Fildes, Robert Alan and Goodwin, Paul and Onkal, Dilek (2016) Use and misuse of information in supply chain forecasting of promotion events. Working Paper. Department of Management Science, Lancaster University, Lancaster.
Use_and_misuse_of_information_Working_Paper_June_2016.4.pdf
Download (724kB)
Abstract
Demand forecasting is a critical component of sales and operations planning (S&OP) and is pivotal in supporting inventory and production planning in supply chains. Because of their relative infrequency the effects of sales promotions can be particularly difficult to forecast - yet these are events where production and inventory planners need clear guidance on the probable uplifts in demand. A widely-documented practice involves judgmentally adjusting a baseline statistical forecast on receipt of shared information from sales, marketing and logistics. However, much of this information will either have no predictive value in estimating demand uplift resulting from the promotion or its predictive diagnosticity will be unknown. Theoretical arguments on ‘system neglect’ and ‘base rate discounting’ suggest that the provision of information with no or unknown diagnosticity would lead to the forecasters being distracted from the underlying base-rate uplift with deleterious effects on forecast accuracy. This study investigates this possibility when forecasters made judgmental adjustments to forecasts via a forecasting support system (FSS) in advance of forthcoming sales promotions. In experiments forecasters were provided with the mean rate of sales uplift achieved through promotions (the base rate), and a baseline statistical forecast, together with both quantitative and qualitative information relating to a range of products that were due to be promoted. The results revealed that forecasters were distracted from the base rate, misinterpreting the diverse information available to them, and this led to underestimates of the uplift achieved by the promotions. By extending earlier findings from field observation to a representative experimental setting, these findings have important implications for the quality of inventory decisions, for the design of organizational S&OP processes, and for the implementation of the FSSs that such processes rely on.