Product Level Evidence of the Bullwhip Effect: Measurement, Aggregation and Impact

Abstract

Empirical estimation of the bullwhip effect has several challenges as discussed in prior literature (e.g., Cachon et al. 2007; Bray and Mendleson 2012; Chen and Lee 2012), although the bullwhip effect has been well studied in the modeling papers. We address these empirical challenges using a large-scale dataset collected from a large Chinese supermarket chain, we estimate the bullwhip effect at product level using various methods, analyze different measurements and aggregations of the bullwhip effect, and examine the impact of the bullwhip effect on supply chain performance in terms of inventory and stockouts. We find that: (1) The bullwhip effect measured using shipment variance is lower than that measured using order variance, and the bullwhip effect measured using material flow is greater than that measured using information flow. (2) The aggregated bullwhip effects, by product, by store and by time, are all lower than the disaggregated bullwhip effect. (3) The bullwhip effect is associated with poor supply chain performance measured by elevated inventory ratio and stockouts. We estimate that, on average, a decrease in the bullwhip effect ratio by 1 can translate into inventory cost savings of $26 and stockout reduction of 0.15 days for a product at a store during a year. These results suggest that order pooling can be an effective way to mitigate the bullwhip effect, and mitigating the bullwhip effect can improve supply chain performance and generate sizable benefits in terms of inventory cost savings and service level improvement.