ECB’s AI‑Powered Inflation Model Beats Forecasts by 20 bps in 2025, Adding 60 Variables to Risk Analysis

The European Central Bank (ECB) has integrated an artificial intelligence (AI)-powered machine-learning model into its monetary policy toolkit since late 2022, enhancing its ability to track inflation risks. This move addresses the increasing complexity and uncertainty in global economic conditions, which complicate traditional forecasting methods.

The model provides a comprehensive assessment of inflation risks, drawing on a robust set of 60 variables. These variables cover a wide array of economic indicators, including expectations, costs, economic activity levels, and financial conditions. This broad data input allows the model to detect intricate and non-linear patterns often missed by conventional economic models.

In 2025, the AI model demonstrated its predictive capability. Actual inflation outcomes in both the second and fourth quarters of that year exceeded the ECB’s baseline projections by 20 basis points (bps). A basis point represents one-hundredth of a percentage point, meaning inflation was 0.20% higher than anticipated. Crucially, the machine-learning model had previously flagged these upward inflation risks, aligning its signals with the eventual real-world data.

The integration of such sophisticated tools indicates a shift in central banking towards data-driven, real-time risk analysis. By processing a greater number of indicators and capturing complex data relationships, the machine-learning model offers a more nuanced understanding of potential inflation trajectories. This capability supports more informed decision-making in monetary policy, moving beyond reliance on a limited set of indicators and restrictive assumptions.

Policymakers will likely continue to expand the application of AI and machine learning, seeking to further refine inflation forecasting and risk assessment in an unpredictable global economy.