[Introduction ...] The results indicate that college dropouts aged between 25 and 65 do, in expectation, not experience significant losses in terms of hourly wages. Further-more, in terms of expectations, college dropouts end up in occupations with higher occupational prestige scores relative to individuals with a college en-trance qualification but no college experience. There seem to be no significant differences in employment status between the two groups. A further descriptive analysis shows that college dropouts are more likely to end up in smaller firms. The rest of the paper is structured as follows. Section 2 reviews the relevant literature on college dropout. Section 3 describes the data set used for the anal-ysis. Section 4 discusses the integration of machine learning techniques into the causal inference framework. Section 5 introduces the method used for treat-ment effect estimation. In Section 6 treatment effect estimation results are pre-sented for hourly wages and occupational prestige scores, and a multinomial logit model is estimated to investigate the relationship between employment and treatment group status. Section 7 concludes by critically discussing the empirical estimation strategy.
Die Studie untersucht ausgewählte Arbeitsmarktergebnisse von Studierenden, die die Universität ohne Abschluss verlassen haben. Datengrundlage sind Stichproben aus dem SOEP. Es werden ökonometrische Schätzverfahren mit Hilfe des maschinelles Lernen verwendet.