Job market paper

Health Screening and Selection: Evidence from Biennial Subsidies in South Korea

Public health screening programs are widely used, but their impact is often limited by low participation among high-risk individuals who stand to benefit most from early diagnosis. I study selection into health screenings and their causal effects using quasi-random variation from South Korea’s National Health Screening Program, which subsidizes 90-100% of screening costs every other year at even-numbered ages. Using nationally representative survey data, I find that subsidy eligibility increases screening completion by 16-19 percentage points (183-295%). Compliers with the subsidies are predominantly from lower socioeconomic backgrounds and are in poorer health than those who always participate regardless of subsidies. Using national health insurance claims data, I find that subsidy eligibility increases both in-situ and invasive cancer diagnoses by 17-19%, as well as treatments for conditions including cancer precursors, hypertension, diabetes, high cholesterol, and osteoporosis. These results demonstrate that subsidies can effectively target high-risk individuals and strengthen the impact of public health screening programs.

Stomach cancer screening rate in South Korea — Fig: Stomach cancer screening rate by age

Working papers

Spousal Spillover in Health Screening: Evidence from National Health Screening Program in South Korea

with Hyuncheol Bryant Kim and Jaehyun Jung

[Manuscript]

Social interactions can strongly shape preventive health behaviors, but little is known about how they shape participation in formal health screening programs. This study investigates spousal spillovers in screening participation using South Korea’s National Health Screening Program, which subsidizes 90–100% of screening costs biennially at even-numbered ages. Exploiting the spouse’s even age as an instrument, we estimate that a spouse’s screening increases one’s own participation by 7.9 percentage points, equivalent to 37% of the direct subsidy effect. Spillovers primarily run from wives to husbands, and evidence points to coordination as a key mechanism, as many couples undergo screening on the same day. These findings highlight the importance of household dynamics in preventive care and suggest that policies engaging families jointly could substantially enhance the effectiveness of public health screening programs.

Health screening rate by spouse's age — Fig: Health screening rate by own age/spouse's age

When Gender Norms Shape the Returns to Education: Evidence from Health Behaviors in Indonesia

with Jaysa Rafi

Education is widely used as a tool for improving health, but its returns depend on cultural and gender norms. This paper examines Indonesia’s primary school construction program and its long-term effects on education and health behaviors. Using geographic and cohort variation in school exposure, we find that the program increased primary school completion for both genders, but downstream outcomes diverged. For women, higher schooling declined, likely due to household budget constraints and son preference, while men were unaffected. Women exposed to the program were less likely to smoke, whereas men’s smoking, associated with social status and masculinity, did not change. Those exposed to the program were more likely to vaccinate their high-parity sons, but not daughters, consistent with stronger son preference for later-born children. These results show that education policies can yield unequal health returns when social norms shape how education translates into behaviors.

Effect of school construction on infant vaccinations — Fig: Effect of primary school construction on infant vaccinations

Misinformation Belief, Health Behavior, and Labor Supply during the COVID-19 Pandemic: Evidence from Tricycle Drivers in Philippines

with Hyuncheol Bryant Kim, Syngjoo Choi, Yasuyuki Sawada, and Takashi Yamano

[Manuscript]

This study aims to analyze the behavioral consequences of people’s beliefs in conspiracy theories and misinformation surrounding COVID-19 vaccines. We employ unique panel data to examine the relationship of belief in misinformation, vaccination behavior, and labor supply of tricycle drivers in the Philippines. We find that individuals with higher risk preference are more likely to hold misinformed beliefs. These beliefs, in turn, are associated with reductions in vaccination and other preventive health behaviors. We also find that beliefs in misinformation delay workplace recovery.