FInanční modely predikce bankrotu

This bachelor's thesis focuses on the prediction of financial failure for medium-sized enterprises in the European Union using probabilistic regression models. The main objective of the thesis is to verify the predictive power of selected financial indicators and to compare the performance of the linear probability model with nonlinear logit and probit models. The empirical analysis is based on corporate accounting data obtained from the Orbis database. Due to the asymmetric distribution of firms in financial distress, the data were methodologically adjusted to ensure stable estimation of nonlinear model parameters and reliable identification of key determinants of financial distress. The predictive ability of the models was evaluated using classification accuracy, ROC analysis, and marginal effects. The results indicate that both the logit and probit models exhibit strong discriminatory power, with the area under the ROC curve (AUC) reaching approximately 0.86. Solvency and profitability are identified as the most significant determinants of financial failure. Firm size, measured by sales volume, does not show incremental predictive value when controlling for solvency, whereas the capital base (Total Assets) remains statistically significant as a financial buffer. The consistency of the results between both nonlinear models confirms their robustness and stability. The thesis contributes to the empirical understanding of the financial structure of medium-sized enterprises in the European Union and confirms that probabilistic regression models serve as an appropriate tool for the early identification of financial distress in this segment.