The conditions under which forecasts from expert judgementoutperform traditional quantitative methods are investigated. It isshown that judgement is better than quantitative techniques atestimating the magnitude, onset, and duration of temporary change. Onthe other hand, quantitative methods provide superior performance duringperiods of no change, or constancy, in the data pattern. Thesepropositions were tested on a sample of real business time series. Todemonstrate how these propositions might be implemented, and to assessthe potential value of doing so, a simple rule is tested on when toswitch from quantitative to judgemental forecasts. The results show thatit significantly reduces forecast error. These findings provideoperations managers with some guidelines as to when and when not theyshould intervene in the forecasting process.
International Journal of Operations & Production Management – Emerald Publishing
Published: Jun 1, 1991