Forecasting and scenario planning in the steel packaging industry: case of tinplate petfood cans

Abstract

Modelling national and global steel markets using macroeconomic variables provides a workable basis for forecasting aggregate steel consumption. At the market sector level, paucity of data and distortion hinder equivalent headway. Neural networks are applied to this problem with results that enhance those of conventional econometric models, but are nevertheless variable. A forecasting and scenario planning methodology is developed to manage the variability in the results of market models, and to manage the uncertainty remaining in the market due to major trends and events that are the realm of judgmental forecasting not statistical models. The result is a scenario planning tool that allows the business decision maker to effectively exploit the information available on the petfood sector of the metal packaging market.