学术报告十三:Yonggan Zhao—Option Pricing When the Underlying Asset Returns Follow a Hidden Markov Process

发布时间:2018-07-05

报告时间:2018年7月6日(星期五)9:00-9:30

报告地点:翡翠湖校区科教楼B座1710室

  :Yonggan Zhao教授

工作单位:Rowe School of Business,Dalhousie University

举办单位:数学学院

报告人简介

淦博士,加拿大风险管理研究教授,于加拿大大豪斯大学商学金融系。200012 毕业加拿大不列大学商学院之后,曾任于新加坡南洋理工大学, 美国普林斯大学的客座研究,西安交大的特聘教授。主要研究方向有投资组合,金融衍生品定价与设计,金融风险管理。在金融经济和管理科学等域里文三十余篇。曾多次主持加拿大国家自然科学基金和社会科学基金目。研究成果曾得加拿大金融年会最佳, 多个冲基金和投机构的投资顾问Journal of Economic and Administrative Sciences, Journal of Management Mathematics任副主

报告简介

We develop a discrete-time derivatives valuation model that incorporates economic strength characterized as a first-order hidden Markov chain. Unlike the standard regime switching models, where one-period logarithmic returns of the underlying asset follow a conditional normal distribution given the current economic strength, we assume these quantities obey a mixture normal distribution to capture the dependency of the underlying asset returns on the expected economic strength.  The optimal martingale measure is obtained by minimizing the mean square error between the model and actual market prices of the derivatives. In contrast with some of the alternative models, such as the Black-Scholes-Merton model, the Hardy model, and the Heston-Nandi model, option prices explicitly depend on the conditional expected return parameters of the underlying asset and the posterior probabilities of the economic strength.  Monthly S\&P 500 index options for the period of January 2012 to September 2017 are used for model estimation and validation. It is found that risk/return profiles under the risk neutral probability measure are drastically different across economic regimes to capture the regime property of the financial market. To compare the model performance with those of the alternative models, we also discuss the model implied volatilities and prediction accuracy for given statistical significance, while we examine the pricing errors to see both the under- and out-performances by various models.   We find that our regime switching model has a superior performance over the alternative models for both the in-sample and out-of-sample data.