GNP forecasts: How credible are they? A sequential test of consistency and rationality

In this study we examine the accuracy in the expectation formation process of a major macroeconomic forecast variable, namely the Gross National Product (GNP). The theoretical foundations are similar to the one used to study exchange rate expectations, i.e. a verification of consistency and rationality in forecast formation. A very reliable and continuos data set, the ASA-NBER survey is used. The Engle-Granger two step cointegration methodology and the Johansen-Juselius canonical correlation's (which has the smallest bias and dispersion) is applied to examine consistency in the gross national product expectation formation process. Our results support (reject) consistency at the short (long) forecast horizon. We then sequentially test for weak and strong form rationality using the Phillips-Hansen fully modified ordinary least squares procedure. This allows for an unrestricted cointegration test correcting for both endogeneity in the data and asymtotic bias in the coefficient estimates. Weak (strong) form rationality is upheld (rejected). This is in line with the literature which rejects orthogonality, but partially supports expectational rationality.