Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. Chris Robinson; Dr. Nirav Mehta


This thesis consists of three policy-motivated chapters in the area of applied microeconomics. In chapter 1, I estimate the impact of English-language courses on the wages of new immigrants. I develop a model of immigrants' investment in language skills which may affect wages directly, as well as change the proportion of pre-immigration skills transferred into the host-country economy. Using unique panel data, LSIC, I find that attending language courses for six months leads to a 0.3 standard deviations gain in language skills, corresponding to an average wage increase of 11.7 percent. The increase in the total return to language skill accounts for 5.5 percent of wage growth, while the remaining 6.2 percent is driven by the transfer of pre-immigration skills. Chapter 2 examines the determinants of school choice and its effect on student outcomes. For any school choice policy to be successful, parents must select schools based on attributes that improve students' academic achievement. Using ECLS-K data, I find that students who move schools for academic reasons suffer a decline in their math performance. I estimate a random utility model of parental school choice and a test score production function to provide an explanation for this finding. Parents seem to select schools based on their socio-economic attributes while ignoring attributes important for test score production. Potentially, this results in worsened academic performance of their children. The Becker (1968) model of crime establishes the importance of the probability of apprehension as a key factor in a rational individual's decision to commit a crime. Most empirical studies based on US data have relied on variation in the number of police officers to estimate the impact of the probability of apprehension or capture. In chapter 3, the probability of apprehension is measured by clearance rates and their effects on crime rates are studied using a panel of Canadian provinces from 1986 to 2005. OLS, GMM, GLS, and IV estimates yield statistically significant elasticities of clearance rates, ranging from -0.2 to -0.4 for violent crimes and from -0.5 to -0.6 for property crimes. These findings reflect the importance of police force crime-solving productivity.