Credit scoring with carbon footprint

<p dir="ltr">This thesis investigates the integration of carbon footprint metrics into credit risk assessment for small and medium-sized enterprises (SMEs) using bank transaction data. By developing a transaction-based carbon footprint estimation framework, the study quantifies SMEs' carbon emissions and examines their predictive value in loan default, delinquency, and screening decisions. Empirical results from logistic regressions and XGBoost models show that several carbon indicators—such as carbon intensity, dependence, and emissions—are significantly associated with credit risk outcomes, even after controlling for traditional financial and firm-level characteristics. The analysis further reveals that these carbon metrics offer incremental predictive value, particularly for borrowers without formal credit scores, thereby supporting financial inclusion. Moreover, the inclusion of carbon variables helps reduce gender disparity in predictive models, suggesting fairness-enhancing properties. These findings demonstrate the viability of incorporating environmental data into lending decisions and highlight the broader potential of carbon indicators in sustainable and inclusive finance.</p>