DEMOCRATIZING DIAGNOSTICS: AN EXPLAINABLE AI FRAMEWORK FOR PREDICTING CASH SHORTAGES IN INDIAN MSMEs

Purpose: This study develops an explainable AI (XAI) framework to predict and diagnose cash shortages in Indian MSMEs, addressing limitations of inaccurate or "black box" existing models. Design: Using a cross-sectional survey of 1,230 MSMEs, we apply XGBoost machine learning with SHAP (SHapley Additive exPlanations) interpretation for global and local explainability. Findings: The XGBoost model achieved outstanding predictive accuracy (AUC=0.93), significantly outperforming logistic regression (AUC=0.79). SHAP analysis identified delayed receivables, low capacity utilization, and absence of formal cash flow planning as key predictors. The framework provides actionable diagnostics for individual MSMEs. Originality: This research offers a triple contribution: methodological innovation through XAI integration, theoretical expansion beyond financial ratios, and practical value via a transparent diagnostic tool for entrepreneurs, lenders, and policymakers to enhance MSME resilience. KEYWORDS: Explainable AI (XAI), SHAP Model, MSMEs, Cash Shortage, Predictive Analytics, Financial Distress.