AI model enhances ECB inflation risk tracking
A new machine learning model developed by the European Central Bank helps experts track inflation risks in real time. It assesses the likelihood of inflation deviating significantly from baseline forecasts, particularly in uncertain economic environments.
Navigating complexity with machine learning
In times of heightened economic and political uncertainty, price dynamics can become more volatile, making monetary policy decisions reliant on a robust assessment of inflation risks.
Traditional economic models, often based on a limited number of indicators and restrictive assumptions, struggle to capture the full spectrum of potential developments.
The ECB's new AI-based tool, a quantile regression forest (QRF) model, addresses these limitations by processing a greater number of economic indicators and detecting complex, non-linear data patterns that conventional models often miss.
This allows for a more comprehensive understanding of how likely inflation is to be higher or lower than the baseline projections, providing crucial insights for the Eurosystem's analytical toolkit.
Real-time risk signals in practice
The QRF model serves a dual purpose: it produces inflation forecasts and offers a detailed assessment of risks surrounding the baseline outlook, leveraging a large set of routinely monitored economic variables.
Integrated into the ECB's monetary policy preparation toolkit since late 2022, the model has proven particularly useful in detecting emerging inflation risks across various components of the Harmonised Index of Consumer Prices (HICP) in real time.
For instance, in 2025, the model effectively identified upside risks for core inflation (HICPX) in the second and fourth quarters, where actual inflation subsequently exceeded ECB/Eurosystem projections by 20 basis points, demonstrating its informative value in volatile periods.
A necessary evolution, not a silver bullet
While the QRF model offers valuable enhancements for inflation risk analysis, it remains a complementary tool, not a replacement for expert judgment.
Its reliance on historical data means it may struggle with truly unprecedented events, requiring continuous adaptation and validation.
Nevertheless, its ability to process vast datasets and identify subtle patterns marks a crucial step forward for central bank forecasting and monitoring economic trends.