This paper builds a computational experiment model in which the manufacturer encourages the group of retailers to increase sale efforts using sales rebate and penalty contracts．We assumed that the retailers have fairness preference and study the impact of external effects among retailers on the incentive effects．The EWA algorithm is applied to the process of the manufacturer adjusting the incentives and penalties dynamically to maximize her own benefits．The aim of this paper is to provide suggestions to manufacturers to make rewards and punishments and use the external effects among retailers reasonably in multi-cycle periods．The results show that: the manufacturer should take measures to minimize the retailers“free rider”behavior，but not to eliminate this behavior completely; the dynamic incentives will be better than the static incentives in multi-cycle incentive periods．Dynamic adjustment of rewards and punishments will greatly affect the level of the average sales efforts of retailers who have been motivated，but cannot affect the evolution trends of sales efforts of retailers who have been punished．