A scenario presented by Philip Lane shows how AI could lift euro area output 3.8% above baseline in 2030

Euro area output could be 3.8% above baseline in 2030 if productivity in IT services increased by 30% simultaneously worldwide, according to a simulation presented by Philip Lane, a member of the Executive Board of the European Central Bank. The scenario, presented on October 30 in San Francisco, also shows the costs of reallocating activity, with lower employment in IT even as real wages increased across several occupational groups.

In brief The simulation assumes an unexpected and nearly permanent 30% global increase in the productivity of IT services. Euro area output would be 3.8% above baseline in 2030 and 4.1% higher in the long term; these are not annual rates or ECB forecasts. Employment in IT would fall by 4.4%, while real wages would rise by 4.3% under the model’s assumptions. Administrative staff outside IT would see employment 0.9% lower; other groups would gain higher real wages without a change in employment. The results depend on assumptions and on the economy’s adjustment. The cited model is a manuscript in preparation, not an assessment of layoffs that have already occurred.

Lane presented the exercise at a Federal Reserve Bank of San Francisco symposium on artificial intelligence and the economy. The calculation explores how a productivity gain concentrated in one sector would spread to the rest of the economy. It does not estimate the effect already produced by AI and does not represent an ECB commitment regarding future developments.

The MANKIO model assumes an unexpected and nearly permanent 30% increase in total factor productivity in IT services, occurring simultaneously in all represented regions. This means more output produced from the same combination of resources, not a 30% increase in the entire economy or in the productivity of every worker.

Under these conditions, euro area output would be 3.8% higher in 2030, five years after the simulated shock, and 4.1% higher in the long term. The differences are measured against the economy’s path without that shock. They are not annual rates that can be repeated or added up from one year to the next. The 5.5% increase in consumption mentioned in the speech refers to the long-term horizon, not to 2030.

The mechanism operates through the prices of IT services. In the model, their relative price falls by approximately one-third, while demand increases by 28%. Firms in other sectors can use cheaper IT services, improving their production capacity. In the sector where productivity has increased, however, the same activity requires less labour.

For IT services, the simulation indicates employment 4.4% below baseline and real wages 4.3% higher. In administrative occupations outside IT, employment falls by 0.9%, while real wages rise by 3.2%. For managers, professionals and technicians in other sectors, the model shows no change in employment but indicates real wages 4.4% higher.

Manual workers and service workers, grouped separately in the exercise, would see real wages 5.4% higher, with no change in employment. The results show that a technology can replace certain tasks while complementing others. Growth in aggregate output alone does not show who gains and who bears the adjustment.

The scale of the effect depends on the model’s assumptions. Lane cites the MANKIO paper as a manuscript in preparation. The size of the shock is linked to an NBER study that uses stock prices to estimate market expectations regarding software engineering productivity. That indicator, too, is not a direct measure of the additional output produced across all firms.

Research by the International Labour Organization on occupations’ exposure to AI starts from job tasks and stresses that exposure does not automatically mean the disappearance of a job. In practice, investment, staff training and the reorganisation of activity will influence both productivity gains and workers’ adjustment. The scenario presented by Lane explains these mechanisms without announcing layoffs or wage increases that have already been decided.