On social preferences at the workplace, one question may prove important: how discriminatory behaviors affect productivity?
This paper brings great insights on the mechanisms that led to output gap between subgroups of workers with different ethnicities, at a plant in Kenya. Using naturel experiments on team composition, the author isolates the effect of taste-based discrimination on the firm’s productivity. Three types of team composition are tracked: i) homogeneous teams, where all three workers (one supplier and two processors) are of the same ethnic group, ii) vertically mixed teams, in which both processors are of different ethnic group than the supplier and iii) horizontally mixed teams in which only one processor is of different ethnic group than the supplier.
The author finds that vertically and horizontal mixed teams are less productive than homogeneous teams. Since suppliers are paid for team productivity and processors for individual performance, this result suggests that the formers are willing to reduce their own pay to lower the pay of non-coethnic coworkers.
The author also looks at this variation of productivity within teams by exploiting a natural (political) experiment and finds that the output gap between homogeneous and diverse teams nearly doubled during political conflict between the two ethnic groups (of interest) in Kenya, which suggests an ethnic discrimination from the supplier.
Finally, the firm intervention, by changing the pay policies (team pay for processors), increased output in horizontally mixed team. The reason was the willingness of the supplier to reduce the horizontal discrimination (supplying more to non-coethnic processor). This change in pay policies also increased overall output.
This paper proves important by carefully identifying and explaining a negative effect of ethnic diversity on productivity in the private sector and by highlighting the distinction between taste-based discrimination and the other types of discrimination that would bias the analysis of firm efficiency.
Comments