Miklós Virág

This article demonstrates that crisis conditions significantly impact the efficacy of corporate bankruptcy prediction models developed with pre-crisis data pertaining to large corporates in the Visegrad Four (V4) countries. Empirical research includes 245,974 firm-year observations and 3091 failure occurrences. Model development was accomplished using a combination of chi-squared automatic interaction detection (CHAID) decision trees and logistic regression (LR) methods, constituting a novel technique in V4-level bankruptcy prediction. Model performance was evaluated by area under the ROC curve (AUROC) analysis. Evidence from V4 large corporates indicates that the classification accuracy of the corporate failure prediction model developed in the pre-crisis period substantially declines during a crisis; thus, it was essential to create a new point-in-time model based only on crisis data. The results indicate that the model design underwent substantial alterations compared to the pre-crisis model. The current ratio is the strongest predictor; however, country and sector classifications also significantly contribute to elucidating corporate failure throughout the crisis era.

This article provides an empirical analysis aimed at evaluating the financial trends and disparities at the sector level within the Balkan and Eastern European region. The dataset encompasses a period of nine years and comprises more than 20 million firm-year observations from 24 industries in 21 countries. It uses 19 financial ratios to assess sectoral performance. In the empirical investigation, trend analysis and the two-step cluster analysis methods were used. Following the global financial crisis, a significant proportion of financial ratios exhibited favorable trends, indicating robust business and economic circumstances. Nevertheless, this trajectory was temporarily disrupted in 2020 due to the onset of the COVID-19 pandemic. By 2021, the financial ratios had reverted back to their historical patterns. Country membership, margin, liquidity, trade turnover, profitability, and leverage ratios are the most effective variables for explaining differences in sectoral performance. Sector membership is a comparatively less influential factor. Although this study effectively identified significant disparities in financial ratio profiles, it does not suggest that companies in the most developed countries in the region attain the most favorable financial performance. Stakeholders who have a vested interest in this region should carefully contemplate the ramifications of the findings from this study.

Research background: Visegrad Four (V4)-level corporate failure modeling is a relatively new research topic. Three empirical studies have been published in scientific journals since 2018 by Slovak researchers in this field. As COVID-19 has affected financial positions and failure behavior patterns of V4 companies, it has become necessary to examine the impact of COVID-19 on V4-level corporate failure prediction, and to reconsider circumstances for modeling. Purpose of the article: This article presents a novel V4-level corporate failure prediction model concentrating on V4 large corporates using long-term historical data. It considers financial data between 2012 and 2020 and failure events occur-ring between 2013 and 2021. Incorporation of COVID-19 financial data and failure experience is a novel research object, as financial data from 2020 and failure events from 2021 have only recently become available. Methods: Model development is accomplished by using a combination of Chi-squared Automatic Interaction Detection (CHAID) and Logistic Regression (logit) methods. Results are evaluated by the Area Under ROC (AUROC) analysis meth-od. A total of 245,974 firm-year observations and 3,091 failure events comprise the modeling database collected from the Moody’s Analytics Orbis Europe data source. Findings & value added: Testing of the pre-COVID-19 model with actual data reveals that results are favorable until 2019 financial data and 2020 failures are included, whereupon predictive power becomes substantially lower for 2020 financial data and 2021 failures. It is thus essential to prepare a new point-in-time model considering only very recent data to lead to a different model design. Testing of this model is successful in that COVID-19 failures indicate that sector classification is significant in predicting overall corporate failure. Results demonstrate that in the current turbulent economic environment failure prediction models must be reviewed at least on an annual basis.

The article provides a comprehensive review regarding the theoretical approaches, methodologies and empirical researches of corporate bankruptcy prediction, laying emphasis on the 30-year development history of Hungarian empirical results. In ex-socialist countries corporate bankruptcy prediction became possible more than 20 years later compared to the western countries, however, based on the historical development of corporate bankruptcy prediction after the political system change it can be argued that it has already caught up to the level of international best practice. Throughout the development history of Hungarian bankruptcy prediction, it can be tracked how the initial, small, cross-sectional sample and classic methodology-based bankruptcy prediction has evolved to today’s corporate rating systems meeting the requirements of the dynamic, through-the-cycle economic capital calculation models. Contemporary methodological development is characterized by the domination of artificial intelligence, data mining, machine learning, and hybrid modelling. On the basis of empirical results, the article draws several normative proposals how to assemble a bankruptcy prediction database and select the right classification method(s) to accomplish efficient corporate bankruptcy prediction.

In our work, we compare the predictive power of different bankruptcy prediction models built on financial indicators calculable from businesses’ accounting data on the database of the first Hungarian bankruptcy model. For modelling, we use data-mining methods often applied in bankruptcy prediction: neural networks (NN), support vector machines (SVM) and the rough set theory (RST) capable of rule-based classification. The point of departure for our comparative analysis is the practical finding that black-box-type data-mining methods typically show better classification performance than models whose results are easy to interpret, i.e. there seems to be a kind of trade-off between the interpretability and predictive power of bankruptcy models. Empirical results lead us to conclude that the RST approach can be a competitive alternative to black-box-type SVM and NN models. In our research, we did not find any major trade-off between the interpretability and predictive performance of bankruptcy models on the database of the first Hungarian bankruptcy model.

In our study we rely on a data mining procedure known as support vector machine (SVM) on the database of the first Hungarian bankruptcy model. The models constructed are then contrasted with the results of earlier bankruptcy models with the use of classification accuracy and the area under the ROC curve. In using the SVM technique, in addition to conventional kernel functions, we also examine the possibilities of applying the ANOVA kernel function and take a detailed look at data preparation tasks recommended in using the SVM method (handling of outliers). The results of the models assembled suggest that a significant improvement of classification accuracy can be achieved on the database of the first Hungarian bankruptcy model when using the SVM method as opposed to neural networks.

A tanulmány a csődmodellezési szakirodalomban alulreprezentált, ugyanakkor kiváló klaszterezési teljesítményt nyújtó többdimenziós skálázás csődmodellezési alkalmazását mutatja be empirikus vizsgálattal alátámasztva. Az empirikus vizsgálat alapján levonható a következtetés, hogy a többdimenziós skálázás a logisztikus regresszióelemzéssel kombinálva klasszifikációs képességben és interpretálhatóságban bármelyik alternatív csőd-előrejelzési módszerrel felveszi a versenyt.

A tanulmány célja nyilvánosan hozzáférhető éves beszámoló adatok alapján az iparági hovatartozást is figyelembe vevő, magas besorolási pontosságú csődelőrejelzési modellek elkészítése. A tanulmány a négy leggyakrabban alkalmazott csődelőrejelzési módszersegítségével, ugyanarra az adatbázisra épít csődmodelleket, elvégzi összehasonlító elemzésüket és értékeli azok megbízhatóságát.