Essays in Policy Evaluation
This dissertation contains three independent papers (chapters). Chapter 1 argues that researchers that use a regression discontinuity design with parametric controls for the assignment variable face significant uncertainty as to the appropriate model. I argue that researchers should incorporate this model uncertainty into their inference through model averaging. I show in a Monte Carlo experiment that Frequentist model averaging, compared to pretesting, leads to more precise estimates and less Type I error. Chapter 2 studies the impact of state-level tax incentives for the film industry, which have exploded since the early 2000s. I study how these incentives affect filming location and the number of jobs and businesses in the film industry. I seek to answer the question: can these incentives create a local film industry? I find that while these incentives have large effects on filming location, they almost never lead to increases in jobs and employment in the film industry. Chapter 3 studies the Prudence Kay Poppink Act, effective 2001, which broadened California's disability discrimination in employment law. This expanded coverage to those with less severe impairments. I study how this expansion affected labor market outcomes for the disabled by comparing the disabled in California before and after the Act to the disabled in other states during the same time period. I augment this difference-in-difference methodology by adding a comparison to the non-disabled as well. I find that this expansion of legal protections increased employment and possibly increased wage and salary rates.