Abstract
We investigate the effect of inaccuracies in the estimation of system parameters and their interaction with two pooling strategies, physical pooling and information pooling. Since managerial attention and resources are often limited, we study using the newsvendor framework whether it is better to estimate system parameters accurately before implementing pooling or vice versa. If the decision maker is not confident about the accuracy of the system parameter estimates, pooling would still be beneficial. Re-optimizing the decision variables under inaccurate parameter estimates may, however, result in higher expected total system cost, which turns out to be particularly sensitive to expected demand. It is, therefore, of paramount importance to be accurate when estimating this parameter before undertaking any further initiative. Finally, information pooling appears to be more robust to errors in parameter estimation than physical pooling.