Long-term Forecasting of Seasonal Goods using De-biased Expert Judgment: Beyond the Pandemic

The problem of forecasting demand for seasonal products became even more challenging for retailers amid the Covid-19 pandemic. They are now faced with the difficult task of forecasting demand in the “new normal”. On the one hand, the pandemic’s future trajectory is hard to predict. On the other hand, an increase in supply constraints forces retailers to confirm orders several seasons in advance. Given this increased uncertainty in future demand, we develop a forecasting model that combines de-biased expert judgment to utilize their knowledge of shifts in consumer behavior, the effects of inflation and other global factors on product demand, and statistical methods well suited for extrapolating repeated patterns from the past. Using data from a premium bicycle manufacturer, we show that accurate demand forecasts for the next three years can be obtained by integrating experts’ estimates of the category growth rate with a seasonal decomposition of the pre-pandemic demand curve. Our method yields a forecast error (MAPE) of 18.05%, on average, on data from early 2022. Moreover, our integrated human-algorithm forecasts perform better than statistical time-series forecasts, indicating that experts play a decisive role in predicting demand for seasonal products during and after the pandemic.