Developing bankruptcy prediction models for Ukrainian insurance companies in post-crisis period

Abstract

In the article there have been applied the methods of  discriminant analysis and logistic regression to build bankruptcy prediction models for Ukrainian insurance companies based on financial reporting data. During the modelling the potential advantages (unambiguous quantification of results and objectification of analysis) and the significant restraints (limited array of observations and quality of information) of using both methods, the analysis and comparison of statistical significance and forecasting power of the developed models are conducted. The ways of improving the processes of bankruptcy forecasting for Ukrainian underwriters are determined. The results of the research show that the majority of insurance companies which function in Ukrainian have a medium financial stability level (have approximately the same probability to go bankrupt or continue functioning), while some companies, which took first places in industry ratings, appear to be classified as potential bankrupts and vice versa. The obtained results allow to assert that prediction of bankruptcy of insurance companies using the chosen methods is not perfect in Ukrainian reality due to the peculiarities of the insurance market; therefore, there is a need for developing and applying alternative methods which are, primarily, less demanding of information base volume.