On the relative worth of recent macroeconomic forecasts

The accuracy of recent forecasts of key macroeconomic variables by a variety of forecasters is analyzed over the period 1980II to 1985I. With the exception of the Bayesian VAR forecasts of cumulative growth rates in some real variables, most of these forecasts are so inaccurate that the forecast MSE exceeds the variance of the variable being forecast. This implies that substituting such a projection into a forecasting model estimated using one of these variables as an explanatory variable will provide inferior forecasts relative to a naive model (incorrectly) dropping the explanatory variable altogether. Put another way: most of these forecasts are so inaccurate that simple extrapolation of historical trends is superior for forecasts more than a couple of quarters ahead.