Demand forecast accuracy and performance of inventory policies under multi-level rolling schedule environments

Abstract

Our incentive is to assess the impact of demand forecast errors on the cost performance of several lot-sizing techniques in a multi-level context. Unpredicted changes in demand keep on plaguing consumer product companies. However, efforts to improve demand forecast accuracy may not be rewarded if lot-sizing techniques perform equally badly as soon as forecast errors affect the demand. With an extensive simulation study we show that it is always worth decreasing the error magnitude: the performance of all techniques improves when the error level is decreased. But the relationship between cost improvement and error level is not linear as bigger cost reductions are obtained when the error decrease is applied to an initial value of error that is moderate. This means that increasing forecast accuracy is more profitable for companies that already have more accurate forecasts than for those who face inadequate forecasts. Although the presence or absence of forecast errors tends to matter more than the error level itself, we show that lot-sizing rules exhibit significant differences in their behavior as the level of error is augmented. This paper also provides a clear description of the rolling procedure when applied to general product structures, demand with forecast errors within the forecast window and positive lead times.