I examine how U.S. college presidents and institutions respond to financial incentives arising from the 2017 Tax Cuts and Jobs Act (TCJA), which introduced a 21% excise tax on nonprofit compensation exceeding $1 million. Using an event study design, I find that total presidential compensation decreased by 26%, a reduction driven by declines in non-taxable and other forms of compensation, and their probability of exit increased by 19%. Presidents adapted their behavior in response to this drop in compensation by prioritizing easily observable financial outcomes instead of student expenditure and success, increasing net incomes by 30% and investment returns by 23%, while the most taxed presidents decreased student net financial aid by 0.56% per additional 1% of their tax burden. The impact of the TCJA also varies by president demographics: female presidents' compensation decreases were twice as large as those found for their male counterparts. I also find that the returns to experience are substantial, as presidents coming from the private sector or government underperform their counterparts with higher education experience.

In the United States, licenses are required for entry into many different occupations. Requirements vary by state and occupation, but many licenses require a minimum number of training or instructional hours. We consider the impact of these hours requirements on students and postsecondary institutions, with a particular focus on cosmetology (also known as hairstyling or beauty)—the field that requires the highest number of training hours in the largest number of states. We implement a difference-in-difference design based on state-level changes in licensing hours for cosmetologists between 2011 and 2019. We ask how and whether changes to hours requirements influence student outcomes and institutional behavior. We find that lowering required hours is likely beneficial for students, as it raises completion, lowers tuition, and expands enrollment among some groups of students. Larger institutions appear to reduce their tuition by less than smaller institutions. We find no detectable effects on student debt or the earnings of cosmetologists.

I evaluate the impacts of an emergency financial aid program on short-term academic outcomes in the context of the Covid-19 pandemic. Using administrative data from a large public college system, I use a regression discontinuity model leveraging the assignment rule for additional funding based on expected economic need to evaluate the effects of marginal increases in aid, and an individual fixed effects model to approximate the effects of additional funding with respect to a universal base grant. The regression discontinuity model finds no significant academic impacts of marginal increases in grant aid, while the individual fixed effects model implies that the maximum level of grant aid increases persistence rates by 2.1 percentage points compared to students who received the universal base grant.

Projecting Subjective Wellbeing and Material Hardship using Google Search and High-Frequency Data Sources (joint with Zachary Parolin)

Real-time information on subjective wellbeing and material hardship of a country’s residents is useful for policymakers and the general public, particularly during moments of rapid economic change. After the onset of the COVID-19 pandemic, the U.S. Census Bureau introduced the Census Household Pulse Survey (CHPS) to provide a rare source of (bi)weekly updates on the social and economic wellbeing of families across the U.S. After the survey’s expiration, however, the U.S. will again lack direct real-time data on hardship and wellbeing. This study investigates whether other high-frequency and non-expiring data sources, such as Google Search Trends, Unemployment Insurance claims, geographic mobility, and other sources, can be used to accurately replicate trends in wellbeing observed in the CHPS and to project trends in wellbeing after the expiration of the CHPS.