Archives: IAB-Veranstaltungen

In 2016, the majority of full-time employed women in the U.S. earned significantly less than comparable men. The extent to which women were affected by gender inequality in earnings, however, depended greatly on socio-economic characteristics, such as marital status or educational attainment. In this paper, we analyzed data from the 2016 American Community Survey using a high-dimensional wage regression and applying double lasso to quantify heterogeneity in the gender wage gap. We found that the gap varied substantially across women and was driven primarily by marital status, having children at home, race, occupation, industry, and educational attainment. We recommend that policy makers use these insights to design policies that will reduce discrimination and unequal pay more effectively.

(joint with Philipp Bach (UHH) and Victor Chernozhukov (MIT))

We study how beliefs about math skills and fit affect occupational choice among Swiss students who are about to apply to apprenticeships. Although there is no gender difference in math skills, we document a substantial gender gap in preferences and search for math-intensive apprenticeships. We conduct a field experiment that randomizes the provision of gender-specific information on math ability and fit in gender-incongruent occupations. The intervention increases both boys' and girls' perceptions of fit in gender-incongruent occupations by 0.09-0.16 standard deviations.

Furthermore, it increases boys' (girls') probability of searching for information about any gender-incongruent occupation over the following two weeks by 44 (27) percent and leads to an increase in their plans to apply for trial apprenticeships and/or apprenticeships in these occupations. The effects on plans to apply for gender-incongruent occupations are driven by boys with low-math skills and girls with high-math skills. Later this year, we will link our survey data to administrative data from the largest apprenticeship application website to evaluate whether the intervention has effects on occupational choice.

Artificial Intelligence (AI) can perform cognitively demanding tasks with more autonomy than previous technologies and is thus expected to have disruptive effects on labor markets. But empirical evidence is limited. Does AI already affect workers’ wages? And how exactly does AI diffuse through labor markets? To answer these questions, we combine novel job vacancy data from Germany with high-quality administrative data and contribute three main findings.

First, using an IV approach, we find that a 10% increase in demand for AI skills implies average AI-induced wage returns of 2%. Second, we identify three key drivers behind our results and find that 95% of AI-induced wage effects are attributed to: (1) Employer Quality, (2) Socioeconomic, and (3) occupational characteristics. Third, we explore mechanisms, suggesting that the primary beneficiaries of AI demand are male workers with: (i) only modest AI exposure, (ii) college education, (iii) 50+ years of age, (iv) occupational mobility, and (v) employment at high-quality firms. Our paper provides valuable insights for policymakers by identifying early winners and losers of growing AI diffusion and offers promising avenues for future research.

1:00 to 1:40 p.m.: Creative Disruption – Technology innovation, labour demand and the pandemic (Prof. Harald Dale-Olsen)

We utilize a new survey on Norwegian firms’ digitalization and technology investments, linked to population-wide register data and show that the pandemic massively disrupted the technology investment plans of firms, not only postponing investments, but also introducing new technologies. More productive firms innovated, while less productive firms postponed investments. In the short-term, both firm productivities and worker wages increase on average, but this is driven by wage growth for skilled workers. New technologies are associated with increased long-term expected labour demand for skilled workers, and reduced demand for unskilled workers, particularly for the more productive firms.

(joint work with Erlin Barth and Alex Bryson)

1:40 to 2:20 p.m.: Did Covid-19 Accelerate the Digital Transformation? (Terry Gregory)

Using longitudinal survey data on technology use by German firms, matched with administrative worker–firm registers, we assess whether the Covid-19 pandemic accelerated

the adoption of cutting-edge technologies. Our data break down technologies by their application and level of sophistication, as well as capturing the timing of investments and whether the pandemic prompted these investments. We do not find evidence for an overall acceleration effect: Cutting-edge investments did not spike, and while they were more common among firms with higher remote work potential, such firms invested at a greater rate even before the pandemic, and also had more ambitious investment plans pre-pandemic. However, we do find that technologies facilitating remote work were adopted at a greater rate due to the pandemic, and these technologies appeared to have helped firms mitigate the negative employment effects of the crisis.

(joint work with Melanie Arntz, Michael Böhm, Georg Graetz, Florian Lehmer and Cäcilia Lipowski)

2:20 to 3:00 p.m.: The Pandemic Push: Digital Technologies and Workforce Adjustments (Christian Kagerl)

Using novel survey and administrative employer-employee data, we demonstrate that the COVID-19 pandemic was a push factor for the diffusion of digital technologies in Germany. About two out of three firms invested in digital technologies, particularly in hardware and software to enable decentralized communication, management and coordination. These investments also fostered additional firm-sponsored training, underscoring the complementary relationship between investments in digital technologies and training. We then show that the additional investments helped firms to insure their workers against the economic downturn. Firms that made such additional investments were able to retain more of their employees on regular working hours and relied less on short-time work schemes. Low and medium-skilled workers benefited the most from the insurance effect of digital investments.

(joint work with Christina Gathmann, Laura Pohlan and Duncan Roth)

Little is known about whether changes in job tasks due to technological progress affect personal wages and whether those changes in job tasks relate to the persistent gender wage gap in contemporary Western societies. Following the task-biased technological change approach, we analyze whether individuals who take on more non-routine job tasks characterized by a low automation risk (complex and autonomous tasks) are rewarded with higher wages. We separately analyze men and women and, due to the rigid German labor market, additionally account for job changes as a potential moderator. We use three-wave panel data covering a period of nine years from the German National Educational Panel Study.

Our results from fixed-effects regressions show substantial heterogeneity in the relationship between changes in non-routine job tasks and wages by gender and job change, which is masked when looking at average wage differentials by non-routine job tasks. While both genders benefit from increased task complexity in job changes, the impact is more pronounced for females, helping to slightly narrow the still persistent gender wage gap. However, when taking on more autonomous tasks in job changes, males experience significant benefits, further contributing to the widening of the gender wage gap. In essence, our findings underscore gender-specific monetary returns to increasing non-routine tasks, particularly highlighting the ability of male job changers to monetarize their newly assigned tasks. 

Joint work with Dr. Alexandra Wicht and Dr. Nora Müller.

Students often face incentives to reach performance goals, for instance, to receive a scholarship, enter a college, or be hired for a job. This paper uses a field experiment to study how incentives to reach performance goals affect students, whether the effects vary for students at different parts of the performance distribution, and whether allowing students to choose their own goal improves their performance. We find that incentives backfire: students offered incentives perform worse than their control counterparts.

These negative effects are mainly driven by mismatched goals: the negative treatment effects are concentrated among low-ability students who are assigned a high goal and among students with high aspirations who are assigned a low goal. The effects are also negative but not statistically significant from zero when students are allowed to choose their own goal. Our results show that incentives for performance goals can harm students' performance, especially among students whose goals are mismatched.