Psychographic segmentation of Italian bank customers

Traditionally, customer’s ‘character’ was used as one of the main criteria in credit granting process (Three ‘C’s of Credit). Credit scoring (risk models targeted at screening individuals) replaced it with hard information, such as credit history and demographics. Recent studies reignited the interest in character/ personality by showing its relevance in modelling consumer indebtedness, and more general financial behaviour. Unfortunately, most literature in this field is mainly based on self-reported financial measures. We overcome this limitation by analysing a unique dataset of Italian bank customers that contains hard information about credit and investment performance, in addition to soft measures of personality, financial knowledge and preferences for different levels of customer support. The purpose of this work is twofold: first, we aim at detecting if there are different psychographic segments of banking customers based on personality traits and other soft information. We experiment with clustering and different kernels in Kernel Principal Components Analysis. The comparison indicates the presence of different clustered structures in the data but some partitions may be difficult to interpret from a managerial perspective. Second, we analyse the investment and credit performance of the different customer typologies. We comment on the model fit and predictive accuracy of models for different segments.