Research 

Before the pandemic, 10% of unemployment insurance (UI) applications were rejected due to the applicant’s failure to meet the earnings requirement. During the pandemic recession, UI relaxed its earnings requirement for the first time under the pandemic unemployment assistance program (PUA). By using an equilibrium search model, I quantify the heterogeneous effects of PUA. I find that PUA increased consumption by 20%, with particularly large benefits at the bottom of the asset and income distribution. I also find that ending the CARES Act at 52 weeks provides the highest increase in welfare per dollar of spending.


Since 2012, the federal government has mandated job search requirements for state unemployment insurance (UI) programs. Following this, many states have increased the number of job search contacts required to collect UI benefits. For example, in 2013, Wisconsin increased its initial job search requirement from two to four jobs minimum. Using synthetic control methods, we estimate the effect of these policies changes on the labor market, and find that increasing job search requirement decreased UI exhaustion rate by about 6 percent and decreased initial claims by 33 percent on average. Then, we propose a two-sided equilibrium search model to quantify such effects on heterogeneous workers’ job search decisions and study the labor market implications of job search requirements. Last, we find the optimal job search requirement.


In the U.S., workers whose past earnings were below a threshold are inel- igible to receive unemployment insurance (UI), which creates a discontin- uous jump in their value of being unemployed. Exploiting this in a regres- sion discontinuity design using administrative panel data, we estimate a sizable local effect from UI eligibility on earnings in the next employer, around $300 or roughly 10% of quarterly earnings. This evidence of a UI treatment effect on re-employment outcomes, however, understates UI’s causal effect and does not distinguish between underlying reasons, either a higher share of production or more productive matches. Withb both a tractable equilibrium and calibrate quantitative model, we interpret the quasi-experimental estimates in the context of endogenous non-compliance and search direction choices. The empirical estimates understate the true causal effect by 4.4% and this high pass-through of UI to earnings implies a low trade-off of between wage and finding rate, essentially very low bar- gaining power.


In this paper we analyze the provision of unemployment insurance in an environment with unobservable employment status and unobservable job offers.  We examine US data characterizing the prevalence of overpayments from concealed earnings fraud and rejection of suitable offers (moral hazard), and calibrate a model to match this data.  There exist novel implications from including fraud from unobservable employment status.  For small increases in the benefit level (10%) the model is consistent with micro evidence on duration elasticities; however, larger increases in the benefit level decrease the unemployment rate and durations.  Similarly, for a range of increases in the potential duration of benefits, the average duration of unemployment decreases.  We calculate that actual occurrences of unemployment insurance fraud amount to 10% of total benefits paid and reduce welfare by around 1%.  We also find the economy is better off relying on minimal welfare payments instead of the current U.S. system of unemployment benefits.


Using data from the Health and Retirement Study, we observe that the incidence of recoveries from mobility limitations is very high between the first and second time that individuals are interviewed, and declines sharply to a stable level in subsequent waves. Based on this observation, we question the initial accuracy of measures of mobility limitations, which could be later ameliorated through individuals’ ability to learn about their health capacities over time. We provide empirical evidence to show that this unusually high incidence of recovery from mobility limitations could be in part the result of respondents’ improvement on health knowledge of questions on mobility limitations or improved experience with the activities instead of caused by real health improvement. Our results are not only relevant to any empirical researchers using mobility limitation indicators in a panel data setting, but also to researchers analyzing the effectiveness of policy interventions on health outcomes using self-reported health measures, since the effects of these policies are likely to be confounded with improvements in health knowledge, and the results be biased to- wards surprisingly large short-run effects and much smaller medium and long-run effects.


This paper outlines a classroom experiment tailored for the first class of undergraduate probability and statistics or econometrics courses. Inspired by the "mark and recapture" method used in ecology to estimate animal populations, our experiment actively engages students from the outset, igniting their interest in the subject matter. Furthermore, it provides instructors with a natural segue to introduce the course syllabus by aligning fundamental concepts — such as probability distribution, estimation, properties of estimators, hypothesis testing, and regression—with specific stages and computations within the experiment. We have implemented this activity in classrooms with approximately forty students, but instructors can adapt it to larger classes with the support of teaching assistants or volunteer students. Additionally, we offer instructors a user-friendly Excel macro that simplifies data entry for the experiment, facilitating seamless sharing of results during class. Five rounds of the activity, including instructions and result discussions, span approximately thirty minutes. We use the remaining class time to introduce the syllabus and establish connections between syllabus modules and the experiment's steps and calculations.

This paper uses panel data from the Health and Retirement Study to estimate the relationship between measures of labor supply flexibility and portfolio-choice decisions by utility-maximizing individuals. Seminal research on portfolio decisions over the lifecycle, and recent research on stochastic dynamic programming models with endogenous labor supply and savings decisions suggest that, other things equal, individuals with more labor supply flexibility are likely to invest more in risky assets, regardless of their age, because of the insurance component that flexible labor supply provides. After controlling for panel sample selection and unobserved heterogeneity I find that labor supply flexibility leads to holding between 12% and 14% more wealth in stocks.


In 2022, there are approximately 1,200 strikes in the United States. Each lasting an average duration of 10.83 days. In response to this phenomenon, numerous states have contemplated implementing alterations to their state unemployment insurance (UI) programs, with the primary objective of extending UI benefits to workers actively participating in strikes. Traditionally, most states, with the exception of New York, have excluded strike workers from eligibility for UI benefits. In this paper, I rely on state-level variations in the stringency of UI provision concerning labor disputes to empirically assess the labor market impact of providing UI benefits to strike workers. Our findings reveal that granting UI benefits to strike workers leads to a 30-50 percent increases in reservation wage and hourly wage of UI recipients. However, it does not yield a significant increase in the frequency of strikes within these states.