Network models to improve credit risk estimation

Abstract:

We propose credit risk measurement models that incorporate not only idiosyncratic risk, as measured by credit ratings or CDS spreads, but also how companies and countries interact via multiple linkage types. The models are based on network science methods that extract adjacency matrices from multilayer networks,based on financial transactions  between companies and/or countries. Based on an adjacency matrix, we develop new network centrality measure that can employed to correct traditional credit risk measures, taking contagion into account. The proposed methodology will be applied to different contexts. First, we will employ the Bank of International Settlements statistics that describe financial flows between country banking systems, and show how credit risk of a country (sovereign risk) can be affected by contagion from other countries, either on the funding or on the lending side. Then we will consider the context of financial technology (f intech)  and, specifically, that of peer-to-peer lending. We will show that including network parameters does improve the accuracy of the credit risks estimated by peer to peer lenders and, in addition, provides useful information on the variables that emphasize the centrality of bad performing companies, potentially triggering contagion.

**********************

This talk is part of a public seminar series organised within the Master in Financial Technology & Computing

Biography:

Paolo Giudici is Professor of Statistics at the University of Pavia, where he teaches Statistics and Data Science (Department of Economics) and coordinates the Phd programme in Financial Technology (Department of Computer Engineering). He has been the academic supervisor of about 160 Master's students and of 12 Phd students, who are currently working in the financial industry, in IT/consulting companies or as academic researchers. Paolo is author of many scientific publications, about Multivariate statistics, Network models, Risk management, and their application to Finance and FINancial TECHnologies., with  an h-index of 25 (google scholar) and of 15 (scopus). Prof. Giudici is Director of the University of Pavia Fintech laboratory (formerly Data Mining laboratory) which, since 2001, carries out research, training and consulting projects, for leading institutions such as the European Commission, the Bank for International Settlements, the Asian Development Bank Institute, the Deutsche Bundesbank, the Italian Banking Association, Cariplo Foundation, Intesa SanPaolo, Unicredit, BancoBPM, UBI, MPS, BPS, Creval, Mediolanum, Nexi, RMS, Reply, Sirti, Accenture, KPMG, Mediaset, Mondadori, SAS Institute, Sky. Paolo is also a Board of Directors member of the Credito Valtellinese Banking Group; Research fellow at the University College London center for Blockchain technologies; Research fellow at the Bank for International Settlements, Basel; Chief Editor of "Artificial Intelligence in Finance", Frontiers; Associate Editor of “Digital Finance”, Springer. Honorary member of the Association of the Italian Financial Industry Risk Managers; Member of the scientific committee of Assofintech.