Forecasting US GNP Growth: The Role of Uncertainty

There are a large number of models developed in the literature to analyse and forecast the changes in output dynamics. The objective of this paper is to compare the forecasting ability of uni- and bivariate models in terms of forecasting U.S. GNP growth at different forecasting horizons, with the bivariate models containing information on a measure of economic uncertainty. Based on point and density forecast accuracy measures, as well as the superior predictive ability (SPA) and equal accuracy tests, we evaluate the forecasting performance of our models over the quarterly period of 1919:2-2014:4, given an in-sample of 1900:1 1919:1. We find that the economic policy uncertainty index should be improving the accuracy of U.S. GNP growth forecasts in the bivariate models. While we find that the Markov switching time varying parameter VAR (MS-TVP-VAR) models in most cases cannot be outperformed its competitive models according to the root mean squared error (RMSE), the density forecast measure reveals that the Bayesian VAR (BVAR) model with stochastic volatility in most cases is the model that produces more accurate forecasts. More importantly, our results highlight the importance of uncertainty in forecasting US GNP growth rate.