Research On The Method Of Testing Financial Crisis Contagion

Under the background of economic globalization and financial liberalization, since the 21st century, the eruption frequency, spread and influence of financial crisis has increased. The far reaching influence, wide spreading range and strong destruction of financial crises are still fresh in our memory. Since the subprime mortgage crisis and the European debt crisis, the global economy is moving slowly with sluggish growth and the risk spillover effects continue to be strengthened. Whether in theory or in practice, it is very important to study the existence and characteristics of financial congtagion from different perspectives.Financial contagion is usually shown as a significant increase in correlation between financial markets. Today, financial innovation is developed and the new financial instruments are popping up. Risk in financial market is becoming more and more complex. The financial asset return series show that excess kurtosis and fat tail characteristics. The correlation between financial markets can not be characterized by a simple linear relationship. The dependence between financial markets shows a nonlinear, non-symmetric, extreme tail and dynamic characteristics. There are some problems on the methods of the existing financial contagion test. We need to develop new statistical or econometric models to study the existence and characteristics of financial crisis contagion and the market risks from different perspectives.Based on the reviewed definitions of financial contagion, we summarize the formation mechanism theory and the transmission mechanism theory on financial crisis, and analyze the formation reasons and infection mechanism of financial crisis. Under the ground of the theorems above, the common test methods of financial contagion are summarized and the shortcomings of the existing test methods are analyzed. The financial contagion is attributed to the transfer of risk in this paper. From the view of the extreme tail dependence and the risk spillover, we use new statistical methods to study the contagion and features of the U.S. subprime mortgage crisis and the European debt crisis.Firstly, the entire sample period is divided into three subsamples:pre crisis, crisis and post crisis. We estimate multivariable multiquantiles conditional autoregressive value at risk model (MVMQ-CAViaR) in precrisis, crisis and post crisis periods, respectively. From the perspective of risk transformation, we study the financial contagion from American equity market to equity markets in China, Japan, Britain, France and Germany during the subprime mortgage crisis, and the degree and direction of the contagion and the changes of stock market risk. The findings show that stock market risk in our sample increases during subprime crisis. There are risk spillovers and financial contagion from American equity market to the makets in Japan, Britain. France and Germany. But. there is no evidence of financial contagion from American equity market to Chinese equity market. After the crisis, there are different characteristics of spillovers of market risk between American and other tested stock markets.Combining local polynomial regression model with quantile association regression model, we develop time varying quantile association regression model and give its estimation and prediction method. The correction of the model is proved via Mento Carlo simulaitons. In the framework of the model, we compare the estimation of accuration between quantile odds raio and quanitle probability ratio. The results show that the quantile probability ratio is more accurate than quantile odds ratio in describing the time vaying dependence between two variables in in-sample estimation and out-sample prediction. However, as for the methodology, two measures describe the different aspects of dependence. In the framework of time varying quantile regression model, from the pespectives of extime risk dependence, we test the financial contagion between international equity markets during American subprime crisis and European debt crisis and analyze the characteristics of the contagion. Meanwhile, we compare the degree of contagion from the crisis original markets to different equity markets. The findings show that there is evidence of contagion from American stock market to all tested markets during subprime crisis with the global characteristics. However, the differences in the degree and time span of the contagion to different stock markets are also found. For European debt crisis, we only find the evidence of contagion from Greece equity market to some tested equity markets. The European debt crisis contagion is with regional characteristics and there are differences in degree and time span for the contagion. In addition, American Subprime crisis are more influential than European debt crisis from the degree and time span of contagion. The dependence between stock markets exhibits asymmetric and time vaying characteristics. Furthermore, the degree of dependence at different quanitle levels varies and with quantile level moved to the center of the distribution, the degree of dependence decreases and the extreme tail dependence is stonger. With regard to the accuracy of in-sample estimation and out-sample prediction, quantile odds ratio is inferior to quantile probability ratio. However, quantile odds ratio decribes different aspects of dependence between two variables and used to analyze the existence of contagion of subprime crisis and European debt crisis. The results are consistent with those of quanitle probability ratio.Furthermore, to prove the effectiveness of quantile probability ratio in the description of dependence between other financial assets, we study the time varying dependence between oil and exchange rate and the impact of subprime crisis on the dependence. The findings show that the dependence between them is enhanced. Different combinations of oil and exchange rate exhibit different dependent characteristics. Moreover, the degree of depdence decreases with the quantile level moved to the center of distribution. In contrast, different from the asymmetric dependence between stock markets, all pairs of oil and foreign exchange are shown to be symmetric tail dependence.At last, we summarize the paper, and give some inspirations about prevention of financial contagion to China. Research prospects are also presented.