Research

Working Papers

• Self-Selection in Randomized Controlled Trials and Screening Decisions: Evidence from the Screening Trial for Lung Cancer (Job Market Paper, draft coming soon)

  Lung cancer is the leading cause of cancer-related deaths worldwide. To combat this, the U.S. Preventive Services Task Force recommends annual screening for high-risk individuals, based on findings from the National Lung Screening Trial (NLST). However, self-selection may reduce the effectiveness of screening in lowering mortality rates. This paper develops a dynamic discrete-choice model that incorporates both NLST participation and lung cancer screening decisions to analyze the factors influencing screening behavior and how self-selection affects health outcomes and costs. Using data from NLST and the National Health Interview Survey (NHIS 2015), the model examines beliefs about lung cancer risk, survival rates, and the costs and benefits of screening, including uncertainties caused by false positives and false negatives. The findings show that individuals with lower lung cancer risk are more likely to screen, and while trial participants who undergo screening have lower mortality rates, those in the general population who screen have higher mortality rates. Counterfactual analysis reveals that lung cancer survival benefits are limited, making 100% screening uptake unlikely. However, targeting underrepresented groups can reduce mortality at minimal cost, and annual lung cancer screening remains the most effective strategy for saving lives.

• Spousal Effects: Will You Quit Smoking Because of Me? with Junyu Zhang (Draft upon request)

  Analyzing Panel Study of Income Dynamics (PSID) data spanning from 1999 to 2019, we use a simultaneous equation model with censored dependent variables, to estimate the spousal effects on smoking behavior as well as the effects of spousal health conditions. By accounting for previous smoking behaviors, state-level cigarette taxes, and unobserved heterogeneity, the model allows us to address simultaneity, homophily, and confounding issues without concerning multiple equilibria. We also consider mental illness records, pregnancy, and health-related occupations, on which smoking awareness and attitude depend heavily. We account for the hidden states of lung disease and heart disease variables, which include: the husband and wife’s respective family history of smoking and their family economic status, state-level COPD prevalence, alcohol use, and past diabetes. Our result highlights a strong and positive spousal effect on smoking behaviors. Specifically, husbands tend to smoke less if wives are associated with a higher likelihood of lung disease, while the corresponding effect is the opposite in wives' equation. Mental illness history increases the propensity and intensity to smoke for both husband and wife, with wives being statistically significant. A wife is less likely to smoke during her pregnancy, an effect that is not significant for the husband.

• Impacts of School Violence on Mental Health and Academic Outcomes for Immigrant and Native-Born Children with Junyu Zhang (Draft upon request)

  This paper investigates how school violence affects academic outcomes through the impacts on mental health for native-born American and immigrant children. We highlight that preventing school violence can improve mental health and academic performance, thus enhancing economic prospects and promoting generational social mobility. We use a simultaneous equations model to examine the structural relationships of bullying involvement, mental health, and academic outcomes. Analyzing the 2022 National Survey of Children’s Health (NSCH2022), our findings reveal that immigrant children are generally less involved as bullies or victims than their native-born children. Yet, children in white immigrant or low-income families are more susceptible to school violence. Controlling factors like parental mental health, attitudes towards children, and life experiences, we find that immigrant children are mentally healthier compared to native-born children, except for white immigrants who experience more significant mental health challenges. Being bullied significantly increases the likelihood of experiencing mental health issues, while being a bully has a converse impact. This effect has no significant difference between immigrant and native-born children. While immigrant children perform better than native-born children academically, those from white immigrant families show worse academic performance. Mental health improves school performance, but this link is weaker in immigrant children.

• Prevalence of Mental Illness and Supply of Medicaid-Funded Services in New York State with Xin Lu and Steven Stern (Draft upon request)

  Both the demand and supply of public mental health services vary within states. It is necessary to measure demand locally to plan for necessary service provision in an efficient way. In this paper, we estimate the prevalence of mental illness within 39 geographical regions in New York State. Our methodology uses the analysis in Johnson et al. (2017). We find that, for the most part, the supply of public mental health services meets the demand for those services across counties in New York.