The effect of correlation between demands on hierarchical forecasting

The forecasting needs for inventory control purposes are hierarchical. For stock keeping units (SKUs) in a product family or a SKU stored across different depot locations, forecasts can be made from the individual series’ history or derived top–down. Many discussions have been found in the literature, but it is not clear under what conditions one approach is better than the other. Correlation between demands has been identified as a very important factor to affect the performance of the two approaches, but there has been much confusion on whether it is positive or negative correlation. This chapter summarises the conflicting discussions in the literature, argues that it is negative correlation that benefits the top–down or grouping approach, and quantifies the effect of correlation through simulation experiments.