Firm training is widely regarded as crucial for protecting workers from automation, yet there is a lack of empirical evidence to support this belief. Using internationally harmonized data from over 90,000 workers across 37 industrialized countries, we construct an individual-level measure of automation risk based on tasks performed at work. Our analysis reveals substantial within-occupation variation in automation risk, overlooked by existing occupation-level measures. To assess whether firm training mitigates automation risk, we exploit within-occupation and within-industry variation. Additionally, we employ entropy balancing to re-weight workers without firm training based on a rich set of background characteristics, including tested numeracy skills as a proxy for unobserved ability. We find that training reduces workers’ automation risk by 3.8 percentage points, equivalent to 8% of the average automation risk. The training-induced reduction in automation risk accounts for 15% of the wage returns to firm training. Firm training is effective in reducing automation risk and increasing wages across nearly all countries, underscoring the external validity of our findings. Training is similarly effective across gender, age, and education groups, suggesting widely shared benefits rather than gains concentrated in specific demographic segments.

Das Produktionspotenzial der deutschen Wirtschaft wächst mittelfristig (2023 bis 2029) mit einer jahresdurchschnittlichen Rate von 0,3% und damit deutlich schwächer als in den Jahren zuvor. Dies ist auf eine ungünstigere Entwicklung aller drei Faktoren (Arbeitsvolumen, Kapitalstock, totale Faktorproduktivität) zurückzuführen. Das potenzielle Wachstum wird insbesondere durch den Rückgang der durchschnittlichen Arbeitszeit gedämpft.

Using a newly collected dataset at the plant level from 2014 to 2018, we provide the first microscopic portrait of robotization in Germany and study the correlates of robot adoption. Our descriptive analysis uncovers five stylized facts: (1) Robot use is relatively rare. (2) The distribution of robots is highly skewed. (3) New robot adopters contribute substantially to the recent robotization. (4) Robot users are exceptional. (5) Heterogeneity in robot types matters. Our regression results further suggest plant size, high-skilled labor share, exporter status, and labor shortage to be strongly associated with the future probability of robot adoption.