International Journal of Scientific & Technology Research

IJSTR@Facebook IJSTR@Twitter IJSTR@Linkedin
Home About Us Scope Editorial Board Blog/Latest News Contact Us

IJSTR >> Volume 6 - Issue 4, April 2017 Edition

International Journal of Scientific & Technology Research  
International Journal of Scientific & Technology Research

Website: http://www.ijstr.org

ISSN 2277-8616

Predicting Bank Financial Failures Using Discriminant Analysis, And Support Vector Machines Methods: A Comparative Analysis In Commercial Banks In Sudan (2006-2014)

[Full Text]



Mohammed A. SirElkhatim, Naomie Salim



Bank Fail, Discriminant Analysis, Data Mining, Support Vector Machines, Sudan Bank System, Predicting Bank Distress



Bank failures threaten the economic system as a whole. Therefore, predicting bank financial failures is crucial to prevent and/or lessen its negative effects on the economic system. Financial crises, affecting both emerging markets and advanced countries over the centuries, have severe economic consequences, but they can be hard to prevent and predict, identifying financial crises causes remains both science and art, said Stijn Claessens, assistant director of the International Monetary Fund. While it would be better to mitigate risks, financial crises will recur, often in waves and better crisis management is therefore important. Analyses of recurrent causes suggest that to prevent crises, governments should consider reforms in many underlying areas. That includes developing prudent fiscal and monetary policies, better regulating the financial sector, including reducing the problem of too-big-to-fail banks, and developing effective macro-prudential policies. Despite new regulations and better supervision, crises are likely to recur, in part because they can reflect deeper problems related to income inequality, the political economy and common human behavior. As such, improvements in crisis management are also needed. This is originally a classification problem to categorize banks as healthy or non-healthy ones. This study aims to apply Discriminant analysis and Support Vector Machines methods to the bank failure prediction problem in a Sudanese case, and to present a comprehensive computational comparison of the classification performances of the techniques tested. Eleven financial and non-financial ratios with six feature groups including capital adequacy, asset quality, Earning, and liquidity (CAMELS) are selected as predictor variables in the study. Credit risk also been evaluated using logistic analysis to study the effect of Islamic finance modes, sectors and payment types used by Sudanese banks with regard to their possibilities of failure. Experimental results are evaluated using accuracy of prediction. Features selection has shown that new groups can be identified from CAMELS ratios and narrowing the data set space to 11 factors instead of eighteen. Discriminant analysis has identified 3 ratios with highest predictive power which are: EAS (Ratio of equity capital to total asset), LADF (Ratio of liquid assets to deposits and short term funds) and RFR (Rain Fall Ratio), the later ratio is a novel one used for the first time by this research.



[1] Rietveld, T., van Hout, R., Rietveld, A.C. and Van Hout, R. (1993) Statistical techniques for the study of language and language behaviour. Germany: Mouton de Gruyter. (references)

[2] Erdogan, A. (2016) ‘Applying factor analysis on the financial ratios of turkey’s top 500 industrial enterprises’, International Journal of Business and Management, 8(9). doi: 10.5539/ijbm.v8n9p134.

[3] Bardos M. (2001): Analyse discriminante: application au ri sque et scoring financier, Dunod.

[4] Altman, E.I., Marco, G. and Varetto, F. (1994) ‘Corporate distress diagnosis: Comparisons using linear discriminant analysis and neural networks (the Italian experience)’, Journal of Banking and Finance, 18(3), pp. 505–529. doi: 10.1016/0378-4266(94)90007-8.

[5] Chang, Y.-W. and Hsieh, C.-J. (2010) ‘Training and Testing Low-degree Polynomial Data Mappings via Linear SVM’, Journal of Machine Learning Research, 11, pp. 1471–1490.

[6] Cortes, C. and Vapnik, V. (1995) ‘Support-vector networks’, Machine Learning, 20(3), pp. 273–297. doi: 10.1007/bf00994018..

[7] Ribeiro, B., Silva, C., Chen, N., Vieira, A. and Carvalho das Neves, J. (2012) ‘Enhanced default risk models with SVM+’, Expert Systems with Applications, 39(11), pp. 10140–10152. doi: 10.1016/j.eswa.2012.02.142.

[8] Chang, Y.-W. and Hsieh, C.-J. (2010) ‘Training and Testing Low-degree Polynomial Data Mappings via Linear SVM’, Journal of Machine Learning Research, 11, pp. 1471–1490.

[9] http://www.CBOS.gov.sd.

[10] Memic, D. (2015) ‘Assessing credit default using logistic regression and multiple Discriminant analysis: Empirical evidence from Bosnia and Herzegovina’, Interdisciplinary Description of Complex Systems, 13(1), pp. 128–153. doi: 10.7906/indecs.13.1.13.

[11] Zavgren, C.V., Dugan, M.T. and Reeve, J.M. (1988) ‘The association between probabilities of bankruptcy and market responses?A test of market anticipation’, Journal of Business Finance and Accounting, 15(1), pp. 27–45. doi: 10.1111/j.1468-5957.1988.tb00118.x.