Consider Regret And Feedback Newsboy Decision-making Bias

This paper establishes a new newsvendor problem model by considering decision maker's regret aversion preference.Based on the present researches, this research proposes the the models considering anticipated regret and experienced regret through integrating the regret theory and newsvendor decision model. The optimal solution of the anticipated regret is closer to the mean of demand distribution than the expected revenue model, so provide theoretical explanation for newsvendor decision bias. And the experienced regret model consider the effect of previous decision on future decision and provide explanation for the variety of order quantities. According to the models proposed, the paper also put forward three heuristic methods, compare the three models to heuristic from Schweitzer and Cachon(2000) in the data analysis section, and find the models proposed in the paper fit the data well.This research simulate the newsvendor inventory decision using a inventory management experiment, manipulate the feedback style in the experiment, then test the proposed heuristics and investigate difference between the two conditions that feedback demand and sale respectively.Main conclusions of the experiment are listed as follows:(1)Pull-to-Center Effect exit in the partial feedback condition and total feedback condition, and is graver for low profit products contrast to high profit products. The average of order quantity in partial feedback condition is lower than total feedback condition, this to say that for the low profit products the newsvendor's order is more closer to the optimal quantity and for high profit products the newsvendor's order deviate from the optimal quantity more farther. The standard deviation of data from partial feedback condition is lower than the total feedback condition, showing that the variety in partial feedback condition is smaller.(2) About models fitting, for the low profit products, the best fitting model is anchor model, the second better models include experiential regret model, anticipated regret model and minimize ex-post inventory error model, and the worst fitting model is demand chasing model and bilateral directions demand chasing model; while for the high profit products, the best fitting models are experiential regret model, anticipated regret model and minimize ex-post inventory error model, the second better fitting models are demand chasing model and bilateral directions demand chasing model, the worst fitting model is anchoring model.(3)About the estimated parameters, for the low profit products the regret parameters estimated with partial feedback data are all lower than the counterpart using total feedback data, indicating the regret effect from counter-factual thinking in partial feedback condition is less salient contrast to total feedback condition; while for the high profit products, there don't exist significant similar pattern. In one feedback condition, partial or total, the stockout regret parameter is always lower than the waste regret parameter, indicating the regret from products left over exert more influence on newsvendor's decision.(4)About the learning in newsvendor decision process, for the high profit products, learning effect push the newsvendor's order toward optimal order quantity, and for the low profit products, in the total feedback condition there exist insignificant adjust learning, which lower the order variety, and there is no positive learning effect in other conditions, which show learning not promote decision performance greatly and that the decision bias are obstinate in newsvendor problem.