A Note on the Use of Markov Chains in Forecasting Store Choice

Ehrenberg's sweeping criticism of Markov brand switching models [3] highlights many shortcomings of these models for aggregate analysis of consumer behavior. While it has been pointed out that some of his criticisms are not entirely correct [13], one of Ehrenberg's themes is unquestionably valid. The models tend to break down empirically due to violations of important Markovian stability assumptions [14]. A situation in which the assumptions of the model appear less restrictive is short-run forecasting of store choice behavior of individual families.