The anchor and adjustment heuristic in time-series forecasting

This paper explores the relevance of the anchor and adjustment heuristic in judgemental time-series extrapolation. Using a database of real time series it examines the proposition that people anchor on the last history value of the series and make insufficient adjustments from it in making their forecast. In contrast to studies in behavioural decision making, analysis shows that the anchor and adjustment heuristic does not describe the behaviour of time-series forecasters. Adjustments from the anchor are often excessive, not insufficient. A number of possible explanations for this exceptional finding are explored.