PUBLISHED ARTICLES
1. The effect of deforestation on COVID-19 transmission to Indigenous peoples in Brazil: A panel fixed-effects analysis before and after vaccination
Co-authored with Carolina Batista, Pedro Henrique Gagliardi, Rudi Rocha, Nicolas Ray
Published at PLOS Global Public Health on April 29, 2025 (free access)
Brazil had the second-largest death toll during the COVID-19 pandemic, with indigenous peoples disproportionately affected among ethnic groups. Parallel to the pandemic, Brazil has recorded the highest rate of deforestation globally, with encroachments into Indigenous territories putting climate stabilization and biodiversity at risk. However, the effects of deforestation on COVID-19 transmission to Brazil’s Indigenous peoples are unknown. This study shows that during the pre-vaccination period, deforestation partially explains COVID-19 transmission among Indigenous populations. Our main results for the pre-vaccination period indicate that a daily increase in deforestation per km2 is associated, on average, with the confirmation of 0.76 (p < 0.004, 95% CI: 0.240 – 1.276) new daily cases of COVID-19 among Indigenous peoples 14 days after deforestation warnings. Our estimates suggest deforestation explains at least 9.6% of all COVID-19 cases among indigenous populations. The association between the two variables disappears after the vaccination program. Our findings provide empirical evidence on the interplay between environmental degradation and negative health outcomes in a vulnerable segment of society in the context of a pandemic. Furthermore, these findings highlight the importance of the One Health approach to building preparedness for future pandemic threats.
Working paper published at CEPR’s COVID Economics.
Short version: VoxEU.org; VideoVox
Media coverage: Folha de S. Paulo, 27/10/20, ClimaInfo, Weekendavisen (Denmark), Uol (15/11/2021)
WORKING PAPERS
1. Tordesillas, Slavery and the Origins of Brazilian Inequality, co-authored with Felipe Valencia
This article documents the long-term impact of slavery on inequality at the receiving end. We focus on Brazil, the largest importer of African slaves and the last country to abolish this institution in the Western Hemisphere. To deal with the endogeneity of slavery placement, we use a spatial Regression Discontinuity Design (RDD) exploiting the Tordesillas Treaty, which established the colonial boundaries between the Portuguese and Spanish empires within modern-day Brazil. We find that the density of slaves in 1872 is discontinuously higher by 3 percentage points on the Portuguese side of the border, consistent with this power’s comparative advantage in transatlantic slavery. We then show how this differential slave rate led to higher income inequality in modern times by 0.04 points of the Gini index, 8% of the national average. In terms of mechanisms, we find higher structural transformation and foreign immigration during intermediate historical times as well as a wider racial income gap, differences in education, employment and prejudice against blacks in modern times.
Media coverage: El Pais, 20/11/17, Il sole 24 ore, Estado de S. Paulo, Nexo
2. Long-Term Effects of a Commodity Boom: Rubber Slavery in the Amazon, co-authored with Daniel Araujo, Dafne Murillo, Hector Paredes and Felipe Valencia
IDB WORKING PAPER No IDB-WP-1764
Can a brief period of economic prosperity leave a legacy of long-term adversity for local populations? This study investigates the enduring impact of the Amazon rubber
boom around 1900 on contemporary income, inequality, Indigenous groups presence, and forest conservation. Identification exploits variation in historical rubber suitability across municipalities and discontinuities around rubber concession boundaries. Municipalities with larger shares of rubber-suitable land experienced an initial economic surge, as evidenced by higher per capita GDP in the 1920 Census, but this prosperity was not sustained by 2010. Increased ethnic mixing, already visible in the 1872 Census, indicates
that earlier economic expansion intensified contact with Indigenous groups. In the long run, rubber-suitable areas show lower population density, higher extinction of Indigenous
groups, and greater income inequality. Consistent with the disproportionate violence and labor coercion inflicted on I ndigenous g roups, o ur Regression Discontinuity a nalysis further documents long-lasting environmental effects, with higher rates of deforestation, coca cultivation, and cattle raising in former rubber concession areas. Together, the results suggest that while the rubber boom generated short-term wealth, it left a legacy of persistent underdevelopment, social transformation, and environmental degradation.
*This project is supported by the Inter-American Development Bank and CAF. New versions to come.
3. Skin Color Politics: Evidence from Brazil, co-authored with Carlos Cavalcante and Kauê Moraes
This paper examines the effects of skin tone on political outcomes. We propose a new method of measuring skin color by using machine learning techniques. We, then, extracted the skin color of all candidates in the Brazilian elections from 2016 to 2022. Measuring racial effects using skin color measures has advantages. First, 25% of all politicians who run for more than one election change their self-declared race, implying a sizable bias and endogeneity. Our results suggest a reassessment of studies relying only on self-declared race, which seems to underestimate the effects of race in politics. Using a regression discontinuity design for close interracial elections, our results show that self-declared whites and non-whites do not have different skin colors. We find that dark-skin candidates spent less than light-skin candidates in 2016, while the opposite process was observed in 2020. In places where darker skin mayors win closed elections, the average income of mixed-race voters is higher. When forming their cabinets, dark-skin mayors chose 25 p.p. less self-declared nonwhite and 12 p.p less self-declared black cabinet members than light-skin mayors. We also find indications that electing darker skin mayors had a role model effect on black students before the COVID-19 pandemic. All those effects are muted or negative for self-declared white and self-declared non-white mayors in tight elections.
4. From Telegraphs to Space: Transport Infrastructure, Development and Deforestation in the Amazon
In this article, I exploit a source of quasi-random variation in observed infrastructure and develop a historical route opened by the Rondon Commission (1915-1917) as an instrumental variable strategy to investigate the impact of a national highway on the development of the Amazon region. Furthermore, I also explore potential transmission mechanisms, such as deforestation. My empirical tests rely on three data sources: night-time satellite data, census micro-data and deforestation satellite data. Using night-time satellite data, I find that for each kilometer’s distance away from the highway, the income of a pixel (0.86 km2) captured by the intensity of night-time light decreases by 0.10%. Transport infrastructure has also affected deforestation in the Amazon, open new frontiers for cattle farming and soybean agriculture. For each kilometer’s distance away from highway BR-364, the average percentage of trees per pixel increases from 5% to 9%, while this coefficient for census group data varies from 2% to 9%. I also observed that in Indigenous areas, the forest coverage by pixel is 24% larger in relation to other areas.
5. Ready to Tax: What Happens When Brazilian Municipalities Invest in Fiscal Capacity?
While the theme of fiscal capacity has gained increased attention in academia, there is little empirical evidence of its impact on development outcomes. This paper contributes to the existing literature by measuring the causality of investments in fiscal capacity of urban property taxes on development and political outcomes. The results show that investing in fiscal capacity increases property tax revenues per capita and property tax revenues over total revenue, as well as increases the provision of public goods that improve welfare, while it has no impact on the re-election of the municipalities’ mayors. I use a combination of the propensity score matching and difference-in-differences (DD) methodologies and find that investments in tax maps, for instance, increase property tax revenues by US$ 1.3 to US$ 1.6 per capita annually, given that the average property tax revenue per capita of the full sample is US$ 9.34. The results also show that the number of households where waste is collected increases 2,210 when a municipality invests in digital tax map in comparison to those that do not. Municipalities that invested in digital real estate database and digital tax map increases the vaccination coverage by 3.3 to 10.8% in comparison to those that did not. For political outcomes, I do not find evidence that investments in fiscal capacity affects re-election. Therefore, the political costs of investing in fiscal capacity might be counter-balanced by the benefits of the higher delivery of public goods.
RESEARCH IN PROGRESS
Colonial boundaries and state capacity, co-authored with Fernando Secco and Felipe Valencia (it will be available soon).
Indigenous health: accessibility and welfare, co-authored with Nicolas Ray
Humberto Laudares 