In Progress

Working papers: 

Who Avoids Health Information? Experimental Evidence on Health Insurance Choice

Does information avoidance, a phenomenon where individuals actively choose to not view costless information that may be possibly distressing or cognitively taxing, help explain why individuals select less generous health insurance plans? Using an online survey experiment to identify individuals’ willingness to pay to avoid health consequence information, I find avoidance is relatively common: 24% of individuals in my sample are willing to pay to avoid health information. Avoiders are older, have lower income, are less likely to perform calculations when selecting plans, and have lower initial knowledge about definitions of insurance terms than non-avoiders. Presented with a hypothetical health issue, avoiders require a 2-3 pp higher likelihood of an adverse health event before selecting a more generous health plan. After randomly assigning avoiders to view facts and images highlighting the illness and mortality consequences of experiencing health issues, I find no measurable treatment effect on respondent preferences for health plans. The paper sheds new light on the fact that avoiders of health information are a distinct group with identifiable traits and finds that, independent of exposure to treatment information, they systematically select less generous health plans.

Pre-Analysis Plan for Experiment

Effects of Expanded School Choice on Racial and Socioeconomic Sorting (with Hayley Abourezk-Pinkstone )

School choice policies can provide additional educational opportunities to students that would be otherwise constrained to their neighborhood school, but the effects of such policies depend on the spread of take-up. If take-up rates vary systematically across students by race or socioeconomic status because of persisting barriers to access, then school choice policies may change the distribution of students across schools and the level of racial or socioeconomic segregation across schools. Using a difference-in-difference approach and an event study, we empirically examine how the California Open Enrollment Act (2010-2016), which increased public school choice for students attending low-achieving public schools, impacted student enrollment patterns across schools by race and socioeconomic status. Total enrollment at treated schools falls by up to 1.4% as a result of the policy, and this effect persists for several years. Further, we find that Hispanic student enrollment and free-and-reduced price meal (FRPM) eligible student enrollment at treated schools fall by up to 5.9% and 6.7% relative to comparison schools, respectively, in a school's first year of treatment, and that these enrollment effects each tend to grow over time. We discuss how our context relates to recent literature that calls into question the validity of TWFE DiD and event study designs when treatment effects are heterogeneous across treatment cohorts or time. Our results are robust to an exercise aimed at eliminating this potential source of bias. Our findings suggest that the Open Enrollment Act did expand public schooling options for low-income students and minority students attending treated schools, enabling them to switch to higher-performing public schools. 

Private Health Insurance Patterns Following Spousal Health Shocks

Using an event study, this paper explores how an individual's health insurance coverage changes after their spouse is hospitalized. I examine longitudinal data spanning 1992 to 2018 from the Health and Retirement Study, which includes insurance status and hospital admissions for individuals and their spouses. For the sample of individuals who did not hold a private health plan in the first survey wave, those whose spouses are hospitalized are 7.4pp more likely to hold a private health plan in the survey wave following the spousal health shock. Furthermore, for all individuals who either newly signed up for private insurance or who already held private insurance, the average overall total annual private insurance plan premium increases by $82 as a result of the spousal hospitalization event, representing a 7.3% increase over their average private plan annual premium. Increased plan coverage persists for multiple years following the spousal hospitalization event. I further identify large initial and persistent take-up in private plan coverage for white respondents, individuals with higher initial asset holdings, and individuals whose spouses held insurance at the time of their own hospitalization. In contrast, I observe a relatively small, short-term increased probability in private plan coverage for those with lower asset holdings, no detectable increase for minorities, and a delayed response for individuals whose spouse did not hold a plan at the time of their hospital admission. These findings suggest that there are more complicated and widespread impacts of hospital admissions than than the existing literature, which focuses on additional financial burden on hospitalized individuals, may suggest. These findings may motivate future research to explore behavioral selection into health insurance in response to particular forms of salient information.