The Impact of Financial Stress on New Energy Vehicles Industry from Cross-correlation to Explainable Machine Learning: Proof from China

This paper employed a combination of the multifractal detrended cross-correlation analysis (MFDCCA) method and interpretable machine learning methods to investigate the correlation between financial stress and the volatility of new energy vehicles stock at both macro and micro levels. The multifractal detrended cross-correlation analysis (MFDCCA) method reveals a significant anti-persistent cross-correlation. Seven benchmark models are applied to identify the contribution of financial stress and its sub-indicators on predicting of the new energy vehicles stock market volatility, while the impact of financial stress on the new energy vehicles industry during the US–China trade war, COVID-19 pandemic, and Russia–Ukraine conflict is explored. Results suggest that financial stress can improve forecasting performance, with support vector regression (SVR) outperforming other models in accuracy, stability, and robustness. SHapley Additive exPlanations (SHAP) is used to explain forecasts, and the empirical analysis reveals that the contributions of financial stress exhibit heterogeneity under different periods. The study’s results can help policymakers, regulators and investors in identifying the current financial risks in the new energy vehicle industry and formulating relevant policies to help its sustainable development by regulating financial stress during various crisis events.