Risk Analysis Based On Copula-Kernel Model

Along with the financial integration and economic globalization, the rapid development of financial markets have showed unprecedented volatility, financial risks are becoming increasingly complex and diversified, then the attendant risk management techniques develop quickly under the pressure. As being the base of control and management risk, the market risk measurement is also an important part of financial theory. VaR is an effective way to measure the financial market risk and have the advantage that the investment risk can be described as a dominant function and export the image data on the maximum possible loss, which is incomparable by the traditional methods. However, the common VaR calculations are based on single assets, and assuming that it follows normal distribution, But a large number of empirical studies have shown that the tail of return distribution of assets is generally thicker than the normal distribution, the extreme values are more likely to occur, while the relationship among assets always shows a nonlinear condition. Therefore, the way of using non-parametric density estimation method to fit the single asset return distribution, combining Copula theory to describe the multi-linear correlation between assets is a meaningful study.It is firstly reviewed the financial risk management technology and related process, issuing the historical context. Analyzing the structure of copula model and the domestic and international research results on VaR, and introducing the Monte Carlo simulation technology and some methods to calculate VaR. Besides Kernel theory is also discussed as the non-parameter estimation process, as well as the expansion of two-dimensional Copula and some comparisons.The paper proposes a Copula-Kernel model, calculate the value of VaR based on two-dimensional along with Monte Carlo simulation, compared with the existing calculation method. Empirical analyses show that the structure of the Copula-Kernel Model in Financial Risk Analysis is feasible and effective.In addition, this paper try to expand Monte Carlo simulation from two-dimensional VaR space to high-dimensional space, and applying it to the multi-asset portfolio analysis, the evidence shows that the implementation and calculation of high-dimensional non-Normal Copula model and the is feasible and meaningful.Finally, the paper summarizes the benefits applying Copula theory to financial markets and extending to high dimension and raise and further research directions...