I remain optimistic that AI has the potential to generally increase productivity in service courts and white collar jobs (that is, most jobs nowadays). Although the productivity gain will probably be much smaller than AI promotors promise (is that not always the case with new technologies?). Last year I wrote about Daron Acemoglu’s estimates That generative AI will only add about 0.66% of productivity growth to the US economy in the following decade. However, it seems that the productivity gains in Europe can be somewhat higher.
Use the Acemoglu model, A team of the IMF Estimated the productivity gain available for European countries of the approval of generative AI. The graph below shows their estimates for productivity gain based on AI acceptance rates in various industries, as well as the productivity gain feasible via Generative AI in different sectors. Note that the graph contains a black line that indicates the European average and a red line for the US.
Productivity gain in Europe versus the US
Source: Misch et al. (2025). NOTE: Black line indicates the European average and the red line is the US.
As you can see, the productivity gain in Europe can approach on average 0.8%, compared to 0.66% in the US. Note that richer countries in Europe, such as Switzerland, France, Germany or Luxembourg, can increase productivity more than countries with a lower income. This reflects (i) the larger service sector in countries with a higher income and (ii) higher labor costs in these countries, which increase the incentive to invest in AI to replace employees.
Suppose you are a business owner in a rich country with many payroll taxes and stifling labor laws that make it incredibly expensive to hire people and even harder to dismiss them again (looking at you, France). In that case you will look forward to AI. It ensures increased flexibility in production, while the costs and workforce are reduced.
If you are a white collar employee in an industry where AI can replace many people (looking at me in a mirror, now), you might be afraid of the future of your career.
But don’t fear because European countries can waste their productivity gains from AI in the way they usually do that: by introducing stifling regulations.
The authors of the research try to estimate the impact that the current and planned EU regulations will have on the acceptance of AI and the costs of implementation and compliance. They emphasize that the graph below shows an upper limit of reducing productivity growth, because not all laws and regulations apply in every country and every job. Even if the resulting decrease in productivity growth is only half as much as shown below, it becomes a serious care.
Note that all combined regulations can reduce by approximately 35% to 40%. Thus, instead of 0.8% in productivity growth, productivity growth would shrink to 0.5%. Even if the regulations reduce growth by half, however, the European productivity gain would still be lower than that of the US, mainly as a result of European over -regulation.
Relatively loss of productivity gain due to European regulations
Source: Misch et al. (2025)
#Europe #benefit #wont