We provide a field experiment to study the efficiency of entrepreneurial program in Gagauzia region, Moldova. This program aims to increase the start-up rate in the region by providing free business and confidence training. Using randomized controlled trial we investigate (1) which type of training increases the willingness of people to engage in start-up activity and (2) what are the underlying behavioral mechanisms of this change. We randomly assign applicants for the entrepreneurship training to three educational tracks: (1) Business and confidence training; (2) Confidence training; (3) No training (control group). We elicit participants' level of confidence before and after the training using the market entry game (Camerer & Lovallo, 1999). We also measure cognitive abilities with cognitive reflection test (Frederick, 2005) and assess risk attitudes (Cettolin & Riedl, 2011) before the training. After the training, we ask about the investment that participants would like to get for their start-ups from third-party and their willingness to take a loan from the bank. These questions are incentivized since the decision about funding their projects and giving the loan is based on the subjects answer and subjects know about it. These are the key dependent variables in our study. We find that participants are under confident on average; however, we observe a positive change in confidence after the confidence training. We find indirect evidence that different educational tracks change investment intentions in different directions. We do not observe that cognitive abilities or risk attitudes are associated with higher intentions of engaging in entrepreneurial activity, taking a larger loan or willingness to invest more. We also plan to provide a follow-up survey in May 2016 to measure the key indicators of entrepreneurial activity: start-up rate, the number of employees, revenue, profit.