10. Conclusions
VMI is a well-established supply strategy that has found favour in a number of market sectors. In many cases this has occurred in response to a feeling by the retailer that it would be a good thing to delegate further responsibility to the vendor. As a concept it can be dated back to the classical contribution of Magee (1958). He first raised the issue of which player should control the supply pipeline. It is intuitively obvious that better sight and hence understanding of both information flow and material flow should lead to better business performance. In particular, it is possible to “condense” the pipeline so that its behaviour approaches that of a single echelon. This is achieved by using a constant (that is one that does not change over time) re-order point at the distributor. As we have demonstrated via a VMI simulation model this is indeed the case.
We have compared the bullwhip performance of a number of VMI supply chains with two-level supply chains. In all cases there is substantial reduction in bullwhip (typically halving the effect). This is true irrespective of the bullwhip measure used. In the paper we have concentrated on the two measures of peak order rate to a step input (a “rich picture” approach) and order rate variance. The latter is widely used in industry. From our perspective it also has the advantage of being predictable from system noise bandwidth. Under certain circumstances this is amenable to an analytic solution, which we have developed but not exploited here. In other cases it is possible to take substantial computational short cuts to calculate variance and availability. ITAE has been used as a composite measure of inventory dynamics. Here VMI also offers a substantial improvement in performance.
Finally, as Berry and Towill (1995) have shown, managers need to be aware that in practice there are potentially two Forrester Effects. The first (demand amplification) is universally known. But the second (rogue seasonality) can equally likely be induced by the system dynamics. Hence a periodicity may appear in the ordering waveforms that is not present in the marketplace demand. Our simulations have shown that particularly in the stock records rogue seasonality is induced by the traditional supply chain in response to typical retail demand waveforms. In contrast this is observed to be less of a problem with the VMI system.