Productivity without the boom: Jefferson on AI, Schnabel on scale and the OECD's ICT shortage
On 13 November two central bankers on opposite sides of the Atlantic gave speeches about the same problem from different ends, and an OECD dataset published the same day added a third view.
Federal Reserve Vice Chair Philip Jefferson asked whether AI is visible in US productivity statistics. US labour productivity growth averaged 2.2% between 2020 and 2025, up from 1.5% in the previous cycle, but the pickup began before chatbots appeared in 2022 and he said it is too early to credit AI. About 25% of US firms used generative AI in at least one business function as of mid-year, rising to about 40% in information and professional services. For now, he argued, the technology raises demand, through equity wealth and data-centre investment, before it raises supply, which adds to price pressure while PCE inflation is 3.6%.
ECB Executive Board member Isabel Schnabel, speaking in Vilnius, started from Europe's weaker productivity growth and gave a reason Jefferson's history also contains. Europe's gap with the US widened in the early 2000s because it adopted ICT more slowly, and one cause was firm size: firms with more than 250 employees account for almost 60% of US employment but about 20% in many euro area countries. Fixed costs of adopting new technology are easier to spread over a large market. Her remedy is scale, through a single European company-law regime.
The OECD's skill-shortage index, as analysed by Euronews, adds the labour side. ICT professionals are hard to find in 18 of 27 European countries and among the five hardest occupations to fill in 12. At the same time an OECD economist said AI may reduce demand for some ICT roles, especially at entry level, and the OECD found the unemployment gap for young graduates widening.
Taken together, the three pieces describe adoption as the bottleneck. Jefferson's evidence says use is spreading but shallow, concentrated in a few business functions and offset by new costs such as debugging AI-generated code. Schnabel's says Europe's firms are too small to absorb those costs as quickly as American ones. The OECD's says the specialists who would do the adopting are scarce, even as the technology may thin their entry-level ranks.

Why it matters
For monetary policy, the implication is the same on both sides: the productivity dividend that would let economies grow faster without inflation is not yet in the data, and it will not arrive evenly. Central banks that wait for it before easing may wait longer in Europe, where the structural obstacles Schnabel lists are not ones interest rates can remove.