Here\u2019s an idea that policy makers like: Take the unemployed, give them entrepreneurship training, and turn them productive, tax-paying business owners. Of course the politicians love the idea. It gives unhappy, out of work people a shot at the American dream and reduces unemployment at the same time. There\u2019s just one problem. It doesn\u2019t work. We\u2019ve had this type of policy in place since 1993, when the federal government created the Self-Employment Assistance (SEA) program to provide small business training and advice to people receiving unemployment insurance. To help turn the jobless into entrepreneurs, participants in the program get entrepreneurship education and advice and receive their unemployment insurance payments without having to look for a job. The gold standard for evaluating whether a government policy works is to run an experiment. If some people are randomly assigned training and assistance and others are not, then observers can see whether the government-provided help has any effect, while ensuring that the group receiving the treatment is no different from the group that didn\u2019t receive it. In the mid-2000s, the Department of Labor designed just such an experiment to see if entrepreneurship assistance and training increases small business ownership and performance. To assess the benefit of the government help, participants in the Growing America through Entrepreneurship (GATE) program were randomly assigned to receive entrepreneurship assessment, training and counseling or to serve in a control group that received none of these. Researchers then observed whether those that got the help had a higher rate of business ownership and small business performance over the following five years. In December of 2009, the Labor Department released the results of this study; and they are instructive. Compared to the control group, the recipients of the entrepreneurship training and assistance: \u2022 Were no more likely to own a business \u2022 Had no lower rate of business closure \u2022 Earned no greater self-employment income \u2022 Had no greater sales \u2022 Had no more employees \u2022 Were no less likely to receive unemployment benefits \u2022 Were no less likely to receive public assistance benefits. In fact, those who received the training and assistance were five percentage points less likely than the control group to have received a business loan, and had invested only invested 70 percent as much in their businesses. And those who received the government\u2019s help were more likely to find getting customers to be a challenge than those who received no assistance. These results followed an earlier experiment in Massachusetts in which the random assignment of entrepreneurship training and assistance did not increase self employment income or the odds of working for oneself five years later. An experimental study in Washington showed that entrepreneurship training and assistance was beneficial over a shorter period of time. And other studies have shown correlations between government help and the odds of being an entrepreneur and entrepreneurial performance. However, to date we have no solid, experimental evidence of the long term benefit of entrepreneurial assessment, training and counseling on the tendency to be an entrepreneur or performance at small business ownership. This result has an important implication: government efforts to help people become entrepreneurs don\u2019t make people more likely to run their own businesses or do a better job at managing them. Would America would benefit more if we gave the resources used to fund these ineffective programs back to small business owners? It's a plausible hypothesis and one worth testing. So here\u2019s what I propose: Give those with a different set of beliefs about how to enhance small business performance a chance to test their ideas. Let\u2019s randomly assign a tax cut to some small business owners and see if the businesses paying less employ more people five years down the road.