PH.D. THESIS

Health-Driven Productivity and Social Security Disability Insurance 

(Job market paper)

This paper explores how an individual's Social Security Disability Insurance (DI) participation decision is related to their health, productivity, and labor market outcomes. Using the objective health indexes available in the Panel Study of Income Dynamics (PSID), I establish that better health conditions are associated with increased wage growth and a reduced probability of future DI participation. To analyze the dynamics underlying these links both theoretically and quantitatively, I develop an equilibrium search model with stochastic health shocks. The novelty of the analysis is incorporating health-driven productivity and endogenous DI participation into the standard Diamond-Mortensen-Pissarides framework. I show that stochastic health-driven productivity in the presence of endogenous job separation plays a pivotal role in accounting for key features of the data on DI participation, as well as the dynamic relationship between health and wages at the individual level. The results emphasize not only the overall health level of the labor force but also the persistence and volatility of the individual-level health conditions are important for understanding the extent of the DI participation. Using the model, I evaluate the current DI program in the U.S. while focusing on policy instruments, such as the length of the waiting periods for the initial application and appeals processes (link).  

Social Security Disability Insurance Participation in an Equilibrium Search Model

The number of Social Security Disability Insurance beneficiaries has been substantially increasing in the U.S. for the last three decades. The literature found that most of this substantial growth is attributed to non-health factors using reduced-form models. In this paper, I construct an equilibrium search model with endogenous participation of the DI program. For a comparison purpose, I also construct an equilibrium search model with exogenous participation. The numerical model with endogenous participation replicates features in the U.S. data better than the one with exogenous participation.  For example, in the model economy, the future beneficiaries have approximately 17.2% lower wages than the future non-beneficiaries when both are employed. Furthermore, the numerical model finds a negative relationship between DI participation and skill accumulation, and a positive relationship between DI participation and skill deterioration. Importantly, DI participation is inversely related to skills levels. Specifically, a 10.0% increase in skill levels would result in a 30.2% decrease in DI participation. Overall, this paper provides insights into employment, unemployment, and DI participation flows when shedding light on the relationship between DI participation and labor market outcomes. 

On the Interplay between Social Security Disability Insurance Participation 

and Labor Market Outcomes

Over the last three decades, the number of Social Security Disability Insurance beneficiaries in the U.S. has almost tripled, while its labor force has increased by less than thirty percent. This is highly surprising because the overall health of the labor force has consistently improved in the U.S.  Motivated by this large gap between the growth rates of the labor force and the number of DI beneficiaries, numerous studies explored the impact of non-health factors. However, the literature has mainly focused on aggregate level data. In this paper, we offer first microdata-based evidence on the impact of certain non-health factors on DI participation. Specifically, using the Annual Socio-Economic Supplement of the Current Population Survey (CPS-ASEC), we evaluate the impact of the replacement rate of DI benefits, unemployment benefits, and unemployment duration. The results show that the replacement rates of DI benefits and unemployment duration have a strong positive impact on DI participation at the individual level, although the strength of the effects differs by age and education.  Using our estimates, we conduct a counterfactual experiment to evaluate the overall impact of the replacement rate of DI. The results show that had the replacement rate of DI benefits remained constant from 2000 to 2020, DI participation in 2020 would have been nearly 20% lower than the actual figure (link).