Research

The rapid evolution of technology is reshaping labor markets by altering skill demands and job profiles. This paper introduces a novel skill-based measure of occupational technology intensity — the Occupational Technology Skill Share (OTSS) — that distinguishes between manual, digital, and frontier technologies, including artificial intelligence (AI). Using natural language processing, generative AI, and supervised machine learning, we develop an AI-powered skill classification that enriches occupation-linked skill labels with standardized GenAI-generated descriptions and structured indicators of technological content, enabling transparent classification by technology intensity. We compute OTSS for all occupations in the German labor market. For the average worker in 2023, manual technologies account for the largest share of skill content (42%), followed by digital (38%) and frontier technologies (20%). Frontier technologies remain concentrated in specialized occupations, while digital technologies are widespread. Linking these measures to administrative data from 2012–2023 shows a broad shift from manual and digital toward frontier skills across occupations, and reveals a non-linear, U-shaped relationship between changes in frontier skill intensity and employment growth.

We provide micro-founded evidence on how establishment-level adoption of frontier technologies shapes the economy-wide occupational employment structure. Linking survey data on technology adoption between 2011 and 2016 by 1,660 German incumbent establishments to administrative worker records, we find that establishments adopting frontier technologies such as AI or smart factories contributed disproportionately to the overall decline in routine jobs. Crucially, these overall effects do not stem from average within-establishment adjustments but from a subset of large and less routine-intensive adopters. Our findings highlight that frontier technology adoption is not routine-replacing per se: establishment size and an initially non-routine-intensive workforce are key conditions for de-routinization to occur.

Digital technologies displace labor from routine tasks, raising concerns that labor is racing against the machine. We develop an empirically tractable task-based framework to estimate the aggregate employment effects of routine-replacing technological change (RRTC), along with the labor and product demand channels through which this aggregate effect comes about, focusing on the role of inter-regional trade. While RRTC has indeed had strong displacement effects in Europe between 1999 and 2010, it has simultaneously created new jobs through increased product demand, resulting in net employment growth. However, the distribution of gains from technological progress matters for its job-creating potential.

Minimum wage levels are rising around the globe. To shed more light on the possible unintended side effects of higher wage floors, we study the impact of a minimum wage introduction on wages and employment in a quasi-experimental setting where the minimum wage is set extraordinarily high during an economic downturn. We identify treatment effects along different wage and skill groups and find positive wage and employment spillover effects for medium-skilled workers with salaries just above the minimum wage. More striking, we find negative wage and employment effects for high-skilled workers who are further up the wage distribution, followed by reduced returns to skills and industry skill supply. We explain these adjustments with a substitution-scale model that predicts negative spillover effects whenever labor-labor substitutions toward high-skilled workers are overcompensated by an overall decline in industry employment. Even though we focus on a specialized industry, we lay out the general conditions under which such unfavorable adjustments may occur in other contexts.

Like many other countries, Germany has experienced rapid population and workforce ageing, yet with substantial variation across regions. In this paper we first use this spatial variation between 1975 and 2014 to estimate quasi-causal supply effects of ageing on regional labour market outcomes, drawing on the identification strategy of Böhm and Siegel (2020). We find in our panel of German labour market regions that workforce mean age has considerable negative effects on the wage returns to age. We also obtain suggestive evidence that relative employment rates of older workers decline when mean age rises. A decomposition of the heterogeneous regional trends using our estimates shows that ageing of rural regions is mainly driven by supply (reflecting local population dynamics) whereas urban ageing is driven by demand (reflecting responses to economic conditions). We discuss the differential implications of these drivers for regional policy.

In light of rapid advances in the fields of Artificial Intelligence (AI) and robotics, many scientists discuss the potentials of new technologies to substitute for human labor. Fueling the economic debate, various empirical assessments suggest that up to half of all jobs in western industrialized countries are at risk of automation in the next 10 to 20 years. This paper demonstrates that these scenarios are overestimating the share of automatable jobs by neglecting the substantial heterogeneity of tasks within occupations as well as the adaptability of jobs in the digital transformation. To demonstrate this, we use detailed task data and show that, when taking into account the spectrum of tasks within occupations, the automation risk of US jobs drops, ceteris paribus, from 38% to 9%.

Demographic ageing is expected to affect labour markets in very different ways on a regional scale. Contributing to this debate, we explore the spatiotemporal patterns of recent distributional changes in the worker age structure and innovation output for German regions by conducting an exploratory space–time data analysis. Besides commonly used tools, we apply newly developed approaches which allow the investigation of joint dynamics of the spatial distributions. Overall, we find that innovation hubs tend to be located in areas with high skill concentrations, but also seem to coincide with favourable demographic age structures. We show that these concentrations are persistent over time due to clusterwise path dependence and spatial contagion forces. The spatiotemporal patterns speak in favour of a demographic polarization process of German regions where the post-reunification East–West divide is increasingly turning into a rural–urban divide.

This paper examines the determinants of skill-selective regional migration in a context where modelling the migration decision as a wage-maximizing process may be insufficient due to persistent employment disparities. Based on a Borjas-type framework it is shown that high-skilled workers are disproportionately attracted to regions with higher mean wages and employment chances as well as higher regional wage and employment inequalities. Estimates from a labour flow fixed-effects model and a general methods of moments (GMM) estimator show that these predictions hold, but only employment disparities induce a robust and significant skill sorting. The paper thus establishes a missing link about why employment disparities may actually be self-reinforcing.

This study contributes to the sparse literature on employment spillovers of minimum wages. We exploit the minimum wage introduction and subsequent increases in the German roofing sector that gave rise to an internationally unprecedented hard bite of a minimum wage. We look at the chances of remaining employed in the roofing sector for workers with and without a binding minimum wage and use the plumbing sector that is not subject to a minimum wage as a suitable benchmark sector. By estimating the counterfactual wage that plumbers would receive in the roofing sector given their characteristics, we are able to identify employment effects along the entire wage distribution. The results indicate that the chances for roofers to remain employed in the sector in eastern Germany deteriorated along the entire wage distribution. Such employment spillovers to workers without a binding minimum wage may result from scale effects and/or capital–labour substitution.

The minimum wage in the roofing sector, which was introduced in 1997 and transferred to a national and generally binding level in 2003, has particularly affected the sector in Eastern Germany, even in an international comparison. However, the concomitant effective cost burden is limited. Based on difference-in-differences estimations in comparison to an ancillary construction sector that is not affected by minimum wages as well as based on a comparison of workers in the roofing sector who have been affected by minimum wage introduction to different degrees, causal effects with regard to employment, worker protection and competition are investigated. The results show that the wage increases associated with the minimum wage introduction only partially compile into earnings increases. Moreover, the overall employment situation does not seem to change even though some negative employment results can be detected for workers affected by binding minimum wages.

We develop a novel firm-level indicator of Artificial Intelligence adoption in Europe (MAP-AI) by extracting information on AI usage from more than three million firm websites from Belgium, France, Germany, and Luxembourg (2016–2024) using a Large Language Model. The indicator captures realized AI adoption as signaled on their website rather than potential exposure. Our method allows to detect not only whether firms adopt AI, but also their role in the AI ecosystem and the type of AI technology they employ. Validation against human-coded benchmarks and external references confirms high accuracy and external validity. We find that the share of AI-active firms grew from 1% in 2016 to 12% in 2024, with acceleration after 2022. We document a structural transformation in the AI ecosystem, as expanding AI adoption increases the share of adopters in overall AI activity, signaling widespread diffusion and more integrated AI use, including generative AI. While adoption is concentrated among larger, younger, knowledge-intensive firms in urban innovation clusters, workforce skills emerge as a key factor associated with AI adoption. Our skill-level analysis shows that foundational data skills form a necessary base for adoption, while a small set of specialized AI skills — such as machine learning and natural language processing — act as strong complements, highlighting human capital as a central driver of AI diffusion across firms.

We investigate the diffusion of frontier technologies across German firms before and during the Covid-19 crisis. Our analysis tracks the nature, timing, and pandemic-related motivations behind technology investments, using tailor-made longitudinal survey data linked to administrative worker-firm records. Technologies adopted after the onset of the pandemic increasingly facilitated remote work and mitigated the negative employment effects of the crisis. Overall, however, investments in frontier technologies declined sharply, equivalent to a loss of 1.4 years of pre-pandemic investment activity. This procyclical adoption pattern is particularly striking since the pandemic created clear incentives to experiment with new technologies. Our findings highlight how short-run fluctuations may influence medium-run economic growth through their impact on technology diffusion.

Within a quasi-experimental setting, we show that the first-time adoption of industry-specific minimum wages led to a doubling in the share of solo self-employment in areas with a strong bite. We explain this with the minimum wage-driven cost shock, which led to reduced labor demand and wages for dependent employment while creating incentives for independent employment. Our results suggest that dependent workers have been involuntarily pushed into more precarious alternative work arrangements, with poorer social security and lower incomes. Such unintended side effects are likely to occur when the minimum wage is set extraordinarily high, especially during an economic downturn.

We study how firm-level technology adoption affects worker outcomes using linked employer–employee data for Germany that combine administrative records with a survey of 2,032 establishments (2011–2016). We classify technologies into established digital and frontier technologies (e.g., AI, AR, 3D printing) and estimate their causal effects using an instrumental-variables strategy that exploits firms’ pre-adoption technological readiness, measured by the historical share of workers with technology-related skills. Technology adoption increases employment stability and wages, with substantially larger gains from established digital technologies. Effects are highly heterogeneous: workers in occupations undergoing task restructuring and those performing complex tasks benefit most. Crucially, returns depend on firms’ deployment strategies. Technologies that augment labor generate sizable gains, whereas substitutive adoption yields limited benefits. These results highlight that the effects of technological change depend on how technologies are integrated into production.

The capabilities of AI have expanded rapidly in recent years; yet, our understanding how AI affects labor demand remains very limited. We use 31 million online job ads from four European countries to investigate how the demand for skills and occupations has shifted in response to AI. Our analysis relies on advanced ML models to extract and classify skills from job ads in multiple languages. We document a significant growth in the demand for AI, data and prediction skills in exposed occupations, but a decline in social skills since 2018. We also observe significant shifts in the skill sets of exposed occupations pointing towards greater skill specialization. Finally, we show that there is some indication that exposed occupations become less labor-intensive, but that employers demand more workers in these occupations.

This paper examines how the adoption of Artificial Intelligence (AI) affects employment, workforce composition, and productivity in European firms. We combine a novel firm-level measure of realized AI adoption, derived from large-scale analysis of firms’ website content, with administrative employer–employee records and firm balance-sheet data. To identify the effects of AI adoption, we exploit variation in the timing of adoption across firms and implement an event-study design that traces firm outcomes before and after the first observed adoption of AI. This approach allows us to characterize the dynamic adjustment of firms and to assess whether adopters and non-adopters exhibit similar pre-trends prior to adoption. We study how AI adoption changes total employment, the composition of the workforce by skill, and firm performance measures such as productivity and revenues. By providing firm-level evidence on the timing and consequences of AI adoption, the paper contributes to the emerging literature on technological change, skill demand, and firm dynamics in Europe.

This project investigates how technological change reshapes labor demand and the content of work over the long run. Using a novel dataset of digitized newspaper job advertisements from Luxembourg spanning nearly two centuries, we apply large language models and natural language processing techniques to extract structured information on job titles, tasks, and skills. We link these data to historical measures of technological breakthroughs to analyze how occupations emerge, evolve, or disappear in response to innovation. Our approach allows us to trace changes in job content, identify shifting skill requirements, and quantify the extent to which different occupations are exposed to technological change. By providing a detailed, task-level perspective on labor market transformation, the project offers new evidence on how technological progress reshapes the nature of work and informs current debates on automation and skill demand.

A large literature studies the impact of minimum wages on employment, focusing on potential displacement effects for workers. In this paper, we ask whether minimum wages encourage firms to automate — to substitute capital for low-wage labor. We study the first-time adoption of industry-specific minimum wages in Germany, whose bite far exceeds that of most minimum wages studied in the literature, using firm balance sheet data linked to administrative records on the firms’ entire workforce. The minimum wages raised the wages of affected workers without reducing employment. Establishments more exposed to the new wage floor increase their fixed assets and capital-labor ratio relative to less exposed establishments in the same industry; the effects are identified from East Germany, where the minimum wages bite deepest. Consistent with automation, the capital response is concentrated among establishments whose workforce performs routine, automatable tasks: these establishments deepen capital by about 16 percent while holding employment constant, whereas establishments with predominantly non-routine work expand employment instead. The response is defensive rather than expansionary — sales do not rise — and it is carried by small incumbent establishments, which retain their routine workforce; new establishments respond instead by entering with markedly fewer routine tasks. The pace of automation thus responds to the price of low-wage labor.