Effects of skills training on employment and livelihood outcomes: A randomized controlled trial with young women in Ghana

We use a randomized controlled trial to examine the short- and mid-term impacts of a best-practice training program on (non-)employment outcomes in Ghana. Overall the program did not affect core labor market outcomes at the extensive (employment) and intensive (hours of work, income) margin, but it (i) induced occupational sorting, with treated individuals more likely to work in their field of specialization, (ii) partially improved job quality (written contracts, medical benefits), and (iii) led to better outcomes on a variety of non-labor market indicators (mental health, delayed marriages, access to finance). We also explore policy stakeholders' expectations and perceptions of program success. We find that stakeholders (i) have overly optimistic prior beliefs about the program's impact on core outcomes and (ii) do not update their beliefs as we would expect from Bayes' rule when presented with information about the program's circumscribed effectiveness. We speculate that this result suggests an obstacle for adaptive programming in development cooperation and could help explain the persistence of some suboptimal labor market interventions.