An Empirical Study of Exactly and Approximately Optimal Inventory Policies

The (s, S) inventory model (infinite horizon with stationary parameters) is explored with a view toward determining the dependency of the optimal policy on the model's parameters and toward evaluating the effectiveness of several numerical approximation methods for obtaining nearly optimal policies. The mode of attack is empirical: We compare the results for 828 cases, which encompass several values for each of the economic parameters, three delivery lags, and three distinct demand probability distributions with varying parameter values. We demonstrate for these cases that close to optimal results can be obtained by applying the Fibonacci search procedure, by using a particular asymptotic approximation, as well as by substituting a certain policy not of the (s, S) form.