George Bulman
Job Market Candidate

Fields:
Economics of Education, Public Economics,
Labor Economics

Expected Graduation Date:
June 2013

Dissertation Committee:

Caroline Hoxby:
choxby@stanford.edu

Nicholas Bloom:
nbloom@stanford.edu

Ran Abramitzky:
ranabr@stanford.edu

Job Market Paper

The Effect of Access to College Assessments on Enrollment and Attainment

This paper examines the college enrollment effects of attending a high school that hosts a college assessment test center on campus. Because the additional cost of taking the exam at a neighboring high school is very small, standard economic models suggest that there should be no effect. To test if this is true, I construct a new data set of every SAT and ACT test center in the United States and exploit within-school variation generated by the opening of new centers over time. Confidential student testing records linked to administrative college enrollment records reveal that students at high schools where new centers open are 8.2 percent more likely to take an exam, with the largest effects at low income schools. Using new centers as an instrumental variable, I demonstrate that 44.3 percent of these marginal test takers subsequently attend a four-year college and complete 5.8 semesters (typical of all matriculates). I then develop a method for inferring the latent distribution of student ability and show that the students induced to matriculate are primarily from the top tercile of the ability distribution. The large response of students to test centers is likely due to factors such as increased awareness about exam dates and an implicit recommendation by the school to take the exam. I replicate this empirical approach to evaluate district policies that compel nearly all students to take a college assessment. While this produces a ten-fold increase in SAT taking relative to new test centers, there are only slightly larger gains in college matriculation. Identifying if, and which students induced to take an exam matriculate and progress through college is potentially important given the magnitude of state and federal spending dedicated to promoting college attendance and the recent proliferation of policies to promote college enrollment through college assessments.

Students who attend schools in low income communities have significantly lower achievement and make smaller gains than their counterparts in high income communities. Three explanations, which are not mutually exclusive, are (i) that schools in low income communities have inadequate resources; (ii) that low income students have lower aptitude or inputs from their families; (iii) that schools in low-income communities are less productive than those elsewhere. Most research has focused on the two former explanations. In this study, I investigate the third by determining whether schools in lower income communities have smaller incentives to raise student achievement. Specifically, I test whether house prices (and, thus, local property tax revenue and school budgets) are more elastic with respect to achievement in higher income communities. Exploiting many regression discontinuities across school boundaries and using nearly universal data on Ohio house prices and characteristics, I demonstrate that that financial incentives for schools to raise achievement are greater in high income suburban districts than in low income urban districts.

With the intent of encouraging students to prepare for college, numerous school districts and a number of states have decided to make the taking of Preliminary SAT (PSAT) and Advanced Placement (AP) exams free. In this paper, I assess the short-term effects of such policies--on the decisions to apply to and enroll in college—and the long-term effects—earnings and employment. I investigate channels for the effects such as students’ matching themselves better to colleges and graduating more quickly. To evaluate the short-term effects of free PSAT and AP exams, I exploit the introduction of such policies in all U.S. schools. To evaluate the long-term effects, I exploit Florida schools’ staggered roll-out of the policies and use the state’s linked secondary, post-secondary, and employment database.

This research will use restricted access IRS tax data to estimate how the U.S. GDP and U.S. taxable income are affected by education-related tax credits, education-related deductions, and Title IV programs (such the Pell Grant). Regarded as an investment, the return to these programs takes the form of higher future incomes among individuals whom they induce to acquire more human capital. We estimate both the average returns and marginal returns on these tax expenditures by exploiting changes in eligibility requirements and federal appropriations levels over time. Average returns are based on the total change generated by each investment, while the marginal returns are those that would occur if the eligibility requirements were relaxed. The problem of student sorting on ability will be addressed in a novel way using access to college aptitude test data and college application data from the College Board.

We investigate whether Teach For America teachers cause students at their schools to apply to college more often, apply to colleges further from home, or apply to more selective colleges. Teach For America teachers typically graduate from a more geographically diverse and elite group of colleges than high school teachers on average. A potential effect of their presence in the classroom is exposing students to college options outside of those that they would typically consider. This may be especially likely in light of the fact that Teach For America teachers are placed in schools in low income communities where students may be less likely to learn about colleges from parents and peers. We construct a history of every Teach For America teacher placed at a high school, and we evaluate their effects using a differences-in-differences methodology with frontier matching and control techniques.