Corporate Bankruptcy Prediction Using Artificial Neural Network

Financial distress is a condition where a company has difficulty paying off its financial obligations to its creditors. Failing to relieve financial distress can lead to bankruptcy, which is very costly to the firm’s creditors and investors, as well as its employees and citizens of the community where it is located. As a result, corporate bankruptcy predication is of great interest to both academics and practitioners. While a number of bankruptcy prediction models have been published, Altman’s Z-score model is still commonly used by practitioners as a standard to evaluate a firm’s financial health after almost fifty years since it was published in 1968. In this thesis, I explore the use of artificial neural networks to improve upon existing predictive techniques. My model shows strong discriminating power in a Receiver Operating Characteristics analysis, and outperforms the hazard model and the distance-to-default model in detecting bankruptcies.