Estimating a Stockkeeping-Unit-Level Brand Choice Model Combining Household Panel Data and Store Data
Published 8/1/2005
Author: Pradeep K. Chintagunta and Jean-Pierre Dubé
View this content
Executive Summary
The marketing literature has addressed the issues of heterogeneity and endogeneity when estimating a choice model with household-level panel data. The former refers to a persistent shock within a panelist’s choice history due to unobserved (to the researcher) taste heterogeneity. The latter refers to a systematic correlation between observed levels of marketing variables, such as shelf prices, and unobserved (to the researcher) causal variables (common shocks) that influence consumer choices. These common shocks are often called “unobserved product characteristics.” Ignoring either of these sources of error can potentially lead to substantial biases in estimated consumer responses to marketing efforts. However, existing methods for resolving such problems have strong data requirements.
Accounting for heterogeneity requires within-panel variation in observed choices and marketing levels with which to identify deviations from the population mean response rates. Accounting for endogeneity requires cross-sectional (across panelist) variation in choices for a given level of marketing effort with which to identify mean valuations for products. These valuations are item- and time-specific fixed-effects that absorb the effect of unobserved product characteristics. Thus, an instrument is required to extract mean response rates to marketing variables from these fixed effects.
When using household data at the stockkeeping unit or the universal product code level, choices for each item in each of the examined time periods cannot be observed. Without such information, it is difficult to control for item- and time period–specific unmeasured characteristics—there is no information on available alternatives during periods during which panelists do not purchase the products. In general, when a product category has many alternatives, each with fairly small shares, the household sample may not contain sufficient choices for each alternative, thus negatively affecting researchers’ ability to control for endogeneity with household data.
Conversely, for those stores from which the panel makes purchases, aggregate store-level data represent the aggregation of purchases by all households visiting the stores. Such data contain the time period–specific, item-level information required to account for endogeneity, as long as each item has some sales in each time period. However, aggregate data does not contain the within-household histories that are used to estimate the distribution of heterogeneity. Similarly, aggregate data does not contain within- and across-household variation in quantity choices that can be used to disentangle brand and quantity choices.
Given the relative merits of household data to estimate the distribution of heterogeneity and store-level data to address the endogeneity problem, the authors propose an integrated estimation procedure that uses the information in both sources. The approach is structural in that it derives the likelihood for both individual and aggregate choices from an economic model of individual consumer demand. The authors provide empirical results from their model using data on the fabric softener market. They extend their approach to situations in which there is variation in purchase quantities chosen by households.
Biography
Pradeep K. Chintagunta is Robert Law Professor of Marketing in the Graduate School of Business at the University of Chicago. He is interested in studying strategic interactions among firms in vertical and horizontal relationships, measuring the effectiveness of marketing activities in pharmaceutical markets, investigating aspects of technology product markets, and analyzing household purchase behavior.
Jean-Pierre Dubé is Associate Professor of Marketing in the Graduate School of Business at the University of Chicago. He received his bachelor’s degree in economics from the University of Toronto and his master’s and doctoral degrees in Economics from Northwestern University. His research applies economic models of empirical industrial organization to study marketing problems. His current areas of research include dynamic oligopoly, price discrimination, competitive advertising, the endogenous formation of market structure, and Internet marketing.
J Marketing Research, Volume 42, Number 3, August 2005
View Table of Contents.