Research

Abstract: This paper examines the impact of skill alignment between college major skills and occupation skills. To measure skill alignment between skills acquired in college by major and skills demanded by occupations, we construct a novel skill alignment measures using Lightcast data (formerly known as Burning Glass Technologies). We use this measure in combination with restricted-access U.S. Census Bureau data to show that skill mismatch converges to zero as individuals gain experience. Next, we examine how our skill alignment measure relates to wages over the early career life cycle, focusing on how disparities in the alignment of ten general skills can affect wage outcomes. This suggests that the sorting process based on skill alignment may be more heavily influenced by certain skills than others. Through a segmented search model, we analyze how skill mismatch influences both the likelihood of job-to-job transitions and skill-alignment patterns of college graduates. Our results contribute to the ongoing debate on how to improve the alignment between higher education and labor market demands, suggesting that search frictions have a minimal role in causing skill misalignment between college educated workers and occupation.

Presented at: 2022 UW-Madison Summer Fellowship Presentation, SEA 2023, SOLE 2024, W.E. Upjohn Institute for Employment Research, University of Wisconsin Joseph L. Krislov Labor Seminar, University of Wisconsin- Whitewater, APPAM 2024, SEA 2024, 2025 AEA/ASSA

Abstract: More states, cities, and municipalities have passed "Ban the Box"-type laws to reduce recidivism and increase employment, but surprisingly little is known about how these laws affect justice-involved individuals and their contact with the criminal justice system. A natural question about the laws is whether the employment channel is the right way to decrease recidivism, especially for individuals with misdemeanor histories, who comprise most of justice contact. In this paper, I analyze how the Fair Chance Hiring Ordinance passed in 2016 in Austin, Texas affects recidivism rates for both those with felony histories and those with misdemeanor histories. Using Texas administrative criminal records and a difference-in-differences estimator, I find an insignificant reduction in recidivism rates for felony arrestees, but a statistically and economically significant reduction for those with misdemeanor histories. Notably, I find larger and more persistent reductions in recidivism rates for Black justice-involved individuals compared to white justice-involved individuals. For robustness, I estimate both a synthetic control difference-in-difference as well as a Cox proportional hazard model.
Presented at: APPAM 2023, SEA 2023

Abstract: We study the relationship between parents' contact with the criminal justice system and children's academic performance. We link administrative data describing parents' criminal justice involvement to K-12 student-level administrative public school data. We find that parents' involvement with the criminal justice system is associated with worse academic performance for their children across several measures. Children of parents who have any criminal charge have lower standardized test scores, lower math and ELA proficiency rates, lower attendance rates, are more likely to experience discipline events in school, are more likely to drop out of high school, and are less likely to attempt or pass an Advanced Placement exam. We find no evidence that controlling for parents' criminal justice involvement shrinks the Black-white standardized test score gap. Furthermore, we find evidence suggesting that white students are more negatively influenced by the criminal justice contact of their parents. The magnitudes of our estimates are substantial, indicating that our results could have implications for intergenerational mobility in the United States. 

Abstract: This paper examines how spouses influence participation in active labor market programs (ALMPs), particularly focusing on the role of marital status. Building on the well-established literature on the added worker effect, I explore how household composition may influence decisions to participate in training programs and contribute to the selection process. By linking administrative unemployment insurance records with survey data, I test comparative static predictions based on both existing literature and my framework within the German context. The results suggest that married individuals are more likely to take up any training than those single or cohabitating. These findings indicates that relationship status may play a key role in ALMP participation.
Presented at: 2021UW-Madison Summer Fellowship Presentation 

Abstract: Given selection into IDR plans, we use a novel instrument, which leverages the randomly assigned student loan servicer to instrument the likelihood a borrower is on income-driven repayment. Using a subsample of the 2% sample of all individuals with credit in the US from the University of California Consumer Credit Panel (UC-CCP), we have estimated the impact of IDR on credit scores, loan balances, having a mortgage, having a small business loan, entering and staying in a licensed career field, and future student loan take-up for borrowers 1-5 years after the end of the first deferment spell. We find that while current IDR enrollment may have initial positive effects on the student loan, individuals, who were more likely to be enrolled in IDR, are more likely to be always current on their student loans.  We find null effects on credit scores. We see increasing student loan balances, partially attributed to an increased likelihood of an individual returning to school and taking out a future student loan. We additionally show IDR decreases in wealth-generating lines of credit such as mortgages and small business loans. Lastly, we are the first to examine how IDR changes the likelihood of an individual working in a licensed profession such as teaching and health care and find positive but insignificant initial point estimates. In later years, IDR is negatively associated with being licensed in a healthcare profession, while the effect remains null for teachers.
Presented at: SEA 2024, 2025 ASSA, MEA 2025, SOLE 2025*
*scheduled

Abstract: This paper studies the job search process of college graduates, focusing on how risk aversion influences search behavior. Using the National Educational Panel Study (NEPS) Starting Cohort 5 (SC5) data, I show that search behavior is indeed correlated with degree of risk aversion; more risk averse individuals send more applications. To capture this behavior and gain insights into the search dynamics as graduation approaches, I develop a search model tailored to college graduates navigating their first job out of college. The model accounts for search effort (number of applications sent) and the role of risk aversion in search intensity leading up to graduation, while accounting for the preference for securing a job by graduation. Moments are matched using simulated method of moments (SMM). The model shows that individuals do not search for a long period of time. In a counterfactual analysis, I show that reducing job search costs leads to an increase in the number of applications sent and job matches with higher wages. This suggests that universities might improve job matching efficiency by subsidizing job search costs, thereby reducing the burden on students.

Presented at: 2021 University of Wisconsin Joseph L. Krislov Labor Seminar