Under-training by Employers in Informal Labor Markets: Evidence from Burundi

Workers obtain limited human capital through on-the-job experience in LMICs, but the reasons for this are unclear. We test whether one friction contributes to low worker productivity: firms unwillingness to train because they do not appropriate the returns from training. We study casual labor markets in Burundi, where employers can train workers in a newly introduced agricultural practice in the region, row-planting—a technique that substantially raises yields. In a first field experiment, in some randomly selected local labor markets (villages), we induce 1/3 of employers to train workers in row planting—leading to a 20 percentage point increase in the share of skilled workers in the village. Training generates meaningful economic returns: employers in treated villages increase their adoption of row-planting by 10 percentage points (20%)—raising farm profitability by 9%. However, employers fail to appropriate most of this surplus: 2/3 of the surplus generated is captured by non-training employers, because many of the trained workers work for others following training. In a second experiment, we randomize employers into a condition that increases the likelihood that the worker will return to work for the employer in the future. Employers receiving this guarantee are 50 percentage points more likely to train the worker. This suggests the wedge between private and social returns from investments meaningfully reduces worker productivity.