随着全球各国金融市场联系的不断紧密，防范金融风险传染已成为各国研究的重要议题。中美股市作为全球最大的两个金融市场，对于两者之间风险相依结构的研究不仅有助于我国金融体系的完善以及防范风险能力的增强，还有利于世界金融市场的健康发展、协调和稳定。 前人的研究主要集中于通过格兰杰检验等线性模型来测度中美股市间风险的相关性，但金融风险多以非线性形式传染，因此本文选取能够较好处理非线性关系的Copula函数，利用其尾部相依系数来表示两国股市的风险相依关系，并在此基础上，通过能较好测度溢出性风险的CoVaR模型表示两国之间的条件风险价值（CoVaR值），从而体现出两国股市之间的风险溢出关系。两种模型分别从股市联动性和相互影响性两方面完善了股市的风险相依结构。 从过程来看，本文选取美股标普500指数、中国内地沪深300指数、中国香港恒生指数作为研究对象，先对数据进行ARMA-GARCH-GED模型的拟合得到各股指的边缘分布，通过对四种Copula函数的参数比较，得到最优函数为t Copula函数，但由于其自身对于测度尾部系数的局限性，仅用其测度市场间的线性相关参数与非线性相关参数，并选用次优的SJC Copula函数测度尾部相依系数，其中前两个系数表示两市场间整体分布的相关关系，尾部相依系数表示两市场尾部分布的相关关系，即市场间风险相关关系。在测度风险相依关系的基础上，本文通过CoVaR模型运用分位数回归的方法计算股指的%CoVaR，从而得出中美股市间的风险溢出程度，并且对于在不同分位数下的条件风险价值进行了测量。 本文的创新之处主要在于：（1）在本文研究背景下，对传统ARMA模型做出了改进，在均值方程加入了对方股指线性项。（2）在分析相关参数时采用了总体与局部分析相结合的方法。（3）在分析尾部相依系数时对静态Copula与时变Copula方法进行了混合应用。（4）引入了广义Granger模型对CoVaR模型结果的准确性加以佐证。 结果表明：（1）整体上看，无论从线性角度还是非线性角度，美股与港股之间的联动性强于内地股市，并且美股与港股的尾部相依系数分别为0.1014和0.0897，显著大于美股与内地股市的0.0088和0.0174，说明美股与港股间同时出现极端风险的概率更大。（2）美股与中国内地股市间下尾相关系数显著大于上尾相关系数，表明出现暴跌的概率大于暴涨的概率。（3）美股与中国内地股市间在出现极端风险时尾部相依系数变化幅度为0.65，显著大于美股与港股的0.4，说明美股与内地股市间的风险关系波动性更强。（4）分时段看，在危机发生前期，股市之间的非线性参数以及尾部相依系数会显著增加，另外风险溢出效应也会增强，三者在危机时期达到极值，相对来说，线性相关参数变化趋势为之相反。（5）风险溢出程度主要取决于两市场的状态，当金融风险直接涉及到两市场时，相互之间的风险溢出价值最大。总体上看，美股与港股之间的上涨风险溢出更强，美股与内地股市的下跌风险溢出更强。（6）在中美贸易争端期间，美股与中国内地股市%CoVaR值仅有0.98%和0.48%，显著小于其它两个危机时期的相对风险溢出值，这从侧面反映出了近几年我国对于金融业改革已初具成效，金融市场风险抵御能力逐渐增强。 综合研究结果，可以对有关部门做出如下建议：（1）稳步推进我国金融市场的对外开放，增加我国金融市场的自由度，降低门槛以促进国内外资本流动，另外我国内地市场应加大与美国市场、香港市场等发达资本市场之间的联系，使金融市场逐步实现国际化。（2）健全国内金融风险预警机制与防范措施，建立合理的风险传染监测指标和预测体系，切实保障金融市场的安全性。（3）应加大金融创新的力度，逐步丰富金融工具的种类并完善相关制度体系，不断拓宽金融交易的媒介，鼓励数字经济、互联网金融的发展。（4）继续坚定不移地推进供给侧改革，大力支持实体经济和民营企业的发展，增强市场对于风险冲击的免疫力。

With the continuous closeness of financial markets in various countries around the world, prevention of financial risk contagion has become an important issue for national research. As the world’s two largest financial markets, the Sino-US stock market research on the risk-dependent structure between the two countries not only contributes to the improvement of China’s financial system and the ability to prevent risks, but also contributes to the healthy development and coordination of the world's financial markets. Previous studies have focused on measuring the correlation between Chinese and American stocks through linear models such as Granger test, but financial risks are mostly transmitted in a nonlinear form. Therefore, this paper selects the Copula function that can better deal with nonlinear relationships. Using the tail dependence coefficient to represent the risk-dependent relationship between the two stock markets, and based on this, the CoVaR model that can better measure the spillover risk represents the conditional risk value (CoVaR value) between the two countries, thus reflecting the two countries. Risk spillover relationship between stock markets. The two models have improved the risk-dependent structure of the stock market from the perspective of stock market linkage and mutual influence. From the process point of view, this paper selects the US stock S&P 500 index, the China mainland CSI 300 index, and the Hong Kong Hang Seng Index as the research object. Firstly, the ARMA-GARCH-GED model is fitted to the data to obtain the edge distribution of each stock index. Comparing the parameters of the four Copula functions, the optimal function is the t Copula function, but because of its own limitations on the measurement tail coefficient, it only uses the linear correlation parameters and nonlinear correlation parameters between the market, and selects the sub-optimal The SJC Copula function measures the tail dependence coefficient, in which the first two coefficients represent the correlation between the two markets, and the tail dependence coefficient represents the correlation between the two market tails, that is, the risk correlation between markets. On the basis of measuring the risk-dependent relationship, this paper uses the CoVaR model to calculate the %CoVaR of the stock index by using the quantile regression method, so as to obtain the risk spillover between the Chinese and American stock markets, and the conditional risk value under different quantiles. Measurements were taken. The main innovations of this paper are as follows: (1) In the context of this paper, the traditional ARMA model has been improved, and the linear term of the other stock index has been added to the mean value equation. (2) A method combining total and local analysis is used in the analysis of relevant parameters. (3) The static Copula and time-varying Copula methods are mixed and applied when analyzing the tail dependence coefficient. (4) Introducing the generalized Granger model to prove the accuracy of the CoVaR model results. The results show that: (1) Overall, the linkage between US stocks and Hong Kong stocks is stronger than that of the mainland stock market, both linearly and non-linearly, and the tail-dependence coefficients of US stocks and Hong Kong stocks are 0.1014 and 0.0897, respectively, which are significantly larger than US stocks and the mainland stock market's 0.0088 and 0.0174 indicates that there is a greater probability of simultaneous extreme risks between US stocks and Hong Kong stocks. (2) The correlation coefficient between the US stock market and the Chinese mainland stock market is significantly larger than the upper tail correlation coefficient, indicating that the probability of a sharp fall is greater than the probability of a sharp rise. (3) The tail-dependent coefficient of the US stock market and the Chinese mainland stock market is 0.65, which is significantly larger than the US stock and Hong Kong stocks, which indicates that the risk relationship between the US stock market and the mainland stock market is more volatile. (4) According to the time period, in the early stage of the crisis, the nonlinear parameters and the tail dependence coefficient between the stock market will increase significantly, and the risk spillover effect will also increase. The three will reach extreme values during the crisis period, relatively linearly related. The trend of parameter changes is the opposite. (5) The degree of risk spillover mainly depends on the state of the two markets. When financial risks directly relate to the two markets, the risk spillover value between them is the largest. On the whole, the risk of rising risk between US stocks and Hong Kong stocks is stronger, and the risk of falling US stocks and mainland stock markets is stronger. (6) During the Sino-US trade dispute, the USCo and China mainland stock market's %CoVaR value is only 0.98% and 0.48%, which is significantly less than the relative risk spillover value of the other two crisis periods, which reflects the recent years in China. The financial industry reform has begun to take effect, and the financial market risk resilience has gradually increased. Based on the results of the comprehensive study, the following suggestions can be made to relevant departments: (1) Steadily promote the opening up of China's financial market, increase the freedom of China's financial market, lower the threshold to promote domestic and foreign capital flows, and the domestic market should increase the links between developed markets such as the US market and the Hong Kong market have gradually made the financial market international. (2) Improve domestic financial risk early warning mechanisms and preventive measures, establish reasonable risk transmission monitoring indicators and forecasting systems, and effectively protect the security of financial markets. (3) It is necessary to intensify financial innovation, gradually enrich the types of financial instruments and improve related institutional systems, continuously expand the medium of financial transactions, and encourage the development of digital economy and internet finance. (4) Continue to unswervingly push forward supply-side reforms, vigorously support the development of the real economy and private enterprises, and enhance the market's immunity to risk shocks.