Forecasting product returns in closed-loop supply chains

Purpose - The purpose of this paper is to provide a generic forecasting approach for predicting product returns in closed-loop supply chains. Design/methodology/approach - The approach is based on Bayesian estimation techniques. It permits to forecast product returns on the basis of fewer restrictions than existing approaches in CLSC literature. A numerical example demonstrates the application of the proposed approach using return times drawn from a Poisson distribution. Findings - The Bayesian estimation approach provides at least 50 percent higher accuracy in terms of error measures compared to traditional methods in all scenarios examined in the empirical part. Hence, more precise results can be obtained when predicting product returns. Research limitations/implications - The flexibility of the proposed approach allows for numerous applications in the field of CLSC research. Areas that depend on the results from a forecasting system, such as inventory management, can embed our estimation procedure in order to reduce safety stocks. Further research should address the incorporation of the quality of returned products and its impact on the actual utilizable amount of product returns. Originality/value - The generic character of the proposed forecasting approach leaves degrees of freedom to the user when adapting it to a specific problem. This adaptability is enabled by the following features: first, an arbitrary function is allowed for capturing the customers' demand. Second, the stochastic timeframe between sale and product return may follow an arbitrary distribution. Third, by adjusting two parameters finite as well as infinite planning horizons can be incorporated. Fourth, no assumptions regarding the joint distribution of product returns are necessary.