Assessing the Impact of Temporary Retail Price Discounts Intervals Using SVM Semiparametric Regression

Abstract

Although the marketing literature has found that temporary retail price discounts cause a significant sales increase, little is known about the specific characteristics of deals that influence the magnitude of the sales spike. In this paper, we analyse the impact of the length of temporary retail price discounts periods on the sales increase using scanner-store daily-sales data for two frequently purchased product categories: ground coffee (a storable category) and yogurt (a perishable category).Wedevelop a robust semiparametric regression model based on support vector statistical theory with several previously proposed predictors along with a daily time description. This model also makes it possible to investigate the impact of temporary retail price reductions on own-andcompeting brand sales, observing brand substitution patterns. The results evidence: (1) which days of the promotional period present a higher contribution to the sales spike; (2) the existence of threshold and saturation effects; and (3) that asymmetric cross price effects apply in both categories.