Production, Manufacturing and Logistics
Inventory control with product returns: The impact of imperfect information

https://doi.org/10.1016/j.ejor.2007.11.063Get rights and content

Abstract

Product returns are characterized by considerable uncertainty on time and quantity. In the literature on inventory management for product return environments best forecasts of future returns are associated with methods that use the most information regarding product return history. In practice, however, data is often scarce and unreliable, while forecasts based on historical data, reliable or not, are never perfect. In this paper we therefore investigate the impact of imperfect information with respect to the return process on inventory management performance. We show that in the case of imperfect information the most informed method does not necessarily lead to best performance. The results have relevant implications regarding investments in product return information systems.

Introduction

The value of information has regained interest due to the upcoming of advanced information technology, e-commerce, and increasing complexity of supply chains (Banker and Kauffman, 2004, Huang et al., 2003, Sahin and Robinson, 2002). Information is undoubtedly viewed as a valuable commodity for the management and coordination of the supply chain.

Product returns seriously complicate all processes in the supply chain, in particular inventory decisions (Fleischmann et al., 1997). Return flows are often characterized by a considerable uncertainty regarding timing and quantity. If one could know exactly how much is going to be returned and when, one would certainly benefit from incorporating this perfect information in the management of production and inventory management.

Product returns appear virtually in all sorts of industries (De Brito and Dekker, 2004). Return intensities are relatively high for distant sellers, remanufacturing environments and for distribution items. Mostard and Teunter (2006) report that for catalogue retailers return rates on fashion items “are generally around 35–40%, but can be as high as 75%” of demand. Rogers et al. (2002) mention that returns on online selling can be as high as 40%. Toktay et al. (2000) reports return intensities of 50% for Kodak reusable cameras, and for car part remanufacturing return intensities can be very close to the demand intensity (Van der Laan, 1997, Driesch et al., 2005). Similar values are reported for distribution items: 31–63% for crates and bins in the food retail industry (De Koster et al., 2002), 97% for reusable Coca-Cola soda bottles (Goh and Varaprasad, 1986). Kelle and Silver (1989) use return intensities of 40–90% in their study which are based on real data on returnable containers. Return intensities in traditional retailing tend to be much lower (around 2–4%), but there are exceptions such as in the publishing industry (41%, Soto Zuluaga, 2006) and in the leasing business.

Many of the standard management systems do not take into account product returns (Kokkinaki et al., 2004). Investments can be made to collect information on product returns. However, ambiguous information, conflicting evidence or even abundance of information, may increase uncertainty rather than reduce it (Zimmermann, 2000). There are many examples of large companies, such as Goodyear, Cisco, Selectron, Sainsbury’s and Nike, that failed after implementing and relying on multi-million dollar advanced data systems (see Worthern, 2003, Lee et al., 2004, Rigby, 2005). To decide on the technology and on the type and amount of information to collect is a very sensitive issue. For instance RFID tags seem very promising with respect to tracking products in the market (Klausner et al., 1999) but how networked RFID-based information sharing can be implemented successfully is still a challenge (see Parkilad and McFarlane, 2004).

The question remains: what is information worth? As confirmed by Ketzenberg et al. (2006), there is not so much literature dealing with the value of information in settings with product returns. Furthermore, the existent literature on the value of information for production and inventory control with product returns assumes ‘perfect’ information, i.e. that the return distribution can really be observed (Kelle and Silver, 1989, Toktay et al., 2000). However, in practice the return distribution cannot be observed. In the best case, we can observe and record when sales, and returns occur, which we can use to estimate the timing of future returns. As put by Chen et al. (2000), even with “complete knowledge of the observed customer demands”, one “must still estimate the mean and the variance of demand”, and that it is unknown. The same applies even more so for returns.

Moreover, the acquisition of accurate data in real life is problematic, even with advanced information systems. Well-regarded companies with point-of-sale (POS) information linked to their inventory systems have their stock-keeping-unit quantities off as much as one-third of the times (Raman et al., 2001). Cattani and Hausman (2000) prove that regular information updates may even lead to less accurate forecasts 30–50% of the time. In addition, return data may be recorded, but only in aggregated form, not linked to the original demand, or not timely available. But often return data is not collected at all.

Summarizing, the data that we need to use to estimate future returns are often not as accurate, not as complete and not as timely as we would hope for. Therefore, besides knowing the value of perfect information, it is at least as important to know the impact of imperfect information.

In this paper, we focus on the impact of imperfect information on inventory control with product returns. Its contribution is fourfold: (i) we analyze four methods to forecast the future returns; (ii) analytically we show, in the case of imperfect information, that the method that uses the most information does not necessarily have the best forecasting performance; and (iii) we show that the impact of imperfect information on inventory related costs can be quite large, even for small errors in parameter estimates.

The remainder of the paper is structured as follows. Next, we give an overview of the relevant literature on the value of information for environments with product returns. Then we introduce an average cost inventory model and the four forecasting methods (Section 3). We investigate forecasting performance through an analytical study in Section 4 and inventory cost performance in Section 5 through a simulation study. In Section 6 we discuss the managerial implications and give recommendations for future research.

Section snippets

Related literature

The literature dealing with product returns has been growing fast in the last years. This literature falls under the general umbrella of closed-loop supply chain management (see Rogers and Tibben-Lembke, 1999, Guide and Van Wassenhove, 2003, Dekker et al., 2004). Inventory Management has received ample attention (see e.g. Van der Laan et al., 1999, Inderfurth et al., 2001, Fleischmann et al., 2002). For a recent review see Van der Laan et al. (2004). However, there are few articles that

The model

We consider a single product, single echelon, discrete time inventory system as defined in Kelle and Silver (1989). We introduce the main notation used in the remainder of this paper in Table 1.

Each unit sold returns with probability p according to some return distribution. The order lead time is L periods. Demands that cannot be satisfied immediately are fully backordered. At the end of each period, overstocks are charged with a holding cost $ h per product, whereas backorders are penalized

Forecasting performance

In this section, we analyze the relative forecasting performance of the exact Methods B and D given imperfect information. Methods A and C contain approximations that preclude an exact analysis. Moreover, Method A is a rather naive forecasting method that we do not expect to perform very well in general (this will be confirmed in Section 5) and the performance of Method C tends to be very close to that of Method D (Kelle and Silver, 1989).

Since, given perfect information, Method D is expected

Cost performance

In order to quantify the impact of imperfect information on inventory related cost performance we conducted a simulation study. The experiments are based on the inventory system that was introduced in Section 3 and are conducted in the following manner.

Summary and discussion

In this paper we investigated the impact of imperfect information on performance with respect to inventory related costs. We analyzed four methods to forecast lead time net demand as proposed by Kelle and Silver (1989). Method A only uses the expectation and variance of demand plus the overall return probability. Methods BD use the same information regarding demands, but different levels of information with respect to returns: Method B requires the return distribution, Method C also uses a

Acknowledgement

The research presented in this paper was supported by the UK–Netherlands Partnership Programme in Science of the NWO/British council. We thank Michael Ketzenberg and Edward Silver for their helpful comments on an earlier draft of the paper.

References (38)

  • F. Chen et al.

    Quantifying the bullwhip effect in a simple supply chain: The impact of forecasting, lead times, and information

    Management Science

    (2000)
  • M.P. De Brito et al.

    A framework for reverse logistics

  • M.B.M. De Koster et al.

    How to organise return handling: An exploratory study with nine retailer warehouses?

    International Journal of Retail and Distribution Management

    (2002)
  • H. Driesch et al.

    Recovery of car engines: the Mercedes-Benz case

  • G. Ferrer et al.

    Value of information in remanufacturing complex products

    IIE Transactions

    (2004)
  • T.N. Goh et al.

    A statistical methodology for the analysis of the life-cycle of reusable containers

    IEE Transactions

    (1986)
  • G. Huang et al.

    The impacts of sharing production information on supply chain dynamics: A review of the literature

    International Journal of Production Research

    (2003)
  • Cited by (88)

    • Forecasting: theory and practice

      2022, International Journal of Forecasting
      Citation Excerpt :

      This lagged relationship between sales and returns is key to the effective characterisation of the returns process. Despite the increasing importance of circular economy and research on closed loop supply chains, returns forecasting has not received sufficient attention in the academic literature (notable contributions in this area include Clottey, Benton Jr., & Srivastava, 2012; de Brito & van der Laan, 2009; Goh & Varaprasad, 1986; Toktay, 2003; Toktay, Wein, & Zenios, 2000). The seminal work by Kelle and Silver (1989) offers a useful framework to forecasting that is based on the degree of available information about the relationship between demand and returns.

    View all citing articles on Scopus
    View full text