中国国债收益率的多标度分析

The Multi-scaling Characters Analysis of China Treasury Bonds Market

首先通过采用一种基于各个国债之间相关性的新测度方法，对17个国债日收益率时间序列进行聚类分析。利用Matlab在临界的阈值δ～0．0595时．将17个国债分成6类。然后，研究了每一个国债日收益率时间序列的价格波动结构．结果表明中国的国债市场也表现出一定程度上的多标度特征；同时还发现属于同一类型国债的多标度特征具有很强的自相似性．而不同类型国债之间的多标度特征则具有相当大的差异性。基于此．本文从6类国债中选取了6个国债进行多标度特征分析，发现国债010103、009908和010107标度函数的波动幅度很大．表现出明显的非线性特征，而其他3个国债标度函数的波动幅度则相对较小．但也表现出一定程度上的非线性特征。

The paper first makes clustering analysis for 17 treasury bonds＇ daily return time series by using a suitable metrical approach based on the correlation among the bonds. Using the Matlab software when the critical threshold value is appropriate 0.059 5, the paper classifies the bonds into six groups. Then,it also studies the price fluctuation structure of each bond＇s daily return series well, and the results show that there is to some extent multi-scallng characters for China treasury bond market. In the meanwhile, the paper finds that there is strong self-similarity among bonds belonging to the same group,while the multi-sealing characters among different bonds are different. On the basis of that,the paper chooses six bonds from the six groups,analyzes their multi-scaling characters and finds that the scaling functions＇ range of volatility of treasury bonds 010103, 009908 and 010107 is large and shows the evident non-linear character, while the scaling functions＇range of volatility of other three bonds is comparatively small, representing non-linear character to some extent.