Return to education in the Dominican Republic, 2010-2016: Endogeneity and sample selection

Although several investigations on the profitability of education in the Dominican Republic have been carried out, they have not considered the two main difficulties faced by the measurement of educational performance: the econometric problem of the endogeneity of the education variable and bias in the sample selection. The present study aims to demonstrate that the existence of these two difficulties alters the results on the profitability of education in the Dominican Republic. Based on a traditional Mincerian model using OLS, reference coefficients were obtained. Applying instrumental variables (IV), the initial model and the other modalities (extended Mincer model and sheepskin) were re-estimated so that the problem of endogeneity was controlled. Likewise, the selection bias of the sample was addressed using the Heckman method. Empirical evidence is presented showing that the existence of a sampling selection bias and the endogeneity problem of the education variable skew downwards the profitability of education. Our report shows that the returns to education estimated using the conventional OLS model with or without controls, are inferior to the values obtained by the instrumental variables model.