Do Generations Matter in Customer Defection?

Abstract

In this paper, we identify the intergenerational effects in customer defection. We develop a mixed (fixed and random) effects discrete-time survival approach to the age-period-generation in order to circumvent the identification and colinearity problems often encountered in the Age-Period-Cohort analyses. We apply our proposed methodology to a panel dataset from an insurance company. We find that there are generation effects in customer defection probability that are independent of the age and period effects. Furthermore, for every equivalent age range, a customer belonging to a younger generation is more likely to sever the relationship with the insurance company than a customer from an older generation. Therefore, customers from the same age range will not express their loyalty to the company in a similar way depending on the generation to which they belong. We also find that the socialization and scarcity hypotheses are irrelevant when customers face high switching costs. There is no calendar effect in customer defection for this company. The changes in loyalty we observed are related to age, generation, and customer lifecycle effects as well as to the control variables in our model.