Francisco Costa

Assistant Professor of Economics at Fundação Getulio Vargas (FGV EPGE) 

Research interests: Environmental/Energy Economics, Development Economics.

francisco.costa [at]
(+55) (21) 3799-5871
Praia de Botafogo, 190/1100, 
Rio de Janeiro, RJ, 22250-900, Brazil.


Homicides and the Age of Criminal Responsibility: A Density Discontinuity Approach, with João de Faria, Felipe Iachan and Bárbara Caballero. [replication files] [Coverage: Época OnlinePortal BrasilExame.comTerraJornal do BrasilEBCTrubuna da BahiaIGO PovoEconomìa, the journal of LACEA, Fall 2018.
A growing body of evidence documents that policies can affect household behaviors persistently, even if they are no longer in place. This paper studies the importance of such ``hysteresis'' -- the failure of an effect to reverse itself as its underlying cause is reversed -- for the welfare evaluation of corrective policies. First, we introduce hysteresis into the textbook framework used to derive canonical sufficient statistics formulas for the welfare effect of corrective policies. We then derive new formulas allowing for hysteresis. We show that, under certain conditions, the persistent effect of a short-run (i.e., temporary) policy becomes a new key statistic for evaluating the welfare effect of such a policy, and also of a long-run (i.e., permanent) version of a similar policy. Second, we estimate the persistent effect of a short-run policy, for which we argue that these conditions are met, in a policy-relevant context: residential electricity use in a developing country setting. We estimate that about half of the dramatic short-run reductions in residential electricity use induced by a 9-month-long policy that was imposed on millions of Brazilian households in 2001 persisted for at least 12 years after the policy ended. Finally, we combine our estimates with our framework to illustrate the implications that hysteresis can have for the welfare evaluation of corrective policies. 

Preserving wilderness ecosystems in developing countries is challenging because their remote location places them far from state control. We investigate this using 30x30 meter satellite data to determine how Amazonian deforestation changes discretely at the Brazilian international border. In 2000, Brazilian pixels were 30 percent more likely to be deforested, and between 2001 and 2005 annual Brazilian deforestation was more than 3 times the rate observed across the border. In 2006, just after Brazil introduces policies to reduce illegal deforestation, these differences disappear. These results demonstrate the power of the state to affect whether wilderness ecosystems are conserved or exploited.

Stop Suffering! Economic Downturns and Pentecostal Upsurge, with Angelo Marcantonio and Rudi Rocha(revised version soon)
This paper estimates the effects of economic downturns on religious conversion to Pentecostal Evangelicalism across Brazilian regions between 1991 and 2010. We find that regions more exposed to economic distress experienced an increase in Pentecostal affiliation during the 1990s, accompanied by a decrease in adherents to other Christian denominations. Our estimates show that this conversion persisted over the following decade. We also show that economic downturns are associated with the growth in the vote share of candidates explicitly connected to Pentecostal churches in national elections in both the short and long run. These results uncover a causal link between economic distress and the political entrenchment of more fundamentalist religious groups in a secular democracy.

Unveiling Physicians’ Choice of Practice Location: Evidence from Observed Migration, with Leticia Nunes and Fabio Sanches
(draft very soon!)
This paper exploits the revealed preferences of all generalist physicians graduated in Brazil between 2001 and 2013 to examine the factors influencing their decision of practice location after graduation. We estimate physicians' locational preferences using a random coefficients discrete choice model. The model allows us to simulate the effects of policies that could be used to increase the fraction of physicians working in underserved areas. We find that wages, though relevant, are not the main factor behind physicians' locational preferences. Health infrastructure, physicians' place of birth and graduation are more important. Affirmative action policies such as quotas in medical schools for students born in poorer areas and the opening of medical schools in underserved areas would improve the geographic distribution of physicians at a lower cost than financial incentives.

Power Plants and Deforestation: Recent Evidence from the Amazon
, with Juliano Assunção and Dimitri Szerman(draft very soon!)
This paper estimates the impacts of the construction of the new hydroelectric power plants (HPPs) built in the Brazilian Amazon between 2003 and 2011 on local deforestation patterns. We assemble a geo-referenced panel dataset combining high-resolution deforestation data with the comprehensive list with the locations of all HPPs’ feasibility studies in the region. We use the synthetic control method to identify the causal impact of the actual construction of each HPP on forest cover on different perimeters around the construction site. We find that the construction process increases deforestation within 15 kilometers from the HPP. Despite stimulating local deforestation, some plants have indirect positive local effects by avoiding deforestation on areas ranging between 15 and 50 kilometers from the HPPs. Overall, the construction of the 11 plants in the Amazon was responsible for 7\% of the observed deforestation in areas within 50 kilometers from the construction site.

Local Economic Impacts of Brazilian Hydroelectric Power Plants, with Juliano Assunção and Dimitri Szerman.
This paper estimates the short- and medium-run effects of the construction of large hydroelectric power plants (HPPs) on the economic development of Brazilian municipalities. We use the synthetic control method to perform one case study for each of the 82 municipalities affected by an HPP between 2002 and 2011. Three main findings emerge. First, the median impact of the construction of HPPs on the local economy follows an inverted U-shape over a five-year horizon (the average construction time is four years). Second, the median impact of the construction in the medium-run is positive but modest, with no effect on the composition of local economic activity. Third, the estimated effects display a lot of dispersion. These results do not provide support for the view that large construction works unequivocally spur local development.

Too hot to learn? Evidence from grade progression and school dropouts, with Diana Goldemberg.
This study analyzes the impact of cumulative heat exposure on educational outcomes of children from 6 to 18 years old. We use data from over 78 thousand schools and 114 million enrollments in the hottest and poorest regions of Brazil between 2007 and 2016. We exploit weather variations across school-grade-year in all 1,794 Northeast municipalities. We find that hotter school days during the academic year reduces grade progression, with extreme heat being particularly detrimental. A one-degree Celsius increase in daily maximum temperatures reduces grade progression rates by 0.43 percent, and increases dropout rates by 3.86 percent. Results are consistent with established links between temperature and cognitive performance and have implications for evaluating the welfare-burden of climate change in developing countries.