Using forecasts and managerial accounting information to enhance closed-loop supply chain management

This paper develops a generic forecasting framework for product returns that combines concepts used in different disciplines. If more-step ahead forecasts of product returns are required, estimating sales data is necessary. This is accomplished by adopting growth curve models based on the extended Kalman filter. In order to capture the process generating product returns more adequately than in the literature, we propose an adaptive Bayesian approach to forecast future returns. The combination of these two concepts enables us to conduct more-step ahead forecasts. We evaluate the robustness of this approach against Holts approach, a Kalman filter based approach, and the model by Toktay et al. (Manag Sci 46:1412–1426, 2000) for varying degrees of misspecification. In addition, we create a link between forecasting accuracy and the economic value added. This enables the user to choose the economically worthwhile forecasting method that trades-off additional operating costs and savings in working capital. Our theoretical and numerical results indicate that our approach operates on high accuracy even in situations when the underlying assumptions are obviously violated.