Working Papers

More Gains Than Score Gains? High School Quality and College Choice. Most measures of high school quality concentrate on how students perform while they are in high school. If schools are intending to prepare their students for later challenges, such as college and careers, a school that produces good high-school students isn't much help if what they learn doesn't help them after they leave. I create a test-score value-added measure of high-school quality and then match it to transcript data from public colleges in Michigan, exploring the relationship between how well high schools raise test scores and how their graduates perform in their first-year college coursework. Overall effects are positive and highly significant, even after accounting for a litany of selection concerns. This may reassure observers who are worried that schools achieve high value added by teaching to the test, reallocating resources toward high-stakes subjects, or even cheating.

The Effect of Labor Market Shocks on College Attendance: Evidence from Plant Closings in Michigan. Human capital theory suggests that students should pursue more education when the local labor market is struggling, particularly if job losses are disproportionately concentrated among high school graduates, because these circumstances are associated with lower opportunity costs of additional schooling. However, credit constraints and incomplete information about financial aid options can prevent students from receiving their optimal level of education. Using WARN Act data on business closings and mass layoffs in Michigan in 2002-2011, I exploit the exogenous variation in local job losses in the year that students graduate from high school, examining changes in college-going rates and substitution between two-year and four-year colleges. A 1-standard deviation increase in per capita job losses is associated with a small but statistically significant 0.2-percentage point (0.3%) increase in the probability of attending college, driven entirely by attendance at two-year colleges; using the WARN job losses as an instrument for local unemployment rates produces similar results. In some subgroups, local job losses are associated with a small but significant decrease in attendance at four-year colleges. This result implies that both standard human capital theory and credit constraints have roles in guiding students' college-going decisions in the face of a turbulent labor market.

Estimating the Effects of a Large For-Profit Charter School Operator (with Sue Dynarski, Brian Jacob, and Silvia Robles). In this paper, we leverage randomized admissions lotteries to estimate the impact of attending a National Heritage Academy (NHA) charter school. NHA is the fourth largest for-profit charter operator in the country, enrolling more than 56,000 students in 86 schools across 9 states. Unlike several of the other large for-profit companies that operate virtual charters, NHA only has standard bricks-and-mortar schools. Our estimates indicate that attending a NHA charter school for one additional year is associated with a 0.04 standard deviation increase in math achievement. Effects on other outcomes are smaller and not statistically significant. In contrast to most prior charter school research that finds the largest benefits for low-income, underrepresented minorities in urban areas, the benefits of attending an NHA charter network are concentrated among non-poor students attending charter schools outside urban areas. Using data from a survey of school administrators in traditional public and charter schools, we document several aspects of school organization, culture and instructional practice that might explain these positive effects.