Multivariate Stochastic Volatility Model with Flexible Dynamic Correlations and Realized Measures

15:00-16:30, Friday, September 25, 2020

Tencent Meeting (Meeting ID: 517 167 322)

Dr. Yijie FEI received his Ph.D. in Economics from School of Economics, Singapore Management University in 2020. His primary fields are Financial Econometrics, Time Series Analysis. His work has appeared in Economics Letters.

Extending stochastic volatility models to multivariate case is not straightforward, especially when correlation structure is allowed to be dynamic. The challenges come from both model setup and parameter inference. In this paper, we make three contributions to this literature. First, we consider a new multivariate stochastic volatility (MSV) model, applying a recently proposed novel parameterization of correlation matrix. This modeling design is a generalization of Fisher's z-transformation to high-dimensional cases and it is fully flexible as the validity of resulting correlation matrix is guaranteed automatically. This allows us to separate the driving factors of volatilies and correaltions. Second, we propose to use a different estimation tool. Like most existing literature on MSV, we work within a Bayesian framework and hence rely on Markov Chain Monte Carlo (MCMC) sampler. However, when dealing with latent variables, traditional single-move or multi-move sampler is replaced by a novel technique called Particle Gibbs Ancestor Sampling (PGAS), which is built upon Sequential Monte Carlo (SMC). Third, we incorporate the information contained in intra-daily realized measures and propose to use a two-stage approach to reduce the estimation bias of some weakly identified parameters. Extensive simulation studies are conducted to confirm the applicability of this method under the current setup and provide guidance on the tradeoff between estimation accuracy and computational cost. The new model is then implemented using two financial datasets and comparison with existing models is discussed.

For more information of the seminar, scan the following QR code(s) to join Tencent QQ group (904 544 292) or WeChat group named "IAER Seminar (2)", please.

QQ Group

WeChat Group (Valid until 9/28/2020)