Predicting Long-term Unemployment Risk

About this event

This paper uses rich administrative and survey data from Sweden to study the predictability and determinants of long-term unemployment (LTU) over the period 1992-2016. We use standard machine learning techniques to predict job seekers' LTU risk and find substantial predictable heterogeneity. Compared to a model using standard socio-demographic variables, a comprehensive model that uses data on income, employment, and benefit histories more than doubles the predictive power. The estimated heterogeneity in LTU risk implies that at least two thirds of the observed duration dependence in job finding is driven by dynamic selection. We apply our prediction algorithm over the business cycle and find significant heterogeneity underlying the cyclicality in average LTU risk, while the role of composition effects is limited. We evaluate the implied value of targeting unemployment policies and how this changes over the business cycle.