Financial Networks and Financial Stability


Published on

The recent global financial crisis has illustrated the role of financial linkages as a channel for the propagation of shocks. It also brought to the fore the concept that institutions may be “too interconnected to fail”, in addition to the traditional concept of being “too big to fail”.

Published in: Economy & Finance
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Financial Networks and Financial Stability

  1. 1. Financial Networks and Financial Stability Kimmo Soramaki Versions of this presentations have been held at the IMF, European Central Bank and Risk Europe 2010
  2. 2. 2003 2004 2003 Growing interest in networks
  3. 3. “ ... need for new and fundamental understanding of the structure and dynamics of economic networks.” “ Meltdown modeling -Could agent-based computer models prevent another financial crisis?” “ Is network theory the best hope for regulating systemic risk?” CFA Magazine, July 2009 Nature, August 2009 Science, July 2009
  4. 4. ... given the fragile condition of the financial markets at the time, the prominent position of Bear Stearns in those markets, and the expected contagion that would result from the immediate failure of Bear Stearns, the best alternative available was to provide temporary emergency financing to Bear Stearns ... Minutes of the Board of Governors of the Federal Reserve System, 14 March 2008 It was the ultra-interconnectedness of the nation’s financial institutions that posed the biggest risk of all [...] every firm was now dependent on the others – and many didn’t even know it . If one fell, it could become a series of falling dominoes . “ Too Big to Fail”, Andrew Ross Sorkin 2009
  5. 5. “ Too big to fail” “ Too interconnected to fail” +
  6. 6. Federal funds Bech, M.L. and Atalay, E. (2008), “The Topology of the Federal Funds Market”. ECB Working Paper No. 986.  Iori G, G de Masi, O Precup, G Gabbi and G Caldarelli (2008): “A network analysis of the Italian overnight money market”, Journal of Economic Dynamics and Control, vol. 32(1), pages 259-278 Italian money market Overnight lending networks
  7. 7. Soramaki, K, M.L. Bech, J. Arnold, R.J. Glass and W.E. Beyeler (2007), “The topology of interbank payment flows”, Physica A, Vol. 379, pp 317-333, 2007. Payment flows in Fedwire
  8. 8. Europe's Web of Debt (Bill Marsh / The New York Times, 1 May 2010)
  9. 9. NETWORK THEORY Financial Network Analysis Biological Network Analysis Graph & Matrix Theory Social Network Analysis Network Science Computer Science Network theory and related fields
  10. 10. Main premise of network analysis: the structure of the links between nodes matters The properties and behaviour of a node cannot be analysed on the basis its own properties and behaviour alone. To understand the behaviour of one node, one must analyse the behaviour of nodes that may be several links apart in the network. Bottom up approach. Generalize and describe. Financial context: network of interconnected balance sheets
  11. 11. <ul><li>Network terminology </li></ul><ul><ul><li>node/vertex </li></ul></ul><ul><ul><li>link/tie/edge/arc </li></ul></ul><ul><ul><li>directed vs undirected </li></ul></ul><ul><ul><li>weighed vs unweighted </li></ul></ul><ul><ul><li>graph + properties = network </li></ul></ul><ul><li>Algorithms/measures </li></ul><ul><ul><li>Centrality </li></ul></ul><ul><ul><li>Flow </li></ul></ul><ul><ul><li>Community/pattern identification </li></ul></ul><ul><ul><li>Distance, shortest paths </li></ul></ul><ul><ul><li>Connectivity, clustering </li></ul></ul><ul><ul><li>Cascades, epidemic spreading </li></ul></ul>-> Financial interlinkages, bilateral positions, exposures -> Systemic importantance -> Liquidity -> Contagion 4 1 2 3 -> Bank/banking group
  12. 12. <ul><li>“ Homophily” </li></ul><ul><ul><li>“ Birds of one feather flock together”, “herd behaviour” </li></ul></ul><ul><ul><li>Ideas, attributes, etc tend to cluster together and enforce each other </li></ul></ul><ul><ul><li>Examples: Some obvious (age, social status), others less (obesity, happiness, divorces) </li></ul></ul><ul><ul><li>How about: risk appetite, portfolio decisions, etc. </li></ul></ul><ul><li>“ Small world phenomenon” </li></ul><ul><ul><li>“ Six degrees of separation” (6.6 on MSN messenger) </li></ul></ul><ul><ul><li>The shortest path between any two nodes is very short </li></ul></ul><ul><ul><li>Implications for contagion? </li></ul></ul><ul><li>“ Robust yet fragile“, “Scale-free networks” </li></ul><ul><ul><li>“ The removal of &quot;small&quot; nodes does not alter the path structure of the remaining nodes, and thus has no impact on the overall network topology. “ </li></ul></ul>Degree (log) Probability (log) Fedwire degree distribution Spread of obesity Nicholas A. Christakis, James H. Fowler New England Journal of Medicine 357 (4): 370–379 (26 July 2007)
  13. 13. A risk-adjusted rate could be designed to address the contribution to systemic risk. Ideally, the rate would vary according to the size of the systemic risk externality , e.g. based on a network model which would take into account all possible channels of contagion. IMF report for the Meeting of G-20 Ministers, April 2010 Systemic importance
  14. 14. <ul><li>Equals “centrality” in network literature </li></ul><ul><ul><li>“ Recently, economists have argued that a bank’s importance within the financial system depends not only on its individual characteristics but also on its position within the banking network” Morten L. Bech, James T. E. Chapman, and Rod Garratt (2008) “ Which Bank Is the “Central” Bank? An Application of Markov Theory to the Canadian Large Value Transfer System ”, FRBNY Staff Report 356 </li></ul></ul><ul><li>Centrality measures in network theory </li></ul><ul><ul><li>degree: number of links </li></ul></ul><ul><ul><li>closeness: distance to other nodes via shortest paths </li></ul></ul><ul><ul><li>betweenness: number of shortest paths going through the node </li></ul></ul><ul><ul><li>eigenvector: nodes that are linked by/to other important nodes are more central </li></ul></ul><ul><ul><li>markov: probablity that a random process is at a node </li></ul></ul>
  15. 15. <ul><ul><li>Centrality in network theory </li></ul></ul><ul><li>The relative importance of a vertex within the graph </li></ul><ul><li>Depends on network process : </li></ul><ul><li>Trajectory: geodesic paths, paths, trails or walks </li></ul><ul><li>Transmission: parallel/serial duplication or transfer </li></ul>
  16. 16. Eyjafjallajökull volcano in Iceland, April 2010. The eruption magnitude was enhanced by interaction of meltwater from the glacier around the vent. Static electricity built up due to friction between ash particles in the dense 7 km high cloud, resulted in frequent massive discharges. (Marco Fulle -
  17. 17. <ul><li>Credit card companies use network/data mining/learning algorithms on customers’ payment behavior to detect card fraud </li></ul><ul><li>Can regulators use similar methods to detect rogue trading, increased riskiness of bank or market disturbances? </li></ul><ul><li>See if any common behavior can be detected </li></ul><ul><ul><li>payment timing by the bank (e.g. earlier than normal to create confidence), payment timing to the banks (e.g. later if the bank is considered risky) </li></ul></ul><ul><ul><li>net outflows (across ancillary securities and payment systems) </li></ul></ul><ul><ul><li>money market activity </li></ul></ul><ul><ul><li>cash withdrawals & deposits </li></ul></ul>Early warning indicators from payment data?
  18. 18. Road ahead FNA Advances in theory More granular and frequent data Improved tools
  19. 19. <ul><li>Advances in theory </li></ul><ul><ul><li>able to identify the contagion channels in different parts of the financial system </li></ul></ul><ul><ul><li>explain the formation and information content of links between financial institutions and their behaviour under normal and stress situations. </li></ul></ul><ul><ul><li>models of systemic risk could make sense of real economic interactions among market participants </li></ul></ul><ul><li>More granular and frequent data </li></ul><ul><ul><li>a key prerequisite for financial network analysis as a surveillance tool </li></ul></ul><ul><ul><li>more granular and frequent, long enough time series for a statistical analysis of different market conditions </li></ul></ul><ul><ul><li>regulators and overseers should continue to develop ways to systematically collect, share and analyse the data from both market sources and financial infrastructures. -> e.g. “Dodd-Frank Wall Street Reform and Consumer Protection Act” </li></ul></ul><ul><li>Improved tools </li></ul><ul><ul><li>Tools for network analysis/data mining have developed substantially over the last few years. Ongoing work: “Financial Network Analyzer” </li></ul></ul>
  20. 20. FNA tool <ul><li>A software tool for the statistical analysis and modelling of financial systems using methods developed in Social Network Analysis (SNA), Network Science and Agent-based Modelling (ABM). </li></ul><ul><li>Open source project sponsored by Norges Bank and the Bank of England.  </li></ul><ul><li>Includes many network creation, editing and analysis commands – as well as tools for data validation and manipulation. </li></ul><ul><li>Is also a platform for the development of simulation, network and agent based models that integrate easily with existing analysis and data manipulation functionality. </li></ul><ul><li>Version 1.3 will (at will be released this week. </li></ul>
  21. 21. Thank you! <ul><li>More information: </li></ul><ul><li>June Risk Magazine article </li></ul><ul><li>June ECB Financial Stability Review </li></ul><ul><li>My blog: </li></ul><ul><li>Contact me: </li></ul>