For an overview, please see my research statement
Temporary employment is widespread and persistent in advanced economies. The data show that: (i) the less-educated are over-represented in temporary jobs; (ii) although training accelerates the transition to permanent employment, training incidence is relatively low among temporary workers. To understand these facts, I build a search model with endogenous contract types and training decisions. In the model, firms hire the low-skilled as temporary workers to avoid the firing costs imposed on permanent contracts, and employers provide little training to temporary workers because the investment horizon is truncated. Consequently, the model endogenously reproduces the persistence of temporary employment, known as the "temporary job trap." I estimate the model on Korean labor market data. The results suggest that current policy depresses output 9.4% below its potential, with 43% of that gap attributable to the lack of training for temporary workers. Counterfactual policy analysis indicates that training subsidies can eliminate nearly 80% of the output loss.
Why does temporary employment persist? Two possible answers exist: (i) previous temporary employment experience may have a causal effect on current employment status (state dependence); (ii) temporary workers may have higher propensities for temporary jobs (persistent latent heterogeneity). This paper distinguishes state dependence from persistent unobserved heterogeneity to understand persistence in temporary employment. For this purpose, I develop a nonparametric framework by extending the dynamic potential outcomes model recently proposed by Torgovitsky (2019). The nonparametric analysis based on British data indicates little evidence of state dependence among British temporary workers: at most 31.3% of temporary workers in a given period would have permanent contracts even if they had been permanent workers in the previous period.