Tinbergen Institute Discussion Paper 12-125/IV/DSF45
The paper proposes a model for the dynamics of stock prices that incorporates increased asset co-movements during extreme market downturns in a continuous-time setting. The model is based on the construction of a multivariate diffusion with a pre-specified stationary density with tail dependence. I estimate the model with Markov Chain Monte Carlo using a sequential inference procedure that proves to be well-suited for the problem. The model is able to reproduce stylized features of the dependence structure and the dynamic behaviour of asset returns.
tail dependence multivariate diffusion Markov Chain Monte Carlo