Estimating a semiparametric asymmetric stochastic volatility model with a Dirichlet process mixture

We extend the asymmetric, stochastic, volatility model by modeling the return-volatility distribution nonparametrically. The novelty is modeling this distribution with an infinite mixture of Normals, where the mixture unknowns have a Dirichlet process prior. Cumulative Bayes factors show our semiparametric model accurately forecasting market returns. During tranquil markets, expected volatility rises (declines, then rises as the shock increases) when the market shock is negative (positive). This asymmetry is muted when the market is volatile. In other words, when times are good, no news is good news, but during bad times, neither good nor bad news matters with regards to volatility.

Estimating a semiparametric asymmetric stochastic volatility model with a Dirichlet process mixture Journal Articles