Network Linkages to Predict Bank Distress

Abstract

Standard-Nutzungsbedingungen:Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden.Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen.Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in AbstractBuilding on the literature on systemic risk and financial contagion, the paper introduces estimated network linkages into an early-warning model to predict bank distress among European banks. We use multivariate extreme value theory to estimate equity-based tail-dependence networks, whose links proxy for the markets' view of bank interconnectedness in case of elevated financial stress. The paper finds that early warning models including estimated tail dependencies consistently outperform bank-specific benchmark models without networks. The results are robust to variation in model specification and also hold in relation to simpler benchmarks of contagion. Generally, this paper gives direct support for measures of interconnectedness in early-warning models, and moves toward a unified representation of cyclical and cross-sectional dimensions of systemic risk.Keywords: bank distress; bank networks; systemic risk JEL codes: G21, G33, C54, D85ECB Working Paper 1828, July 2015 1 Non-technical summaryIn recent years, considerable effort has been invested into research on analyzing the build-up of vulnerabilities for banks and timely predicting possible bank distress events that could impact the soundness of the European banking system. The motivation behind this lies with the recent global financial crisis of 2007-2009 and the subsequent sovereign debt crisis in Europe, which brought a large number of European banks to the brink of collapse and prompted government interventions that led to unprecedented bailout costs. An important contribution to the enormous impact of the financial crises came from the high interconnectivity of the European banking system, which functions as a potential propagation mechanism of bankspecific vulnerabilities, allowing for risks and contagion to flow through the system. Much of the empirical literature proposes early-warning models based on conventional statistical modeling methods, such as multivariate logit/probit models that deliver distress predictions for individual banks. However, these models do not account for possible bank interdependencies. This paper addresses this shortcoming by providing a general-purpose framework that enables combining potential contagion effects with bank distress models. In particular, the paper proposes a two-step estimation where the bank failure model of Betz et al. (2014) is complemented with different contagion variables that account for possible vulnerabili...

Conditional graphical models for systemic risk estimation

“…et al (2012) 526 Diebold et al (2014) Hautsch et al (2013) andPeltonen et al (2015),527 is mainly a simplified method for model selection and parameter es-528 timation, based on conditional independence restrictions, which al-529 lows computations to localise on subsets of the graph. This, in turn, 530 simplifies model interpretation.…”

Bank distress in the news: Describing events through deep learning

“…We conclude that at µ = 0.9 with vector-level evaluation and at µ = 0.875 with aggregated evaluation the model has decent predictive performance by capturing up to 33% of available Usefulness and 12% in the more conservative leave-N-banks-out sampled exercise. To relate the results we may confer Betz et al [5] who obtain U r of 19-42% and Peltonen et al [29] with 58-64%. The latter incorporates network linkages, which we currently do not model, although this is possible to extract from text as well (cf.…”

“…Following previous studies [5,29], we make use of a skewed preference µ ≈ 0.9 (i.e., missing a crisis is about 9 times worse than falsely signaling one). From the viewpoint of policy, highly skewed preferences are particularly motivated when a signal leads to an internal investigation, and reputation loss or other political effects of false alarms need not be accounted for.…”

RiskRank: Measuring interconnected risk

“…Yet, this provides little information on the vulnerability of one entity and its impact on others. In this vein, Peltonen et al Peltonen et al [33] have explicitly modeled the vulnerability of one bank as a function of the vulnerability of its neighbors through tail-dependence networks. While being a starting point, this provides no structured approach to accounting for both dimensions simultaneously.…”