Redundancy of Centrality Measures in Financial Market Infrastructures

Abstract

The concept of centrality has been widely used to monitor systems with a network structure because it allows identifying their most influential participants. But this monitoring task can be difficult if the number of system participants is considerably large or if the wide variety of centrality measures currently available produce non-coincident (or mixed) signals. This document uses principal component analysis to evaluate a set of centrality measures calculated for the financial institutions that participate in four financial market infrastructures of Colombia. The results obtained are used to construct general indices of centrality, using the strongest measures of centrality as inputs, and leaving aside those considered redundant.