Executive Summary
The extant literature on retail category pricing optimization concentrates on grocery retailing. In contrast, this article focuses on the problem of determining profit-improving, store-level prices of failure-related “hard-part” product categories at a U.S. specialty automotive part retailer with 3400 stores. Key institutional differences between automotive hard-part and grocery retailing include the following: No syndicated data are available for hard parts; each subclass of a hard-part category contains variants that are ordered by quality (typically, “Good,” “Better,” and “Best”); there is intra- but no intersubclass competition; market shares of variants and their prices are not positively correlated within a subclass; a consumer enters the market only infrequently; and at a purchase occasion, a consumer buys one and only one variant within a subclass and one and only one unit of that variant. Using two years of weekly sales histories from 800 stores, the authors develop store-level demand models for 23 subclasses of a hard part and employ these with available product cost data to set prices of variants of each subclass at each store that better satisfy demand and increase profit. The model-recommended prices for ten subclasses are tested in a field experiment involving 500 stores, leading to a projection of an annual increase of more than $610,000 in the retailer’s profit from these ten subclasses if the new prices are applied at all stores. Subsequently, the retailer has decided to roll out the proposed modeling approach to a second hard-part category in the first quarter of 2006.
Taking advantage of the rare access to a national, store-level database and product cost data, the empirical analysis also yields new insights into (1) asymmetric price competition across quality variants, (2) deviations of actual from optimal prices, and (3) the impact of observable store characteristics on estimated price response parameters of demand that run counter to previous grocery retailing-based findings. Specifically, the empirical findings include the following:
- There is evidence of asymmetric cross-price effects, but opposite to the case in grocery retailing, the effect of a percentage price change in Better or Good variants on sales of the Best variant is greater than the effect of a percentage change in the price of the Best variant on sales of the Good or Better variants.
- There is evidence of neighborhood price effects; that is, variants that are closer to each other in price (and quality) have larger cross-price effects than those that are priced farther apart; however, counter to what has been observed in grocery retailing, this research finds that the impact of a change in Good’s price on Better is stronger than the impact of Best’s price on Better.
- Unlike grocery products, for which optimal prices are typically higher than actual prices, the current study finds in the case of AAR (the automotive retailer under study) that optimal prices are higher than actual prices for some product subclasses but lower for others.
- Effects of consumer demographics or store characteristics are more likely to generalize across grocery product categories than automotive hard-part categories, because in the latter case, one product subclass effectively serves a different automobile market from another.
Bigraphy
Murali K. Mantrala is Sam. M. Walton Distinguished Professor of Marketing in the College of Business at University of Missouri–Columbia. Murali is interested in investigating and developing solutions to marketing decision problems, especially in the domains of retail category management, sales force incentives planning, and pharmaceuticals marketing. Some of his related work has previously appeared in Journal of Marketing Research, Marketing Science, Journal of Marketing, and Interfaces. He serves on the editorial board of Marketing Science and is a coeditor of the book Retailing in the 21st Century: Current and Emerging Trends (Springer 2006).
P.B. Seetharaman is Associate Professor of Marketing at Rice University. Seethu’s research primarily deals with the application of econometric models to marketing data, with a recent interest in the application of game-theoretic models to understanding strategic marketing interactions among firms. Seethu’s article in the collaborative research special issue of Journal of Marketing Research is a special tribute to the late Dick Wittink who had his research eye focused on real business problems rather than on econometric problems of little practical consequence.
Rajeeve Kaul is Director of Price and Product Optimization at AutoZone, a Fortune 300 automotive aftermarket retailer. His education includes MS and MBA degrees from University of Massachusetts, Amherst. His role at AutoZone is to develop next-generation solutions and provide recommendations for improving product assortment and pricing strategy, thus improving return on inventory investment. His prior job experiences includes GE, for which he developed six-sigma and cost-of-failure models, contributing significantly to business profitability, and Advanta, for which he created pricing decision support solutions for the mortgage banking industry.
Srinath Gopalakrishna is an Associate Professor of Marketing in the College of Business at the University of Missouri–Columbia. Srinath is interested in the application of quantitative modeling techniques to marketing problems, especially in the area of business-to-business marketing and sales management. His research has examined the effectiveness and appropriate deployment of business-to-business marketing communications, such as advertising, direct mail, and trade shows, in conjunction with personal selling. He is also interested in measuring the economic returns from marketing expenditures and studying the effectiveness of sales incentive programs. His research has been sponsored by the Center for Exhibition Industry Research, the Society for Incentive Travel Executives, and other organizations. His work has appeared in Marketing Science, Journal of Marketing Research, Journal of Marketing, and other journals.
Antonie Stam is Leggett & Platt Distinguished Professor of Management Information Systems in the Management Department at the University of Missouri. Professor Stam has served in visiting professor and research scientist roles in Belgium, Austria, Finland, and France, and he has consulted with companies and organizations in the United States, China, and Finland. His primary research interests include information systems, decision support systems, applied artificial intelligence, multicriteria decision making, and applied statistics. He has published in journals such as Management Science, Decision Sciences, Journal of the American Statistical Association, Operations Research, Public Opinion Quarterly, Multivariate Behavioral Research, International Journal of Production Research, and others.
J Marketing Research, Volume 43, Number 4, November 2006
View Table of Contents.