Estimation of financial failure by artificial neural networks and multivariate statistical analysis techniques: An application in the İstanbul Stock Exchange

Financial failure is a very important issue for in terms of businesses to continue their activities. Especially it negatively affects the business and bussiness's creditors, employees, suppliers, consumers and all stakeholders. At this point, estimation models that predict financial failure are developed so that the business and its stakeholders are not exposed to the negative effects of financial failure. Financial failure prediction models, it prohibitive attribute the failure to result in bankruptcy. The aim of this reseach is to develop forecasting models that can predict the financial failure of companies traded in the manifacturing sector on the Istanbul stock exchange one year in advance, using logistic regression analysis and artifical neural networks method and to determine the appropriate model by comparing the predictive power of the developed models. Within the scope of the research, the financial ratios calculated by using the income statements and balance sheets of 140 manufacturing sector enterprises traded in the Istanbul stock exchange for the years 2015 - 2020 were used as independent variables in the models. Logistic regression analysis was performed using IBM SPSS Statistics 21, and artificial neural network method was performed using MATLAB (R2021b) program. As a result of the research, it was concluded that the artificial neural network model (95.7 %) had a higher power to predict financial failure one year in advance compared to the logistic regression model (92.1 %)