Forecasting Accuracy of the McGraw-Hill Anticipatory Data

Abstract

Abstract This study analyzes the forecast accuracy of the McGraw-Hill (M-H) anticipations with one- to four-year horizons of investment, sales and capacity for major manufacturing industries for the period 1948–71. Average absolute percentage errors are presented for each anticipation series. In comparison to the BEA survey of one-year investment anticipations, the M-H data are generally less accurate. On the other hand, in forecasting experiments the M-H series are shown to be more accurate predictors than several naive statistical models and yield lower forecast errors than the accelerator and neoclassical econometric models of investment behavior.