Toward a One-Number Forecast: Cross-Temporal Hierarchies

The 2022:Q2 issue of Foresight included an article by Simon Clarke entitled "One- Number Forecasting: A Solution for Silo Behavior?" Simon argued in his paper that while the goal of achieving consensus on the forecasts is laudable, forcing consensus through the forecasting process is counterproductive, glossing over the reality that forecasts serve different purposes within organizations, require different horizons, and demand different types of data. There are, in short, better ways to reconcile diverse forecasts. Nikos Kourentzes agrees and offers a comprehensive perspective for doing so: that is, for achieving consensus while enabling individual forecasts across the organization. The challenge, he argues, is creating a unified view of the product hierarchy from which various functional forecasts can be derived in a coherent way. The unified view requires reconciliation not merely across products and locations in a hierarchy, but over time buckets as well; e.g., weekly to monthly. The key construct that we're seeking is the cross-temporal hierarchy. Copyright International Institute of Forecasters, 2022