Systemic risk measurement in banking using self-organizing maps

For further discussion and perspectives on systemic risk, see Kaufman (1990, 1991, 2002) Kaufman, G. (1990) Are some banks too large to fail? Myth and reality. Contemporary Policy Issues 8 (4): 1–14; Kaufman, G. (1991) Too large to fail, too costly to continue. Consumer Finance Law. Quarterly Report, Spring; Kaufman, G. (2002) Too big to fail in banking: What remains. Quarterly Review of Economics and Finance 42(3): 423–436; Bernanke, B. (2010) Causes of the recent financial and economic crisis. Speech before the Financial Crisis Inquiry Commission, Washington, D.C., September 2nd; Jagtiani, J.A., Kolari, J.W., Lemieux, C.M., and Shin, G.H. (2000) Predicting inadequate capitalization: Early warning system for bank supervision. In Emerging Market Series (S&R-2000-10R), Federal Reserve Bank of Chicago; Stern, G. (2009a) Addressing the too-big-to-fail problem. Statement presented before the Committee on Banking, Housing, and Urban Affairs, U.S. Senate, Washington, DC, May 6; Stern, G. (2009b) Better late than never: Addressing too-big-to-fail. Brookings Institution, Washington, DC, March 31; and Stern, G., and Feldman, R. (2009) Addressing TBTF by shrinking financial institutions: An initial assessment. Federal Reserve Bank of Minneapolis, The Region, June: 8–13. Some authors suggest that TBTF companies avoid market discipline to some extent by receiving favorable treatment from regulatory authorities and rating agencies (e.g., lower deposit costs that smaller banks and high ratings on bonds issued in the securitization process). See, for example, Kane, E. (2000) Incentives for banking megamergers: What motives might regulators infer from event-study evidence? Journal of Money, Credit and Banking 32(3): 671–701. Other authors propose a measure of bank undercapitalization in times of financial stress, which is similar in spirit to stress tests under the Federal Reserve’s Supervisory Capital Assessment Program (SCAP). See, for example, Acharya, V.V., Pedersen, L. H., Philippon, T., and Richardson M. (2010) Measuring systemic risk. Working paper, New York University. Also, some authors propose a value at risk (i.e., maximum dollar loss) measure of the financial system in times of financial stress. See, for example, Adrian, T., and Brunnermeier, M. (2011) CoVar. Working paper, Federal Reserve Bank of New York and Princeton; and Allen, L., Bali, T.G., and Tang, Y. (2012) Does systematic risk in the financial sector predict future economic downturns? Review of Financial Studies 25(10): 3000–3036. For excellent surveys of different systemic risk measures developed over the years, see Gramlich, D., Miller, G., Oet, M., and Ong, S. (2010) Early warning systems for systemic banking risk: Critical review and modeling implications. Banks and Bank Systems 5(2): 199–211; Bisias et al (2012) Bisias, D., Flood, M., Lo, A. W., and Valavanis, S. (2012) A survey of systemic risk analytics, Working paper no. 0001, Office of Financial Research, U.S. Department of the Treasury; and Hansen, L. P. (2013) Challenges in identifying and measuring systemic risk. Working paper, University of Chicago and NBER. Most experts agree that large banks were the source of most systemic risk in the recent global financial crisis. For discussion and evidence on the role of big banks in the financial system crisis, see Laeven, L., Ratnovski, L., and Tong, H. (2014) Bank size and systemic risk. International Monetary Fund, Staff Discussion Note SDN/14/04. For descriptions and applications of trait recognition methods to predicting bank failures, see studies by Kolari, J., Caputo, M., and Wagner, D. (1996) Trait recognition: An alternative approach to early warning systems in commercial banking. Journal of Business Finance and Accounting 23(9–10): 1415–1434; Jagtiani, J., Kolari, J., Lemieux, C., and Shin, G.H. (2003) Early warning models in bank supervision: Simpler could be better. Economic Perspectives 27 (3rd quarter), Federal Reserve Bank of Chicago: 49–60; Kolari, J., Glennon, D., Shin, H., and Caputo, M. (2002) Predicting large U.S. commercial bank failures. Journal of Economics and Business 54(4): 361–387; and Jagtiani, Kolari, Lemieux, and Shin1. On February 10th, 2007 Treasury Secretary Geithner announced the SCAP and associated stress tests. On May 7, 2009, the Federal Reserve identified 10 out of 19 large institutions that failed initial stress tests and were required to raise $74.6 billion in additional capital. For alternative approaches to estimate ex ante bank undercapitalization, see Acharya, Pedersen, Philippon, and Richardson1. Hansen2. While the bank failure literature is voluminous, a limited number of studies on the prediction of large bank failure exist. See, for example, Pettway, R.H. (1976) Market tests of capital adequacy of large commercial banks. Journal of Finance 31(3): 865–875; Pettway, R.H. (1980) Potential insolvency, market efficiency, and bank regulation of large commercial banks. Journal of Financial and Quantitative Analysis 15(1): 219–236; Peavy, J.W., III, and Hempel, G.H. (1988) The Penn Square Bank failure: Effect on commercial bank security returns – A note. Journal of Banking and Finance 12(1): 141–150; Kolari, Glennon, Shin, and Caputo4; Harada, K., Takatoshi, I., and Takahashi, S. (2013). Is the distance to default a good measure in predicting bank failures? A case study of Japanese major banks. Japan and the World Economy 27(August): 70–82; Beltratti, A., and Stulz, R.M. (2010) The credit crisis around the globe: Why did some banks perform better during the credit crisis? Working paper, Bocconi University, Ohio State University, NBER, and ECGI; and Jin, J., Kanagaretnam, K., and Lobo, G. (2011) Ability of accounting and audit quality variables to predict bank failure during the financial crisis. Journal of Banking and Finance 35(11): 2811–2819. Recent work suggests that decreasing liquidity was a predictor of large bank distress and failure in the recent 2008–2009 financial crisis. See, for example, Cooke, J.B., Koch, C., and Murphy, A. (2015) Liquidity mismatch helps predict bank failure and distress. Federal Reserve Bank of Dallas, Economic Letters 10(4): 1–4. See http://www.fdic.gov/. Management is normally assessed by on-site bank examiners and is not available from Call Report data. See, for example, Bernhardsen, E. (2001) A model of bankruptcy prediction. Working paper, Financial Analysis and Structure Department, Norges Bank, Oslo; De Laurentis, G., Maino, R., and Molteni, L. (2010). Developing, validating and using internal ratings. Chichester: Wiley; and Hammer, P.L., Kogan, A., and Lejeune, M.A. (2012). A logical analysis of banks financial strength ratings. Expert Systems with Applications 39 (9): 7808–7821. See, for example, du Jardin, P. (2010) Predicting bankruptcy using neural networks and other classification methods: The influence of variable selection techniques on model accuracy. Neurocomputing 73(10–12): 2047–2060; Hol, S. (2006) The influence of the business cycle on bankruptcy probability. Discussion papers no. 466, Statistics Norway Research Department; and Levy-Yeyati, E., Martínez Pernía, M.S., and Schmukler, S.L. (2010) Depositor behavior under macroeconomic risk: Evidence from bank runs in emerging economies. Journal of Money, Credit and Banking 42(4): 585–614. See Kohonen, T. (2001). Self-organizing maps (3rd ed.). Berlin: Springer. See Silven, O., Niskanen, M., and Kauppinen, H. (2003) Wood inspection with nonsupervised clustering. Machine Vision and Applications 13(5): 275–285. Closely related to the present study, see studies of bank and corporate failure prediction by Boyacioglu, M. A., Kara, Y., and Baykan, O.K. (2009) Predicting bank financial failures using neural networks, support vector machines and multivariate statistical methods: A comparative analysis in the sample of savings deposit insurance fund (SDIF) transferred banks in Turkey. Expert Systems with Applications 36(2): 3355–3366; Kiviluoto, K., (1998), Predicting bankruptcies with the self-organizing map, Neurocomputing 21(1–3): 191–201; and Serrano-Cinca, C. (1996) Self-organizing neural networks for financial diagnosis. Decision Support Systems 17(July): 217–228. For work on SOMS produced for macroeconomic financial imbalances among European countries, see López Iturriaga, F., and Pastor, I. (2013) Self-organizing maps as a tool to compare financial macroeconomic imbalances: The European, Spanish and German case. The Spanish Review of Financial Economics 11(2): 69–84. See, for example, Kohonen, T., (1993) Physiological interpretation of the self-organizing map algorithm. Neural Networks 6(6): 895–905; and Kaski, S., and Kohonen, T. (1994) Winner-take-all networks for physiological models of competitive learning. Neural Networks 7(6–7): 973–984. From the U-matrix, it is possible to construct a colored SOM for each variable or feature in the present case. Some researchers use component places to better understand the correlations between variables and relative contributions of variables to the SOM. Component planes provide a sliced version of the SOM wherein the relative distribution of observations for each feature (i.e., variable) is mapped. This analysis is beyond the scope of the present study but could be valuable to regulators and others applying SOMs to banking institutions. To conserve space, component planes of the 36 features are available upon request from the authors. See Davies, D.L., and Bouldin, D.W. (1979), A cluster separation measure. IEEE Transactions on Pattern Analysis and Machine Intelligence 1(2): 224–227. See Ingaramo, D.A., Leguizamón, G., and Errecalde, M. (2005) Adaptive clustering with artificial ants. Journal of Computer Science and Technology 5(4): 264–271. See Kolari, Caputo, Wagner4. See Kohonen14.