Research
Publications
- Joelle Fong and Yongjoon Youn (2023). Assessing patterns and stability of ADL hierarchical scales for functional disability assessment. In The Gerontologist (Oxford University Press).
Background and Objectives
This study examined the stability over time of activities of daily living (ADL) items in 3 comparable longitudinal data sets and evaluated ADL loss sequences for older adults in the United States, South Korea, and Japan.
Research Design and Methods
Data from the U.S. Health and Retirement Study, and its 2 international sister surveys, were analyzed. Participants were community-dwelling adults aged 60 and older. For each data set, Rasch analysis was implemented to determine if the ordering of items remained stable across multiple waves (2006–2014), such that a single ADL hierarchy may be derived from multiwave data.
Results
Data fitted the Rasch model well. Item calibrations were sufficiently stable across measurement periods in each data set, reflecting a stable frame of reference. Results were also robust to sample variations. The derived ADL hierarchies based on scaled logit scores revealed that “dressing” and “bathing” were relatively more difficult items for older adults in all study populations.
Discussion and Implications
Scale stability is essential when exploiting longitudinal data to analyze patterns in ADL disabilities. The consistency in ADL scales across measurement periods supports their use as screening tools and identifying those at risk for transitions in care. Interventions to reduce dependency in bathing and dressing can help improve independent functioning for community-dwelling older adults.
Working Papers
- “Unlocked or Locked? Labor Market Responses to Health Insurance Decoupling”
Approved U.S. Census Bureau FSRDC Project No. 3447
Employer-sponsored health insurance (ESI) creates a non-portable component of compensation that can affect worker mobility, firm wage setting, and the allocation of labor across employers. I study these effects using restricted Census data linked to detailed employer insurance costs and plan characteristics. I estimate how greater insurance portability changes workers’ career paths, including whether, when, and where they move and how their earnings evolve after separation. On the employer side, I examine how firms adjust compensation and employment when the cost and outside value of ESI change, and how these adjustments are distributed across workers within firms. I then use matched worker-firm histories to distinguish worker composition from firm pay premia and study how insurance portability changes sorting across employers. Finally, I use the empirical evidence from the worker and employer analyses to discipline a structural framework that quantifies how insurance portability reshapes workers’ outside options and firms’ wage-setting when switching employers no longer carries an insurance-related cost.
- “From Continuity to Fragmentation: Evidence from Physicians’ Premature Deaths”
This paper leverages the sudden and premature deaths of U.S. primary care physicians to study the value of care continuity and the relationship with the doctor for patients. The sudden and premature death of a PCP mitigates concerns regarding anticipation by patients in variations used in the literature, such as retirement and reallocation. Utilizing administrative Medicare claims from mid-2000 to 2019, linked to over 3000 PCPs who passed away prematurely, this paper identifies patients who lost their PCPs due to the sudden and premature deaths and tracks their health care utilization and health outcomes thereafter. Preliminary results using an event study framework with a matched sample reveal several patterns: (1) healthcare utilization of patients of the deceased physicians increased relatively for several quarters after the physician’s death: They had more primary care visits, more referral from primary care physicians, more specialist visits, and more lab tests and images, which collectively results in a relatively 10% increase in the patients’ total healthcare spending. (2) the patients’ healthcare became relatively more fragmented: they are more likely to simultaneously utilize several primary care physicians (measured by either the number of PCP seen per quarter, or the HHI index of primary care), and received more duplicated lab tests and images as a result. (3) While receiving relatively more healthcare resources, patients’ health outcomes became worse. They are about 0.05 pp more likely to die, which is about a 9% increase in annual mortality rate (Medicare population annual mortality rate is about 5%). They are also more likely to visit the ER and be hospitalized through the ER. This paper plans to dig into the mechanism of these patterns, which may have implications for the literature on continuity of care, fragmentation of care, and quality of care.
Work in Progress
“The Circular Flow of AI Talent Production and Exchange (with Sujin Lee, Tian Lou, Aiday Sikhova, and Bruce Weinberg)”
“AI-Related Education Is Increasing and Widespread but Unequal (with Sujin Lee, Tian Lou, Aiday Sikhova, Bruce Weinberg, and Saleh Zakerinia)”
We develop a comprehensive approach for measuring AI-related education and examine how higher education prepares the workforce for an AI-driven economy. Using a taxonomy of six dimensions of AI-related skills (Core, Foundations, Hardware, Infrastructure, Applications, and Governance) and large language models, we classify all degree and certificate programs in the Integrated Postsecondary Education Data System (IPEDS) as AI-related or not, and use these classifications to document national patterns in AI-related credentials. Using data for Ohio’s public postsecondary institutions, we apply our classification to university courses based on descriptions in course catalogs, link these data to student-level enrollment records, and summarize patterns in students’ AI-related courses. Finally, using UMETRICS grant data at Ohio State University, we identify instructors engaged in funded AI research and examine which students take courses taught by these instructors. By measuring AI-related education across credentials, coursework, and exposure to AI research, we capture a broader range of AI-related skills developed through higher education than do approaches focused on individuals engaged in developing frontier AI technologies. Our approach provides a broad understanding of the extent to which higher education develops AI-related human capital and how its development varies across fields of study, types of institutions, and student populations. Our framework also lays the groundwork for future research on how AI-related skills shape workers’ career trajectories, firm performance, and the broader labor market.
“The Demand Side Selection in Public Insurance”
“Killing Me Softly With Dinner: The Costs of Circadian Misalignment”
“Intergenerational Dilemma: Balancing Windfalls and Burdens of Public Transfers”
Over the past few decades, the global demographic landscape has rapidly transformed with the onset of an unprecedented era of population aging. Among others, South Korea has emerged as the world’s fastest-aging country, exhibiting the lowest total fertility rates and the most rapid growth in the elderly population (Bloom et al., 2023). Despite this demographic shift, the nationwide development of old-age income security measures was introduced relatively late, in the early 2000s, in response to an elderly poverty rate exceeding 40% (Chung, 2024). To provide financial security to the ever-increasing elderly population, the Korean government introduced the Basic Pension (BP) system in 2008. This non-contributory pension unconditionally transfers cash to 70% of the population aged 65 or older, providing windfalls for the elderly, while funding is sourced from shrinking younger generations.
The adult children of financially insecure elderly parents may face particularly constrained economic decisions across various life-cycle stages, as they may have to bear the dual responsibility of increasing BP funding burdens and providing financial support to their financially insecure parents. If BP benefits completely crowd out financial support from adult children, the burden on the children will be eased, but the elderly parents will have no windfall gains (Becker, 1974). However, if the benefits only partially crowd out financial support from adult children, the parents will experience windfall gains while the adult children remain burdened with continuing to provide both public and familial financial support for their parents (Cox et al., 2004).
How should we achieve an intergenerational balance between the windfalls that benefit the elderly and the burdens that affect their children? While existing studies mainly focus on potential crowding-out effects for policy evaluations (Nikolov and Bonci, 2020), the literature is silent on “how recipients utilize the benefits they receive.” For example, if windfall gains of elderly parents are shared with their children, both parents’ and children’s net gains may vary. In this paper, I present new evidence on the importance of the impact of recipients’ shared benefits, which I term the spillover effect, alongside the crowding-out effect to understand net gains from public and family transfers in Korea, crucial for balancing windfalls and burdens in intergenerational public and household finances.
I find evidence that BP benefits displaced the transfer burdens of adult children for elderly single households and resulted in net gains for both family members and recipients in elderly couple households living without children. However, elderly couples living with their children experienced net gains, with the wealth of their children being redistributed from non-coresident to coresident children. The findings suggest that considering how recipients utilized their benefits can alter policy evaluations, contrasting with current literature.
Technical Reports
Tian Lou, Bruce Weinberg, Harrison Yongjoon Youn, and Julia Lane (2026). Characterizing AI “Headlight” Employers. In Industries of Ideas Technical Documentation.
Tian Lou, Audrey Yu, Harrison Yongjoon Youn, Bennett Otten, and Richard Hendra (2026). Tracking Earnings and Employment. In Industries of Ideas Technical Documentation.
Audrey Yu, Bennett Otten, Richard Hendra, Tian Lou, Harrison Yongjoon Youn, Bruce Weinberg, and Julia Lane (2026). Measures of Workforce Dynamics. In Industries of Ideas Technical Documentation.