Forecast Evolution in the Final Order Problem with Product Returns

Abstract

We consider the final order problem of a spare parts provider who faces customer demands and receives product returns. The returns can be remanufactured for reuse, or can be disposed. There are forecasts for both demands and returns over the remainder of the product life. We consider an evolution of forecasts and examine its influence on remanufacturing policies and costs. We prove the structure of the optimal policy. Using stochastic dynamic programming, we find the following results: (i) In many instances, there exists a flexibility effect of forecast evolution. Under this effect, we place larger final orders and keep more returns than without forecast evolution. This gives us the ability to respond to new information. (ii) There is a pull-away-from-center effect when there is no forecast update. If we can update the forecasts, then the last buy quantity is less sensitive to the costs, i.e. we buy less at lower costs and more at higher costs compared to the last buy without a forecast update. In a numerical study, we find that a consideration of forecast evolution yields significant cost savings, especially if we have time-dependent demands. The largest part of the savings is obtained from updating the demand forecasts; updating the return forecasts has only a small influence.