Modeling employment dynamics with state dependence and unobserved heterogeneity

We extend existing work on the dynamics of labor force participation by distinguishing between full-time and part-time employment and by allowing unobserved heterogeneity in the effects of previous employment outcomes, children and education on labor supply behavior. In addition, unobserved heterogeneity may feature autocorrelation and correlated random effects. Our results reveal significant variation in the effects of children and education on labor supply behavior. Moreover, the omission of random coefficients and autocorrelation biases estimates of state dependencies. On average, temporary shocks that increase the rate of part-time employment lead subsequently to lower rates of non-employment than do shocks that temporarily increase the rate of full-time work.