We propose a clustering method for large dimensional data to classify the 365 largest euro area financial institutions according to their business model. The proposed clustering approach is applied to granular supervisory data on banks’ activities and combines also dimensionality reduction and outlier detection. We identify four business models, namely wholesale funded, securities holding, traditional commercial and complex commercial banks while identifying as outliers the banks that follow idiosyncratic business models. Evidence is provided that the sets of banks following the distinct business models differ with respect to performance and risk indicators.

中文翻译：

---

欧元区银行的商业模式：高维度的聚类分析

我们针对大型数据提出一种聚类方法，以根据其业务模型对365个最大的欧元区金融机构进行分类。拟议的聚类方法适用于银行活动的精细监管数据，并且还结合了降维和离群值检测。我们确定四种商业模式，即批发融资，证券持有，传统商业和复杂商业银行，同时将遵循特殊商业模式的银行识别为离群值。提供的证据表明，遵循不同业务模型的银行组在性能和风险指标方面有所不同。