Detection Strategy Of Demand Distribution Change For Newsvendor

The changes in demand distribution significantly influence the performance of inventory systems and supply chains.Both responding too early and too slowly to the changes can lead to inappropriate adjustments,mismatch between supply and demand,and higher inventory costs.Thus,how to construct a detection strategy of demand change that balances between responding too early and too slowly and minimizes inventory costs becomes a theoretical and practical problem.Due to the problem above,we firstly investigate the detection strategy under the assumption that decision maker is rational.Then,we study the order decision behavior of irrational newsvendors who are influenced by multiple irrational factors.Given the transitional probability of demand change,we investigate the detection problem of demand change under the assumptions that demand distribution is constant,normal,uniform,and exponential separately and propose the optimal detection strategies for every case.Further,we apply the analysis of the optimal detection strategies to explore the issues about the trust and information falsify among supply chain members under buyback contracts.Studies show that:when demand distribution is constant,the optimal detection strategy depends on cost parameters only.When demand distribution is uniform,the optimal strategy depends on demand parameters only.When demand distribution is normal or exponential,the optimal strategy depends on both of cost and demand parameters.When the magnitude of demand change increases,the optimal detection strategy for normal distribution case becomes more and more similar to that for constant case,while the optimal strategy for exponential case becomes more and more different to that for constant case.Under a coordinated buyback contract,the supply chain members respond synchronously to demand changes of normal and exponential distribution.Under a uncoordinated buyback contract,the members may tend to falsify demand information.When the demand distribution is uniform,no matter what the cost parameters are,the supply chain members would respond to demand change synchronously and share demand information honestly.Given the prior demand,we study the optimal strategy for detecting the change of normal demand distribution.Based on the probability of appearing the prior demand,we develop a rule to detect the demand change and optimize it by using the hypothesis test and expected utility theory.Studies show that:the optimal detection strategy that based on the prior demand depends on both demand and cost parameters.When the magnitude of a demand change is sufficiently large,the optimal detection strategy coincides with that taking the mean of the means of the demand before and after demand change as the threshold for judging the demand change.And,when deciding to share the prior demand,the supply chain members under coordinated buyback contract are honestly,while the members under uncoordinated buyback contracts might tend to falsify the demand information instead.Manufacturer can adjust the wholesale price or the buyback contract to encourage the retailer to share demand information honestly.Given the transitional probability of demand change and the prior demand,we investigate the optimal strategy for detecting the change of normal distribution.Deriving the probabilities of that both the possible distributions appear in the current period from the transitional probability,viewing the prior demand as a signal from the underlying demand distribution,we optimize the detection rule proposed in the prior chapter and obtain an optimal detection strategy.Compared with the classical forecasting method the exponential smoothing method,the optimal strategy can save more inventory costs.Also,we analyze the impacts of the transitional probability,the magnitude of demand change,excess cost and shortage cost on the optimal detection strategy.And use the analysis results to conduct the decisions of sharing demand information among the supply chain members under buyback contracts.Studies show that:the optimal strategy that relies on the transitional probability and the prior demand is more efficient to save inventory costs than those rely on single type of information.The proposed detection strategy outperforms the exponential smoothing method in sense of saving inventory costs.When the value of transitional probability is very small or very large,the impact of changing the transitional probability on optimal order decision is more dramatic than in other situations.When the magnitude of demand change is very large,it's nearly optimal to take the mean of the means of the demand before and after demand change as the threshold for inferring the demand change.The optimal detection threshold increases with excess cost and decreases with shortage cost.When sharing demand information among supply chains,whether a retailer falsifies the prior demand to the manufacturer depends on the value of transitional probability that manufacturer has obtained before.That means the status of information distribution along the supply chains influence the trust among the supply chain members.Given the mean of demand,we study the newsvendor problem where multiple irrational factors involve.In the heuristic framework of mean anchoring and adjustment,by using expected utility theory and signal detection theory,we combine the stochastic choice and demand chasing into newsvendor problem and develop a decision model for irrational newsvendors.With the proposed model,we investigate the individual difference of the order decisions of irrational newsvendors.Studies show that:the normative newsvendor model is a limit case of the proposed irrational newsvendor model where the parameter of stochastic choice limits to 0.The proposed model verify more empirical findings than the normative newsvendor model.Relative to regret and overconfidence,bounded rationality is a secondary behavior factor that influences order decisions.Managers in practice should pay different attention on bounded rationality and regret.Higher profit margin may lower the quality of order decision.When a decision maker is insensitive to the difference between the prior demand and the order quantity,the actual order quantity excess the expected-profit maximizing quantity with higher probability.Higher computation speed can improve the decision quality,and lead to longer period for adjusting orders.This dissertation investigates both rational and irrational newsvendor problems in contexts of different demand information being available.The core of all the developed methods is to balance the excess cost and shortage cost.Thus,the proposed optimal detection strategy can save more inventory costs than the exponential smoothing methods.With the optimal detection strategies,it's easy to understand the response of a rational newsvendor to demand changes,the order behavior of an irrational newsvendor and the individual order behavior difference when demand uncertainty exists.