上海财经大学 > 科学研究 > 学术交流 > 学术报告
  • 统计与管理学院2017年学术报告第4

     

    【主  题】Extended Unified GARCH-Ito Model

    【报告人】 Xinyu Song, Ph.D.candidate

    University of Wisconsin-Madison

    【时  间】 2017年03月13日(星期一)09:30-10:30

    【地  点】 上海财经大学统计与管理学院大楼1208室

    【摘  要】Traditionally low frequency models such as GARCH and high frequency models such as Ito diffusion process are developed and applied independently to financial data, however, the inter-correlation between data at different time scales cannot be ignored, therefore it follows that a unified model that can accommodate both a continuous-time Ito process and a GARCH process by embedding a discrete-time GARCH volatility in its continuous-time instantaneous volatility will have advantageous in explaining the dynamic evolutions of financial data. We propose a unified model where the instantaneous volatility has both integrated volatilities and squared log returns as innovations so that the proposed model would be able to catch up rapid changes in volatility process. For the proposed unified model, integrated volatilities in the continuous process are constructed using realized volatility estimators based on high frequency data and are treated as proxy for low frequency volatilities in the discrete process. Quasi-maximum likelihood estimators are used to estimate parameters. Asymptotic behaviors of proposed estimators have been studied. Simulation study has been done to check finite sample performances for proposed estimators and has confirmed that proposed estimation method is capable of achieving higher efficiency compare to other traditional estimation method.

    【邀请人】 冯兴东