Publication Type
Working Paper
Version
publishedVersion
Publication Date
7-2023
Abstract
Modeling multivariate stochastic volatility (MSV) can be challenging, particularly when both variances and covariances are time-varying. In this paper, we address these challenges by introducing a new MSV model based on the generalized Fisher transformation of Archakov and Hansen (2021). Our model is highly exible and ensures that the variance-covariance matrix is always positive-definite. Moreover, our approach separates the driving factors of volatilities and correlations. To conduct Bayesian analysis of the model, we use a Particle Gibbs Ancestor Sampling (PGAS) method, which facilitates Bayesian model comparison. We also extend our MSV model to cover the leverage effect in volatilities and the threshold effect in correlations. Our simulation studies demonstrate that the proposed method performs well for the MSV model. Furthermore, empirical studies based on exchange-rate returns and equity returns show that our MSV model outperforms alternative specifications in both in-sample and out-of-sample performances. Overall, our novel MSV model and inferential method over a more reliable and exible framework for modeling time-varying variances and covariances of financial assets.
Keywords
Multivariate stochastic volatility, Dynamic correlation, Leverage effect, particle filter, Markov chain Monte Carlo
Discipline
Econometrics
Research Areas
Econometrics
First Page
1
Last Page
49
Publisher
Paper No. 09-2023
Citation
CHEN, Han; FEI, Yijie; and Jun YU.
Multivariate stochastic volatility models based on generalized fisher transformation. (2023). 1-49.
Available at: https://ink.library.smu.edu.sg/soe_research/2683
Copyright Owner and License
Authors
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.