Exploring mechanisms of consumer behavior

Analysis of extensive microdata

Abstract

In a developed, excess demand economy, simply providing high quality products at reasonable prices is not enough. Consumers cannot be expected to select a specific supplier’s goods and services if that supplier does not understand the consumer (customer) or respond to their specific needs, and provides products and services that exceed consumer needs. This paper explores the mechanisms of consumer behavior that occur in brand selection by examining large amounts of consumer-level data, particularly historical data of household purchases that was collected at supermarkets. After an initial analysis with a brand selection model, based on the stochastic utility maximization process of consumer behavior theory, the data is examined with a nonparametric model that eliminates theory and assumptions as much as possible. Case studies are then used to introduce a model extrapolated from the results in two beneficial and practical directions. One is a semi-parametric model that seeks nonlinear reaction to effective consumer-marketing activities, particularly pricing and discounts. The other is a hierarchical Bayes model that incorporates effective individual parameters, customized by the consumer, into marketing activities, such as customer relationship management.