Predictive Modeling of Debt Capacity for MSMEs Using Financial Indicators and Machine Learning

Accurately estimating debt capacity is critical for improving credit access and financial sustainability among Micro, Small, and Medium Enterprises (MSMEs). Traditional assessment methods often rely on limited financial ratios and subjective judgment, leading to inefficiencies and credit constraints. This study proposes a machine learning-based framework to predict the debt capacity of MSMEs using firm-specific financial, operational, and demographic attributes. Multiple supervised learning models are developed and evaluated to capture nonlinear relationships and complex interactions among variables that influence borrowing potential. Model performance is assessed using standard evaluation metrics to identify the most reliable predictive approach. The findings demonstrate that machine learning techniques significantly enhance prediction accuracy compared to conventional methods, offering robust and scalable decision support for lenders and policymakers. The proposed approach contributes to data-driven credit risk assessment and supports inclusive financing by enabling more objective and timely debt capacity evaluation for MSMEs.