Prediction of financial strength ratings using machine learning and conventional techniques

  • Received November 1, 2017;
    Accepted December 19, 2017;
    Published December 26, 2017
  • Author(s)
  • DOI
  • Article Info
    Volume 14 2017, Issue #4, pp. 194-211
  • Cited by
    7 articles

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

Financial strength ratings (FSRs) have become more significant particularly since the recent financial crisis of 2007–2009 where rating agencies failed to forecast defaults and the downgrade of some banks. The aim of this paper is to predict Capital Intelligence banks’ financial strength ratings (FSRs) group membership using machine learning and conventional techniques. Here the authors use five different statistical techniques, namely CHAID, CART, multilayer-perceptron neural networks, discriminant analysis and logistic regression. They also use three different evaluation criteria namely average correct classification rate, misclassification cost and gains charts. The data are collected from Bankscope database for the Middle Eastern commercial banks by reference to the first decade of the 21st century. The findings show that when predicting bank FSRs during the period 2007–2009, discriminant analysis is surprisingly superior to all other techniques used in this paper. When only machine learning techniques are used, CHAID outperform other techniques. In addition, the findings highlight that when a random sample is used to predict bank FSRs, CART outperform all other techniques. The evaluation criteria have confirmed the findings and both CART and discriminant analysis are superior to other techniques in predicting bank FSRs. This has implications for Middle Eastern banks, as the authors would suggest that improving their bank FSR can improve their presence in the market.

view full abstract hide full abstract
    • Figure 1. An evaluation chart for the five predictive statistical modelling techniques
    • Figure 2. MLP feed-forward NN architecture (one hidden layer)
    • Figure 3. Gain charts for machine learning techniques using 2007–2009 testing sub-sample1 and 33% testing sub-sample2
    • Figure 4. Gain charts for conventional techniques using 2007–2009 testing sub-sample1 and 33% testing sub-sample2
    • Table 1. Descriptive statistics for banks, by country and whether rated by CI based on size (ln total assets)
    • Table 2. A synopsis of CI bank FSRs numerical ratings and rating categories
    • Table 3. Correlation matrix for predictor variables
    • Table 4. Descriptive statistics for the 14 financial indicators
    • Table 5. Group statistics for the 14 financial indicators
    • Table 6. Descriptive statistics for non-financial indicators
    • Table 7. Classification results for the three machine learning modelling techniques
    • Table 8. Classification results for the two conventional modelling techniques
    • Table 9. Error rates, estimated misclassification costs and gain chart ranking for all the five modelling techniques