Access Type

Open Access Dissertation

Date of Award

January 2017

Degree Type


Degree Name




First Advisor

Allen C. Goodman


This dissertation examines the relationship between health and labor market outcomes using the 1996-2010 longitudinal RAND Health and Retirement Study (HRS). First, it estimates the relationship between health and labor supply using a simultaneous equation model, treating health as endogenous. The effect of health may be overestimated because people may adjust their non-employment status by their health conditions (justification hypothesis). By using a full information maximum likelihood method, we can conduct a true test of exogeneity on the health variable, taking into account the correlation between two labor supply and health equations’ time-varying error components (unobserved heterogeneity). The results confirm that health is endogenous to labor supply, and has a positive and significant effect on the males’ (0.6833, p<0.01) and females’ (0.6833, p<0.01) labor supply. The reverse effect of labor supply on health is also positive and significant (0.2981, p<0.01, males; 0.0305, p<0.05, females). The finding indicates that it is impossible to determine the direction of bias in the health effect for both males and females. Second, this dissertation examines the impact of health insurance coverage on the labor supply. To address the possible endogeneity of health insurance coverage to labor supply, I estimated the model for a group of married people who have spousal health insurance. The finding indicates that individuals with spouse’s health insurance are more likely to exit the labor market (-0.5527, p<0.01, males; -0.7601, p<0.01, females). Third, this study examines the short- and long-term impacts of cancer on the labor market outcomes for a sample of married people. The effect of cancer is negative and significant for those women cancer survivors who are diagnosed two years or fewer prior to interview (-0.08, p<0.01), and for those males who have survived for five years or more (-0.07, p<0.10). Employed men in the years immidiately following diagnosis, work 2.76 hours (p<0.10) less a week than other employed men. Employed women following three to five years since diagnosis, work 4.70 hours (p<0.05) less per week. Forth, using survival analysis, strong evidence was found that cancer survivors are more likely to experience a longer non-employment spell than a non-cancer group.