Single Period Inventory Problem Under Different Information Settings: A Behavioral Study

Theoretical newsvendor model can help managers solve the single periodinventory problem in whether the demand information is known or unknown.However, previous researches on behavioral newsvendor have shown that subjectshow significant deviate result from normative solution. This research focuses oncomparative study of the newsvendor behavioral biases under different settings. Atthe current stage of the behavioral operation research, this is the first experimentalnewsvendor study that considers the influence of unknown demand information tonewsvendor decision.In this experiment, half of the subjects were set in the known demandinformation setting, while the other half were set in the setting in which underlyingdemand information and lost sales were unobservable. In an experimentalnewsvendor setting, participants have to determine how many products to order for60periods. In this study, the newsvendor games were conducted by110Chinesestudents.Previous experimental newsvendor study suggests that under unknowndemand distribution, decision makers tend to lower their order quantities.Compared with the literature, we find that the effect of “pull-to-center”-thephenomenon which newsvendor average order falls between average demand andoptimal order-is more significant when demand information is limited. We furtheranalyze the behavioral biases (such as the bias of demand chasing, the bias ofadjustment, etc.) in the two information scenarios.Besides, we compare five modelsfor either scenario and find that the EWA-learning model can best fit the behaviorin both scenarios. The results may help managers understand how to improve theirdecisions under different demand information settings.