Patrizia Pulejo

In this paper we present a method for estimating the effect of college education independent of individual ability. We use a matching procedure to compare graduates and high school diploma holders (the latter to be used as counterfactuals) in order to identify both treatment and ability parameters in a two-regime equation model. A sensitivity analysis suggests that our matching results are robust with respect to a possible bias due to unobserved heterogeneity. Using Istat’s Survey on Italian Graduates, we estimate that the average effect of one year of college attendance on labour income is close to 7.5%. Unlike the results reported in recent literature, the effect of education is not higher for better endowed individuals. Keywords: Ability Bias, Matching, Treatment Effects Estimation, Two-Regime Regression Model.