Essays on Measuring Climate Change Damages and Adaptation

Abstract

Through changes in average temperature, precipitation patterns, and extreme weather events, climate change is already causing severe ecological and economic damages. Further warming is expected to have a profound effect on the functioning of ecological and human systems worldwide. While it is a top priority to limit carbon emissions and mitigate future climate change, it is also essential to prepare for damages from climate change in the remainder of this century. Research is needed to understand these impacts, and whether it is possible to adapt to these changes.

In this thesis, I measure damages and adaptation to recent climate change in three essays. First, in joint work with Sylvia Klosin, I develop a novel debiased machine learning approach to measure continuous treatment effects in panel settings. We demonstrate benefits of this estimator over standard machine learning or classical statistics approaches. We apply this estimator to measure the degree of damages from climate change in U.S. agriculture, and find that extreme heat is significantly more damaging than linear models suggest. In the second essay, I measure the degree of adaptation to extreme heat in U.S. agriculture using flexible modeling of weather variables and a debiased machine learning estimator. I demonstrate that my double machine learning approach works well in high-dimensional settings. Applying this estimator to the past thirty years of crop yields, I find evidence of considerable adaptation to extreme heat. Finally, I examine the equity of adaptation to increasing wildfire risk in California. I study how electric utilities’ power shutoff decisions correlate with community socioeconomic status and health risk factors.