"My research focuses on how labor market dynamics are different across ability. In this dissertation, I explore the impact of individual ability on the cyclicality of employment rates over the business cycle and non-employment duration. Then I provide a human capital model with learning-by-doing and heterogeneous learning ability to explain the differences in observed labor market dynamics across ability. In chapter one, I discuss the Armed Forces Qualification Test as a measure of ability. Then, I provide some empirical evidence which shows that high ability workers are more attached to the labor market than low ability workers. Data from the National Longitudinal Survey of Youth 1979 shows that high ability workers have higher employment rates, less pro-cyclical employment rates, and shorter non-employment duration than low ability workers. At the same time, workers with high ability have higher wage levels and wage growth rates than low ability workers. I suggest that a human capital model with learning-by-doing and heterogeneous learning ability can explain high labor supply from high ability workers, because current labor supply increases future human capital which delivers higher labor income in the future. In the second chapter, I empirically document that employment rates of high ability workers are less volatile than those of low ability workers over the business cycle. Less pro-cyclical employment rates for high ability workers remains even after controlling for education and average life-time wage. Moreover, the impact of ability on employment cyclicality decreases over the life cycle. In order to explain my empirical findings, I provide a life-cycle model with human capital accumulation through learning-by-doing. Heterogenous learning ability is introduced into the model to generate different wage profiles across ability. In this model, high ability workers can accumulate more human capital than low ability workers during any given period of employment. Therefore, high ability workers have more incentive to provide their labor than low ability workers because they can have higher return in the next period by accumulating more human capital. In the recession, all workers reduce their labor supply because the aggregate productivity falls. However, high ability workers decreases labor supply much less than low ability workers because working in the current period increases their future human capital. I calibrate the model to match employment and wage profiles over experience. Then, an aggregate productivity shock is introduced into the model to perform the business cycle analysis. The simulated data mimics the observed pattern in the data. In the last chapter, I explore the duration difference across ability groups instead of the incidence of employment. I find that high ability workers experience shorter non-employment duration than low ability worker even after conditioning on education and average life-time wage. I test whether the main reason of short non-employment duration for high ability workers is induced by low propensity to change their previous occupation, industry, or employer. Although workers who do not change occupation, industry, or employer have much shorter non-employment duration than workers who find new jobs in different sectors, the ability effect on non-employment duration is not generated by this sorting mechanism. Lastly, I build a human capital model with learning-by-doing and heterogeneous learning ability. Since high ability workers accumulate more human capital than low ability workers during the same employment period, high ability workers have a lower reservation wage conditional on current human capital stock. Hence, high ability workers are more likely to accept the wage offer than low ability workers after separation. I calibrate the model to match employment and wage profiles and average non-employment duration simultaneously. The calibrated model can qualitatively reproduce short non-employment duration for high ability workers"--Pages v-vii.