An empirical study of the transition from paid work to self-employment

We explore the relationship between the probability of a transition from paid work to self-employment and three explanatory variables: paid income, predicted income, and income for ability. We use panel data for heads of households from the PSID SRC sample for eight pairs of years. Our results show that therelationship between paid income and self-employment is not linear. We then break up paid income into two components: a)predicted income based on human capital, demographic, and locational variables, and b) income for ability. Again, we find nolinear relationship between self-employment and either predicted income or income for ability. We then test for curvilinear relationships between these three variables (i.e., paid income, predicted income, and income for ability) and the transition to self-employment. We find that individuals with low incomes are more likely to take up self-employment. Further, income for ability is a stronger predictor of the transition to self-employment than predicted income. We show that the relationship between ability and self-employment is U shaped: very low ability and very high ability individuals are more likely to take up self-employment than medium ability individuals. We use prospect theory to explain this result.