| Risk is ubiquitous in the financial market.The value at risk is one of the most commonly used risk measurement tools in the field of financial measurement.The value at risk of investment portfolio is also one of the most important indicators required by financial institutions for global capital allocation and portfolio management.Value-at-risk measures market risk by providing a point estimate of the most serious financial losses that the portfolio may suffer within a fixed time frame.At the same time,the volatility of different assets and markets are interconnected,and it is crucial to predict the correlation between asset returns.Considering the linkage effect between different assets and value-at-risk is a point estimator to measure market risk,this paper studies the interval estimation of the value-at-risk of the asset portfolio under the framework of the DCC-GARCH model.We take the correlation between the returns of different assets into consideration in the calculation of the risk value.Based on residual Bootstrap we propose a method of value-at-risk interval estimation for portfolio.Monte Carlo simulation was used to study the limited sample nature of the method,and the effects of different error distribution assumptions and different Bootstrap sample numbers were evaluated.Finally,it is compared with the existing method based on the univariate model.The simulation results show that when the error distribution is multivariate normal distribution and multivariate Laplace distribution,the coverage rate in each case reaches more than 85%,and when the error distribution is multivariate t distribution,the simulation coverage rate is about 80%.In each case,the simulation coverage of this method is much higher than the existing univariate model-based methods,indicating that this method can be applied to the interval estimation of the portfolio value at risk.In the empirical part,we took the Shanghai Stock Exchange Index,the S&P 500 Index and the Hang Seng Index as examples,constructed a virtual asset index portfolio,and used the rolling window method to obtain the confidence interval for the one-step forecast of the insurance value during the test period,with a good prediction effect. |
