Bayesian analysis of random coefficient stochastic volatility models

We develop Bayesian estimation, model comparison and prediction for a Stochastic Volatility models, where some of the parameters in the conditional variance follow stochastic process. In particular, we consider the following Stochastic Volatility models: SV model, where log of the conditional variance follows random coefficient autoregressive process, SV model, where autoregressive parameter follows a random walk and SVX model, where influence of exogenous variable on the volatility is also random. The empirical analysis investigates whether there is evidence of some form of parameter instability in the conditional variance and it also helps us to verify whether random coefficient SV models perform better in predicting the volatility of exchange rates than standard SV models.