Safety Stock Levels for Products with Seasonality or High Variability in Demand

Abstract

     This research considered methods for setting safety stock levels for products with seasonal demand or high variability in demand. The standard method was compared to the seasonal safety stock calculation method and the method proposed in Krupp (1997). Thirty three data sets of actual sales of agro-industrial products were used in the study. The results from the randomized complete block design analysis revealed that the standard method set higher safety stock levels and thus yielded a lower percentage of being out of stock during the replenishment lead time than the other two methods did, whereas the seasonal safety stock calculation method had higher out of stock levels, at the significance level of .05. It could be noticed that the seasonal safety stock calculation method and the method proposed by Krupp seemed to be suitable for many products with fairly apparent seasonality in demand. The seasonal safety stock calculation method seemed to be suitable for the seasonal demand data with no trend or low trend. The method proposed by Krupp seemed to be suitable with data sets with low MAPE, whereas the standard method seemed to be suitable with data sets having high variability in demand. Applying the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), it was found that when the weight w for the safety stock level measure was set low (or in other words the weight for the out of stock level measure was set high), the standard method would be best. When the weight w was set higher, the percentage of products that were suitable with the method proposed by Krupp and the seasonal safety stock calculation method became higher. The percentage of products that were suitable with the method proposed by Krupp was almost always higher than that with the seasonal safety stock calculation method.