Managerial Applications of Neural Networks: The Case of Bank Failure Predictions

TLDR

Neural networks model nonlinear discriminant functions through interconnected processing units. The study introduces a neural‑net method for discriminant analysis in business research. The authors compare the neural‑net approach to linear classifiers, logistic regression, kNN, and ID3 using bank default data. Empirical results demonstrate that neural nets offer superior predictive accuracy, adaptability, and robustness for bank failure prediction, though limitations as a general modeling tool are noted.

Abstract

This paper introduces a neural-net approach to perform discriminant analysis in business research. A neural net represents a nonlinear discriminant function as a pattern of connections between its processing units. Using bank default data, the neural-net approach is compared with linear classifier, logistic regression, kNN, and ID3. Empirical results show that neural nets is a promising method of evaluating bank conditions in terms of predictive accuracy, adaptability, and robustness. Limitations of using neural nets as a general modeling tool are also discussed.