Demand forecasts with judgement bias in a newsvendor problem

We explore the impact of judgement bias on demand forecast accuracy and profit to identify the driving force of decision-makers to be biased. We study the accuracy-maximising forecasts and the profit-maximising forecasts, which results in the least error of demand forecasts and minimum deviation from the optimal order decision, respectively. Under the assumption that period-to-period demand is independent over time, we find that both types of forecasts are biased. It implies that a newsvendor has the motivation to be biased to obtain either a more accurate demand forecast or a higher profit. Moreover, the decision error under the accuracy-maximising forecasts can be lower than that under unbiased demand forecasts and be bounded by twice of the error under profit-maximising forecasts. It suggests that the biased accuracy-maximising forecasts can perform satisfactorily in both forecasts and decisions. We further relax the assumption of independent demand by considering correlated and trended demand processes, and show the robustness of the positive impact of judgement bias. We then propose a method to solve pure data-driven newsvendor problem and examine its performance with empirical evidence. Our paper contributes to the literature on behavioural operations management by investigating the rationality of judgement bias and its implications.