Descriptive statistics results.

With growing uncertainty in global trade, improving access to domestic capital markets has become an important way to manage financial risk spillovers. This study examines how the registration system reform affects the finance sector's risk spillovers to other 10 sectors in China’s stock market using a dual machine learning model. The findings include: (1) The finance sector's risk spillovers vary over time and are heterogeneous. Risk spillovers rapidly intensify two months after the outbreak of the COVID-19 pandemic, with the average net ∆CoVaR value changing from −0.001 to −0.006. The finance sector mainly accepts risk from the public utility sector and exports risk to the other 9 sectors, with the highest spillovers going to the communication and information technology sectors, showing extreme net ΔCoVaR values around −0.02. (2) The registration system reform increases the finance sector's risk spillover effect, and this conclusion remains the same after a series of robustness tests. (3) Sector heterogeneity tests show that the reform boosts the finance sector's risk spillovers to cyclical sectors and sectors with a low proportion of strategic emerging companies but reduces risk spillovers to midstream and supportive sectors. Finally, some suggestions and implications are proposed.

伴随全球贸易不确定性日益加剧，提升国内资本市场可及性已成为管控金融风险溢出的重要路径。本研究采用双重机器学习模型，考察中国股票市场中注册制改革对金融行业向其余10个行业的风险溢出效应的影响。研究结果如下：(1) 金融行业的风险溢出效应具有时变性与异质性。新冠疫情爆发后两个月，风险溢出效应迅速加剧，平均净ΔCoVaR值由-0.001变动至-0.006。金融行业主要承接公用事业行业的风险，并向其余9个行业输出风险，其中对通信与信息技术行业的溢出效应最强，净ΔCoVaR极值约为-0.02。(2) 注册制改革强化了金融行业的风险溢出效应，该结论经一系列稳健性检验后依然成立。(3) 行业异质性检验显示，改革提升了金融行业对周期性行业以及战略新兴企业占比较低行业的风险溢出效应，但降低了对中游及配套行业的风险溢出。最后，本研究提出若干政策建议与研究启示。