Assessment of the Bankruptcy Risk in the Hotel Industry as a Condition of the COVID-19 Crisis Using Time-Delay Neural Networks

Abstract

In this paper we demonstrate a new conceptual framework in the application of multilayer perceptron (MLP) artificial neural networks (ANNs) to bankruptcy risk prediction using different time-delay neural network (TDNN) models to assess Altman’s EM Z“-score risk zones of firms for a sample of 100 companies operating in the hotel industry in the Republic of Serbia. Hence, the accuracies of 9580 forecasting ANNs trained for the period 2016 to 2021 are analyzed, and the impact of various input parameters of different ANN models on their forecasting accuracy is investigated, including Altman’s bankruptcy risk indicators, market and internal nonfinancial indicators, the lengths of the learning periods of the ANNs and of their input parameters, and the K-means clusters of risk zones. Based on this research, 11 stability indicators (SIs) for the years under analysis are formulated, which represent the generalization capabilities of ANN models, i.e., differences in the generalization errors between the preceding period and the year for which zone assessment is given; these are seen as a consequence of structural changes at the industry level that occurred during the relevant year. SIs are validated through comparison with the relative strength index (RSI) for descriptive indicators of Altman’s model, and high correlation is found. Special focus is placed on the identification of the stability in 2020 in order to assess the impact of the COVID-19 crisis during that year. It is established that despite the fact that the development of bankruptcy risk in the hotel industry in the Republic of Serbia is a highly volatile process, the largest changes in the analyzed period occurred in 2020, i.e., the potential applications of ANNs for forecasting zones in 2020 are limited.

Sustainability of Transport Sector Companies: Bankruptcy Prediction Based on Artificial Intelligence

“…According to Shetty et al [17], the 1990s ushered in a new phase in the evolution of business failure prediction models, introducing innovative methods, particularly artificial intelligence algorithms, including neural networks (Špiler et al [41]) and decision trees. These artificial intelligence-based techniques offer promising alternatives to traditional statistical models, addressing their principal shortcomings (Dong and Chen [42]).…”