This dissertation contains two chapters that chronicle the degree to which the way school and teacher inputs are allocated within and across American high schools is contributing to disparities in academic and labor market success. A third chapter examines whether such disparities might be partially redressed later in individual's careers via the acquisition of vocational re-training.
Recent research using matched student-teacher data has confirmed that teaching quality plays an important role in producing student test score improvement in elementary and middle schools. However, researchers have struggled to assess the extent to which teaching quality is inequitably distributed across students, as well as the degree to which the importance of good teaching depends on the school. The first chapter uses ten years of administrative data from North Carolina public high schools to address these issues directly. I estimate a flexible education production function in which student achievement reflects student inputs, teacher quality, school quality, and a school-specific scaling factor that allows the impact of teaching quality to vary across schools. The existence of nearly 3,000 teacher transfers, combined with an exogenous mobility assumption, allows separate identification of each teacher's quality from both school quality and school sensitivity to teacher quality. I find that teaching quality is surprisingly equitably distributed both within and across high schools. Schools predominantly serving underprivileged students employ teachers who are only slightly below average, and most students receive a mix of their school's good and bad teachers. Overall, I find that teacher and school inputs contribute only 4% to the achievement gap between the top and bottom deciles of an index of student background. Finally, I find that schools that disproportionately serve disadvantaged students tend to be more sensitive to teacher quality, so that policies that incentivize further equalization of teacher quality would be likely to both increase average performance and reduce inequality.
The second chapter examines the impact of high school quality on three subsequent outcomes of concern to economists and policymakers: high school graduation, college enrollment, and adult wages. We isolate the causal impact of school/neighborhood combinations from student sorting among schools by exploiting panel data from three national longitudinal surveys. Each survey features a stratified sample design that allows school averages of all observed student-level characteristics to be estimated. This facilitates a decomposition of each of our three outcomes into the within- and between-school contributions of both observed and unobserved student and family characteristics, as well as the contributions of observed and unobserved school and neighborhood variables that vary only across schools. Instead of attempting to disentangle school averages of individual-level unobservable inputs from school-level unobservable inputs, we estimate upper and lower bounds on the contribution of school quality to student outcomes. On the one hand, the vast majority of the variation in students' outcomes can be attributed to some combination of student inputs, parent inputs, and quality of schooling prior to high school. On the other hand, the small fraction of the variance attributable to differences in school quality translates into substantial impacts on high school graduation and college enrollment, since large numbers of students seem to be near the decision margin.
The third chapter exploits panel data from the 1979 National Longitudinal Survey of Youth to investigate the efficacy of vocational education as a means of refraining adults. Using Altonji, Elder and Taber's (2005) method to correct for selection into training, I estimate a 10-year profile of returns to vocational training for those changing jobs. While fairly imprecise, the estimates I obtain suggest that the average spell of vocational training increases hourly wages by 6 percent for the first couple of years after training and around 8–10 percent thereafter, with the effects persisting for at least 10 years. I also find positive effects of training on employment, weeks worked, and income, but these results are both less precise for each year and less consistent across years. Under further assumptions regarding the cost of training and the pattern of returns after ten years, I also show that the estimated returns are sufficient to make vocational training a sound investment.