Cruising to victory or a dead heat? Central Bank Championships in forecasting ability 2021 and 2022

Summary – in principle, a dead heat between the central banks

To the report's start page
Cruising to victory or a dead heat? Central Bank Championships in forecasting ability 2021 and 2022

Summary – in principle, a dead heat between the central banks

Published: 19 January 2024

It is difficult to make forecasts when conditions change drastically. This becomes clear when we study the forecasts of ten central banks for inflation in 2021 and 2022. We observe large forecasting errors across the board – especially for 2022. Neither the central banks nor other forecasters predicted how quickly inflation would rise and inflation forecasts accordingly showed major forecasting errors.

Simple measures that summarise forecasting errors show that the Riksbank’s forecasting ability for 2021 was good, but that it was not the best for 2022 – but also not the worst. Inflation rose relatively late in Sweden and the Riksbank did not adjust its forecasts sufficiently in late 2021 and the beginning of 2022. The forecasting errors for inflation in 2022 were therefore relatively large during this period. The conclusion in Hassler, Krusell and Seim (2023) that the Riksbank should have raised its preparedness when inflation began to rise abroad, is thus a relevant criticism if one only looks at these two forecast dates.

The simple measures that summarise forecasting errors compare inflation target variables that have different characteristics. For example, some countries are more dependent on electrical heating, while others may be more sensitive to supply disruptions. Other countries have closer ties to Ukraine and thus may have been more directly affected by the war. One way to account for these different conditions is to standardise the forecasting errors. When we do so, it turns out that the "forecast competition" is basically a draw.

Finally, it should be noted that what we study is the average of the inflation forecasts for 2021 and 2022. It may be the case that an evaluation of forecasts for individual quarters or certain forecast periods would differ from the results presented in this Commentary. Different approaches may also yield slightly different results, and the study should therefore not be interpreted as the Riksbank's overall assessment, or evaluation, of either its own forecasting ability or that of other central banks.