CONSTRUCTION ELEMENTS OF BANKRUPTCY PREDICTION MODELS IN MULTI–DIMENSIONAL EARLY WARNING SYSTEMS
Abstract: A consequence of the inevitability of the occurrence of internal crises in companies is the taking of preventive action in place of purely remedial measures. In this respect a significant role is played by Early Warning Systems (EWS), which provide early warning information and financial threat assessments relating to the continuation of operations and bankruptcy not only for individual companies as such but also for companies as a whole. The limitations of existing models used for EWS purposes have led to the elaboration of new models, estimated on one of the largest hitherto drawn up teaching sets, constituting more than five hundred bankrupt companies. These models also distinguish themselves through the application of innovative methods and precise instruments; the structural concept of these models for multi–dimensional EWS purposes, accompanied by elements used for predicting, is presented in this article.
Key words: bankruptcy, bankruptcy prediction, early warning system, logistic regression model
JEL Codes: G33
Thinking of early warning systems one has in mind a solution which takes into account the systematic diagnosis of a company’s financial standing, appropriately early identification and interpretation of warning signals and assessment of the risk of financial threat to the continuation of activities, company bankruptcy and companies as a whole (classes, groups, departments, sections).The result of such assessments may be the taking of action in support of company restructuring processes; this is because if restructuring is performed effectively and in a timely manner, it can save a company, and the expenses incurred by the State for this objective are much lower than the economic and social costs caused by the collapse of companies .
Bearing the above in mind a concept of an institutionalised instrument was elaborated; its basic function in the first component, constituting the core of the Early Warning System, is the current monitoring and analysis of company standing as a whole and informing about the need to take remedial action in terms of entities (as a whole), in the face of their deteriorating economic and financial standing.
The remaining elements are: programming, implementation and evaluation components.
more in full version
The elaboration of the concept of an institutionalised instrument, whose core is the national Early Warning System, required a new approach to three basic problems. The first involved the elimination of delay between the appearance of signs of threat to the financial continuation of operations and company bankruptcy, and the obtaining of analyses and assessments in this area. The solution which permits a switch to quarterly frequency analyses leads to more frequent updating of the estimated level of threat, giving the possibility of taking implementation measures.
The second solved problem involved the prolongation by one year of the period of predicting the level of threat of bankruptcy by applying prediction methods in estimating the probability of bankruptcy. This was achieved by forecasting prediction model variables, and not the measure of threat itself.
Thirdly, limitation through imitation transfer and the disadvantages of known bankruptcy prediction models have led to the basis of estimation of new models based on Firth’s logistic regression on a unique set of more than three thousand companies, with the application of innovatory methods and tools of own construction and high levels of prediction capacity, which speaks in favour of the uniqueness of this solution concerning the modelling of level of risk of bankruptcy.
The results of empirical studies have been presented in separate publications [e.g. 9, pp. 27–31].
ELEMENTY KONSTRUKCJI MODELI PREDYKCJI UPADŁOŚCI W WIELOWYMIAROWYCH SYSTEMACH WCZESNEGO OSTRZEGANIA
Abstrakt: Konsekwencją nieuchronności występowania kryzysów wewnętrznych w przedsiębiorstwach jest podejmowanie działań zapobiegawczych w miejsce działań tylko o charakterze sanacyjnym. Istotną rolę spełniają w tym względzie Systemy Wczesnego Ostrzegania (SWO), dostarczając wyprzedzających informacji oraz ocen zagrożenia finansowego kontynuacji działalności i upadłości nie tylko dla pojedynczych przedsiębiorstw, ale i ich zbiorowości.Ograniczenia zastosowania istniejących modeli dla potrzeb SWO, skłoniły do opracowania nowych modeli, estymowanych na jednym z największych jak dotychczas zbiorze uczącym, liczącym ponad pięćset przedsiębiorstw upadłych. Modele te wyróżnia także zastosowanie innowacyjnych metod i narzędzi szczegółowych, a ideę ich konstrukcji dla potrzeb wielowymiarowego SWO wraz z elementami prognozowania zawarto w niniejszym artykule.