Demystifying AI in Financial Services: An Explainable Machine Learning Framework for Tourism SME Credit Risk Assessment in Emerging Markets

Credit risk assessment for Small and Medium-sized Enterprises (SMEs) remains a critical challenge in emerging markets, particularly within volatile sectors such as tourism and hospitality. While advanced artificial intelligence (AI) and machine learning (ML) models have demonstrated superior predictive accuracy over traditional credit scoring methods, their "black-box" nature hinders their adoption by financial institutions that require regulatory transparency. Drawing upon the intersection of technology management and tourism innovation, this paper proposes an Explainable AI (XAI) framework tailored for assessing the creditworthiness of tourism SMEs. By integrating non-traditional digital footprint data (e.g., Online Travel Agency metrics, digital booking volumes) with standard financial indicators, we develop a predictive model using Extreme Gradient Boosting (XGBoost). We then apply SHapley Additive exPlanations (SHAP) to interpret the model’s outputs, transforming opaque predictions into actionable, transparent insights. Utilizing a synthesized dataset of 8,542 tourism-related SMEs across Southeast Asia, the experimental results demonstrate that the XGBoost model achieves an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.89, outperforming traditional Logistic Regression. More importantly, the SHAP integration allows for both global and local interpretability, revealing that digital operational metrics are increasingly vital indicators of SME resilience. This study contributes to the ICCMETS domain by providing a practical, transparent algorithmic management tool that promotes financial inclusion for tourism SMEs while satisfying the regulatory and risk management prerequisites of modern financial institutions.