Robust Linear Discriminant Analysis with Modified One-Step M-Estimator Qn Scale for Classifying Financial Distress in Banks: Case Study

Nabila Putri
Parmikanti, Kankan
Gusriani, Nurul

Eksakta : Berkala Ilmiah Bidang MIPA
Vol. 25 No. 02 (2024): Eksakta : Berkala Ilmiah Bidang MIPA (E-ISSN : 2549-7464)

The COVID-19 pandemic has significantly disrupted the banking sector, leading to a decline in profit growth as an indicator of financial distress. Bank financial health can be evaluated using the RGEC (Risk Profile, Good Corporate Governance, Earnings, Capital) analysis. While Linear Discriminant Analysis (LDA) ideally requires normality and homogeneity of covariance matrices, financial data often fail to meet these assumptions. Therefore, this study employs robust linear discriminant analysis using the Modified One-Step M-Estimator with Qn scale estimator (MOM-Qn) to classify ‘distress’ and ‘non-distress’ bank conditions. Given these challenges, this study acts as a preventive measure for banks to evaluate financial health simultaneously. The objective is to provide a robust discriminant function for more accurate and stable classification, particularly in the presence of outliers. It focuses on conventional private banks listed on the Indonesia Stock Exchange (IDX) during December 2021-2022. The results show a classification accuracy of 69.23% and a Press’s Q value of 11.53846, indicating the method’s effectiveness in classifying real financial data.