Forecasting The Canadian Unemployment Rate

Title: Forecasting The Canadian Unemployment Rate
Authors: Abdikader, Omar
Date: 2019
Abstract: This purpose of this paper is to forecast the Canadian unemployment rate (UER) using a seasonal ARIMA model, a seasonal random walk model, and an autoregressive distributed lag (ARDL) model containing the West Texas Intermediate (WTI) spot price of oil as a leading indicator variable. The seasonal ARIMA or SARIMA model will be developed based on the Box-Jenkins method, and the seasonal random walk model will be determined based on the level of integration required to make the unemployment rate series stationary. One-step ahead and 3-steps ahead forecasts are carried out. The performance of each model is assessed based on three forecasting evaluation criteria: the root mean squared error (RMSE), the mean average (MAE), and the mean absolute percentage error (MAPE). The statistical significance of differences in forecast accuracy is then evaluated using the Diebold-Mariano test and the Clark-West test to determine the overall best model at each forecast horizon. The results obtained show that the models that most accurately forecast the Canadian unemployment rate are the ARIMA(||1,3,5||,0,0)(0,1,1)12 and the ARIMA(||3,4,5||,1,0)(0,1,1)12. Overall, the results obtained in this study provide little evidence for the use of the oil price as a leading indicator to forecast the Canadian unemployment rate due to the poor performance of the ARDL models compared to the ARIMA models.
CollectionÉconomie - Mémoires // Economics - Research Papers
Omar Abdikader.pdf531.68 kBAdobe PDFOpen