Speaker: 金鑫 (Xin Jin)

Title: The Autoregressive Inverse-Wishart Multivariate Stochastic Volatility Model and its Factor Extension

Schedule: June 19, Mon 1:30-3:30 PM

Location: 诚明楼 (Chengming Hall）RM 315

 

Introduction: 金鑫，上海财经大学经济学院副教授，加拿大多伦多大学经济学博士学位，研究方向为金融计量经济学、时间序列计量经济学等，研究成果发表于Journal of Econometrics, Journal of Applied Econometrics, Journal of Financial Econometrics 等国际权威期刊，并主持国家自然科学基金项目。

 

Abstract: The development of Wishart/Inverse Wishart based multivariate stochastic volatility models is hindered by inefficient sampling methods. In this paper, we introduce a new proposal for the latent covariance matrix in the AIW model which demonstrates far better scalability. We then introduce various extensions to the basic AIW model that possess long-memory and accommodate leverage effects as well as allow for Markov-switching parameters. We make a third contribution by fitting the AIW model into a factor structure and provide an a posteriori identification procedure to produce identified observationally equivalent factor-loading representation while avoiding order dependence. When evaluated using real datasets with dimensions n = 10,60,355,1000, the new AIW based models are shown to be particularly superior in covariance matrix forecasting. In the case of ultra high dimensions, such strength of AIW models can be better utilized through a divide and conquer strategy.