A Survey Study of Financial Contagion and Interbank Network
Cross-border interbank market and financial contagion: A survey By Ye Bai, Christopher J. Green, Victor Murinde Bai: Nottingham University Business School, University of Nottingham Green: Department of Economics, Loughborough University Murinde: Birmingham Business School, University of Birmingham ESRC Reference: ES/N013344/1: Delivering Inclusive Financial Development and Growth
Research Project Workshop SOAS University of London 3rd 4th March, 2017 1. Introduction The aims: to identify key results and questions arising from the theoretical and empirical literature The focus: financial contagion via interbank network In particular: cross-border contagion and low income countries in Africa The background: Contagion banks interconnected as a complex network Channel - interbank network cross-border activities and risk exposures of international financial intermediaries vs. supervision mainly at national level
Structure of the presentation 2. To review theoretical literature that identify how contagion is channelled through interbank market 3. To discuss two main methodologies used to measure/identify contagion 4. To review both theoretical and empirical literature to see whether a particular type of interbank network structure is more vulnerable to contagion
5. To review literature on cross-border contagion 6. To review the recent development of cross-border banking in low income countries in Africa 7. To conclude 2. Theoretical literature: contagion channelled through interbank market Contagion: one bank being hit by an idiosyncratic shock, which then is transmitted to other banks.
Direct contagion: the overlapping claims between different regions or sectors of the banking system (Allen and Gale, 2000; Freixas, Parigi and Rochet, 2000; Allen, Babus and Carletti, 2009) Unforeseen liquidity shocks Credit shocks Indirect contagion: moral hazard, asymmetric information, and monopoly power (Acharya et al., 2008) markets assume the existence of the direct contagion effects o With asymmetric information about the quality of counterparties and of the underlying collateral, adverse selection problems may arise (Allen and Gale, 2000) one failure = a signal of systematic failures (mistakenly) (Freixas, et
al., 2000; Blavarg and Nimander, 2002) liquidity hoarding o Acharya and Merrouche (2013) -lender with a higher demand for liquidity charges a higher price; riskier UK settlement banks held more reserves than expected after 9/8/2007; the rollover risk of banks have also increased o Berrospide (2013) - evidence for the precautionary motive of liquidity hoarding for U.S. commercial banks during 2007/08 financial crisis o Acharya and Skeie (2011) - the rise in spreads and a collapse in maturities in the term interbank market are explained by lending banks precautionary demand for liquidity. 3. Empirical literature: measure/identify contagion The early applied literature: a number of possible contagion channels and several ways to measure contagion risk
the estimation of autocorrelation (Schoenmaker, 1996) the survival time test (Calomiris and Mason, 2000) event study method (Docking, Hirschey and Jones, 1997; Slovin, Sushka and Polonchek, 1999) Simultaneous study of the overall risk exposure (network theory) So far the evidence is inconclusive o Contagion is reported: Upper and Worms (2004) for Germany, Mistrulli (2005) for Italy, van Lelyveld and Liedorp (2006) and van Lelyveld (2014) for the Netherlands and Degryse and Nguyen (2007) for Belgium o Limited scope of contagion: Mistrulli (2007), Sheldon and Mauer (1998) and Mller (2006) for Switzerland, Blavarg and Nimander (2002) for Sweden, Furfine (2003) for the US, Wells (2004) for the UK, Lubly (2005) for Hungary and Elsinger, Lehar and Summer (2006) for Austria. Counterfactual simulation method
o Advantage: allows for counterfactual analysis vs. Information-induced empirical studies: lack of evidence o Disadvantages: 1) Limited data access to bilateral exposures (Kho and Stulz, 2000; Burnside, Eichenbaum, and Rebelo, 2001) hence have to use maximum entropy to assume uniform distributionlead to bias when applied to cross-border exposures 2) The absence of behavioural foundations 3) Plagued by data problems (Upper, 2011): (excluding a range of items: the off-balance sheet exposure, intraday exposures, exposures of foreign banks, long time series) The method of extreme value theory (EVT) and market-based distance-to-default (DD) measure o EVT: to capture the information of large, extreme shocks transmitted across financial systems
o DD: a market-based measure of bank distress, defined as the difference between the current market value of assets of a firm and its estimated default point, divided by the volatility of assets (Gropp and Vesala, 2004; Gropp, Vesala, and Vulpes, 2004; Gropp, Lo Duca and Vesala, 2006; Duggar and Mitra, 2009) does not require specification of a particular channel of contagion (Duggar and Mitra, 2009) to utilise the more widely available market-based data to focus on cross-border contagions (e.g. Gropp and Vesala, 2004; Gropp and Moerman, 2004; Gropp, Lo Duca and Vesala, 2006; Cocozza and Piselli, 2011 try to identify potential risks of cross-border contagion using a sample of large banks in the EU) potentially omit important linkages with other major banking centres Instead, Chan-Lau, et al. (2007): test for contagion risk among
selected 24 biggest banking groups in the world. Their results reveal several key trends among major global banks. The nature of DD has led to some criticisms on this measurement: 1) Two critical parameters are unobservable: the asset drift and the asset volatility 2) DDs are risk-neutral 3) Can only be applied to listed banks 4. The Vulnerability of Different Network Structures Theoretical literature a simple weighted graph (Allen and Gale, 2000; Freixas et al., 2000; Leitner, 2005; Shin, 2008)
Interbank networks display a knife-edge or robust-yet-fragile property (Haldane, 2009)whether it is easier for the failure of one bank to spread to other banks within a particular type of interbank network structure. Existing theory predicts two opposite results: 1) A less complete banking structure implies a higher probability of contagion (Allen and Gale, 2000; Babus, 2006) 2) A more complete banking structure implies a higher contagious effect (Brusco and Castiglionesi, 2007; Hasman and Samartin, 2008) Empirical literature a tiered network structure From the network topology perspective (e.g. Caballero, 2015) o Nodes o weighted and directed links o betweenness,
o clustering coefficient o hub centrality, o authority, o indegree o outdegree Typically interbank networks: a core-periphery (or scale-free or tiered) structure (Boss et al., 2004 for Austria, Embree and Roberts, 2009 for Canada, Lubly, 2006 for the Hungry, Sokolov et al., 2012 for the Australia, Craig and Peter, 2014 for Germany) o a few money centre (core) banks exploit their cost of funding advantage o due to business ties with small cash-rich institution or overseas investors o by undertaking lending transactions with domestic banks (periphery) which are lack of such market access
Scale-free networks are more vulnerable to contagion when compared with randomly formed network (Lenzu and Tedeschi, 2012, Iori, et al. (2008) Cont et al. (2012) scale-free networks are more resilient to shocks as the level of tier increases ((Boss et al., 2004 ; Degryse and Nguyen, 2007; Teteryatnikova, 2014) 5. Cross-border Banking and Contagion Cross-border financial contagion defined as the situation when an exogenous, sudden and idiosyncratic shock that hits the foreign liabilities (entirely or partly) of a countrydomino effects startimpact other banking systems worldwide (Degryse, Elahi and Penas, 2009) Sources of cross-country contagion risk: 1) bank deposit runs triggered by rumours about banking system where
banks have overseas subsidiaries, branches, etc. (Freixas, et al., 2000) 2) liquidity and credit risk in interbank markets (Chan-Lau, 2010; EspinosaVega and Sole, 2010; Garratt et al., 2011). 3) legal enforcement for the vulnerable net settlement can be particularly difficult for international transactions 4) International banks are less able to rely on the financial safety net in the host country due to the inadequacy of an efficient recovery and resolution regime for international banks (Bertay, Demirg-Kunt and Huizinga, 2016) Mixed empirical evidence: Adding more links increases the channels through which the risk of financial contagion can spread to the international financial system (Degryse and Nguyen (2004, Gropp and Moerman, 2004; Hartmann, Straetmans, and
De Vries, 2005; Gropp, Lo Duca, and Vesala, 2009) Further increases strengthen the resiliency (Nier, Yang, Yorulmazer and Alentorn, 2007). The relative importance of the two depends on the level of connectivity and the amount of capital in the system (Nier, et al., 2007). Similar core-periphery (or tier) network structure (Gropp and Vesala, 2004; Garratt, et al., 2011; Chan-Lau, Mitra, and Ong, 2007) Different approaches are applied in limited cross-border contagion studies 1) Use equity prices to measure cross-border contagion (Gropp and Moerman, 2004; Hartmann, Straetmans and de Vries, 2005; Gropp, Lo Duca and Vesala, 2006; Duggar and Mitra, 2009; Cocozza and Piselli, 2011). 2) Use bank cross-border exposures, but focuses on the effects on
a single country (Van Lelyveld and Liedorp (2006) for the Netherlands, Degryse and Nguyen (2007) for Belgium), McGuire and Tarashev (2007) failure originating from emerging countries ) 3) Apply a network approach to Bank of International Settlements country-level banking statistics (Hattori and Suda, 2007; von Peter, 2007; McGuire and Tarashev, 2008; Degryse, et al., 2009; Espinosa-Vega and Sole, 2010; Minoiu and Reyes, 2013) 4) Use simulated network structure with topology statistics (Hattori and Suda, 2007; Nier et al., 2007; Battiston et al., 2009) analyse the.) 5) Gabrieli, Salakhova and Vuillemey (2015): critical impact of the underlying network structure on contagion; cross-border contagion vary over time between 2009 and 2010 and vary over different banking sectors
6. Limited attention to interbank and financial crisis in LICs banking systems in LICs have on average shown resilience to 2007/08 financial crisis Enhanced supervision framework and strengthened prudential requirements play important roles (IMF, 2012) However, banking system fragility still exist and others can arise as financial deepening increases and financial systems become more sophisticated On one hand, relaxed ownership restrictionsincreasing presence of foreign banks in Africa, e.g. the share of total assets owned by foreign banks in Uganda leading the region at 73.7% (Kodongo, et al., 2015) positive progress for financial markets in developing countries (Cihk and Podpiera, 2005)
a source of vulnerability too (Claessens Demirg-Kunt and Huizinga, 2001; Koutsomanoli-Filippaki et al., 2009; Giannetti and Ongena, 2009; Gormley, 2010). In particular, cross-border banking has increased in recent times in the East African region For example, Kenya o With more advanced financial markets, institutional and legal systems than others in the region o 11 Kenyan banks operate across East Africa region by the end of 2012 o All have operations in Rwanda, Tanzania and Uganda o Foreign subsidiaries and branches contributed about 9.8% of Kenyan banks total assets between 2007 and 2011 o A well-established and relatively liquid overnight interbank market (Green, et al., 2017)
o Unlike typical core-periphery structure, the strongest network relationships are those directly among smaller banks and much less as between large and small banks (Oduor, Sichei, Tiriongo and Shimba, 2014) 7. Conclusion Interbank network exposure is an important channel for contagion (national or international) Empirical evidence is mixed Majority of the studies/approaches are restricted by data availability problem Most of the studies focus on industrial countries with very limited attention paid to other countries especially countries at the early stage of financial development Network theory helps our understanding of the interconnections between banks but data issue causes concerns All these elements have made supervision difficult but also more
important Going Forward Completion of survey paper Empirical study of the interbank network structure in Eastern Africa To identify how the growth of cross-border banking in the region affect the risk of contagion Fieldwork for key data required: Kenya, Rwanda, Tanzania and Uganda Perhaps including Malawi and Zambia? Thank You !! Questions/ Comments are welcome
The sender's address The senders mailing address Phone number and e-mail address Date being sent Heading 234 South 3rd Street Drew, MS 38737 June 6, 2001 Inside Address The inside address gives the name and address of the person or...
Chapter 21: Metamorphism The Limits of Metamorphism Low-temperature limit grades into diagenesis The boundary is somewhat arbitrary Diagenetic/weathering processes are indistinguishable from metamorphic Metamorphism begins in the range of 100-150oC for the more unstable types of protolith Some zeolites are...
An object that wraps a row in a database table or view, encapsulates the database access, and adds domain logic on that data. Putting DB access logic in the domain object. Each active record is responsible for DB access as...
BE HONEST Homework: Ailanthus Melcon Due Friday Weekly Reading #3 Due Friday Unit 1 test Monday 9/25 (Content, Skills, Reading )and Tuesday 9/26 (Writing & Binder Quiz) If you have not taken your Quiz, Tomorrow Lunch is the LAST CHANCE...
Describe the evolution of computers and digital media. ... An electronic device, operating under the control of instructions stored in its memory, that can accept data, process the data according to specified rules, produce results, and store the results for...
DNA : Cytoplasm. Cell Membrane. Ribosomes. ... necklace to your friend in another state. Look like stack of pancakes. ... Understand the diffusion lab that we conducted in class. What did the bag represent? What moved into bag and how...
Learning Objective: To explore some of the characters of the novel Exam questions often feature a character study. For that reason, you should be able to provide key quotes and character. So far, we've looked at Crooks, Slim and Curley...
Ready to download the document? Go ahead and hit continue!