An econometric approach to test competitive market structures

Pallassana Krishnan Kannan, Purdue University

Abstract

Identifying the structure of competition is important in the formulation of many marketing strategies such as new product introduction, product repositioning, advertising, and pricing. Prior research in the area of competitive market structure has followed two distinct directions. One line of research describes the "aggregate nature of competition based on aggregate measures of the market." However, aggregate measures often give little insight into individual consumer decision processes, which is crucial in understanding consumer buying behavior and predicting variations in their responses to marketing programs. The second line of research considers the decision processes that an individual consumer uses to make product choices. This research usually involves a micro-level analysis of individual decision making process. However, the results of such research often are not readily applicable for marketing decision making. The aggregation of the implications of the individual level models over the total population of consumers is difficult and beset with problems. This dissertation attempts to solve some of the problems identified above. To deal with heterogeneity in individual choice, our model incorporates consumer choice history through consumer specific loyalty variables. To deal with heterogeneity in market structure, the market is segmented to identify groups of consumers homogeneous with respect to the choice structure characterizing their decision processes. The existence of multiple structures in the market is explicitly incorporated in the model as a hypothesis to be tested. Using nested multinomial logit models at the segment level, our methodology determines groups of consumers who are homogeneous in the structure characterizing their choice processes and examines the effect of marketing mix variables and consumer loyalties on competition within each segment. Based on the information available on consumer purchase history, heuristic algorithms are developed to perform segmentation and estimation efficiently. Tests are proposed to confirm or reject the hypothesis of multiple structures in the market. Empirical analysis of two different consumer scanner panel data on coffee reveals that it is possible to identify with significantly different market structures.

Degree

Ph.D.

Advisors

Wright, Purdue University.

Subject Area

Management

COinS