Intergenerational Mobility in the Land of Inequality (with A. Fonseca, P. Pinotti, B. Sampaio and L. Warwar)
Revised and Resubmitted, Review of Economics and Statistics
See our Mobility Atlas Brazil
We provide the first estimates of intergenerational income mobility using population-wide administrative data for a large developing country, namely Brazil. We measure formal income from tax and payroll data, and we train machine learning models on census and survey data to predict informal income. We develop novel methods to quantify and characterize the estimation bias resulting from income imputation and other sources of measurement error, and show that such bias remains negligible in our context. A 10 percentile increase in parental income rank is associated on average with a 5.5 percentile increase in child income rank, and only 2.5% of children born to parents in the bottom quintile reach the top quintile. Mobility varies widely by gender, race, and geographical areas, and causal place effects explain 57% of variation in mobility across regions. Parental income impacts several long-term outcomes beyond income, including education, teenage pregnancy, crime victimization, and mortality.
The Kids Aren't Alright: Parental Job Loss and Children's Outcomes Within and Beyond Schools (with C. Melo and B. Sampaio)
Revise and Resubmit, Review of Economic Studies
We study the impacts of parental job loss and unemployment benefits on children's education in Brazil, using rich individual-level data on employment, school enrollment, and unemployment insurance for the entire population. Leveraging mass layoffs for identification, we find that parental job loss has a significant adverse impact on children's educational outcomes. School dropouts and age-grade distortion increase by up to 1 and 2 percentage points. The effect is concentrated on disadvantaged families, persisting for at least six years and leading to lower high school completion rates. We further show that children aged 14-17 are more likely to work informally and to commit crimes following parental displacement. In turn, children in advantaged families are more likely to move to lower-quality schools due to parental displacement. Using a clean regression discontinuity design, we show that access to unemployment benefits mitigates some of the adverse impacts of parental job loss on children. Our findings indicate that the income losses following parental displacement are an important mechanism of the effects on children, highlighting the importance of policies that provide income support for displaced workers. Other explanations related to family rupture, migration to poorer neighborhoods, and changes in household production do not receive much support from the data.
Job Loss, Unemployment Insurance and Health: Evidence from Brazil (with G. Amorim, A. Fonseca, and B. Sampaio)
Revise and Resubmit, Review of Economics and Statistics
We study the effects of job loss and unemployment insurance (UI) on health among Brazilian workers. We construct a novel dataset linking individual-level administrative records on employment, hospital discharges, and mortality for a 17-year period, rarely available in the context of developing countries. Leveraging mass layoffs for identification, we find that job loss increases hospitalization (+33%) and mortality risks (+23%) for male workers, while women are not affected. These effects are pervasive over the distribution of age, tenure, income and education, and men's children are also negatively affected. Remarkably, about half of these impacts are driven by external causes associated with accidents and the violent Brazilian context. Using a regression discontinuity design, we show that access to UI partially mitigates the adverse effects of job loss on health. Our results indicate that the health costs of job loss are only partially explained by the income losses associated with job displacement.
Small Children, Big Problems: Childbirth and Crime (with R. Hsu Rocha, P. Pinotti and B. Sampaio)
May 2025 submitted
We investigate the effect of having a child on parents' criminal behavior using administrative data from Brazil. Fathers' criminal activity increases sharply by up to 10% during the pregnancy period, and by up to 30% two years after birth, while mothers experience only a transitory decline in criminal activity around childbirth. The effect on fathers lasts for at least six years and explains 5% of the overall male crime rate. Domestic violence within the family also increases after childbirth. These results are based on a difference-in-difference design leveraging variation in the timing childbirth, and further supported by an analysis leveraging the exogeneity of twin births. The generalized increase in fathers' crime stands in sharp contrast with previous evidence from developed countries, where criminal behavior by both parents markedly decreases with childbirth. Our findings can be explained by the costs of parenthood and the prevalence of poverty among newly formed Brazilian families. Consistent with this explanation, we provide novel evidence that access to maternity benefits largely offsets the increase in crime by fathers after childbirth.
A Few Bad Apples? Criminal Charges, Political Careers, and Policy Outcomes (with G. Daniele, M. Le Moglie, P. Pinotti and B. Sampaio)
June 2025 submitted
We study the prevalence and effects of individuals with past criminal charges among candidates and elected politicians in Brazil. Individuals with past criminal charges are twice as likely to both run for office and be elected compared to other individuals. This pattern persists across political parties and government levels, even when controlling for a broad set of observable characteristics. Randomized anti-corruption audits reduce the share of mayors with criminal records, but only when conducted in election years. Using a regression discontinuity design focusing on close elections, we demonstrate that the election of mayors with criminal backgrounds leads to higher rates of underweight births and infant mortality. Additionally, there is an increase in political patronage, particularly in the health sector, which is consistent with the negative impacts on local public health outcomes.
Job Displacement, Unemployment Benefits and Domestic Violence (with S. Bhalotra, P. Pinotti and B. Sampaio)
Review of Economic Studies (2025)
We estimate impacts of male job loss, female job loss, and male unemployment benefits on domestic violence in Brazil. We merge employer-employee and social welfare registers with administrative data on domestic violence cases brought to criminal courts, use of public shelters by victims and mandatory notifications of domestic violence by health providers. Leveraging mass layoffs for identification, we find that both male and female job loss, independently, lead to large and pervasive increases in domestic violence. Exploiting a discontinuity in unemployment insurance eligibility, we find that eligible men are not less likely to commit domestic violence while benefits are being paid, and more likely to commit it once benefits expire. Our findings are consistent with job loss increasing domestic violence on account of a negative income shock and an increase in exposure of victims to perpetrators, with unemployment benefits partially offsetting the income shock while reinforcing the exposure shock.
Access to Justice and Social Protection (with L. Germinetti, F. Gerard, J. Naritomi and B. Sampaio)
American Economic Association, Papers and Proceedings (forthcoming)
Governments in developing countries are expanding social protection policies, yet coverage remains imperfect. This paper explores how the justice system influences coverage and the consequences of unequal access to justice for targeting. Using administrative microdata from Brazil, we document how two distinct groups – displaced workers and the elderly poor – resort to the courts to secure social protection. Using the justice system for this purpose correlates with key individual characteristics – notably, income and geographical distance from courts – indicating significant barriers to accessing justice with implications for program targeting.
The Effect of Job Loss and Unemployment Insurance on Crime in Brazil (with P. Pinotti and B. Sampaio)
Econometrica (2022) [Lead Article]
We investigate the effect of job loss and unemployment benefits on crime, exploiting unique individual-level data on the universe of workers and criminal cases in Brazil over the 2009-2017 period. We find that the probability of criminal prosecution increases on average by 23% for workers displaced upon mass layoffs, and by slightly less for their cohabiting sons. Using causal forests, we show that the effect is driven entirely by young and low tenure workers, while there is no heterogeneity by education and income. Regression discontinuity estimates indicate that unemployment benefit eligibility completely offsets potential crime increases upon job loss, but this effect completely vanishes immediately after benefit expiration. Our findings point at liquidity constraints and psychological stress as main drivers of criminal behavior upon job loss, while substitution between time on the job and leisure does not seem to play an important role.
The Employment Effects of Lump-Sum and Contingent Job Insurance Policies: Evidence from Brazil
Review of Economics and Statistics (2022)
Lump-sum job displacement policies (e.g. severance pay) are often presented as a better alternative to contingent policies (e.g. unemployment insurance) in the context of developing countries, under the rationale that the former are less harmful to formal employment as they do not incentivize substitution from formal to informal jobs. First, this paper provides original evidence on the employment effects of lump-sum income in the context of a developing country with high labor informality. Using Brazilian data, a regression discontinuity (RD) design shows that a transfer equivalent to 15 days of earnings (i) increases the duration out of a formal job by 1.9 weeks, (ii) reduces monthly earnings in the next job by 1.6%, and (iii) reduces total earnings in the formal labor market by 3.6% over a three-year period. Second, the paper studies the impact of a one-month extension in unemployment insurance (UI) on a comparable sample of displaced workers. UI is shown to have a stronger impact on the duration out of a formal job compared with a lump-sum transfer. In addition, a novel exercise matching administrative and survey data shows that 57% of the decrease in formal employment caused by UI is compensated by an increase in the incidence of informal employment. However, workers receiving the UI extension partially recover the initial employment loss over time in such a way that the adverse impact on employment over a three-year period is similar compared with the lump-sum transfer. Moreover, UI is found to be less harmful to re-employment wages, possibly because it improves the worker's bargaining power as it offers insurance against the duration of joblessness. Overall, the UI extension is less detrimental to total earnings in the formal labor market over a three-year period. Hence, although these findings indicate that contingent job insurance policies have a stronger impact on the initial duration out of a formal job and indeed incentivize informal employment, they do not support the notion that lump-sum policies are less harmful to formal employment and earnings in the medium term.
Corruption and Legislature Size: Evidence from Brazil (with S. Fiorin)
European Journal of Political Economy (2020)
This paper studies whether and how legislatures affect political corruption. Using a regression discontinuity design in the context of Brazilian municipalities, we find a positive causal impact of council size on corruption levels, as detected by random federal audits. This indicates that an extra councilor represents an additional political actor potentially interested in diverting public resources, which we define as a rent extraction effect. However, we find further evidence that, in some contexts, larger councils enhance the representation of opposition parties and effectively increase monitoring over the executive, attenuating the rent extraction effect. Namely, in municipalities where opposition parties are typically underrepresented, the additional seat in the council is absorbed by the opposition and corruption outcomes do not worsen. In addition, only in such context, mayors are more commonly sentenced for misconduct in office by judicial authorities, whose investigations anecdotally often originates from councilors denouncing mayors to local courts. Overall, our findings show that legislature size is detrimental to corruption outcomes but less so where the representation of opposition parties improves with the enlargement of the legislature.
Let the Water Do the Work: Climate Adaptation Policies and Individual Welfare
(with Y. Barreto, B. Carrillo, D. Da Mata, L. Emanuel & B. Sampaio)
Dry Lives: Climate Adaptation and Mortality in the Semi-arid Regions of Brazil
(with C. Imbert, B. Sampaio, G. Ulyssea & A. Fonseca)
Parenthood and Productivity
(with C. de Holanda, B. Ferman, A. Fonseca, B. Sampaio & L. Warwar)
Conditional Cash Transfers and Intergenerational Mobility
(with A. Fonseca, P. Pinotti, B. Sampaio & L. Warwar)
Family Matters: Politics, Public Procurement, and Jobs in Brazil
(with A. Fonseca, M. Gehrke, P. Pinotti & B. Sampaio)
Household response to health shocks and its intergenerational consequences
(with B. Sampaio and M. Eliúde)