论文部分内容阅读
考虑正负资产收益对分位数冲击的不对称性提出间接TARCH-CAViaR模型.CAViaR一般模型中递归分位回归方程的非线性和非连续可微性是参数估计的一个难题,基于含有尺度参数的不对称拉普拉斯分布作为误差过程,指出将尺度参数固定为常数会导致不对称拉普拉斯分布随机变量的方差存在最小正值的限制,与实际金融数据分布不符;进而提出采用贝叶斯分析和马尔科夫链蒙特卡罗模拟方法,估计间接TARCH-CAViaR模型的参数,并可获得尺度参数的合理估计.实证研究中对上证指数的市场风险演化模式进行了测算,动态分位检验、后验测试表明模型VaR预测效果理想,消息冲击曲线表明上海股市好坏消息对市场风险的冲击是不对称的,且这种影响作用在不同置信水平市场风险中表现得有显著差异.
The indirect TARCH-CAViaR model is proposed considering the asymmetry of positive and negative return on quantile impact.Nonlinearity and discontinuity of regression equation of recursive quantile in general model of CVA RR is a difficult problem in parameter estimation, Asymmetric Laplace distribution is regarded as the error process. It is pointed out that fixing the scale parameter as a constant results in the minimum positive variance of the variance of the unsymmetrical Laplacian random variable, which is inconsistent with the actual financial data distribution. Furthermore, Yayitz analysis and Markov chain Monte Carlo simulation method to estimate the parameters of indirect TARCH-CAViaR model and to get a reasonable estimate of scale parameters.The Empirical Study estimates the evolution of market risk in Shanghai Stock Exchange Index, The tests and the posterior tests show that the model VaR has a good forecasting effect. The news impulse curve shows that the impact of the Shanghai stock market on the market risk is asymmetric, and this effect has significant differences in the market risk of different confidence levels.