Probabilistic Classification of PN17 Companies Based on Financial Indicators: A Logit Model Approach

PN17 classification is synonymous with companies that fail to comply with the Bursa Malaysia laws, thus serving as an early warning for potential financial issues. Accurately determining the financial status of PN17 companies often requires an extensive review of the reports and procedures. This study employed a logit model to identify the factors contributing to PN17 classification with a combination of financial distress indicators: stock volatility, leverage, liquidity, profitability, and probability of default (PD). A sample of financial data from 46 companies listed on Bursa Malaysia, comprising both PN17 and non-PN17 companies, from 2017 to 2022 was analysed using logistic regression. The results reveal that stock volatility, liquidity, and leverage are statistically significant to PN17 classification at a 95% confidence interval. The logit function obtained from the logistic regression analysis was able to classify PN17 and non-PN17 companies with 85.1% accuracy. Nonetheless, the accuracy of classifying PN17 (31.9%) is lower compared to non-PN17 (96.4%). This may be due to class imbalance bias (PN17 and non-PN17) and the negative relationship found in leverage, which is contradicted by the financial theory where higher leverage means higher risk. The ROC-AUC analysis supports the struggle in classifying PN17 due to the counterintuitive relationship in leverage. This suggests a complex relationship in leverage that requires further investigation into its linearity, bimodality, and relations with other predictors. Overall, PN17 classification can be improved with stock volatility as the strongest predictor, followed by liquidity and leverage. Meanwhile, profitability and PD were found to be insignificant to PN17 classification. These findings highlight the gaps in the existing literature regarding the predictive power of the logit model when incorporating additional financial indicators (stock volatility and PD) and understanding the relationships of the financial factors in classifying the minority PN17 class. Regulators might consider giving more weight to stock volatility for identifying financially distressed in PN17 companies. Future research may consider data augmentation methods and other model approaches to address the class imbalance bias and counterintuitive relationship in leverage in order to produce a more robust model.