The returns to education in self-employment are addressed in four different specifications of the relationship between log income and years of schooling. The specifications range from a standard Mincer equation with a constant percentage increase in income for an additional year of schooling to the most flexible specification with dummy variables for different combinations of years of schooling and fields of study. Based on the more flexible specifications, important non-linearities and heterogeneity in the returns to education in self-employment are found. These results are robust across different estimation methods: OLS, Heckit correction models handling sample selection, and IV dealing with the potential endogeneity of years of schooling. Moreover, the results are robust to the use of different sample years, different definitions of self-employment, and different income measures for the self-employed.
- Returns to education
- Sample selection