Code available at: GitHub

This paper examines the quality of projections of real GDP growth taken from the Bank of Canada Monetary Policy Report (MPR) since they were first published in 1997. Over the last decade, it has become common practice among the central banking community to discuss forecast performance publicly. The assessment we undertake is on annual forecasts as well as the average prediction over the policy horizon. We find that the MPR is more accurate than a naïve forecast model and marginally superior to a consensus of professional forecasters. The accuracy of the MPR annual predictions, measured by the root-mean-square prediction error (RMSPE), improves from 1.6 to 0.6 percentage points as more data become available. On a two-year average basis, the RMSPEs are about 1.0 percentage point for forecasts made in April and October. Our results also suggest that the bias present in MPR forecasts is often not statistically significant for both annual and two-year projections. Nonetheless, we found a tendency to overpredict growth at the beginning of the forecast cycle. Finally, at the beginning of the forecast cycle, the MPR correctly predicts the sign of the change in annual real GDP growth roughly 50 per cent of the time, improving to about 75 per cent at the end of the cycle. The sign of change is correctly predicted roughly 60 per cent of the time for the two-year average prediction.