论文部分内容阅读
文献中,在险值估计方法一般基于线性假设,但是该假设在实际中很难满足,需要为此提出非线性的在险值估计方法。与以往传统模型一般假定变化发生在“时间”点上不同,门限双自回归(TDAR)因状态空间的不同而建立不同模型来对非对称性、结构变点等非线性现象进行刻画,并同时允许均值和波动率过程的结构变化。本文首次基于TDAR建立TDAR-VaR方法,并对上证指数和香港恒生指数进行了实证研究和对杠杆效应进行了分析。实证分析发现TDAR-VaR较好地预测了市场风险。
In the literature, the method of risk estimation is generally based on linear assumptions, but this assumption is difficult to meet in practice. Therefore, we need to propose a non-linear method of VaR estimation. Different from the conventional model, which assumes that the change occurs at the “time” point, the threshold dual autoregression (TDAR) establishes different models due to the different state space to describe the non-linear phenomena such as asymmetry and structural change point. While allowing structural changes in the mean and volatility processes. This article first establishes the TDAR-VaR method based on TDAR, and conducts empirical research on the Shanghai Composite Index and Hong Kong’s Hang Seng Index and analyzes the leverage effect. Empirical analysis shows that TDAR-VaR better predicts market risk.