Your browser is not supported

Our website does not support the browser you are using. For a better browsing experience update to a compatible browser like the latest browsers from Chrome, Firefox and Safari.

Evaluating the Reserve Bank’s forecasting performance

This Bulletin article looks at the accuracy of our forecasts, if they were unbiased and how they compare with forecasts from other frameworks.

Thomas Bohm, Marea Sing


This paper attempts to answer 3 questions about our forecasts:

1. Are our forecasts unbiased? 

Unbiased forecasts are neither persistently too high or too low compared to the actual data outturns.

2. How accurate are our forecasts?

The accuracy of forecasts measures how large average forecast errors are, irrespective of whether the forecast was too high or too low. 

3. How do our forecasts compare to:

  • a benchmark forecast — we compare our average forecast errors to those from a simple forecasting model, which acts as a reference point.
  • the big 4 private banks’ forecasts over the COVID-19 pandemic period.

Key findings

  • The central forecast is an important input into monetary policy decisions for the Monetary Policy Committee (MPC), and an important component of our  publications. The forecast process is structured to ensure that the economic outlook and associated risks are comprehensively considered. 
  • Forecast error analysis furthers our understanding of macroeconomic relationships in New Zealand, allows for correcting of any persistently too high or too low forecasts, and ensures we continue to provide high-quality forecasts for monetary policy decision-making.
  • Our forecasts for inflation, the unemployment rate, and GDP growth did not tend to be persistently too high or too low on average over the period 2002 to 2022. In contrast, the forecasts for house price growth and changes in the exchange rate tended to be too low. However, these persistent misses were steadily decreasing until the outset of the COVID-19 pandemic. The forecasts for the short-term interest rate tended to be too high. This is mainly due to parts of the economy not having evolved as expected, which required monetary policy to keep the policy interest rate lower than forecast.

  • Unsurprisingly, average forecast errors are larger as the forecast horizon increases and during periods of heightened volatility and uncertainty (‘crisis’ periods), such as the COVID-19 pandemic and the global financial crisis. 

  • Our larger forecast errors during the COVID-19 pandemic is comparable to those of the big 4 private banks — ANZ, ASB, BNZ, and Westpac. None of the forecasters foresaw the strong inflation increases that have occurred since the June 2021 quarter. In New Zealand, and globally, the mismatch between robust demand and constrained supply has caused inflation to pick up much more than expected. All forecasters missed the extent of house price growth over the COVID-19 pandemic period.