RESEARCH

Abstract. 

Abstract. Student loan debt has risen sharply over the past two decades and now constitutes a large share of overall household liabilities. At the same time, homeownership gaps have emerged, and individuals who have held student debt are less likely to own homes by middle age. This paper studies how, and to what extent, mortgage underwriting rules affect access to homeownership for potential homebuyers with student debt. We exploit 2017 policy changes by Fannie Mae and Freddie Mac that relaxed how student loan obligations enter debt-to-income (DTI) calculations. Using individual-level credit bureau and mortgage servicing data (CRISM–CCP), we implement difference-in-differences and triple-differences designs comparing potential homebuyers differentially affected by the reform. Our primary comparison contrasts potential homebuyers with student loans enrolled in income-driven repayment (IDR) plans with those in standard repayment, alongside broader comparisons between potential homebuyers with and without student debt. To interpret these effects, we develop a structural lifecycle model of housing, borrowing, and default in incomplete markets. The model incorporates mortgage underwriting constraints, including DTI and loan-to-value (LTV) limits, and allows student loan obligations to affect borrowing capacity through required payments and balance sheet dynamics. This framework captures how interactions across financial obligations shape access to mortgage credit, the timing of homeownership, and housing wealth accumulation.

Abstract. Trade liberalization can reshape local workforce skill composition by altering the skill premia, in turn affecting both human capital accumulation and worker mobility. Using Brazil’s 1990s tariff reductions and the local labor market exposure measure of Dix-Carneiro and Kovak (2015, 2017), we show that overall population responses to the trade shock are limited, but mask large differences across education groups. More exposed labor markets experience slower growth in the number of high school-educated individuals across all age cohorts. We interpret these patterns as the joint result of two adjustment margins. Trade-induced declines in the skill premium between high school and non-high school workers reduce incentives for younger cohorts to acquire education, while also encouraging more educated adults to move out from the most adversely affected regions. To quantify these mechanisms, we develop an overlapping generations model in which forward-looking agents make education and location choices in the presence of a skill agglomeration externality. The model allows us to assess how trade-induced changes in local labor demand generate persistent divergence in regional skill composition and heightened spatial inequality. We will further use the model to evaluate the extent to which subsequent expansions in education access and quality in Brazil may have mitigated these effects.

Abstract. This paper examines the effects of a large need-based scholarship in Michigan, the Tuition Incentive Program (TIP). TIP covers in-district community college tuition and fees, can be stacked on top of most other financial aid sources (such as the Federal Pell Grant), bases eligibility on categorical participation in Medicaid, does not require annual certification of need, and communicates eligibility early in students’ academic career. Despite these advantageous features, participation in the program remains low. Using linked administrative data on 1.2 million Michigan high school graduates, we find null effects on enrollment with a regression discontinuity design but positive effects using a matching approach. These contrasting results suggest that students on different margins may vary in their awareness of the program; barely eligibles are the least likely to know about the program. To investigate these findings, we develop and estimate a structural model of college choice with endogenous aid application and take-up, in which some students lack information about the program. Information depends on individual- and school-level factors driving awareness, such as attachment to the safety net and presence of high school counselors. Preliminary simulations suggest that aid effectiveness is moderated by awareness and application barriers; policies that improve each of these (individually and especially simultaneously) greatly increase the effect of TIP on enrollment.

Abstract.

Abstract. Understanding how need-based student aid programs interact or "stack" to determine a total aid package is crucial for maximizing financial aid. If students lack this knowledge, when new programs are introduced, they may make changes to their take-up decisions that reduce their total aid receipt. This paper examines whether schools' adoption of last-dollar local student aid programs causes students to substitute away from Michigan's first-dollar Tuition Incentive Program, when doing so can reduce their total aid receipt.  A difference-in-differences approach that leverages the staggered adoption of these local aid programs suggests that TIP take-up decreases by about 15% among 2-year enrollees and their TIP aid received falls by around $167 per pupil, or about 20%. However, differential trends in outcomes before program adoption threatens a causal interpretation of these results. With these caveats in mind, our results suggest that the adoption of last-dollar local programs may have inadvertent consequences for students who misunderstand differences between first- and last-dollar aid programs with the most harm concentrated among the neediest students.

Abstract. In the United States, social welfare programs often have low take-up, where only a fraction of eligible beneficiaries receive the resources that they are eligible for. We examine this problem through the lens of Michigan’s Tuition Incentive Program (TIP), a state grant aid program that provides free community college to low-income students based on their childhood participation in Medicaid. To investigate TIP take-up, we conduct a large-scale mixed-methods study using comprehensive data on over one million Michigan public school students over 11 cohorts, and 55 interviews with program administrators, high school counselors, and financial aid staff. We find that while one third of Michigan high school graduates are eligible for TIP, its take-up rate is only 14 percent, limiting its impact on college affordability. Our results show that TIP take-up is higher for students who have early and consistent Medicaid enrollment, and for students in schools with a high proportion of TIP-eligible students. We identify key barriers that play a significant role in shaping take-up: the presence of administrative burdens, and the constraints faced by front-line administrators in alleviating these burdens when administrative responsibility for grant aid is fractured and ill-defined. Other safety net programs could experience similar challenges.

• EPI Policy Brief: Michigan’s Tuition Incentive Program: An Initial Look at Take-up (with Kathy Michelmore, Nathan Sotherland, Kevin Stange, and Marissa Thompson). 2025.