Essays in Public Economics and Estimation

Abstract

This dissertation has two main themes. The first is using the tools of public and labor economics to estimate how policies and other forces impact the lives of disadvantaged populations so that we may better understand and address both problems faced and their possible solutions. The second applies ideas from econometrics and applied microeconomics to improve how we interpret data. The intersection of these themes in research offers new perspectives and insights into some of society's most pressing problems. The first chapter studies racial disparities in police use of force, in particular how they vary. Despite efforts to address systemic racial inequities in the criminal justice system, in many areas, our understanding of the problems faced is poor. I combine novel data from the hundreds of independently operated police departments covering New Jersey with empirical Bayes techniques to estimate how racial disparities in police use of force vary across departments. I find significant variation but also note that these disparities are difficult to predict, suggesting a role for more idiosyncratic departmental features. The second chapter, co-authored with Christina Korting, Jordan Matsudaira, Zhuan Pei, and Yi Shen, examines the econometric role of graphs, focusing on regression discontinuity designs (RDDs). We experimentally evaluate people's abilities to conduct "visual inference" with graphs to determine the presence of discontinuities in RDDs. Moreover, we compare visual and econometric inferences' performances, highlighting the value of graph-based visual inference and the existence of complementarities between them. Visual inference compares favorably with conventional econometric procedures, and we believe that there is an untapped frontier of graph-based econometrics to be explored. The third chapter studies migratory responses to local minimum wages. Using the universe of tax filers in the US, I find that local minimum wages result in negatively selected out-migration from treated counties. Disemployment effects are one explanation, but I observe some pass-through to home values, and these may temper the positive effects of higher wages and earnings. There are no apparent spillovers to neighboring untreated areas, which may limit spatial distortions from these policies.