Research

[Abstract] 

Nursing home care accounts for the largest share of Medicare's post-acute spending, yet the health consequences of shortening these stays remain contested. This paper estimates the causal effect of early nursing home discharge on patient health outcomes. To address endogenous discharge timing, I exploit Medicare's skilled nursing facility benefit, which imposes a substantial daily copayment after 20 days of care. I combine difference-in-differences and instrumental variables strategies, leveraging variation in copayment exposure to instrument for early discharge. I find evidence that copayment exposure increases early discharge by 29 percentage points, removing roughly five days of care. Estimates suggest that copayment-induced early discharge increases 30-day hospital readmission by 8.4 percentage points (40 percent above a 20.9 percent baseline) and 30-day mortality by 1.6 percentage points (84 percent above a 1.9 percent baseline). Clinical assessments show that early-discharged patients go home in greater pain and still on active medications. These results demonstrate that cost-sharing applied during ongoing healthcare episodes induces patients to exit care before clinical stabilization.

Presented at: 2026 MSU Grand River Conference (East Lansing, MI); 2026 Yonsei Emerging Researchers Forum (Seoul, South Korea); 2026 Southern Economic Association Conference (Houston, TX; scheduled)

[Abstract] 

The demand for nursing home care is high and rising. This paper examines the impact of nursing home admissions on patients and their spouses using stacked difference-in-differences and event-study designs. For patients, admission increases Medicaid participation, with stronger effects for unmarried individuals. It also increases reliance on informal caregiving, underscoring the role of informal care availability. Among married couples, we find little evidence that spouses adjust their labor supply. However, spouses experience worse mental health, particularly among those whose partners were initially relatively healthy. Overall, these findings show that nursing home admissions generate substantial economic and health spillovers, highlighting the importance of accounting for impacts on both patients and their families when evaluating their consequences.

Presented at: 2025 American Society of Health Economists Conference (Nashville, TN)

[Abstract] 

A large literature examines how losing health insurance affects initial hospital entry, but less is known about how coverage shapes hospital discharge destination. I study this question exploiting a 2011 Medicaid enrollment freeze in Arizona that substantially reduced Medicaid coverage among adults aged 50 to 64 hospitalized for non-deferrable conditions, more than doubling the uninsurance rate. Using administrative inpatient

discharge records in a difference-in-differences framework and an estimator that accommodates state-specific trends and seasonality (Lee and Wooldridge, 2026), I find that discharges to skilled nursing facilities fall by 15 percent and in-hospital death rises by 22 percent compared to the pre-period mean, with no offsetting increase in home health or routine discharge. The in-hospital mortality increase is concentrated among unmarried patients, suggesting the importance of informal insurance through family against coverage loss. These findings show that the consequences of coverage loss extend into the hospital stay itself, and fall hardest on patients without alternative sources of insurance.

Presented at: 2025 Southern Economic Association Conference (Tampa, FL), as "The Role of Medicaid on Patient Discharge Destinations"

[Abstract] 

The U.S. public health insurance system provides fragmented coverage for long-term care. For many low-income older adults under age 65, Medicaid is the primary source of coverage for nursing home and home health services. This paper examines whether the Affordable Care Act (ACA) Medicaid expansion affects the use of different types of care, including formal care (nursing home and home health care) and informal care (family caregiving). I focus on a subgroup of older adults who would not have public insurance absent the ACA: low-income non-disabled adults age 50 to 64. Using a difference-in-differences framework, I find that Medicaid expansion increases nursing home use within two years of implementation and is accompanied by a decline in informal caregiving. Over time, however, home health care use rises persistently and is accompanied by increases in informal caregiving. These contrasting dynamics indicate that institutional formal care substitutes for informal care in the short run, whereas home-based formal care complements informal caregiving in later periods. Effects on formal care are most pronounced among individuals who are unmarried or without children. Overall, the results imply that public insurance expansions generate dynamic adjustments across institutional and home-based formal care, and family caregiving, highlighting the interconnected roles of formal and informal care in response to changes in financial protection.

Presented at: 2024 Association for Public Policy Analysis and Management Conference (National Harbor, MD); 2024 Southern Economic Association Conference (Washington, DC); 2024 National Tax Association Conference (Detroit, MI); 2024 MSU Red Cedar Conference (East Lansing, MI)

[Abstract] 

We introduce lwdid, a Stata command that implements the rolling difference-in-differences (DiD) estimator proposed by Lee and Wooldridge (2026a). The rolling approach transforms a panel-data DiD problem into a sequence of cross-sectional estimation problems, allowing flexible analysis of staggered adoption designs with heterogeneous treatment effects. The command accommodates both large- and small-sample settings. In designs with a limited number of treated or control units, it implements the extension developed in Lee and Wooldridge (2026b), where conventional large-sample asymptotic inference may be unreliable. For large panels, lwdid provides computationally efficient inference via a multiplier bootstrap based on influence functions. By avoiding repeated model estimation, this approach substantially reduces computation time. For settings with few treated or control units, the command supports several inference methods, including exact inference using the t distribution under normality, HC3-based heteroskedasticity-robust inference, and randomization inference. Overall, lwdid makes the rolling DiD framework easily implementable in a wide range of empirical applications.

[Abstract] 

This paper evaluates the effect of in utero exposure to the 2004 Indian Ocean Tsunami on short-term childbirth outcomes in Indonesia. Exploiting variation in the timing of exposure, we find that the probability of a successful pregnancy drops by 5.9 percentage points (pp), while miscarriages increase by 5.5 pp for those exposed in the earliest stage of pregnancy. We find suggestive evidence that post-disaster health investments by households may have shielded later cohorts from harmful effects. Our results suggest the importance of considering fetal loss in developing countries and highlight that facilitating household investment in health through various policies may mitigate negative birth effects in the aftermath of natural disasters.

[Abstract] 

We investigate the effect of education on public job choice using quarter of birth as an instrumental variable. We find that an additional year of education increases the probability of public sector employment by 1.87 percentage points. However, this positive effect is driven by females, whites, and those with high school degrees or less. For those with college or higher degrees, we observe a decrease in public employment with more education. Our results imply that highly-educated individuals go into high-risk, high-return occupations.

[Abstract] 

In Washington State, drivers under 70 years old may renew their licenses online, but those 70 and older must renew in person. Using administrative records covering the universe of Washington driver's license renewals from 2005 to 2023, we measure a significant decrease in the share of drivers renewing their licenses at age 70 compared to individuals in their 60s. We exploit this discontinuity, along with variation in driving distance to the nearest licensing office, to ask whether the in-person renewal requirement screens out riskier drivers or simply those for whom a trip to the office is most costly.