Abstract: When ordering under uncertainty of demand distribution, a newsvendor may make wrong judgments of demand distribution, which raises expected inventory costs. How to identify the demand distribution and make right inventory decisions is the main concern of this paper. Assume that the demand follows two different types of normal distribution, which have the same variance but different means. By introducing signal detection theory, an optimal ordering strategy is proposed based on historical demand information and nontraditional demand information to minimize inventory costs for the newsvendor, and it is compared with the intuition rule merely based on the probability of demand distribution. The results show that, excluding the two extreme cases where the probability of demand distribution is either too large or too small, the proposed optimal strategy is better than the intuition rule in controlling inventory costs. It means that with few historical demand information, the newsvendor can effectively adjust the result of demand distribution detection and thus improve his order decisions. Also, the managerial implications of this study in relation to the integration of historical demand information and nontraditional demand information and the exchange of demand information are discussed.

Key words: newsvendor model, demand distribution detection, ordering decision, expected inventory costs