By Sabrina Genz and Claus Schnabel
Although numerous studies have investigated the aggregate employment effects of automation and digitalization, relatively little is known about the effects at the level of individual workers and along the gender dimension. Moreover, existing studies (in particular those on robots) often focus on the male-dominated manufacturing sector and thus provide few insights on gender heterogeneity. In contrast, this paper looks at individual workers and analyzes whether digitalization affects males and females differently, thus contributing to labor market inequality.
There are various reasons why exposure to new technology and its employment effects may vary by gender. Females disproportionately work in administrative support and service occupations while men are more likely to work in blue-collar jobs. Even within the same occupation, men often conduct different tasks than women. Gender segregation in the labor market and self-selection into different professions and task bundles within the same occupation both contribute to heterogeneous automation and digitalization risks for men and women. Despite early considerations of technology and gender, the topic remains under-researched, especially concerning digitalization.
In this paper, we analyze employment developments of male and female workers in Germany, a country in which gender equality with respect to digital transformation has gained considerable political attention. Using novel linked employer-employee data from 2011 to 2016 and applying a matching approach, we examine establishments and their workers after the first-time introduction of digital technologies. We compare workers in investing establishments with similar workers in establishments that do not make such an investment.
We find that the employment stability of incumbent workers is lower in investing than in non-investing establishments and that this difference is more pronounced for women than for men. We further document substantial heterogeneity in the employment effects across occupational tasks. Specifically, we differentiate workers across the two dimensions routineness and manual versus cognitive occupations. Females in non-routine occupations drive the difference in the employment response between investing and non-investing establishments. Females in non-routine manual jobs are on average four months less employed with their investing employer than their peers at non-investing employers. Since these females are only 76 days more employed at other employers, they spend on average 24 days more in unemployment than females in non-investing establishments. A similar pattern arises for females in non-routine cognitive jobs. These effects on (non-)employment are much smaller or even non-existent for male workers.
Taken together, our analysis suggests that employment adjustment processes following the first-time introduction of autonomous and self-contained technologies are most pronounced among females. Our results are consistent with evidence from Portugal and the U.S. documenting that women move out of occupations exposed to automation more quickly than men (Cortes et al., 2020). In other words, digitalization seems to be an example of gender-biased technological change.