Robust Clustering of Banks in Argentina

VARGAS, JOSE M.

Revista de Economía y Estadística

Facultad de Ciencias Económicas, UNC

Lugar: Córdoba; Año: 2019

0034-8066

bstractThe purpose of this paper is to classify and characterize 64 banks,active as of 2010 in Argentina, by means of robust techniques used oninformation gathered during the period 2001-2010. Based on the strategycriteria established in Wang, 2007 and Werbin, 2010, seven variables wereselected. In agreement with bank theory, four ?natural? clusters were ob-tained, named ?Personal?, ?Commercial?, ?Typical? and ?Other banks?, us-ing robust K-means clustering as implemented in R package RSKC, Kondo,Salibian-Barrera, and Zamar, 2016. In order to understand this grouping,projection pursuit based robust principal component analysis, Croux andRuiz-Gazen, 2005, was conducted on the whole set showing that essen-tially three variables can be attributed the formation of different clusters.In order to reveal each group inner structure, we used R package mclust tofit a finite Gaussian mixture to the data, selecting the best model throughBayesian Information Criterion from ten possible models. Thi