Modeling Employment Dynamics with State Dependence and Unobserved Heterogeneity

Abstract

This paper extends existing work on labor force participation dynamics by distinguishing between full-time and part-time employment and allowing unobserved heterogeneity in the effects of previous employment outcomes, children and education on employment dynamics. The results reveal significant autocorrelation in unobservables, and significant variation in the effects of children and education on labor supply preferences. Moreover, omission of random coeffcients or autocorrelation can bias significantly estimates of policy effects. On average, policies temporarily incentivizing part-time and full-time employment are equally effective tools for reducing non-employment. However, non-employment among women with young children is more responsive to policies encouraging part-time rather than full-time work.

Keywords: discrete labor supply, unobserved heterogeneity, state dependence, repeated multinomial choice

JEL Classification: C15, C25, J6, J22