Using clustering ensemble to identify banking business models

Abstract

Summary The business models of banks are often seen as the result of a variety of simultaneously determined managerial choices, such as those regarding the types of activities, funding sources, level of diversification, and size. Moreover, owing to the fuzziness of data and the possibility that some banks may combine features of different business models, the use of hard clustering methods has often led to poorly identified business models. In this paper we propose a framework to deal with these challenges based on an ensemble of three unsupervised clustering methods to identify banking business models: fuzzy c‐means (which allows us to handle fuzzy clustering), self‐organizing maps (which yield intuitive visual representations of the clusters), and partitioning around medoids (which circumvents the presence of data outliers). We set up our analysis in the context of the European banking sector, which has seen its regulators increasingly focused on examining the business models of supervised entities in the aftermath of the twin financial crises. In our empirical application, we find evidence of four distinct banking business models and further distinguish between banks with a clearly defined business model (core banks) and others (non‐core banks), as well as banks with a stable business model over time (persistent banks) and others (non‐persistent banks). Our proposed framework performs well under several robustness checks related with the sample, clustering methods, and variables used.

The profitability and distance to distress of European banks: do business choices matter?

“…Particularly, we use as proxies for relationship banking the number of branches per total assets (which indicates the proximity of the bank to its customers), net margin (a higher value of which suggests the provision of customized, high value-added services) and the number of employees per total assets (in order to capture the relatively greater time and effort expected to be invested by relationship banks). The method used to identify business models follows the approach developed by Marques & Alves (2020), which may be summarized in the following way. Firstly, we perform principal component analysis on a selection of business model variables related with the assets and funding structures, diversification, size and capital.…”

“…Bearing this in mind, this paper is also the first to employ the method proposed by Marques & Alves (2020) to identify discrete banking business models, which combines the outputs of three clustering methods (Fuzzy C-Means, Self-Organizing Maps, Partitioning Around Methods).…”

“…A detailed description of the clustering ensemble method, including each internal valuation criteira used, can be found inMarques & Alves (2020).…”

Forecasting Commodity Market Returns Volatility: A Hybrid Ensemble Learning GARCH‐LSTM based Approach

“…Ensemble methods are among the most widely used techniques to improve prediction performance. They have thus found several applications in the financial world, such as credit scoring (Abellán & Castellano, 2017), bankruptcy prediction (Davalos et al, 2014), and identification of banking business models (Marques & Alves, 2020). Lahmiri and Boukadoum (2015) use it to forecast the intraday volatility of the S&P 500.…”