Date of Award

Spring 4-2018

Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Tepper School of Business


Rebecca Lessem

Second Advisor

Yaroslav Kryukov

Third Advisor

Maryam Saeedi


Chapter 1: Health, Health Insurance and Employment: A Dynamic Model This paper studies the complex feedback loop of health status, health insurance and employment status. Using a discrete choice dynamic programming model where working-age men make employment decisions each period, I model the effect of a health shock on an individual's choice to work full-time, part-time or not work. If he chooses to work, he may not find a job and instead be unemployed. His labor force status, in turn, impacts his probability of health insurance coverage, as well as his subsequent health status. Data from the Panel Study of Income Dynamics from 1999-2015 are used to estimate the model. Counterfactual analyses show that expanding health insurance to all individuals slightly improves overall health but decreases labor force participation. Insurance affects employment decisions through health and expected out-of-pocket medical expenditures. Healthy men obtain more utility from working than not working and have higher wages than unhealthy men. Although there are more healthy individuals with higher wages, this is not enough to o set the negative effect from the increase in medical expenditures. This increase in medical expenditures only occurs for the unhealthy. Thus, the decline in labor force participation is driven by unhealthy men exiting full-time jobs, which provide insurance at higher rates than part-time jobs. This supports the theory that some individuals, particularly unhealthy ones, choose to work to obtain health insurance.

Chapter 2: Consumer Plan Choice in the Florida Medicaid Market In this paper, I model an individual's health insurance plan decision in the Medicaid market in Florida. Since 2014, most Florida Medicaid recipients have been required to enroll in a Managed Medical Assistance plan that is offered at no cost to the enrollee. All plans must o er the same core benefits but have varying extended benefits and physician networks. Using monthly plan and county specific enrollment data from the Florida Agency on Health Care Administration, I determine what Medicaid enrollees value most in a health insurance plan. I adapt a Berry, Levinsohn and Pakes (BLP 1995) random coefficients discrete choice demand model to a setting without prices and without an outside good. Using a variety of characteristics for plan size, I find that most people prefer to enroll in larger, well-known plans, and that black and Hispanic individuals most prefer large physician networks. In addition, enrollees seem to prefer plans with higher Medicaid report card ratings of ability to keep adults healthy. This characteristic, likely a proxy for plan reputation, has the largest effect in highly impoverished areas where there is a larger network of Medicaid recipients, but a negative effect in areas with lower population densities.

Chapter 3: Close Contests and Future Voter Turnout1 Voter turnout is persistent across election cycles, but understanding what factors can change persistent voting behavior and for whom remains unexplored. This paper asks how close state Electoral College contests for presidential elections influence future voting behavior. Specifically, do these close contests deferentially affect those who supported losers, winners, or those who did not vote in the previous presidential election. It is likely that after the election, voters update their beliefs on the degree to which they influence on the state and therefore national election. The beliefs depend not only on whether or not one's preferred candidate won the state Electoral College votes but also on the margin by which the candidate won or lost. Using data from the American National Election Studies, we use within state variation in voter turnout and Electoral College closeness from 1948-2012 to analyze individual voting behavior across election cycles. Our findings suggest that those who report not voting in the previous election are 3.5 percentage points more likely to vote than those who did not vote in states where the contest for electoral votes was not close. The results further show that females and low- income individuals who voted for non-victors in states with slim margins of victory were less likely to participate in the subsequent election than those who chose a victor in the previous close contest.