Patrick Baylis

In Chapter 2, I document the health consequences of forest fires in the Brazilian Amazon and examine whether access to public healthcare can mitigate potential adverse effects. I link comprehensive hospitalization and mortality records with satellite information on fire locations and wind patterns to identify the causal impact of fires on health for individuals across all age groups. I find that upwind fires increase hospital admissions for infants and children, which is driven by respiratory and circulatory diseases, as well as child and adult mortality. Access to public healthcare services, particularly community health centers and clinical doctors, strongly mitigates the effects, suggesting an essential role for healthcare delivery in mitigating the impacts of climate change.In Chapter 3, we study the impacts of fires in the Brazilian Amazon on electoral performance. We combine satellite data on fires and wind direction with administrative data on municipal elections at the polling station level to assess whether voters punish incumbent mayors running for reelection in locations adversely affected by fires before the elections. The empirical strategy explores exogenous variation in wind direction at the time of fire detection to estimate the differential effects of fires that generated more air pollution due to the direction in which the wind was blowing when they took place. We find that upwind fires significantly decrease incumbents' vote shares. The effects are robust across different specifications and are more adverse for places with lower educational levels, lower income, and more urbanized. Chapter 4 evaluates the impacts of a large cash transfer program in Brazil in which prenatal care was required for benefit receipt. Using administrative data from the program and the Brazilian Ministry of Health, we compare women at different stages of pregnancy when the program was implemented to identify the degree to which additional income and access to prenatal care affected birth outcomes and prenatal care utilization. We find that the benefit reduced the incidence of low birth weight, preterm births, and the likelihood of starting prenatal care late. Better prenatal care, in combination with the additional income, appears to drive improvements in birth outcomes.

View record

In Chapter 1, I study how newspaper closures affect environmental monitoring and enforcementactivities conducted by the U.S. Environmental Protection Agency. First, I propose a modelof regulatory behaviour where newspapers influence the regulator’s optimal probability of detectingenvironmental violations, by working as vehicles of information dissemination about firms’environmental performance. I then use a difference-in-differences estimator to estimate the impactof newspaper closures on enforcement activities conducted by the Environmental ProtectionAgency. I find that a daily newspaper closure leads to a drop in the number of inspections, detected violations, and enforcement actions that is within a range of 8-22 percentage points of their yearly averages.In Chapter 2, I investigate two mechanisms that are behind the drop in monitoring and enforcement activities following a newspaper closure. First, consistent with the information mechanism I propose in the model presented in Chapter 1, I find evidence that newspapers serve as an informant to the regulator about potential sources of environmental violations, reducing the cost of detecting a violation. Second, I find that newspapers inform regulated facilities about enforcement activities conducted in other facilities within the same state and industry. The second information mechanism enhances compliance spillover effects of EPA activities, improving the effectiveness of regulation enforcement.Chapter 3 uses a longitudinal survey from Indonesia to examine whether the effects of droughtson individual consumption are mitigated by the ability to migrate. By instrumenting between surveymigration with pre-existing migrant networks, we show that one standard deviation drop inannual precipitation in the origin location reduces consumption by 1.82% for non-migrant individuals, but that migrants are actually able to increase consumption. Given a one standard deviation drop in annual precipitation, the ability to migrate leads to an increase of about 13.9% in consumption over the medium run. Migration is a potentially important way to mitigate the costs of being exposed to extreme weather events that are a result of climate change, and these results suggest that removing barriers to migration is a promising strategy for mitigating climate damages.

View record

This dissertation presents three essays on Environmental and Development Economics. The first studies how economic activity erodes biodiversity and the role of decentralized institutions for achieving sustainable development. The second explores how structural change affects the geography of agriculture. The third focuses on legacies of coal power plant investments. This blend of short-term, spatial, and long-term analyses produces a comprehensive body of work that helps deepen our understanding of how economic activity shapes the environment. Chapter 2 documents biodiversity loss triggered by infrastructure expansion in India. Combining new data on infrastructure-driven deforestation with one million birdwatching diaries, and exploiting within-birdwatcher travel for identification, I find that infrastructure development drives 20% of total species loss and that species diversity does not recover in the medium run. Fortunately, species loss is more than halved when local institutions empower marginalized communities who are excluded from project planning. Informed consent by tribes is a key mechanism, underscoring the importance of grassroots institutions for balancing development and conservation. Chapter 3 studies the land use implications of internal migration in India. Using household microdata and a shift-share instrument for migration, we document sharp declines in crop production among migrant-sending households. Guided by a spatial equilibrium model, we find that non-migrant households living in the same village, as well as in more remote villages, contrastingly expand farming and adopt technology. Over half of aggregate production losses are cushioned by these spillovers, leading to a spatial reallocation of food production from urban to remote areas. Chapter 4 quantifies the long-run health effects of India’s coal power plants. Using an atmospheric dispersion model to compute an exogenous measure of cumulative exposure to power plant pollution, we find that a one standard deviation increase in this measure increases child mortality by 1.3 per 1,000 births. These effects are largely driven by exposure in utero, as well as exposure to private power plants built between 1992-2005. We find no evidence of differential economic development between more and less exposed districts, underscoring pollution as the main mechanism.

View record