Measuring the systemic importance of interconnected banks

Abstract

We propose a method for measuring the systemic importance of interconnected banks. In order to capture contributions to system-wide risk, our measure accounts fully for the extent to which a bank propagates shocks across the system and is vulnerable to propagated shocks. An empirical implementation of this measure and a popular alternative reveals that interconnectedness is a key driver of systemic importance. That said, since the two measures incorporate the impact of interbank borrowing and lending on system-wide risk differently, they can disagree substantially about the systemic importance of individual banks. Keywords: Systemic risk, Shapley values, Interbank marketJEL Classification: C15, G20, G28, L14.* We thank Marek Hlavacek and Jörg Urban for excellent research assistance, Christian Upper and Goetz von Peter for providing us with codes for deriving interbank matrices, and Jeremy Staum and Kostas Tsatsaronis for insightful comments on an earlier draft of the paper. We have also benefited from presentations at the BIS, the Bank of England, the 2011 LSE-Bank of England conference on Systemic Risk and the Geneva Finance Research Institute conference on Financial Networks. The views expressed in this paper are ours and do not necessarily reflect those of the Bank for International Settlements. IntroductionIt is commonly thought that interconnectedness is a key driver of systemic importance. Yet, the literature has produced few concrete insights to support this view. Who is more systemically important -the lender or the borrower in an interbank market transaction? How should the two counterparties share the rise in system-wide risk associated with the interbank link? These fundamental questions remain unanswered despite recent policy initiatives to strengthen the financial system by tightening the regulatory requirements for interconnected banks. 1 To assess whether interconnectedness is indeed a key driver of systemic importance and how it affects different interbank market participants, we propose a measurement methodology and implement it empirically.We measure each bank's systemic importance as its share in the overall level of system-wide risk. And we explore two approaches, which decompose the same quantum of system-wide risk but allocate it differently across individual institutions. Adopting the perspective of a macroprudential regulator, we think of system-wide risk as the expected credit losses that the banking system as a whole may impose on non-banks in systemic events. These events, in turn, are characterised by aggregate losses exceeding a critical level.The first approach measures systemic importance as the expected losses that a bank imposes on its own non-bank creditors in systemic events. This approach equates systemic importance with the expected participation of individual banks in systemic events. Thus, we label it the participation approach (PA).Importantly, a bank's participation in systemic events is conceptually different from its contribution to system-wide risk. Consider, ...

Bank networks: Contagion, systemic risk and prudential policy

“…Given the vectors l and b, the matrix X obtained by this method yields the maximum density possible, that is, the market will be as complete as possible in the sense of Allen and Gale (2000). To obtain the maximum entropy solution we follow Drehmann and Tarashev (2011) and employ the RAS algorithm, a technique of bi-proportional matrix balancing developed in the context of input-output analysis for the purpose 21 Equivalently, one can think of the process as stopping when |L−B| ≤ ε, where ε is an arbitrarily small threshold. 22 The generic element x ij of matrix X is a proxy -generated by the algorithm -of l ij .…”

“…For brevity we decided to report the analytical description of the ex ante metrics and their performance in the numerical analysis in Appendix D. In the main text we focus on one ex post metric, the Shapley value (see Shapley (1953), and Bluhm et al (2013) and Drehmann and Tarashev (2011) for applications to banking), and we comment on the comparison with other ex ante metrics. The Shapley value, borrowed from the literature on cooperative and non-cooperative game theory, provides the contribution (through permutations) of each bank to an aggregate value.…”

An Overview of Macroprudential Policy Tools

“…DSGE-models with financial frictions can suggest an optimal mix of macroprudential and monetary policies (e.g., Kannan, Rabanal, and Scott, 2009;Quint and Rabanal, 2013). Or some historically derived indicators of (excessive) procyclicality and systemic risks, e.g., a notion of a "credit gap," can suggest specific dynamic provisioning surcharges (Drehmann et al 2011). And a Pigouvian tax on SIFIs can be made to depend on measures reflecting the size of interconnectedness externalities (Kocherlakota, 2013).…”

Banks’ Vulnerability and Financial Openness across Central and Eastern Europe

“…More recently, authors developed measures to identify SIFIs based on interbank positions (Drehmann and Tarashev, 2013), sovereign interlinkages (Correa et al, 2014), cross-border linkages (Minoiu et al, 2015) or network analysis (Cont et al, 2013;Hautsch et al, 2015;Betz et al, 2016). Also, this paper fits to research on regulatory incentives which highlight that financial stability can be significantly influenced by regulatory regimes (Weiβ et al, 2014), deposit insurance arrangements (Anginer et.…”