In complex systems, homogeneity has been documented as a source of fragility. Likewise, in the financial sector, it has been documented as a contributing factor for systemic risk. We assess homogeneity in the Colombian case by measuring how similar banks are regarding the structure of their overall financial statements, and their lending, investment, and funding portfolios. Distances among banks and an agglomerative clustering method yield the hierarchical structure of the banking system, which exhibits how banks are related to each other based on their financial structure. The Colombian banking sector displays homogeneous features, especially among the largest banks. Also, it seems size is a crucial determinant in the banking sector’s hierarchical structure. Results are robust to a principal component analysis feature selection procedure that reduces the dimensionality of the dataset. Results enable studying to what extent the banking sector is homogeneous, to identify banking firms that have a(n) (un)common financial structure, and, thus, to better examine systemic risk.
|Number of pages||42|
|Journal||Revista de Economia del Rosario|
|Publication status||Published - 1 Jul 2020|
- Machine learning
- Systemic risk