by Stephen K. McNees
July/August 1992

Opinion about the reliability of economic forecasts ranges widely. Some argue that they are literally worthless, at the same time that most forecasters can point to a sequence of near perfect predictions. How much confidence should one place in economic forecasts? The errors vary with many factors.

A crucial determinant of the size of forecast errors is the forecast period; some periods are very difficult to predict while others are relatively easy. By far the largest errors were the sustained underestimations of the acceleration of inflation in 1973-75 and again in 1978-80. In addition, variations in the difficulty of predicting different variables can be illustrated by examining the forecasts. The results show drastic differences among variables in the forecasters’ ability to outperform a naive standard of comparison. Finally, much of the interest in forecast accuracy stems from the wish to know "Who is the best forecaster?" Even a cursory examination of the information in this article shows that no single forecaster dominates all outliers for all, or even most, of the variables.

Full-text article