Three Essays Using Natural Experiments to Measure Causal Effects on Educatio

This dissertation covers two topics. The first in two essays, each using variation in the timing of statewide deregulation of bank branching to estimate causal effects of credit availability on high school graduation. Between 1970 and 1999, 39 states reduced legal geographic restrictions on the location of bank branches. These bank branching deregulations represent quasi-random increases in credit availability to residents of the deregulating state relative to residents of regulated states. In the first essay, using data on state of residence and educational attainment from the Current Population Survey March Supplement, I estimate bank branching deregulation's impact on high school graduation by regressing a binary indicator of high school attainment on a treatment variable. This treatment variable indicates whether an individual's graduation from high school was plausibly affected by bank branching deregulation. I find bank branching deregulation significantly increases the likelihood of graduating high school by 1 percentage point. In the second essay, I continue to investigate how bank branching deregulation affects high school graduation. In place of the CPS data, I use National Longitudinal Survey of Youth 1979 data to mitigate both measurement error introduced when identifying treatment and omitted variable bias. I also use the difference-in-discontinuities method, which should improve similarity of unobservable characteristics by finding the local average treatment effect. I again find positive and significant effects of bank branching deregulation on high school graduation. The third essay comprises the second topic covered in this dissertation, namely, does big-time college football promote other university outcomes. Football Bowl Subdivision colleges and universities support their athletics departments to the tune of $20 million per university per year. Advocates of university sponsored athletics suggest there are large advertising effects of fielding big-time college football and men's basketball teams. Using athletic success to measure changes in the advertising effect fails to control for unobservable, institution-specific factors that influence university outcomes.Therefore, we use conference changes to represent quasi-random shocks to the aforementioned advertising effect. Using difference-in-differences methods we estimate average treatment effects of conference switching with Integrated Postsecondary Education Data System (IPEDS) data for 90 NCAA D1 FBS universities over the years 2000 to 2015. Difference-in-differences estimations provide weak evidence that conference switching reduces applications and ACT scores of incoming students and increases state appropriations, calling into question the subsidization of these athletics programs.