Offering casual labor workers in rural Malawi a higher piece rate increased output but did not attract more productive workers

An observed positive relationship between compensation and productivity cannot distinguish between two channels: (1) an incentive effect and (2) worker selection. In this paper by Guiteras and Jack (latest version: January 2014), the authors use a simplified Becker-DeGroot-Marschak mechanism, which provides random variation in piece rates conditional on revealed reservation rates, to separately identify the two channels in the context of casual labor markets in rural Malawi. Workers choose the minimum piece rate at which they are willing to accept a one-day contract to perform a simple task: sorting beans by type and quality. A piece rate offer is then generated randomly, determining whether the worker is given a contracttand, if so, the piece rate. Random assignment to a quality monitoring treatment provides exogenous variation in the workers’ incentives to trade off quantity of output for quality. A higher piece rate increases output in this setting, but does not attract more productive workers. Among men, the average worker recruited at higher piece rates is actually less productive. Local labor market imperfections appear to undermine the worker sorting observed in well-functioning labor markets. 

By Raymond Guiteras and B. Kelsey Jack (2014)