Transformations  in
Business & Economics

THE RESEARCH OF RELIABILITY OF BANKRUPTCY PREDICTION MODELS IN LITHUANIAN COMPANIES
Rytis Krusinskas, Ausrine Lakstutiene, Jurgita Stankeviciene

ABSTRACT.The abundance of research investigating and assessing various bankruptcy prediction models that best characterize the situation of a company and allow to predict probability of its bankruptcy, confirms the relevance of this topic and discloses the depth of the scientific problem. To assess reliability of bankruptcy prediction models in Lithuanian companies, this research applies models of linear discriminant analysis, logistic regression and decision tree. The empirical research consists of two parts. In the first part, Lithuanian bankrupted companies are analyzed applying models of Altman, Springate, Taffler and Tisshaw, Zavgren, Chesser, Grigaravicius. It has been found that during five-year period, logistic regression model developed by Chesser is the most accurate and model of linear discriminant analysis proposed by Taffler and Tisshaw is the least accurate. In the second part, bankruptcy prediction models have been applied in case of operating Lithuanian public companies; the research results allow recommending logistic regression models for bankruptcy prediction in Lithuanian companies. The models of decision tree applied in the second part of the research predict probability of bankruptcy differently, therefore, application of the decision tree developed by Mackevicius and Silvanaviciute (2006) is recommended in case of economy growth. However, in case of economic downturn, model of Frydman, Altman, Kao (1985) is more accurate .

KEYWORDS: probability of bankruptcy, bankruptcy prediction models, classical statistical models, models of artificial intelligence, Lithuania.

JEL classification: G33, G39, C29, C53.