The amendment to the German Trade and Crafts Code in 2004 offers a natural experiment to asses the causal effects of this reform on the probabilities of being self-employed and transition into and out of self-employment, using cross-sections (2002-2006) of German microcensus data. This study applies the difference-in-differences technique in logit models for four occupational groups. Easing the educational entry requirement has fostered self-employment significantly for less qualified craftsmen, almost doubling the entry probability, even as exit rates remained unaffected. Weaker effects occur for other occupational groups. These findings have implications for the design of regulations with educational requirements.