Executive Summary
As a determinant of consumer behavior and a basis for market segmentation, the concept of household life cycle has been widely accepted and used in both academic and applied marketing research. However, there is considerable disagreement about the two interrelated steps in operationalizing the life-cycle concept: the classification of life stages (i.e., the main types of households in a particular population) and the identification of life paths (i.e., the sequences that households follow through the various life stages). In the extant life-cycle models, households are classified into life stages on the basis of various a priori definitions, reflecting the authors’ own beliefs about the typical compositions and evolutions of households in a population. Although extant life-cycle models postulate specific life paths, they are all based on snapshots of a cross-section of households at a point in time and therefore can not provide any empirical validation regarding how households actually move from one life stage to another. Thus, these models fail to quantify the transition probabilities between life stages, making them incapable of predicting future stages given current stages and of understanding how households evolve through these life stages.
In this study, the authors take advantage of the Panel Study of Income Dynamics (PSID), a nationally representative longitudinal panel of approximately 8,000 U.S. households tracked annually from 1968 through present. The authors develop a hidden Markov model (HMM) in which the stages of the household life cycle are taken as latent, unobservable states and the process that governs the transitions from the states at time t – 1 to the states at time t is assumed to be first-order Markov. The authors fit the HMM model to the Panel Study of Income Dynamics data, thereby using the manifest demographic profiles of the households over time to identify empirically the most commonly observed life stages and life paths in the United States over the past three decades.
The authors apply the HMM results to classify the participants of the Consumer Expenditure Survey (CEX) into life stages, which enables them to study the impact of the household life cycle on lifestyles (as manifest through different patterns of consumption expenditures rather than psychographics). The authors fit a Type-2 Tobit regression model to 35 major categories of expenses and assets obtained from the CEX and find substantial and meaningful differences in expenditure incidences and quantities across the 13 life stages identified by the HMM model.
Based on widely available demographic variables, the parameter estimates of the HMM model offer marketers a valuable life-stage classification scheme that can be readily applied (without any additional model estimation) to classify any other sample of households into the same typology of life stages, following the standard latent class procedure. Marketers can also develop regression models similar to the one used is this article to analyze the CEX data and to predict the annual expenditures of a household in a wide range of categories as a function of life stages, income, and other demographics (e.g., household head ethnicity, education, gender). For example, to target customer acquisition campaigns more effectively, a marketer can first classify prospects into life stages and then determine the expected annual consumption needs of each prospect, using the demographic information available in rented mailing lists or geodemographic databases. The marketer can also apply the results of this study to his or her current customer database to determine the total consumption needs for each customer (and thus the firm’s share of wallet). Finally, the estimated transition probability matrix of the HMM model provides a means to project a household’s expected life paths, which, when combined with a consumption–life-stage regression model, can be used to project a customer’s lifetime potential value.
Biography
Rex Y. Du is Assistant Professor in Marketing in the Terry College of Business at the University of Georgia. He receives his doctoral degree from Duke University. He is a winner of the Alden G. Clayton Award for the 2003 Marketing Science Institute Dissertation Proposal Competition. He has published in Marketing Science and Journal of Marketing Research. His research interests include customer valuation, customer equity management, return on marketing, consumer life cycle, household portfolio and budgetary allocation, and channel management.
Wagner A. Kamakura is Ford Motor Company Professor of Global Marketing in the Fuqua School of Business at Duke University. He holds a PhD in Marketing from the University of Texas at Austin, an MS in Industrial Engineering from the University of Sao Paulo (Brazil), and a BS in Mechanical Engineering from the Technological Aeronautics Institute (Brazil). He is a coauthor of Market Segmentation: Conceptual and Methodological Foundations and has authored and coauthored numerous articles in academic journals. He has served as the editor of Journal of Marketing Research, an area editor of Marketing Science, and an associate editor of Journal of Consumer Research. He is currently a member of the editorial boards of Journal of Marketing Research, International Journal of Research in Marketing, Journal of Marketing, Journal of Retailing, and Marketing Science. His current research interests are in market segmentation and market structure; database marketing; and the modeling of customer satisfaction, retention, and profitability.
J Marketing Research, Volume 43, Number 1, February 2006
View Table of Contents.