| Abstract [eng] |
The purpose of this thesis is to evaluate the accuracy of a modified bankruptcy prediction model which was achieved by utilizing macroeconomic variables, in the construction sector. The main components of the work are introduction, exposition (consisting of scientific literature analysis, research methodology, and research itself), conclusions, and proposals. During the analysis of the scientific literature, the concept of bankruptcy and its essential elements – causes and consequences, are presented. Also, the examination of the existing bankruptcy forecasting methodologies along with their shortcomings and advantages, and the analysis of previous scientific research is carried out in order to distinguish previously identified macroeconomic variables and forecasting models that were significant for bankruptcy prediction. Taking into account the analysis of the scientific literature, a work research methodology is created and a bankruptcy prediction model development study adapted to the construction sector is carried out. First of all, the analysis of the construction sector reveals an increased risk of bankruptcy in the aforementioned sector due to its specifics (cyclicality, long value chains, long cash conversion period, etc.). Then, when conducting a study of the significance of macroeconomic indicators, it is determined that the number of bankruptcies in the construction sector significantly correlates with housing prices, registration of construction companies, building permits, production indexes, and the balance sheet indicator of the government sector. From the selected financial indicators, a primary bankruptcy prediction model is created, reaching 83.05 percent classification accuracy (96.31 percent for healthy and 54.4 percent for bankrupt/undergoing bankruptcy companies). Finally, combining the original model and significant macroeconomic variables results in a modified bankruptcy prediction model reaching 82.34 percent classification accuracy (93.57 percent for healthy and 58.08 percent for bankrupt/undergoing bankruptcy companies). Significant independent variables forming the modified model: total debt ratio, asset turnover, profitability before taxation, and housing price index. The conducted study does not show an overall increase in the accuracy of the bankruptcy prediction of construction sector companies after applying macroeconomic variables, but taking into account the determined significance of the analyzed variables and the increase in the accuracy of the classification of bankrupt/undergoing bankruptcy companies by applying the bankruptcy prediction model modified with macroeconomic variables, it can be concluded that the study shows the predictive potential of such indicators, which should be investigated further. The work is summarized by the presented conclusions and suggestions. In the author's opinion, this work proves that, in order to increase the accuracy of bankruptcy prediction models, it is important to take into account the potential of using other types of variables (not only financial indicators), the importance of applying specific variables and other restrictions taking into account the specifics of the situation, etc., i.e. the results of the work and the obtained conclusions can be a guideline for conducting other researches in a similar field and for practical analyzes of bankruptcy risk. |
