摘要

In this paper, we define multivariate normal tempered stable (NTS) processes by evaluating multivariate Brownian motions at a random clock and construct multivariate general normal tempered stable (GNTS) processes by setting different parameters on the random clock for different Brownian motions. Then under the risk-neutral probability measure, for multiple assets we introduce multivariate NTS models (resp., multivariate GNTS models), along with their stochastic volatility cases of NTSSV (resp., GNTSSV) models by evaluating multivariate NTS (resp., multivariate GNTS) processes at a new random clock, a time integral of CIR process. In these financial models, assets' returns present non-Gaussian characters and jump components and the stochastic volatility is inherited by each asset and the correlation between any two different assets is also affected stochastically by the random clock instead of being constant or deterministic.