VaR (Value at Risk)在风险管理领域一直深受银行业和金融机构的重视,GARCH模型对VaR的测量是一个重要研究领域。然而在实际应用中,利用传统参数GARCH模型建模时需要指定条件分布形式,一旦分布指定错误将会导致模型失效。因此,我们在标...VaR (Value at Risk)在风险管理领域一直深受银行业和金融机构的重视,GARCH模型对VaR的测量是一个重要研究领域。然而在实际应用中,利用传统参数GARCH模型建模时需要指定条件分布形式,一旦分布指定错误将会导致模型失效。因此,我们在标准的GARCH(1,1)模型下,结合累积经验分布函数对残差进行修正,避免了传统参数分布由于事先指定错误带来的模型风险。经过实证研究发现,我们采用的方法比指定参数分布下的标准GARCH(1,1)模型在测量VaR方面有了很大改进,其失败频率和相对误差都显著降低。因此,文中采用这种创新的尾部分布形式在估计VaR值方面具有一定的实际应用价值。展开更多
The paper is going to introduce some methods about select distributional model、select an estimation technique、iterate to minimize objective function、record estimated parameters、select one or more models which had ...The paper is going to introduce some methods about select distributional model、select an estimation technique、iterate to minimize objective function、record estimated parameters、select one or more models which had low value of the objective function and test of fit of selected model by empirical distribution function、mean of residual life function、minimum distance and minimum chi square.This should prove especially useful to those readers who want to set up a computer system to perform the model fitting operation.展开更多
文摘VaR (Value at Risk)在风险管理领域一直深受银行业和金融机构的重视,GARCH模型对VaR的测量是一个重要研究领域。然而在实际应用中,利用传统参数GARCH模型建模时需要指定条件分布形式,一旦分布指定错误将会导致模型失效。因此,我们在标准的GARCH(1,1)模型下,结合累积经验分布函数对残差进行修正,避免了传统参数分布由于事先指定错误带来的模型风险。经过实证研究发现,我们采用的方法比指定参数分布下的标准GARCH(1,1)模型在测量VaR方面有了很大改进,其失败频率和相对误差都显著降低。因此,文中采用这种创新的尾部分布形式在估计VaR值方面具有一定的实际应用价值。
文摘The paper is going to introduce some methods about select distributional model、select an estimation technique、iterate to minimize objective function、record estimated parameters、select one or more models which had low value of the objective function and test of fit of selected model by empirical distribution function、mean of residual life function、minimum distance and minimum chi square.This should prove especially useful to those readers who want to set up a computer system to perform the model fitting operation.
