Using Support Vector Semiparametric Regression to estimate the effects of pricing on brand substitution.

This paper analyses the sales impact of temporary retail price discount on consumer goods product categories with different perishability rates, and provides some empirical findings regarding how the deal discounts of competing brands affect the sales substitution effects among them. We focus on the cross-price effects by considering both the asymmetric cross-price effect and the neighborhood cross-price effect. To test these effects we use Support Vector Machine-Semiparametric Regression (SVM-SR) and we highlight the benefits of this methodology on the study of promotional effects with aggregate scanner data. The empirical findings confirm the existence of cross-promotional effects of different magnitude, depending on the considered category. In addition, other significant promotional effects are evidenced, such as saturation levels -- especially in the lower-priced brands of the categories -- and a higher weekend sales acceleration effect in the storable category. Our conclusions are of key relevance for retail managers, since developing a superior knowledge on sales substitution effects within the categories carried is critical for designing and developing better pricing, promotional, and category management schemes. [ABSTRACT FROM AUTHOR]