We use randomized program offers and multiple follow-up survey waves to examine the effects of entrepreneurship training on a broad set of outcomes. Training increases short- run business ownership and employment, but there is no evidence of broader or longer- run effects. We also test whether training mitigates market frictions by estimating heterogeneous treatment effects. Training does not have strong effects (in either relative or absolute terms) on those most likely to face credit or human capital constraints, or labor market discrimination. Training does have a relatively strong short-run effect on business ownership for those unemployed at baseline, but not at other horizons or for other outcomes.
That is from Dean Karlan's new paper "Behind the GATE Experiment: Evidence on Effects of and Rationales for Subsidized Entrepreneurship Training" January (2012). He concludes:
Although a substantial amount of money is spent on subsidizing entrepreneurship training around the world, we know very little about its effectiveness and whether it alleviates market frictions. We provide new estimates on the average and heterogeneous treatment effects of entrepreneurship training from Project GATE, the largest and broadest entrepreneurship training experiment ever conducted. We find evidence that the training increased average business ownership in the short-run, but that the marginal businesses were unsuccessful and failed to produce tangible or subjective benefits at any of the three follow-up horizons (6-, 18-, and 60- months). We also find no evidence that training shifts the distribution of firms in important ways (e.g., disproportionately creating very successful firms) that might be missed by analysis of average treatment effects. Although we find higher attrition among the control group, bounds analyses confirm that only extreme forms of biased attrition would change these results.