Job Market Paper (Click Here)

Identifying Outlying Groups in Emergency Room Visit Patterns: A Procedural Framework with Bayesian Data Augmentation for Binary Regression (with Jihnhee Yu)

Abstract

Healthcare utilization exhibits significant disparities, making it an important area of study. Among healthcare services, emergency room visits are unique, as emergency room provides care regardless of a patient’s ability to pay or health insurance status, resulting in diverse utilization patterns and increased social costs. Advanced analytical techniques that identify and classify outlying trends across different groups are valuable for uncovering the factors driving disparities in ER utilization and developing targeted strategies to improve healthcare access and outcomes. We introduce a data analytical procedure for detecting outlying groups in binary regression, utilizing Bayesian data augmentation, group residual analysis and outlier insight extraction, with a focus on identifying disparities in emergency room utilization patterns. Across various simulation scenarios, the proposed procedure consistently classifies outlying groups with high accuracy validating its effectiveness.

We apply this method to data from the National Health Interview Survey (2011–2016). Our analysis segments populations based on demographic and socioeconomic factors to identify outlying groups. The results show that outlying groups for both emergency room and primary care physician visits—often substitutes for emergency room services—share common traits in terms of income, education, and race. Low-educated groups, in particular, exhibit relatively unnecessary use of emergency room services, which may be attributed to insufficient health informationseeking behavior compared to other groups. Our findings suggest that addressing the information gap among these populations has the potential to reduce healthcare disparities and lower the social costs associated with unnecessary emergency room visits

Working papers

Impact of Young Adult Health Insurance Eligibility on Parental Labor Force Participation: New evidence from State-dependent mandates

Abstract

In this paper, I delve into the effects of state-dependent mandates on parental labor market participation. While state-dependent mandates have been proven to boost health insurance enrollment rates among young adults, there’s limited evidence about their impact on policyholders—their parents. This paper explores how parental assistance facilitated by these mandates influences parents’ labor force choices.

The analysis reveals that state-dependent health insurance mandates are associated with a modest decline in labor force participation, reducing extensive margin outcomes by approximately 4%. In contrast, the effects on the hours worked and full-time status are statistically significant at the 1% level, with an increase of 1.72 working hours per week and 2.0%. These results suggest that while some parents may reduce participation at the extensive margin, others increase their hours once employed. Notably, single parents exhibit a different pattern. They increased their working hours by approximately 1.6 hours per week, with no significant change in their extensive margin. This indicates that single parents may face stronger incentives to maintain employer-sponsored health insurance coverage for their dependents. The contrasting results imply the presence of altruistic motives in parental decision-making.

Publications

Kwon, H., Yu, J., & Li, M. (2025). Identifying outlying groups through residual analysis and its application to healthcare expenditure. Journal of Applied Statistics, 1-22.

Kwon, H., & Cho, H. C. (2018). The effect of international oil price on LNG price in South Korea and Japan. Geosystem Engineering, 21(6), 297-308.