Abstract
Despite its theoretical and empirical merits, segmentation may fall short of the mark in certain respects, especially with regards to the assumed temporal stability of a particular market segmentation. For example, competitive structures and customer competence and perceived needs have been rapidly changing in the retail financial services market in recent years. The household financial services market also forms the basis for the empirical example in the paper. The two main purposes of the study were (1) to develop a conceptual model of segment stability and (2) to empirically determine how stable a clustering-based behavioral segmentation of the market for household retail financial services in the Netherlands was over a four-year period (1993-1996). To this end seven research hypotheses were formulated concerning segment stability in terms of size and discriminating characteristics such as financial behaviors, psychological variables, demographics and socioeconomic variables. An explanatory model was drawn up and applied in order to validate the segmentation and discriminate among behavioral groups. Finally, the research hypotheses were tested utilizing clustering-based analysis of portfolio behaviors. A five-cluster solution was chosen for the 1993, 1995 and 1996 panels. Statistically significant differences among clusters on theoretically important background variables were found. While segment sizes, typologies and means for background descriptors were quite stable over time, household segment switching was in the order of 40-50 percent for each year, increasing with time. Predictive discriminant analyses based on significant segment descriptors from the explanatory model classified under 40 percent of all households correctly. This was significantly higher than could be expected from pure chance models.