The Effect of Forecast Redundancy on Judgments of a Consensus Forecast's Expected Accuracy

Abstract

This paper examines the effect of redundancy among multiple individual forecasts on a decision maker's assessment of the expected accuracy of a consensus forecast. In the paper, forecast redundancy is defined as the existence of positive correlation among the errors of individual forecasts. Statistical theory indicates that the accuracy of a consensus forecast is a function of two factors: (1) the accuracy of individual forecasts composing the consensus, measured by their forecast error variances, and (2) the correlation among errors of these forecasts. Positive correlation among forecast errors reduces the accuracy of the consensus forecast. In this paper, I address the issue of whether decision makers' expected accuracy judgments reflect the negative impact of forecast redundancy. In addition, tests examine whether expected accuracy judgments are more likely to reflect the effect of individual forecast accuracy than forecast redundancy. Decision makers face two types of expected accuracy judgments when