Research

with Nagisa Tadjfar

Draft Paper

Abstract: Low-income students are less likely to attend elite universities than equally qualified high-income peers, in large part because they apply at lower rates. We study whether this reflects lack of exposure to students who have attended top universities, and how exposure affects students’ perceptions. Using UK administrative data, we exploit “breakthrough” events when a school first sends a student to a top university. Applications from that school to that university subsequently rise by 30%. This access promotes upward mobility: marginal entrants graduate at typical rates and earn £4,000 more annually than matched control students, despite coming from relatively poor backgrounds. To understand why students who lack exposure might not apply, we turn to a field experiment in British schools. We find that a primary barrier is students' beliefs about their social fit. At baseline, low-income students are more pessimistic about their social fit at elite universities, but not their chances of receiving an offer or graduating. Students randomly assigned to view short videos of undergraduates discussing their experiences are 6 percentage points more likely to apply to the speaker's university. This treatment makes students more positive about their social fit at that university, with no effect on other beliefs. Finally, when matched with mentors, students primarily seek out information about social life. Our findings highlight perceptions of the social environment at elite universities as a central barrier to applications and illustrate scalable treatments to promote access and social mobility.

with Nagisa Tadjfar

Draft Paper

Abstract: Do high-stakes standardized tests expand or inhibit opportunity for low-SES students? We answer this question in the context of the UK's staggered elimination of pre-university exams in favor of teachers' predicted exam grades. Eliminating testing increases the university enrollment of low-income students by 3 percentage points (7%), while leaving wealthy students' enrollment unchanged. Marginal students induced to enroll in university attain employment at better firms and, in expectation, earn £50,000–£100,000 more over their careers, in net present value. Paradoxically, standardized exams exhibit no calibration bias against marginal low-income students—accurately predicting their university success—whereas teacher-supplied grades are systematically biased in their favor. Despite proper calibration, standardized tests inhibit low-SES students by deterring human capital investment. When tests are eliminated, 5% of low-income students shift into academic tracks. These findings highlight how disparate impacts can arise even when screening algorithms are unbiased. When the measurement of information itself poses a direct disutility, standardized tests generate disparities that commence earlier in the pipeline.

with Phi Adajar and Nagisa Tadjfar

Slides

Competition for talent can cause labor market unraveling, where institutions create inefficient matches by extending offers before candidate ability is fully revealed. We study the impacts of early offers in the UK college admission system on college sorting and match quality. We build a theoretical model of student and university choices, generating three predictions about the resulting match, which we then validate empirically in this setting. First, universities with lower student quality are more likely to give early offers; we find the universities in the lowest quintile of yield are 9.4pp more likely to use early offers than the highest quintile. Second, early offers divert high-ability students away from the most competitive universities; in our context, students are 9.6pp less likely to attend an elite university. Finally, also consistent with our model, we find that universities preferentially target high-achieving students with early offers. These early offers also benefit students directly: students who accept early offers are 6.2pp more likely to graduate on time, though after three more years, this gap is statistically indistinguishable from zero. Universities collectively banned these offers in 2021; to understand the impact of this ban, we build a structural model to evaluate the effects on student match quality and labor market outcomes, and compare this system to alternative market designs.

with Taha Choukhmane, Cormac O’Dea, Jonathan Rothbaum, and Lawrence Schmidt

The elasticity of intertemporal substitution is a key parameter in models in macroeconomics and public finance, but credible estimates of this parameter require exogenous variation in the intertemporal price of consumption. We use variation in the formula by which employers match their employees’ retirement savings contributions in the United States to credibly estimate this parameter jointly with inertia in savings contributions. We link administrative data on earnings and retirement plan contributions for the US population with data on the retirement savings policies at over 100,000 firms. We make use of bunching at kink points in the budget set induced by employer matching, employee responses to moves across firms, employee responses to within-firm plan changes, and responses to automatic enrollment policies to separately identify inertia, risk aversion, and the elasticity of intertemporal substitution. We develop a life-cycle model to exploit these different sources of quasi-experimental variation to estimate the level and heterogeneity in the elasticity of intertemporal substitution across the population.

with Aroon Narayanan

Do different CEOs within the same firm systematically set different markups, or are markups determined solely by firm-level optimization? To answer this question, we estimate a Two-Way Fixed Effects (TWFE) model of firm markups on CEO and firm dummies. We use the De Loecker et al. (2020) framework to estimate firm-year level markups, and use CEO movements between firms to identify CEO effects on markups. We address limited mobility bias using the leave-out estimator of Kline et al. (2020). To enable meaningful comparisons across different connected sets of firms and CEOs, we apply the normalization procedure of Best et al. (2023). After applying these corrections, we estimate that CEO effects explain 10-15% of the overall variance in markups.

IFS Briefing Note BN234, with Barra Roantree

PDF