Increased bullwhip in retail: A side effect of improving forecast accuracy with more data?

Can there be side effects of improved forecast accuracy? In this study of the Belgian food retailer Colruyt Group, we show how adding explanatory variables (such as promotions, weather forecasts, national events, etc.) increases forecast accuracy compared to methods using only historical sales data. Furthermore, when using these sales forecasts to determine inventory levels and order decisions in a numerical experiment, we see that these more accurate forecasts require less inventory to maintain a target service level, indicating that more accurate predictions may reduce stockouts and operational costs related to high inventories. These are expected findings. We also found the use of explanatory variables makes the sales forecasts (and consequently the replenishment) more responsive towards changes in customer demand patterns. This creates a higher bullwhip effect regarding the variability of the supermarket’s replenishment orders -- a less desirable outcome of more accurate forecasting using explanatory variables