The empirical performance of ECB/Eurosystem staff inflation projections since 2000

This article provides a technical evaluation of the performance of ECB/Eurosystem staff inflation projections since 2000. It complements the existing literature by examining the influence of HICP components as well as conditioning variables on the properties of HICP inflation projections, also taking into account potential time variation in forecast performance. The article shows how, from the low projection errors over the period leading up to the pandemic, Eurosystem/ECB staff forecast accuracy deteriorated in the face of atypical post-pandemic shocks before improving again since late 2022. However, it finds that the accuracy of Eurosystem/ECB staff projections of headline HICP inflation is broadly comparable to real-time market-based and private professional forecasts even after including the post-pandemic period of high inflation. The HICP forecast accuracy is comparable across main HICP components, including HICP excluding energy and food (HICPX), although HICPX inflation projections tend to show smaller errors than headline inflation projections. The article finds that ECB/Eurosystem staff inflation projections are unbiased overall but exhibit specific periods over the last 25 years in which this unbiasedness broke down. It also points to some rigidities in ECB/Eurosystem staff inflation projections, in particular for HICPX, which might explain part of this occasional bias. Finally, the article underscores the contribution of not only oil price assumptions but also other conditioning assumptions to the rigidities, occasional bias and reduced accuracy of ECB/Eurosystem staff projections of HICP inflation. JEL Classification: C53, E37, E58