Who Pays the Price? Inequity in Municipal Borrowing Costs: Evidence from Machine Learning

• In Event: Algorithmic Governance: Machine Learning and Spatial Analytics in Public Policy

Abstract

Municipal bond markets serve as the primary mechanism through which local governments finance critical public infrastructure, yet systematic disparities in borrowing costs across jurisdictions raise profound equity concerns within the field of public finance. While theories of fiscal federalism suggest that bond pricing should reflect objective credit risk, recent scholarship indicates that structural inequities may distort these financial outcomes. This study explores the potential for hidden inequities in municipal bond markets by developing an out-of-sample validation framework to test whether conventional econometric models and gradient boosting machine learning algorithms can accurately predict gross underwriting spreads and credit ratings for historically disadvantaged jurisdictions.Drawing on a comprehensive nationwide panel of municipal bond issuances spanning over 20 years, integrated with U.S. Census demographic records, we train and evaluate predictive performance on held-out observations. Building on the foundational frameworks of structural fiscal disparities, our results reveal that borrowing costs in lower-income and majority-minority jurisdictions vary significantly across different modeling approaches. We find that conventional linear methods often fail to capture the complex, non-linear burdens placed on these communities, leading to systematic underestimation of the "equity tax" paid by marginalized areas.To recover causal interpretability, we utilize SHAP-based feature decomposition to identify the specific interaction effects between racial composition, median income levels, and negotiated sale structures that drive asymmetric prediction errors. Our analysis reveals that the intersection of high minority populations and specific market entry methods creates a unique disadvantage that is often overlooked in traditional risk assessments. Extending the recent work on systemic inequity in municipal bonds, these findings demonstrate that structural biases are embedded deeper in the market than previously recognized. By exposing these discrepancies, this research provides critical policy implications for credit rating governance and offers a roadmap for designing an equity-centered municipal finance system that ensures fair capital access for all communities regardless of demographic composition.

Authors

• Yoon-Jung Choi, Florida International University
  Non-Presenting Co-Author

• Il Hwan Chung, Sungkyunkwan University
  Presenting Author

• Byeong Ho LEE
  Non-Presenting Co-Author