This paper utilizes the self-employed to analyze the observed increase in the educational earnings premium in the 1980’s. The paper compares the predictions of the signaling and human capital models in response to an exogenous demand shock such as a skill-biased technological change. Since the self-employed have no incentive to invest in a costly signal to show to employers their productivity, a change in the schooling equilibrium should not affect their earnings. Four testable hypotheses are derived. The findings suggest that the signaling model may indeed predict the observed changes in the schooling premium that are not consistent with the predictions of the human capital model.