Abstract:
We evaluate three labor market policies (vocational training, on the job training and firm-worker matching) in a representative sample of urban labor markets in Uganda. The evaluation is based on a four year panel of workers and firms and exploits a randomized control trial and structural model. The reduced form impacts comparing on-the-job and vocational training show that: (i) both forms of training have significant impacts on the extensive margin of finding wage employment (by around 25% relative to the control group); (ii) both treatments have significant impacts on hours workers, hourly wages and total earnings (by at least 30%). Bounds estimates of the productivity impacts suggest both forms of training have positive productivity impacts, with vocational training having the larger impact on worker productivity. We verify the impacts on hourly earnings and productivity bounds by implementing a practical skills test to workers in all treatment groups: in line with the earlier evidence, this shows vocational training to significantly raise practical skills, both relative to the control group and relative to the workers that were assigned to on-the-job training. We find little evidence that there are informational frictions in these urban labor markets relating to firms not having information on workers willing to work, or on workers willing to work and being trained. The final part of analysis develops and estimates a structural model of worker job search, where workers make two endogenous choices: (i) search effort; (ii) their reservation wage. Using monthly data on labor market histories for workers in our evaluation, we use the experiment to identify the structural model and shed light on how training impacts these two endogenous outcomes through two mechanisms: (i) worker beliefs over the arrival of job offers; (ii) the distribution of offered wages. We use these structural estimates to estimate the lifetime benefits of these various training and matching routes into the labor market.