The role of major markets in predicting the U.S. municipal green bond market performance: New evidence from machine learning models

Abstract

This study aims to examine the impact of major markets (conventional bond, stock, and energy commodity markets) on the performance of U.S. municipal green bonds, accounting for economic risk factors. We also investigate the interactions between major markets and economic risk factors. In our analysis, we employ machine learning techniques, such as Random Forest (RF), Extreme Gradient Boosting (XGBoost), CatBoost, and Light Gradient Boosting Machine (LightGBM) models and choose the best one depending on its performance measures. The analysis considers two sub-periods: the pre-COVID-19 and COVID-19 crisis periods. The Shapley Additive Explanations (SHAP) method is used to better interpret machine learning predictions. The main results indicate that, on the one hand, higher values of conventional bond market performance lead to higher municipal green bond market performance levels for the pre-COVID-19 period. On the other hand, compared to the impact of the conventional bond market, the stock and energy commodity markets have a more significant impact on the U.S. municipal green bonds during the COVID-19 crisis period. Last but not least, we find that major markets interact frequently with some macroeconomic risks and uncertainties during the COVID-19 crisis. Our research provides new insights into the investment potential of municipal green bonds.