Analysis Of Financial Risk Of Portfolio Based On Pair Copula-GJR-CVaR Credit Model

It is of great significance for portfolio risk measurement and selection that one knows the real probability distribution and dependence of financial asset returns.Systematic research on the measurement of portfolio coupled risk(such as market risk and credit risk) is discussed in this thesis, based on research on Financial Mathematics,Analysis of Financial Time Series theory, and etc., associated with the present research of portfolio theory at home and abroad.This paper uses Pair Copula based on vine structure to describe the inner-dependence structure of portfolio returns, which can not only effectively solve problems resulting from the assumption that the portfolio returns follow a multivariate normal distribution, but also have less limitations than multivariate Copula on the research and applications, showing high accuracy and flexibility. Considering the leptokurtosis, fat tails, volatility clustering effect of financial time series, this paper builds the marginal distribution of each asset return using GJR model, describes the credit default via probability of default estimated by ‘Distance-to-Default’ according to KMV model, then constructs the multivariate distribution of portfolio returns based on Pair Copula-GJR-CVa R credit model. Combining Monte Carlo method to simulate each asset’s loss scenarios, CVa R as a coherent measure of risk is used to estimate the risk of the portfolio. The result of an empirical study shows that CVa R estimated by multivariate pair Copula model much better than multivariate t-Copula model, which reflects the advantage of coupled risks measurement of the portfolio applying multivariate Pair Copula-GJR-CVa R credit model.The main innovative points are summarized as follows:(1) Apply Pair Copula decomposition model to describe the dependence structure of portfolio returns,breaking through the limitations of traditional multivariate Copula method;(2)Propose a method based on Pair Copula-GJR-CVa R credit model to estimate coupled risks of portfolio, and emphasize that credit risk should not be ignored when measuring the risk of the portfolio, then find a simple way to measure credit risk by assessing the probability of default using market prices;(3) The empirical study verifies that CVa R as a coherent measure of risk is better than Va R in terms of risk measurement.