Three Essays in Applied Microeconomics

Public Deposited

I study three topics in applied microeconomics. My first chapter concerns the effect of daily school start times on academic achievement in Florida. Exploiting the sharp discontinuity in school start time relative to sunrise, I track children who move between schools on either side of the time zone boundary in Florida. Consistent with children's sleep schedules becoming later after entering puberty, I find that the benefits of later start times are concentrated among girls older than 12, and boys older than 10. I conclude that reordering start times so that elementary students started first, then middle school students followed by high school students would increase math achievement for high schoolers by 0.064 standard deviations and reading scores by 0.044 standard deviations without negatively impacting younger students or affecting the overall distribution of start times in each district.", "My second chapter studies how to measure the quality of \\textit{categorization workers --- judges, doctors, patent and benefits examiners --- who's job is to decide whether a case fits a given criteria. I call examiners who can perfectly rank cases according to their colleagues' understanding of the standard \\textit{consistent, and show that in institutional settings where different examiners study the same case consistency is identified. This chapter has implications for the fundamental fairness of the justice system, by estimating the extent to which defendants sentences depend on the identity of the judge who makes the decision. It is also relevant for understanding potential biases in examiner-assignment instrumental variables research designs, which have become increasingly popular in economics over the past several years. I show that my method can quantify the extent to which the monotonicity assumption is violated, and the implications for estimating treatment effects. Using a sample of Canadian refugee judges, I find evidence of widespread differences in judges in how they rank claimants. Judges with the same overall approval rate disagree on the correct decision about 13\\% of the time, roughly halfway between being perfectly consistent, and flipping coins to make decisions. I find little evidence that this will induce very much bias in treatment effect estimates using examiner assignment designs, but find that marginal treatment effect estimates may be very biased.", "Finally, my third chapter studies the effect of incarceration on the children of the incarcerated. Previous, correlational research has shown that the children of incarcerated individuals are very often incarcerated themselves. This may reflect either selection --- the children of the incarcerated are coming from a disadvantaged background in many ways that are observable and unobservable to the researcher --- or treatment --- parental incarceration actually \\textit{affects child outcomes --- but pre-existing observational designs were unable to separate these two explanations. I study a natural experiment in Ohio, where the county courts randomly assign cases to judges. Judges differ in their propensity to incarcerate defendants, and I use this variation as an instrumental variable for incarceration. I find that parental incarceration \\textit{reduces the likelihood of the child committing crimes or being incarcerated as an adult, with the reductions concentrated among children who's parent is a first time-defendant and accused of relatively minor crimes. I find similar effects for sibling incarceration, and conclude that incarceration of a family member increases the salience of punishment for children, reducing future criminality."]

Last modified
  • 11/24/2019
Date created
Resource type
Rights statement