A behavioural and experimental analysis of retailer order amplification in single-supplier multi-retailer supply chain

This project seeks to better understand how different factors may independently and in combination influence order amplification in single-supplier multi-retailer supply chains. Retailer order amplification often takes place when supply shortages - due to insufficient production capacity, uncertain production yields, or supply chain glitches - lead to fierce retailer competition. The resulting retailer order amplification can lead to problems such as excessive supplier capital investment, inventory gluts, low capacity utilization, and poor service, among others (Armony and Plambeck, 2005; Gonçalves, 2003; Lee et al., 1997a; Sterman, 2000).
Lee et al. (1997a, 1997b) characterize the mechanisms leading to retailer order amplification. Under scarce supply, the supplier rations the allocation of available supply to satisfy retailers’ orders, while at the same time retailers amplify orders in an attempt to secure more units (Lee et al., 1997a, 1997b). Cisco System’s 2001 inventory write-off provides an instructive example. In the summer of 2000, Cisco began to experience shortages of several key components. As delivery delays to customers increased, they started to place orders with multiple distributors to improve their chances of receiving their orders. Because Cisco failed to recognize the magnitude of customers’ order amplification, it maintained its ambitious sales forecast and engaged in strong capacity expansion through long term contracts with its OEMs. When additional capacity became available and delivery delays went back to normal levels, customers began to cancel duplicated orders with distributors, leaving Cisco with significant excess capacity, rigid long-term contracts and a remarkable amount of inventory. By the spring of 2001, Cisco took a US$ 2.2 billion inventory write-off (Byrne and Elgin, 2002).

Despite the fact that retailer order amplification has been identified in the literature for almost a century (Mitchell, 1924), there has been little research analyzing it systematically. For instance, the available research does not explain (a) how different factors may independently or in combination influence retailer order amplification in single-supplier multi-retailer supply chains, or (b) the impact that such factors may have on the upstream supplier demand estimation or capacity investment. To start addressing these gaps, this proposal builds on Armony and Plambeck’s (2005) analytical work on the impact of duplicate orders on upstream suppliers’ demand estimation and capacity investment by extending it in two significant ways. First, we develop a System Dynamics (SD) simulation model encompassing more realistic (and endogenous) decision policies for (i) supplier’s capacity investment, (ii) retailers’ inventory management, and (iii) retailers’ order cancellations. Second, we run a decision-making laboratory experiment based on the developed model to study how human subjects playing the role of retailers make ordering decisions. Based on subjects’ decisions and on available information cues, we econometrically test the results obtained to shed light on the decision rules used by subjects in the retailer role.
The results of the project promise to improve our understanding of the endogenous dynamics leading to retailers’ order amplification decisions as well as the decision-making biases leading retailers and suppliers to engage in such behaviors during supply shortages. Furthermore, the project promises to provide practical implications that may be useful for supply chain managers.