skip to main content

CORPORATE FINANCIAL DISTRESS PREDICTION USING STATISTICAL EXTREME VALUE-BASED MODELING AND MACHINE LEARNING

*Dedy Dwi Prastyo orcid scopus  -  Department of Statistics, Institut Teknologi Sepuluh Nopember, Indonesia
Rizki Nanda Savera  -  Department of Statistics, Institut Teknologi Sepuluh Nopember, Indonesia
Danny Hermawan Adiwibowo  -  Bank Indonesia, Indonesia
Open Access Copyright (c) 2023 MEDIA STATISTIKA under http://creativecommons.org/licenses/by-nc-sa/4.0.

Citation Format:
Abstract
The industrial sector plays a leading role in an economy such that the financial stability of companies from this sector be a big concern. Two financial ratios, i.e., the Interest Coverage Ratio (ICR) and the Return on Assets (ROA), are used to determine the corporate financial distress conditions. This work considers two schemes for determining financial distress. First, a company is categorized as distressed if either ICR<1 or ROA<0. The second scheme is for when both ICR<1 and ROA<0 are met. The proportion of distressed and non-distressed companies is imbalanced. Our work views the distressed companies (minority class) as a rare event, causing the proportion to be extremely small, such that the Extreme Value Theory can be employed. The so-called Generalized Extreme Value regression (GEVR), developed from GEV distribution, predicts the distressed labels. The GEVR's performance is compared using machine learning with and without feature selection. The feature selection in GEVR uses backward elimination. The model for prediction employs a drift or windowing concept, i.e., using past-period predictors to predict the current response. The empirical results found that the GEVR, with and without the feature selection, provides the best prediction for financial distress.
Fulltext View|Download
Keywords: Classification; feature selection; financial distress; GEVR; imbalanced data; machine learning.

Article Metrics:

  1. Bradley, P.S. and O.L. Mangasarian, O.L. 1998. Feature selection via concave minimization and support vector machines, Proceeding of the Fifteenth International Conference on Machine Learning (ICML)
  2. Calabrese, R. and Osmetti, S.A. 2013. Modeling small and medium enterprise loan defaults as rare even: the generalized extreme value regression model, Journal of Applied Statistics, Vol. 40, No. 6, pp. 1172-1188
  3. Calabrese, R and Giudici, P. 2015. Estimating bank default with generalized extreme value regression models, Journal of the Operational Research Society, Vol. 66, pp. 1783-1792
  4. Friedman, J., Hastie, T., and Tibshirani, R. 2010. Regularization path for generalized linear models via coordinate descent, Journal of Statistical Software, Vol. 33, No. 1, pp. 1-22
  5. Härdle, W.K., Prastyo, D. D., and Hafner, C. M. 2014. "Support vector machines with evolutionary model selection for default prediction," in The Oxford Handbook of Applied Nonparametric and Semiparametric Econometrics, J. S. Racine, L. Su, and A. Ullah (Eds). New York: Oxford University Press
  6. Haerdle, W.K., and D. D. Prastyo. 2014. "Embedded predictor selection for default risk calculation: a Southeast Asian industry study," in Handbook of Asian Finance, Vol. 1: Financial Market and Sovereign Wealth Funds, D. L. K. Chuen and G. N. Gregoriou (Eds). San Diego. Academic Press, 2014, pp 131-148
  7. Haerdle, W.K., Werwatz, A., Mueller M., and Sperlich S. 2004. "Nonparametric density estimation," In Nonparametric and Semiparametric Models. Berlin: Springer. pp.39-83
  8. Hanley, J.A. and McNeil, B. J. 1982. The meaning and use of the area under a receiver operating characteristic (ROC) curve, Radiology, Vol. 143, No. 1, pp. 29-36
  9. Kowarik, A. and Templ, M. 2016. Imputation with the R package VIM, Journal of Statistical Software, Vol. 74, No. 7, pp. 1-16
  10. Natasha, A., Prastyo, D. D., and Suhartono. 2019. Credit Scoring to Classify Consumer Loan using Machine Learning, AIP Conference Proceedings 2194, 020070
  11. Prastyo, D.D., Miranti, T. and Iriawan, N. 2017. Survival analysis of companies' delisting time in Indonesian Stock Exchange using Bayesian multiple-period logit approach, Malaysian Journal of Fundamental and Applied Sciences, Vol. 13, No. 4-1, pp. 425-429
  12. Prastyo, D.D., Rucy, Y. N., Sigalingging, A. D. C., Suhartono, et al. 2018. Micro and Macro Determinants of Delisting and Liquidity in Indonesian Stock Market: A Time-dependent Covariate of Survival Cox Approach, MATEMATIKA: Malaysian Journal of Industrial and Applied Mathematics, Vol. 34, pp. 73-81
  13. Randhawa, K., Loo, C.K., Seera, M., Lim, C.P., et al. 2018. Credit card fraud detection using AdaBoost and majority voting, IEEE Access, Vol. 6, pp. 14277-14284
  14. Tibshirani, R. 1996. Regression shrinkage and selection via the Lasso, Journal of the Royal Statistical Society, Series B, Vol. 58, pp 267-288
  15. Wang, C., Han, D., Liu, Q., and Luo, S. 2019. A deep learning approach for credit scoring of peer-to-peer lending using attention mechanism LSTM, IEEE Access, Vol. 7, pp. 2161-2168
  16. Zou, H. and Hastie, T. 2005. Regularization and variable selection via the elastic net, Journal of the Royal Statistical Society, Series B, Vol. 67, No. 2. pp 301-320
  17. Zhu, J., Rosset, S., Hastie, T., and Tibshirani, R. 2004. 1-norm support vector machine, Proceeding of Advances in Neural Information Processing System 16, 2004

Last update:

No citation recorded.

Last update: 2024-05-18 12:33:32

No citation recorded.