CORPORATE BANKRUPTCY PREDICTION USING MACHINE LEARNING METHODS: THE CASE OF THE USA

Abstract

Many of the international hospitals and health institutions have faced serious financial difficulties with the Covid-19 epidemic that emerged at the end of 2019. In the coming years, these financial difficulties are expected to cause more bankruptcies. Being able to predict company bankruptcies is important to protect company partners, investors, company creditors and the continuity of the healthcare industry. Thanks to accurate forecasts in this area, company managers can prevent company bankruptcies by taking the necessary precautions and investors can limit their losses. This study aims to build a model that can predict bankruptcies in the health sector by using artificial neural networks (ANN). For the sample of this study, 23 companies in the health sector that declared bankruptcy between 01.01.2018 and 31.12.2020 in the USA, and 23 companies that were in the same period and the same sector but had no financial problems were selected as the control group. 30 financial ratios belonging to these companies were used as input data of the research. In the study, artificial neural networks (ANN) were chosen as the method. According to the results of the research, the correct classification rate of the artificial neural network models created using the training set data was 100%. The correct classification rate of artificial neural network models created using test set data was 90%. According to the results of the research, ANNs are promising for the prediction of company bankruptcies with their high classification success and ease of use. Therefore, it is recommended to be used by both researchers and investors.