Financial Distress Prediction in Indonesian Companies Using Artificial Neural Networks (ANN)

Abstract

Aims: This study aims to predict financial distress in hotel, restaurant, and tourism subsector companies listed on the Indonesia Stock Exchange using artificial neural networks (ANN) and to examine the influence of financial ratios—current ratio (CR), return on assets (ROA), earnings per share (EPS), debt-to-assets ratio (DAR), and shareholders’ equity ratio (SER)—on financial distress. Population and Sample: The population consists of 23 hotel, restaurant, and tourism companies listed on the Indonesia Stock Exchange during 2018–2022. Training data were obtained from 40 public companies (20 bankrupt and 20 non-bankrupt) across various countries, while test data were drawn from the financial statements of the Indonesian companies in the selected subsector. Methodology: The study employed a multilayer perceptron ANN with a backpropagation algorithm to predict financial distress. The ANN architecture was optimized using training and testing datasets, followed by logistic regression analysis to test the simultaneous and partial effects of the financial ratios on financial distress. Results: The ANN model achieved high prediction accuracy with an optimized 25-30-1 architecture. Results indicate that 15 companies are predicted not to experience financial distress, while 8 companies are predicted to face financial distress. Simultaneously, CR, ROA, EPS, DAR, and SER significantly affect financial distress. But partially, only ROA and SER have significant effects. Conclusion: ANN provides an effective early warning tool to predict financial distress in hotel, restaurant, and tourism subsector companies. By monitoring ROA and SER in particular, companies can better anticipate financial difficulties and adopt appropriate financial strategies to reduce bankruptcy risk.