Structural Breaks, Biased Estimations, and Forecast Errors in a GDP Series of Canada versus the United States

Abstract

A structural break was suspected for the Canadian gross domestic product (GDP) time series when the reporting system switched from the Standard Industrial Classification system to the North American Industry Classification System system in 1997, as was previously detected for the United States. Any failure to identify in-sample breaks not only will produce biased parameter estimates but may adversely affect the model’s out-of-sample forecasting performance. This study investigated the possibility of poor forecast performance and biased estimation in the presence of the 1997 structural break in Canadian GDP. We confirmed the detected break in Canadian GDP data (1973–2014). All statistics indicated that the coefficients were not stable over time. Three models were employed to provide more accurate forecasts of GDP. The results demonstrate gains in forecasting precision when out-of-sample models accounted for structural breaks. Decision and policy makers might benefit from more precise GDP anticipation if the models were corrected for the 1997 break.