Modeling Credit Card Share of Wallet: Solving the Incomplete Information Problem

• Published in: Journal of marketing research 2012-10, Vol.49 (5), p.655-669
• Main Authors: CHEN, YUXIN, STECKEL, JOEL H.
• Format: Article
• Language: English
• Subjects: Bank credit cards; Banks; Consumer behavior; Credit cards; Customers; Induced substructures; Information processing; Marketing; Modeling; Parametric models; Princesses; Studies; ZIP codes
• ISSN: 0022-2437; 1547-7193
• DOI: 10.1509/jmr.06.0005

Despite the powerful technologies that have enabled the assembly of transactional databases and the processing of information about individual customers and their buying patterns, database marketers have been limited by what is known as the "incomplete information problem"— that is, marketers have incomplete information about a customer's behavior in the product category of interest. A company's transactional database can only be built from customers' transactions with that company. Any transactions that have been made with other companies are missing. The authors present a modeling approach designed to solve the incomplete information problem. They use transactions conducted with a single supplier in that category to infer consumers' behavior with other suppliers in that category. In particular, armed with prior knowledge of the parametric form of consumer interpurchase time distributions, they uncover elements of the stochastic process that dictates which supplier a consumer chooses on a particular purchase occasion. The authors focus on interpurchase times because they form the core of the incomplete information problem. If a company uses its observed interpurchase times to estimate interpurchase times in the category for a specific consumer, its estimate will be biased upward. Combined with the model developed in this study, a familiar analysis of transaction profitability can be used to build a new type of lifetime value: lifetime category value of the customer.