Gaute Torsvik, University of Oslo

Firm-level evidence on gender differences in pay and productivity


This project uses data from the customer service unit of an insurance firm to examine gender differences in pay and productivity. I will present material from two papers (both in progess) based on these data. The first, a project with Magne Mogstad and Jack Mountjoy, considers the effect of replacing a low-powered partly team-based sales bonus with a highpowered individual bonus? The results show a small increase in average sales, but exhibit a large gender difference in the reform effect. The pay reform induced male workers to increase their sales by more than 10 per cent on average, but did not have any effect on average sales among female workers. The reform also raised the sales target agents had to pass in order to obtain a bonus and made incentives beyond the target much steeper, giving especially high productivity workers incentives to increase effort. The results show that female workers respond accordingly, while there is no heterogeneity in the reform effect among male workers. The results from this study suggests that paying for personal performance raises the average gender gap in productivity and pay and increases inequality among female workers. Evidence from surveys conducted in the company substantiate and help explain the gender difference in response to individual incentives.

In a paper, co-authored with Arild Aakvik and Julian Vedeler, we examine how motherhood affects productivity and pay. Several studies have shown that there is a motherhood penalty in terms of labour market income. The mechanisms behind this penalty are only partially uncovered: While previous research shows that parenthood is associated with a decline in work hours of female employees, there is scarce evidence on the impact of parenthood on female worker productivity. Our data allow us to examine the impact of pregnancy and parenthood on female worker productivity. We find that weekly work-hours fall sharply during pregnancy, and never recovers to its pre-pregnancy levels. However, when accounting for this drop in actual work-hours, there is no effect of pregnancy and parenthood on female worker productivity.

Gaute Torsvik's research fields are Labour economics, Public economics and Development economics