Estimation of SVJD Models with Bayesian Methods and Power-Variation Estimators
35 Pages Posted: 1 Aug 2018
Date Written: March 30, 2017
Methodology is proposed of how to utilize high-frequency power-variation estimators in the Bayesian estimation of Stochastic-Volatility Jump-Diffusion (SVJD) models. Realized variance is used as an additional source of information for the estimation of stochastic variances, while the Z-Estimator is used as an additional source of information for the estimation of asset price jumps. The models are estimated by a combination of a MCMC algorithm and a SIR Particle Filter. The performance of the models is evaluated on simulated times series as well as real world financial time series of the 4 major foreign exchange rates.
Keywords: Stochastic Volatility, Bayesian Inference, MCMC, Particle Filters, Realized Variance, Bipower Variation, Z-Estimator, Jump Clustering, Self-Exciting Jumps, Hawkes Process
JEL Classification: C11, C14, C15, C22, G1
Suggested Citation: Suggested Citation