金融数学学术报告

题 目: GARCH Modelling of Term Structures
报告人: Dr. Shuling Chen （陈曙玲）
University of New South Wales, Sydney
and
Senior researcher
Bureau of Crime Statistics and Research, Attorney General Department
Australia
时 间：2004年8月6日（星期五），上午10点
地 点：理科大楼A座次16楼统计系会议室

摘要：
The development of economic and tatistical models describing the term structure of interest rates has been onsidered important. Term structure of interest rates is dependent on two ariables, the time of evolution and he length of time to maturity (or enchmark, the level of maturity). nivariate GARCH models in each enchmark, with some exogenous nnovation variables and residuals in Student-t distribution, showed that he parameters of the individual GARCH models are functionally dependent on he length of time to maturity (benchmark). The function patterns are lausible with the financial economics. In this talk, a GARCH odel is presented, with the new parameters describing the benchmark dependence of he GARCH coefficients in both mean equations and conditional variance quations of the multivariate GARCH ignoring the correlation structure. This model not only reduces the number of parameters to estimate but also rovides a consistent interpolation and prediction of the term structure nterest rates in any possible benchmarks. We refer this model as a erm structures GARCH model (TS-ARCH).

Diagnoses of TS-GARCH model are onsidered, and comparing with the Pure Diagonal Correlation model, the onstant Conditional Correlation (CCC) odel and Dynamic Conditional orrelation (DCC) model in terms of the onditional variance structure.