Profiling banks: how to use cluster analysis with payment system data

Marc Glowka

Abstract

In this paper, payment profiles for participants are identified by applying clustering techniques to TARGET2 data. Payment profiles describe the general payment behavior of participants and are, therefore, relevant for several risks, such as liquidity, credit and operational risks, that payment systems face. After presenting the challenges of applying cluster analysis techniques to payments data, general payment profiles are derived from the pooled results of multiple runs of the k-means clustering algorithm with different similarity measures. Ten different payment profiles with different general intraday payment behaviors were determined in this way. By identifying the deviations of each participant from their profile on a daily basis, the stability of the payment profiles was checked for robustness.