Mixed Uncertainty Information and Volatility in the Green Bond Market Based on a New Two-regime GARCH-MIDAS Extension Model

Purpose This study pioneers a novel two-regime Generalized Autoregressive Conditional Heteroskedasticity–Mixed Data Sampling (GARCH-MIDAS) extension model to disentangle the dual dynamics of traditional drivers and mixed uncertainty indicators in green bond market volatility. Unlike conventional single-regime frameworks, our approach uniquely integrates macroeconomic fundamentals, financial market conditions, and bond performance with multidimensional uncertainty indices (economic policy, geopolitical risks, and market volatility) through a regime-switching mechanism. Design/methodology/approach This paper applies the GARCH-MIDAS model to analyse the data. First, traditional and uncertainty variables are incorporated into the single-regime GARCH-MIDAS model to study their impact on the green bond market's long-term components. Second, the two-regime GARCH-MIDAS model is implemented to study the combined impact of traditional and uncertainty factors at different frequencies on the green bond market. Findings Our findings suggest that uncertainties have a lasting influence on the green bond market. The results also indicate that traditional factors are still the leading drivers of the green bond market. Notably, the two-regime model captures more information than the benchmark GARCH-MIDAS model. Research limitations/implications While this paper advances understanding of China's green bond market, three limitations warrant attention. First, the reliance on nationally specific indicators necessitates regional calibration for global applications given divergent policy frameworks. Second, the monthly frequency of uncertainty indices may understate persistent low-frequency shocks, while the 2018–2023 sample period's concurrent crises (trade tensions and the COVID-19) could amplify measured uncertainty effects. Third, although we incorporate multidimensional uncertainties (economic policy, geopolitical, financial), other sector-specific factors, such as climate technology risks remain unexplored. Originality/value The first marginal contribution of this paper is the construction of a multidimensional indicator system for the green bond market that integrates traditional factors and uncertainty variables. The second marginal contribution lies in effectively addressing the problem of varying data frequencies in the green bond market and other indicator systems by constructing the GARCH-MIDAS model, thereby avoiding information loss or result distortion caused by converting data of different frequencies to the same frequency. The third marginal contribution is the extension from the original single-regime model to a two-regime model, which links traditional indicator variables with uncertainty variables, addresses the limitations of single-regime models in explaining multiple variables, and examines the combined impact of traditional and uncertainty variables on the green bond market.