There is no robust empirical support for the effect of financial incentives on the decision to work in self-employment rather than as a wage earner. In the literature, this is seen as a puzzle. We offer a focus on the opportunity cost, i.e. the wages given up as an employee. Information on income from self-employment is of inferior quality and this is not just a problem for the outside researcher, it is an imminent problem of the individual considering self-employment. We also argue that it is not only the location of an income distribution that matters and that dispersion and (a)symmetry should not be ignored. We predict that higher mean, lower variance and higher skew in the wage distribution in a particular employment segment reduce the inclination to prefer self-employment above employee status. Using a sample of 56,000 recent graduates from a Dutch college or university, grouped in approximately 120 labor market segments, we find significant support for these propositions. The results survive various robustness checks on specifications and assumptions.