Tracing the Effects of Price Promotions across Categories

Raju (1992) argued that a retailer's performance is determined not so much by the sales of any one particular brand, but rather by sales across all brands in a category. While the fast majority of research on the effectiveness of marketing efforts has been at the brand level, retailers understand that marketing programs aimed at enhancing the sales of a particular brand are bound to be less beneficial for a category as a whole than those that take into account the correlations in demand between the different brands. The resources poured into Category Management (CM) in recent years are evidence of the intensified focus by retailers on demand at the category rather than the brand level (Dhar, Hoch, and Kumar 2001). We argue, however, that even this category level focus is likely to be too narrow. Using a similar logic as described above, marketing programs that take into account possible demand correlation across categories are likely to be more potent for the retailer than those that focus on enhancing the performance of a single category. In the current paper, we seek to contribute to the emerging research stream on CM by investigating whether and how price promotions in different categories are likely to influence demand across category boundaries. The effects of promotions across categories have been called 'the essence of merchandising' (Walters 1991) since it has been shown that failing to take demand interdependencies into account may have large profitability implications (Chen et al. 1999). For example, a price promotion can boost sales in a particular category but this sales increase may come at the at the expense of sales in other categories, hurting overall performance (Mulhern and Leone 1991). It is therefore essential for retailers to understand the nature of across-category cannibalization that may result from the use of price promotions. Demand interdependencies can prove beneficial to retailers if they enhance the effectiveness of price promotions across complementary categories. Retailers can take advantage of such complementarity through the development of coordinated cross-category marketing programs. Overall, we suggest that CM is likely to be more effective once dependencies across different categories are better understood (Chib, Seetharaman, and Strijnev 2001; Song and Chintagunta 2001). In addition, our findings are likely to be of interest to manufacturers operating in multiple categories. Currently, a small body of research exists that has investigated the effects of price promotions across categories. Mulhern and Leone (1991) and Walters (1991) found that sales of a brand in one category might be positively affected by promotional activity of a brand in a complementary category. Studies by Chintagunta and Haldar (1998) and Manchanda, Ansari, and Gupta (1999), in contrast, have a category-level focus. Manchanda et al. (1999) find significant effects of category-level price changes and price promotions on consumer purchase incidence probabilities across complementary categories. The study by Chintagunta and Haldar (1998) investigates whether a consumer's purchase timing in one category is affected by purchases made in another category. They find that purchase timing is negatively related across substitute categories (i.e., the purchase in the substitute category is delayed) while a positive effect is found for complementary categories (i.e., the purchase in the complementary category occurs earlier). The goal of the current study is to generate insights regarding the category-level impact of price promotions across different categories. We begin by developing a conceptual framework that describes the components of cross-category effects. Within this framework, hypotheses are derived for potential moderators of these components. We believe that a major contribution of our research lies in uncovering these moderators and, thereby, developing insights that can assist retailers in making their CM efforts more effective. Furthermore, by applying recent time-series techniques on a database covering 79 product categories, our study is intended to yield generalizable findings on the size and frequency with which cross-category price-promotion effects occur. In doing so, we seek to fill a relevant gap in our understanding of promotional impact as called for by Blattberg et al. (1995, p. G125). In addition, we seek to capture the dynamics of demand interdependencies, as called for by Mulhern and Leone (1991). In doing so we extend the work of Chintagunta and Haldar (1998) mentioned above, by deriving a measure of dynamic cross-category price-promotion effectiveness that is able to capture not only changes in purchase timing, but also a wide variety of effects induced by consumer and manufacturer behavior. The methodology we employ allows us to capture these processes and estimate long-run effects (see, for example, Bronnenberg et al. 2000; Dekimpe and Hanssens 1999; Srinivasan et al. 2000). Finally, we extend the work of Chintagunta and Haldar (1998) and Manchanda et al. (1999) by allowing the intra- and inter-category effectiveness of price promotions to differ across brands. The remainder of the paper is organized as follows: In section two we describe the conceptual framework. Section three describes the methodology and the measurement and hypothesis testing procedures. Results are presented in section four. Finally, section five contains conclusions and directions for future research.