本研究运用结构函数和层次模型对资产波动多时间标度条件下的自相似性和层次结构特征进行的研究发现:在结构函数和相对结构函数度量框架下,资产波动仅在一定时间标度和波幅上表现出了自相似性;当运用普适性更为广泛的归一化相对结构函数度量时,资产波动在整个时间标度范围内表现出显著的自相似性。此外,资产波动在SL层次结构模型中表现出离散特征,而在新构建的归一化层次结构模型度量框架中具有显著幂次函数关系,表明资产波动间具有不同于由最高激发态(最强间歇结构)控制的SL层次结构的逐级传递的“相邻”层次结构特征。
The multiscale self-similarity and hierarchy properties of asset volatility are investigated by structure functions and hierarchical models. The available empirical results show that volatility manifests self-similarity only for certain scales and amplitudes measured by the ordinary and relative structure functions. While the normalized relative structure function with more universality is further utilized, volatility exhibits the significant self-similarity at the whole scales. In addition, the significant power relations based on the new normalized hierarchical model, contrary to the scattered phenomena of SL model, conclude that the volatilities imply the ‘neighboring' hierarchy different from the highest stimulated states (the strongest intermittent structures) of SL hierarchy.