Easy-to-use estimators for waste, on shelf availability and number of orders in a periodic review inventory system with perishable items

Abstract

We consider a periodic review inventory system with perishable items having a fixed case pack size and a shelf life between three and nine days. To assist retailers in reducing food waste, we develop easy-to-use estimators for the waste fraction, the on shelf availability and the number of orders. Since the estimators are easy to use, managers can develop their own logic and tools to support assortment, buying and replenishment decisions for perishable items. We assume stochastic discrete demand, consumers buying the oldest items first and a replenishment policy which considers information on the ages of the inventory on the shelf. The estimators can be used for classification, evaluation and optimization purposes. The average relative cost error when using the easy-to-use estimators to determine the optimal safety buffer as well as the optimal case pack size in the base setting is equal to 0.40 % and its standard deviation is equal to 1.19 %. Relaxing the assumptions in the model leads to average relative cost errors which are still less than 1 % except for two scenario's: when demand shows a very strong week pattern and when a large part of customers do not buy the oldest item first. For those scenario's additional research is needed. While the estimators perform well for classification and cost optimization in most scenario's, the estimators may be far off when used as stand-alone estimators. For those situations we suggest to use a more advanced estimator. Hence we also determine the added value from using more advanced estimators.