Incorporating contemporaneous residual relationships: A new approach

A direct (and easy to use) measure for capturing any information contained in the contemporaneous relationships among the residuals of different equations is derived herein from the moment generating function of the joint probability density function of the residuals.     

STATISTICAL TESTS OF THE ACCURACY OF ALTERNATIVE FORECASTS: SOME RESULTS FOR U.S. UTILITY BETAS

The common approach for assessing the relative forecast accuracy of various predictors is simply to compare their simple ranks based on some measure(s) of forecast accuracy (such as the mean square errors). However, conventional methodologies for conducting a t-test or F-test of forecast accuracy are inappropriate and unreliable. A statistical methodology is proposed in this paper that rectifies most of the problems involved with conducting such conventional tests of statistical significance. Using this statistical methodology on the mean square errors of six beta predictors for five forecast horizons for fifty utilities, it was found that there are statistically significant differences between the forecast accuracies of the beta predictors for each and every forecast horizon, that the statistical rankings of the beta predictors were significantly different from and more congruent than the simple rankings as the forecast horizon lengthened, and that the ordinary least squares predictor was consistently ranked as one of the poorest beta predictors for all of the forecast horizons.