Small and Medium Sized Enterprise Credit Customer’s Insolvency Prediction by using Two Group Discriminant Analysis: A Case Study of Bangladesh

This study develops a discriminant function to predict the creditworthiness of Small and Medium Enterprises (SMEs) in Bangladesh, aiming to distinguish between default and non-default borrowers. Data were collected from 20 SME credit customers of a large commercial bank, evenly split between defaulters and non-defaulters. Six independent variables related to financial and socio-economic characteristics were analyzed to build the predictive model. The discriminant analysis identified Loan Position Against Portfolio (LPAP), loan amount, and number of employees as the most significant factors influencing credit risk classification. The estimated discriminant function demonstrated statistical significance at the 1% level, indicating a strong model fit. When applied to the dataset, the model correctly classified 95% of the original sample cases and maintained a 70% accuracy rate under cross-validation, confirming its robustness and practical utility. The function enables calculation of a discriminant score (Z-score) for new loan applicants, which can be used to predict their likelihood of default. Positive scores indicate higher default risk, while negative scores suggest lower risk. Implementing this discriminant function can improve credit risk management by providing a systematic, data-driven tool to assist banks in loan approval decisions, risk-based pricing, and resource allocation. The model offers a cost-effective approach to reduce non-performing loans and enhance portfolio quality. However, the study is limited by its small sample size and scope, which calls for further research with larger and more diverse datasets. Future studies could also explore the inclusion of additional variables and advanced modeling techniques to improve predictive accuracy and adaptability across different economic conditions.