Design Features of Forecasting Systems Involving Management Judgments

Abstract

A set of forecasting system design features is proposed to detect, measure, and reduce bias in forecasts and to determine the need for retention of forecasting methods used by an organization. The system is flexible enough to be incorporated into a variety of forecasting environments as part of a marketing control system. Use is made of several statistics that measure forms of forecasting bias including lack of time-series pattern recognition and over-optimism (over-pessimism). The system also uses a criterion for retention of a forecasting method that is based on composite forecasting model procedures. An empirical application illustrates the advantages of the proposed design features in correcting management judgment forecast methods which are prone to bias and in assessing the need for retention of forecasting methods.