Retailer’s Optimal Inventory Policies for Cross-Border E-Commerce

Abstract

This study considers a supply chain formed by multiple suppliers and a cross-border retailer facing a non- stationary demand process. We build a multi-period inventory model with (s, S)-type inventory policy and (s, Q)-type inventory policy. Using this model, demands can be forecasted on the basis of two demand processes, i.e., ARIMA and average demand process. Performances of the two inventory policies, (s, S)-type and (s, Q)-type, are assessed and compared in terms of average delivery time, stock-out frequency, and cost of selling. Through the analysis of 6489 purchase orders of an online shop in Taiwan, covering a period from January 2012 to July 2017, the results present a near-optimal (s, S)-type inventory policy for a cross-border distribution network with multiple suppliers. The model is a synthesis of two components: (i) the inventory policy analysis at a retailer, and (ii) order demand forecasting. We use action research to analyze the performances of inventory models in a cross-border retailer. The results indicate that the semiannual average method using (s, S)-type inventory policy best suits the case company for demand forecasting, as it can decrease the order delivery time from 7.08 days to 0.63 days, and decrease the stock-out frequency from 100.00% to 9.49%. The key contribution of the findings is the seamless integration of the two components to analyze order history data for cross-border supply chains between retailer and suppliers. We anticipate that the research findings may enhance our understanding of inventory control and provide insights into cross-border retailers’ future inventory policies decision.