A0323
Title: Long-term effects of early adverse labor market conditions: A causal machine learning approach
Authors: Petru Crudu - Ca Foscari University of Venice (Italy) [presenting]
Abstract: The causal effects of completing education are estimated during adverse labour market conditions on labour market, health, and family outcomes measured after more than three decades after concluding education. A novel database is constructed that combines historical administrative regional unemployment rates with detailed SHARE microdata for European cohorts completing education between 1960 and 1990. To estimate the causal effects, the generalized random forest is used, a machine learning estimator specifically designed for causal inference enabling uncovering the heterogeneity and non-linearity of the effects. Results show that a one percentage point increase in the unemployment rate at the time of completing education causes a reduction of 5\% in earnings and 2\% in self-perceived health after more than three decades. Heterogeneity analysis shows a clear educational gradient, university-educated people are able to hedge from early unfortunate events. Further, evidence that the systematic divergence in life course trajectories could be explained by search theory and human capital models is presented. To further validate the causal link, an instrumental variable approach is also employed based on exogenous timing and location of unemployment rates. As an implication, it is possible to argue that policies aiming to increase employment opportunities for less-educated young individuals may have long-term benefits.
