ECB Paper Introduces Deep-π*, a Neural Network Estimate of US Trend Inflation

What the ECB Published

The European Central Bank released a paper titled "Deep-π*: a neural network estimate of trend inflation" on 4 November 2026. Most empirical models treat trend inflation as a latent random walk driven by unobserved shocks, so they cannot show which economic conditions move it. Deep-π* instead makes the innovations to the trend a possibly nonlinear function of many macroeconomic indicators, estimated with neural networks. Shapley values then attribute each movement of the trend to the indicators behind it.

Two loss functions define two measures: a current trend, which fits inflation today, and a forward-looking trend, which best predicts inflation over the next year.

Findings in US Data

Employment-related variables were the largest contributors to the post-pandemic rise of the current trend. The forward-looking trend rose earlier, with real activity, employment, and money and credit as the main drivers from the start of the pandemic. It peaked eight to twelve months before the current trend. Both trends looked through most of the movements in energy and commodity prices.

The contributions of the indicators are nonlinear, with sizeable interactions between them. Re-estimated every month during the surge, the model forecast inflation more accurately than the standard UC-SV model and a random walk. In early 2021, when professional forecasters expected inflation of 3% or less, the forward-looking trend was already signalling an increase in persistent inflation.

Notes for EA trading

Research papers like this one are not scheduled data releases, but inflation and central bank communication can still affect market sentiment. Around major inflation data and policy events, volatility can rise and spreads can widen. Traders running EAs may want to consider pausing them or reviewing their risk settings during such periods.