Research

Heat Before Birth: Primary Healthcare and the Long Run Cognitive Cost of Prenatal Temperature in Brazil. Evidence from 2007 to 2019 Brazilian national birth data

We estimate the long run cost of in utero exposure to high temperatures on adolescent cognitive achievement in Brazil and identify the institutional channels through which the effect operates. Combining more than 20 million SAEB test scores between 2007 and 2017 with high resolution ERA5 climate reanalysis and administrative birth and mortality records, we exploit within municipality variation in the average maximum temperature during the 40 weeks of gestation across birth cohorts under municipality by month, state by year, cohort, and test year fixed effects together with municipality specific linear trends. A one degree Celsius increase in average maximum prenatal temperature reduces adolescent test scores by 1.13 percent of a standard deviation, with effects of similar size on mathematics and Portuguese, a profile concentrated in the second and third trimesters, and inference that survives a Conley spatiotemporal correction at spatial bandwidths up to 100 kilometers. The penalty falls disproportionately on children of less educated mothers, on non white students, and on cohorts born in humid municipalities, and the year of birth specific coefficient declines steadily across cohorts born between 1988 and 2008.

We test three candidate channels. The evidence is strongest for healthcare access. The staggered municipal rollout of Brazil's Programa Saúde da Família  is associated with a reduction of roughly one third to two fifths in the long run heat penalty in covered municipality years, with an event study that exhibits parallel pretrends and a gradual post adoption buildup peaking five years after adoption. The agricultural income channel operating through heat tolerant C4 cropping, the mechanism emphasized in the prior developing country literature, delivers weaker and less robust evidence and an implied mitigation roughly an order of magnitude below benchmarks documented in Asian smallholder settings. Birth outcomes provide evidence on the immediate health channel, with a dose response on gestation length, birth weight, and preterm and low birth weight delivery that concentrates on the upper tail of the temperature distribution. The pattern of results is consistent with the long run cognitive cost of prenatal heat being sensitive to the institutional environment in which the heat shock lands, and identifies community based primary healthcare as the channel for which the moderation evidence is strongest. We cannot rule out that part of the estimated moderation is shared with adjacent Brazilian social programs that expanded over the same window, and we treat the PSF coefficient as a reduced form moderation effect rather than as a structural channel decomposition.

Acute Heat and High-Stakes Cognition: Within-Student Evidence from Cambodia, 2014–2023

Exposure to acute heat depresses cognitive performance, but whether long-run adaptation to chronically hot climates buffers or amplifies this cost remains an open question. We exploit within-student variation in daily heat-index conditions across the seven to eight subject exams that each Cambodian Grade 12 student sits over consecutive days, combining administrative records on 731,279 candidates of the National Examination from 2014 to 2023 with ERA5-Land reanalysis data matched to each school's location. A one-standard-deviation increase in the local heat-index z-score on an exam day reduces a student's subject score by about 0.018 standard deviations, conditional on student, subject, and year fixed effects, with effects robust to controls for precipitation, time-of-day scheduling, and district-specific trends. Strikingly, the effect is roughly five times larger in schools located in the hottest baseline quartile than in the coolest — the opposite of what an acclimatization story would predict, but consistent with hot-climate students operating closer to physiological and cognitive thresholds. Negative effects are concentrated in the social sciences track and are somewhat larger for male and rural students. The results suggest that adaptation to chronic heat does not protect performance from acute exam-day shocks; in tropical, lower-resourced education systems, procedural reforms to high-stakes testing — exam-hour scheduling, contingency rules for heat events — deserve as much policy weight as classroom cooling.

EU Regulations to Fight Nutrient Surpluses: the Case of Nitrate Vulnerable Zones

Co-authors:  Santiago Guerrero

Nutrient balance indicators can act as a signal for the potential environmental impact of agriculture on water and air. This article estimates the impact of the designation of nitrate vulnerable zones (NVZ) in the European Union (EU), the most comprehensive regulation for curbing water pollution caused by nitrates from agricultural sources in the EU mandated by the Nitrates Directive. Using the OECD agri-environmental indicators database on nutrient balances covering 47 countries and a set of policy variables constructed from EU official documents, we leverage the timing of NVZ designation across EU countries to identify sharp sustained reductions of 12%–18% in Nitrogen pollution, and 20%-25% in Phosphorus. Nonetheless, such reductions are only observed in countries that implemented the Nitrogen Vulnerable Zones in the whole territory as opposed to countries who implemented them partially. The impacts of the NVZ policy are mainly explained by reductions in fertilizer use. The results are robust to numerous robustness checks. They suggest a successful policy tool for tackling Nutrient surpluses.

More than just friends: heterogeneous peer effects in health habits 

This article explores peer effect heterogeneity in adolescent adjusted Body Mass Index (BMI), physical activity and fast food consumption. In particular, this paper makes an original contribution by studying heterogeneous peer effects based on friendship intensity. Adolescents are assumed to interact through a social network, where they have strong and weak friendships. To identify both types, I use Add-Health's wave II friendship roster questionnaire to calculate a friendship score for every friend listed by each student in the sample: friends with a high score are classified into the strong friendship network and the rest into the weak friendship network. We expect strong friendships to have a greater effect on individuals' observed health outcomes. I use a 2SLS strategy to estimate the econometric model, which exploits the network structure as an exogenous source of variation to identify the causal effect of peers choices on individual outcomes. Results provide evidence that supports the heterogeneous peer effect hypothesis: only strong peers have a significant impact on adjusted body weight, physical activity and diet choices. I conduct three different robustness tests: 1) using different threshold levels of the friendship score to define the strong and weak friendship networks; 2) including romantic partners; and 3) studying long-run peer effects (e.g. the effect of peers during wave II on outcomes in wave IV). The robustness tests offer a similar insight to the main results.

The Causal Effect of Physical Activity on Health  in Early Adulthood: A Gene By Environment Instrumental Variables Approach. 

Co-author: Lise Rochaix

This article explores the effect of Physical Activity (PA) on subjective health status and non-communicable diseases (NCDs) during early adulthood. In particular, it analyzes the respective impacts of leisure time physical activity (LTPA) and heavy work-related physical activity (WTPA). To deal with a potential endogeneity bias, we combine an instrumental variables approach. We instrument LTPA by using the Physical activity polygenic score and heavy WRPA by using the academic performance polygenic score . Results suggest that LTPA has a significantly positive effect on the number of NCDs and self-reported health status. WRPA has the opposite effect: it decreases subjective health status and increases the number of NCDs. Lastly, we explore potential heterogeneities between LTPA and WRPA, and we find that only sedentary workers experience a positive effect on health by increasing their LTPA. 

Eating like an American: Impact of Free Trade Agreements on Unhealthy and Healthy Food Imports in the Americas. (2020)

 This paper investigates the impact of FTAs with the US on unhealthy and healthy food imports in the Americas by analyzing a rich dataset of bilateral exchange from the MIT Observatory of Economic Complexity. The identification strategy relies on a Difference-in-Differences estimation. The results show that FTAs play a substantial role in reshaping the food environment of a domestic country. Depending on the region, the DiD estimates range from  $16 to $75 USD increase in per capita unhealthy food imports after implementing an FTA with the US. I also find evidence that FTAs increase the availability of healthy food, though to a lesser extent ($13 to $54 per capita). This suggests that FTAs disproportionately contributed to increasing imports of foods with poor nutritional content in the Americas. In order to relax the parallel trends assumption and provide robustness checks, I employ an augmented synthetic control method (ASCM)  for the CAFTA and NAFTA regions. I then extend the SCM to consider all countries in the Americas that signed an FTA with the USA within a staggered adoption context. The SCM results support DiD findings.  In contrast to previous research, which are single country studies that have focused on either Canada or Mexico and only analyze sugar categories, this paper provides three original contributions. First, I analyze both types of food imports: healthy and unhealthy. Second, this research considers all countries in the Americas that have signed FTAs with the US and not just a subset. Lastly, I use two novel synthetic control methodologies that correct for poor pre-treatment fit and which are flexible to the staggered adoption setting.