ECB Paper Finds Output Gap Estimates Subject to Large Revisions Driven by GDP Data and Potential Gro

Key Takeaways

• An ECB Occasional Paper (No 399) analyzes revisions to output gap and potential growth estimates from the European Commission, Eurosystem, IMF, and OECD.
• It finds that real GDP data revisions and potential growth revisions drive most output gap revisions, and that real-time slack estimates are unreliable.
• This is a research publication, not a binding rule, but compliance and risk teams that rely on macroeconomic inputs in models, stress tests, and planning should treat it as a caution on model risk.

Deep Dive

An ECB Occasional Paper (No 399) analyzes revisions to output gap and potential growth estimates from the European Commission, Eurosystem, IMF, and OECD. It finds that real GDP data revisions and potential growth revisions drive most output gap revisions, and that real-time slack estimates are unreliable. This is a research publication, not a binding rule, but compliance and risk teams that rely on macroeconomic inputs in models, stress tests, and planning should treat it as a caution on model risk.

Using annual vintages for up to 20 euro area countries (2002-2025) and a balanced panel of 11 countries (2007-2025), the author reports differences in reliability across institutions. European Commission estimates revise least in the balanced sample, followed by the Eurosystem and the IMF, while OECD estimates revise most. Eurosystem estimates have become markedly more stable over the past decade, likely due to the Annual Supply Side Review and more systematic cross-country review. IMF estimates look stable for individual years but involve more extensive revisions to the historical series. Revisions stem mainly from real GDP data revisions and potential growth revisions, while nowcast errors play a smaller role. Panel regressions suggest that around half of GDP data revisions and more than one-third of GDP forecast errors pass through into potential growth revisions. As a result, weak growth may later be reclassified as lower potential rather than a larger negative gap, which changes the policy diagnosis from insufficient demand to weaker supply capacity.

Compliance and model risk teams should:

(1) inventory internal models and scenarios that use output gap or potential growth inputs and document the source institution and vintage;

(2) test sensitivity of outputs to plausible revisions, including use of multiple institutional sources;

(3) supplement output gap measures with less frequently revised indicators, such as survey-based capacity utilization and demand as a factor limiting production, as the paper recommends;

(4) record in model validation files the uncertainty and limitations of real-time estimates; and

(5) separate changes in economic assessment from changes in estimation methodology when interpreting updated inputs.

The overall impact is low and indirect, with no implementation timeline. The main exposure is model risk and governance risk: overreliance on point estimates of economic slack could distort scenario design, capital and provisioning forecasts, and assessments of the monetary policy path or fiscal space. Teams should monitor whether supervisors or policymakers cite this work in future guidance on model uncertainty or scenario design.

Source: European Central Bank