Forecasting with Many Predictors

The calibration of the two parameters allows to shrink the number of coefficients to be estimated and to determine the regressors to consider in the factor derivation process. [...] The Canadian confidence dataset is thus a panel of 88 time series that contain the raw data for all questions and all answering options to four confidence surveys managed by the Conference Board of Canada and the Bank of Canada. [...] The responses to nine of the questions are available in the form of balance of opinions (the difference in percentages between the opposite options of each question), except for the responses on labour shortages (one variable), ability to meet the demand (arranged in two variables) and inflation expectations (arranged in four variables which are in percentages). [...] Once again, Figure 2 illustrates the relationship between Canadian GDP and simple- sum aggregates of the two surveys managed by the Bank of Canada, to obtain intuition about the likely forecasting ability of the data underlying these survey (the figure depicts negatives value for the index associated to the SLO, which fits better with the GDP growth dynamics. [...] The entries recorded in the tables report theMSFErelative for a specific case, derived as the ratio of the mean-squared forecast error obtained with the model considered to the one obtained with each of two benchmarks: the univariate AR benchmark (Panel A of each table) or the factor model using only standard Cana- 21 dian macroeconomic and financial variables as predictors (labelled CA, Panel B o