Forecasting Canadian Inflation Using Dynamic Model Averaging

Awards

Author(s):

Category:

Institution:

Region:

Winner Category:

Year:

Abstract:

This paper forecasts CPI inflation and core-CPI inflation measures for Canada using Dynamic Model Averaging (DMA) over the period 1990Q1 to 2019Q2. DMA combines a state space model with a markov chain model which allows the significance of the predictors and the model itself to change over time. We compare the forecasting performance of DMA against traditional forecasting models. There are two key results. First, we find that DMA improves upon forecasts produced by standard econometric models for the short run for both CPI and core-CPI inflation. Second, we find that DMA’s forecasting performance is more ambiguous in the medium and the long run.

Attached Documents:

Record copies