Excess loyalty - how often does it occur?

Abstract

In many fast moving consumer goods (FMCG) markets small brands suffer twice in that they have less customers who buy them less often. This pattern is well known as “Double Jeopardy” and is predicted by a popular purchase model - the NBD-Dirichlet (Goodhardt, Ehrenberg and Chatfield, 1984). This model provides loyalty benchmarks for competing brands in a product category, but Fader and Schmittlein (1993) found that nearly 78% highshare brands in a Japanese dataset and 68% of those in a US dataset have behavioural loyalty in excess of Dirichlet predictions. This paper is intended to replicate and extend Fader and Schmittlein’s result (1993) and Habel et al (2005) by using two methods - the double jeopardy relationship and a regression of polarization to market share.