基于谱聚类的SHIBOR非对称波动研究

Research of SHIBOR's asymmetric volatility based on spectral clustering

以8种上海银行间同业拆放利率(SHIBOR)的日结算数据为对象,采用多路归一化谱聚类算法对建模对象的代表性进行分析,同时界定长短期产品,然后针对流动性最高且独立性较好的隔夜利率,运用EGARCH模型对其进行建模,考察利率波动的非对称效应.结果表明:非预期正的利差抖动引起的波动上升大于同幅度非预期负的利差抖动引起的波动上升,SHIBOR依据产品时限的不同,呈现出时限越长平稳性越好的特点.

With settlement data of eight kinds of Shanghai interbank offered rate（ SHIBOR） as object,using multiple normalized spectral clustering analysis of modeling the representation of the object was analyzed,defining the short-term and long-term products at the same time,and then with overnight interest rates the highest liquidity and better independence were selected as the research object,using EGARCH sequence of data modeling,the asymmetric effects of interest rate fluctuations were explered. The study found that： spreads jitter caused by the fluctuation in the unexpected was rising more than an unexpected negative interest rate differentials with amplitude jitter caused by higher volatility,SHIBOR time limit according to the product,represents a feature of the longer the time limit the better.