Contemporaneous healthcare consumption may yield future health benefits that lower longer-run expenditures. In this study I leverage exogenous variation in the expansion of Medicaid under the Affordable Care Act to assess how comprehensive insurance affects immediate healthcare spending for the near-elderly (aged 60-64) and future healthcare spending once they receive Medicare at age 65. Using panel data from the Health and Retirement Study, I provide evidence that Medicaid coverage increases immediate expenditures by roughly 112 percent; but, after receiving Medicare, the same individuals are healthier and consume 77 percent less care measured by total expenditures. I then develop and estimate a life-cycle model of near-elderly individuals that incorporates the Medicaid expansions as well as endogenous health capital and mortality. I find that the Medicaid expansions were valued at slightly above net government costs for the near-elderly population and that modest reforms to increase Medicare's generosity could generate considerable welfare improvements relative to program costs.

Balancing the demands of employment and schooling is a challenging task for an increasing number of students who have to pay their way through college. However, flexible learning and working environments could play an important role in easing many of the frictions associated with performing both activities simultaneously. Using detailed data from gig workers enrolled in online college education, we analyze how labor supply and study efforts respond to changes in labor market conditions and college activities/tasks. Our findings indicate that average weekly college activities reduce weekly working hours by 1.7 hours, representing a "short run" opportunity cost of only $41 per week. We also show that study time is not particularly sensitive to changes in labor market conditions, where a 10% increase in average weekly pay reduces study hours by only 2%. Consistent with these results, we find that workers take advantage of their flexible schedules by changing their usual working hours when college activities are more demanding. Finally, we do not find adverse effects of work hours on academic performance in this context. Overall, the evidence suggests that combining flexible working and learning formats could constitute a suitable path for many (low-SES) students who work to afford an increasingly expensive college education.

• Selected for presentation at NBER Economics of Education Program Meeting (2023).

Utilities increasingly sell electricity using complex menus of time-constant and time-varying price schedules. We study how to design such a menu to maximize social welfare in a second-best environment where the marginal private and external costs of generating electricity vary over time, institutional constraints prevent mandating time-varying pricing, and consumer behavior is distorted by frictions. We develop a model of plan choice, consumption, and intertemporal substitution with time-varying marginal social costs, and estimate it using administrative data from a large utility. We provide evidence of substantial intertemporal substitution in response to time-varying price incentives, and selection across plans based on multidimensional heterogeneity. While the current menu’s time-varying plans substantially shift consumption from high-price to low-price hours, we find that they reduce social welfare. This loss is mitigated by information frictions. We show how to redesign the menu to simultaneously improve outcomes for consumers, the utility, and the environment.

• Selected for presentation at NBER Environment and Energy Economics Program Meeting (2025).

We examine how elders choose insurance plans that either supplement (i.e., Medigap and Medicare Part D) or substitute (i.e., Medicare Advantage) Traditional Medicare (TM) with a focus on potential distortions due to cognitive decline associated with Alzheimer's disease and related dementias (ADRD). First, we use administrative data to provide new descriptive evidence on insurance plan choice patterns among individuals with ADRD. Specifically, we find that beneficiaries with ADRD are (1) much more likely to enroll in TM and stand alone Part D prescription drug plans, but there is considerable heterogeneity by age, sex, race, and health status; (2) are much less likely to switch plans year-over-year, which is driven primarily by older and sicker individuals; and (3) are more likely to choose more expensive plans when, ex-post, cheaper plans were available. Next, we develop a model of nested insurance plan choice where Medicare beneficiaries first choose between Traditional Medicare or Medicare Advantage (MA) and then select specific health and prescription drug plans within TM or MA. We are currently estimating the model using administrative data from the Centers for Medicare and Medicaid Services linked with data from the Medicare Current Beneficiaries Survey.

Student loans play a critical role in increasing accessibility to secondary education for students in the United States; however, borrowers are much more likely to default on student loans than other forms of debt. To address high levels of delinquency and default, income-driven repayment (IDR) provides insurance against earnings risk by lowering monthly payments to a share of discretionary income. In this project we aim to first empirically quantify the extent to which the availability of IDR affects college major decisions, initial career choices, labor supply, and the likelihood of defaulting on student loan debt using several data sets from the National Center for Education Statistics. Initial descriptive analysis suggests that students who enroll in IDR are less likely to enter delinquency but are more likely to choose low paying careers conditional on a rich set of observable characteristics and academic outcomes.  Our research design combines reduced-form analysis with a structural model of repayment plan and career choice that incorporates earnings risk and information frictions about the availability of IDR.