Abstract: In the previous research, a first-order autoregressive(AR(1))process was adopted by most researchers to describe the demand process. However, it is difficult to explain the managerial insights of the demand process characteristics, such as the demand correlation coefficient on the bullwhip effect. Our research considers a demand model where the demand depends on its price and the price follows an AR(1)process, we derive the analytical expressions of the bullwhip effect with minimum mean-squared error(MMSE), moving average(MA)and exponential smoothing(ES)techniques and deduce the conditions under which the retailer should choose the best forecasting technique. Results show that: first, the market demand scale does not influence the retailer's choice. Second, when the product price sensitivity coefficient is small, or when the price correlation coefficient is small, the retailer should choose MMSE technique. Third, for products with a large product price sensitivity coefficient and a large price correlation coefficient, the retailer should choose MA technique.

Key words: supply chain management, forecasting technique, AR(1)process, bullwhip effect, price sensitive

摘要： 一阶自回归(AR(1))序列模拟需求过程是传统文献采用的经典模型,然而上述文献关于需求过程参数(如需求自回归系数)对牛鞭效应的影响分析缺乏实践意义,为了更符合企业的实际决策过程,本文建立了需求依赖于价格、而以AR(1)序列模拟价格过程的需求函数模型,分析了最小均方差、移动平均和指数平滑预测下的牛鞭效应,确定了零售商的预测技术选择条件。研究表明:(1)产品市场规模不影响零售商预测技术的选择;(2)当产品价格敏感系数较小或价格自回归系数较小时,零售商应选择最小均方差预测技术;(3)当产品价格敏感系数和价格自回归系数均较大时,零售商应选择移动平均预测技术。

关键词: 供应链管理, 预测技术研究, 一阶自回归, 牛鞭效应, 价格敏感