Research On The Mechanism Of Financial Contagion Based On Dynamical Interdependece

The late 20th century and early 21st century is a peak period of financial crisis. The financial crisis becomes increasingly frequent and serious. With a higher level of international economic integration, the financial contagion effects have gradually revealed. The most obvious example is USA sub-prime crisis in 2007, which have global infection, and lead to the global financial crisis in 2009. Therefore, more and more researchers work over the financial contagion. The study of the financial contagion theory is gradually differentiated out of the financial crisis theory to become to an independent direction of international economics. The research on the machenism of financial contagion has very large theoretical significance. The global financial turmoil had a significant impact on China's economy, so the research of financial contagion can help managing financial contagion, and has a very large theoretical and practical significance.The existing theory of financial contagion is not sufficient to resolve nonlinear transmission characteristics of financial contagion. The empirical study is mainly through the variance-covariance matrix or the linear correlation coefficient method to analyze the degree of market-related, or using VAR or other methods. These methods could not repit the nonlinear characteristics of financial contagion, and could not be real-time during the period of financial crisis. Therefore, we try to use nonlinear interdependence to characterize nonlinear dynamics of the multiple financial time series. We hope this method could be a better empirical method for the nonlinear machenism of financial contagion.In this paper, we propose a financial contagion model based on the nonlinear dynamics. The financial market is treated as a dynamic system with different levels of volatility. The dynamic systems are coupled, and the coupling of these markets is due to freedom cross-border movement of trade links, or of the press, funds and financial assets. The cross-border transmission of financial shocks and financial contagion could be seen as fluctuations nonlinear interdependence between the nonlinear dynamic systems. So, we can use nonlinear interdependence to depict the nonlinear characteristics of the financial contagion.We use the nonlinear mutual prediction algorithm to study the nonlinear interdependence of the dynamical systems. Before the nonlinear mutual prediction, we have to test the existence of nonlinear of the testing system. And we choose the optimal time delayτand the best embedding dimension d . On this basis we use the nonlinear mutual prediction algorithm to predict. The prediction accuracy of original nonlinear mutual prediction algorithm is not high enough, and during the financial crisis, the samples are small, so we use support vector machine to improve the original algorithm.We use the improved nonlinear mutual prediction algorithm on the Southeast Asian financial crisis. We use the financial market time series data of different countries (or regions) to nonlinear mutual predict and gain the nonlinear mutual prediction measure between every two sequences. The forecast results of the original algorithm have strong random, so we range the nonlinear mutual predict measure from three aspects to improve, to make it more suitable for analytical requirements of financial contagion.We use the improved nonlinear mutual predict measure on the global financial turmoil in 2007-2009. By analyzing the obtained nonlinear mutual prediction measures, we conclude that : First, the improved nonlinear mutual prediction algorithm by support vector machine did improve the prediction accuracy; Second, during the financial crisis, the nonlinear mutual prediction measures of the financial market time series data of between different countries (or regions) occurred strong fluctuations, such fluctuations show that during the financial crisis, nonlinear interdependence between the countries (or regions) dramatic shock, that is the coupling of nonlinear dynamic systems of the countries dramatic fluctuations. That is during the financial crisis, the contagion happened through various channels between the countries (or regions). Due to the consequences of infection are different, the coupling shocks. Finally, we compared the similarities and differences on financial contagion of the Southeast Asian financial crisis and the global financial turmoil in 2007-2009 from a dynamics point of view. On the basis of empirical research, we calculate the nonlinear mutual predict measure between China and main trading partners. This nonlinear mutual predict measure could be used as an early warning indicationr of financial contagion. So it is an important tool for China's response to the financial contagion.