Lateral transshipment could be used as an effective and fast replenishment policy in inventory systems. However, in decentralized supply chain systems, imperfect transshipment planning can make total costs increasingly. In this paper, we consider a single-item, multi-location, two-echelon supply chain system with lateral transshipment existence. A continuous review order-up-to policy is assumed for the inventory control of the item. Two objectives are investigated: total costs of the supplier, total costs of the retailers' coalition. A stochastic multi-objective decision model is established to describe systems behavior. Based on stochastic quasi gradient algorithm (SQGA) and genetic algorithm (GA), a solution procedure is developed to obtain the with-probability-1 (w.p.1) optimal order-up-to levels. An extensive numerical experiment shows the model and algorithm is very efficient. We also test cost parameters sensitivity to the model.