Financial Distress Prediction Using MARS and Logistic Regression: Evidence from Indonesia

The increasing uncertainty in the business environment has intensified the need for reliable financial distress prediction models, particularly within the manufacturing sector, which plays a strategic role in economic development. The study aims to compare the effectiveness of logistic regression and multivariate adaptive regression splines (hereinafter – MARS) in predicting financial distress among manufacturing companies listed on the Indonesian stock exchange. The study employs a quantitative research design with purposive sampling, using data from 70 manufacturing firms and 210 firm-year observations over the 2022–2025 period. Financial distress is examined using four key financial indicators, namely current ratio, total liabilities to total assets, return on assets, and sales to total assets. The findings reveal that both models are statistically valid; however, MARS outperforms logistic regression in terms of predictive accuracy, achieving an overall classification rate of 82.4 % compared to 65.7 %. Logistic regression revealed a statistically significant effect of return on assets only on financial distress (p = 0.003; Exp(B) = 0.006), whereas MARS showed that all financial indicators under consideration contributed to the predictive model. These findings highlight the importance of profitability as a primary determinant of financial distress and suggest that MARS provides a more robust framework for developing early warning systems and supporting financial decision-making. The practical significance of the study lies in the potential for businesses, investors, creditors, and regulatory authorities to use the results to identify financial risks early, which is important for the economy’s stable development.