Cross-Border spillover of imported sovereign risk to China: Key factors identification based on XGBoost-SHAP explainable machine learning algorithm

•Quantifying China's imported sovereign risk by using QVAR model.•Developing countries are major sources of China's imported sovereign risk in normal condition.•Novel use of XGBoost-SHAP explainable machine learning algorithm to explore risk spillover channels.•Transmission channels of imported sovereign risk are heterogeneous under different conditions. This paper examines the imported sovereign risk from G20 countries to China and the transmission channels. To explore how sovereign risk transmit in different conditions, the study constructs QVAR model to measure China's imported sovereign risk and use XGBoost-SHAP machine learning algorithm to explain the key channels. The research reveals key findings. First, external sovereign risk importing to China in upper and lower conditions are asymmetric. Second, developing countries have more significant risk spillovers to China comparing to advanced countries in normal condition. Third, there are significant differences in risk transmission channels under different quantile conditions.