the program continue to support all Castro. 2020. Projected 2020 Poverty Impacts workers or only some? If the program is tar- of the COVID-19 Global Crisis in Latin geted, how should it choose who to support America and the Caribbean. Washington, DC: and for how long? In case of longer spells of World Bank. nonemployment, should the program continue to support jobs or shift to supporting work- Dix-Carneiro, R., and B. K. Kovak. 2017. “Trade ers in case their jobs have been eliminated? Liberalization and Regional Dynamics.” Amer- These decisions are difficult, and a combina- ican Economic Review 107 (10): 2908–46. tion of instruments might be needed to avoid Dix-Carneiro, R., and B. K. Kovak. 2019. “Margins large increases in poverty and unemployment of Labor Market Adjustment to Trade.” Journal when job retention programs are abruptly of International Economics 117: 125– 42. terminated. Elsby, M. W., B. Hobijn, and A. Sahin. 2013. “Unemployment Dynamics in the OECD.” Review of Economics and Statistics 95 (2): 530–48. References Fernald, J. G., R. E. Hall, J. H. Stock, and M. W. Acemoglu, D., and P. Restrepo. 2017. “Secular Watson. 2017. “The Disappointing Recovery Stagnation? The Effect of Aging on Economic of Output after 2009.” Working Paper 23543, Growth in the Age of Automation.” American National Bureau of Economic Research, Cam- Economic Review 107 (5): 174–79. bridge, MA. Overview 19

Fernandes, A., and J. Silva. 2021. “Labor Market Regis, P., and J. Silva. 2021. “Employment Dynam- Adjustment to External Shocks: Evidence for ics: Timeline and Myths of Economic Recov- Workers and Firms in Brazil and Ecuador.” ery.” Background paper written for this report. Background paper written for this report. World World Bank, Washington, DC. (See also annex Bank, Washington, DC. (See also annex 1A for 1A for additional details on this background additional details on this background paper.) paper.) Fieler, A. C., M. Eslava, and D. Y. Xu. 2018. “Trade, Robertson, R. 2021. “The Change in Nature of Quality Upgrading, and Input Linkages: Theory Labor Market Adjustment in Latin America and Evidence from Colombia.” American Eco- and the Caribbean.” Background paper writ- nomic Review 108 (1): 109–46. ten for this report. World Bank, Washington, Gerard, F., J. Naritomi, and J. Silva. 2021. “The DC. (See also annex 1A for additional details Effects of Cash Transfers on Formal Labor on this background paper.) Markets: Evidence from Brazil.” Background Shimer, R. 2005. “The Cyclical Behavior of Equi- paper written for this report. World Bank, librium Unemployment and Vacancies.” Amer- Washington, DC. (See also annex 1A for addi- ican Economic Review 95 (1): 25–49. tional details on this background paper.) Silva, J., and L. Sousa. 2021. “Job Creation and Mion, G., R. Proenca, and J. Silva. 2020. “Trade, Destruction in Small and Large Firms in Bra- Skills, and Productivity.” Unpublished paper. zil and Ecuador.” Background paper written Moreira, S. 2018. “Firm Dynamics, Persistent for this report. World Bank, Washington, DC. Effects of Entry Conditions, and Business (See also annex 1A for additional details on Cycles.” Unpublished paper. this background paper.) Moreno, L., and L. Sousa. 2021. “Early Sousa, L. 2021. “Economic Shocks and Employ- Employment Conditions and Labor Scarring ment Adjustments in Latin America.” Back- in Latin America.” Background paper writ- ground paper written for this report. World ten for this report. World Bank, Washington, Bank, Washington, DC. (See also annex 1A DC. (See also annex 1A for additional details for additional details on this background on this background paper.) paper.) Packard, T., U. Gentilini, M. Grosh, P. O’Keefe, Utar, H. 2018. “Workers beneath the Floodgates: R. Palacios, D. Robalino, and I. Santos. 2019. Low-Wage Import Competition and Workers’ Protecting All: Risk Sharing for a Diverse and Adjustment.” Review of Economics and Sta- Diversifying World of Work. Washington, tistics 100 (4): 631–47. DC: World Bank. Vijil, M., V. Amorin, M. Dutz, and P. Olinto. Packard, T., and J. Onishi. 2021. “Social Insurance 2021. “The Distributional Effects of Trade and Labor Market Policies in Latin America Policy in Brazil.” Background paper written and the Margins of Adjustment to Shocks.” for this report. World Bank, Washington, DC. Background paper written for this report. World (See also annex 1A for additional details on Bank, Washington, DC. (See also annex 1A for this background paper.) additional details on this background paper.) Vegh, C. A., and G. Vuletin. 2014. “The Road Ramey, V. 2012. “Comment on Fiscal Policy in to Redemption: Policy Response to Crises in a Depressed Economy.” Brookings Papers on Latin America.” IMF Economic Review 62 Economic Activity 2012 (1): 279–90. (4): 526–68. 20 employment in Crisis

Annex 1A: Background papers written for this report

TABLE 1A.1 Background papers written for this report Country coverage, data, and Author and title Key research questions time frame

Dynamics of labor market adjustment Regis, P., and J. Silva. How large and long-lasting is the destruc- • Brazil, Chile, Ecuador, and Mexico “Employment Dynamics: Timeline tion of formal jobs? What is the timeline • Longitudinal administrative and Myths of Economic Recovery.” of the observed contraction in jobs after employer-employee data comple- a crisis? Do both formal and informal jobs mented by cross-sectional national shrink? accounts and household survey data • Since 1986 for Brazil, since 2006 for Chile and Ecuador, and since 2020 for Mexico Silva , J., and L. Sousa. “Job Creation How do job creation and job destruction • Brazil and Ecuador and Destruction in Small and vary across small and large employers in • Longitudinal administrative Large Firms in Brazil and Ecuador.” the formal sectors of Brazil and Ecuador? employer-employee data comple- What are the relative contributions of mented by cross-sectional national large and small firms to overall unemploy- accounts and household survey data ment flows? Do the adjustment mech- • Since 1986 for Brazil and since 2006 anisms vary across small and large firms for Ecuador and across the two countries? Sousa, L. “Economic Shocks and How do worker flows in Latin American • Argentina, Brazil, Chile, Ecuador, Employment Adjustments in Latin economies differ across the wage and Mexico, Paraguay, and Peru America.” household income distributions? How do • Labor force survey (panel data) these flows respond to economic shocks, • 2005 Q1–2017 Q4 including the global financial crisis of 2008–09? What types of labor transitions are more and less cyclical? How do labor market adjustments differ when respond- ing to cyclical shocks versus changes in growth trends? How does cyclicality vary across the income distribution?

Impact of crises on workers, firms, and places Fernandes , A., and J. Silva. “Labor How does exposure to foreign demand • Brazil and Ecuador Market Adjustment to External shocks affect workers’ employment and • Longitudinal matched Shocks: Evidence for Workers and wages? What are these shocks’ short-, employer-employee data Firms in Brazil and Ecuador.” medium-, and long-run effects on labor • 2004–17 market outcomes at the individual level? How were workers able to adjust to these negative shocks according to the characteristics of their sectors and local labor markets? Moreno, L., and L. Sousa. What is the impact of initial labor market • LAC 17 countries “Early Employment Conditions and conditions on the earnings and career • Labor force and household surveys Labor Scarring in Latin America.” trajectories of worker cohorts in Latin (cross-sectional) America? • Since 1980

(Continued on next page) Overview 21

TABLE 1A.1 Background papers written for this report (Continued) Author and title Key research questions Country coverage and type of data

Policy responses to crises Artuc, E., P. Bastos, and E. Lee. What are the effects of trade shocks on • Brazil “Trade Shocks, Labor Mobility, and labor mobility and welfare in Brazil? To • Longitudinal social security records Welfare: Evidence from Brazil.” what extent are workers mobile across • 1994–2015 sectors, places, and occupations? Beyond their effect on wages, how do trade shocks affect the quality of job matching? Gerard, F., J. Naritomi, and J. Silva. What are the effects of the expansion of • Brazil “The Effects of Cash Transfers on welfare programs on formal labor mar- • Administrative records on Bolsa Formal Labor Markets: Evidence kets? What are the expansion’s spillover Familia recipients and formal from Brazil.” effects to nonbeneficiaries? What is the workers multiplier effect of Bolsa Familia benefits? Packard, T., and J. Onishi. “Social How do social protection systems in Latin • All Latin American countries Insurance and Labor Market Poli- America affect the margins of adjustment • Administrative data cies in Latin America and the Mar- to shocks? gins of Adjustment to Shocks.” Robertson, R. “The Change in How has real wage flexibility evolved in • Brazil, Chile, Colombia, and Mexico Nature of Labor Market Adjust- the LAC region since the 1980s? Is the • Quarterly manufacturing data ment in Latin America and the documented margin of adjustment to (cross-section) Caribbean.” slowdowns and crises in the 2000s dif- • 1980–2017 ferent from that of the 1980s and 1990s? Do the two margins differ in relative importance? Vijil, M., V. Amorin, M. Dutz, and P. What is the within-country distributive • Brazil Olinto. “The Distributional Effects impact of competition policies? What • Household expenditure surveys, of Trade Policy in Brazil.” was the distribution of the welfare gains labor market surveys, and local con- caused by Brazil’s tariff liberalization in sumption price data the 1990s? • 1991–99 Note: The LAC 17 countries = Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Peru, Paraguay and Uruguay.

The Dynamics of Labor Market Adjustment 2

Introduction as the unemployment rate), it considers the mechanisms of adjustment and their cycli- Crises in Latin America and the Caribbean cality and heterogeneity across firms and (LAC) depress labor demand. How- workers. The chapter answers three ques- ever, different labor market dynam- tions: (a) What are the principal margins ics may hide behind similar reductions of adjustment of labor markets in Latin in labor demand. In the months after America? (b) What are the key mechanisms the COVID-19 (coronavirus) pandemic driving these adjustments? and (c) Do cri- reached the LAC region, unemployment ses affect the employment structure beyond grew by about 9.8 percentage points in their effects on job flows? Colombia, 7.6 percentage points in Costa Important to these questions is how firms Rica, 2.7 percentage points in Brazil, adjust their wage bill (wages times employ- 1.5 percentage points in Mexico, and ment) in response to crises. There are three 1.3 percentage points in Paraguay. More key dimensions of adjustment. First, firms than 65,000 formal jobs were lost in El Sal- can try to adjust wages. Second, firms can vador between March and May 2020, and adjust the hours worked by existing employ- more than 350,000 workers lost their jobs ees (the intensive margin). And third, firms in the Dominican Republic between March can adjust their number of employees (the 1 and June 2020. Do these statistics mean extensive margin). Several studies show that Colombia has been more affected by the that in the face of adverse shocks, firms COVID-19 crisis than the other countries? rarely adjust wages. Kaur (2019) finds sig- Not necessarily. The unemployment rate nificant downward rigidity of wages in on its own does not fully characterize the India. Significant downward rigidity of real impact of a crisis on labor markets. wages has also been documented in Mex- By complementing key research with ico (Castellanos, Garcia-Verdu, and Kaplan new results, this chapter seeks to further 2004), South Africa (Erten, Leight, and the understanding of how labor markets in Tregenna 2019), and many other countries. the LAC region adjust to economic crises. Downward real wage adjustment is most Rather than focusing on one statistic (such likely to occur through inflation, which has

23 24 employment in Crisis

been relatively low in Latin America over the informal employment, including in the gig past two decades. economy, as a temporary stopgap between The existing literature provides little con- formal jobs. sensus about whether, a priori, firms can be What does the heterogeneity of the infor- expected to react to a shock by adjusting mal sector mean for unemployment rates along the intensive margin or the extensive during a crisis? Economies with higher lev- margin. Empirically, both margins have been els of subsistence self-employment or of gig shown to be important. Van Rens (2012) workers could expect to see lower levels of argues that in the presence of negative shocks, unemployment in response to crises. If jobs the intensive margin (hours) is just as affected are lost in a formal sector with low or inad- as the extensive margin (employment) in equate unemployment insurance, the infor- Organisation for Economic Co-operation mal sector might end up absorbing workers and Development (OECD) countries. Taskin who would otherwise become unemployed (2013) finds a similar result for Turkey and (who would perhaps join Uber or start selling the United States, which is surprising because refreshments on the street). In this way, an Turkey has a much larger informal sector economy that might at first glance appear than the United States. In India, the lack of to be more sheltered from crises or to have downward real wage flexibility induces firms smoother adjustment mechanisms could sim- to reduce employment (Kaur 2019). ply have lower reservation wages because Institutions also affect firms’ adjustment of higher (and countercyclical) informality strategies. Regulations around firing are or the limited availability of unemployment associated with slower adjustment to crises benefits. in some Latin American countries (David, This chapter begins by considering the Pienknagura, and Roldos 2020) and in role of key potential margins of labor market Italy (Belloc and D’Antoni 2020), Japan adjustment across six countries in the LAC (Liu 2018), and other countries. This slow region through an analysis of labor flows and employment adjustment often comes at the transitions in the region. The estimates pro- cost of future hiring. Enforcement of these duced by Sousa (2021) consider four adjust- regulations helps make the formal sector ment mechanisms: unemployment, exiting more attractive to workers than the informal the labor force, shifting to informality, and sector (Abras et al. 2018), and this difference shifting to part-time work. The paper’s remains a critical component of Latin Ameri- results indicate that exits from the labor force can labor markets. and shifts to part-time work do not appear The level and composition of informal- to be significant margins of adjustment, but ity in the labor market is key to how crises formal and informal job flows have a strong impact workers in Latin American econo- negative correlation in five of the six countries mies. Informality may function as a buffer analyzed. That is, reductions in formality are for employment during bad times, because often accompanied by increases in informal- the costs of entering the informal sector ity, and vice versa. However, even in econo- are lower than the costs of entering formal mies with large informal sectors that absorb employment (Arias et al. 2018). However, this some excess labor, unemployment remains a buffer functions in a nuanced way, because significant margin of adjustment to economic the informal sector is highly heterogenous. shocks in the LAC region. Large gross flows Informal workers range from the subsistence to unemployment, in turn, represent signifi- self-employed, for whom unemployment is cant reductions in household income, increas- not an option, to informal dependent working vulnerability and poverty. Labor income ers, who may well lose their jobs in response represents 60 percent of household income to negative shocks, to relatively skilled work- among the poorest 40 percent of households ers more typical of the formal sector, who use in the LAC countries, and among households The Dynamics of Labor Market Adjustment 25

not living in poverty, a job loss by the main fluctuations look very similar for small and earner would push 55 percent into poverty.2 large firms. This chapter then digs into the mechanisms Are worker flows more cyclical for underlying the cyclicality of unemployment. low-skilled or informal workers than for Fluctuations in unemployment are deter- formal or high-skilled workers? In LAC mined by changes in transitions into and out economies, which feature a combination of of unemployment—job loss and job finding large informal sectors and varying skill levels rates, respectively. During economic down- among workers, there appears to be a hier- turns, these changes are driven by increased archy in adjustment costs, in which informal job destruction (as existing positions are workers, who have fewer job protections, eliminated), reduced job creation (as new face a greater likelihood of job loss regardless positions are not created), and lower levels of skill. Workers in the lower income quin- of job reallocation, or churning, as fewer tiles are more likely in general to experience workers voluntarily leave their positions to negative labor transitions than workers in look for better matches. Each of these factors the higher income quintiles, but overall, this will require different policy tools to address. project’s results suggest that high-skilled The relative contribution of each varies employment is more responsive to growth across labor markets, with some research in shocks than low-skilled employment. This high-income economies finding that cycli- finding is consistent with the greater cyclical unemployment is driven by reductions in cality of job loss for formal sector workers job finding rates and other studies (such as and employees of large firms, because these Shimer [2005] and Elsby, Hobijn, and Sahin workers are more likely to be skilled. [2013]) finding that it is driven instead by The final section of this chapter considers increased job separation rates. how the margins of employment adjustment New evidence produced in the context of affect the structure of the labor market. this research project shows that the cycli- Various labor market dynamics can lead cality of job loss across formal and infor- to changes in the composition of the work- mal employment is low. Instead, in most force, and the macroeconomic aftereffects countries analyzed, adjustment in employ- of a crisis on the employment structure can ment during the 2008–09 global financial shape its medium- to long-term effects on crisis was driven by a drop in net job finding employment and wages. Do such effects rates, which was larger for formal work- occur in Latin America? New research pro- ers than for informal workers (Sousa 2021). duced in the context of this project shows Zooming in on the formal sector through that crises do indeed have significant effects the use of worker-firm linked administrative on the employment structure and that these data from Brazil and Ecuador, another study effects can last for several years (Regis and showed that a reduction in job creation, Silva 2021). In the three countries analyzed rather than an increase in job destruction, is (Brazil, Chile, and Mexico), the shrinkage what drives lower employment in the formal of formal employment caused by crises was sector during downturns (Silva and Sousa strong and long-lasting. In two of these 2021). Moreover, although larger firms tend countries, informality seems to have been to be more productive and more resilient a long-term buffer for employment; in the than smaller firms, they also exhibit greater other one, employment stagnated or fell. cyclical fluctuations in labor demand. That At the same time, Artuc, Bastos, and Lee is, although large firms themselves are resil- (2021), considering the effects of crises on ient, employment at large firms may not be labor mobility and welfare, find that because the most resilient to economic shocks. Once of reduced job flows, job match quality falls the higher death rates of small firms during during slowdowns and crises, reducing esti- crises are taken into account, employment mated utility. These results suggest that a 26 employment in Crisis

crisis has the potential to push the labor mar- This section discusses both job flows and ket into a new equilibrium between formal employment transitions. While job flows and informal employment, with long-term measure the number of workers that switch implications for welfare and productivity. between two labor market states in a given period, employment transitions measure the Labor market flows: rates at which workers switch between these Unemployment versus states. Microdata are used to compute the informality cyclicality of labor transitions (in other words, the deviations from their natural rates or Economic crises are significant and frequent trends). The states of employment considered in the LAC region, and they represent a major are formal private wage employment, formal obstacle to the region’s economic develop- public employment, informal wage employment and poverty reduction. These economic ment, self-employment, unemployment, not shocks reduce overall demand, reducing working, and working part-time. Following demand for labor and eventually leading to Moscarini and Postel-Vinay (2012), cyclical- quantitative adjustments in employment. In ity is measured by constructing growth series this section, evidence is presented regarding (for which growth is measured as quarterly four margins of this quantitative adjustment: year-to-year GDP growth) and quarterly unemployment, exit from the labor force, labor transition series. Economic growth and shifting to informality, and shifting to part- transitions are detrended by applying sea- time work. sonal adjustments and a Hodrik-Prescott fil- The goal of this section, which is based ter to obtain the cyclical component of these on Sousa (2021), is to characterize the short- figures. A transition is procyclical when the term impacts of cyclical growth fluctuations correlation between the growth cycle and on the labor market by analyzing how labor the respective transition series is positive market flows vary with the business cycle. and countercyclical when the correlation is Sousa estimates worker flows using data on negative. more than six million labor market transitions constructed from labor force survey Unemployment panels from the Labor Database for Latin America and the Caribbean (LABLAC) In the LAC region, despite differences project (a joint project of the Center for Dis- among countries’ labor markets, unem- tributive, Labor and Social Studies and the ployment is strongly countercyclical in all World Bank). The chapter’s analysis focuses countries—a not-so-obvious result, given on urban workers, limiting the extent to the region’s level of informality. Figure 2.1 which the results are affected by differences plots quarterly fluctuations in GDP growth across countries in levels of subsistence or and unemployment rates since 2005 in low-productivity primary sector activities. six of the largest Latin American econo- The countries considered, those for which mies. In each of these economies, unem- sufficient data were available, are Argentina, ployment clearly spikes during significant Brazil, Chile, Ecuador, Mexico, and Peru. downturns (including the 2008–09 global The quarterly panels for each country were financial crisis). In more recent years, Bra- constructed by linking data from consecutive zil and Ecuador have experienced signifi- surveys between the first quarter of 2005 and cant downturns, both of which triggered the fourth quarter of 2017 across individ- large upticks in unemployment. For exam- uals between the ages of 15 and 64. These ple, between the fourth quarters of 2014 quarterly survey panels were used to calcu- and 2016, Brazil shed 2.6 million jobs, and late quarterly labor flows and transition rates the country’s unemployment rate increased from the individual-level data. from 6.5 percent to 12.0 percent. Although The Dynamics of Labor Market Adjustment 27

FIGURE 2.1 Quarterly fluctuations in unemployment and GDP growth, 2005–17

a. Argentina b. Brazil 15 1.5 6 2.5 10 1.0 4 2.0 5 1.5 0.5 2 1.0 0 0 0.0 0.5 –5 –2 –0.5 0.0 –10 –4 GDP growth rate –0.5 –1.0 Unemployment rate –15 –6 –1.0 –20 –1.5 –8 –1.5 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

c. Chile d. Ecuador 6 2.0 6 2.0 4 1.5 4 1.5 2 1.0 2 1.0 0 0.5 0 0.5 –2 0.0 –2 0.0 –4 –0.5 –4 –0.5 GDP growth rate –6 –1.0 –6 –1.0 Unemployment rate –8 –1.5 –8 –1.5 Q1 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q1 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

e. Mexico f. Peru 8 1.5 6 1.0 6 4 4 1.0 0.5 2 2 0.0 0 0.5 0 –2 –2 –0.5 –4 0.0 –4 –6 –1.0

GDP growth rate –6

–8 –0.5 –1.5 Unemployment rate –10 –8 –12 –1.0 –10 –2.0 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

GDP growth Unemployment (right axis)

Source: Sousa 2021. Note: This figure shows the cyclical components of unemployment and GDP growth. These calculations are based on unemployment rates and gross domestic product estimates from national authorities, seasonally adjusted and detrended using a Hodrik-Prescott filter. There is a break in the comparability of the Chilean unemployment series between 2009 and 2011. Because of a change in methodology, the Brazilian unemployment data series ends in 2015. GDP = gross domestic product. these spikes in unemployment quickly follow particularly slow: growth reentered posi- economic downturns, employment recovers tive territory in the first quarter of 2010 and more gradually than the overall economies. remained above trend until the end of 2012, Mexico’s employment recovery following but unemployment continued above trend the 2008–09 global financial crisis was until the end of 2011. 28 employment in Crisis

The spikes in unemployment during sig- from the labor force shows that these out- nificant downturns are obvious upon visual flows are not countercyclical in the six coun- inspection of figure 2.1. To measure the cycli- tries analyzed (table 2.2). Net outflows from cality of unemployment across the business the labor force are cyclical in Mexico and cycle, table 2.1 reports the correlation and Peru, but in both cases they are procyclical: ordinary least squares coefficients between net outflows from the labor force increase the two detrended series (Sousa 2021).3 Neg- with economic growth and fall during down- ative correlations reflect countercyclicality— turns. This result suggests that labor force that is, a decrease in growth associated with exit is not driven by workers’ discourage- an increase in unemployment, or vice versa. ment; rather, workers tend to hold on to their This countercyclicality is statistically signif- employment when jobs are harder to find, icant in five of the six countries for which perhaps to compensate for family members’ data are available. For example, in Mexico, weaker job-finding prospects. a decrease of 1 percentage point in detrended GDP growth is associated with an increase Informality in the detrended unemployment rate of 7.9 percent. In Peru, however, unemployment Not all labor market flows emerging from cri- does not have a strong cyclical component. A ses are job losses. In Latin America in partic- rule of thumb known as Okun’s Law states ular, the significant informal sector dampens that in the United States, each 1 percent the Okun’s Law relationship between GDP change in real output is associated with a growth and unemployment (David, Lambert, deviation from the natural rate of unemploy- and Toscani 2019). Fully characterizing the ment of about 0.5 percent, although this esti- quantitative margin of adjustment during mate may vary by the frequency with which slowdowns in these economies thus requires changes are measured (Aguiar-Conraria, going beyond unemployment dynamics. The Martins, and Soares 2020). region’s low access to unemployment insurance benefits likely lowers reservation wages and shortens job searches. Although these Exit from the labor force factors lower unemployment in the short A secondary indicator of slack labor demand term, they may also result in lower-quality is increased exit from the labor force, reflect- job matches.4 In the LAC region, where ing an increase in discouraged workers. Anal- the majority of jobs are informal— ysis of the cyclical component of net outflows self-employment (23 percent) or informal

TABLE 2.1 Cyclical components of GDP growth, the unemployment rate, and net flows out of the labor force Unemployment rate Flows out of labor force Countries Correlation OLS OLS Argentina −0.557 −0.061 *** −1391.3 Brazil −0.444 −0.098 ** −291.5 Chile −0.489 − 0.110 ** −1403.9 Ecuador −0.520 −0.159 *** −687.3 Mexico −0.533 −0.079 *** 12980.1 ** Peru −0.143 −0.035 2519.0 ** Source: Sousa 2021. Note: This table reports the correlation coefficients and the coefficients from simple OLS regressions of the detrended unemployment rates and detrended GDP growth rates, lagged by one quarter, of the six countries studied. GDP = gross domestic product; OLS = ordinary least squares. Significance level: * = 90 percent, ** = 95 percent, *** = 99 percent. The Dynamics of Labor Market Adjustment 29

dependent wage work (35 percent)5—a Figure 2.2 shows clear patterns across sev- reduction in positions in the formal sector eral of the countries analyzed in net flows into may lead to more workers entering the infor- formality and informality during economic mal sector.6 Is informality doing the dirty downturns. In 2008–09, as the effects of the work of maintaining employment in Latin global financial crisis reverberated across American economies? these countries, each experienced a steep loss Earlier research has shown that although of formal employment. This loss was accom- informal employment is not always inferior panied by an increase, although smaller in to formal employment in the LAC region, it magnitude, in informal and independent likely absorbs displaced formal sector work- work. The same pattern is seen during later ers during downturns. Using Mexican data, downturns in Argentina, Brazil, Chile, and Maloney (1999) presents one of the earliest Ecuador. During the recovery period after analyses of employment transitions between these downturns, the trends reverse: informal formal and informal jobs. He concludes that and independent work declines as net flows “the labor market for relatively unskilled into formality increase. workers may be well integrated with both Despite the clear reversal in net flows formal and informal sectors, offering desir- between formality and informality during able jobs with distinct characteristics.” crises and recovery periods, these flows are Bosch and Maloney (2008) find that the only weakly countercyclical in general— sectoral composition of employment is itself that is, there is only a weak correlation cyclical: informal employment is generally over the full business cycle between work- countercyclical, while the reverse is true for ers transitioning into formal and infor- formal employment. Additionally, Bosch and mal work and fluctuations in lagged GDP Maloney (2010), using survey panel data growth (table 2.2). However, this finding for Argentina, Brazil, and Mexico and con- has some exceptions. The net flows suggest tinuous time Markov transition processes, that in Argentina and Mexico, independent find evidence supporting voluntary entry work (self-employment) may be an inferior into informal employment, especially self- employment option used as a buffer during employment. Even so, when considering tran- low-growth periods. This observation is sition rates across the business cycle, they find supported by procyclical net flows from increased likelihood of transitioning from self-employment into unemployment in both formal to informal wage employment during countries, suggesting that formerly self-em- downturns, especially for young workers. ployed workers increasingly search for depen- Indeed, net worker flows across the six dent employment during periods of higher countries studied suggest that informal growth. A similar pattern is found among and independent work act like a buffer for informal wage workers in Ecuador. employment during economic downturns in In the relationship between growth and the LAC region. Figure 2.2 shows net flows, formality, Peru stands apart: the analysis in the difference between the number of new figure 2.2 suggests that the country expe- workers entering and exiting a sector in each riences higher flows into formality during quarter, for formal private sector jobs and downturns. The results in table 2.2 further informal or independent jobs. There is a support this observation. Peru’s net flows into strong negative correlation between formal formality and independent work are both and informal job flows in five of the six coun- cyclical; the country’s flows into formality are tries analyzed. That is, in these countries, countercyclical (flows into formality fall as reduced net flows into formality are often growth increases) and its flows into indepen- accompanied by increased net flows into dent work are procyclical (flows into indepen- informality and independent work, and vice dent work increase as growth increases). Net versa. flows from Peru’s formal sector to informal 30 employment in Crisis

FIGURE 2.2 Quarterly net flows into formal and informal employment, 2005–17

a. Argentina b. Brazil 0.90 0.50 0.70 0.30 0.50

0.30 0.10 (thousands) 0.10 –0.10 ers –0.10 Work –0.30 –0.30

–0.50 –0.50 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

c. Chile d. Ecuador 0.50 0.30

0.30 0.10

0.10 –0.10 (thousands) –0.10 ers

Work –0.30 –0.30

–0.50 –0.50 Q1 Q4 Q2 Q4 Q5 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q4 Q2 Q2Q4 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1 Q3Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

e. Mexico f. Peru 2.50 0.40 2.00 0.30 1.50 0.20 1.00 0.50 0.10 0.00 0.00 –0.50 –0.10 –1.00

Workers (thousands) –0.20 –1.50 –2.00 –0.30 –2.50 –0.40 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Downturn Formal salaried employment Informal salaried employment Self-employment

Source: Sousa 2021. Note: This figure’s calculations are based on the cyclical components of net flows of full-time jobs (new hires minus job losses) in the formal private sector and the informal dependent and independent (self-employed and employers) sectors. These flows are estimated on the basis of the number of workers experiencing employment changes between two consecutive quarters of observation. The cyclical component is estimated with seasonal adjustments and a Hodrik-Prescott filter. The shaded areas in the figures labeled “downturn” represent quarters of negative gross domestic product growth according to estimates from national authorities. The Dynamics of Labor Market Adjustment 31

TABLE 2.2 Cyclicality of net flows across sectors and out of employment, 2005–17 OLS coefficients of the cyclical components of net flows and lagged growth Net flows into Argentina Brazil Chile Ecuador Mexico Peru Note Formality 474.0 1153.3 160.5 30.4 888.7 −2639.7 *** Informality −382.3 −177.8 −697.4 −121.9 −2879.4 465.1 Informality (wages) −571.8 −435.9 −506.7 15.6 −5439.9 165.9 Independent work 163.7 −413.9 −943.4 −702.0 ** −2629.9 705.5 *

OLS coefficients of the cyclical components of net flows between employment sectors and lagged growth Quarter 1 Quarter 2 Argentina Brazil Chile Ecuador Mexico Peru Note Formal (private Informal −443.5 −102.0 −851.4 −46.9 −2660.1 1812.1 *** sector) Unemployment 5.7 −305.7 415.4 44.7 277.6 55.3 Out of labor force −101.9 −555.4 −12.5 −123.1 393.3 299.0 Informal (wage) Formal (private sector) 506.8 −95.4 −243.7 110.2 1759.8 −997. 3 ** Unemployment 175.6 298.8 304.5 681.6 ** 776.6 971.6 ** Out of labor force −252.22 373.3 448.2 −117.0 4413.8 −293.9 Independent Formal (private sector) −63.3 197.4 1095.1 −63.2 900.3 −814.8 ** work Unemployment 421.5 * 33.7 −595.9 −39.4 2158.3 267.8 Out of labor force −756.6 ** −100.9 −101.6 −94.3 6580.4 860.0 Source: Sousa 2021. Note: These calculations are based on the cyclical components of net flows (formal to informal work minus informal to formal work, and so forth) of full-time jobs. The sample analyzed is limited to workers who were in the formal private sector, in the informal wage sector, or independent (self-employed and employers) in the first quarter of observation. Flows are estimated as the number of workers who changed their employment status between two consecutive quarters of observation. The cyclical component of each flow is estimated with seasonal adjustments and a Hodrik-Prescott filter. OLS = ordinary least squares. Significance level: * = 90 percent, ** = 95 percent, *** = 99 percent. work are also procyclical, while the country’s (current firing costs and future hiring costs), flows from informality (both dependent and while preserving valuable firm-specific independent work) to formality are counter- human capital. However, employees whose cyclical. The question of why this might be hours are reduced will see their incomes fall the case for Peru (and not the other countries but not be able to tap into unemployment studied) may be worth exploring further in insurance (in the countries where this mech- future research. A key distinctive feature anism exists). Labor regulations restrict the of Peru compared with the other countries extent to which this option is feasible in the studied is Peru’s high share of self-employed LAC region’s formal sector.7 (These labor workers and low number of formal salaried regulations are discussed in more detail in employees (Jaramillo and Nopo 2020). chapter 3.) Even so, the option may be an additional margin of adjustment available to the informal sector and independent work- Adjusting hours worked ers. For example, rather than becoming fully A temporary reduction in hours could be unemployed, self-employed workers may an effective alternative to layoffs when a instead reduce their work hours in response firm faces a temporary decrease in demand. to lower demand for their services. By reducing the hours of employees rather Analysis of net flows into part-time work than letting them go, the firm can main- in the LAC region suggests that this option tain the employment links it has previously is not a significant margin of adjustment in established, reducing firm adjustment costs the labor market, either in general or in the 32 employment in Crisis

formal or informal sectors. Figure 2A.1 in What are the main margins of annex 2A shows the net flows into part-time adjustment in Latin America? work among workers who remain in the same The analysis above, which measures the sector (formal or informal), thus isolating the cyclicality of labor transitions across job adjustment of existing employment. In Argen- types, shows that despite the evidence of tina, Brazil, Mexico, and perhaps Peru’s for- informal employment serving as a buffer for mal sector, net flows into part-time work employment in the LAC region, unemploy- seem to spike in the early part of the 2008–09 ment in the region is strongly countercyclical global financial crisis, but these flows show (despite large differences in labor markets similar spikes outside crisis periods. Of the among countries). In contrast, exits from the net and gross flows into part-time and full- labor force or shifts to part-time work do not time work across the six countries and two appear to play a big role in Latin American sectors, only flows into part-time work in labor markets’ adjustment to crises. Impor- the informal sector of Ecuador are correlated tantly, the results presented in this section with the cyclical component of growth. until now reflect only the short-run adjust- Despite the differences in regulations ments to crises: unemployment is the main between the formal and informal sectors, in margin of adjustment during that period, four of the six countries analyzed, net flows but in the medium to long run, workers may into part-time work in the two sectors are transition from unemployment to informal- positively correlated: they move together, ity, as has been shown by Dix-Carneiro and suggesting similar patterns of fluctuations in Kovak (2019). job finding and churning rates. This correla- This section’s findings reflect significant tion is particularly strong in Mexico (with changes in Latin America’s flows to unem- a correlation coefficient of 0.59) and less ployment during the global financial cri- strong in Argentina (0.24). With a correlation sis, despite the relatively low estimates for coefficient of −0.23, Chile has the strongest these shifts given by Okun’s Law for Latin negative correlation between net flows into America.9 According to that rule of thumb, part-time work in the two sectors.8 the acceleration of GDP growth in the There is evidence that Argentina’s infor- region by 1 percentage point is associated mal sector makes some adjustments through with a contemporaneous (or 1-year lagged) the reduction of hours (panel a of figure 2.3). 0.2 percentage point reduction in the unem- At the beginning of the global financial cri- ployment rate. Existing data on the LAC sis, the number of part-time workers flowing region as a whole do not allow for a pre- into full-time work in Argentina fell below cise estimate of this elasticity when restrict- trend, while the number of full-time working the sample to crisis years. However, ers flowing into part-time work greatly sur- this estimate can be made for Brazil and passed the trend. At the end of the crisis, the Mexico, where we find an elasticity of about data reflected a short but large reversal: flows 0.5 during crises in the 2000s. In line with into full-time work grew significantly above this estimate, the latest projections of the trend. As shown in panel b of figure 2.3, impact of the COVID-19 pandemic on the adjustment to part-time status in the formal region predict a 9.1 percent drop in regional sector is not strongly correlated with growth. GDP and a rise of 4 to 5 percentage points However, transitions across full- and part- in the unemployment rate, which would time status are highly cyclical for indepen- correspond to a record 44 million workers dent workers. Transitions into full-time work unemployed. are strongly procyclical (increasing during The important role of the informal good times and decreasing during bad times), sector reflected in this section’s findings while transitions to part-time work are helps explain why, at the aggregate level, strongly countercyclical (increasing during the unemployment rate seems to be less bad times and decreasing during good times). The Dynamics of Labor Market Adjustment 33

FIGURE 2.3 Part-time work as a margin of adjustment in Argentina, 2005–15

a. Flows to part-time and full-time status b. Correlation between ows and GDP growth, cyclical components 0.50 Countercyclical Procyclical To full time 0.32 0.30 To part time –0.31 0.10 Self-employed To full time –0.10 –0.10 –0.36 Workers (thousands) To part time Worker ows

–0.30 Informal salaried

To full time –0.19 –0.50 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 To part time –0.03

Downturn Part–time Full–time Formal salaried –0.40 –0.20 0.00 0.20 0.40 Correlation coe cient

Source: Sousa 2021. Note: Panel a reports the cyclical component of flows from full-time to part-time jobs (labeled “Part-time”) and from part-time to full-time jobs (labeled “Full-time”) in the informal and independent (self-employed and employers) sector. Part-time work is defined as working fewer than 30 hours per week. The flows are estimated as the number of workers who change employment statuses between two consecutive quarters of observation. The cyclical components of these flows are estimated with seasonal adjustments and a Hodrik- Prescott filter. The shaded areas in the figures labeled “downturn” represent quarters of negative gross domestic product (GDP) growth according to estimates from national authorities. Panel b reports the correlation between each of the cyclical components of the quarterly labor flows and GDP growth lagged by one quarter. The correlations represented by colored bars are statistically significant, with confidence levels of 90 percent or higher. elastic to changes in output in Latin Amer- Job destruction and job creation ica than in advanced economies and why in times of crisis this elasticity is highly heterogeneous in the region. Although flows into unemployment As seen in the previous section, unemployment are an important margin of adjustment is a primary margin of labor market adjust- in the LAC region, flows into informality ment in the LAC region. The region’s high complement them as part of the adjust- informality rates and strong protections in the ment mechanism. This channel, informal formal sector imply large differences between employment, is significantly more limited the two types of jobs in employers’ adjustment in advanced economies. costs. Eliminating jobs in the formal sector can This section’s results are in line with recent be expensive because of the contractual and findings in the international trade literature. legal obligations of the employer, while job Dix-Carneiro and Kovak (2019) document destruction in the informal sector is relatively that the regions most exposed to foreign cost-free for employers, especially for jobs that competition following Brazil’s trade liberal- require low levels of firm- or industry-specific ization (in which the country suffered a nega- human capital. tive trade shock) faced medium-run increases This observation suggests that the destruc- in unemployment above the national aver- tion of informal jobs and the creation of age but lower than they would have faced formal jobs would be the margins of adjust- if informality did not absorb some workers ment most responsive to slowdowns. Indeed, displaced by trade. Bosch and Maloney (2008) find that countercyclical unemployment during recessions 34 employment in Crisis

in Brazil and Mexico is driven more by job unemployment and on how these dynamics separations among informal workers than differ across types of jobs and employers. by separations from formal employment. They also find decreased hiring in the formal Job loss and job finding rates during sector during downturns. Similarly, Bosch economic downturns and Esteban-Pretel (2012) find that cyclical variation in unemployment is explained As noted earlier in this chapter, the work- mostly by changes in the separation rate of er-based analogues for job creation and job informal workers, while cyclical variation in destruction are the concepts of job finding the share of formal employment is explained and job loss, respectively, which are based by changes in the transition rate from infor- on worker data rather than establishment mal to formal jobs. data. This alternate perspective allows for the Job flows can be decomposed into job inclusion of flows into and out of the informal creation flows and job destruction flows sector. Rates of job loss for full-time work- (Davis and Haltiwanger 1992). Measure- ers by sector of employment are presented ments of gross and net job flows in devel- in figure 2.4. As the global financial crisis oping countries are relatively rare (Ochieng rippled across the LAC countries, job losses and Park [2017] is a recent exception), so notably increased in most sectors of most the measurements presented in this section countries. As one might expect, the sector are especially valuable. For each firm ana- least affected was independent workers, who lyzed, the net change in jobs is calculated are largely low-skilled and self-employed. per quarter. Firms that lost more employees Table 2.3 reports the correlation coefficients than they gained are net job destroyers and between the cyclical components of each of contribute to job destruction flows. Con- these trends, showing strong correlations versely, firms that end the quarter with more in job loss between the two informal sec- employees than they began with are net job tors (informal wage work and independent creators and contribute to gross job creation work). The table also reveals a strong posi- flows. Because these flows are measured at tive correlation between formal and informal the establishment level, they are limited to salaried work in four of the countries ana- formal sector firm employment. For a wider lyzed—showing that job losses by dependent view of employment dynamics, this section workers exhibit similar responses during the also includes worker survey–based concepts business cycle in both sectors. of job finding flows and job loss flows. This Significantly, in most countries analyzed, alternative metric is particularly relevant in adjustment in employment during the 2008– the LAC region because it is the only way to 09 global financial crisis appears to have measure the informal sector. been driven by a drop in net job finding rates, Job separation rates are not necessarily which was larger for formal workers than the critical adjustment mechanism for labor for informal workers. This trend was partic- markets in Latin America: many workers ularly pronounced in Mexico and Peru. The who would otherwise voluntarily quit their net job finding rate of a sector—the worker jobs postpone that decision during down- side of net job creation—is calculated as the turns, while other employment opportunities number of new workers joining the sector in are scarce. This section addresses the fol- excess of the number of workers leaving the lowing questions: Which jobs are at greater sector in a particular quarter, as a share of risk during crises: those in the formal sector the sector’s employment. Table 2.2, in the or those in the informal sector? And within previous section, revealed the low cyclicality the formal sector, are jobs more at risk in of job loss across the three sectors—formal large or small firms? The section focuses private sector dependent work, informal on the underlying dynamics that determine dependent work, and independent work. The Dynamics of Labor Market Adjustment 35

FIGURE 2.4 Quarterly job loss, formal and informal sectors, 2005–17

a. Argentina b. Brazil 1.20 0.45 0.40 1.00 0.35 0.80 0.30 0.25 0.60 0.20 0.40 0.15 rkers (thousands) 0.10 Wo 0.20 0.05 0.00 0.00 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

c. Chile d. Ecuador 1.00 0.30 0.90 0.80 0.25 0.70 0.20 0.60 0.50 0.15 0.40 0.10 rkers (thousands) 0.30

Wo 0.20 0.05 0.10 0.00 0.00 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

e. Mexico f. Peru 3.50 0.30

3.00 0.25 2.50 0.20 2.00 0.15 1.50 0.10 rkers (thousands) 1.00 Wo 0.50 0.05 0.00 0.00 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Downturn Formal salaried employment Informal salaried employment Self-employed

Source: Sousa 2021. Note: This figure shows the quarterly numbers of workers experiencing job loss, defined as transitioning from employment to unemployment. The analysis is limited to full-time workers, and the trends are seasonally adjusted. The shaded areas labeled “downturn” represent quarters of negative gross domestic product growth according to estimates from national authorities. 36 employment in Crisis

TABLE 2.3 Correlation of job loss across sectors

Formal and Informal sector Country informal wage work Independent work Informal wage work Argentina −0.094 0.167 0.277 Brazil 0.594 0.458 0.566 Chile 0.428 −0.087 0.335 Ecuador 0.227 0.272 0.323 Mexico 0.402 0.022 0.355 Peru 0.136 0.519 0.082 Source: Sousa 2021. Note: This table reports the correlation coefficients between seasonally adjusted detrended quarterly job loss flows, by sector of employment. The analysis is limited to full-time workers.

Figure 2.5 shows the lower net job finding slowdown that hit commodity-exporting rates in the formal sector across the six coun- economies in South America beginning in tries throughout the business cycle. That is, 2015, gross job destruction surpassed gross as a share of the workforce, new job growth job creation, resulting in a net destruction of is lower in the formal sector than in the infor- formal employment. In other words, upon mal sector. entering a period of slower growth, employers first stop generating new positions. As the downturn continues or worsens, they then Job creation and job destruction during begin to reduce overall employment (through economic downturns layoffs, early retirements, or simply not fill- In the formal sector, decomposing the con- ing vacant positions).10 In contrast, during tributions of job creation and job destruc- periods of growth, gross job creation exceeds tion across small and large firms shows gross job destruction as existing firms grow that employment adjustments are driven by and new firms enter the labor market. lower job creation and higher volatility of But how do different types of employers employment in large firms. The employment adjust across the business cycle? Literature response to crises by small firms is more from the OECD and other high-income coun- muted. However, this difference between tries has found significant differences in how small and large firms goes away when firm small and large firms respond to crises, with births and deaths—which are substantially implications for the quality of employment more common among small firms—are for workers. Moscarini and Postel-Vinay taken into account. The following analysis is (2009, 2013) argue that less attractive firms done using administrative data sets from Bra- (those that pay less or provide lower quality zil and Ecuador on repeated cross sections employment, often smaller firms) are better and employer panels with full longitudinal able to retain good workers during down- information. These data span the past four turns because they face less competition for decades in Brazil and the past two decades in talent during those periods. This finding is Ecuador and include several business cycles. consistent with Moscarini and Postel-Vinay As shown in figure 2.6, a reduction (2012), who find that employment growth in gross job creation is what drives lower is more cyclical in large firms than in small employment in the formal sector during firms across several high-income economies, downturns in Brazil and Ecuador—job cre- a surprising result because small firms are ation falls faster and more significantly than more credit constrained. They also posit that job destruction. Notably, during the recent large firms can poach workers from smaller The Dynamics of Labor Market Adjustment 37

FIGURE 2.5 Quarterly net job finding rates, formal and informal sectors, 2005–17

a. Argentina b. Brazil 0.07 0.020 0.05 0.015 0.03 0.010 0.005 0.01 0.000

- nding rate –0.01 –0.005 –0.03 –0.010 Net job –0.05 –0.015 –0.07 –0.020 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

c. Chile d. Ecuador 0.04 0.05 0.03 0.04 0.02 0.03 0.01 0.02 0.000 0.01 - nding rate –0.01 0.000 –0.02 –0.01 Net job –0.03 –0.02 –0.04 –0.03 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

e. Mexico f. Peru 0.020 0.06 0.015 0.05 0.04 0.010 0.03 0.005 0.02

nding rate 0.000 0.01 0.000 –0.005

Net jo b- –0.01 –0.010 –0.02 –0.015 –0.03 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q1Q3 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Downturn Formal employment Informal employment

Source: Sousa 2021. Note: This figure shows the net job finding rates for the formal and informal sectors, calculated as the quarterly job finding rate (the flow out of unemployment into the sector) minus the quarterly job loss rate (the flow from the sector into unemployment) as a share of employment in that sector. The analysis is limited to full-time jobs, and the net job finding rate is seasonally adjusted. The shaded areas labeled “downturn” represent quarters of negative gross domestic product growth according to estimates from national authorities. 38 employment in Crisis

FIGURE 2.6 Gross job flows in Brazil and Ecuador, formal sector

a. Brazil b. Ecuador

2,000 100

80 1,500

1,000

Workers (thousands) 40

500 20 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Gross job creation Gross job destruction

Source: Silva and Sousa 2021. Note: This figure is based on a balanced panel of private sector firms. The shaded areas indicate economic downturns. Series are seasonally adjusted. In panel a, the data are from the first quarter of all years shown.

firms during high-growth periods, when magnitude of their effects differs, both types labor markets are tight. Haltiwanger et al. of firms follow broadly similar patterns in (2018) find that large firms are more sensi- terms of job creation, suffering large swings tive to unemployment levels than small firms across the business cycle. in the United States. Additionally, they find To assess the relative contributions of evidence of a “cyclical firm wage ladder” large and small firms to overall unemploy- wherein workers are more able to move to ment, these gross job flows are converted better employers (defined as firms that pay into job creation and job destruction rates— higher wages) during good times and less able calculated as total job creation or destruction during bad times. as a share of employment. Panels c and d of Applying a similar empirical approach to figure 2.7 show the net job creation rates for formal sector employment information from large and small firms in Brazil and Ecuador.11 Brazil and Ecuador, Silva and Sousa (2021) They show that large firms exhibit partic- differentiate job creation and job destruction ularly sharp peaks and valleys across the rates between large and small firms in the business cycle: that is, large firms are more formal sector. Panels a and b of figure 2.7 cyclical in employment than small firms. Even show the gross job creation and job destruc- taking into account their larger employment tion rates of large and small firms in each level, the fluctuations in the net employment country—in both countries, large firms of large firms are more responsive to demand increasingly account for more of gross job factors than those of small firms. flows. This trend is partially by construction, Finally, this report applies the methodol- because as large firms grow, they account for ogy of Moscarini and Postel-Vinay (2012) to an increasing share of total employment (large estimate the differential gross job creation firms account for 40 percent and 35 percent rate and differential gross job destruction of formal employment in Brazil and Ecua- rate between large and small firms. A pos- dor, respectively, and small firms account itive differential rate indicates that large for 30 percent and 32 percent). Although the firms have higher rates (of job creation or The Dynamics of Labor Market Adjustment 39

job destruction) than small firms, while employment in large firms, differential job a negative differential rate indicates that creation rates in both Brazil and Ecuador large employers have lower rates. In accor- show greater cyclicality in job creation for dance with the sharper peaks and valleys of large firms (figure 2.7, panels e and f).

FIGURE 2.7 Gross job flows and differential rates in large and small firms in the formal sector

Private sector gross job creation and job destruction flows in large and small firms a. Brazil b. Ecuador 200 1,200 60 15

50 1,000 150 10 40 800 100 30 600 5 50 20 400 0 10 0 Workers in large rms (thousands) Workers in large rms (thousands) Workers in small rms (thousands) Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Workers in small rms (thousands) 199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Gross job creation, large rms Gross job destruction, large rms Gross job creation, small rms (right axis) Gross job destruction, small rms (right axis)

Private sector net job creation rates in large and small firms c. Brazil d. Ecuador 1.0 2

0.5 1

0 0 Rate Rate −0.5 −1

−1.0 −2 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 19961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Large rm Small rm

Private sector net job creation rates in large and small firms e. Brazil f. Ecuador 0.4 2 0.2 0 0 Rate Rate −0.2 −2 −0.4 −0.6 −4 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Dierential gross job creation rate Dierential gross job destruction rate

Source: Silva and Sousa 2021. Note: This figure is based on a balanced panel of private sector firms. Small firm job flows are reported on the right axis in panels a and b. Firm size is defined each year: small firms are those with 20 or fewer employees, and large firms are those with more than 250 employees. The shaded areas indicate economic downturns. Series are seasonally adjusted. In panels a, c, and e, the data are from the first quarter of all years shown. 40 Employment in Crisis

However, the mechanisms of job loss dif- contradicted by the results above: the biggest fer across countries, as is illustrated by the job losses in the formal sector are in large differential job destruction rates in Brazil and firms, in both absolute and relative terms. Ecuador. In Brazil, differential job destruc- However, it is important to differentiate tion rates show significant variation. During between firm resilience and job security. The downturns they are positive, reflecting higher results above do not take into account firm job destruction rates in large firms than in births and deaths. Births and deaths are dis- small firms, and during recoveries they are proportionately likely to happen among small negative, suggesting the opposite. In Ecua- firms (in part by construction, because firms dor, however, the differential job destruction often begin small and may gradually shrink rate is stable around 0, implying that large before death). Once firm births and deaths and small firms destroy jobs at similar rates are taken into account, the difference in job across the business cycle. losses between large firms and small firms Taken together, these results show that becomes less noticeable (figure 2.8). The among formal firms, job losses are substan- lower resilience of small firms, exhibited by tially higher for large firms than for small their higher fluctuations in deaths and births, firms during downturns. In other words, in explains a significant part of their fluctua- Brazil and Ecuador, employment in large tions in net job creation during downturns. firms is more cyclical than employment in small firms. However, deconstructing job Job transitions across different types of flows into job creation and job destruc- workers tion shows that the mechanisms underlying changes in employment in small firms and Another source of complexity in job creation large firms differ across the two countries. and destruction trends is heterogeneity across In Brazil, large firms both create and destroy workers. Because more experienced workers jobs at higher rates than small firms. In Ecua- may be more productive and more expensive dor, the higher employment response among for firms to replace (Jovanovic 1979), losing large firms is driven by these firms’ more dra- workers with lower levels of human capital matic swings in job creation over the course (including firm-specific human capital) of the business cycle relative to small firms. would result in lower transaction costs for Large firms are thought to be more resil- employers. For example, Robertson and Dut- ient than small firms, and, partially for this kowsky (2002) find that adjustment costs reason, to be better employers that provide in the manufacturing sector in Mexico are more job security. This idea appears to be higher for nonproduction workers (who are

FIGURE 2.8 Net job creation rates in Brazil and Ecuador’s formal sectors

a. Brazil b. Ecuador 1.0 2

0.5 1

0.0 0

–0.5 −1

Net job creation rate –1.0 −2 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 Q1Q3 199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Large rm Small rm

Source: Silva and Sousa 2021. Note: This figure shows the net job creation rates in the private sector, including firm births and deaths. It is based on a balanced panel of private sector firms. The shaded areas indicate economic downturns. Series are seasonally adjusted. In panel a, the data are from the first quarter of all years shown. T he Dynamics of Labor Market Adjustment 41

typically higher-wage workers), workers with result holds for both men and women. None more job-specific training, and workers in of the job transitions analyzed was cyclical more unionized sectors. As a result, when a for low-skilled women in any of the six coun- firm experiences falling demand, lower-wage tries studied. production workers are laid off at higher rates than these higher-wage workers. A changing employment This finding implies that quantitative structure and the disappearance adjustment will tend to take place through of good jobs reductions in employment of workers with lower adjustment costs. In high-income econ- Crises translate into fewer job opportunities omies, low-earning workers exit the labor over time, beyond the business cycle (Artuc, force during downturns at disproportionate Bastos, and Lee 2021). Crises reduce worker rates (Carneiro, Guimarães, and Portugal churning, that is, job-to-job flows, leading to 2012; Solon, Barsky, and Parker 1994), a reduction in job match quality. The ratio- and displacement is higher for low-skilled nale for this effect is that when fewer job and young workers than for other types of opportunities are available, workers are less workers (Devereux 2004; Teulings 1993). able and less willing to leave their current Research in LAC countries suggests that jobs for new jobs. Using a structural model similar trends hold there. Studying the 2009 for Brazil, Artuc, Bastos and Lee (2021) find economic crisis in Mexico, Campos-Vazquez that adverse external shocks lower internal (2010) and Freije, López-Acevedo, and Rodrí- mobility across jobs, which in turn reduces guez-Oreggia (2011) both find larger job loss worker welfare over the life of the worker. rates for young and unskilled workers. Crises are also gradually shrinking the The LAC region’s combination of large number of good job opportunities in the LAC informal sectors and workers of varying skill region. This effect occurs because crises in the levels suggests there may be a hierarchy in region do not just shape worker flows tempo- adjustment costs, in which informal work- rarily; they also have significant after-crisis ers, who have fewer job protections, face a effects on the structure of employment that greater likelihood of job loss regardless of last for several years (Regis and Silva 2021). skill. Among formal workers, lower-income In many macroeconomic studies of crises, workers would be more likely to experi- the analysis focuses on short-term impacts, ence job loss than higher-income workers. such as negative deviations in employment Indeed, in five of the six countries analyzed or real wages in the short run (concomitant and across the business cycle, lower-wage or the following year). This focus is to be workers are more likely to enter a spell of expected, given that direct measures on the unemployment than higher-wage workers basis of aggregate data at the national level regardless of whether they are in the formal from past crisis episodes are generally not or informal sectors (figure 2.9). Mexico is the available. Quarterly data on employment in only exception: in that country, transitions LAC countries are available only from the into unemployment are higher in the middle late 1990s, and these data are not separated of the wage distribution than at the bottom into formal and informal employment. Long or the top. monthly time series for employment are even However, low-skilled workers’ increased more limited. An alternative strategy to mea- likelihood of transitioning into unemploy- sure longer-term effects would be to examine ment does not indicate that their employment high-quality administrative data combined is more vulnerable to fluctuations in eco- with national accounts. This was the path nomic growth. Instead, cyclical employment that Regis and Silva (2021) followed. transitions—those that are correlated with Regis and Silva (2021) investigate the lagged economic growth—are more common longer-term effects of crises on jobs by among skilled workers than among unskilled compiling time series of total, formal, and workers. (See table 2A.1 of annex 2A.) This informal employment dating back as far 42 Employment in Crisis

FIGURE 2.9 Quarterly share of workers entering unemployment per wage decile, formal and informal sectors, 2005–17

a. Argentina b. Brazil 0.09 0.09 0.08 0.08 0.07 0.07 0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03 Transition rate 0.02 0.02 0.01 0.01 0.00 0.00 123 4 567 8910 123 4 567 8910

c. Chile d. Ecuador 0.09 0.09 0.08 0.08 0.07 0.07 0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03 Transition rate 0.02 0.02 0.01 0.01 0.00 0.00 123 4 567 8910 123 4 567 8910

0.09 e. Mexico 0.09 f. Peru 0.08 0.08 0.07 0.07 0.06 0.06 0.05 0.05 0.04 0.04 0.03 0.03 Transition rate 0.02 0.02 0.01 0.01 0.00 0.00 1 2 3 4 567 8910 123 4 567 8910 Wage decile Wage decile Formal salaried employment Informal salaried employment Self-employment

Source: Sousa 2021. Note: The quarterly transition rate from employment to unemployment is defined as the share of workers employed in quarter t who transition into unemployment in quarter t+1.

as the 1980s for three countries: Brazil separation of each formal worker was used. (1985–2019), Chile (2006–19), and Mexico Because these data cover all formal workers, (1994–2019). Following Jorda (2005) and the team took the employment series available Jorda, Singh, and Taylor (2020), the study from the national accounts, harmonized it estimates impulse response functions of total, across time with their database, and inferred, formal, and informal employment to crises. through the difference between the two, the The authors created a new monthly linked countries’ total informal employment.12 In the employer-employee database using annual context of this study, crises were defined on administrative data from the social secu- the basis of quarterly GDP for each country, rity records of each country (which include which was standardized on the interval [0,1] administrative longitudinal data on the coun- (where 0 represents the deepest recession tries’ formal labor markets). To create these and 1 represents the largest expansion) and data, information on the months of hiring or used to define the economy’s long-run trend. The Dynamics of Labor Market AdjustmenT 43

GDP was below or above its long-run trend This study has three key results. First, crises as the cycle approached the values of 0 or 1, have caused multi-year reductions in employ- respectively. The business cycles thus defined ment in Brazil, Chile, and Mexico (figure 2.10). were then used to obtain a recession dummy. Second, in these three countries, formal Its duration corresponded to the time between employment is shrinking in a strong and last- the cycle’s peak and trough. ing manner. It takes LAC economies multiple

FIGURE 2.10 Impulse response functions, by type of employment, during the 30 months after the beginning of the recession

a. E ect on total employment b. E ect on formal employment c. E ect on informal employment BRAZIL 0.0 0 3

−0.5 −1 2 −2 −1.0 Percent −3 1 −1.5 −4 0 −2.0 −5 0102030 0102030 0 10 20 30

CHILE 0.0 0 8 –0.5 6 −2 −1.0 4

Percent −4 −1.5 2

−2.0 −6 0 0 10 20 30 0102030 0102030

MEXICO 0.0 0 1.5

–0.5 1.0 −2 −1.0 0.5

Percent −4 −1.5 0.0

−2.0 −6 –0.5 0102030 0102030 0102030

ALL COUNTRIES 0.0 0 4

−1 –0.5 3 −2 2

Percent –1.0 −3 1 –1.5 −4 0 0102030 0102030 0102030 Months since recession started Months since recession started Months since recession started

Source: Regis and Silva 2021. Note: Lags are measured in months after the beginning of a crisis. Shaded areas correspond to the 95 percent confidence intervals (1.96 standard-error-bands, lighter blue) and 68 percent confidence intervals (1 standard-error-bands, darker blue) around response estimates. 44 Employment in Crisis

years to recover from the contraction in for- strongly inelastic to changes in output: the mal employment induced by a c risis. In Brazil, accumulated change after one year is 10 or more than 30 months after the beginning of fewer basis points (IMF 2019). a recession, formal employment remained well below its initial level and— especially Conclusion c oncerning—showed little in the way of recovering. Overall employment remained lower for The results presented in this chapter show 20 months after the start of a recession, with that in general, and despite the presence of signs of movement toward recovery only in large informal sectors, unemployment is a Chile and Mexico; formal employment also significant margin of adjustment for labor remained lower for more than 20 months, markets in the LAC region during economic with a recovery trend only in Chile; and infor- downturns. Although job loss is present in mality remained higher, showing little in the both the formal and informal sectors, a signif- way of reversal in Brazil or Chile. These find- icant driver of unemployment is a slowdown ings suggest that exposure to sluggish labor in job creation in the formal sector. At the markets not only moves people temporarily same time, there is evidence that the infor- into informality but also leads to more funda- mal sector acts as a buffer for employment mental structural changes. After a severe crisis, during downturns, absorbing workers who employment may not recover to what it had might otherwise enter the formal sector or been before; the crisis may very well push the become unemployed. A third potential mar- labor market into a new equilibrium. Third, gin of adjustment, reduction in hours, does whereas in Brazil and Chile informality works not seem to be important in most countries in as a shock absorber in terms of employment either the formal or informal sectors. from the outset of a crisis, this is not the case Economic crises impact workers’ welfare in Mexico, where informal employment stag- not only through these margins of adjust- nates for about 20 months before it starts to ment but also by reducing the number of job increase. The delayed effect on informality opportunities available. Fewer opportuni- may be caused by formal workers looking for ties mean reduced churning, which in turn other formal work before eventually giving up means lower-quality job matches. These infe- and shifting to informality. Overall, after three rior matches reduce the productivity growth years, the average recession in Brazil, Chile, and lifetime earnings of workers, resulting in and Mexico causes a net loss of 1.5 million real welfare reductions. Also, the impact of a jobs, with a 3 percent contraction of formal crisis on the employment level can last long work and an expansion of the informal. The after the recession ends, and the crisis may current crisis could be even worse and cause a cause permanent adjustments to the structure contrac-tion in formal employment of up to 4 of employment in an economy. percent. This chapter exploited key sources of avail- This study’s findings help us to rationalize able data to provide empirical measures of the differences across the LAC countries in the labor market adjustments throughout the LAC lag between a change in output and its effect region. The first section of the analysis was on the unemployment rate. In Colombia, based on labor flows, using labor force survey unemployment exhibits a fast and elastic reac- data for six LAC countries (Argentina, Brazil, tion to output shocks: on average, the unem- Chile, Ecuador, Mexico, and Peru), while the ployment rate decreases by 45 basis points second and third sections featured analysis of following a 1 percent increase in output. In monthly linked employer-employee databases Brazil, unemployment is also sensitive to out- that were newly developed for this study using put shocks, but it reacts more slowly: on aver- information on the months of hiring or sepa- age, the unemployment rate decreases by 10 ration for individual workers in Brazil, Chile, basis points following a 1 percent increase in Ecuador, and Mexico. Although the adminis- output, but the accumulated change after one trative data from the first section provide rich year is around 40 basis points. Unemployment detail on worker flows and job destruction in in Argentina, Chile, and Peru, in contrast, is The Dynamics of Labor Market AdjustmenT 45

the formal sector, the new survey data paint FIGURE 2.11 Estimates of Okun’s Law for countries in the LAC a broader picture of the full labor market, region, 1991–2018 including the unemployed and the significant proportion of workers in the LAC region who 0.1 work informally. 0.0 Because of the large amounts of data –0.1 required for these analytics, many of the –0.2 smaller or more impoverished countries in the –0.3 LAC region cannot be included in this analysis. –0.4

However, estimates of Okun’s Law for a wider relationship –0.5 set of LAC countries show that the countries –0.6 included in this analysis run the gamut of –0.7 results in the region. The relationship between Estimated unemployment-in ation y r Chile Peru Bolivia Belize economic growth and unemployment rates Jamaica Mexico Panama Ecuador Urugua Honduras Paraguay varies widely across the region (figure 2.11). Costa Rica Argentina El SalvadoGuatemala The estimate of Okun’s Law is relatively high Dominican Republic Trinidad and Tobago in Bolivia (−0.63) but close to 0 in neighboring Paraguay, while Chile and Jamaica have Source: Authors’ calculations on the basis of IMF indicators. Note: Okun’s Law estimates represent the inverse relationship between inflation and unemploy- estimates comparable to that of the United ment. These estimates represent the slope of that relationship. States (−0.48) (Aguiar-Conraria, Martins, and Soares 2020; Ball, Leigh, and Loungani 2017). Although the lack of data prevents a detailed analysis, like Chile, Mexico, and Peru, can per- examination of the mechanisms of labor mar- haps shed some light on the adjustment mech- ket adjustment in each of the LAC countries, anisms in other countries in the region with the countries that are included in this chapter’s similar Okun’s Law values.

Notes 1. These statistics are according to the Insti- 6. Formal self-employment work is becoming tuto Salvadoreño de Seguro Social and the more common in some countries in the LAC Inter-American Development Bank Labor region, notably Brazil. However, this study Market Observatory. combines such workers with informal depen- 2. These statistics were computed by the authors dent workers when discussing informality, using SEDLAC (Socio-Economic Database for because the vast majority of formal self-em- Latin America and the Caribbean) data (World ployed workers in the region do not pay taxes Bank and CEDLAS). on their earnings, contribute to social security, 3. In this table and in subsequent figures, the or receive employment benefits. standard measure of the cyclicality of a vari- 7. As part of its response to the COVID-19 able is used, which is the unconditional cor- crisis, Brazil introduced a temporary wage relation of its deviations from the trend with subsidy program that allowed the reduction of a filtered measure of GDP. To detrend a series, hours for formal sector workers under certain we use a Hodrik-Prescott filter. conditions. 4. Indeed, research in high-income countries has 8. Brazil, Ecuador, and Peru show weaker found that job match quality is lower during correlations, with coefficients of 0.10, 0.15, downturns; that is, individuals take short- and −0.14, respectively. er-lasting and lower-paying jobs during such 9. Okun’s so-called Law gives the elasticity periods (Bowlus 1995), and employment real- between unemployment and output. This elas- locates to lower-paying industries and firms ticity is typically negative but less than one in (Moscarini and Postel-Vinay 2012). absolute terms, which means that labor mar- 5. These values are for 2018 and are based on kets do not fully adjust to every shock in cycli- LAC Equity Lab tabulations using SEDLAC cal output in the first year after it occurs. A data (World Bank and CEDLAS). sensible explanation for this partial adjustment 46 Employment in Crisis

is the expectation by employers that these in Developing Countries.” Policy Research cyclical changes are, by definition, temporary. Working Paper 4648, World Bank, Also, whereas this paper’s results on cyclicality Washington, DC. are derived from the analysis of individual-level Bosch, M., and W. F. Maloney. 2010. “Labor transitions, Okun’s Law simply compares the Dynamics in Developing Countries: m acro series of the unemployment rate with Comparative Analysis using Markov that of real GDP growth. Processes: An Application to Informality.” 10. Lower job creation can also result from fewer Labour Economics 17 (4): 621–31. firm births, and higher job destruction can Bowlus, A. J. 1995. “Matching Workers and also result from more firm deaths. Figure 2.5 Jobs: Cyclical Fluctuations in Match Quality.” is a balanced panel, so it does not include Journal of Labor Economics 13 (2): 335–50. these effects, but the patterns of gross job cre- Campos-Vazquez, R. 2010. “The Effects of ation and destruction are similar when using Macroeconomic Shocks on Employment: The unbalanced panels in both countries. Case of Mexico.” Estudios Economicos 25 (1): 11. N et job creation equals job creation minus job 177–246. destruction as a share of employment. Carneiro, A., P. Guimarães, and P. Portugal. 12. Although this study reports results from the 2012. “Real Wages and the Business Cycle: impulse response function to crises, it also cov- Accounting for Worker, Firm, and Job Title ers the effects of other large exogenous shocks, Heterogeneity.” American Economic Journal: including import growth among trading part- Macroeconomics 4 (2): 133–52. ners, changes in the terms of trade for com- Castellanos, S. G., R. Garcia-Verdu, and D. modities, and movements in the exchange rate. S. Kaplan. 2004. “Nominal Wage Rigidities in Mexico: Evidence from Social Security Records.” Journal of Development Economics References 75 (2): 507–33. Abras, A., R. K. Almeida, P. Carneiro, and C. David, A., F. Lambert, and F. G. Toscani. H. L. Corseuil. 2018. “Enforcement of Labor 2019. “More Work to Do? Taking Stock Regulations and Job Flows: Evidence from of Latin American Labor Markets.” IMF Brazilian Cities.” IZA Journal of Development Working Paper 19/55, International Monetary and Migration 8 (1). Fund. https://www.imf.org/en/Publications/ Aguiar-Conraria, L., M. M. F. Martins, and M. WP/Issues/2019/03/09/More-Work-to-Do- J. Soares. 2020. “Okun’s Law across Time and Taking-Stock-of-Latin-American-Labor- Frequencies.” Journal of Economic Dynamics Markets-46661. and Control 116. David, A. C., S. Pienknagura, and J. E. Roldos. Arias, J., E. Artuc, D. Lederman, and D. 2020. “Labor Market Dynamics, Informality Rojas. 2018. “Trade, Informal Employment and Regulations in Latin America.” IMF and Labor Adjustment Costs.” Journal of Working Paper 20/19, International Monetary Development Economics 133: 396–414. Fund. https://www.imf.org/en/Publications Artuc, E., P. Bastos, and E. Lee. 2021. “Trade /WP/Issues/2020/01/31/Labor-Market- Shocks, Labor Mobility, and Welfare: Evidence Dynamics-Informality-and-Regulations-in- from Brazil.” World Bank, Washington, DC. Latin-America-48893. Ball, L., D. Leigh, and P. Loungani. 2017. “Okun’s Davis, S., and J. Haltiwanger. 1992. “Gross Law: Fit at 50?” Journal of Money, Credit and Job Creation, Gross Job Destruction, and Banking 49 (7): 1413–41. Employment Reallocation.” Quarterly Journal Belloc, F., and M. D’Antoni. 2020. “The Elusive of Economics 107 (3): 819–63. Effect of Employment Protection on Labor Devereux, P. J. 2004. “Cyclical Quality Turnover.” Structural Change and Economic Adjustment in the Labor Market.” Southern Dynamics 54: 11–25. Economic Journal 70 (3): 600–15. Bosch, M., and J. Esteban-Pretel. 2012. “Job Dix-Carneiro, R., and B. K. Kovak. 2019. Creation and Job Destruction in the Presence “Margins of Labor Market Adjustment to of Informal Markets.” Journal of Development Trade.” Journal of International Economics Economics 98 (2): 270–86. 117: 125– 42. Bosch, M., and W. Maloney. 2008. “Cyclical Elsby, M. W., B. Hobijn, and A. Sahin. 2013. Movements in Unemployment and Informality “Unemployment Dynamics in the OECD.” T he Dynamics of Labor Market Adjustment 47

Review of Economics and Statistics 95 (2): Working Paper 14740, National Bureau of 530–48. Economic Research, Cambridge, MA. Erten, B., J. Leight, and F. Tregenna. 2019. Moscarini, G., and F. Postel-Vinay. 2012. “The “Trade Liberalization and Local Labor Market Contribution of Large and Small Employers Adjustment in South Africa.” Journal of to Job Creation in Times of High and Low International Economics 118: 448–67. Unemployment.” American Economic Review Freije, S., G. López-Acevedo, and E. Rodríguez- 102 (6): 2509–39. Oreggia. 2011. “Effects of the 2008–09 Moscarini, G., and F. Postel-Vinay. 2013. Economic Crisis on Labor Markets in “Stochastic Search Equilibrium.” Review of Mexico.” Policy Research Working Paper Economic Studies 80 (4): 1545–81. 5840, World Bank, Washington, DC. Ochieng, H. K., and B. Park. 2017. “The Haltiwanger, J. C., H. R. Hyatt, L. B. Kahn, and Heterogeneity of Job Creation and Destruction E. McEntarfer. 2018. “Cyclical Job Ladders in Transition and Non-Transition Developing by Firm Size and Firm Wage.” American Countries: The Effects of Firm Size, Age Economic Journal: Macroeconomics 10 (2): and Ownership.” East Asian Economic 52–85. Review 21 (4): 385–432. IMF (International Monetary Fund). 2019. Regis, P., and J. Silva. 2021. “Employment Regional Economic Outlook: Stunted Dynamics: Timeline and Myths of Economic by Uncertainty.https://www.imf.org/en/ Recovery.” World Bank, Washington, DC. Publications/REO/WH/Issues/2019/10/22/ Robertson, R., and D. H. Dutkowsky. 2002. wreo1019. “Labor Adjustment Costs in a Destination Jaramillo, M., and H. Nopo. 2020. “COVID- Country: The Case of Mexico.” Journal of 19 and External Shock: Economic Impacts Development Economics 67 (1): 29–54. and Policy Options.” COVID-19 Policy Shimer, R. 2005. “The Cyclical Behavior of Document 5, United Nations Development Equilibrium Unemployment and Vacancies.” Program in Latin America and the Caribbean, American Economic Review 95 (1): 25–49. New York. Silva, J., and L. Sousa. 2021. “Job Creation and Jorda, O. 2005. “Estimation and Inference of Destruction in Small and Large Firms in Brazil Impulse Responses by Local Projections.” and Ecuador.” World Bank, Washington, DC. American Economic Review 95 (1): 161–82. Solon, G., R. Barsky, and J. A. Parker. 1994. Jorda, O., S. R. Singh, and A. M. Taylor. “Measuring the Cyclicality of Real Wages: 2020. “Longer-Run Economic Consequences How Important Is Composition Bias?” of Pandemics.” Working Paper 26934, National Quarterly Journal of Economics 109 (1): 1–25. Bureau of Economic Research, Cambridge, MA. Sousa, L. 2021. “Economic Shocks and Jovanovic, B. 1979. “Job Matching and the Employment Adjustments in Latin America.” Theory of Turnover.” Journal of Political World Bank, Washington, DC. Economy 87 (5): 972–90. Taskin, T. 2013. “Intensive Margin and Extensive Kaur, S. 2019. “Nominal Wage Rigidity in Margin Adjustments of Labor Market: Turkey Village Labor Markets.” American Economic versus United States.” Economics Bulletin 33 Review 109 (10): 3585–3616. (3): 2307–19. Liu, Y. 2018. “Job Creation and Destruction Teulings, C. 1993. “The Diverging Effects of the in Japan: Evidence from Division-Level Business Cycle on the Expected Duration of Employment Data.” Journal of Asian Job Search.” Oxford Economic Papers 45: Economics 58: 59–71. 482–500. Maloney, W. F. 1999. “Does Informality Imply Van Rens, T. 2012. “How Important Is the Segmentation in Urban Labor Markets? Intensive Margin of Labor Adjustment? Evidence from Sectoral Transitions in Mexico.” Discussion of Aggregate Hours Worked in World Bank Economic Review 13 (2): OECD Countries: New Measurement and 275–302. Implications for Business Cycles by Lee Moscarini, G., and F. Postel-Vinay. 2009. “Large Ohanian and Andrea Raffo.” Journal of Employers Are More Cyclically Sensitive.” Monetary Economics 59 (1): 57–63. 48 employment in Crisis

Annex 2A: Additional analysis of employment transitions

FIGURE 2A.1 Quarterly net flows into part-time work, formal and informal sectors, 2005–17

a. Argentina b. Brazil 1.00 0.40 0.80 0.30 0.60 0.20 0.40 0.10 0.20 0.00 0.00 –0.10 –0.20 –0.20

Workers (thousands) –0.40 –0.60 –0.30 –0.80 –0.40 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

c. Chile d. Ecuador 0.30 0.20 0.15 0.10 0.10 0.05 –0.10 0.00 –0.05 –0.30 –0.10 Workers (thousands) –0.15 –0.50 –0.20 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

e. Mexico f. Peru 400 0.30 300 0.20 200 100 0.10 000 0.00 –100 –0.10 –200 –0.20

Workers (thousands) –300 –400 –0.30 –500 –0.40 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Downturn All workers Formal employment Informal employment

Source: Sousa 2021. Note: These panels show the cyclical components of net flows into part-time work (the number of workers moving from full-time to part-time work minus the number of workers moving from part-time to full-time work) among workers employed in formal dependent employment or informal employment (dependent or self-employed). This analysis is limited to workers who did not transition between formality and informality at the same time as they transitioned from part-time to full-time work and who did not change sectors. Flows are estimated as the number of workers who transitioned between two consecutive quarters of observation. The cyclical component is estimated with seasonal adjustments and a Hodrik-Prescott filter. The shaded areas labeled “downturn” represent quarters of negative gross domestic product growth according to estimates from national authorities. The Dynamics of Labor Market Adjustment 49

TABLE 2A.1 Cyclicality of employment transitions, by gender and skill level Women Men Country Initial state Final state High-skilled Low-skilled High-skilled Low-skilled Argentina Formal Informal 0.19 0.12 0.17 0.14 Formal Unemployed −0.04 −0.01 −0.06 −0.15 Informal Formal 0.09 0.23 0.45 0.93* Informal wage Formal 0.09 0.31 0.41 0.71 Informal wage Unemployed −0.34 −0.29 −0.69** 0.09 Self-employed Formal 0.08 −0.02 0.22 0.81* Self-employed Unemployed 0.40 0.17 −0.33 −0.41 Brazil Formal Informal 0.29 0.19 0.30 0.15 Formal Unemployed 0.02 0.05 −0.22* −0.20 Informal Formal 0.86** 0.80 0.60* 0.93* Informal wage Formal 0.94 0.93 0.28 0.94 Informal wage Unemployed −0.10 −0.41 −0.28 −0.04 Self-employed Formal 0.88** 0.80 0.78** 0.77 Self-employed Unemployed −0.01 0.09 −0.29* − 0.11 Chile Formal Informal 0.13 0.00 −0.72 −0.23 Formal Unemployed 0.22 0.58 −0.03 −0.13 Informal Formal 1.14 1.25 0.33 1.03 Informal wage Formal 0.27 0.78 0.53 1.15 Informal wage Unemployed −0.09 −0.65 0.69 0.44 Self-employed Formal 2.43 2.45 0.51 1.66 Self-employed Unemployed −0.53 −0.73 0.95 0.87 Ecuador Formal Informal 0.12 0.72 −1.25 0.29 Formal Unemployed −0.58 −0.79 0.14 0.16 Informal Formal 0.47 0.40 −0.33 −0.53 Informal wage Formal 0.62 0.08 −0.03 −0.82 Informal wage Unemployed 0.34 −0.93 0.33 −0.75 Self-employed Formal −0.09 −0.21 −0.20 −0.08 Self-employed Unemployed −0.27 0.10 0.08 −0.07 Mexico Formal Informal − 0.11 −0.44 0.22 −0.17 Formal Unemployed −0.38** 0.02 −0.19 0.07 Informal Formal 0.17 0.25 0.86*** 1.04** Informal wage Formal −0.06 0.28 0.72** 0.90 Informal wage Unemployed −0.10 −0.14 −0.47** −0.06 Self-employed Formal 0.29 0.86 0.59* 0.71 Self-employed Unemployed 0.00 −0.40 0.00 −0.34 table continues next page 50 employment in Crisis

TABLE 2A.1 Cyclicality of employment transitions, by gender and skill level (Continued) Women Men Country Initial state Final state High-skilled Low-skilled High-skilled Low-skilled Peru Formal Informal 1.71 2.83 2.65** 1.78 Formal Unemployed 0.67 0.80 0.09 −0.07 Informal Formal −0.48 −1.27 −1.26 −1.32 Informal wage Formal −1.27 −1.65 −1.18 −0.88 Informal wage Unemployed 0.86 0.98 1.37 0.29 Self-employed Formal 0.48 −0.23 −0.90 −0.86 Self-employed Unemployed 0.63 0.52 −0.55 0.06 Source: Sousa 2021. Note: This table shows the coefficients of an ordinary least squares regression of quarterly transition rates on lagged gross domestic product growth rate, by gender and skill group. “Low-skilled” workers have secondary education or less, and “high-skilled” workers have tertiary or higher education. Transition rates are defined as the share of workers with one type of employment in quarter t who transitioned into another type of employment in quarter t+1. Significance level: * = 90 percent, ** = 95 percent, *** = 99 percent. The Impact on Workers, Firms, and Places 3

Introduction If unemployment is persistent, the associated The previous chapter showed how crises in human capital decay will be greater and will Latin America and the Caribbean (LAC) lead to a larger decrease in long-term growth change aggregate employment dynamics and potential. Notwithstanding the size of a the employment structure. Crises lead to shock, if its effects are largely heterogeneous higher unemployment (more than they lead across workers, with some losing much more to increases in informality), with particularly than others, targeting scarce support toward prominent job losses in the formal sector. As the workers who lose the most may yield good job opportunities shrink, the overall larger gains. The stakes are very high for economic structure is altered. Job loss caused Latin America in terms of not only growth by crises is particularly painful in the LAC potential but also social stability; some recent region because of its sluggish recovery pro- studies have linked job displacement with ris- cesses. The region’s slow job creation depends ing violence (Dell, Feigenberg, and Teshima on demand-side factors, like firms and loca- 2019). Furthermore, the previous chapter tions, not just on workers. Although the evi- showed that quantitative adjustments to cri- dence presented thus far suggests that crises ses affect lower-skilled workers more than have detrimental impacts at the aggregate higher-skilled workers. Scarring can amplify level, how severe are their impacts on indi- this effect, further eroding the earnings of vidual workers? How do sectors and firms lower-skilled workers and increasing inequal- adjust employment and wages in response ity in an already highly unequal region. to crises? Which margins of adjustment are Crises can decrease individual welfare, used, beyond shedding jobs, and what are but they can also increase efficiency in their medium- to long-run effects on effi- the short and medium run. During a cri- ciency? And how do the characteristics of sis, employer-employee matches and the localities shape crisis impacts? job-specific human capital arising from These questions are important to the LAC them, which often take a long time to build region’s crisis response agenda, particularly and would regain viability when the economy because of their long-lasting implications. goes back to normal, may be permanently

51 52 employment in Crisis

dissolved because of the temporary shock. But to what extent do these results apply This loss may slow the ramping up of pro- to labor markets in the LAC region, where duction later on, and it implies a loss in pro- the share of college-educated workers is far ductivity. However, job loss spurred by an smaller and where informality continues to economic crisis can also have an important be a significant employment option? Moreno cleansing effect and lead to increased produc- and Sousa (2021) estimate the extent of the tivity at both the firm and market levels. It scarring caused by the initial labor conditions can be a good thing—provided that new jobs new entrants face in the first decade of their are created after the crisis is over. working lives in four Latin American econo- This chapter begins with a careful char- mies. Their findings confirm that entering the acterization of the scarring caused by invol- labor market during a crisis does have long- untary and exogenous job loss—that is, job term consequences in the LAC region. How- loss unrelated to the worker’s performance ever, this scarring is found in employment or preferences—by looking at the long- outcomes (lower participation rates, higher term wage losses of displaced workers fol- unemployment rates, and higher likelihood lowing firm closures. The literature on the of informality) rather than in a long-term LAC region finds large, short-run effects of effect on earnings, and it is most prominent such displacements; for example, Amarante, among workers with only secondary educa- Arim, and Dean (2014) find wage losses in tion. Similarly, Fernandes and Silva (2021) excess of 14 percent one year after job sep- find stronger scarring effects in employment aration for high-tenure Uruguayan work- and wage outcomes for lower-skilled workers ers. Arias-Vázquez, Lederman, and Venturi than for higher-skilled workers in the formal (2019) add to this growing literature, finding sector in Brazil and Ecuador. One explana- large and long-lasting wage effects of dis- tion for this effect is that competition is lower placement caused by firm closure. Two years for skilled jobs because of the relative scarcity after the closure of a plant, wages tend to be of college graduates in the LAC region. That 11 percent lower for displaced workers than is, the analysis suggests that scarring is likely for nondisplaced workers. Four years after exacerbating the region’s high level of wage the closure, the wage gap is 6 percent. Wages inequality. do not fully recover until nine years later. Switching gears to consider efficiency, this Next, this chapter considers the scarring chapter shows that there are three main ways caused by the initial conditions faced by in which LAC firms and sectors adjust to cri- new entrants into the labor force in the LAC ses that can alter their efficiency in the long region. It considers whether there are long- term. First, worker adjustment varies depend- term employment and wage consequences ing on workers’ employer or firm characteris- of entering the labor market during a down- tics: workers for larger, better-managed firms turn, generating what the popular press calls cope better with crisis effects (Fernandes and “a lost generation.” This question is particu- Silva 2021). This finding has implications larly salient for the LAC region, given its high for firms’ productivity and labor demand. In rates of youth unemployment and its invest- the LAC region, the adjustment mechanisms ments in increasing and improving educa- to crises include cleansing effects; scarring, tional outcomes at the secondary and tertiary which reflects a lack of opportunities; and levels. Are these investments in the region’s shocks to distortions that led to rents, with human capital stock undermined by frequent potential positive effects on efficiency in the crises? Previous research that has found evi- long run. dence of long-term effects of economic down- In addition, the results presented in this turns on labor market entrants has focused chapter lead to questions about how institu- on high-income economies.1 Studies of coun- tional and market factors external to work- tries such as Japan, Sweden, and the United ers affect scarring and, in general, long-run States find evidence of negative long-term prospects for job recovery. The chapter’s wage effects for new graduates. results show that employment in protected The Impact on Workers, Firms, and Places 53 firms, defined as those that face less com- Workers: A bigger toll on the petition, is less affected by crises than is unskilled employment in less-protected firms. In sectors where a few firms hold a large percent- The goal of this section is to improve under- age of the market, shocks do not lead to any standing of the long-term implications of cri- downward real wage adjustments. Instead, ses for workers in the LAC region. A primary they can lead to increases in employment, focus of the chapter is characterizing the inci- the opposite of what normal economic dence and magnitude of labor scarring in the mechanisms would bring about. By the same region. Scarring refers to the long-run effects logic, employment responds less to negative of job loss on a worker’s earnings through export shocks in state-owned firms than in the decay of the worker’s human capital and private-owned firms. changes in the worker’s quality of employ- The chapter closes by considering the ment. Since the lasting wage effects of job third piece of the triangle, places. Findings losses were first documented in the United indicate that workers in less formal locali- States by Jacobson, LaLonde, and Sullivan ties cope better in the wake of crises than (1993a, 1993b), studies from around the workers from other localities. The presence world have found that the wage effects of los- of a large informal sector may protect some ing a job are long-lasting. Decay of human workers against shocks. For instance, this capital and reallocation of workers and firms study finds smaller employment and wage across sectors are two important channels losses in response to crises for formal, pri- through which crises can have long-term vate sector workers who live in localities implications on the welfare and economic with higher rates of informality (Fernandes growth prospects of the LAC region. and Silva 2021). This finding suggests that This section answers two sets of ques- informality can be an important buffer for tions. First, how extensive is scarring in Latin employment in the medium to long run, America and the Caribbean, and what forms when workers may transition from unem- does it take? The section looks at scarring ployment to informality; such an effect was across three dimensions that can cause it: job shown by Dix-Carneiro and Kovak (2019) loss, worse initial conditions at workforce in the case of adjustment to trade liberal- entry, and crisis-driven adjustment. And sec- ization. Indeed, transitions from unemploy- ond, how does scarring vary across different ment to informality are two times more types of workers? likely in the Brazilian data than transitions Scarring implies a reduction in human cap- from unemployment to formality. ital and worker productivity, leading to worse Finally, findings indicate that work- employment outcomes and lower wages over ers in localities with more (alternative) job time. Human capital can be thought of as opportunities bounce back better from cri- taking two forms. General human capital ses. Losses in employment (and sometimes includes skills that are valuable in many wages) for formal workers are larger and sectors of the economy (such as general edu- longer-lasting in localities with larger pri- cation, literacy, and some computer skills). mary sectors, smaller service sectors, fewer Specific human capital is related to a specific large firms, and production highly concen- industry or firm and is generated through trated in the same sector where the workers employment experience and on-the-job train- were employed before the crisis (Fernandes ing that make a worker more productive in and Silva 2021). In such cases, these work- that firm or industry. Scarring results from ers’ persistent losses of earnings may reflect a decay of either (or both) of these types of the lack of opportunities for them in the human capital. When workers lose their rebound, not just scarring in the traditional jobs, they lose the firm-specific skills and sense of a persistent loss of human capital relationships they had learned and built in associated with a period of unemployment those jobs. Burdett, Carrillo-Tudela, and or lower-quality employment. Coles (2020) find that lost human capital is 54 employment in Crisis

the most important factor in determining the Jacobson, LaLonde, and Sullivan (1993a, costs of job loss to workers, but Carrington 1993b) show that US workers incur pro- and Fallick (2017) suggest that additional longed periods of lost wages after losing their research is necessary to assess the extent of jobs. They also demonstrate that a sample of this factor’s contribution. workers in Pennsylvania experienced losses However, scarring does not strictly require of approximately 25 percent of their predis- job loss. To the extent that on-the-job training placement earnings that lingered for five to and experience generate human capital, keep six years after the loss of their jobs (Jacob- existing human capital from eroding, and son, LaLonde, and Sullivan 1993a). Other signal a worker’s quality to other employers, studies find the same phenomenon in other the quality of workers’ early job matches countries.2 Longer or more frequent spells can have significant effects on their human of unemployment have larger negative wage capital accumulation and career trajectories. consequences (Arulampalam 2001; Gregg Workers who first enter labor markets in bad and Tominey 2005; Gregory and Jukes times have been shown to have lower earn- 2001). In addition, scarring from job loss can ings than otherwise similar workers who first last across generations. Oreopoulos, Page, enter labor markets in good times. and Stevens (2008) find that Canadian sons This section reviews the literature and whose fathers were displaced had annual reports the results of two new background earnings 9 percent lower than those of simi- papers, developed in the context of this flag- lar children whose fathers did not experience ship research project, on labor market scar- similar employment shocks. However, most ring in Latin America. Two of the papers of the research in this area has focused on use matched worker-firm data to look at developed countries. scarring following job loss and scarring One of the more arresting symptoms of caused by firms’ exposure to crises. The third economic crises is mass layoffs caused by firm uses labor force surveys to look at scarring closures. Looking at mass layoff events helps beyond those employed in the formal sector address the potential endogeneity of workers and consider how conditions at the time of who leave firms voluntarily, because the rea- entry into the labor market impact workers’ son for leaving a firm (even in the context of employment outcomes over the first decade a mass layoff) has a significant effect on sub- of their working lives. sequent earnings and employment (Flaaen, This section also examines whether dif- Shapiro, and Sorkin 2019). However, there ferent types of workers are affected in differ- are surprisingly few studies on scarring of ent ways by crises. Scarring-based losses will displaced workers following firm closures in not be incurred by every worker; rather, they Latin America. Amarante, Arim, and Dean will be concentrated among certain groups of (2014) and Kaplan, González, and Robert- workers. This section identifies some of those son (2007), studying Uruguay and Mexico, groups, who merit special attention from pol- respectively, are important exceptions. Amar- icy makers in order to reduce the economic ante, Arim, and Dean (2014) find wage losses and social costs of crises. Understanding the in excess of 14 percent one year after job sep- differences in responses to crises across types aration for high-tenure Uruguayan workers. of workers is important, because it allows The reduction in wages is even greater for governments to target support where it is workers who separated during downturns. needed most. In Mexico, Kaplan, González, and Robert- son (2007) use administrative matched firm- worker data to follow workers who left firms Severity of long-term effects during “mass layoffs”—events in which a Scarring caused by job loss large share of the firms’ employment leaves Losing a job has significant costs in the short the firms. They find large reductions in wages term and the long term. The earliest studies for these workers and greater reductions for on this topic focused on the United States. workers who separated during downturns. The Impact on Workers, Firms, and Places 55

Building on this approach, Arias-Vázquez, Austria and Kahn [2010] for the United Lederman, and Venturi (2019), in a back- States). Liu, Salvanes, and Sørensen (2016) ground analysis for this report, find that find that the mechanism behind these long- workers who were displaced from their jobs term negative effects for college-educated when their plants closed suffered large and workers is cyclical skill mismatch—that is, long-lasting wage losses. They use Mexican the dearth of job opportunities when these data from 2005 to 2017 (retrospective infor- workers enter the job market leads them to mation on employment at the worker level accept worse employment matches. The study from Mexico’s National Survey of Occupa- finds that the match quality of these workers’ tion and Employment) that allow them to first jobs explains most of their losses of long- identify job losses resulting from firm clo- term earnings from graduating in a recession. sures. Their results indicate that it took an There is scant research on the scar- average of 10 years for workers who were ring of new labor force entrants in the LAC displaced by a plant closing to recover their region. Cruces, Ham, and Viollaz (2012) wages. This period is significantly longer use pseudopanels and birth cohorts to find than the average recovery period for work- a strong but short-lived negative impact on ers who voluntarily quit (three years) or for wages for Brazilian workers exposed to down- workers who closed their own businesses turns early in their labor market experience. (four years). Figure 3.1 illustrates the mag- Martinoty (2016) finds that downturns lead to nitude and duration of the impact of a plant changes in the composition of youth employ- closing on the real wages of displaced work- ment (because some youth choose to remain ers compared with nondisplaced workers. in education longer). All education groups Initially, in their first two years after dis- who enter the labor market during a down- placement, wages are 11 percent lower for turn show evidence of long-term scarring. the displaced workers than for nondisplaced To build more evidence on the topic of workers. The wage gap narrows to 6 percent scarring in the LAC region, Moreno and after the fourth year following the plant closing, and the displaced workers’ wages recover completely only after ten years. FIGURE 3.1 Effect on wages of displacement caused by plant closings in Mexico Scarring caused by entering the labor market during a downturn 0.00 As mentioned earlier in this chapter, scarring –0.02 does not strictly require job loss. To the extent that on-the-job training and experience gener- –0.04 ate human capital, keep existing human capital –0.06 from eroding, and signal a worker’s quality to other employers, the quality of workers’ early –0.08 job matches can have significant effects on their human capital accumulation and career –0.10 trajectories. In this way, young workers may –0.12 be especially vulnerable to the impacts of cri- % di erence between log wages of displaced worker vs nondisplaced –0.14 ses on local labor markets, because entering 0123456789 the labor market is more difficult in bad times Year of plant Number of years after plant closing and doing so has long-term effects on earnings closing (Hardoy and Schone 2013). Workers who first enter labor markets in Source: Arias-Vázquez, Lederman, and Venturi 2019. Note: This figure plots the percentage difference between the log wages of displaced workers and bad times have been shown to have lower nondisplaced workers (i.e., the estimated coefficient on the lagged displacement variable) on the verti- earnings than otherwise similar workers cal axis against the year since job displacement on the horizontal axis (where 0 is the year in which the plant closed). The solid line plots the coefficients (the wage gap). The dotted lines plot the 95 percent who first enter labor markets in good times confidence intervals. Standard errors are clustered at the state level. All regressions control for years of (as shown by Brunner and Kuhn [2014] for education, gender, marital status, age, age squared, state, survey period, and industry fixed effects. 56 employment in Crisis

FIGURE 3.2 Unemployment rates by cohort, Argentina and Colombia

a. Argentina b. Colombia 30 35

25 30 25 20 20 15

Percent 15 10 10

5 5

0 0

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

2004 2006 2007 2008 2007 2008 2009 2010 2009 2010 2012 2014 2011 2012 2013 2014

Source: Moreno and Sousa 2021. Note: This figure reports national unemployment rates by year of entering the labor market. The stars on the lines for the 2010 and 2012 cohorts in Colombia show their unemployment rates four years after entry.

Sousa (2021) test for cohort effects on labor used to consider informality as an outcome, market outcomes related to the economic which is not possible when using administra- conditions during the year of entry into the tive records. labor force. Figure 3.2 shows the unemploy- Across the four countries with subna- ment rates for cohorts in their first years of tional series long enough to use for this labor market participation in Argentina and analysis—Argentina, Brazil, Colombia, and Colombia. Initial unemployment levels dif- Mexico—Moreno and Sousa (2021) find fered noticeably across these cohorts: Colom- that initial employment conditions can lead bian youth entering the labor market in 2010 to scarring, which is experienced as a combi- faced significantly higher unemployment nation of higher unemployment, lower labor rates than those entering in 2012. The figure market participation, and higher informality, also illustrates that this difference can con- depending on the country. However, their tinue for years after entry. The 2010 cohort results do not reflect significant scarring did not catch up to the 2012 cohort’s lower through wage effects; such effects are notice- unemployment rate until four years after able only in the first three years of labor mar- entering the labor market. ket participation and only for specific subsets Following the methodology of Genda, of workers (Brazilian women and low-skilled Kondo, and Ohta (2010), Moreno and Sousa Mexican men). Figure 3.3 presents the results (2021) exploit spatial and temporal vari- for low-skilled workers in Mexico. In Mex- ations in local unemployment rates at the ico, new entrants with only secondary educa- time a cohort enters the labor market. By tion who enter the labor market during worse using repeated cross-sections and subna- periods have lower participation rates and tional unemployment rates, the same cohort potentially higher rates of discouragement, as is observed over time and across different suggested by the combination of lower par- initial labor market conditions. Because this ticipation and a lower unemployment rate. methodology relies on survey data, it can be This result continues for these workers’ first The Impact on Workers, Firms, and Places 57

FIGURE 3.3 Employment and wage effects of higher local unemployment at labor market entry in Mexico

a. Women with only a secondary b. Men with only a secondary education education 1.05 2.0 1.05 2.0 1.04 1.5 1.04 1.5 1.03 1.03 1.0 1.0 1.02 1.02 1.01 0.5 1.01 0.5 1.00 0.0 1.00 0.0

Odds ratio 0.99 –0.5 0.99 –0.5 0.98 0.98 –1.0 % change in wages –1.0 0.97 0.97 0.96 –1.5 0.96 –1.5 0.95 –2.0 0.95 –2.0 1–34–6 7–9 10–12 1–34–6 7–9 10–12 Number of years in the labor market Number of years in the labor market

Active in the labor market Unemployed Active in the informal market Wages (right axis)

Source: Moreno and Sousa 2021. Note: The logistic odds ratio is reported for binary outcomes, and Heckman coefficients are reported for log wages. The solid bars (and solid dots) indicate that the corresponding coefficient is statistically significant. The standard errors are calculated by bootstrapping (rep = 50) and are clustered by year and region. The model used includes individual characteristics and fixed effects for region, cohort, and year. The open bars and portions of line without a solid dot are not statistically significant.

nine years in the labor market. Interestingly, Scarring passed through from despite a significant difference in labor force crisis-affected firms participation rates between men and women Most of the existing research on scarring in Mexico, this result holds for both groups. focuses on the overall effects of displacements The results for Mexico also show higher in both good and bad economic times. How- rates of informality in later years for work- ever, there are many reasons to suspect that ers who enter the labor market during a displacement during a crisis may not have the downturn. This same result is found in same effects as displacement during better Colombia and for some types of workers times. This distinction is particularly relevant in Brazil. Although, as discussed in the pre- to the LAC region, where crises are frequent. vious chapter, informality is not necessar- Davis and von Wachter (2011) find that earn- ily an inf erior employment option in the ings losses from job displacement are higher LAC region, this result echoes the theory in recessions than in expansions, and Amar- mentioned earlier that informality serves as ante, Arim, and Dean (2014) find larger wage a buffer for employment during economic reductions for workers who separated during crises. One role of that buffer is to absorb new downturns than for other workers. McCar- entrants into the labor market who might, thy and Wright (2018) find that Irish workers in better times, find formal employment. who lost their jobs during the global finan- By contrast, in Argentina, informality rates cial crisis of 2008–09 incurred much larger for men who enter the labor market during and long-lasting earnings losses than workers downturns are lower and unemployment who lost their jobs between 2005 and 2007. rates are higher, suggesting that informality Focusing on the US labor market, Carrington is not an effective buffer for this group. (1993); Farber (2003); Howland and Peterson 58 employment in Crisis

(1988); and Jacobson, LaLonde, and Sullivan the crisis. One possible reason for this dif- (1993b, chapter 6) suggest that labor market ference has to do with wage flexibility. Fig- conditions can affect the wages workers earn ure 3.4, panel b, shows that the real wages in their new jobs (if they find new jobs) after of more adversely affected workers in Brazil involuntarily leaving their previous jobs. And fell significantly after the crisis. This reduc- Kaplan, González, and Robertson (2007) tion in real wages meant that workers in show that scarring in Mexico is much worse more affected firms took pay cuts (either at during times of crisis than at other times. their original firm or at their new jobs). In One of the key challenges in identifying Ecuador, however, real wages did not fall for the effects of crises on worker scarring is that workers in more affected firms. These coun- examples are rare of firm-specific changes in tries have important differences in their labor demand ensuing from crises. Fernandes and regulations that could affect crises’ impacts Silva (2021), in a background paper written on workers. In particular, rigidity of wages for this report, exploit heterogeneity in firms’ implies that any adjustment in the labor exposure to the global financial crisis of market must occur through quantities (i.e., 2008–09 to measure the potential of a severe employment) rather than through prices (i.e., demand shock to generate long-term employ- wages), so employment adjustments in labor ment and wage effects on workers. Their markets with more rigid wages may be larger. strategy is based on the idea that firms were Other papers find long-lasting effects differentially exposed to the external shocks from other shocks as well. Increased com- caused by the crisis given the differences in petition from China, for example, has led their predetermined portfolios of export des- to long-lasting losses in employment and tinations. Using full-population longitudinal wages in high-income countries (for the employer-employee data from social security United States, Autor, Dorn, and Hanson records linked with firm export customs data [2013] and Autor et al. [2014]; for Denmark, by destination, for two LAC countries (Brazil Utar [2014]; and for Germany, Dauth, Find- and Ecuador), they compared the employment eisen, and Suedekum [2016] and Yi, Müller, outcomes of formal workers initially in firms and Stegmaier [2016]). The international that faced larger labor demand shifts with trade literature shows that the dynamics of those of workers in firms that faced smaller labor market adjustment to trade shocks are shifts over the eight years following the crisis. different in developed countries than they They find that the effect of the global are elsewhere (Autor et al. 2014; Dauth, financial crisis was stronger and longer- Findeisen, and Suedekum 2017; Dauth et al. lasting for workers initially in more nega- 2019; Utar 2018). For Brazil, Dix-Carneiro tively affected firms. Figure 3.4 shows the and Kovak (2017, 2019) focus on the adverse study’s results. The effects of the global effects of a large tariff liberalization reform. financial crisis on workers’ average number They find that a worker whose region ini- of months formally employed are shown in tially faced a 10-percentage-point-larger panel a, and the effects on workers’ average decline in tariffs compared with other regions real monthly wages are in panel b. Each point worked, on average, 9.9 fewer months in the in the figure represents the regression coeffi- formal sector between 1990 and 2010. This cient on the crisis for the relevant year. effect is large: those months represent 8 per- The response in Brazil differs from that cent of the average number of total months in Ecuador. The effect on months worked in worked in the formal sector during that Brazil is more negative at first but fades over 21-year period (125 months). Like Fernandes time (although never fully disappearing). In and Silva (2021), Dix-Carneiro and Kovak Ecuador, the opposite is true: the immediate (2017, 2019) find that average employment negative effect grows over time. Employment and wage outcomes fail to recover even a did not recover in Brazil or in Ecuador (nor decade and a half after an initial shock. It is did real wages in Brazil) even nine years after striking that the temporary shock studied by The Impact on Workers, Firms, and Places 59

FIGURE 3.4 Dynamic effects of the global financial crisis on workers

a. Average number of months employed b. Average real monthly wages BRAZIL 0.00 0.00

−0.02 −0.02

−0.04 −0.04 −0.06

−0.06 −0.08 rm-level demand shocks

Eects of global nancial crisis–induced −0.08 –0.10 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017

ECUADOR 0.000 0.02

−0.005 0.00

−0.010 −0.02

−0.015 −0.04 rm-level demand shocks

Eects of global nancial crisis–induced −0.020 −0.06 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017

Source: Fernandes and Silva 2021. Note: This figure shows the effects of global financial crisis–induced firm-level demand shocks (experienced from 2008 to 2009) on workers’ average number of months formally employed in the years 2009 to 2017 (panel a) and average real monthly wages from 2009 to 2017 (panel b). More negative estimates imply larger reductions in the respective outcome among workers employed at the time of the global financial crisis in firms facing larger external demand reductions (relative to those employed in less affected firms). The vertical lines represent 95 percent confidence intervals on the basis of robust standard errors clustered by firm. The samples include about 3 million worker-year observations for Brazil and about 800,000 worker-year observations for Ecuador.

Fernandes and Silva (2021) led to the type of within types of workers in the United States. effects normally associated with large, per- Workers who were initially similar experi- manent shocks such as trade liberalization. enced different employment and earnings outcomes after exposure to different crisis-related local shocks. Yagan (2019) finds Incidence of long-term effects that low earners initially bore more of the As shown in chapter 2, the job losses caused employment effects of those shocks. The lit- by crises are not evenly spread across types erature has also shown that the long-term of workers. In related recent research, impact on labor markets of Chi na’s export Yagan (2019) shows that localized shocks explosion was large and that the job losses it during the global financial crisis were asso- caused were concentrated among low-skilled ciated with rising (wage) earnings inequality workers in high-income countries (Autor 60 employment in Crisis

et al. 2014; Autor, Dorn, and Hanson 2013; each of these individuals from 2009 to 2017. Dauth, Findeisen, and Suedekum 2016; Utar Figure 3.5 summarizes their results. 2014; Yi, Müller, and Stegmaier 2016). The The results for Brazil indicate that only evidence on this phenomenon’s effects on less educated workers experienced strong women versus men and young workers versus employment and wage responses to the cri- old workers is more mixed. sis, whereas workers with higher education Job and worker characteristics— were not significantly affected (figure 3.6). especially the skill distribution, but also the Relative to older workers, younger work- occupational composition and labor force ers saw significantly smaller responses to participation rates—are different in the LAC the firm export shock in terms of months countries than in high-income countries. worked. Responses to the global financial Therefore, this section considers scarring crisis were similar across male and female across workers with different demographic workers and across workers with high and characteristics and levels of labor market low previous participation in formal labor experience. It presents new evidence, draw- markets. ing on two of the background papers for For Ecuador, the paper finds a strong this report, Fernandes and Silva (2021) and employment response to the global financial Moreno and Sousa (2021). Both papers dis- crisis only for less educated workers, whereas aggregate scarring across different types of workers with higher education were not workers, thus providing a closer look at how significantly affected (figure 3.6). Interest- scarring operates across these types. ingly, the wages of workers with low previous To the extent that low-skilled workers and participation in the formal sector showed a workers from low-income households are significant negative response to gross domes- more likely to experience quantitative adjust- tic product declines in the workers’ firms’ ments leading to unemployment or lower export destinations, but there was no such quality employment, scarring can have long- response for workers with high previous par- term implications for wage inequality. If ticipation in the formal sector. Employment slowdowns imply disproportionately lower responses to the global financial crisis were access to on-the-job training and employ- similar across male and female workers, ment experience for low-skilled workers, workers in different age groups, and workers their human capital accumulation will suffer with different levels of previous participation more than that of high-skilled workers. This in the formal sector. Taken as a whole, these effect could lead to lower wage growth for findings suggest that less skilled workers in low-skilled workers, exacerbating the wage Brazil and Ecuador and older workers in inequality between the two groups. The Brazil were most severely hurt by the global results presented below suggest that scarring financial crisis. may contribute to increasing inequality in the Turning back to the scarring of new entrants already highly unequal LAC region. Low- on the basis of initial labor market conditions, skilled workers, at any given moment, are Moreno and Sousa (2021) find that workers more likely to experience not just job loss and with only secondary edu cation experienced higher unemployment but also longer-term the most negative long-term effects on their impacts from shocks. employment outcomes because of initial labor What do this research project’s findings market conditions (table 3.1). Although the suggest? Fernandes and Silva (2021) estimate specific mechanisms that cause scarring and the differences in changes in employment the length of the scarring may vary across and wages following the global financial cri- men and women, the overall story is similar. sis–induced drop in demand for individuals For both men and women, scarring occurs for working for Brazilian and Ecuadorean firms those with lower levels of schooling, is unlikely across occupation and demographic groups. for those who are college-educated, and is far They follow the employment outcomes of The Impact on Workers, Firms, and Places 61

FIGURE 3.5 Heterogeneity in effects of the global financial crisis across workers

a. Average number of months employed b. Average real monthly wages

BRAZIL

Overall

Men Women

No higher education ** * Higher education

Ages 16–34 Ages 35–48 ** Ages 49–65

Low labor force attachment High labor force attachment

–0.05 –0.03 –0.01 0.00 –0.02 –0.01 0.00

ECUADOR

Overall

Men Women

No higher education ** ** Higher education

Ages 16–34 Ages 35–48 Ages 49–65

Low labor force attachment ** High labor force attachment

–0.08 –0.03 0.00 0.02 –0.10 –0.05 0.00 0.05 E ects of global nancial crisis– E ects of global nancial crisis– induced rm-level demand shocks induced rm-level demand shocks

Source: Fernandes and Silva 2021. Note: This figure shows the effects of global financial crisis–induced firm-level demand shocks (experienced from 2008 to 2009) on Brazilian and Ecu adorian workers’ average number of months formally employed from 2009 to 2017 (panel a) and average real monthly wages from 2009 to 2017 (panel b). More negative estimates imply larger reductions in the respective outcome among workers employed at the time of the global financial crisis in firms facing larger external demand reductions (relative to those employed in less affected firms). Each bar shows the coefficient from a regression for the subsample on the basis of the worker characteristic listed on the y-axis in each subsample. The full samples have about 3 million worker-year observations for Brazil and about 800,000 worker-year observations for Ecuador. ***, **, and * indicate the significance levels, 1 percent, 5 percent, and 10 percent, respectively, of the t-test that indicates whether the difference in coefficients across worker categories is significant. 62 Employment in Crisis

FIGURE 3.6 Dynamic effects of the global financial crisis on workers by skill

a. Average number of months employed b. Average real monthly wages BRAZIL 0.000 0.000

−0.020 −0.005

−0.040 −0.010 Eects of global nancial crisis– induced rm-level demand shocks −0.060 −0.015 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017

ECUADOR 0.020 0.040

0.000 0.020

−0.020 0.000

−0.040 −0.020 −0.060 Eects of global nancial crisis–

induced rm-level demand shocks −0.040 −0.080 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017 Higher education No higher education

Source: Fernandes and Silva 2020. Note: The figure shows the effects of global financial crisis–induced firm-level demand shocks (2008–09) on Brazilian and Ecuadorian workers’ with and without higher education in terms of average number of months formally employed from 2009 to 2017 (panel a) and average real monthly wages from 2009 to 2017 (panel b). Negative estimates imply larger reductions in the respective outcomes among workers employed at the time of the global financial crisis in firms that face larger external demand reductions (relative to less affected firms). The lines join the coefficients for each year. Dashed lines mean that the coefficient on the global financial crisis effect is not statistically significantly different from zero. The full samples have about three million worker-year observations for Brazil and about 800,000 worker-year observations for Ecuador.

more likely to occur through employment out- and Parker 1994), and displacement is more comes than through lower wages. likely for low-skilled and young workers than These results are in line with research on for other types of workers (Devereux 2004; high-income countries. Because more expe- Teulings 1993). Research in LAC countries rienced workers may have higher produc- suggests similar trends. Studying the 2009 tivity and be more expensive for their firms economic crisis in Mexico, Campos-Vazquez to replace (Jovanovic 1979), losing workers (2010) and Freije, López-Acevedo, and with lower levels of human capital (includ- Rodríguez-Oreggia (2011) find larger job loss ing firm-specific human capital) implies rates for young and unskilled workers than lower transaction costs for employers. In for other workers. high-income economies, low-earning work- The LAC region’s combination of large ers exit the labor force in disproportionate informal sectors and workers of varying numbers during downturns (Carneiro, Gui- skill levels suggests there may be a hierar- marães, and Portugal 2012; Solon, Barsky, chy in adjustment costs, in which informal The Impact on Workers, Firms, and Places 63

TABLE 3.1 Presence of negative effects on employment and wage scarring, by gender and education Women Men Secondary Tertiary Secondary Tertiary education education education education Lower labor force Colombia [4–12] N.A. Brazil [4–6] N.A. participation Mexico [1–9] Mexico [1–3] Higher unemployment Argentina [1–6] Colombia [1–3] Argentina [4–6] N.A. Brazil [4–12] Brazil [4–9] Higher informality Brazil [4–9] N.A. Colombia [4–12] Colombia Colombia [4–12] Mexico [10–12] [10–12] Mexico [7–9] Lower wages Brazil [1–3] Brazil [1–3] Mexico [1–3] N.A. Source: Moreno and Sousa 2021. Note: This table reports the countries and years (in brackets) since entering the labor market for which there is evidence of scarring (defined as a statistically significant coefficient). N.A. indicates that no evidence of scarring was found.

workers, who have fewer job protections, That structure can indicate the prevailing face a greater likelihood of job loss regardless degree of market power held by certain firms. of skill. Among formal workers, low-skilled Market power is a function of factors both and low-income workers would be more outside a firm, such as market concentration, likely to experience job loss than high-skilled and inside a firm, such as productivity. This or high-income workers. section highlights frontier research in this area in Latin America and the Caribbean and Firms: The cost of limited market presents new results. competition Cleansing effects The previous section established that crises affect welfare in the long run. This section Job loss spurred by an economic crisis can will discuss how crises also affect efficiency. reduce productivity by destroying employ- Economists are increasingly coming to appre- er-employee matches and the job-specific ciate the importance of firm heterogeneity in human capital arising from them. However, determining and driving workers’ outcomes. macroeconomic shocks result in microeco- Some studies suggest that industry structure nomic reallocation at the worker and firm lev- (Koeber and Wright 2001) and local wage els. At these defining moments, workers’ and bargaining practices (Janssen 2018) matter to firms’ fates are linked. Firms can adjust their these outcomes, but the relationship between number of employees, the number of hours firm and local labor market characteris- worked by these employees, and the wages tics and labor market scarring has received offered to them, and workers can choose either very little attention. This section focuses on to accept these offers or to search for other the roles of two topics that are particularly options. From these interactions, a new short- important in the LAC region: market power run equilibrium is formed. This study shows and labor market conditions. that this equilibrium depends on local labor As a crisis ripples through an economy, market conditions as well as on the ability of the supply side and the demand side of the firms to adjust jobs and wages, which is linked labor market are affected. On the supply to labor market regulations. Because firms are side, less skilled and more vulnerable workers a key transmission channel of the effects of bear larger costs and lasting adverse effects. crises to individual workers, these effects also On the demand side, firm-level responses depend on the structure of the product market, depend on the structure of the local market. existing rents, and rent-sharing mechanisms. 64 employment in Crisis

In the transition to a new equilibrium, adjust employment and wages in response many workers will lose their jobs or see their to these shocks (Fernandes and Silva 2021). earnings fall, some firms will go out of busi- Negative external demand shocks also ness, and new entrants to the labor market increase the skill content of production—the will face more challenging starts to their share of skilled labor in total employment careers. As explored earlier in this chapter, increases—in countries such as Argentina the impacts of a crisis leave scars on work- (Brambilla, Lederman, and Porto 2012), ers and firms. Many workers do not fully Brazil (Mion, Proenca, and Silva 2020), and recover even in the long run: their earnings Colombia (Fieler, Eslava, and Xu 2018). do not bounce back, and their careers follow Through different mechanisms, crises in different, worse tracks. The losers lose a lot. Latin America impact the structure of the Less-skilled workers and those with lower overall economy. Firm-level demand shocks earnings are the most adversely affected during the global financial crisis induced in the LAC region. In the labor market, firm exit—not on impact, but around two employer-employee matches and the job- years after the crisis in Brazil and Ecuador specific human capital arising from them may (Fernandes and Silva 2021). Debt overhang be permanently dissolved because of a tem- problems have also been documented, with porary shock. This dissolution may slow the the potential to kill off less resilient firms ramping up of production later, and it implies and increase the market share of more resil- a loss in productivity. A crisis can also have ient ones. In addition to their effects on persistent effects on technology inputs, which already-existing firms, crises have persistent can be a margin of adjustment for firms, and effects on firms that are created during bad on the structure of the economy, if the crisis times. Demand is a key driver of firms’ capa- kills off some firms and increases the market bilities, and if firms start while demand is share of others. Large economic disruptions low, they will have more difficulty developing can free workers and other inputs of produc- their networks of clients and learning how to tion from low productivity firms, allowing work well with them. New evidence from the them to move to more productive firms as the United States indicates that firms that are economy recovers. Similarly, crises can spur born in times of crisis end up stunted—that is, reallocation of firms out of sectors with very they grow slowly throughout their life cycles low productivity. These changes can have even when times improve (Moreira 2018). persistent effects on the economy. These changes have persistent implications Because of these types of effects, crises can for the economy that firms cannot revert. decrease individual welfare, but they can also By inducing the exit of less efficient firms, increase efficiency and increase productivity bad economic times can have a cleansing (at both the firm and market levels) in the effect. Suppose a labor market is subject to short and medium runs. large friction, so that very-low-productivity Although the Covid-19 crisis, because of firms can survive by hiring workers for very its nature, has led to increasingly rapid tech- low wages. Given the market’s large friction, nological advances driven by digitization workers who receive these firms’ low-wage (Beylis et al. 2020), how can crises with- offers take them, because the other option out this characteristic also have persistent of continuing to look for work despite the effects on technology? Evidence from Brazil low job-matching rate is worse. Therefore, shows that firms adjust to external demand low-productivity firms can essentially trap shocks through changes in productivity, resources that could be more efficiently used skill demand, product appeal, and markups elsewhere. In this context, large economic dis- (Mion, Proenca, and Silva 2020). Firm-level ruptions can have a cleansing effect by free- demand shocks during the global financial ing workers from such firms and allowing crisis caused more-affected firms to reduce them to reallocate themselves into more pro- their capital-to-worker ratio in countries such ductive firms as the economy recovers. Sim- as Ecuador, whereas in Brazil firms simply ilarly, crises could allow reallocation out of The Impact on Workers, Firms, and Places 65

low-productivity sectors that have been living Silva 2021). Figure 3.7 shows that in Brazil, at the margin of survival. But these effects are a drop in demand for a firm’s exports caused possible only if jobs are created after the crises by the crisis was associated with significantly are over. Hence, the positive effects of crises lower productivity for that firm, regardless on aggregate productivity are not guaranteed. of whether this productivity was measured Importantly, the positive effects of cri- using value added per worker or TFP. In ses on firm-level productivity are also not Ecuador, the relationship between productiv- guaranteed. New evidence produced in the ity and decreases in export demand caused by context of this report shows that the global the crisis goes in the same direction. One pos- financial crisis of 2008–09 led to a reduction sible reason for the difference in productivity in firm productivity in Brazil and Ecuador effects across the two countries—long-lasting (contrary to evidence from the United States decline in Brazil versus temporary decline in of job recovery because of increases in total Ecuador—is the documented difference in factor productivity [TFP]) (Fernandes and the countries’ employment adjustment: the

FIGURE 3.7 Dynamic effects of the global financial crisis on firms

a. Value added per worker b. Total factor productivity BRAZIL 0.00 0.00

−0.01 −0.01

−0.02

−0.02 −0.03 Eects of global nancial crisis– induced rm-level demand shocks −0.03 −0.04 2009 2010 2011 2012 2013 2014 2009 2010 2011 2012 2013 2014

ECUADOR 0.01 0.00

0.00 −0.01

−0.01

−0.02 −0.02 Eects of global nancial crisis–

induced rm-level demand shocks −0.03 −0.03 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017

Source: Fernandes and Silva 2021. Note: This figure shows the effects of global financial crisis–induced firm-level demand shocks (experienced from 2008 to 2009) on firms’ productivity from 2009 to the year listed on the x-axis. More negative estimates imply larger reductions in the respective outcomes for firms facing larger external demand reductions (relative to less affected firms). Vertical lines represent 95 percent confidence intervals on the basis of robust standard errors clustered by firm. Each graph shows the coefficients from a single regression that is based on about 90,000 firm-year observations for Brazil and about 23,000 firm-year observations for Ecuador. 66 employment in Crisis

recovery was more “jobless” in Ecuador. output in order to increase the market price This difference is also consistent with possi- of their goods. In the extreme case, a market ble stronger labor hoarding by Ecuadorean served by a monopolist has less output than a firms in anticipation of a recovery in demand, market served by a large number of tiny com- caused by the difficulty of firing and then petitive firms. As such, total employment in rehiring workers or the overhead character a concentrated market may be lower than it of some types of labor. Two other mecha- would be in a more competitive market. nisms were also at play. First, results show At the same time, market concentration that firms in Brazil increased their share of is associated with larger differences between skilled workers during the crisis, but such output prices and firm-level costs (i.e., higher adjustment did not take place in Ecuador. markups). In some cases, higher markups Second, Brazilian firms did not adjust their generate rents that firms may share with their capital-to-worker ratios, but firms in Ecua- workers. A firm-specific “premium” is the dor registered a strong and lasting reduction term used to describe what happens when a in this measure. firm shares higher-than-competitive earnings with its workers in the form of additional pay. Firms might also offer workers high- Protected sectors and firms: Market er-than-competitive wages in order to retain concentration dampens positive them or to encourage loyalty to the firm and cleansing reduce shirking. Recent work demonstrates Less competitive market structures may the importance of firm-specific premiums dampen the positive cleansing effects that can be shared with workers through bar- described earlier in this chapter. If protected gaining (Card, Cardoso, and Kline 2016). firms, defined as those facing less competition, Several studies suggest that higher market adjust less during a crisis, their opportunity to concentration is associated with lower wage experience a cleansing effect is smaller. Rather inequality because firms earning premiums than becoming more agile and productive, if pass those premiums along to their work- these firms gain more market share and fur- ers. For example, Magalhães, Sequeira, and ther crowd out the competition during eco- Afonso (2019) find that inequality (measured nomic downturns, they may trap additional as the ratio of “skilled” workers’ wages to resources that could be used more efficiently less-skilled workers’ wages) is lower when the elsewhere. This dynamic is particularly con- local industry is more concentrated (as mea- cerning in the LAC region, which exhibits sured by the Herfindahl-Hirschman index). high inequality and low productivity growth. However, lower product market concen- It is well accepted that perfect compe- tration may also be correlated with more tition is rare in output and factor mar- competition among workers—but it can help kets. The imperfect competition in output cushion the adverse effects of job loss and markets often implies that firms have some scarring. When there are more firms in an market power. Market power is often mea- area that value a specific set of skills, workers sured with concentration indices, such as the with those skills incur fewer losses follow- Herfindahl-Hirschman index, which calcu- ing job loss (Green 2012; Neffke, Otto, and lates the sum of the squares of the market Hidalgo 2018). Additionally, Yang (2014) shares of each firm in a defined market. Econ- finds that increased agglomeration, defined omists often associate higher measures of as an increase in the number of firms within concentration—resulting from the presence a given sector in a region, reduces unemploy- of fewer firms with larger market shares— ment rates within that sector but is associated with more market power held by those firms. with higher unemployment rates within the Although firms with more market power region. A smaller number of studies focus on tend to make the same kinds of input choices employment adjustment in response to shocks (such as capital-labor ratios) as firms facing in the presence of wage premiums. Orazem, stronger competition, firms with more mar- Vodopivec, and Wu (2005) show that, at least ket power have an incentive to reduce their for Slovenia, firms with higher profits shed The Impact on Workers, Firms, and Places 67

fewer workers than less profitable firms when into public sector organizations (Colonnelli, exposed to negative output shocks. Prem, and Teso 2020). In Ecuador, where In this context, product market concen- political connections have been shown to tration can affect the size and distribution cause misallocation of procurement con- of labor market adjustments to crises. This tracts, firms that form links with the bureau- effect is particularly relevant in Latin Amer- cracy experience an increased probability of ica, where there is some evidence of resource being awarded a government contract (Bru- misallocation in favor of politically connected gués, Brugués, and Giambra 2018). In Costa firms. The value of political capital has been Rica, Alfaro-Urena, Manelici, and Vasquez studied in different contexts in the region. (2019) find large changes in rents collected For example, during the Pinochet regime in by a firm when the firm starts supplying mul- Chile (1973–90), firms with links to the dic- tinational corporations. Overall, De Loecker, tatorship were relatively unproductive and Eeckhout, and Unger (2020) show that mark- benefited from resource misallocation, and ups are very high in Latin America compared these distortions persisted as the country with other continents. transitioned into democracy (González and Although the existing literature shows a Prem 2020). In Brazil, after anticorruption link between product market concentration audits, municipalities experienced an increase (and agglomeration), wages, and unemploy- in economic activity concentrated in the sec- ment, few studies focus on the relationship tors that were most dependent on government between concentration and labor market relationships. Evidence of patronage in Brazil scarring. To fill this gap, Fernandes and Silva was also found in the selection of workers (2021) use detailed, microlevel data to

FIGURE 3.8 Effects of the global financial crisis on Brazilian workers depending on sectoral concentration and state ownership

a. Average months employed b. Average real wage

0.25 0.02 *** *** 0.00 0.20 –0.02 0.15 –0.04 *** 0.10 –0.06 –0.08 0.05 –0.10 0.00 –0.12 –0.14

E ects of global nancial crisis– –0.05 induced rm-level demand shocks –0.16 –0.10 –0.18

Low sector High sector Low sector High sector State owned State owned concentration concentration Not state owned concentration concentration Not state owned

Source: Fernandes and Silva 2021. Note: The figure shows the effects of global financial crisis–induced firm-level demand shocks (2008) on Brazilian workers’ average number of months formally employed and average real monthly wages from 2009 to 2017 in low- and high-concentration sectors in the first set of bars. High-concentration sectors have a Hirshman-Herfindahl index of employment concentration within the sector above the median. The second set of bars in the figure show the effects of global financial crisis–induced firm-level demand shocks (2008–09) in firms that are publicly owned and in firms that are privately owned. Negative estimates imply larger reductions of the respective outcome among workers employed at the time of the global financial crisis in firms that face larger external demand reductions (relative to less affected firms). Each graph shows the coefficients from two different regressions, one for the first set of two bars, and one for the second set of two bars. ***, **, and * indicate the significance levels, 1 percent, 5 percent, and 10 percent, respectively, of the t-test that indicates whether the difference in coefficients across worker categories is significant. 68 employment in Crisis

evaluate changes in worker-level employment private firms, little empirical evidence com- and wages according to the type of market pares the differences in employment adjust- structure and the type of firm. The results are ment to shocks between state-protected presented in figure 3.8. They show, for exam- firms and unprotected firms. ple, that workers who at the time of the crisis To fill this gap, Fernandes and Silva (2021) work in markets with low market concentra- compare the change in employment following tion incur larger scarring effects. In these less a negative shock in state-owned firms and in concentrated sectors, lower wages and lower private firms. They find that state-owned employment are seen in response to shocks, firms adjusted less along the intensive margin as economic theory predicts. Conversely, in (workers’ months worked) in response to the sectors in which a few firms control a large global financial crisis of 2008–09 because share of the market, a negative demand shock these firms were shielded from the shock. does not result in employment losses—rather, it increases employment—and wages do not Places: The role of local adjust. These results are consistent with the opportunities and informality idea that firms with more market power are more protected from negative shocks. So far, this chapter has explored the potential Although the workers in these sectors are differences in scarring caused by differences better insulated from crises than other work- in personal (demographic) characteristics ers, the costs of this protection are borne by and firm characteristics. A third dimension the economy as a whole. that could affect labor market scarring is the The existence of imperfect competition in properties of the local labor markets in which output markets is well accepted by economists workers live and work. Recent research has (as asserted by Card et al. 2018), and con- paid increasing attention to the significant cerns about imperfect competition in Latin costs workers bear when moving between American product markets have emerged in cities or industries. Dix-Carneiro (2014) policy discussions (OECD 2015). These pol- estimates that the median direct mobility icy discussions hinge critically on the source cost that workers face when switching sec- of the concentration in these product markets. tors ranges from 1.4 to 2.7 times the aver- There are several potential sources, includ- age annual wage. Artuc, Chaudhuri, and ing economic barriers to entry (such as large McLaren (2010) and Artuc and McLaren fixed costs), political barriers to entry (such (2015) obtain similar estimates for the cost as government protections or regulation), and of changing sectors. The costs of changing firm-level productivity differences that result locations, by some estimates, can be as high in the most productive firms driving out less as 7 times the average annual income (Bayer, productive firms (Melitz 2003). Keohane, and Timmins 2009; Bishop 2008; Recent work suggests that firm rents are Kennan and Walker 2011).3 As a result, prevalent in developing countries and that workers, especially low-wage workers, tend they are frequently associated with polit- to be more closely tied to their local labor ical connections, both in Latin America markets than was previously thought. (Brugués, Brugués, and Giambra 2018) and Such ties to local labor markets means that elsewhere (Rijkers, Freund, and Nucifora the characteristics of those markets might 2017). State-owned firms are an extreme play a significant role in shaping labor market example of firms with strong political con- scarring. For example, Meekes and Hassink nections. These firms are less profitable in (2019) show that the local housing market general and have higher labor-to-capital significantly affects workers following job ratios than private firms (Dewenter and losses—the effect of displacement on their Malatesta 2001). Although a large liter- probability of changing homes is negative. ature describes the differences in employ- Lessons from the previous section already ment choices between state-owned and suggest that the amount of concentration in The Impact on Workers, Firms, and Places 69

an industry might affect employment and weakens the link between increased labor wages. Living in a “company town” offers demand and the need for their firms to raise very different opportunities than does living wages. This effect generates results similar in a large and diversified metropolitan area. to those of labor market scarring—but, on The degree of informality in the local labor the other hand, job losses in the formal sec- market also matters. tor might be lower in these localities than in Informality might affect scarring in sev- those with less informality. eral important ways, and not all of them Recent work for this study sheds light on point in the same direction. For many work- this apparent contradiction in several ways. ers, especially in the LAC region, the infor- First, Fernandes and Silva (2021) find that mal labor market is considered to be a safety formal workers in municipalities with higher net that replaces the traditional role of social levels of informality lost less in the global insurance. Workers who lose formal sector financial crisis. In particular, within a given jobs and find themselves without family or country, the paper finds smaller employment government support have little choice but to and wage losses for formal workers hit by accept any available job opportunity. When shocks who live in less formal local labor mar- the legal requirements for starting or operat- kets, suggesting that the medium-run effects ing a business are not strictly enforced, these of crises on unemployment are lower for for- workers may have an easier time setting up mal workers in places where informality is informal businesses (such as the street mar- higher. Ironically, the effects of crises may ket shops, or tianguis, often seen in Mexico) even shift from negative to positive for work- than regaining formal employment. Informal ers in municipalities with high informality sector employment can therefore offer oppor- rates. These results are consistent with the idea tunities for human capital accumulation that (a) in the presence of informality, formal through experience and entrepreneurship. firms may employ informal workers and (b) However, the presence of a large informal those firms reduce their employment of infor- sector may indicate imperfect enforcement mal workers, rather than formal workers, in of the minimum wage or other employment the face of negative export shocks (figure 3.9). laws. In many cases, informal sector entre- In this context, informality may provide de preneurs offer services to exporting firms facto flexibility for firms and workers to cope and formal sector firms that help those firms with adverse shocks. This result is in line with smooth out demand fluctuations. Because of findings on the effect of trade liberalization in the various firm-level costs to adjust employ- Brazil by Dix-Carneiro and Kovak (2019) and ment, formal sector firms prefer to keep rel- Ponczek and Ulyssea (2018). atively constant formal labor forces. In some This study also finds that workers in local- cases, employment laws even motivate firms ities with more (alternative) job opportunities to hire fewer workers during peak times than bounce back better from crises. Specifically, they otherwise would. Instead, when demand larger and longer-lasting losses in employment increases, exporting formal sector firms sub- (and sometimes wages) in response to a crisis contract with informal firms or hire work- are found for formal workers in localities with ers informally to meet the surge in demand. larger primary sectors, smaller service sectors, When demand contracts, the formal sector fewer large firms, and production highly con- firms simply reduce their orders and employ- centrated in the same sector where they were ment for the informal sector or reduce their employed precrisis (Fernandes and Silva 2021). use of informal workers, keeping their formal These workers’ persistent earnings losses employment more or less constant. Therefore, may reflect their lack of opportunities in the formal workers in areas with high informal- rebound, not just “scarring” in the traditional ity may find wage growth elusive, because sense of a persistent loss of human capital the essentially infinite supply of informal associated with a period of unemployment or workers available to perform unfilled jobs lower-quality employment. 70 employment in Crisis

FIGURE 3.9 Effects of the global financial crisis on Brazilian workers depending on local labor market informality

a. Average months employed b. Average real wage 0.01 0.01 0.00 0.00 –0.01 –0.01 –0.02 –0.02 –0.03 –0.03 –0.04 –0.04 –0.05 –0.05 –0.06 –0.06

E ects of global nancial crisis– ** *** *** induced rm-level demand shocks –0.07 –0.07 Low High Low High Low High Low High Informality rate Share of agriculture Informality rate Share of agriculture

Source: Fernandes and Silva 2021. Note: The figure shows the effects of global financial crisis–induced firm-level demand shocks (2008) on Brazilian workers’ average number of months formally employed and average real monthly wages from 2009 to 2017 in high and low informality. High informality municipalities are those with a level of informality above the median. The share-of-agriculture bars in each panel show the effects of global financial crisis–induced firm-level demand shocks (2008–09) in firms that are located in municipalities with high and low shares of employment in agriculture. Negative estimates imply larger reductions of the respective outcome among workers employed at the time of the global financial crisis in firms that face larger external demand reductions (relative to less affected firms). Each panel shows coefficients from two different regressions, one for the first set of two bars, and one for the second set of two bars. ***, **, and * indicate the significance levels, 1 percent, 5 percent, and 10 percent, respectively, of the t-test that indicates whether the difference in coefficients across worker categories is significant.

Conclusion slowed by market friction. The findings presented in this chapter suggest that both chan- Crises can lead to long-term effects on labor nels are in play in Latin America. markets through two key channels. First, to The evidence presented in this chapter the extent that crises lead to interruptions indicates that crises leave scars—not on every in employment that destroy or reduce the worker, but on many of them. Some work- human capital stock of a region, they may ers recover from displacement and other have long-run effects on that region’s growth livelihood shocks, while others end up per- 4 prospects. The accumulation of human manently scarred. For example, in Brazil and capital is integral to the continued eco- Ecuador, the effects of a crisis on worker nomic growth and social gains of a region. employment and wages are still present for Although a country’s human capital stock is an average of nine years after the beginning a key determinant of growth through its role of the crisis. It is striking that a temporary as an input into production (Mincer 1984), shock can lead to the type of effects normally researchers have also emphasized its contri- associated with large permanent shocks, such bution to growth through increasing innova- as trade liberalization.5 This finding suggests tion, productivity, and technological uptake that success in raising the LAC region’s long- (Benhabib and Spiegel 1994; Nelson and term economic growth rate will depend on Phelps 1966; Romer 1990). whether the region’s crisis-response measures The second channel by which crises can effectively prevent unnecessary destruction of lead to employment changes with long-term human capital and business capabilities. effects on growth is by inducing a reallo- For lower-skilled workers (those with no cation of workers and firms (for example, college education), the earnings losses caused across sectors or occupations) or altering by crises are persistent. Workers with higher firms’ use of technology (for example, the education suffer minimal impacts from crises way firms combine their various inputs). Such on their wages and very short-lived impacts reallocation results both in job destruction on their employment. Interestingly, however, (which hits at once) and in job creation being the responses to the global financial crisis The Impact on Workers, Firms, and Places 71

of 2008–09 were similar across male and of production: the share of skilled labor in female workers and across workers with high total employment has been shown to increase and low previous participation in the formal in the presence of negative external demand labor market. New entrants to the labor mar- shocks in countries such as Arg entina ket during a crisis face a worse start to their (Brambilla, Lederman, and Porto 2012), careers, from which it is difficult to recover. Brazil (Mion, Proenca, and Silva 2020), and Although the specific mechanisms for and Colombia (Fieler, Eslava, and Xu 2018). the lengths of scarring may vary between Second, firms that are born in times of crisis men and women, the overall story across end up stunted, growing slowly throughout the LAC countries is similar—for both men their life cycles even when times improve. and women, scarring is more likely for those If firms start in a period when demand is with lower levels of schooling compared with low, they are less able to develop client rela- those with college educations, and it is far tionships and to learn how to work well with more likely to occur through employment clients, and this impairment stays with them outcomes than through lower wages. for a long time. If jobs are slow to recover In this context, the extent to which after a crisis because of firm scarring in this resources are targeted toward those most broader sense, policies to address labor- affected by crises and whether those resources market scarring may not be enough to solve can effectively reach them is crucial. However, the problem, as discussed in the next chapter. if workers in localities with more job opportu- Another key mechanism at play is cleans- nities bounce back better from crises, as this ing. Less productive firms can more easily study finds—either because they have more turn unprofitable and be scrapped in a reces- informal job opportunities or because the set sion than can highly productive ones. How- of employers available to them (more large ever, the units in place may not experience firms and larger service sectors with compat- the full impact of a fall in demand if the fall ible jobs) is more amenable to them finding is made up for with a reduction in the job new jobs—the observed losses in earnings creation rate. Destruction and reallocation among other workers may not be so much effects are a good thing, but only as long as scarring, in the traditional sense of a loss of new jobs are created after the crisis is over. human capital, but rather a symptom of those Once the firm and spatial dimensions of labor workers’ lack of opportunities. In this case, market adjustment to crises are included, the approaching the issue solely from a work- importance of the demand side is clear. Impor- er-based perspective will not solve the issue. tantly, results show that protected sectors and The effects on workers described above firms adjust less during a crisis (Fernandes have important implications for equity and and Silva 2021), suggesting that these sectors poverty. And, as this report shows, they can experience less of a cleansing effect. also affect efficiency in the long run. Crises do more than just destroy jobs; they also alter productivity down the road. Con sistent Notes with the slow pickup of jobs after a crisis are 1. See, for example, Brunner and Kuhn (2014) for two facts. First, crises can have persistent evidence on Austria; Genda, Kondo, and Ohta effects on the technology use of preexisting (2010) for Japan and the United States; Kwon, firms and on the size and capability of new Milgrom, and Hwang (2010) for Sweden and firms formed during the crisis. Preexisting the United States; and Kahn (2010) for the firms adjust their skill demand, product United States. 2. These studies include Burda and Mertens appeal, and markups in response to demand (2001); Couch (2001); Davis and Von Wach- shocks. For example, Fernandes and Silva ter 2011; Fallick (1996); Flaaen, Shapiro, and (2021) find that in response to the global Sorkin 2019; Hyslop and Townsend (2019); financial crisis, Brazilian and Ecuadorian Kletzer (1998); Krolikowski 2017; Lachowska, firms increased their capital per worker Mas, and Woodbury 2018; Menezes-Filho ratio. Crises also increase the skill content (2004); and Ruhm (1991a, b). 72 employment in Crisis

3. There are important differences in the determi- Benhabib, J., and M. M. Spiegel. 1994. “The nants of mobility across the diverse countries Role of Human Capital in Economic Develop- of the LAC region. A “spatial development” ment: Evidence from Aggregate Cross-Country flagship report analyzing these determinants Data.” Journal of Monetary Economics 34 (2): in the region is under way. 143–73. 4. Crises can also lead to firm scarring by Beylis, G., R. Fattal Jaef, M. Morris, A. Rekha imposing a fixed cost on firms, from which Sebastian, and R. Sinha. 2020. Going Viral: financially constrained firms might not be COVID-19 and the Accelerated Transformation able to recover, and a loss of firm-specific of Jobs in Latin America and the Caribbean. human capital. World Bank Latin American and Caribbean 5. The international trade literature shows Studies. Washington, DC: World Bank. large and lasting dynamic labor market Bishop, K. C. 2008. “A Dynamic Model of Loca- adjustments to trade shocks (see, for exam- tion Choice and Hedonic Valuation.” Unpub- ple, Autor et al. 2014; Dauth, Findeisen, lished, Washington University in St. Louis, 5. and Suedekum 2017; Dauth et al. 2019; Brambilla, I., D. Lederman, and G. Porto. Dix-Carneiro and Kovak 2017, 2019; and 2012. “Exports, Export Destinations, and Utar 2018). Skills.” American Economic Review 102 (7): 3406–38. References Brugués, F., J. Brugués, and S. Giambra. 2018. “Political Connections and Misallocation of Alfaro-Urena, A., I. Manelici, and J. P. Vasquez. Procurement Contracts: Evidence from Ecua- 2019. “The Effects of Multinationals on Work- dor.” Research Department Working Papers ers: Evidence from Costa Rica.” Unpublished 1394, CAF Development Bank of Latin paper. America, Caracas, Venezuela. Amarante, V., R. Arim, and A. Dean. 2014. “The Brunner, B., and A. Kuhn. 2014. “The Impact of Effects of Being out of the Labor Market on Sub- Labor Market Entry Conditions on Initial Job sequent Wages: Evidence for Uruguay.” Journal Assignment and Wages.” Journal of Popula- of Labor Research 35 (1): 39–62. tion Economics 27 (3): 705–38. Arias-Vazuez, FJ, D. Lederman, and L. Venturi. Burda, M. C., and A. Mertens. 2001. “Estimat- 2019. “Transitions of Workers Displaced due ing Wage Losses of Displaced Workers in to Firm Closure.” Unpublished paper. Germany.” Labour Economics 8 (1): 15–41. Artuc, E., and J. McLaren. 2015. “Trade Policy and Burdett, K., C. Carrillo-Tudela, and M. Coles. Wage Inequality: A Structural Analysis with 2020. “The Cost of Job Loss.” Review of Eco- Occupational and Sectoral Mobility.” Journal nomic Studies 87 (4): 1757–98. of International Economics 97: 278–94. Campos-Vazquez, R. 2010. “The Effects of Artuc, E., S. Chaudhuri, and J. McLaren. 2010. Macroeconomic Shocks on Employment: The “Trade Shocks and Labor Adjustment: A Case of Mexico.” Estudios Economicos 25 (1): Structural Empirical Approach.” American 177–246. Economic Review 100: 1008–45. Card, D., A. R. Cardoso, J. Heining, and P. Kline. Arulampalam, W. 2001. “Is Unemployment 2018. “Firms and Labor Market Inequality: Really Scarring? Effects of Unemployment Evidence and Some Theory.” Journal of Labor Experiences on Wages.” Economic Journal Economics 36 (S1): 13–70. 111 (475): 585–606. Card, D., A. R. Cardoso, and P. Kline. 2016. Autor, D. H., D. Dorn, and G. H. Hanson. 2013. “Bargaining, Sorting, and the Gender Wage “The China Syndrome: Local Labor Market Gap: Quantifying the Impact of Firms on the Effects of Import Competition in the United Relative Pay of Women.” Quarterly Journal of States.” American Economic Review 103 (6): Economics 131 (2): 633–86. 2121–68. Carneiro, A., P. Guimarães, and P. Portugal. Autor, D. H., D. Dorn, G. H. Hanson, and 2012. “Real Wages and the Business Cycle: J. Song. 2014. “Trade Adjustment: Worker- Accounting for Worker, Firm, and Job Title Level Evidence.” Quarterly Journal of Heterogeneity.” American Economic Journal: Economics 129 (4): 1799–1860. Macroeconomics 4 (2): 133–52. Bayer, P., N. Keohane, and C. Timmins. 2009. Carrington, W. J. 1993. “Wage Losses for “Migration and Hedonic Valuation: The Case Displaced Workers: Is It Really the Firm That of Air Quality.” Journal of Environmental Matters?” Journal of Human Resources Economics and Management 58: 1–14. 28 (3): 435–62. The Impact on Workers, Firms, and Places 73

Carrington, W. J., and B. Fallick. 2017. “Why Do American Economic Review 107 (10): Earnings Fall with Job Displacement?” Indus- 2908–46. trial Relations 56 (4): 688–722. Dix-Carneiro, R., and B. K. Kovak. 2019. “Margins Colonnelli, E., M. Prem, and E. Teso. 2020. of Labor Market Adjustment to Trade.” Journal “Patronage and Selection in Public Sector of International Economics 117: 125– 42. Organizations.” American Economic Review Fallick, B. C. 1996. “A Review of the Recent Empir- 110 (10): 3071–99. ical Literature on Displaced Workers.” Industrial Couch, K. A. 2001. “Earnings Losses and Unem- and Labor Relations Review 50 (1): 5–16. ployment of Displaced Workers in Germany.” Farber, H. S. 2003. “Job Loss in the United States, Industrial and Labor Relations Review 54 (3): 1981–2001.” Working Paper 471, Indus- 559–72. trial Relations Section, Princeton University, Cruces, G., A. Ham, and M. Viollaz. 2012. Princeton, NJ. “Scarring Effects of Youth Unemployment Fernandes, A., and J. Silva. 2021. “Labor Mar- and Informality: Evidence from Brazil.” ket Adjustment to External Shocks: Evidence Center for Distributive, Labor and Social for Workers and Firms in Brazil and Ecuador.” Studies (CEDLAS) working paper, Econom- Background paper written for this report. ics Department, Universidad Nacional de la World Bank, Washington, DC. Plata, Argentina. Fieler, A. C., M. Eslava, and D. Y. Xu. 2018. “Trade, Dauth, W., S. Findeisen, and J. Suedekum. 2016. Quality Upgrading, and Input Linkages: Theory “Adjusting to Globalization: Evidence from and Evidence from Colombia.” American Worker-Establishment Matches in Germany.” Economic Review 108 (1): 109–46. Düsseldorf Institute for Competition Econom- Flaaen, A., M. D. Shapiro, and I. Sorkin. 2019. ics (DICE) Discussion Paper 205, Heinrich “Reconsidering the Consequences of Worker Heine University, Düsseldorf, Germany. Displacements: Firm versus Worker Perspec- Dauth, W., S. Findeisen, and J. Suedekum. 2017. tive.” American Economic Journal: Macro- “Trade and Manufacturing Jobs in Germany.” economics 11 (2): 193–227. American Economic Review 107 (5): 337–42. Freije, S., G. López-Acevedo, and E. Rodrí- Dauth, W., S. Findeisen, J. Suedekum, and guez-Oreggia. 2011. “Effects of the 2008–09 N. Woessner. 2019. “The Adjustment of Labor Economic Crisis on Labor Markets in Mex- Markets to Robots.” University of Würzburg, ico.” Policy Research Working Paper 5840, Würzburg, Germany. World Bank, Washington, DC. Davis, S. J., and T. M. Von Wachter. 2011. Genda, Y., A. Kondo, and S. Ohta. 2010. “Long- “Recessions and the Cost of Job Loss.” Work- Term Effects of a Recession at Labor Market ing Paper 17638, National Bureau of Economic Entry in Japan and the United States.” Journal Research, Cambridge, MA. of Human Resources 45 (1): 157–96. De Loecker, J., J. Eeckhout, and G. Unger. 2020. González, F., and M. Prem. 2020. “Losing “The Rise of Market Power and the Macro- Your Dictator: Firms during Political Transi- economic Implications.” Quarterly Journal of tion.” Journal of Economic Growth 25 (2): Economics 135 (2): 561–644. 227–57. Dell, M., B. Feigenberg, and K. Teshima. 2019. Green, C. P. 2012. “Short Term Gain, Long “The Violent Consequences of Trade-Induced Term Pain: Informal Job Search Methods and Worker Displacement in Mexico.” American Post-Displacement Outcomes.” Journal of Economic Review: Insights 1 (1): 43–58. Labor Research 33 (3): 337–52. Devereux, P. J. 2004. “Cyclical Quality Adjust- Gregg, P., and E. Tominey. 2005. “The Wage Scar ment in the Labor Market.” Southern Eco- from Male Youth Unemployment.” Labour nomic Journal 70 (3): 600–15. Economics 12 (4): 487–509. Dewenter, K., and P. Malatesta. 2001. “State- Gregory, M., and R. Jukes. 2001. “Unemploy- Owned and Privately Owned Firms: An ment and Subsequent Earnings: Estimating Empirical Analysis of Profitability, Leverage, Scarring among British Men.” Economic Jour- and Labor Intensity.” American Economic nal 111 (475): 607–25. Review 91: 320–34. Hardoy, I., and P. Schone. 2013. “No Youth Dix-Carneiro, R. 2014. “Trade Liberalization Left Behind? The Long-Term Impact of Dis- and Labor Market Dynamics.” Econometrica placement on Young Workers.” Kyklos 66 (3): 82: 825–85. 342–64. Dix-Carneiro, R., and B. K. Kovak. 2017. “Trade Howland, M., and G. E. Peterson. 1988. “Labor Liberalization and Regional Dynamics.” Market Conditions and the Reemployment 74 employment in Crisis

of Displaced Workers.” Industrial and Labor Mismatch and the Long-Term Effects of Grad- Relations Review 42 (1): 109–22. uating in a Recession.” European Economic Hyslop, D. R., and W. Townsend. 2019. “The Review 84: 3 –17. Longer-Term Impacts of Job Displacement on Magalhães, M., T. Sequeira, and Ó. Afonso. 2019. Labour Market Outcomes in New Zealand.” “Industry Concentration and Wage Inequal- Australian Economic Review 52 (2): 158–77. ity: A Directed Technical Change Approach.” Jacobson, L. S., R. J. LaLonde, and D. G. Sul- Open Economies Review 30: 457–81. livan. 1993a. “Earnings Losses of Displaced Martinoty, L. 2016. “Initial Conditions and Life- Workers.” American Economic Review 83 (4): time Labor Outcomes: The Persistent Cohort 685–709. Effect of Graduating in a Crisis.” Mimeo. Jacobson, L. S., R. J. LaLonde, and D. G. Sulli- McCarthy, N., and P. W. Wright. 2018. “The van. 1993b. The Costs of Worker Dislocation. Impact of Displacement on the Earnings of Kalamazoo, MI: W. E. Upjohn Institute for Workers in Ireland.” Economic and Social Employment Research. Review 49 (4): 373–417. Janssen, S. 2018. “The Decentralization of Wage Meekes, J., and W. H. J. Hassink. 2019. “The Bargaining and Income Losses after Worker Role of the Housing Market in Workers’ Resil- Displacement.” Journal of the European Eco- ience to Job Displacement after Firm Bank- nomic Association 16 (1): 77–122. ruptcy.” Journal of Urban Economics 109: Jovanovic, B. 1979. “Job Matching and the The- 41–65. ory of Turnover.” Journal of Political Econ- Melitz, M. J. 2003. “The Impact of Trade on omy 87 (5): 972–90. Intra-industry Reallocations and Aggregate Kahn, L. B. 2010. “The Long-Term Labor Market Industry Productivity.” Econometrica 71 (6): Consequences of Graduating from College in 1695–1725. a Bad Economy.” Labour Economics 17 (2): Menezes-Filho, N. 2004. “Costs of Displacement 303–16. in Brazil.” University of São Paulo, Brazil. Kaplan, D. S., G. M. González, and R. Robert- Mincer, J. 1984. “Human Capital and Economic son. 2007. Mexican Employment Dynamics: Growth.” Economics of Education Review Evidence from Matched Firm-Worker Data. 3 (3): 195–205. Washington, DC: World Bank. Mion, G., R. Proenca, and J. Silva. 2020. “Trade, Kennan, J., and J. R. Walker. 2011. “The Effect Skills, and Productivity.” Unpublished paper. of Expected Income on Individual Migration Moreira, S. 2017. “Firm Dynamics, Persistent Decisions.” Econometrica 79 (1): 211–51. Effects of Entry Conditions, and Business Kletzer, L. G. 1998. “Job Displacement.” Journal Cycles,” Working Papers 17-29, Center for of Economic Perspectives 12 (1): 115–36. Economic Studies, U.S. Census Bureau. Koeber, C., and D. W. Wright. 2001. “Wage Moreno, L., and L. Sousa. 2021. “Early Employ- Bias in Worker Displacement: How Industrial ment Conditions and Labor Scarring in Latin Structure Shapes the Job Loss and Earnings America.” Background paper written for this Decline of Older American Workers.” Journal report. World Bank, Washington, DC. of Socio-Economics 30 (4): 343–52. Neffke, F. M. H., A. Otto, and C. Hidalgo. 2018. Krolikowski, P. 2017. “Job Ladders and Earnings “The Mobility of Displaced Workers: How the of Displaced Workers.” American Economic Local Industry Mix Affects Job Search.” Jour- Journal: Macroeconomics 9 (2): 1–31. nal of Urban Economics 108: 124–40. Kwon, I., E. M. Milgrom, and S. Hwang. 2010. Nelson, R. R., and E. S. Phelps. 1966. “Invest- “Cohort Effects in Promotions and Wages: ment in Humans, Technological Diffusion, Evidence from Sweden and the United and Economic Growth.” The American Eco- States.” Journal of Human Resources 45 (3): nomic Review 56 (1/2): 69–75. 772–808. OECD (Organisation for Economic Co-opera- Lachowska, M., A. Mas, and S. A. Woodbury. tion and Development). 2015. “Competition 2018. “Sources of Displaced Workers’ and Market Studies in Latin America: The Long-Term Earnings Losses.” Working Case of Chile, Colombia, Costa Rica, Mexico, Paper 24217, National Bureau of Economic Panama and Peru.” Organisation for Eco nomic Research, Cambridge, MA. Co-operation and Development, Paris. Liu, K., K. G. Salvanes, and E. Ø. Sørensen. 2016. Orazem, P. F., M. Vodopivec, and R. Wu. 2005. “Good Skills in Bad Times: Cyclical Skill “Worker Displacement during the Transition: The Impact on Workers, Firms, and Places 75

Experience from Slovenia.” Economics of How Important Is Composition Bias?” Quar- Transition 13 (2): 311–40. terly Journal of Economics 109 (1): 1–25. Oreopoulos, P., M. Page, and A. H. Stevens. 2008. Teulings, C. 1993. “The Diverging Effects of “The Intergenerational Effects of Worker the Business Cycle on the Expected Duration Displacement.” Journal of Labor Economics of Job Search.” Oxford Economic Papers 26 (3): 455–83. 45: 482–500. Ponczek, V., and G. Ulyssea. 2018. “Is Informal- Utar, H. 2014. “When the Floodgates Open: ity an Employment Buffer? Evidence from the ‘Northern’ Firms’ Response to Removal of Trade Trade Liberalization in Brazil.” Unpublished Quotas on Chinese Foods.” American Economic manuscript. Journal: Applied Economics 6 (4): 226–50. Rijkers, B., C. Freund, and A. Nucifora. 2017. Utar, H. 2018. “Workers beneath the Floodgates: “All in the Family: State Capture in Tunisia.” Low-Wage Import Competition and Workers’ Journal of Development Economics 124 (C): Adjustment.” Review of Economics and Sta- 41–59. tistics 100 (4): 631–47. Romer, P. M. 1990. “Endogenous Technological Yagan, D. 2019. “Employment Hysteresis from Change.” Journal of Political Economy 98 the Great Recession.” Journal of Political (5, Part 2): 71–102. Economy 127 (5): 2505–58. Ruhm, C. J. 1991a. “Are Workers Permanently Yang, X. 2014. “Labor Market Frictions, Scarred by Job Displacements?” American Agglomeration, and Regional Unemployment Economic Review 81 (1): 319–24. Disparities.” Annals of Regional Science Ruhm, C. J. 1991b. “Displacement Induced Job- 52 (2): 489–512. lessness.” Review of Economics and Statistics Yi, M., S. Müller, and J. Stegmaier. 2016. “Indus- 73 (3): 517–22. try Mix, Local Labor Markets, and the Inci- Solon, G., R. Barsky, and J. A. Parker. 1994. dence of Trade Shocks.” US Census Bureau, “Measuring the Cyclicality of Real Wages: Suitland, MD.

Toward an Integrated Policy Response 4

Introduction economic growth potential. To mitigate this damage, policy makers should design and Few doubt that a better policy framework to deploy instruments that cushion the effects mitigate, manage, and help people recover of crises on workers in the short run; shock from crises is crucial for countries in Latin impacts spread unequally across workers and America and the Caribbean (LAC) to succeed firms, and many will not regain their lost job, in lifting their long-term growth rates and wages, or clients. But policy makers should increasing their people’s well-being. Macro- pay just as much attention to efficiency and economic frameworks across the countries resilience, promoting peoples’ ability to in the region changed dramatically in the bounce back when exposed to an adverse 1990s, as did social protection and labor shock (which can be aided by healthy eco- policies in the early 2000s. But changes in policy have been relatively limited since then. nomic growth). The COVID-19 pandemic and the associated This chapter builds on the findings from sluggishness of the global economy may be the first three chapters to identify the neces- prolonged, and structural changes in labor sary elements of an effective policy response markets are ongoing. Given these circum- to crises in Latin America and the Carib- stances, crisis response has moved to the fore- bean as revealed by these broader lenses. It front of policy debate in the LAC region. discusses the policy implications of the pre- Considering the evidence presented in ear- vious chapters’ findings, assesses the ability lier chapters on the importance of demand to of existing systems to tackle the challenges crisis adjustment and the triangle of work- of crisis response, and discusses potential ers, sectors and firms, and locations, how reforms, although it does not evaluate the can policies mitigate the impacts of crises impacts of the different policy responses pro- on workers and promote better recovery? posed. The reported policy results and imple- This study shows that crises have a mean- mentation details are based on the existing ingful negative effect on welfare in the LAC literature, and the new evidence presented on region and that the documented labor mar- the effectiveness of reforms is based on past ket scarring is likely to affect the region’s crises in the region.

77 78 employment in Crisis

The evidence featured in the previous of these impacts or the economy’s ability to chapters suggests that the success of pol- bounce back. This study highlights, for exam- icy responses to the crisis triggered by the ple, the dichotomy between protected and COVID-19 pandemic will depend on whether unprotected firms in the LAC region (caused response measures effectively prevent unnec- by the former group’s market power) and the essary destruction of human capital and oth- region’s low geographic mobility of workers, erwise viable enterprises and on the quality both of which magnify the welfare effects of of complementary domestic policies and shocks. It also highlights pockets of rigidity in reforms beyond the labor market. Cushion- labor regulations that are slowing transitions ing the crisis’s short-term impact through of workers across jobs. Hence, competition macroeconomic and social protection and policies, regional policies, and labor regu- labor policies will be crucial to avoiding pov- lations are a third key dimension of a policy erty and excessive job destruction, given the response to crises. These important structural employment and wage losses documented in issues are also causing LAC labor markets’ this report. Strong, prudent macroeconomic poor adjustment to crises, and might require frameworks and automatic stabilizers are the interventions at the sector and locality levels first line of defense to shield labor markets that interact with social protection needs and from crises. Prudent fiscal and monetary incentives. policies can lower the likelihood and sever- Local labor markets’ characteristics ity of certain types of crises and provide the and sector product markets’ conditions fiscal space needed to give support and avert determine the magnitude of the impacts of system-wide financial strain when crises crises on workers. In terms of labor mar- do occur. ket regulations and institutions, this study In addition to macroeconomic poli- documents limited adjustment through cies, the typical automatic stabilizer used reductions in hours, more adjustment in Organisation for Economic Co-opera- through unemployment, and an informal tion and Development (OECD) countries sector that serves as a buffer in some coun- is reliable unemployment insurance, which tries. With respect to product market con- many LAC countries still lack. This type of ditions, the study finds that initially similar social protection and labor program is key workers experience different employment to cushioning the impacts of crises on for- and earnings outcomes after exposure to mal workers. However, many workers in the the same crisis because of differences in LAC region earn their living in the informal their sectors’ competition structures. It also economy, and the best way to protect their documents that the localities where workers consumption is through responsive cash live affect the impacts they face. Its results transfer programs. Targeted on the basis of suggest that employment and wage losses household need rather than whether the lost following shocks are smaller for formal job was formal or informal, these programs workers who live in areas with more infor- reduce the extent to which labor market mality. Why do crisis effects pass through adjustment translates into short- and long- to workers more in some places than in oth- term impacts on the poor and vulnerable. ers? One factor is that the geographic mobil- Because reemployment is crucial to avoid ity of workers is lower than expected by scarring, active labor market programs to economists and policy makers. Constraints support reskilling and job search are a third on movement create friction in the styl- vital element of effective social protection ized labor market adjustment mechanisms and labor systems. that can magnify welfare losses, as shown Although social protection and labor sys- in earlier chapters. The policy response to tems can cushion workers from the impacts crises needs to tackle these structural issues of crises, they do not address the struc- squarely, according to the weight each issue tural issues that determine the magnitude has in each country or setting. Toward an Integrated Policy Response 79

Given the complexity of labor markets’ policy options that governments in the region adjustment to economic crises in the LAC could consider in order to improve the mixed region, this report argues that countries record of their employment support programs can improve their responses by advancing and to reduce the (short- and long-term) on three fronts. The combination of poli- impacts of crises on workers. A distinc- cies needed is truly cross-sectorial, including tion is made between short-term, transitory macroeconomic, social protection and labor, programs enacted during crises to avoid competition, and regional policies, and it will excessive job losses (including employment determine the speed of adjustment and the retention schemes, temporary employment trajectory of workers’ recoveries. programs, and demand stimulus programs) This chapter starts with a discussion of the and longer-term programs deployed by gov- public policy “shield” that determines how ernments to reskill workers and ease their a crisis affects workers and their families— transitions between jobs. The latter set of the country’s macroeconomic frameworks programs are discussed in light of the LAC and automatic stabilizers. Strong, sound region’s mixed track record at reducing the macroeconomic policies can decrease the duration of unemployment and improving frequency of crises, for example, by protect- the quality of job matches. Because even brief ing against fiscal imbalances and domestic crises can leave lasting scars on workers, the inflationary pressures. They can also lessen formidable challenge for governments is to the severity of crises by reducing the size distinguish the crises requiring only transi- of the needed adjustments and shaping the tory responses from those requiring more adjustments’ composition. The LAC region sustained support and to respond accord- has significantly improved its macroeco- ingly as workers and the economy adjust. nomic frameworks in the past few decades, This chapter also discusses insights from the resulting in fewer domestic crises and existing evidence and policy experiences and significantly lower inflation rates. Even so, its recommends ways to improve the region’s fiscal policy in particular remains weak and, responses to the current COVID-19 crisis. in many countries, unsustainable, with small Next, the chapter shifts its focus to the tax bases and relatively generous entitlement efficiency effects of crises and how to tackle programs. Many countries in the region also structural issues that can worsen labor lack social protection and labor programs market adjustment—in particular, how to that provide adequate automatic stabilizers break rigidities, address the insider story, (such as unemployment insurance). and respond to the lack of opportunities in The chapter next turns to the question some places and regions. Chapter 3 showed of how governments can use labor markets how factors beyond the labor market affect and social protection policies to alleviate or the magnitude of crises’ impacts on workers. revert the impacts of crises on workers and Structural challenges in the LAC region act to the economy. To begin answering this ques- slow and even prevent necessary labor mar- tion, the chapter considers why most people ket adjustments, hence weakening economic in the LAC region are not covered by any for- recoveries. These structural issues can change mal income assistance for unemployment. It the nature—and the impact on people—of assesses the region’s existing income support systemic shocks from transitory to long-term. programs (including unemployment insur- The policy implication of these findings and ance plans, cash transfers, and other social the related literature is that even if a country’s assistance benefits), their unintended (posi- macroeconomic policies and its social protec- tive and negative) consequences as currently tion and labor system are in pristine order, designed, and how they could be reengineered the best outcomes for workers in the wake to provide a more effective crisis response. of crises can be achieved by complementing The chapter’s discussion of social protection these measures with sectoral and place-based and labor systems concludes by highlighting policies to deal with the structural issues 80 employment in Crisis

that currently impede strong recoveries. In the transition to a new equilibrium, This reform agenda would involve address- many workers will lose their jobs or see their ing inefficiencies in labor market adjustment earnings fall, some firms will go out of busi- caused by legislation, product markets’ struc- ness, and new entrants to the labor market will tures, workers’ lack of geographic mobility, face more challenging starts to their careers. and depressed areas. Addressing these struc- As explored in chapter 3, the impacts of a cri- tural challenges will require changes to legal sis leave scars on workers and firms. Many frameworks and regulations as well as tar- workers do not fully recover even in the long geted public investment. run: their earnings do not come back, and their careers follow different, worse tracks. Three key policy dimensions The losers lose a lot. Less skilled workers and those with lower earnings, are the most So far, this report has unpacked the welfare adversely affected in the LAC region. From a and efficiency stories caused by the triangle labor market perspective, employer-employee of workers, sectors and firms, and localities matches and the job-specific human capital and has examined from there the mecha- arising from them, which often take a long nisms at play. time to build and would regain profitability An exogenous shock is transmitted to the when the economy returns to normal, may be labor market through shocks in supply and permanently dissolved because of a tempo- demand, altering the labor market’s normal rary shock. This loss may slow the ramping functioning and generating job losses and up of production later, and it implies a loss in transitions to informal jobs in excess of those productivity. A crisis can also have persistent registered in normal times. These excess effects on technology inputs, which can be a flows affect the size and composition of margin of adjustment for firms, and on the employment. Chapter 2 describes this adjust- structure of the economy, as the crisis kills ment process in the LAC region, considering off some firms and increases the market share the various margins by which the labor mar- of others. These changes can have persistent ket can adjust to shocks and the factors that implications for the economy that firms can- help determine the severity of the adjustment. not revert. What does this process mean for work- Considering the triangle of workers, sec- ers, and how do the characteristics of sectors and firms, and locations, how can pol- tors, firms, and localities affect the size and icies mitigate the impacts described above? nature of crisis impacts? A macroeconomic Figure 4.1 presents a framework for thinking shock results in microeconomic realloca- about the relevant policy areas. Given the tion at the worker and firm levels. At these characteristics of labor market adjustments defining moments, workers’ and firms’ fates in the LAC region identified in chapters 2 are linked. Firms can adjust their numbers of and 3, this chapter discusses what policies are employees, hours worked, and wages offered, needed in the region to mitigate the negative and workers can choose to either accept these impacts of crises and provide better responses offers or search for other options. From these to them. The analysis makes clear that labor interactions, a new short-run equilibrium market policies alone are insufficient. Strong, is formed. This study shows that the new prudent macroeconomic frameworks and equilibrium depends on local labor market automatic stabilizers (the shield in figure 4.1) conditions as well as on the ability of firms are the first line of defense to protect labor to adjust jobs and wages, which is linked to markets from crises. These preventative mac- labor market regulations. Because firms are roeconomic measures—along with social a key channel transmitting the effects of cri- protection and labor systems—mitigate the ses to individual workers, these effects also impacts of external shocks on a country’s depend on existing rents, rent-sharing mech- economy and reduce the chance of domes- anisms, and the structure of product markets. tic shocks by stabilizing the macroeconomic Toward an Integrated Policy Response 81

FIGURE 4.1 How adjustment works and a triple entry of policies to smooth it

Income support s k Active labor market programs r o WORKERS w e

SHOCK m a r f

o c

e Competition policies o

r Local investment + place-based policies

a Labor regulations

t S

Source: World Bank environment. Prudent fiscal and monetary impact of a shock on aggregate demand policies prevent certain types of crises and and employment and, therefore, the size ensure sufficient fiscal space to provide sup- and composition of the shock’s effects on port and avert system-wide financial strain labor markets. These policies can lessen the when other types of crises occur.1 severity of crises by shrinking the necessary In order for economies to effectively adjustment and shaping its composition. protect themselves from external shocks, They will affect both the change in the total macroeconomic stabilizers are key. As size of the labor market caused by a crisis discussed in later sections of this chapter, and the labor market dynamics between the countries’ social protection and labor systems formal economy, the informal economy, and can include nationally administered income outright unemployment (for example, by protection arrangements, such as unemploy- preventing excessive destruction of formal ment insurance and other forms of income jobs), as described in chapter 2.2 support for affected individuals. These Social protection and labor policies are key “safety net” programs ideally function coun- to cushioning the impacts of crises on work- tercyclically, expanding and increasing the ers. In addition to providing an automatic support they provide in bad times, in order to stabilizer (unemployment insurance), when safeguard and stimulate consumption, which organized into coherent and coordinated sys- would provide a demand stimulus to limit the tems, they protect the incomes and consump- damage from a crisis and help speed recovery. tion of households through safety nets and Overall, automatic stabilizers operating promote reemployment through active labor in the aggregate help households smooth market programs. Targeted on the basis of their consumption, reducing the immediate household need rather than whether a lost 82 employment in Crisis

job was formal or informal, these programs arrow in figure 4.1). These structural issues lessen the extent to which labor market may also be behind LAC labor markets’ adjustment translates into short- and long- poor adjustment to crises, and they might term impacts on workers (as illustrated by the require interventions at the sector and local- top arrow of figure 4.1). ity levels, in addition to worker-level and The labor scarring documented in this economy-wide interventions, and that inter- study and its adverse impact on countries’ act with social protection needs and incen- potential productivity imply that the LAC tives (as illustrated by the vertical arrows in region could achieve greater long-term figure 4.1). LAC countries’ policy responses growth if crisis-induced, worker-level human need to tackle these structural issues squarely, capital decay was reduced. This change according to the weights they hold in each would require cushioning the short-term country or setting. impact of crises through both short-term income support to protect welfare and social Aggregate: Stronger protection and labor policies to build human macroeconomic stabilizers capital and promote faster, higher-quality transitions for displaced workers moving into As illustrated in figure 4.1, the first shield new jobs. The speed and extent of scarring in against a crisis is the strength of a country’s the region require that social protection and macroeconomic frameworks and automatic labor systems provide more than just income stabilizers. These policies filter the extent support. They should also help people to to which an exogenous shock affects the renew and redeploy their human capital. It domestic labor market and, especially rele- is in this broader sense that reforms to the vant to Latin America and the Caribbean, region’s existing social protection and labor the extent to which domestic conditions policies and systems are urgently needed. can lead to a crisis situation. This section These transformations will, in turn, affect looks at how the LAC region has improved labor market flows and provide a responsive in terms of its macroeconomic frameworks, safety net that contributes meaningfully and leading to fewer domestic crises, although it effectively to countries’ automatic stabilizers, still lacks sufficient automatic stabilizers. as detailed below. Although social protection and labor pro- Stronger macroeconomic frameworks grams cushion workers from the impacts of crises, they do not address the structural Few would disagree that avoiding crises in issues that determine the magnitude of these the first place is an important priority to limit impacts in the first place. For example, this their effects, which occur at both the aggre- report highlights the dichotomy between gate and the individual levels. As documented protected and unprotected firms in the LAC in this study, the LAC region experiences fre- region (caused by the lack of contestability quent crises. During one-third of the quarters and competition, high levels of concentra- between 1980 and 2018, one or more coun- tion, and market power held by the former tries in the region was in an economic crisis group of firms) and the sluggish mobility of (as mentioned in chapter 1). However, pru- labor across economically lagging and lead- dent fiscal and monetary policies can lower ing localities, both of which serve to magnify the likelihood of certain types of crises, and the welfare effects of shocks. This study also macroeconomic policies, including demand highlights pockets of labor market rigidity stimulus interventions and exchange rate that are slowing transitions between jobs. depreciations, provide a first line of response Hence, competition policies, regional pol- to them. In recent decades, the LAC coun- icies, and labor market regulations are a tries have made important strides in strength- third key policy dimension determining the ening their macroeconomic frameworks and effects of crises (as illustrated by the bottom improving their governance and institutions. Toward an Integrated Policy Response 83

These efforts have reduced the frequency of increased the extent to which labor markets crises in the region, especially those of domes- adjust to crises along the quantitative mar- tic origin. However, some crises persist, most gin (employment). In a recent paper, Gam- notably the crisis in the República Bolivari- betti and Messina (2018) show that real ana de Venezuela but also recent political or wage flexibility declined in Brazil, Chile, economic crises in Argentina, Brazil, Haiti, Colombia, and Mexico from 1980 to 2010. and Nicaragua. This result is in line with previous research In a critical point to understanding how that has found evidence of falling real wage crises impact workers, the region’s stron- flexibility in LAC countries over this period ger macroeconomic frameworks have also (see Lederman, Maloney, and Messina altered the nature of the region’s labor mar- [2011] for the region as a whole, Messina ket adjustments. Because of these stron- and Sanz-de-Galdeano [2014] for Brazil and ger frameworks, crises in the LAC region Uruguay, and Casarín and Juárez [2015] for now occur in the context of relatively low Mexico). inflation. In contrast, the 1980s and early The resulting changes in LAC labor mar- 1990s were characterized by high infla- kets’ adjustment to economic shocks can be tion in most countries in the region. Latin illustrated by the differences in responses American monetary policies in the 1990s to crises in Brazil and Mexico during the and early 2000s were increasingly aimed 1990s versus during the 2000s. The panels at keeping inflation low (Céspedes, Chang, in figure 4.2 track fluctuations in the real log and Velasco 2014). For example, follow- gross domestic product (GDP), the inflation ing the Tequila Crisis, Mexico moved from rate, mean real wages, and the unemploy- a fixed exchange rate policy to a floating ment rate before and after the first quarter exchange rate policy and instituted rela- of negative growth (identified as t = 0 in the tively strict inflation-targeting rules. Since graphs) for four crises: the 1994 Tequila Cri- the early 2000s, most countries in the sis and the 2008–09 global financial crisis in region have succeeded in taming inflation. Mexico and the 1990 and 2015 recessions The unweighted average inflation rate in the in Brazil. The figure shows that inflation region was 69.6 percent in the 1980s and increased significantly during the Tequila 30.0 percent in the 1990s, but it fell to only Crisis in Mexico, whereas it remained flat 5.4 percent in the 2000s.3 during the 2008–09 downturn. Similarly, in Although this new macroeconomic con- Brazil, the recession of 1990 was character- text has reduced the number of domestic ized by a spike in inflation, whereas inflation crises in the LAC region, it also has impli- remained flat in the more recent crisis. cations for how the region’s labor markets Given the region’s lower inflation rates adjust to the crises that do occur. Low during more recent crises, one would expect inflation reduces the downward flexibility that real wages have not adjusted as much in of real wages, while firms’ ability to cut these crises as in earlier crises. As shown in the nominal wages of existing workers is figure 4.2, real wages fell more significantly limited by contracts (formal and informal during the earlier crises in both Brazil and agreements) and by labor legislation, such Mexico. This difference suggests that the as binding statutory minimum wages that price margin is becoming less central to labor are set high relative to average earnings.4 market adjustments. These findings are con- Hence, firms operating in contexts where firmed in Robertson (2020). inflation is low and stable cannot rely on If real wage adjustment was an important inflation to help erode real wages during a margin of labor market adjustment during crisis. Rather, firms can reduce their labor earlier growth shocks, then, given the cur- costs only through quantitative adjustments, rent context of lower inflation rates, quan- such as reducing their numbers of positions. titative adjustment has likely taken on As a result, the reduction in inflation likely increased importance, as the cases of Brazil 84 employment in Crisis

FIGURE 4.2 Wage and unemployment responses during crises in the 2000s versus crises in the 1990s, Brazil and Mexico

a. Brazil b. Mexico REAL GDP PER CAPITA 1.3 1.3 1.2 1.2 1.1 1.1 1.0 1.0 0.9 0.9

Index (equals 1 at 0.8 0.8 beginning of recession) 0.7 0.7 –5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

INFLATION RATE 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 0

Index (equals 1 at –1 –1

beginning of recession) –2 –2 –3 –3 –5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

REAL WAGES 1.8 1.8 1.6 1.6 1.4 1.4 1.2 1.2 1.0 1.0 0.0 0.8 Index (equals 1 at

beginning of recession) 0.6 0.6 0.4 0.4 –5 –4 –3 –2 –1 0 123 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5

UNEMPLOYMENT RATE

1.8 1.8 1.6 1.6 1.4 1.4 1.2 1.2 1.0 1.0 0.0 0.0 Index (equals 1 at 0.6 0.6 beginning of recession) 0.4 0.4 –5 –4 –3 –2 –1 0 123 4 5 –5 –4 –3 –2 –1 0 123 45 Number of quarters from beginning of recession Number of quarters from beginning of recession

Crises in the 1990s Crises in the 2000s

Source: World Bank. Note: Episodes of recession in the 2000s were the 2008–09 global financial crisis for Mexico and the 2015 recession for Brazil. Episodes of recession in the 1990s were the Tequila Crisis for Mexico and the 1991 crisis for Brazil. All series are indexed to the year when log real gross domestic product fell, and initial output falls at t = 0, which is indicated by dashed lines. Toward an Integrated Policy Response 85

and Mexico suggest. With inflation spik- suggests that there has been a statistically ing, unemployment increased only margin- significant reduction in the sensitivity of ally during the Tequila Crisis in Mexico. In wages and a statistically significant increase contrast, inflation remained relatively flat in the sensitivity of unemployment to output in Mexico during the 2008–09 crisis while fluctuations. unemployment grew substantially more than in the earlier crisis. In Brazil, the recession Restoring fiscal space of 1990 was also characterized by a spike in inflation and low increases in unemploy- Although the LAC region has made signifi- ment, but the recession of 2015 saw lower cant strides in reducing inflation, it contin- inflation and a more significant increase in ues to struggle with another key aspect of the unemployment.5 The changing importance macroeconomic shield—fiscal policy. Prudent of wage adjustments relative to quantitative fiscal policies prevent certain types of crises adjustments is reflected in figure 4.3, which and ensure the fiscal space needed to provide reports the sensitivity of unemployment and support and avert system-wide financial strain wages to growth shocks during crisis years when other types of crises occur. This issue is for Brazil, Colombia, and Mexico during particularly concerning in the face of the fiscal the 1990s and the 2000s. In line with the adjustment that might be needed in the LAC results shown in figure 4.2, figure 4.3 region. The region, especially the Atlantic

FIGURE 4.3 Sensitivity of unemployment and wages to output fluctuations

a. Average elasticity between real wages and output over crises

BRAZIL COLOMBIA MEXICO 1.5 1.5 1.5

1.0 1.0 1.0

erage elasticity 0.5 0.5 0.5 Av

0.0 0.0 0.0

b. Average elasticity between employment and output over crises

1.2 1.2 1.2

0.9 0.9 0.9

0.6 0.6 0.6 erage elasticity

Av 0.3 0.3 0.3

0.0 0.0 0.0 1990s 2000s

Source: World Bank calculations using data on wages and employment from Gambetti and Messina (2018) updated to 2018. Note: This figure reports the dynamic betas during crisis years estimated on the basis of Okun’s law by rolling regressions, following the methodology of IMF (2010). The difference in means between crises in the 1990s and crises in the 2000s is statistically significant at the 5 percent level for each country shown. 86 employment in Crisis

subregion, has experienced steady growth in problems with this landscape, and the policy public spending in recent years, which has agenda going forward are discussed in detail translated into sizable fiscal deficits and public in the next section. The lack of unemploy- debts (Vegh et al. 2018). Taking a longer-term ment insurance adds to the region’s lack of perspective on fiscal policy will involve tack- effective consumption-smoothing mecha- ling complex issues such as removing energy nisms, leaving an important gap in the crisis subsidies, modernizing tax policies, and rais- responses countries are equipped to deploy. ing the efficiency of social spending, including Although unemployment insurance is com- the financial sustainability of old-age pensions monly used as an automatic stabilizer, other as populations rapidly age. policies can fill this role. In the COVID-19 crisis, for example, LAC countries’ fiscal policies have been strongly countercyclical. Automatic stabilizers (and the lack Strategies like furloughs, job retention subsi- thereof) dies and the expansion of cash transfer pro- Automatic stabilizers can buffer the impacts grams have represented an important share of crises on households by increasing dis- of spending in response to the crisis. Mak- posable income, attenuating the decline in ing some of these instruments a permanent employment and consumption in response part of their economies’ automatic stabilizers to negative demand shocks. In a nutshell, could lower losses and adjustment costs in they constitute a de facto demand stimulus. the wake of future shocks. This change could The most intensively used automatic stabi- be implemented by making these programs lizers in high- and middle-income countries state-contingent and automatically activated are income support systems for those who when, for example, unemployment rises lose jobs, including severance pay (lump-sum above a determined threshold.6 Indeed, a payments at dismissal) and unemployment dynamic, or “adaptive,” system of state-con- insurance (periodic payments contingent on tingent support is one of the original motiva- nonemployment and searching for a job). tions for countries to invest in national social These policies provide liquidity to workers protection and labor “safety net” systems upon dismissal and can smooth their con- (Bowen et al. 2020; Grosh et al. 2008). sumption during their job searches. The LAC region has a long institutional Policy actions history of providing social insurance to cover threats to income and consumption The LAC region has improved significantly from old age, disability, and the untimely in terms of its macroeconomic frameworks death of households’ primary income earn- in recent decades, and it needs to continue ers. However, nationally administered, in this direction of prudence and sound countercyclical income support plans to macroeconomic management in order to cover labor market dislocation (whether pure keep up and maintain this progress. Fiscal risk-pooling unemployment insurance or policy—a key instrument for managing crises mixed saving and risk-pooling approaches) and providing demand stimulus to support are relatively rare in the LAC region. Two- recovery—remains an area of concern for thirds of the countries in the region do not the region, and the region’s more recent track yet offer national-level, countercyclical record leaves space for improvement. The income support plans for displaced work- reforms still needed involve tackling difficult ers. And among the few existing programs, issues, such as tax policies, energy subsidies, only Brazil’s, Chile’s, and Uruguay’s are suf- the efficiency of social spending, and the ficiently well established or have sufficient financial stability of old-age pension systems. coverage and payout volumes to contribute In addition, the LAC region still lacks suf- significantly to stabilizing their economies. ficient automatic stabilizers. The need for The landscape of unemployment insur- these stabilizers is made more urgent by the ance plans in Latin America, the current shift in the labor market’s main adjustment Toward an Integrated Policy Response 87

margin to quantitative adjustment. The lim- policy interventions that absorb the impacts ited availability of countercyclical, income of these systemic shocks and cushion house- support plans for people negatively affected holds from them. As documented in the pre- by labor market adjustments is making it vious three chapters, crises often bring about more difficult to manage crises and is mag- job dislocation or other negative impacts on nifying crises’ effects. Most people who lose livelihoods, which come along with long-last- their jobs (formal or informal) in downturns ing losses in earnings. In this context, having are largely unprotected. strong social protection and labor systems in Figure 4.4 provides a more complete char- place to protect people’s welfare and prevent acterization of the possible policy areas of the depletion of human capital is key. Crises focus to achieve stronger macroeconomic also generate important labor reallocation frameworks and create automatic stabilizers effects, and effective social protection and (first policy dimension). labor systems can help people redeploy into new jobs. Social protection and labor Do these systems exist in the LAC systems: Cushioning the region? To answer that question, this sec- impact on workers and tion describes the array of publicly provided preparing for change risk-sharing and coping instruments available in the region and discusses their key gaps. It The deep and lasting impacts on individuals then presents an agenda for reform to fill and economies of labor market adjustments these gaps and to increase the coherence and to crises provide a powerful rationale for coordination between interventions so that

FIGURE 4.4 Stabilizers and macroeconomic frameworks: Policy reforms

s k r o Prudent macroeconomic frameworks to avoid crises w e • Normalized in ation implies labor market adjustment on quantitative SHOCK m a employment, with long-lasting scarring r f

c i Monetary and scal stabilization policies to manage crises

o • Create scal space with a broader, long-term perspective (tax policy,

o energy subsidies, social spending eciency, and nancial sustainability of c

e pensionsystems)

r c

a Automatic stabilizers to smooth crises

• Create or reform unemployment insurance (UI)

e • Make short-time compensation (STC) programs a permanent part of the

i economy’s automatic stabilizers b

t S • Give UI and STCs the ability to adapt to changing conditions more swiftly

Source: World Bank. 88 employment in Crisis

they can operate as “systems” to cushion the various instruments fully integrated into a short-term impacts of crises, prevent lasting coherent and coordinated plan: participating human capital losses, and facilitate the rede- workers who lose their jobs make scheduled, ployment of working people through reskill- limited withdrawals from their individual ing and reemployment support. savings accounts, and a risk-pooling “solidarity fund” underpins their protection should they exhaust their unemployment savings Cushioning the short-term impact: before finding a new job. In Panama and Income support during unemployment Peru, income support for unemployment is Landscape of income support for limited to individual savings accounts, with unemployment in Latin America and no risk-pooling mechanism. Mexico (with the Caribbean the notable exception of Mexico City and Only about a third of the countries in the LAC Yucatán) and most other countries in Cen- region offer national unemployment income tral America and the Caribbean do not support plans. Job displacement income sup- have any form of unemployment insurance, port—programs specifically designed to sus- in sharp contrast with countries at similar tain the income and consumption of laid-off income levels in other regions. For example, workers and their families—in the form of all countries in Europe and Central Asia unemployment insurance is, therefore, rel- have mandatory, risk-pooling unemploy- atively rare in the region. Workers with for- ment insurance (see map 4.1). mal employment contracts in Brazil, Chile, Most countries in the LAC region rely and Uruguay have access to large risk pools instead on severance pay mandates, which offered by a national unemployment insur- are specific to the employment relationship ance plan (that is, one not specific to a work- and are financed fully and paid directly by er’s firm, occupation, or sector). In addition, firms (table 4.1). The legal coverage and Argentina, The Bahamas, Barbados, Colom- generosity of this form of protection can be bia, Ecuador, and the República Bolivariana uniform across all regulated employment de Venezuela offer unemployment insurance relationships, or it can vary by contract in the form of contributory risk-pooling plans type, by sector, or even by province. As a (table 4.1). risk-sharing instrument, the distinguishing Individual unemployment savings feature of severance pay is that it pools the accounts are also available in Chile, Colom- risk of income loss from involuntary dismiss- bia, and Ecuador. Only in Chile are these als solely within firms.

TABLE 4.1 Landscape of formal unemployment income support in the LAC region “Risk-pooling” within firms Savings (self-insurance) National risk pooling Severance pay mandates on employers Funded severance and/or individual Unemployment insurance/benefits unemployment savings accounts Most countries Argentina Argentina Brazil The Bahamas Chile Barbados Colombia Brazil Ecuador Chile Panama Colombia Peru Ecuador Uruguay Venezuela, RB Sources: Fietz 2020; Packard and Onishi 2020. Note: Argentina’s individual unemployment savings accounts are available only to registered workers in the construction sector. The government of Mexico City administers a job-seeker benefit, but only to residents and certain groups deemed vulnerable. Toward an Integrated Policy Response 89

MAP 4.1 Unemployment insurance throughout the world

Countries with a national unemployment income protection plan Available Not available No data

IBRD 45553 | FEBRUARY 2021

Unemployment insurance: Falling short of However, informal employment prac- an adequate crisis response tices are only part of the problem. Workers Job displacement support is effectively out of with formal but more precarious contracts reach for most workers in the LAC region. may be statutorily excluded from coverage Only about 12 percent of unemployed work- by unemployment income support pro- ers in the region have received unemployment grams (Fietz 2020). And even among for- benefits (ILO 2019). This rate of effective cov- mally employed workers with “standard” erage falls far below that observed in develop- employment contracts, effective coverage is ing and emerging-market countries in Central disappointingly low. Demanding eligibility and Eastern Europe and in some countries in requirements that fail to reflect the patterns Asia and the Pacific (see figure 4.5). of employment and tenure achieved even Widespread, unregulated informal by many formal workers impede effective employment practices are the principal cul- coverage. Regulatory and administrative prit for the small share of workers who can failures often mean that contributions are access unemployment income support in the not received. And the transaction costs of LAC region. This low rate of access is the securing benefits can be prohibitively high, case even in countries with comprehensive particularly if the benefits are meager. sets of unemployment insurance instruments, Beyond the exceptional case of Barbados, such as Argentina, Brazil, and Ecuador (ILO whose system delivers benefits to 88 percent 2020). For informal workers, any kind of of unemployed workers, only in The Baha- displacement income support is generally mas, Chile, and Uruguay do national unem- weak—informal urban workers often fall ployment insurance arrangements appear into poverty during crises—and 55 percent of to provide widespread, effective coverage all workers in the LAC region hold this status (figure 4.5, panel b). Outside these three (Messina and Silva 2020). countries, even in the remaining few LAC 90 employment in Crisis

FIGURE 4.5 Effective coverage of unemployment benefits, selected countries, latest available year

a. Estimates by region b. Estimates by LAC country

Africa 5.6 Barbados 88.0

Central and Western Asia 12 Chile 45.6 Latin America and Uruguay 30.1 the Caribbean 12.2

Americas 16.7 Bahamas, The 25.7

World 21.8 Brazil 7.8

Asia and the Paci c 22.5 Argentina 7.2

Northern America 28.5 Venezuela, RB 5.1

Europe and Central Asia 42.5 Colombia 4.6 Northern, Southern and Dominican Republic 4.2 Western Europe 46.2

Eastern Europe 56.6 Bolivia 3.0 0204060 80 100 020406080 100 Percent Percent

Source: ILO 2019. Note: The figures for Bolivia and for the Dominican Republic refer only to mandated severance payments. LAC = Latin America and the Caribbean.

countries that offer unemployment insur- economy (Fietz 2020). Furthermore, in coun- ance, coverage remains too low. tries that offer plans centered around individ- In addition to limited access and the result- ual savings, low-income workers with limited ing low levels of effective coverage, unem- saving capacity will find it difficult to manage ployment insurance arrangements (whether long unemployment spells. Only in Chile’s pure risk pooling or individual savings) in the national Seguro de Cesantia are parameters LAC region have three main flaws as crisis set that effectively combine protection and response mechanisms. First, the insurance positive incentives for unemployed workers value of their benefits is limited. In Ecua- to find new jobs. (Holzmann et al. 2012; dor, for example, formally employed workers Hartley, van Ours, and Vodopivec 2012). must contribute to the unemployment scheme Second, in most LAC countries that have for at least 24 months before they are eligible unemployment income support arrange- for benefits, and upon losing their jobs they ments, these arrangements are offered as are required to wait another 60 days before an array of overlapping and uncoordi- drawing on their unemployment insurance. nated instruments. For example, a formally This program follows a pattern observed employed worker in Brazil with a standard in several middle-income countries where, contract is entitled to receive severance pay, out of concern for moral hazard, the bene- an unemployment insurance benefit, and fits from the risk-pooling mechanism are set access to the full balance in the worker’s at inadequate levels or the eligibility condi- individual employer-sponsored severance tions to receive benefits are made prohibi- savings account. Access to the risk-pooling tively stringent and bear little relation to the benefit is not sequenced with savings, nor actual patterns of formal employment in the is access to the savings account limited by Toward an Integrated Policy Response 91

a schedule of withdrawals. For workers LAC countries with broad-based national who earn the statutory minimum wage— unemployment insurance plans—Argentina, or close to it—this lack of coordination Brazil, and Uruguay—spending on unem- combines with the benefits’ parameters to ployment insurance benefits is only weakly deliver a total payment upon dismissal well correlated with detrended GDP growth above what they were earning from their job (figure 4.6). (Almeida and Packard 2018; Fietz 2020). As a result, there is growing evidence of collu- Severance pay mandates: Shallow and sion between employees and employers and unreliable risk pools of induced dismissals: Pinto (2015) shows In contrast to the dearth of national unem- that the rate of dismissal spikes at the de ployment insurance, almost all LAC countries jure vesting period for unemployment insur- rely heavily on mandated, employer-financed ance and that 6 percent of those dismissed severance pay. In the early stages of their “without just cause” return to the same firm economies’ structural shift away from agri- after a period similar to the maximum pay- culture, LAC governments—like those of out period for unemployment insurance. The many developing countries—lacked the prospect of this “unemployment bonanza” capacity to collect taxes and administer has been identified as the source of Brazil’s risk-sharing programs. Mandating instead high rates of employee turnover (Da Silva that employers pay severance had three sub- Teixeira, Balbinotto Neto, and Soares Lei- stantial social advantages: (a) it discouraged vas 2020; Portela Souza et al. 2016). Brazil’s frivolous or unfair dismissals; (b) it gave unemployment insurance appears to have a employees greater bargaining power in nego- very limited consumption smoothing effect to tiations with dominant employers in what are support job searching: Gerard and Naritomi still relatively concentrated and oligopolistic (2019) show that as soon as covered unem- markets (tending toward monopsony in the ployed workers in Brazil receive the benefit, rural, large-scale agriculture and resource their consumption spikes, although their extraction sectors); and (c) it provided house- displacement still generates a long-term loss holds with some protection from destitution of consumption of around 14 percent. at a time when social assistance transfers Third, when poorly designed and unco- were rarely provided or meager in size. ordinated, unemployment insurance has However, as the capacity of LAC gov- contributed to perverse aggregate out- ernments to collect taxes and administer comes, delivering only a muted “stabilizer” risk-sharing arrangements has grown, the response during downturns and surges in deficiency of mandated severance pay as spending during sustained periods of eco- the sole or even the primary instrument of nomic growth. If quantitative adjustments income protection for unemployment has have come to predominate labor markets’ become apparent. The deficiencies of sever- responses to crises and are leading to long- ance pay are particularly clear in the context of run consequences through the scarring they systemic shocks such as the LAC region’s cri- cause, the lack of accessible, state-contingent ses, whose impacts overwhelm these relatively unemployment support programs becomes shallow, firm-level risk-pooling arrangements. an even bigger problem. As automatic stabi- In Argentina, according to workers’ reported lizers, these programs should also contribute reasons for their employment ending, of the to countercyclical fiscal policy. Indeed, in 22.8 percent who reported firm bankruptcy as many OECD countries, spending on unem- the reason in 2018, only 33.1 percent actually ployment insurance and other transfer pro- received severance pay (World Bank 2020). grams automatically falls in good times and This rate is about the same as in 2010 and increases in bad times (as unemployment and is an improvement on earlier coverage rates. poverty experience their cyclical increases), However, it leaves two-thirds of the workers cushioning workers. By contrast, in three covered by law effectively without the income 92 employment in Crisis

FIGURE 4.6 Economic cycle, unemployment, and spending on labor policies and programs

a. Argentina, 2012–19 0.20 12

0.18 10 0.16 8 0.14 0.12 6 P 0.10 4 % GD 0.08 2 Annual % 0.06 0 0.04 0.02 –2 0.00 –4 2012 2013 2014 2015 2016 2017 2018 2019 Spending on active LMP (left axis) Spending on passive LMP (left axis) Unemployment rate (right axis) GDP growth (right axis)

b. Brazil, 2000–19

1.4 14 12 1.2 10 1.0 8 6 P 0.8 4

% GD 0.6

2 Annual % 0.4 0 –2 0.2 –4 0.0 –6 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Spending on active LMP (left axis) Spending on passive LMP (left axis) Unemployment, total (right axis) GDP growth (right axis)

c. Uruguay, 2009–19 0.6 10 9 0.5 8 0.4 7 6 P 0.3 5 % GD

4 Annual % 0.2 3 0.1 2 1 0.0 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Active LMPa (left axis) Passive LMPb (left axis) Unemployment rate (right axis) GDP growth (right axis)

Source: Social Protection and Jobs Global Practice, with data from Brazil’s Ministry of the Economy, Argentina’s Administración Nacional de Seguridad Social and Ministry of Labor and Employment, and Uruguay’s Banco de Previsión Social. Note: GDP = gross domestic product; LMP = labor market programs. a. Projoven, Proimujer, Trabajadores en Seguro Desempleo, etc. b. Seguro de Desempleo. Toward an Integrated Policy Response 93

support during unemployment they were available in Latin America, two were intro- promised. In countries where the labor code duced following the crises of the late 1990s. and specialized labor courts impose the bur- Past crises and the 2020 pandemic shock den of proof on firms in disputes over sever- dramatically demonstrated the usefulness of ance pay and where severance liabilities are having unemployment income support sys- borne by business owners rather than busi- tems with deep and extensive risk pools and nesses themselves, large severance mandates that provide a platform and channel for addi- combine with restrictions on contracting and tional, extraordinary support measures when dismissals to have a chilling effect on new for- these are needed. The pandemic crisis is likely mal job offers (Holzmann et al. 2012). This to motivate proposals for national unemploy- effect can partly explain the LAC region’s pat- ment insurance plans in the countries that do tern of formal employment shrinking during not have them yet and proposals to expand crises mainly through a reduction in new job coverage in those that do. Augmenting unem- offers. ployment support through the creation or Relying solely on severance pay or redesign of unemployment insurance is there- unemployment insurance is not as good fore a key policy action to be considered. of a risk-sharing arrangement as using a In Latin America and the Caribbean, combination of both (Schmieder and von several countries have recently implemented Wachter 2016). More reliable, robust, and changes to social insurance plans that ease incentive-compatible income support for job eligibility requirements and increase benefits displacement can be achieved when employ- (ILO 2020). In Brazil and Chile, for exam- ers’ severance obligations are prefunded ple, the unemployment insurance system has and prudently managed; when their regu- also served as the platform for implement- lar contributions to this prefunding can be ing subsidized furlough measures and other made transparently in the form of individ- employee-retention programs. These systems ual savings accounts (preferably managed make all the difference in the quality of labor by independent third parties); and when this markets’ adjustment to crises. For example, prefunding can be integrated with a larger, in the United States, extensions of unemploy- country-level risk-pooling mechanism, as is ment insurance benefits introduced during the case in several OECD member countries. deep recessions have been shown to improve Many countries, including Chile in Latin the quality of worker-job matches—and the America (as discussed in detail later in this impact on match quality is greater among chapter), combine severance, individual sav- people who are more likely to be liquidity ings, and risk-pooling plans into effective constrained, such as women, nonwhites, and protection systems that discourage frivolous less educated workers (Farooq, Kugler, and dismissals, eliminate collusive gaming of the Muratori 2020). In the pandemic-induced system by firms and employees, tolerate sys- contraction of 2020, the United States has temic as well as idiosyncratic shocks, and used unemployment insurance top-ups and strongly incentivize job search efforts (Cas- benefit extensions more intensively than sub- tro, Weber, and Reyes 2018; Robalino and sidized furloughs or other retention subsidies, Weber 2014). more so than in other OECD member countries (Furman et al. 2020). Given the burden What can be done? it forces onto individual firms, as well as the A history of frequent systemic shocks com- possibly irrecoverable loss of viable busi- bined with the emergence of a significantly nesses and business-specific human capital, sized middle class has created more demand this policy choice is controversial. But the for robust unemployment insurance mecha- expansion of access to (already widespread) nisms in Latin American countries than in unemployment insurance has contained the other regions (De Ferranti et al. 2000). Among human costs of this sharp contraction and the handful of national unemployment plans prevented its unprecedented rates of job 94 employment in Crisis

destruction from leading to destitution (Fur- often result in poor employment matches; man et al. 2020). and (d) the unanticipated costs incurred out Why have even administratively sophis- of pocket or from postponing health and ticated middle- and upper-middle-income education investments by unemployed people countries in the LAC region held back from without effective income support as they look offering nationally administered income sup- for new employment. port plans for the unemployed? Setting aside Second, in the LAC region, the reluctance the valid concerns that arise in most coun- to augment unemployment income support tries over moral hazard and other distortions is often anchored in the argument that such to incentives, the reluctance to offer unem- a scheme is unnecessary given the ample ployment income support is usually based on work opportunities in the informal economy three arguments. First is a fiscal justification: and the expectation that informal employ- offering unemployment insurance could be ment operates as a countercyclical income costly, especially in a time of sizable fiscal safety net. Historically, Latin American deficits, and the contingent liabilities could labor markets have been characterized by be fiscally explosive in light of the region’s high levels of informality (Perry et al. 2007). relatively frequent crises. On average, the Indeed, the LAC region is more informal LAC countries that offer unemployment than expected given its level of development insurance spend about 0.3 percent of their (Robertson 2020). However, changes in the GDP on these plans. In the OECD, spending size of the region’s informal economy are on such plans ranges from about 0.3 percent not always countercyclical, nor is relying on to 1.8 percent of GDP.7 This level of spending informal employment as a safety net with- on a single social protection program could out costs: growing evidence shows that spells indeed be prohibitive for some LAC coun- of informal employment can cause scarring tries, given their smaller tax bases, limited (Cruces, Ham, and Viollaz 2015). Further- enforcement capacity, and ongoing struggles more, the 2000s saw important growth to extend poverty-relief programs and human in the number of formal jobs in the region. capital–building services. The trade-off is Between 2002 and 2015, the share of formal more acute for the low-income countries in employees in total employment increased the region. However, evidence suggests that from 47 percent to 55 percent. This change the introduction of some key features can was caused by a reduction in the share of reduce the cost of such programs while still self-employed, from 24 percent to 20 percent, keeping their level of support and respon- and of informal employment from 29 percent siveness to crises at acceptable levels. Chile’s to 25 percent. In contrast, between 1995 Seguro de Cesantia is an exemplary case of and 2002, the share of self-employed work- a well-established national income support ers had remained stable (Messina and Silva plan in Latin America that is responsive in 2020). Hence, although informality remains times of crisis and financially robust.8 high in the region, a large share of the labor Furthermore, concerns for fiscal costs in market is formal in most LAC countries. This the immediate term can be myopic, ignoring shift has changed the aspirations of the Latin the true, full costs of prolonged labor market American middle class. adjustment and forgone productivity from: Moreover, as pointed out by Antón, Trillo, (a) workers clinging to jobs that are no longer and Levy (2012) and by Levy (2018), fur- viable or that they are not particularly good ther formalization might be discouraged by at for fear of losing acquired rights to sever- the region’s current architecture of social ance; (b) firms’ incentives to dismiss newer, protection and labor systems. These sys- younger, possibly better-skilled employees tems may not yet offer the protection that rather than older workers with costlier sev- working people really value, making them erance entitlements; (c) job searches driven a tax that workers and firms seek to avoid solely by the urgent need for income, which or evade. Although the extent of mandated Toward an Integrated Policy Response 95

nonwage costs in most LAC countries—the objectives of the country’s social protec- “tax wedge”9—leaves little room to add ben- tion and labor regulation from protecting efit programs, the perception that the “effec- jobs to protecting working people, facil- tive” benefits of formality are low can lead to itating the reallocation of workers into increased informality. more productive occupations (Pinelli et al. The third argument for the reluctance to 2017). Labor reforms in France from 2016 augment unemployment income support is to 2018 substantially reduced restrictions that it is not necessary in countries where on the dismissal of permanent workers labor laws make employee dismissals almost and brought the mandatory total finan- impossible. Indeed, some of the world’s exist- cial compensation for dismissed working employment protection laws and insti- ers in 2018 below the average for OECD tutions were designed when displacement member countries, comparable to the level from formal employment was relatively rare. offered in the Scandinavian countries But in a growing number of middle-income (OECD 2019). and high-income countries, the emphasis of Finally, in the current crisis, several unemployment support programs has shifted countries that have well-established unem- from preserving employment relationships ployment insurance systems have offered to protecting working people as they tran- retention subsidies rather than direct support sition from job to job. Pursued most ambi- of workers through expanded unemployment tiously by several countries in Europe, this insurance. The argument for doing so is that policy shift entails loosening restrictions on if workers are displaced, human capital, and dismissals, aligning protections for workers therefore long-term growth potential, may on different types of employment contracts, be permanently lost. The magnitude of the and considerably augmenting unemployment human capital destruction (scarring effects) income support and public employment ser- avoided thanks to these programs depends on vices. The general direction of reform has (a) the estimated losses in productivity caused been to shift from protecting workers from by periods of unemployment or nonemploy- change to supporting them for and through- ment; (b) the unemployment permanently out labor market changes. In Europe, several averted, that is, the workers in the employ- countries have combined reforms of labor ment retention program who would other- regulations with improvements in the reliabil- wise have been dismissed (either directly or ity of non–firm-specific income support for indirectly when their firm went bankrupt or unemployment. closed for lack of liquidity); and (c) the unem- This shift has occurred in several coun- ployment temporarily averted, that is, the tries. For example, from 2011 to 2015, workers who are supported now by the pro- Portugal reduced its severance mandates gram but will be let go after the program sup- and considerably improved the reliability port period, or even before then if their firm of non–firm-specific income support for goes bankrupt. In terms of the government’s unemployment as it eased the restrictions costs per worker, without this program, and penalties around employee dismiss- unemployment insurance would have to be als (OECD 2017). Labor policy reforms paid in full to each laid-off worker. With in Spain in 2012 took the country’s the employment retention program, how- worker protection, unemployment, and ever, some of the costs are supported by the reemployment support in a similar direc- firm (and part by wage reductions assumed tion, substantially loosening restrictions by workers). In addition to the size of these on dismissals and severance entitlements, programs, a second key choice when design- although these entitlements remain some ing them is their duration and their coordi- of the highest among OECD member nation with unemployment insurance. Costs countries (OECD 2013). Italy’s labor and potential distortions to the economy policy reforms in 2014–15 also shifted the increase as the effects of a crisis extend from 96 employment in Crisis

transitory to permanent or accelerate struc- responses to crises in the LAC region (Grosh, tural changes that were only incipient prior Bussolo, and Freije 2014; Packard and Oni- to the crisis. shi 2020). Although these programs are not meant to serve as the lone source of insur- Social assistance cash transfers: A vital but ance nor to act as a channel for large-volume overstretched source of income support in fiscal stimulus, even in the LAC countries crises that have national unemployment insurance In the LAC region’s context of pervasive plans, governments responded to the global informality and difficult-to-access or missing financial crisis of 2008–09 by rapidly and unemployment insurance, social assistance substantially expanding these programs transfers have become the key instruments (Grosh, Bussolo, and Freije 2014), much as to smooth the consumption of most people they have done in 2020 in response to the coping with livelihood shocks and to protect COVID-19 pandemic (Gentilini et al. 2020). them from impoverishment or deeper pov- For people who work informally and their erty. Conditional cash transfer (CCT) pro- dependents, social assistance transfers are grams have become a foundation of social vitally important—and they are typically protection systems in most LAC countries; the only access these people have to an effec- consequently, they are intensively used for tive, efficient risk-pooling mechanism, albeit crisis response. Introduced in the late 1990s, through governments’ general tax and expen- they have grown considerably since the turn diture systems rather than through explicit of the century (World Bank 2015). Despite social insurance arrangements (Packard a substantial increase in spending and inno- et al. 2019). But as consumption-smoothing vations in targeting and delivery, these instruments in the wake of a shock to live- programs remain a resource primarily for lihoods, these transfers fall short. This households living in poverty or close to the should not come as a surprise, because most poverty line. Another prominent feature of of these programs were not set up for crisis LAC countries’ safety nets is their “uncondi- management, even if they have succeeded in tional” cash transfers, most of which are also preventing poor people from falling deeper targeted. In addition to poverty-targeted pro- into poverty, divesting precious assets, grams, many governments in the region have or postponing human capital investment added “categorical” transfers (such as child (World Bank 2018). allowances and social pensions)—cash pay- In general, existing cash transfer programs ments targeted to all members of dependent have four structural limitations that makee it groups, such as children, the elderly, and peo- difficult to quickly increase the support they ple of all ages living with disabilities. offer in response to a shock: (a) tight poverty Figure 4.7 presents a breakdown of targeting using “static” beneficiary registries, selected LAC countries’ spending on social (b) meager benefit amounts relative to recip- assistance transfers by program type. In all ient households’ incomes, (c) de jure budget countries featured except Colombia and Nic- restrictions on the number of eligible house- aragua, where spending on price subsidies holds who are admitted to the programs, dominates, CCTs and other cash and food and (d) LAC countries’ lagging investment in transfers are the largest categories of spend- identification and delivery systems. ing on social assistance. For people who Tight poverty targeting and static identi- are informally employed or who earn their fication systems: Most social assistance cash livelihoods in the informal economy, these transfer programs in Latin America were “noncontributory” transfers are the only designed to alleviate poverty rather than protection. In this context, it is not surpris- prevent impoverishment. As such, eligibil- ing that social assistance and labor-inten- ity for them is based on whether the house- sive public works (“workfare”) are the main hold is already poor and whether it includes Toward an Integrated Policy Response 97

FIGURE 4.7 Level and composition of government spending on social assistance transfer programs, selected LAC countries

a. Expenditure as a share of GDP per capitab. Program composition of total expenditure 3.5 100 ) 3.0 80 2.5 ogram (% 2.0 60

1.5

% of GDP 40 1.0

e of spending by pr 20 0.5 Shar 0.0 0.0

Chile, 2015 Peru, 2015 Chile, 2015 Peru, 2015 Bolivia,Brazil, 2015 2015 Brazil, 2015 Ecuador, 2015 Jamaica,St. Lucia, 2018Mexico, 2013 2015Panama, 2015 Bolivia, 2015 Ecuador, 2015 Jamaica,St. Lucia, 2018Mexico, 2013 2015Panama, 2015 Granada, 2015Honduras, 2014 Uruguay, 2015 Granada, 2015 Uruguay, 2015 Argentina, 2017 Colombia,Costa 2015 Rica, 2014Guatemala, 2013 Nicaragua,El 2013 Salvador, 2014 Argentina, 2017 Colombia,Costa 2015Rica, 2014Guatemala,Honduras, 2013 2014 Nicaragua,El 2013 Salvador, 2014

Dominican Republic, 2018 Dominican Republic, 2018 CCT UCT Social pension School feeding Public works In kind Fee waivers Other SA

Source: World Bank ASPIRE data set: https://www.worldbank.org/en/data/datatopics/aspire. Note: The data shown are from the latest year for which comparable data are CCT = conditional cash transfer; GDP = gross domestic product; LAC = Latin America and the Caribbean; SA = social assistance; UCT = unconditional cash transfer.

members in a dependent group (such as chil- livelihoods above but close to the poverty line dren, the elderly, or the disabled). (De Ferranti et al. 2000; Jalan and Ravallion Although these programs, as they have 2003; Subbarao et al. 2013). These programs been designed and administered by most were the de facto unemployment insurance countries in the LAC region, provide a life- for the majority of such workers worldwide line to many households, they are not the best during the crises of the 1980s and 1990s, and suited to helping households manage the risks even in the global financial crisis of 2008–09 associated with transitory systemic shocks. in several European countries (Packard and The programs cover too few households even Weber 2020). But outside Argentina and among the lowest-earning segments of the Chile, the capacity to deploy these programs population; their registries and delivery sys- quickly and effectively is still limited in the tems cannot cope well with rapid increases in LAC region. the number of households that need support; Meager benefits: Social assistance cash and their benefit amounts are low relative to transfers are typically designed to comple- conventional levels of consumption-smooth- ment rather than replace earned income. ing social insurance (figure 4.8). They are typically much lower than the level The recent policy push to make social of income replacement conventionally con- assistance cash transfer programs respon- sidered adequate for consumption smooth- sive (“adaptive”) to natural disasters (Bowen ing (anywhere from 40 to 70 percent of et al. 2020; Williams and Berger-Gonzalez prior income). In light of the explicit aim of 2020) has substantially loosened these con- enabling and even incentivizing households straints. In earlier crises, labor-intensive to make human capital investments (such as public works—also known as “workfare” taking up and maintaining good nutrition, programs—helped smooth the consumption timely preventative health care, and regular of some workers, mainly those employed school attendance), these benefit amounts in the informal economy and earning are low relative to the households’ earned 98 employment in Crisis

FIGURE 4.8 Insufficient support, with many left behind

a. Conditional cash transfer coverage b. Amount of bene ts received

Uruguay 2012 Bolivia 2012 Bolivia 2012 Honduras 2013 Mexico 2012 Ecuador 2016 Philippines 2015 Mexico 2012 Brazil 2015 Argentina 2013 Argentina 2013 Jamaica 2010 Brazil 2015 Colombia 2014 Paraguay 2011 Dominician Republic 2014 Panama 2014 Chile 2013 Peru 2014 Average, 15.6 Ecuador 2016 Costa Rica 2014 Peru 2014 Dominician Republic 2014 Average, 40.3 Panama 2014 Colombia 2014 Guatemala 2014 Chile 2013 Costa Rica 2014 Philippines 2015 Honduras 2013 Bangladesh 2010 Jamaica 2010 Paraguay 2011 Timor-Leste 2011 Timor-Leste 2011 Bangladesh 2010 02040 60 80 100 020 40 60 80 100 % of population in the poorest quintile % of bene ciaries’ post transfer welfare among the poorest quintile

Source: World Bank 2018.

income. A large body of research suggests benefits to be loosened or lifted altogether. that, on the whole, these benefits support Because these rations are politically and even positive incentives to return to work (Fisz- legislatively cumbersome to loosen quickly, bein and Schady 2009; Garganta and Gaspa- and because policy makers have many other rini 2015). However, this finding may imply demands to manage with limited fiscal space, that the amounts transferred are insufficient it is difficult for social assistance cash trans- to robustly smooth consumption when liveli- fers, as they exist today in most countries in hoods are destroyed by a crisis.10 the LAC region, to substitute for coverage by Rationed benefits: Few social assistance unemployment insurance plans. Argentina’s cash transfer programs in Latin America and family allowance program is an instructive the Caribbean are entitlements. This distinc- exception (box 4.1). tion matters because budget allocations for Lagging investment in identification and most of these programs are discretionary, delivery systems: The foundations of most imposing limits on the amount of benefits social assistance programs are identification that can be paid out to eligible households in databases, known in most countries as social any given year. Being eligible does not guar- registries, which enable these programs to antee entry; there have to be slots available in identify those in need. However, the share the program, which depends on other house- of the population covered by these registries holds’ entry and exit into the program and on in the LAC region is low and limited to the the government’s budget. In Argentina and chronically poor and vulnerable. In the 2020 Chile, this aspect of the programs has been pandemic, the LAC countries have relied changed. But even in Brazil—which is famous heavily on cash transfers to get money into for its Bolsa Familia CCT program—prior to the hands of vulnerable people quickly, some the recent expansion in coverage in response in more effective ways than others. A key to COVID-19, more than one million eligi- determinant of the success of these efforts ble families were waiting for the rations on is the share of the population covered by the Toward an Integrated Policy Response 99

BOX 4.1 Family allowances as de facto unemployment insurance

Argentina’s safety net is particularly effective at mit- make it difficult for workers to gain coverage under igating impoverishment, especially among families that program, and even if they do, benefit levels have with children and those that depend on informal live- been allowed to deteriorate. Additionally, Argenti- lihoods (World Bank 2020). For a growing majority na’s labor code mandates that employer-financed of Argentine working people and their dependents, severance be paid to workers dismissed because family allowances, specifically the noncontributory of economic difficulties or the insolvency of a Asignación Universal por Hijo (AUH), have become firm. However, in 2018, only a third of people de facto unemployment insurance. Indeed, for who reported losing a formal job for these reasons microentrepreneurs, other self-employed workers, received severance pay, about the same share who and informal employees with children, AUH is effec- received severance in 2009. tively the only rapidly available means of sustaining The registration, identification, and delivery consumption in the wake of shocks. systems developed by the Argentine government to But even for people who have formal jobs, AUH quickly shift working people and their families from can act as a more reliable instrument for consump- the country’s contributory family allowances to the tion smoothing than Argentina’s official unem- AUH are, for this reason, serving a vital unemploy- ployment support program. This is ironic, because ment insurance function that the country’s other Argentina is one of the few countries in Latin Amer- social protection instruments are no longer able to ica to offer nationally administered unemployment serve. income protection. A growing scarcity of formal job offers and a trend toward shorter employment spells Source: World Bank 2020.

social registry of the social safety net pro- Years of investment in developing infor- gram, which ranges from almost 100 percent mation management and benefit delivery of the population in Argentina and Uruguay systems are making the LAC region’s cash to about 5 percent in Bolivia (figure 4.9, transfer programs more responsive. Figure panel a). When a greater share of the pop- 4.10 illustrates the expansion of cash trans- ulation is covered by these social registries, fer programs in response to crises in Latin governments are better able to expand bene- America. At the time of the global financial fits to the newly poor and newly vulnerable. crisis, countries that had CCTs and other Another determinant of success is whether transfer programs expanded these programs these registries have dynamic intake systems “vertically” (by increasing the amount paid that allow programs to quickly expand to by the program to existing recipients) and include previously nonpoor groups. Coun- “horizontally” (by extending coverage to tries with initially weaker safety nets were previously uncovered households). Brazil did less able to provide robust income protec- both by expanding the coverage of Bolsa tion through this route (figure 4.9, panel Familia to a total of 12 million families and b). In general, the LAC region lacks reliable increasing the benefit amount by 10 percent. and robust income protection, in addition to Colombia’s government expanded the sufficient job search support to curb human coverage of Familias en Acción to new capital losses (as will be discussed in the fol- households, and Mexico’s Progresa (which lowing subsection). later became Opportunidades and then Pros- How can the LAC region do better for pera) increased the amount paid to existing its workers and communities in terms of beneficiaries (Grosh, Bussolo, and Freije its social protection and labor responses to 2014). With the benefit of greater experience, crises? governments in the region have repeated this 100 employment in Crisis

FIGURE 4.9 Coverage of social registries and support received through social assistance programs during the COVID-19 (coronavirus) pandemic

a. Population covered by social registries b. Responsiveness of cash transfer programs to address income and job losses during lockdowns 100 100

80 80

60 60 ent ent rc rc Pe Pe 40 40

20 20

0 0

Chile MexicoEcuador ColombiaHondurasCosta Rica Paraguay Chile, RSH Ecuador, RS Guatemala El Salvador Haiti, SIMAST Mexico, SIFODEHonduras, RUP Colombia, SISBEN Uruguay, SP Registry Dominican Republic Argentina, SP Registry Brazil, Cadastro Unico Bolivia, PREGIPS (hh 2015) Dominican Republic, SIUBEN Dynamic registries (for citizen enrollment) Households experiencing a decrease in wage income Static registries (census enrollment) in the past 12 months Respondents receiving government assistance after job loss or reduced wage income

Sources: Morgandi et al. 2020; World Bank High Frequency Phone Surveys, first and second rounds, Poverty and Equity Global Practice. Note: The medium blue bars in panel a denote dynamic registries that are open to citizen enrollment, enabling rapid expansion; the dark blue bars denote static registries populated by census sweeps. hh = household survey; PREGIPS = Registro Integrado de Programas Sociales del Estado Plurinacional de Bolivia (Bolivia´s system for selecting beneficiaries of social programs); RS = Registro Social (Ecuador´s system for selecting beneficiaries of social programs); RSH = Registro Social de Hogar (Chile´s system for selecting beneficiaries of social programs); RUP = El Registro Único de Participantes (Hondura´s system for selecting beneficiaries of social programs); SIFODE = Sistema de Focalización de Desarrollo (Mexico´s system for selecting beneficiaries of social programs); SIMAST = Information System of the Ministry of Social Affairs and Labour (Haiti’s social registry database); SISBEN = Sistema de Selección de Beneficiarios de Programas Sociales (Colombia´s system for selecting beneficiaries of social programs); SIUBEN = Sistema Único de Beneficiarios (Dominican Republic´s Colombia´s system for selecting beneficiaries of social programs); SP = social protection.

FIGURE 4.10 Expansion of cash transfer programs in response to crises strategy to rapidly provide support to help households through the economic impact of the COVID-19 pandemic (Morgandi et al. 2020). Box 4.2 discusses Brazil’s response to Vertical expansion the pandemic along these lines. As argued earlier, although Latin America’s national social assistance cash Temporarily increased transfer programs fall short of the ideal bene t countercyclical safety net, they still have amount amount considerable social and economic value in a crisis. What are the effects of the expansion

Benefit of these programs? Can such cash transfers Regular social Regular protection system Horizontal perform a “stabilizer” function? On the basis bene t(s) parameters expansion of a rare quasinatural experiment, Gerard, amount Naritomi, and Silva (2021) show that a social assistance program’s expansion has aggregate benefits (for the entire local economy) Core bene ciaries of Those not in receipt of in addition to individual-level benefits. These the social protection regular bene ts, but positive effects on employment and income system aected by a shock work as automatic stabilizers for the econ- Population omy and counter crisis-induced inequality. Social assistance cash transfer programs Sources: Bowen et al. 2020; Morgandi et al. 2020; Williams and Berger-Gonzalez 2020. Toward an Integrated Policy Response 101

BOX 4.2 Brazil’s social protection response to the COVID-19 (coronavirus) pandemic

Social protection and labor measures were at ance program, and (b) by advancing payments of the center of the fiscal response package to the regular entitlements, including special withdrawals COVID-19 crisis in Brazil. These measures explic- from the FGTS (Fundo de Garantia do Tempo de itly targeted several vulnerable groups, includ- Serviço) employer-sponsored savings accounts. ing the preexisting poor, families working in the As a second line of response, Brazil launched two informal economy who became temporarily poor temporary social protection and labor programs because of the crisis, low-income single mothers, to address specific vulnerabilities in the formal and and formal workers at risk of losing their income informal labor markets generated by COVID-19 that from dismissal (figure B4.2.1). The measures had were not covered by the expansion of existing pro- two objectives: to make social distancing possible grams. One program was an emergency cash trans- for those economically affected by it and to mitigate fer program for the poor (defined as participants in the negative impacts of the crisis on welfare and the Bolsa Familia program) as well as those outside human capital. formal wage employment but normally ineligible for The first line of Brazil’s policy response expanded social assistance, such as nonpoor informal workers the existing flagship social protection and labor pro- and formal freelance self-employed workers. And grams, including the Bolsa Familia conditional cash for formal sector workers, Brazil introduced a tem- transfer and unemployment insurance, both hori- porary wage subsidy for those who were furloughed zontally (by adding new beneficiaries) and vertically or had their hours temporarily reduced, under the (by providing greater benefits to existing beneficia- requirement that their firms maintain the employ- ries). These goals were achieved in two ways: (a) by ment relationship for a certain amount of time after relaxing budget constraints so that already eligible the program ends. These measures targeting workers and newly eligible families could gain coverage by were complemented by subsidies to firms and other Bolsa Familia through the automatic absorption of measures; altogether, they have amounted to 4.1 per- new claims submitted to the unemployment insur- cent of Brazil’s gross domestic product.

FIGURE B4.2.1 Brazil’s COVID-19 (coronavirus) social protection and labor response strategy for two major vulnerable groups COVID-19 Evolution of

Time

Low-wage formal Wage subsidies for furloughed workers at risk of formal workers dismissal Unemployment Insurance

Informal workers at risk of poverty and Emergency cash transfers the extreme poor for informal workers Bolsa Familia

Time

Source: World Bank 2020. Note: This figure illustrates Brazil’s response to protect two major vulnerable groups from the COVID-19 crisis: low-wage formal employees and low-income families working in the informal economy. 102 employment in Crisis

inject funds into local economies, potentially increase in formal employment could occur raising the demand for labor, including in the through either the creation of new jobs or the formal sector. The paper links administra- formalization of jobs that were previously tive records on the universe of Bolsa Familia informal (although in the latter case, the recipients and formal workers in Brazil to program expansion would not be associated provide evidence of the program’s effects on with an increase in overall employment). The formal labor markets. Using variation across program’s expansion did in fact have positive municipalities in the program’s expansion and significant effects on GDP (figure 4.11, in 2009, the paper finds that this expansion panel b), suggesting that the expansion led to increased formal employment. Its evidence is employment creation, not just the conversion consistent with the large multiplier effect of of informal jobs to formal jobs. Local gross Bolsa Familia benefits, which dominate the products increased by 1.5 percent as a con- negative effects on the formal labor supply sequence of the program expansion. These at the individual level, as is also documented results highlight the importance of account- using variation caused by income eligibility ing for both the individual and the aggregate thresholds. effects of social assistance welfare programs Importantly, Gerard, Naritomi, and Silva in policy debates. (2020) also show that the program expan- To deliver emergency social transfers in sion had positive aggregate effects beyond response to crises and to maximize their pos- its effects on individual beneficiaries through itive effects, the LAC countries must ensure spillovers to nonbeneficiaries. Taking advan- their cash transfer programs are sufficiently tage of their linked data on formal employ- responsive and adaptive to the needs created ment and the poor and vulnerable, they by systemic shocks, and they must substan- investigate whether the additional formal tially augment their delivery capacity. Key to employment they observe comes from ben- this change is expanding population registries eficiaries or nonbeneficiaries. Panel a of to cover all the poor and vulnerable— figure 4.11 presents their results: the effect indeed, extending the registries as far up the of the program’s expansion is positive and income distribution as is practicable—and significant among noneligible families. This sharing their information among all social

FIGURE 4.11 Positive effects of welfare transfers on local formal employment

a. Spillovers within municipalities: E ect on formal employment among nonbeneciaries b. E ect on local GDP per capita

0.06 0.06 m

gra 0.04 0.04

0.02 0.02 amilia p ro 0.00 0.00

–0.02 –0.02 t of an (exogenous) expansion t of an (exogenous) of the Bolsa F

E ec –0.04 –0.04 2007 2008 2009 2010 2011 2012 2007 2008 2009 2010 2011

Source: Gerard, Naritomi, and Silva 2020. Note: This figure shows the effect of a 2009 expansion of the Bolsa Familia program on the number of formal employees among nonbeneficiaries (panel a) and local GDP per capita (panel b). The vertical lines represent 95 percent confidence intervals on the basis of robust standard errors clustered at the municipality level In panel a, the data are from the first quarter of all years shown. Toward an Integrated Policy Response 103

programs rather than keeping registries in Latin America. In many cases, because program-specific. Moreover, because most of these programs’ administrative capacity LAC countries use social assistance trans- and broad registries, they are one of the few fers as an instrument in crisis response, these options to deliver benefits to the population countries’ governments must better manage quickly. However, the support they provide the processes of enrollment, registration, in response to crises is insufficient, and many and recertification for these transfers and are left behind, because these programs are make these processes rapid and effective. The targeted at the preexisting poor. Going for- administrators of the region’s national cash ward, it will be important to improve the transfer programs aspire to make more flex- capacity of these programs to countercycli- ible enrollment and exit processes for benefi- cally increase their level of benefits and their ciaries so that previously nonpoor families can coverage of vulnerable populations.12 receive benefits when needed and so that those There are three main policy priorities to whose incomes grow beyond the programs’ improve the dynamism of social assistance eligibility thresholds have every incentive to cash transfers. The first is to improve the move into sustainable, productive livelihood adaptability of these programs—that is, activities. However, because CCT programs their capacity to be responsive to households are designed to address chronic poverty, suffering the impacts and repercussions of they typically have lengthy intake processes various shocks, including hurricanes, earth- to identify, enroll, and recertify beneficiaries quakes, and tsunamis as well as economic that are implemented uniformly across all crises. This reform will include establishing prospective beneficiary households.11 comprehensive and dynamic social registries Thus, although the LAC countries with that are usable by all social programs, such national CCT programs and other cash trans- as Brazil’s Cadastro Unico (Lindert et al. fer programs have used them to effectively 2020). The second priority is to move from help households adjust to shocks, these pro- budgeted programs and rationed cupos to grams are not yet a sufficiently nimble policy protection guarantees, that is, from merely instrument to address the needs of poor and assisting the chronically poor to building nonpoor vulnerable groups during transient safety nets that can expand to catch all who systemic shocks (such as financial crises or are vulnerable to impoverishment before they long recessions). Because of the demands they become poor. (Packard et al. 2019). And the make on administrative capacity, cash trans- third is to prevent the emergence of assistance fer systems on their own are not yet effective “ghettos” by structuring benefits to incen- substitutes for fuller national risk-sharing tivize the return to work (with support from arrangements. augmented reemployment services). Crises are very costly for some workers, The COVID-19 global pandemic has and for most people, policy responses to spurred governments to quickly enact many these crises have failed to compensate for parts of this agenda (see box 4.3). Many of the costs or to offer effective remedies. One the needed changes and additions to the LAC key reason for this failure is that social pro- region’s social protection and labor systems tection and labor systems in the LAC region were already under way prior to the 2020 are not yet fully in place, so they are certainly pandemic, especially in the countries more not yet able to provide a dynamic safety net vulnerable to climate change and other nat- that responds robustly to shocks and crises ural disasters (Bowen et al. 2020; Williams (Packard and Onishi 2020; Williams and and Berger-Gonzalez 2020). Many of the Berger-Gonzalez 2020). changes that make a social protection system responsive to households suffering the What can be done? impacts of natural disasters also improve Social assistance cash transfers are an import- the system’s function as part of a country’s ant source of income support during crises automatic stabilizers against other systemic 104 Employment in Crisis

BOX 4.3 Latin America and the Caribbean’s social protection and labor responses to the COVID-19 (coronavirus) contraction of 2020

Gentilini et al. (2020) have kept a record of the middle”—with generous one-off emergency transfers. social protection and labor measures taken by Latin Among the largest of these transfers are the Ingreso American and Caribbean (LAC) countries since the Familiar de Emergencia programs launched in Argen- COVID-19 (coronavirus) pandemic was declared in tina and Chile. early March 2020. The paper tracks an unprecedented Figure B4.3.1 is a stylized diagram showing how expansion of cash transfers, changes to social insur- all these social protection and labor measures have ance plans to ease eligibility requirements and increase been used by the LAC countries to respond to the benefits, extensive use of employee furlough programs economic fallout from the pandemic. These coun- underpinned by public financing, the deployment of tries responded relatively quickly and with a wide grants and “soft” loans to small and micro enterprises, array of instruments, having benefited from years of and the launch of employment-intensive public works prior investment in making their social protection (although this instrument was used relatively rarely, and labor systems adaptive. given the need for social distancing—avoiding congre- Many of these measures inspired by the crisis gations of people—and other public health imperatives are transitory. However, several permanent changes of the pandemic). In all LAC countries, strict con- have also been made, especially in how household finements and closures have hit the livelihoods of the data are gathered and used and in how benefits are informally employed particularly hard and have wiped delivered. These changes had long been planned, out many informal businesses. Several Latin American but they were accelerated in order to help people countries have sought to ease the plight of these non- cope with the economic consequences of vital public poor but nonetheless vulnerable people—the “missing health measures.

FIGURE B4.3.1 Stylized social protection and labor policy responses to the COVID-19 pandemic

Number of COVID-19 cases

Ex ante Time

Ensure preparedness of Adjust service delivery to ensure safety of sta and bene ciaries social protection and labor delivery mechanisms Ensure continuity of bene ts and services to existing bene ciaries

Fee waivers for health services for poor and vulnerable Subsidize health insurance coverage to Scale-up of safety nets to existing bene ciaries and additional poor vulnerable households safety-net bene ciaries Death bene ts to uninsured poor households who lose family members to the virus

Additional lump-sum payments for social pensions

Anticipation of future pension bene ts

Tax and incentives for rms to retain workers Extension of unemployment insurance bene ts

Safe public works and strategies to facilitate diversi cation of livelihoods

Retraining and intermediation for those out of work

Service delivery systems Safety nets Pensions Labor market

Sources: Morgandi et al. 2020; Williams and Berger-Gonzalez 2020.

Source: Gentilini et al. 2020. Toward an Integrated Policy Response 105

economic shocks. In advanced economies and response to crisis-induced threats to employ- in several LAC countries (such as Brazil and ment is intermediation and job search sup- Peru), very large fiscal packages have been port, for people who have lost their jobs, and deployed to deal with the current COVID-19 the creation of affordable opportunities for crisis. LAC countries as a group have acted upskilling and reskilling. countercyclically in response to it. The concern now is to not withdraw this support Landscape of reemployment and reskilling too quickly. Although some countries have support in Latin America renewed their support as the pandemic has Noncash reemployment services provide a continued, others have not, and this renewal vital complement to income support for peo- (or lack thereof) has fiscal implications. ple who have been displaced by crises and other categories of shocks. Reemployment services include programs that help displaced Preparing workers for change: (short- workers to renew their skills. Such reskilling term) retention subsidies and (long- and reemployment support measures (some- term) reemployment and reskilling times collectively known as active labor support market programs, or ALMPs) are necessary In addition to the short-term impacts of crises in two ways. First, along with the combined on workers, this study highlights that during and coherent use of individual savings and crises, workers suffer unemployment, loss risk pooling for income support, intensively of durable earnings, and worse career starts engaging people who have lost jobs with sup- that are hard to recover from. These effects port to help them find new work has been are long lived. What can be done to miti- shown to lower the risks of moral hazard and gate them? Past research on crises suggests perverse labor supply incentives that arise that the persistence of these effects depends almost inevitably from offering unemploy- on how a crisis is handled and how well, in ment insurance (Fietz 2020). And second, consequence, workers adjust. Historical evi- these programs help to compensate for dence shows that the longer a crisis lasts, people’s bounded rationality, behavioral lim- the more difficult it is for workers to move itations, and less-than-perfect information from declining to expanding sectors. The about new work prospects and the demand previous sections of this chapter emphasized for skills. However, the global evidence to the importance of automatic stabilizers and date on the effectiveness of ALMPs can be effective income support as policy responses discouraging. to help workers maintain their level of con- A recent review of the more rigorous evi- sumption. This section focuses instead on dence from impact evaluations of skill train- employment assistance and on reemployment ing, wage subsidy, and job search assistance and reskilling schemes. It also discusses job programs by McKenzie (2017) shows that retention programs (furloughs with pay and these programs’ impacts are modest at best other short-term schemes that keep workers in most circumstances.13 Public employment matched with their jobs, helping to restore services are typically the least resourced employment to previous levels and to prevent branches of national social protection and the loss of human capital specific to certain labor systems. Most governments are unable sectors and firms), given their prominence in to offer interventions specifically suited the response to the current crisis in countries to particular shocks, to the varying needs like Argentina; Brazil, especially; and Chile. of different groups of job seekers (such as Increasing the speed and quality of job young people, parents, or the elderly), or to matches or investments in new skills can particular industries or places. The track mitigate the effects of crises on workers and record of traditional ALMPs has also been improve prospects for the affected regions’ marred by governments’ tendency to deploy future growth. The traditional first line of them instead of—rather than in support 106 employment in Crisis

of—necessary structural, institutional, and system (most obviously the unemployment regulatory reforms. A further limitation on insurance or other income support plan); the provision of these services in response to (c) monitoring of their implementation and crises is that in countries across the world, evaluation of their impact; and (d) adequate public employment services suffer from poor resources from national budgets. funding and low investment in their imple- Figure 4.12 conceptually organizes shocks mentation capacity, and private programs experienced by labor market participants and suffer from limited supply. Even countries proposes sets of interventions (other than in Latin America with long track records of income support) that are best suited to getting administering public employment assistance people back to work after each type of shock. programs, such as Argentina, Colombia, and Crises such as the 2008–09 global financial Peru, fail to fund them properly (ILO 2016). crisis, because they affect an entire country, This lack of resources leads to low coverage are classified as transient systemic shocks and difficulties in implementing and tailoring (on the top left). They differ from permanent programs to the needs of different groups. systemic shocks (on the top right), which A renewed policy emphasis on reemploy- consist of disruptions driven by structural ment support will require four elements transformations (such as climate change, the rarely associated with traditional ALMPs: (a) widespread adoption of new technologies, specificity to the shocks that caused unem- and changes in trade policy) that destroy cer- ployment or to the particular needs of job tain occupations and create new ones with seekers; (b) coherence and coordination with different skill sets. Shocks also differ from other parts of the social protection and labor those that are transient but idiosyncratic

FIGURE 4.12 Employment and reemployment policies, by the nature of the shock causing displacement

• Bank hours • Reskilling • Job sharing • Mobility assistance • Furlough • Earnings insurance • Retention subsidies • Credit

Transient and Permanent and systemic systemic Financial crises Structural and Natural disasters technological change

Information and coordination

Transient and Permanent and idiosyncratic idiosyncratic Structural Occupational unemployment transition and churn and structural • Information • Reskilling exclusion • Intermediation • Productive inclusion • Search support • Credit • Skills renewal • Demand subsidy

Source: Adapted from Packard et al. 2019. Toward an Integrated Policy Response 107

to individuals or households (bottom left), counseling, various types of training, job such as increased competition, more flexi- search assistance, intermediation, and var- ble dismissal procedures, structural churn, ious forms of wage subsidies. However, the and cyclical fluctuations that lead to more combination of services required to support frequent separation and reemployment or to those transitioning between similar types variations in earnings. Finally, permanent of jobs when an individual firm downsizes idiosyncratic shocks (bottom right) call for will be different from those needed to sup- policies to facilitate longer-unfolding transi- port people displaced by structural changes, tions from lower- to higher-productivity jobs, such as trade liberalization or the widespread particularly in lagging areas and regions, or adoption of new technologies, which affect out of long-term unemployment and inactiv- whole industries and places. Probably the ity (Packard et al. 2019; Robalino, Romero, most difficult interventions are those needed and Walker 2020). However, as this study to facilitate transitions out of very-low-pro- shows, crises can leave long-lasting scars, and ductivity activities (such as subsistence agri- thus some of their impact is not transitory culture or own-account work in household but long-term (tending toward permanent). enterprises). In these cases, if ensuring access Therefore, programs typically associated to quality public services and adequate con- with permanent systemic shocks, such as nective infrastructure is not enough, tradi- reemployment and reskilling support, should tional active labor measures may need to be considered in response to crises. be combined with “demand-side” interventions to mobilize investment and create new Mixed record of reemployment assistance job opportunities (Robalino, Romero, and services Walker 2020). This approach is being fol- A recent review by Card, Kluve, and Weber lowed and evaluated in several countries (2017) synthesizes the findings of more than (box 4.4). 200 recent studies of active labor market programs. The authors distinguish between Beyond short-term income support: Policy three postprogram time horizons and use actions to get people back to work and regression models to determine the programs’ reskill them estimated effects, for studies that model the Employment policies are the traditional probability of employment, and the sign and response to the challenges of reemployment significance of the estimated effects for all and reskilling. However, most LAC countries the studies in their sample. They conclude spend very little on active labor measures: that the average effects of ALMPs are close about 0.5 percent of GDP. Even those that to zero in the short run but that the effects spend at higher levels have rather poor track become more positive two to three years records of performance (McKenzie 2017). For after completion of the programs. The time instance, among 90 youth employment pro- profile of these impacts varies by the type of grams in the LAC region that were rigorously program, with larger average gains shown evaluated, only 30 percent had positive by programs that emphasize human capital effects on employment rates or earnings, and accumulation. There is also systematic het- these effects were small (Kluve et al. 2016; erogeneity across groups, with larger impacts Robalino and Romero 2019). Moreover, there for women and for participants coming from were no significant differences in effectiveness long-term unemployment. between types of programs (for example, The combination and intensity of inter- training versus job search assistance). Most ventions required to successfully get people active labor measures managed by public back to work could be different for each employment offices have not been evaluated. type of shock. The standard reemployment But their institutional capacity is usually lack- services (dealing with information and skill ing, they face severe constraints in terms of constraints) should continue to include human and financial resources, and their 108 employment in Crisis

BOX 4.4 Permanent, systemic shocks: Responses to job dislocation caused by structural changes

Even the best-performing, modern, generously a job in a tradable sector creates between 0.5 and funded public employment services will struggle 1.5 extra jobs in nontradable sectors [Ehrlich and to meet the needs of people who have lost jobs in Overman 2020]; thus, the loss of a tradable job the wake of systemic shocks that bring permanent might lead to additional job destruction in down- consequences. Thankfully, there are examples of stream or upstream sectors.) The most affected countries that have responded to help the people workers will be difficult to identify, because they who bore the brunt of immediately disruptive but are likely to work in sectors not initially affected ultimately beneficial structural changes. The inter- by the shock. ventions used include targeted labor adjustment Critics of the program emphasize that the best assistance programs that create appropriate incen- reskilling is delivered on the job. They have proposed tives to return to work and can minimize mobility an alternative of “wage insurance”—time-bound costs and accelerate employment transitions. payments made directly to workers in order to The Trade Adjustment Assistance Program in the reduce the difference between what they earned United States is a federal program that helps workers in the jobs just lost and in their new jobs, up to a by providing job search assistance, training, wage ceiling. Wage subsidies instead of classroom train- subsidies to their prospective new employers, health ing could encourage workers to seek reemployment insurance for the unemployed, and reallocation rapidly and thus improve their access to on-the-job allowances. The program helps workers who have learning (Vijil et al. 2018). been displaced because their firm relocated to In Austria, the Austrian Steel Foundation has another country or because of trade liberalization helped displaced workers find new work since the (for workers in the import-competing industry privatization of the country’s steel industry. It offers as well as for those employed by downstream or a wide range of services, including vocational orien- upstream producers). Evaluations of this program tation programs, small-business start-up assistance, show mixed results, including limited effectiveness extensive training and retraining programs, formal at helping trade-affected workers obtain reemploy- education, and job search assistance. The foundation ment at a suitable wage (Schochet et al. 2012). is financed by all participants: the trainees them- Evidence on the targeting of beneficiaries on selves, the steel firms, local government (through the basis of their sector of employment, as done in unemployment benefits), and the remaining workers this program, is not encouraging. It shows that (a) in the steel industry, who pay a solidarity share of regional mobility frictions are higher than sectoral their gross wages to the foundation. The program mobility frictions and (b) a crisis that is initially has increased the probability of participants being transmitted through one sector quickly spreads employed (Winter-Ebmer 2000). to other sectors. (For instance, estimates in the United States and the European Union suggest that Source: Vijil et al. 2018.

existing staff have weak incentives to respond depends on its ability to adapt its services to to the needs of job seekers and employers. very different profiles and to the demands of Several lessons from the international different workers. To do this, reemployment experience can be used to guide the reform assistance services have to set up registration of active labor measures in the LAC region. and statistical profiling systems to help them First, the evidence shows the importance of identify the types of constraints facing indi- moving from solitary interventions to provid- viduals. Additionally, modern monitoring ing an integrated package of services. Even and evaluation practices are key to assessing individuals affected by the same type of shock the results of programs and to introducing seldom face identical constraints to accessing corrections when they are needed. The fis- a new job. Thus, the success of a program cal sustainability of larger, more effective Toward an Integrated Policy Response 109

programs will also require diverse sources promoting short-cycle higher education proof financing. When governments make grams, expanding access to low-income risk-pooling structures more widely avail- students, and conditioning these programs’ able to cover shocks with uncertain and cat- funding on students’ employability. astrophic losses, it is reasonable to expect the All things considered, a more complete resources contributed to these programs by characterization of the possible policy areas people and firms to meet the needs caused by of focus to achieve a stronger social pro- more foreseeable and less costly shocks. Most tection response to crises in Latin America active labor measures today are financed (second policy dimension) is illustrated in from general budget expenditures. Given the figure 4.13. Evidence from multiple contexts nature of shocks and losses and the degree of shows that each priority area included in that the market failures involved, this source of figure can make a real difference in labor funding is appropriate for some but not nec- market adjustment. essarily all needs. More robust and coordinated employment assistance is needed, with Structural: Greater competition greater focus on these programs’ results and and place-based policies unintended consequences. In terms of reskilling policies, is import- Chapters 2 and 3 of this report document the ant to support working people in the face of importance of demand factors and the three change. This support involves strengthening structural issues that magnify the impacts of technical education and vocational training, crises on welfare and efficiency in the LAC

FIGURE 4.13 Addressing crises’ impacts and preparing for change: Policy reforms

WORKERS

s k Income support r o Active labor market programs

w e

SHOCK m a r f

o Cushion the short-term impact of shocks on workers

o • Augment unemployment income through the creation or redesign of c

e unemployment insurance

o r

c • Improve the capacity to assist with social assistance programs

+ Beyond short-term income support

r • O er robust and coordinated employment services to get workers back to

i the workforce quickly l

b Support working people for change: Enhance their skills a •

t S

Source: World Bank. 110 employment in Crisis

region: labor rigidity, which complicates can improve outcomes with minimal distor- employment transitions; the dichotomy in the tions or efficiency costs (World Bank 2012). region between protected and unprotected Some of the most contentious regulatory firms (caused by factors such as lack of com- instruments are restrictions on firms’ con- petition and excessive market power among tracting and dismissal decisions, known protected firms); and low levels of mobil- collectively as employment protection legis- ity among workers. In light of these issues, lation (EPL). EPL is part of the institutional LAC countries’ policy actions may need to framework around the labor market. Other go beyond traditional social protection and elements of this framework are (the existence labor reforms in order to make a difference of and) rules for unemployment insurance, (third policy dimension) is illustrated in ALMPs, and governance structures such figure 4.1. What does this mean? as tripartite collective bargaining (between The following section discusses the key trade unions, employers or business associa- institutional impediments to employment tions, and the government as the convener). transitions and the key agenda items to reform This institutional framework affects both them. It then discusses the insider story and the functioning of labor markets and firms’ some practical examples of how better com- productivity (Betcherman 2014). Within this petition policies can change the status quo framework, labor regulations determine the and bring needed dynamism to employment types of employment contracts permitted; recovery after crises. It concludes with a dis- employers’ ability to adjust wages, benefits, cussion of how to address the spatial dimen- and hours; working times and conditions; sion of labor market adjustment through a forbidden employment practices; and the dual-dimension policy response, including rules governing the hiring and dismissal both well-designed regional development of workers (Kuddo, Robalino, and Weber policies supporting job creation in depressed 2015). Designed to protect or redistribute regions and place-based policies to reduce the income to workers, these regulations are costs of mobility between regions or neigh- normally intended to address a flaw in the borhoods. Helping people to overcome the labor market (such as imperfect information, structural and, especially, spatial constraints uneven market power between employers they face to employment is a necessary focus and workers, discrimination, or inadequa- of an augmented deployment of active labor cies of the market to provide insurance for market policies. employment-related risks). In the LAC region, where job displacement income support has limited coverage, Facilitating employment transitions: some governments took the policy stance to Labor market regulatory rigidity address the risk of job loss and other employ- Chapter 2 showed that in countries with ment shocks in the formal sector by prevent- very different labor market regulations, the ing or slowing adjustments rather than by natures of the markets’ adjustments and their helping affected workers manage and recover consequences for firms’ productivity and sur- from these shocks. This approach relies vival also differ. This section discusses the heavily on restricting dismissals, mandating key areas in which the LAC region has labor employer-provided severance obligations, market rigidities and how to address those and limiting the use of flexible employment rigidities for better crisis response. A long contracts, such as fixed-term contracts or and fierce debate among economists on the outsourcing. benefits and costs to employment outcomes Evidence shows that when these regulations of labor market regulations is slowly mov- are set at overly restrictive levels they can gen- ing toward a consensus: when policy makers erate undesirable economic and social impacts avoid extremes of either too little or too much that exacerbate the labor market imperfec- regulation, reasonable levels of regulation tions they were originally intended to address Toward an Integrated Policy Response 111

(Betcherman 2014). The LAC region has some adjustment, fewer job offers will be made, examples of regulations set at extreme levels lengthening unemployment spells.14 Overly compared with countries in other regions. restrictive employment regulations affect In Bolivia and the República Bolivariana de employers’ decisions on how to adjust to Venezuela, for instance, the labor code does demand shocks, altering how workers are real- not allow contract termination for “economic located over the business cycle.15 In the LAC reasons” (that is, poor performance or market region, the devil is in the details of this issue— downturns), limiting grounds for dismissal to labor markets are rigid only in some countries disciplinary reasons. In Ecuador, the use of and only in some key dimensions. The region’s fixed-term contracts as well as outsourcing are regulations vary considerably according to severely limited. In Suriname, employers must widely used indicators of the extent of labor seek approval from the Ministry of Labor market regulation, such as the OECD’s EPL to dismiss a worker. Mexico, Panama, and index (which the Inter-American Develop- Peru have similarly restrictive procedures for ment Bank [IADB] expanded to cover many dismissals. As long as there is a requirement LAC countries). Even in countries where reg- to give employees reasonable advance notice ular employment contracts are restricted sim- of dismissal, firms should be given more flex- ilarly to or less than the OECD average (such ibility in their human resource decisions. To as Colombia, Panama, Peru, and Uruguay, as prevent abuse and discriminatory practices by shown in figure 4.14), fixed-term (temporary) firms, ministries of labor can implement risk- employment is more restricted and collective based, ex post audits and apply severe penal- dismissals are substantially more difficult. ties where infractions are found (Packard and However, enforcement capacity makes all the Onishi 2020). difference as to whether regulation as written In an environment with critical pockets does in fact constrain employment practices of overly rigid labor regulation, high costs and create significant friction in the labor mar- of job destruction, and slow labor market ket’s adjustment (Kanbur and Ronconi 2018).

FIGURE 4.14 Employment protection legislation in OECD member countries and selected Latin American countries, 2014 or most recent data

Most rigid 6

0 Most exible Italy Spain Peru Chile France Mexico AustriaGreecePoland Turkey JapanIreland Tunisia Bolivia Belgium Portugal Sweden Thailand Uruguay Australia Panama Canada Germany Argentina Colombia Denmark Paraguay Costa Rica Netherlands Korea, Rep. Venezuela, RB OECD average UnitedNew States Zealand United Kingdom Regular contracts Temporary contracts Collective dismissals

Sources: OECD Stat, Employment Protection Legislation (version 3), augmented for Latin America and the Caribbean for the years 2013 and 2014 by the Inter-American Development Bank’s Database of Labor Markets and Social Security Information Systems (SIMS). Note: These countries’ employment protection regulations are ranked on a scale from 0 to 6, where 0 = most flexible and 6 = most rigid. OECD = Organisation for Economic Co-operation and Development. 112 employment in Crisis

Going deeper than these aggregate indi- the use of part-time and temporary workers, ces of employment protection legislation, the repeated use of fixed-term contracts, and more granular indicators of labor regulation outsourcing), (b) the regulation of work time show how specific regulatory instruments (that is, what defines the working day and are used with differing intensities across the business days), (c) rules for dismissal proce- LAC countries. The World Bank’s Doing dures (such as requirements for third-party Business project’s Employing Workers data notification and even approval for single or set can distinguish among the regulation of collective redundancies), and (d) the actual hiring practices, working hours, the handling financial costs of dismissal (severance pay, of redundancies, and dismissal costs.16 In the payouts from accumulated leave, and other four panels of figure 4.15, a single compos- financial penalties required of firms). These ite index of overall labor regulation rigidity indices are constructed with principal com- constructed on the basis of all of these indi- ponent analysis (PCA) and normalized on a cators is plotted against separate subindices scale ranging from −3 (least rigid) to 3 (most that capture (a) restrictions on hiring (i.e., on rigid), with 0 assigned to the LAC region’s

FIGURE 4.15 Regulation of employment in the LAC countries, circa 2019

a. Restrictions on hiring practices b. Regulation of working time (rigidity of hours) 3 3 HND HND MEX PAN MEX 2 PER 2 PAN PER SUR SUR LCA BHS LCA 1 ECU 1 CHL GUY KNA ECU KNA BRA CHL BRB BRB GUY 0 0 BHS BRA BLZ NIC DMA GTM TTO ARG CRI GTM CRI DMA ATG ARG SLV TTO SLV –1 COL JAM –1 DOM ATG GRD COL DOM NIC HTI URY Rigidity Index (PCA) 2019 GRD HTI (normalized annual) JAM g. –2 URY –2

LM Re –3 –3 –3 –2 –1 0123 –3 –2 –1 0 123 Di culty of Hiring Index (PCA) 2019 (normalized annual) Rigidity of Hours Index (PCA) 2019 (normalized annual)

c. Procedural di culty of dismissals d. Financial costs of dismissal to employer 3 3 PAN PAN MEX HND MEX 2 PER 2 PER CHL SUR 1 KNA 1 ECU LCA GUY SUR BHS CHL LCA GTM BRA KNA BRB BHS GUY BLZ 0 0 BRA CRI CRI SLV GTM DMA BRB ARG GRD TTO DOM SLV ARG –1 JAM TTO –1 NIC NIC HTI ATG COL URY DMA Rigidity Index (PCA) 2019 GRD HTI (normalized annual) URY JAM g. ATG –2 COL –2

LM Re –3 –3 –3 –2 –1 012 3 –3 –2 –1 0123 Di culty of Dismissal Index (PCA) 2019 (normalized annual) Dismissal Cost Index (PCA) 2019 (normalized)

Sources: Packard and Onishi 2020; indices constructed by Maratou-Kolias et al. 2020 using Employing Workers data from the Doing Business project, World Bank. Note: Following Packard and Montenegro 2017, five indices of de jure regulations are constructed using principal component analysis (PCA): a composite “overall labor market regulation rigidity” index of all the Doing Business project’s Employing Workers labor market regulation indicators (plotted on the vertical axis in each panel) and four indices for different subsets of regulation measures. PCA values have been normalized to the LAC regional mean values (indicated by blue horizontal and vertical lines, respectively) to create a scale ranging from −3 (most flexible) to 3 (most rigid). LAC = Latin America and the Caribbean; LM Reg. = labor market regulations. Toward an Integrated Policy Response 113

mean values. Thus, countries in the top half What can be done? of each panel have more rigid regulations Reducing the intensity with which countries overall, and countries’ placement in the right regulate firms’ human resource decisions is or left quadrant of the panel (their distance likely to affect workers. With broader access from the mean) indicates what specific aspect to national unemployment insurance proof labor regulation explains their rigidity (or grams, more dynamic safety net transfers, lack thereof). and a robust system of reemployment support As shown in figure 4.15, overall labor services, labor market adjustments, including regulation rigidity in the sample of LAC regulatory changes, will be smoother (Ander- countries for which the Employing Work- sen 2017; Bekker 2018). Similarly, the princi- ers indicators are collected is driven mostly ples of protecting working people rather than by restrictions on hiring practices (panel a) protecting employment and of decoupling and the procedural difficulty of dismissals protections from where and how people work (panel c). Consistent with the OECD-IADB in response to the permanent effects of eco- EPL index presented earlier, the statutory nomic transformations might not help with financial costs of dismissal, such as paid a crisis in the short-run but might be appli- notice periods, severance, and payout of cable in the medium run.18 The effects of this unused leave (panel d) and rigidity of hours change will also depend on whether broader (panel b) appear to be lesser contributors to access to national unemployment insurance overall regulatory rigidity, although their programs is available and on the employment contribution varies considerably across dynamism of the economy. countries.17 Figure 4.16 shows a stylized plot compar- Rigid regulation of regular employment ing countries in the LAC region with those in contracts and large disparities between the other regions according to (a) the flexibility of protections extended by these contracts and their labor market regulation (along the hor- those offered with nonstandard forms of izontal axis, the inverse of the rigidity index employment can create an insider-outsider in figure 4.15) and (b) the extent to which key labor market even within formal employ- protections are available outside the employment. This effect can add to the formida- ment relationship (along the vertical axis, ble obstacles caused by the formal/informal an index of public spending on education, divide: a large set of protections are associ- health, social assistance, and labor market ated with formal employment, but also a support programs as a percentage of GDP). large tax wedge. Betcherman (2014) shows In the upper right quadrant of figure 4.16 are that employment protection legislation has Denmark, New Zealand, the United King- been found to have an equalizing effect dom, and other countries that have shifted among covered, full-time workers of prime their policy stance to combine greater labor working age but that it leaves groups such market flexibility with more robust human as youth, women, and the less skilled dis- capital and social protection services to help proportionately outside its coverage and its people navigate transitions between jobs. benefits (Betcherman 2014; Heckman and Although a handful of LAC countries are in Pagés 2004). The attempt to mitigate the that same quadrant (with high flexibility and impact of overly rigid regulation with spe- high protection), many lag behind in at least cial forms of contracts only exacerbates these one of these dimensions. adverse distributional impacts. Youth and The current overreliance in some coun- women are disproportionally likely to be tries on employment protection regulations hired on temporary contracts, which leave instead of income support and reemployment them without access to many benefits and services comes at a cost: it damages the job protections against dismissal (Gatti, Goraus, prospects of many people, particularly young and Morgandi 2014; Kuddo, Robalino, and people and people of all ages who prefer or Weber 2015). need to combine work with study or care 114 Employment in Crisis

FIGURE 4.16 Flexibility of labor regulation and spending on human capital and labor programs in selected countries in LAC compared to other regions

A “exicurity” approach to labor policy requires that governments invest more in protection than most do currently

Low exibility High exibility

High High protection High protection

DNK

NOR SWE FIN BEL FRA ISL AUT NZL NLD ESP DEU AUS CAN ISR USA ITA PRT ARG JPN EST HUN GBR LUX SVN CRI Protection SVK IRL CZE CHL DMA URY HRV UKR LVA TUR CHE LTU GEO KOR BLZ POL BRA BLR HND GRC MDA SRB COL MEX ECU GUY TTO NIC JAM PAN RUS BIH KSV SLV PER UZB ROM BRB ALB DOM MNE GRD PRY LCA KAZ GTM SUR KNA ARM AZE KGZ HTI BHS ATG Low exibility TJK High exibility

Low Low protection Low protection

Low Flexibility High Latin America and the Caribbean All other regions Median Mean

Source: Onishi and Packard 2020, based on Packard et al. 2019. Note: All other regions = East Asia and Pacific, Europe and Central Asia, Middle East and North Africa, and North America. The horizontal axis shows the inverse of labor market regulation principal component analysis of the composite rigidity index using the Doing Business project’s Employing Workers data set; the vertical axis charts the index of “protections” (measured as government spending on health, education, and social protection support) that are accessible outside the employment relationship.

responsibilities. It is also associated with lon- people with more physical and human capi- ger average job searches and thus slower labor tal working in countries with extensive social market adjustments. Exploiting the extensive insurance systems can take more time after country coverage of uniform survey micro- losing jobs to find better matches). However, data in the International Income Distribution the significant associations between the dura- Database (I2D2) and the Employing Workers tion of job searches and de jure restrictions indicators, Packard and Montenegro (2021) on hiring practices, working hours, and dis- analyze the association between the average missal procedures are less ambiguous. The duration of unemployment and various forms size and statistical significance of these asso- of labor regulations, controlling for economic ciations increases when the country’s capac- growth, average educational attainment, and ity and efforts to enforce regulations are other relevant factors. Figure 4.17 shows included in the analysis. statistically significant coefficients for sev- International experience shows that loos- eral labor regulation variables (including an ening restrictions on firms’ hiring and dis- enforcement capacity indicator from Kanbur missal decisions needs to be accompanied and Ronconi [2018]). The interpretation of by establishing more effective protections the positive, significant association between outside the employment contract, including duration of unemployment and factors like reemployment support such as income and home ownership, educational attainment, job search assistance (Kuddo, Robalino, and the levels of employers’ social insurance and Weber 2015; see also the OECD Jobs contributions is ambiguous (for instance, Strategy in footnote 18). The goal is not Toward an Integrated Policy Response 115

FIGURE 4.17 Labor market regulation instruments and the duration of unemployment

Employer social security taxes

Dismissal procedures

Dismissal costs

Rigidity of hours

Difficulty of hiring Estimated coe cient Enforcement effort (index)

Completed education

Owns home

–0.20–0.15 –0.10–0.05 00.050.100.15

All Women Men Ages 15–25 Ages 26–55 Age 56 and older

Sources: Packard and Montenegro 2021; data are from the International Income Distribution database (I2D2) and the World Bank’s Employing Workers data set. The “Enforcement effort” index was adopted from Kanbur and Ronconi (2018). Note: Estimated coefficients on the variable that controls for economic growth have been omitted from the figure. “All” indicates estimated coefficients for the whole sample of individuals. For the variables “Employer social security taxes,” coefficients are statistically significant at the 1 percent level for all groups; “Dismissal procedures” coefficients are statistically significant at the 1 percent level for all groups. “Dismissal costs” coefficients are not statistically significant. “Rigidity of hours” are significant at the 5 percent level for All, Women, and those aged 26 to 55; at the 10 percent level for men and those aged 15 to 25; and not significant for those 56 and older. “Difficulty of hiring” coefficients are statistically significant at the 1 percent level for All, Men, and those aged 15 to 25 and 26 to 55; at the 10 percent level for Women; and not significant for those 56 and older. “Owns own home” is statistically significant at the 1 percent level for all groups.

deregulation but smarter regulation that employment practices. The frequency of reflects the risks and opportunities avail- “nonstandard” forms of work in the LAC able in modern, diverse labor markets. As region (including self-employment) poses long as firms are required to give employ- additional challenges. In Argentina, Brazil, ees reasonable advance notice of dismissal, Chile, and many other LAC countries, formal they should be given more flexibility in their but nonstandard forms of work appear to be human resource decisions. To prevent abuse growing at the expense of formal dependent or discrimination, ministries of labor can employment. The profile of those in nonstan- implement risk-based, ex post audits and dard employment has also changed dramat- apply severe penalties in such cases. Greater ically since the mid-1990s: people in formal regulatory flexibility for firms must be nonstandard employment today are younger accompanied by more concerted and robust and better educated than before (Apella and reemployment support. Without responsive Zunino 2018). income and employment support in place to How should LAC countries make pro- help absorb the shock of unemployment and tection accessible to not only formal work- assist with job searches, loosening labor reg- ers (regardless of their contract type) but ulations would simply shift the risk burden also informal workers? The challenge is from firms onto working people and raise to restructure the countries’ social protec- the likelihood of segmentation and abusive tion and labor systems so that support is 116 employment in Crisis

accessible no matter where or how people evade taxes and statutory contributions, or work (Packard et al 2019). Many countries— disguise actual employment relationships spanning the full range of economic and as self-employment would be substantially institutional development—are giving seri- reduced. ous, if cautious, consideration to the fiscal The disadvantage to this approach is that viability and social benefits of universal organizing social protection and labor sup- basic income plans to achieve full protection port in this way demands much more from (Gentilini et al. 2020), though such exten- governments than the LAC region’s present sive safety nets may still be far beyond the systems. But this is the challenge of eco- fiscal and administrative capacity of most nomic and institutional development. Gov- LAC countries. The region’s social protec- ernments in the region would have to apply tion and labor systems today are greatly taxation instruments far more effectively and improved from the truncated welfare states efficiently than they do today and augment of the 1980s and 1990s. However, there is their administrative capacity, especially by still much to be done to achieve efficiency, more quickly adopting digital technologies, effectiveness, and sustainability. to enable information management as well as A good place to start to improve the benefit delivery. adjustment of Latin America’s labor mar- Liberalizing labor contracts initially has kets is with extensive reforms to regulatory a positive effect on employment under the restrictions on hiring and dismissals, sever- new, more flexible type of contract (Ben- ance, and other employer-specific benefits tolila, Dolado, and Jimeno 2012). But this (including firm- or sector-specific health effect gradually fades as the stock of perma- insurance). The idea behind Europe’s recent nent workers is replaced with workers under labor and social protection policy reforms flexible contracts. In Latin America, the was to transform unfunded, firm-specific effect of these regulatory changes has been protection to nationally administered evaluated in some countries using panel data plans consisting of portable “backpacks,” (household or firm) and time-series models, delinked from specific jobs, that working and the results have been mixed. Although people could “carry” with them from job Kugler (2004) in Colombia, Mondino and to job. Instead of making contributions at Montoya (2004) in Argentina, and Saavedra levels consistent with previous severance and Torero (2004) in Peru identify a negative obligations, employers and workers could effect on employment of new job security make contributions into individuals’ savings rules, De Barros and Corseuil (2004) for Bra- accounts (either stand-alone unemployment zil; Downes, Mamingi, and Antoine (2004) savings accounts or accounts combined for three Caribbean countries; and Petrin and with retirement savings, per Feldstein and Sivadsadan (2006) for Chile find no signifi- Altman [1998]). These savings would be cant effect. underpinned by a risk-pooling mechanism that guaranteed benefits proportionate Breaking the insider story: More than to workers’ contribution history but with just labor market reform a guaranteed minimum benefit financed by the broadest-based taxes, much as the Results in the previous chapter show that LAC region’s CCTs and other social assis- effects of crises are lessened for firms with tance transfers are today. The main advan- more market share and for those with state tage of this approach is that it opens access ownership. In particular, Fernandes and Silva to effective and efficiently priced protec- (2020) show that product market concentration to a larger share of working people. tion affects the magnitude and distribution Protection would no longer be segmented of crises’ effects on workers. Shocks cause by type of employment. Also, the present greater losses in employment and wages in incentives for employers and job seekers to sectors with low market concentration (many game the system, underdeclare earnings, players). In contrast, in sectors where a few Toward an Integrated Policy Response 117

players hold a large share of the market in product markets and the ways crises (high market concentration), shocks actually affect workers. First, protectionism gener- increase employment and wages do not adjust ates rents for employers. Rents can influence (the reverse of what normal economic mech- the distribution of losses between workers anisms would bring about). Whether a cer- and firms, and in sectors with more market tain labor market regulation helps or hinders power (particularly in monopolies), firms are employment outcomes is also determined able to adjust their prices less in the wake of by the extent of product and service market shocks. This factor affects the distribution of concentration and the bargaining power of losses across firms and translates into higher employers relative to that of workers. The prices for consumers, including client firms aggregate data are directionally consistent downstream, which may harm employment with this observation, including the fact dynamics. Market power can also be driven that large firms in Latin America (often by the structure of the market for a partic- protected firms in the energy, commodities, ular good, as well as by the distribution of and retail sectors) have been more resilient to firm-specific technology. The latter effect crises and have recovered faster from them may mean that firms’ reactions to crises have (see chapter 1). some stickiness, in the sense that they gen- This section discusses (a) the institutional erate persistent changes that firms cannot peculiarities in Latin America that are giv- revert. ing rise to this insider-outsider labor market The second channel through which eco- and allowing segmentation to persist and (b) nomic rents affect adjustments to crises is the types of complementary policy responses by changing the willingness of workers to beyond the labor market that could address adjust and to incur adjustment costs. Firms concerns about crises’ effects on workers. may pass economic rents on to their workers, Market power in product and service mar- increasing the workers’ reservation wages, kets, defined as the ability to drive prices and which in turn makes it harder for them to returns above competitive levels, is coming find well-enough paying replacement jobs if under increasing scrutiny for causing adverse their current ones are lost. At a time when socioeconomic outcomes beyond just higher reallocation is difficult but needed, this effect prices for consumers. Increasing concen- may increase pressure for additional protec- tration in product and service markets in tionism and favoritism, a further impediment high-income countries often translates into to efficient resource allocation. concentration in labor markets and to dom- In light of these effects, lowering bar- inant employers engaging in exploitative riers to competition in product markets practices (Azar et al. 2019). In many Latin could increase job creation and productivity American countries, market concentration growth. It would shrink the share of firms is linked with close ties between large firms earning above-normal returns on capital— and governments. An extreme example is the and, therefore, the share of workers pro- prevalence of state-owned enterprises in the ducing and sharing in those above-normal region: many of the region’s largest firms are returns. This change, in turn, would lead to partially or totally owned by governments. higher labor mobility, because the rents these Familial and social ties between the politi- workers earn are discouraging them from cal and entrepreneurial elite throughout the leaving their firms. Compensation policies region are another important factor. Close to smooth the adjustments for these workers links between business and political elites could also make their transitions less costly. result in protectionism and favoritism in At the same time, there are too few large domestic markets, favoring incumbents and firms in Latin America. Not all sources of strangling new entrants (Clarke, Evenett, and economic market power in the region should Lucenti 2005; De Leon 2001; OECD 2015). be eliminated. Some degree of product mar- Two main mechanisms may create the ket power is desirable in order to create pos- relationship between a lack of competition itive incentives to innovate. However, these 118 employment in Crisis

incentives could be improved in other ways, Bergman, and Kim 2018). Concentration is such as instituting transparent intellectual often accompanied by restrictive practices, property and patent regulations, that do not such as the proliferation of local licensing sacrifice the benefits of market contestability requirements or the extensive use of non- and competition. Regulators’ focus should be compete clauses, even in industries that hire on the abuse of market power by dominant mostly low-skilled people (Naidu, Posner, companies to restrict competition, the for- and Weyl 2018). These restrictions on com- mation of cartels (illegal agreements between petition combine with declining labor mobil- firms not to compete), and the removal of ity to put downward pressure on earnings unnecessary anticompetitive regulations (Konczal and Steinbaum 2016). that decrease labor market dynamism without encouraging innovation. A recent study What kind of reforms? using data for Brazil, Chile, China, Estonia, Breaking the insider story of the LAC region India, Indonesia, Israel, the Russian Federa- would require policy levers outside the labor tion, Slovenia, and South Africa, found that market. Possible areas of focus include reducing barriers to entrepreneurship (such changes to competition laws, subsidies, pro- as barriers to entry and antitrust exemp- curement, and the level of state participation tions) to a level consistent with the current in various industries. These types of policies best practices among OECD members would could complement other policy responses lead to 0.35 percent to 0.4 percent higher to address concerns about crises’ effects on annual GDP growth per capita (Wölfl et al. workers (as discussed in Baker and Salop 2010). Protectionism and favoritism protect [2015]). Calls are growing for competition some large firms at the cost of preventing policy regulators to look beyond their tra- new entrants from effectively competing and ditional toolkit. These regulators have been themselves growing into large firms. criticized for keeping too strict of a focus on As economies stabilize after crises, it is consumer prices (ignoring, for example, the essential, in order to improve labor market emergence of monopsonist employment prac- performance, for policy makers to safeguard tices). Taking into account a wider range of against threats to competitive and contest- socioeconomic indicators can reveal contestable product and service markets. Structural ability and competition problems that do not changes brought by the advent of technology manifest immediately as higher prices. have contributed to growing concentration. The market power of firms is growing in Addressing the spatial dimension of many parts of the world (Diez, Leigh, and crises’ impacts on workers Tambunlertchai 2018). Ensuring competitive and contestable markets has long been a Results from the previous chapter show that challenge in low- and middle-income coun- shocks’ impacts on workers vary in size and tries, where governance institutions are weak persistence depending on local economic and can be especially vulnerable to oligopo- conditions. For example, in Brazil, employ- listic pressures and collusion problems. How- ment and wage losses in response to shocks ever, many of the same pressures and dangers are higher for formal workers who live in of market concentration are increasing in more informal local labor markers than high-income countries as well (Aznar, Mari- for those in less informal places. Why do nescu, and Steinbaum 2017). A growing body shocks pass through to workers more in some of research from the United Kingdom, the places than in others? The way the places’ United States, and other high-income coun- structural features interact with the shocks tries shows that as local-level employer con- is important. Evidence suggests that the rel- centration grows, wages stagnate and that evant features include the place’s sectorial the negative impact on wages of a given level composition, the types of firms, and the size of concentration is increasing (Benmelech, of the informal sector. In general, workers in Toward an Integrated Policy Response 119

places with fewer alternative job opportuni- costs (Brand 2015), and transportation ties end up more severely scarred by crises. To costs (Zarate 2020). the extent that shocks lead to labor scarring Many of the root causes of low labor and that mobility across regions is limited, mobility are imperfections in housing and shocks can have permanent effects that differ credit markets (Bergman et al. 2019). Place- across space. based policies can promote regional mobility, Is a spatially differentiated approach for example, by addressing the lack of afford- needed to address scarring in the LAC region able housing, improving land-use policies, as a result of temporary shocks such as cri- adjusting zoning rules, or developing mort- ses? Regional policies are generally consid- gage finance. But many areas are already con- ered for issues associated with permanent gested, the noneconomic costs of moving are shocks, such as trade liberalization or techno- difficult to compensate for, and many people logical change. However, if jobs are perma- do not want to move. Thus, a second policy nently lost, effects are spatially concentrated, dimension is also important: well-designed and worker mobility is low, workers will regional development policies to support job be scarred by crises. In these cases, retrain- creation. Such policies can increase long-run ing and other ALMP polices will not be growth and benefit a region’s development. enough, nor will private incentives on their They can also help reintegrate displaced own. Instead, these places will need to be workers by creating more jobs where those revitalized by, for example, increasing public workers are. And they could address business investment in them. climate, infrastructure, and profit opportuni- This report shows that even temporary ties locally so that income-generating oppor- shocks have spatially differentiated long-term tunities are spread throughout the country consequences. Two aspects of these shocks (to the level that makes sense on the basis of make regional policies useful to address the local resources, population, and such). their long-term consequences. The first is the Finally, they could generate local multiplier presumption that the effects of these shocks effects, stimulating consumption and demand are localized. The second is that the shocks and, through this channel, employment. will have structural and permanent effects. Regional policies are therefore needed in a Many spatially differentiated effects can be broader sense in order to strengthen regions’ addressed with nonspatial policies. For exam- economic opportunities. As discussed in ple, ALMPs will be cheaper to institute than box 4.5, whether these policies achieve this place-based policies. But some structural objective depends on the modality of the pol- issues cannot be addressed except by the lat- icies and on the characteristics of the region. ter type of policy, such as a lack of opportu- Evidence regarding the ability of place- nities in a place. The previous chapters show based policies to reduce the costs of mobility the importance of being proactive to prevent between regions or neighborhoods is growing scarring. How can policy makers proactively but remains limited (box 4.6). confront scarring with place-based policies, How large are these effects? Artuc, Bas- and why should they? tos, and Lee (2020) develop a reduced-form Two dimensions are relevant to a regression and structural model for the regional policy response to crises. If welfare effects of changes in external demand on losses are linked with a lack of geographic welfare (i.e., a worker’s lifetime utility) and mobility, removing barriers to mobility and for the role of mobility in these effects. The enhancing connectedness across regions analysis shows that a shock-induced welfare can be a good solution. Transitioning reduction would be lessened if mobility, par- between locations brings a wide range of ticularly across regions (within a country), costs, including search costs to determine was higher. In other words, the reduction of where to go, the cost of moving to the new mobility costs across regions would do more location, a wide range of psychological to mitigate a crisis-induced welfare reduction 120 employment in Crisis

BOX 4.5 How well have regional policies performed at strengthening economic opportunities?

Local job creation efforts often involve (a) invest- in in the city center and thus are inaccessible to work- ments in infrastructure and in local public goods ers who live on the periphery. and services, (b) direct subsidies to firms, or (c) Evidence is more mixed on the effectiveness of the relocation of public sector employment or large direct subsidies or discretionary grants from the public agencies to depressed areas. Neumark and government to firms in disadvantaged areas. These Simpson (2015) provide an overview of the litera- grants aim to support employment at individual firms ture on these types of policies, updated by Ehrlich or to attract new employers to an area. The two key and Overman (2020) in the context of the Euro- concerns with these programs are that they finance pean Union (EU). Overall, the evidence suggests activities that recipient firms would have undertaken that investment in transportation infrastructure and anyway and that the new activity in the targeted areas in local public goods and services in a mix of firm comes at the cost of activity displaced from nontar- subsidies and training, as is done by the EU’s cohe- geted areas. Some studies suggest that subsidies, if sion funds, has on average been effective at fostering well designed, increase local employment, mainly at growth in recipient localities and thus at reducing small firms. This increase, in turn, can generate pos- disparities across places in economic opportunities itive multipliers (i.e., additional jobs) by increasing (Becker, Egger, and Ehrlich 2010; Giua 2017; Mohl productivity (Greenstone, Hornbeck, and Moretti and Hagen 2010; Pellegrini et al. 2013). 2010) or demand for local goods and services. Esti- However, the effects of these programs vary con- mates from the United States and the EU suggest that siderably across areas: they are high in regions with each job in a tradable sector creates between 0.5 and high human capital and high-quality local govern- 1.5 extra jobs in nontradable sectors (Ehrlich and ments but low elsewhere, yielding different trade-offs Overman 2020). But not all evidence is as encourag- between spatial inequality and aggregate efficiency ing. First, positive local effects may be offset by gen- (Becker, Egger, and Ehrlich 2013). They also have eral equilibrium effects in the form of higher wages diminishing returns: the effectiveness of these pro- and prices. Second, some programs show evidence of grams decreases as transfers increase (Becker, Egger, substantial deadweight and displacement of existing and Ehrlich 2012; Cerqua and Pellegrini 2018). And jobs (Bronzini and de Blasio 2006). This evidence there is no evidence that their effects last for long is particularly strong concerning enterprise zones, (after the region loses eligibility for the program) which some countries have moderated by requiring (Barone, David, and de Blasio 2016; Becker, Egger, that supported firms demonstrate that they do not and Ehrlich 2018; Di Cataldo 2017). Recent liter- predominantly serve local markets and by requiring ature has emphasized the importance of thinking a certain percentage of workers in the zones to live about a region’s transportation network (Redding locally (see, for example, Mayer, Mayneris, and Py and Turner 2015) and incremental changes in road [2017] and Neumark and Simpson [2015]). infrastructure (Gibbons et al. 2019), finding positive Decisions about public sector employment, local effects from these changes on employment, the including the relocation of large public agencies to number of establishments, and, to a smaller extent, depressed areas, can also affect the spatial allocation the productivity of incumbent firms. These studies of employment. Evidence suggests that public sector show sizable local effects but they do not all iden- jobs have positive multiplier effects on employment in tify the aggregate effects from these improvements’ services and that the relocation of large public agen- impacts on the entire network. In a more recent cies has positive effects on overall local employment paper, Zarate (2020) shows that informal workers (Faggio and Overman 2014). However, more recent are more responsive to transportation costs than evidence points to negative effects of such moves on are their formal counterparts and that therefore the private sector employment in manufacturing (What former tend to work closer to their residences. As a Works Centre for Local Economic Growth 2019). result, investment in transportation infrastructure Note that general national-level policies, such as in Mexico City reduced informality by increasing funding for schools or training or even a nationwide access to formal jobs, which tend to be concentrated minimum wage, also affect spatial disparities. Toward an Integrated Policy Response 121

BOX 4.6 Evidence on the effects of place-based policies on mobility and labor market outcomes

Traditional place-based policies include mobil- postal codes within a metro area (the Small Area ity subsidies, rental assistance programs in Fair Market Rents program). They find that chang- high-opportunity areas, and informational interven- ing payment standards did not increase the rate of tions. Evidence on a mobility subsidy for unemployed moves to high-opportunity areas. In another pro- job seekers in Germany shows that it extended those gram that increased payment standards specifically job seekers’ search radiuses and increased their prob- in high-opportunity neighborhoods, only 20 percent ability of moving to a more distant region. The proof voucher recipients with children moved. In terms gram also led to higher job-finding rates and higher of programs providing information to those con- wages; the latter was mainly due to an improve- sidering moving, the available evidence is also not ment in workers’ job matches (Caliendo, Künn, and very encouraging, although it focuses mostly on the Mahlstedt 2017a, b). In a related paper on a Roma- United States. For example, Bergman et al. (2019) nian program to reimburse unemployed individu- find only limited effects from the provision of infor- als for the expenses of migration, the program was mation to families about the quality of schools observed to be effective at improving labor market associated with various rental units on a website outcomes (Rodríguez-Planas and Benus 2006). commonly used by voucher holders. Results from Bergman et al. (2019) use a randomized con- Schwartz, Mihaly, and Gala (2017) on light-touch trolled trial to evaluate the effects of an alternative counseling also indicate limited effects. Lagakos, approach: a US program called Creating Moves to Mobarak, and Waugh (2018) show that subsidizing Opportunity offers services to reduce barriers to mobility for rural-to-urban migration in Bangladesh moving into high-upward-mobility neighborhoods, has aggregate welfare impacts similar to uncondi- such as customized home search assistance, landlord tional cash transfers. They show that the welfare engagement, and short-term financial assistance. gains are highest for the poorest households, which The intervention increased the share of families had a greater propensity to migrate even before the who moved into high-upward-mobility areas from policy intervention. They also find that the main 15 percent in the control group to 53 percent in obstacle to mobility is the disutility of moving rather the treatment group.1 The researchers also evalu- than a failure or distortion in the housing market, ate the effects of more traditional programs offer- which implies that targeted policies to incentivize ing higher voucher payment standards in high-rent mobility would be ineffective.

than the reduction of mobility costs across of job opportunities created there and the sectors. region’s internal churning (that is, job switch- What are the driving mechanisms of this ing within the local labor market), both of effect? Artuc, Bastos, and Lee (2020) empha- which lead to a loss of welfare. size an important motivation for mobility: the Considering these new channels, Artuc, number of job opportunities provided by dif- Bastos, and Lee (2020) explore the role of ferent sectors and regions. First, if a worker mobility frictions faced by workers by quanti- has more job opportunities to choose from, the fying the effects of potential policies meant to best of those opportunities will likely deliver mitigate these frictions. The paper shows the greater welfare. Second, even if that worker is effects of higher mobility of workers across hit by a negative labor demand shock in the regions and sectors compared with two future, the worker will more likely be able to alternative scenarios: higher mobility across find another job without having to move to a only sectors and higher mobility across only different region or sector. A negative tempo- regions. Mobility frictions 20 percent lower rary shock in a region will reduce the number across regions and sectors reduce welfare 122 employment in Crisis

losses from the same benchmark shock by transportation infrastructure or by the level 16.5 percent, and the welfare-enhancing of competition in transportation and dis- effect of a policy of the same magnitude is tribution services, were approximated by greater when the policy targets regional fric- location fixed effects and thus were shown tions than when it targets sectoral fric tions. to lead to significant differences in tariff In particular, although the reduction in pass-through rates between metropolitan welfare is mitigated by 13.4 percent when regions. Similarly heterogeneous effects regional frictions alone are targeted, it is across localities of trade liberalization mitigated by only 2.3 percent when sectoral shocks have been observed in China (Han frictions alone are targeted. et al. 2016), India (Marchand 2012), and Why is there more pass-through to Mexico (Nicita 2009). For instance, after workers in some places than in others? Vijil et China’s entry to the World Trade Organi- al. (2020), studying Brazil from 1991 to zation, the market structure at the city level 1999, show that some metropolitan areas (as measured by the share of private sector were almost bypassed by the trade liber- participation in distribution services and in alization shock that occurred during that the production of final goods impacted by period because of their low internal market the shock, as proxies for the level of competi- integration. Market structures, influenced, tion) led to differences in the transmission of for instance, by the quality or quantity of tariff prices between cities: prices responded

FIGURE 4.18 Tackling structural issues that worsen the impacts of crises on workers

Competition policies

s • Address protectionism and unfair market conditions k r through better competition laws, fewer subsidies, o less state participation, and stronger procurement SHOCK w e practices

f Labor regulations WORKERS

c i • Reduce the pockets of labor rigidity (fewer

o restrictions on HR decisions) to speed up

o adjustments and shorten transitions

s Competition policies

e Local investment + place-based policies

l Labor regulations i

t S

Source: World Bank. Toward an Integrated Policy Response 123

more to the trade shock in cities that ben- financial strain if other types of crises occur. efited from higher competition (Han et al. Moreover, nationally administered income 2016). protection arrangements, such as unemployment insurance, have smoothed consumption What can be done? and served as automatic stabilizers in most Chapter 3 shows that the LAC region faces OECD countries. The costs of these programs structural issues that affect the magnitude of and the LAC region’s smaller tax base might the impacts of crises on workers. The policy necessitate a different approach for expanding implication of these findings and the related these programs in the region, such as combin- literature is that even if macroeconomic and ing individual savings and risk pooling. labor market policies are in pristine order, Other potential alternative mechanisms better outcomes to crises could be achieved include making the job retention schemes for workers by complementing these poli- currently employed for the COVID-19 cri- cies with sectoral and place-based policies sis permanent features of the LAC coun- to address the structural issues that impede tries’ respective economies by making them strong recoveries from crises and have state-contingent and automatically activated long-lasting productivity effects, as described when, for example, unemployment reaches a in this report. This change would involve certain rate or a recession worsens. By com- addressing the inefficiencies in labor market plementing existing adjustment-assistance adjustment caused by labor market legisla- mechanisms with countercyclical, publicly tion, product market structures, the lack of financed income support for affected individ- geographic mobility, and depressed areas. All uals, the LAC region could achieve smoother, things considered, a more complete charac- better-quality adjustments to crises and faster terization of the possible policy areas of focus recoveries from them. to tackle these structural issues (third policy Some crises, however, are unavoidable, dimension 3) is illustrated in figure 4.18. and better outcomes from them could be achieved if, in addition, the region tran- Conclusion sitions into augmented social protection programs that reduce scarring effects. The This chapter presented this study’s policy existence of these effects implies that the implications and the current context in the region could increase its long-term growth LAC region. It argued that LAC countries’ if crisis-induced, worker-level human capi- policy responses to crises need to squarely tal decay were reduced. This change would tackle three key dimensions of adjustment. require income support to cushion the short- These dimensions are not inconsistent, and term impacts of crises and protect welfare they have different weights in each country or as well as social protection policies aimed setting. They call for a triple entry of policies. at building human capital and promoting The first line of response to crises is poli- faster, better-quality transitions across jobs cies that lead to fewer crises and that smooth for displaced workers. Social protection sys- out their impacts at the aggregate level. tems provide more than just income support; Reducing the number of crises requires cre- they also help build human capital. For these ating a more stable macroeconomic environ- reasons, the second line of response includes ment and establishing adequate automatic deep reforms to the LAC region’s existing stabilizers that provide countercyclical, pub- social protection and labor programs. licly financed income support for people Traditionally, crises were viewed as transient negatively affected by labor market adjust- (as opposed to permanent) systemic (affecting ments. Prudent macroeconomic (fiscal and the whole economy) shocks. Although per- monetary) policies prevent certain types of manent systemic shocks such as trade liberal- crises and ensure the fiscal space needed ization and technological change also affect to provide support and avert system-wide employment and productivity, they do so 124 employment in Crisis

over long time horizons. Cycle-independent benefits to incentivize the return to work are (“secular”) forces cause some jobs to become key steps to ensure that social protection bet- permanently nonviable; these jobs will not ter cushions the short-term impacts of crises. rebound in the same firm, sector, or locality. Although it is clear that without job vacan- In contrast, effects from exchange rate fluc- cies, placements will not occur, a normal eco- tuations or changes in the terms of trade are nomic rebound from a crisis will include job more likely to be temporary. Understanding creation, and active searching is key to plac- is emerging that crises may have sticky effects ing workers in these new jobs. More robust on labor markets and productivity that differ and coordinated employment services, with a from those generated by technology or global- greater focus on results and unintended con- ization. However, because crises’ effects occur sequences, are thus needed. while changes in trends and structural factors But will macroeconomic stabilizers and and the normal churning of the economy are reforms to social protection and labor sys- already happening, distinguishing these effects tems be able to spur enough job creation to and better deploying programs to assist work- generate better recoveries? In light of the ers is difficult. evidence presented in this report, the LAC The standard advice in the presence of region urgently needs to tackle structural adverse permanent systemic shocks is to pro- issues in order to improve its response to cri- tect workers, not jobs—to prepare workers ses. The needed reforms include addressing for change rather than prevent the change the sectoral and spatial dimensions behind from coming about. Allowing sectoral or poor labor market adjustments. Without spatial restructuring is bound to increase addressing these fundamental challenges, efficiency; in contrast, retention subsidies recoveries in the region will remain char- and temporary employment programs delay acterized by sluggish job creation. In this (while the support lasts) but do not avoid context, competition policies, regional pol- job destruction. However, this advice does icies, and labor regulations are a third key not apply to systemic shocks that are only dimension of the policy response. This study temporary. In these cases, temporary reten- highlights, for example, the dichotomy tion programs may avoid the dissolution of between protected and unprotected firms in employer-employee matches that took a long the LAC region and the impact of low geo- time to build but are threatened by a tempo- graphic mobility among workers, both of rary shock, and they may stem productivity which serve to magnify the welfare effects losses from the unnecessary destruction of of crises. It also highlights pockets of labor job-specific human capital.19 rigidity that hinder necessary transitions When crises lead to permanent changes and adjustments in the labor market. in labor demand or supply, however, reskill- The policy implications of these findings ing initiatives and demand stimulus may be and of the related literature are that even if more appropriate responses. In addition, even macroeconomic, social protection, and labor though crises are systemic shocks, they gen- systems are pristine and flawlessly imple- erate highly heterogeneous effects across ini- mented, they are insufficient unless comple- tially similar workers, so adaptive programs mented with sectoral and place-based policies usually deployed to deal with more individual that address the underlying structural issues or idiosyncratic shocks (such as customized impeding strong recoveries from crises. The intermediation and job search support) may existing literature and policy experiences sug- be adequate to address them. Moving from gest that place-based policies could address budgeted programs and rationed cupos to the lack of geographic mobility and maximize protection guarantees (i.e., from assisting workers’ relocation potential. Reducing pock- only the chronically poor to offering benefits ets of labor rigidity (by loosening restrictions to all people in need), preventing the emer- on the human resource decisions of firms and gence of assistance “ghettos,” and structuring individuals) could speed up adjustments Toward an Integrated Policy Response 125

and shorten transitions. Similarly, addressing delayed investment. Third, some LAC coun- protectionism and unfair market conditions tries have exhibited strong policy responses through better competition laws, lower sub- to the crisis, although the effectiveness of sidies, less state participation, and stron- these responses has varied considerably. ger procurement practices could promote On the other hand, this crisis is not so stronger recoveries. LAC countries’ policy different from others before. A large part of responses need to tackle these issues, which the crisis’s effects on the LAC region derive will have different weights depending on the from the global recession, the sharp fall in country, the period, and other circumstances. demand for many months, and the possi- The COVID-19 pandemic is a convulsive, bility of financial crises in some countries. catastrophic crisis that is exacting a savage The region has a notable history of frequent toll on labor markets in the LAC region. and often severe economic slowdowns. What The region is experiencing an extraordinary happens to workers during these slowdowns rate of destruction of employment, massive is largely determined by aggregate demand negative income shocks, and rising levels of fluctuations (although some domestic cri- poverty. Between 35 and 45 million people in ses have been self-inflicted by economic the region may become newly poor in 2020 mismanagement). as a result of the pandemic, and although the This deep crisis arrived just as many gov- region’s middle class has grown significantly ernments in the LAC region were grappling since 2000, the crisis could reduce it by 5 with known structural challenges. It has percent, pushing out 32 to 40 million people accelerated some long-running structural (Diaz-Bonilla, Moreno Herrera, and Sanchez shifts that have been changing the nature Castro 2020). This shrinking of the middle of work, magnifying the crisis’s potential class and increase in poverty are driven by to further reduce employment opportuni- losses of labor earnings; the crisis is projected ties in what were traditionally considered to be the most severe labor market recession “good jobs”—the standard, stable, protected in some LAC countries’ history. Millions of employment associated with the formal workers in the region have lost their jobs, and sector (Beylis et al. 2020). millions more have seen significant reduc- The employment dynamics already tions in their earnings. And these losses are observed in many LAC countries will lead not expected to be evenly spread across the this crisis to cause sizable labor scarring income distribution—rather, the crisis could effects. Sector and location characteristics increase inequality substantially, pushing the are likely to further magnify these effects region’s Gini coefficient from 51.5 to as high for some workers. However, the three- as 53.4 (Diaz-Bonilla, Moreno Herrera, and dimensional policy framework presented in Sanchez Castro 2020). this study provides a roadmap that could Although this crisis—which was triggered lead to a more resilient recovery. How pub- by the public health imperatives of mitigat- lic and business policies address the current ing a global pandemic—is exceptional in challenges will shape the progress of the some ways, it is also yet another in a long LAC countries’ economies and the well-being series of aggregate demand shocks that have of their workers and citizens for decades. hit the LAC countries. On one hand, the The challenge is immense, but now is the crisis has several distinctive factors. First, defining time to take it on. the lockdown caused by the pandemic was bad for many jobs and worse for those for which home-based work is not an option (or Notes for workers who lack quality access to the 1. Although monetary and fiscal stabilization internet). Second, the prolonged uncertainty policies (including the management of a coun- around this crisis, particularly around the try’s capital account, exchange rate policy, form in which employment will rebound, has fiscal rules, and sovereign welfare funds and 126 employment in Crisis

the adjustment of its interest rate) are power- 8. This program’s combination of individual ful tools to respond to crises, they are not the savings accounts and risk pooling provides main focus of this study. effective financial support while incentivizing 2. These safety-net stabilizers work best when job searches and reemployment (Hartley, van supported by monetary and fiscal measures, Ours, and Vodopivec 2011). Four features of including exchange rate policies and capital the plan are particularly attractive. First, its account management; interest rates and other “hybrid” insurance model is able to address levers of monetary policy; fiscal rules and the needs of workers who change jobs fre- sovereign precautionary savings funds; and quently as well as those of the long-term access to global financial and risk-sharing unemployed (although it is debated whether markets and international risk-pooling mech- the maximum payout period from the anisms (such as the International Monetary risk-pooling component is adequate, given Fund and the World Bank). Each of these the observed duration of unemployment examples is supported by a vast academic and spells among the country’s lowest-paid work- policy literature. The discussion of such mea- ers). Second, the plan provides better levels sures in this report, however, is limited to the of compensation and consumption smooth- measures most directly linked to labor mar- ing than Chile’s purely noncontributory flat ket outcomes. unemployment benefit. Third, the plan’s ben- 3. These estimates are based on the World efits are indexed to protect their value from Economic Outlook. Inflation rates are com- inflation and to stabilize replacement rates at pounded. The values for Argentina for their starting levels. And fourth, the system 1981–97 are from the World Development has a sound financial basis, underpinned by Indicators. reserves that serve as an additional channel 4. Nominal downward wage rigidity is a feature of fiscal support to weather crises. Since its of most economies, and the LAC region is inception, the plan has included an automatic no exception (see Castellanos, García-Verdú, extension of benefits triggered when the and Kaplan [2004]; Dickens et al. [2007]; national unemployment rate rises above a cer- Holden and Wulfsberg [2009] and the refer- tain level. In the current COVID-19 contrac- ences therein; and Schmitt-Grohé and Uribe tion, the plan has also served as the platform [2016]). Moreover, the evidence suggests that for additional protections, such as subsidized the region’s lower inflation in recent decades furloughs. increased the downward rigidity of nominal 9. The “tax wedge” for formally employed work- wages. Hence, to the extent that the 2011–16 ers is the difference between take-home pay slowdown was marked by low and relatively and the total amount that the law requires stable inflation, real wage adjustments are that the employer and employee pay (including likely to have been lower during that period income taxes, statutory contributions to social than during the slowdowns and crises in the insurance, and other mandated benefits) in 1980s and 1990s, which were characterized order to have the employee (Summers 1989). by large increases in inflation. 10. There are no clear-cut guidelines to set 5. However, there have been some episodes of appropriate transfer amounts, and the appro- significant inflation spikes and corresponding priate level of benefits depends on the pro- reductions in real wages in the region since gram’s objective. The transfer values of CCT the early 2000s, including the 2004 bank- programs should therefore reflect such pro- ing crisis in the Dominican Republic, which grams’ twin objectives of reducing current resulted in a significant and long-term wage poverty among beneficiaries and providing correction. incentives for human capital accumulation 6. The discussion here focuses on job displace- (Grosh et al. 2008). One of the most generous ment and other livelihood losses caused CCT programs in the LAC region is Bolivia’s by shocks to aggregate demand. However, (which combined two CCT programs, the state-contingent support programs can help Bono Juancito Pinto and the Bono Juana households cope with a wide range of shocks. Azurdy), which provided 36 percent of prior 7. These figures are taken from the Interna- income, followed by Honduras’s Bono Pro- tional Labour Organization’s LABORSTA grama de Asignación Familiar, later renamed database. Bono 10,000 and now Bono Vida Mejor. Toward an Integrated Policy Response 127

11. Although the length of time CCT programs 17. However, these cross-country measures take to enroll new beneficiaries varies from fail to capture several mandated dismissal country to country, the Jamaican Program payments that are unique to particular coun- for Advancement through Health and Edu- tries, such as Ecuador’s deshaucio and the cation (PATH) sets its service standard for jubilacion patronal, which are described in the enrollment of new beneficiary households Gachet, Packard, and Olivieri (2020). to take no longer than four months after the 18. One such agenda is the new OECD Jobs household’s application. As of 2017, PATH Strategy, details about which can be found has been able to meet this service standard here: https://www.oecd.org/employment about 60 percent of the time. Its lengthy /jobs-strategy. intake processes imply considerable financial 19. Note that evidence shows that retention and political costs. subsidies (such as wage subsidies or pay- 12. Although this report focuses on social pro- roll tax rebates) can be effective at protecting tection policies and worker-level scarring, employment in firms going through difficul- scarring also occurs at the firm level. Min- ties, and these subsidies can be used to stim- imizing this scarring will require policies ulate the hiring of unemployed young people such as (a) providing financing to firms, and informal workers. But they may also lead (b) enhancing insolvency proceedings (as a to higher wages among incumbent workers key doing-business reform), and (c) improv- rather than greater employment, and they ing managerial capabilities for business con- can crowd out employment in nonsubsi- tinuity planning (to help businesses navigate dized firms and sectors. Similarly, temporary crises), as well as more generally reducing employment programs can be effective at barriers to firm entry and exit. Insolvency keeping people at work. But workers in tem- reforms aimed at preserving viable enterprises porary employment programs tend to have under temporary distress could also support low levels of job satisfaction, and their con- workers and reduce unnecessary frictional tracts are most often a palliative rather unemployment. And improving managerial than a steppingstone to more permanent capabilities would enhance firms’ productiv- employment. ity growth, which is likely to lead to firm survival and some employment retention (rather than the alternative of firm exit and conse- References quent job destruction). Almeida, R. K., and T. G. Packard. 2018. Skills 13. This review of the evidence concluded that and Jobs in Brazil: An Agenda for Youth. most traditional ALMPs have had at best Washington, DC: World Bank. modest impacts on employment; a typical Andersen, T. M. 2017. “The Danish Labor Market, intervention leads to a 2-percentage-point 2000−2016.” IZA World of Labor 2017: 404. increase in employment, which is usually not Antón, A., F. H. Trillo, and S. Levy. 2012. The statistically significant (McKenzie 2017). End of Informality in México? Fiscal Reform 14. For empirical evidence on the impacts of for Universal Social Insurance. Washington, social protection and labor policies and their DC: Inter-American Development Bank. possible distortions to labor markets, see Apella, I., and G. Zunino. 2018. “Nonstandard Frolich et al. (2014). Forms of Employment in Developing Coun- 15. Furthermore, changes in labor demand tries: A Study for a Set of Selected Countries in and supply factors driven by technologi- Latin America and the Caribbean and Europe cal change, regional and global integration, and Central Asia.” Policy Research Working and population aging threaten to render the Paper 8581, World Bank, Washington, DC. institutional apparatus of labor regulation Artuc, E., P. Bastos, and E. Lee. 2020. “Trade increasingly ineffective or counterproduc- Shocks, Labor Mobility, and Welfare: Evi- tive, as is discussed at length in Packard et al. dence from Brazil.” Background paper written (2019). for this report. World Bank, Washington, DC. 16. These data are available at https://www Azar, J., H. Hovenkamp, I. Marinescu, E. Pos- .doingbusiness.org/en/data/exploretopics ner, M. Steinbaum, and B. Taska. 2019. /employing-workers/reforms. “Labor Market Concentration and Its Legal 128 employment in Crisis

Implications.” OECD Seminars, Organisa- Betcherman, G. 2014. “Labor Market Regula- tion for Economic Co-operation and Devel- tions: What Do We Know about Their Impacts opment, Paris. https://www.oecd.org/els/emp in Developing Countries?” Policy Research /OECD-ELS-Seminars-Marinescu.pdf. Working Paper 6819, World Bank, Washing- Aznar, J., I. Marinescu, and M. I. Steinbaum. ton, DC. 2017. “Labor Market Concentration.” Work- Beylis, G., R. F. Jaef, R. Sinha, and M. Mor ris. ing Paper 24147, National Bureau of Economic 2020. Going Viral: COVID-19 and the Research, Cambridge, MA. Accelerated Transformation of Jobs in Latin Baker, J., and S. Salop. 2015. “Antitrust, Com- America and the Caribbean. Washington, DC: petition Policy, and Inequality.” Georgetown World Bank. Law Faculty Publications and Other Works Bowen, T., C. del Ninno, C. Andrews, S.Coll- 1462. https://scholarship.law.georgetown.edu Black, U. Gentilini, K. Johnson, Y. Kawasoe, /facpub/1462/. A. Kryeziu, B. Maher, and A. Williams. Barone, G., F. David, and G. de Blasio. 2016. 2020. Adaptive Social Pro tection: Building “Boulevard of Broken Dreams: The End of Resilience to Shocks. Washington, DC: World EU Funding (1997: Abruzzi, Italy).” Regional Bank Group. Science and Urban Economics 60: 31–8. Brand, J. E. 2015. “The Far-Reaching Impact of Becker, S. O., P. H. Egger, and M. von Ehrlich. Job Loss and Unemployment.” Annual Review 2010. “Going NUTS: The Effect of EU of Sociology 41: 359–75. Structural Funds on Regional Performance.” Bronzini, R., and G. de Blasio. 2006. “Evaluating Journal of Public Economics 94 (9): 578–90. the Impact of Investment Incentives: The Case Becker, S. O., P. H. Egger, and M. von Ehrlich. of Italy’s Law 488/1992.” Journal of Urban 2012. “Too Much of a Good Thing? On the Economics 60 (2): 327–49. Growth Effects of the EU’s Regional Policy.” Caliendo, M., S. Künn, and R. Mahlstedt. 2017a. European Economic Review 56 (4): 648–68. “The Return to Labor Market Mobility: An Becker, S. O., P. H. Egger, and M. von Ehrlich. Evaluation of Relocation Assistance for the 2013. “Absorptive Capacity and the Growth Unemployed.” Journal of Public Economics and Investment Effects of Regional Trans- 148: 136–51. fers: A Regression Discontinuity Design with Caliendo, M., S. Künn, and R. Mahlstedt. 2017b. Heterogeneous Treatment Effects.” American “Mobility Assistance Programmes for Unem- Economic Journal: Economic Policy 5 (4): ployed Workers: Job Search Behavior and 29 –77. Labor Market Outcomes.” Discussion Paper Becker, S. O., P. H. Egger, and M. von Ehrlich. Series 11169, Institute for the Study of Labor, 2018. “Effects of EU Regional Policy: Bonn, Germany. 1989–2013.” Regional Science and Urban Card, D., J. Kluve, and A. Weber. 2017. “What Economics 69: 143–52. Works? A Meta-Analysis of Recent Active Bekker, S. 2018. “Flexicurity in the European Labor Market Program Evaluations.” Wor king Semester: Still a Relevant Policy Concept?” Paper 21431, National Bureau of Economic Journal of European Policy 25 (2): 175–92. Research, Cambridge, MA. Benmelech, E., N. Bergman, and H. Kim. 2018. Casarín, D., and L. Juárez. 2015. “Downward “Strong Employers and Weak Employees: How Wage Rigidities in the Mexican Labor Market Does Employer Concentration Affect Wages?” 1996–2011.” Unpublished manuscript. Bank Working Paper 24307, National Bureau of of Mexico. Economic Research, Cambridge, MA. Castellanos, S. G., R. García-Verdú, and D. S. Bentolila, S., J. J. Dolado, and J. F. Jimeno. 2012. Kaplan. 2004. “Nominal Wage Rigidities “Reforming an Insider-Outsider Labor Mar- in Mexico: Evidence from Social Security ket: The Spanish Experience.” IZA Journal of Records.” Journal of Development Economics European Labor Studies 1 (4): 1–29. 75 (2): 507–33. Bergman, P., R. Chetty, S. DeLuca, N. Hendren, Castro, R., M. Weber, and G. Reyes. 2018. L. F. Katz, and C. Palmer. 2019. “Creating “A Policy for the Size of Individual Moves to Opportunity: Experimental Evidence Unemployment Accounts.” IZA Journal of on Barriers to Neighborhood Choice.” Work- Labor Policy 7: 9. ing Paper 26164, National Bureau of Eco nomic Cerqua, A., and G. Pellegrini. 2018. “Are We Research, Cambridge, MA. Spending Too Much to Grow? The Case of Toward an Integrated Policy Response 129

Structural Funds.” Journal of Regional Science Implications.” Working Paper WP/18/137, 58 (3): 535–63. International Monetary Fund, Washington, Céspedes, L. F., R. Chang, and A. Velasco. DC. 2014. “Is Inflation Targeting Still on Target? Downes, A., N. Mamingi, and R. M. B. Antoine. The Recent Experience of Latin America.” 2004. “Labor Market Regulation and International Finance 17 (2): 185–208. Employment in the Caribbean.” In Law and Clarke, J., S. Evenett, and K. Lucenti. 2005. Employment: Lessons from Latin America “Anti-competitive Practices and Liberalising and the Caribbean, edited by J. J. Heckman Markets in Latin America and the Caribbean.” and C. Pagés, 517–52. Chicago: University of World Economy 28 (7): 1029–56. Chicago Press. Cruces, G., A. Ham, and M. Viollaz. 2012. Ehrlich, M. V., and H. G. Overman. 2020. “Scarring Effects of Youth Unemployment and “Place-Based Policies and Spatial Disparities Informality: Evidence from Argentina and Bra- across European Cities.” Journal of Economic zil.” Center for Distributive, Labor and Social Perspectives 34 (3): 128–49. Studies (CEDLAS) working paper, Economics Faggio, G., and H. G. Overman. 2014. “The Department, Universidad Nacional de la Plata, Effect of Public Sector Employment on Local Argentina. Labour Markets.” Journal of Urban Econom- Da Silva Teixeira, G., G. Balbinotto Neto, and ics 79: 91–107. P. H. Soares Leivas. 2020. “Evidence on Farooq, A., A. D. Kugler, and U. Muratori. Rule Manipulation and Moral Hazard in the 2020. “Do Unemployment Insurance Benefits Brazilian Unemployment Insurance Program.” Improve Match Quality? Evidence from Recent International Journal of Social Science Studies U.S. Recessions.” Working Paper 27574, 8 (1). National Bureau of Economic Research, Cam- De Barros, R. P., and C. H. Corseuil. 2004. “The bridge, MA. Impact of Regulations on Brazilian Labor Feldstein, M., and D. Altman. 1998. “Unemploy- Market Performance.” In Law and Employ- ment Insurance Savings Accounts.” Working ment: Lessons from Latin America and the Paper 6860, National Bureau of Economic Caribbean, edited by J. J. Heckman and Research, Cambridge, MA. C. Pagés, 273–350. Chicago: University of Fernandes, A., and J. Silva. 2020. “Labor Market Chicago Press. Adjustment to External Shocks: Evidence for De Ferranti, D., G. Perry, I. Gill, and L. Serven. Workers and Firms in Brazil and Ecuador.” 2000. Securing Our Future in a Global Background paper written for this report. Economy. Washington, DC: World Bank. World Bank, Washington, DC. De Leon, I. 2001. Latin American Competition Fietz, K. M. 2020. “Unemployment Insurance in Law and Policy: A Policy in Search of Identity. Latin America and the Caribbean: A Compara- Kluwer Law International: London. tive Review of Current and Leading Practices.” Di Cataldo, M. 2017. “The Impact of EU Social Protection and Jobs, Human Develop- Objective 1 Funds on Regional Development: ment Department for Latin America and the Evidence from the U.K. and the Prospect of Caribbean, World Bank, Washington, DC. Brexit.” Journal of Regional Science 57 (5): Fiszbein, A., and N. R. Schady. 2009. “Condi- 814–39. tional Cash Transfers: Reducing Present and Diaz-Bonilla, C., L. Moreno Herrera, and Future Poverty.” Policy Research Report, D. Sanchez Castro. 2020. Projected 2020 World Bank, Washington, DC. Poverty Impacts of the COVID-19 Global Frolich, M., D. Kaplan, C. Pagés, J. Rigolini, Crisis in Latin America and the Caribbean. and D. Robalino, eds. 2014. Social Insurance, Washington, DC: World Bank. Informality and Labor Markets: How to Dickens, W. T., L. Goette, E. L. Groshen, Protect Workers While Creating Good Jobs. S. Holden, J. Messina, M. E. Schweitzer, Oxford: Oxford University Press. J. Turunen, and M. E. Ward. 2007. “How Furman, J., T. Geithner, G. Hubbard, and M. S. Wages Change: Micro Evidence from the Inter- Kearney. 2020. “Promoting Economic Recov- national Wage Flexibility Project.” Journal of ery after COVID-19.” Economic Strategy Economic Perspectives 21: 195–214. Group, Aspen Institute, Washington, DC. Diez, F., D. Leigh, and S. Tambunlertchai. 2018. Gachet, I., T. Packard, and S. Olivieri. 2020. “Global Market Power and Its Macroeconomic “Ecuador’s Labor Market Regulation: A Case 130 employment in Crisis

for Reform.” Unpublished paper. World Bank, The Design and Implementation of Effective Washington, DC. Safety Nets. Washington, DC: World Bank. Gambetti, L., and J. Messina. 2018. “Evolving Han, J., R. Liu, B. U. Marchand, and J. Zhang. Wage Cyclicality in Latin America.” The 2016. “Market Structure, Imperfect Tariff World Bank Economic Review 32 (3): 709–26. Pass-Through, and Household Welfare in Garganta, S., and L. Gasparini. 2015. “The Impact Urban China.” Journal of International Eco- of a Social Program on Labor Informality: The nomics 100: 220–32. Case of AUH in Argentina.” Journal of Devel- Hartley, G. Reys, J. C. van Ours, and M. Vodop- opment Economics 115: 99–110. ivec. 2011. “Incentive Effects of Unemploy- Gatti, R. V., K. M. Goraus, M. Morgandi, ment Insurance Savings Accounts: Evidence E. J. Korczyc, and J. J. Rutkowski. 2014. from Chile.” Labor Economics 18 (6): Balancing Flexibility and Worker Protection: 798–809. Understanding Labor Market Duality in Heckman, J. J., and C. Pagés. 2004. “Law and Poland. Washington, DC: World Bank. Employment: Lessons from Latin America and Gentilini, U., M. Almenfi, P. Dale, G. Demarco, the Caribbean—An Introduction.” In Law and I. Santos. 2020. “Social Protection and and Employment: Lessons from Latin Amer- Jobs Responses to COVID-19: A Real-Time ica and the Caribbean, edited by J. J. Heckman Review of Country Measures.” Social Protec- and C. Pagés, 517–52. Chicago: University of tion and Jobs Global Practice, World Bank, Chicago Press. Washington, DC. Holden, S., and F. Wulfsberg. 2009. “How Strong Gentilini, U., M. Grosh, J. Rigolini, and R. Is the Macroeconomic Case for Downward Yemtsov, eds. 2020. Exploring Universal Basic Real Wage Rigidity?” Journal of Monetary Income: A Guide to Navigating Concepts, Economics 56 (4): 605–15. Evidence, and Practices. Washington, DC: Holzmann, R., Y. Pouget, M. Vodopivec, and World Bank. M. Weber. 2012. “Severance Pay Programs Gerard, F., and J. Naritomi. 2019. “Job around the World: History, Rationale, Sta- Displacement Insurance and (the Lack tus, and Reforms.” In Reforming Severance of) Consumption-Smoothing.” Working Pay: An International Perspective, edited by Paper 25749, National Bureau of Economic R. Holzmann and M. Vodopivec. Washington, Research, Cambridge, MA. DC: World Bank. Gerard, F., J. Naritomi, and J. Silva. 2021. “The ILO (International Labour Organization). 2016. Effects of Cash Transfers on Formal Labor What Works: Active Labour Market Policies Markets: Evidence from Brazil.” Background in Latin America and the Caribbean. Studies paper written for this report. World Bank, on Growth with Equity. Geneva: International Washington, DC. Labour Organization. Gibbons, S., T. Lyytikäinen, H. G. Overman, and ILO (International Labour Organization). 2019. R. Sanchis-Guarner. 2019. “New Road Infra- World Social Protection Report 2017–19: structure: The Effects on Firms.” Journal of Universal Social Protection to Achieve the Urban Economics 110: 35–50. Sustainable Development Goals. Geneva: Giua, M. 2017. “Spatial Discontinuity for the International Labour Organization. Impact Assessment of the EU Regional Pol- ILO (International Labour Organization). 2020. icy: The Case of Italian Objective 1 Regions.” “Panorama Laboral 2020 América Latina y Journal of Regional Science 57 (1): 109–31. el Caribe.” Regional Office for Latin Amer- Greenstone, M., R. Hornbeck, and E. Moretti. ica and the Caribbean, International Labour 2010. “Identifying Agglomeration Spillovers: Organization, Geneva. Evidence from Winners and Losers of Large IMF (International Monetary Fund). 2010. World Plant Openings.” Journal of Political Econ- Economic Outlook, April 2010: Rebalancing omy 118 (3): 536–98. Growth. Washington, DC: International Mon- Grosh, M., M. Bussolo, and S. Freije, eds. 2014. etary Fund. Understanding the Poverty Impact of the Jalan, J., and M. Ravallion. 2003. “Estimating Global Financial Crisis in Latin America and the Benefit Incidence of an Antipoverty Pro- the Caribbean. Washington, DC: World Bank. gram by Propensity-Score Matching.” Jour- Grosh, M., C. Del Ninno, E. Tesliuc, and A. Ouer- nal of Business & Economic Statistics 21 (1): ghi. 2008. For Protection and Promotion: 19–30. Toward an Integrated Policy Response 131

Kanbur, R., and L. Ronconi. 2018. “Enforcement Outcomes: Evidence from France.” Journal of Matters: The Effective Regulation of Labour.” Economic Geography 17 (4): 709–52. International Labour Review 157 (3). McKenzie, D. 2017. “How Effective Are Active Kluve, J., S. Puerto, D. Robalino, J. R. Romero, Labor Market Policies in Developing Coun- F. Rother, J. Stöterau, F. Weidenkaff, and tries? A Critical Review of Recent Evidence.” W. Witte. 2016. “Do Youth Employment Pro- World Bank Research Observer 32 (2): grams Improve Labor Market Outcomes? A 127–54. Systematic Review.” Discussion Paper 10263, Messina, J., and A. Sanz-de-Galdeano. 2014. Institute for the Study of Labor, Bonn, Germany. “Wage Rigidity and Disinflation in Emerg- Konczal, M., and M. Steinbaum. 2016. “Declin- ing Countries.” American Economic Journal: ing Entrepreneurship, Labor Mobility, Macroeconomics 6 (1): 102–33. and Business Dynamism: A Demand-Side Messina, J., and J. Silva. 2020. “Twenty Years Approach.” Working Paper, Roosevelt Insti- of Wage Inequality in Latin America.” World tute, New York, NY. Bank Economic Review 35 (1): 117– 47. Kuddo, A., D. Robalino, and M. Weber. 2015. Mohl, P., and T. Hagen. 2010. “Do EU Structural Balancing Regulations to Promote Jobs: From Funds Promote Regional Growth? New Evi- Employment Contract to Unemployment dence from Various Panel Data Approaches.” Benefits. Washington, DC: World Bank. Regional Science and Urban Economics 40 Kugler, A. D. 2004. “The Effect of Job Security (5): 353–65. Regulations on Labor Market Flexibility. Mondino, G., and S. Montoya. 2004. “The Evidence from the Colombian Labor Market Effects of Labor Market Regulations on Reform.” In Law and Employment: Lessons Employment Decisions by Firms: Empiri- from Latin America and the Caribbean, edited cal Evidence for Argentina.” In Law and by J. J. Heckman and C. Pagés, 183–228. Chi- Employment: Lessons from Latin America cago: University of Chicago Press. and the Caribbean, edited by J. J. Heckman Lagakos, D., A. M. Mobarak, and M. E. Waugh. and C. Pagés, 351-400. Chicago: University of 2018. “The Welfare Effects of Encourag- Chicago Press. ing Rural-Urban Migration.” Working Morgandi, M., M. Ed, B. Wilson, A. Williams, Paper 24193, National Bureau of Economic and T. Packard. 2020. “Potential Responses to Research, Cambridge, MA. COVID-19 in Latin American and Caribbean Lederman, D., W. F. Maloney, and J. Messina. Countries.” Social Protection and Jobs Global 2011. “The Fall of Wage Flexibility.” World Practice, World Bank, Washington, DC. Bank, Washington, DC. Naidu, S., E. A. Posner, and E. G. Weyl. 2018. Levy, S. 2018. Under-Rewarded Efforts: The Elu- “Antitrust Remedies for Labor Market Power.” sive Quest for Prosperity in Mexico. Washing- Research Paper 850, Coase-Sandor Institute ton, DC: Inter-American Development Bank. for Law & Economics, University of Chicago, Lindert, K., T. G. Karippacheril, I. Rodriguez Chicago. Caillava, and K. Nishikawa Chavez. 2020. Neumark, D., and H. Simpson. 2015. “Place- Sourcebook on the Foundations of Social Pro- Based Policies.” In Handbook of Regional tection Delivery Systems. Washington, DC: and Urban Economics vol. 5B, edited by World Bank. G. Duranton, V. Henderson, and W. Strange, Maratou-Kolias, L., K. M. Fietz, M. Weber, and 1197–1287. Amsterdam: Elsevier T. Packard. 2020. “Quantifying and Validat- Nicita, A. 2009. “The Price Effect of Tariff Liber- ing the Cliffs of the Labor Market Regulation alization: Measuring the Impact on Household ‘Plateau’: A Global Review of Labor Market Welfare.” Journal of Development Economics Institutions.” Jobs Group, Social Protection 89: 19 –27. and Jobs Global Practice, World Bank, Wash- OECD (Organisation for Economic Co-oper- ington, DC. ation and Development). 2013. “The 2012 Marchand, B. U. 2012. “Tariff Pass-Through and Labour Market Reform in Spain: A Prelimi- the Distributional Effects of Trade Liberaliza- nary Assessment.” Organisation for Economic tion.” Journal of Development Economics 99: Co-operation and Development, Paris. 265–81. OECD (Organisation for Economic Co-operation Mayer, T., F. Mayneris, and L. Py. 2017. “The and Development). 2015. “Competition and Impact of Urban Enterprise Zones on Estab- Market Studies in Latin America: The Case lishment Location Decisions and Labor Market of Chile, Colombia, Costa Rica, Mexico, 132 employment in Crisis

Panama and Peru.” Organisation for Eco- Theory, a New Statistic, and Evidence from nomic Co-operation and Development, Chile.” Working paper 12757, National Bureau Paris. http://www.oecd.org/daf/competition of Economic Research, Cambridge, MA. /competition-and-market-studies-in-latin Pinelli, D., R. Torre, L. Pace, L. Cassio, and -america2015.pdf. A. Arpaia. 2017. “The Recent Reform of the OECD (Organisation for Economic Co-operation Labour Market in Italy: A Review.” European and Development). 2017. “Labour Market Economy Discussion Paper 072, Directorate Reforms in Portugal 2011–2015: A Prelimi- General for Economic and Financial Affairs, nary Assessment.” Organisation for Economic European Commission, Brussels. Co-operation and Development, Paris. Pinto, R. de Carvalho Cayres. 2015. “Three OECD (Organisation for Economic Co-operation Essays on Labor Market Institutions and and Development). 2019. “Special Feature: Labor Turnover in Brazil.” Departamento de Government Expenditures by Functions of Economia, Pontifícia Universidade Católica do Social Protection and Health (COFOG).” In Rio de Janeiro, Rio de Janeiro, Brazil. Government at a Glance 2019. Paris: OECD Portela Souza, A., G. Ulyssea, R. Paes de Barros, Publishing. D. Coutinho, L. Finamor, and L. Lima. 2016. Packard, T., U. Gentilini, M. Grosh, P. O’Keefe, “Rede de Proteção ao Trabalhador no Bra- R. Palacios, D. Robalino, and I. Santos. 2019. sil: Avaliação Ex-Ante e Proposta de Rede- Protecting All: Risk Sharing for a Diverse and senho.” Center for Learning on Evaluation Diversifying World of Work. Washington, Results (CLEAR), Fundação Getúlio Vargas DC: World Bank. Escola de Economia de São Paulo, Sao Paulo, Packard, T. G., and C. E. Montenegro. 2017. Brazil. “Labor Policy and Digital Technology Use: Redding, S. J., and M. A. Turner. 2015. “Trans- Indicative Evidence from Cross-Country Cor- portation Costs and the Spatial Organiza- relations.” Policy Research Working Paper tion of Economic Activity.” In Handbook of 8221, World Bank, Washington, DC. Regional and Urban Economics vol. 5, edited Packard, T. G., and C. E. Montenegro. 2021. by G. Duranton, J. V. Henderson, and W. C. “Labor Market Regulation and Unemploy- Strange, 1339–98. Amsterdam: Elsevier. ment Duration: Indicative Evidence from Robalino, D., and J. Romero. 2019. “A Purchaser Cross-Country Correlations.” Social Protec- Provider Split in Public Employment Services? tion and Jobs, Human Development Depart- Lessons from Healthcare Systems.” Social Pro- ment for Latin America and the Caribbean, tection and Jobs Global Practice, World Bank, Washington, DC. Washington, DC. Packard, T., and J. Onishi. 2020. “Social Insur- Robalino, D. A., J. M. Romero, and I. Walker. ance and Labor Market Policies in Latin 2020. “Allocating Subsidies for Private America and the Margins of Adjustment to Investments to Maximize Jobs Impacts.” Shocks.” Background paper written for this Jobs Working Paper 45, World Bank, Wash- report. World Bank, Washington, DC. ington, DC. Packard, T., and M. Weber. 2020. “Managing the Robalino, D., and M. Weber. 2014. “Designing Employment Impacts of the COVID-19 Cri- and Implementing Unemployment Benefit Sys- sis Policy Options for the Short Term.” World tems in Middle- and Low-Income Countries: Bank: Washington DC. Key Choices between Insurance and Savings Pellegrini, G., F. Terribile, O. Tarola, T. Mucci- Accounts.” Social Protection and Labor Dis- grosso, and F. Busillo. 2013. “Measuring the cussion Paper 1303, World Bank, Washington, Effects of European Regional Policy on Eco- DC. nomic Growth: A Regression Discontinuity Robertson, R. 2020. “The Change in Nature of Approach.” Papers in Regional Science 92 (1): Labor Market Adjustment in Latin America 217–33. and the Caribbean.” Background paper writ- Perry, G., W. Maloney, O. Arias, P. Fajnzylber, ten for this report. World Bank, Washington, A. Mason, and J. Saavedra-Chanduvi. 2007. DC. Informality: Exit and Exclusion. Washington, Rodríguez-Planas, N., and J. Benus. 2006. “Eval- DC: World Bank. uating Active Labor Market Programs in Petrin, A., and J. Sivadasan. 2006. “Job Secu- Romania.” Working Paper 2464, Institute for rity Does Affect Economic Efficiency: the Study of Labor, Bonn, Germany. Toward an Integrated Policy Response 133

Saavedra, J., and M. Torero. 2004. “Labor Mar- Brazil’s Productivity Agenda, edited by M. ket Reforms and Their Impact over Formal Dutz. Washington, DC: World Bank. Labor Demand and Job Market Turnover: The Vijil, M., V. Amorin, M. Dutz, and P. Olinto. Case of Peru.” In Law and Employment: Les- 2020. “The Distributional Effects of Trade sons from Latin America and the Caribbean, Policy in Brazil.” Background paper written edited by J. J. Heckman and C. Pagés, 131–82. for this report. World Bank, Washington, DC. Chicago: University of Chicago Press. What Works Centre for Local Economic Growth. Schmieder, J. F., and T. von Wachter. 2016. “The 2019. Local Multipliers. London: What Works Effects of Unemployment Insurance Benefits: Centre for Local Economic Growth. New Evidence and Interpretation.” Working Williams, A., and S. Berger-Gonzalez. 2020. Paper 22564, National Bureau of Economic “Towards Adaptive Social Protection Systems Research, Cambridge, MA. in Latin America and the Caribbean: A Syn- Schmitt-Grohé, S., and M. Uribe. 2016. “Down- thesis Note on Using Social Protection to Mit- ward Nominal Wage Rigidity, Currency Pegs, igate and Respond to Disaster Risk.” World and Involuntary Unemployment.” Journal of Bank, Washington, DC. Political Economy 124 (5): 1466–1514. Winter-Ebmer, R. 2000. “Long-Term Schochet, P. Z., R. D’Amico, J. Berk, S. Dolfin, and Consequences of an Innovative Redundan- N. Wozny. 2012. “Estimated Impacts for Partici- cy-Retraining Project: The Austrian Steel pants in the Trade Adjustment Assistance (TAA) Foundation.” Working Paper 2000-29, Program Under the (2002) Amendments.” Department of Economics, Johannes Kepler Office of Policy Development and Research, University, Linz, Austria. Employment and Training Administration, Wölfl, A., I. Wanner, O. Roehn, and G. Nicoletti. US Department of Labor, Washington, DC. 2010. “Product Market Regulation: Extending Schwartz, H. L., K. Mihaly, and B. Gala. 2017. the Analysis Beyond OECD Countries.” Eco- “Encouraging Residential Moves to Opportu- nomics Department Working Paper 799, nity Neighborhoods: An Experiment Testing Organisation for Economic Co-Operation and Incentives Offered to Housing Voucher Recip- Development, Paris. ients.” Housing Policy Debate 27 (2): 230–60. World Bank. 2012. World Development Report Subbarao, K., C. del Ninno, C. Andrews, and 2013: Jobs. Washington, DC: World Bank. C. Rodriguez-Alas. 2013. Public Works as a World Bank. 2015. The State of Social Safety Safety Net: Design, Evidence, and Implemen- Nets 2015. Washington, DC: World Bank. tation. Washington, DC: World Bank. World Bank. 2018. World Development Report Summers, L. H. 1989. “Some Simple Economics 2019: The Changing Nature of Work. Wash- of Mandated Benefits.” AEA Papers and Pro- ington, DC: World Bank. ceedings 79 (2). World Bank. 2020. “Additional Financing for Vegh, C. A., G. Vuletin, D. Riera-Crichton, D. Argentina Children and Youth Protection Proj- Friedheim, L. Morano, and J. A. Camarena. ect (P167851).” Project Appraisal Document, 2018. Fiscal Adjustment in Latin America and World Bank, Washington. DC. the Caribbean: Short-Run Pain, Long-Run Zarate, R. D. 2020. “Spatial Misallocation, Gain? Washington, DC: World Bank. Informality, and Transit Improvements: Vijil, M., V. Amorim, M. Dutz, and P. Olinto. Evidence from Mexico City.” University 2018. “Productivity, Competition and of California at Berkeley and World Bank, Shared Prosperity.” In Jobs and Growth: Washington, DC. ECO-AUDIT Environmental Benefits Statement

The World Bank Group is committed to reducing its environmental foot- print. In support of this commitment, we leverage electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Steward- ship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine–free (PCF), or enhanced elemental chl orine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility. A region known for its volatility, Latin America and the Caribbean (LAC) has suffered severe economic and social setbacks from crises—including the COVID-19 pandemic. These crises have taken their toll on careers, wage growth, and productivity. Employment in Crisis: The Path to Better Jobs in a Post-COVID-19 Latin America provides new evidence on the effects of crises on the region’s workers and firms and suggests several policy responses that can bolster long-term and inclusive economic growth. This report has three key findings. First, crises lead to persistent employment losses and accelerate structural changes away from the formal sector. This change occurs more through reductions in the creation of formal jobs than through job destruction. Second, some workers recover from crises, while others are permanently scarred by them. Low-skilled workers can suffer up to a decade of lower earnings caused by crises, while high-skilled workers rebound fast, exacerbating the LAC region’s high level of inequality. Formal workers suffer smaller employment and wage losses in localities with higher rates of informality. And the reduced job flows caused by crises decrease welfare, but workers in localities with more job opportunities, whether formal or informal, bounce back better. Third, crises’ cleansing effects can increase efficiency and productivity, but these effects are dampened by the LAC region’s less competitive market structure. Rather than becoming more agile and productive during economic downturns, protected sectors and firms gain market share and crowd out others, trapping valuable resources. This report proposes a three-pronged mix of policies to improve the LAC region’s responses to crises:

• Create a more stable macroeconomic environment to smooth the impacts of crises, including automatic stabilizers such as unemployment insurance and short-term compensation programs;

• Increase the capacity of social protection and labor programs to respond to crises and coalesce these programs into systems that complement income support with reemployment assistance and reskilling opportunities; and

• Tackle structural issues, including the lack of product market competition and the spatial dimension behind poor labor market adjustment—a “good jobs and good firms” agenda.

ISBN 978-1-4648-1672-7

SKU 211672