Overview
During the fire season across Brazil's Legal Amazon from 2009 to 2019, more than a third of all respiratory disease hospitalizations were among children under 14. Deforestation is a classic externality where the people clearing trees capture the economic gains, but surrounding communities bear the health costs from smoke and changes in local climate. Existing studies have linked deforestation to hospitalizations, malaria, diarrhea, and child stunting, but they mostly look at one country or one health outcome at a time, and none use causal identification across multiple countries simultaneously.
I'm proposing an instrumental variables strategy to estimate whether deforestation causes worse health outcomes for young children in tropical developing countries. The core challenge is that deforestation is endogenous. It clusters in poor areas with bad governance, so it's hard to separate the effect of tree loss from underlying poverty. My proposed instrument exploits the fact that unusually dry years cause more burning during land clearing. Whether a given year is dry is effectively random with respect to child health, but it strongly predicts how much fire and smoke communities near deforestation experience.
The data will come from three sources: the Hansen Global Forest Change Dataset for satellite-derived tree cover loss, NASA FIRMS for active fire data, and DHS Surveys for child health outcomes across tropical developing countries.
This is still in the early stages. I have the research design and data sources planned out, and I'm working on assembling the data and running the analysis.