BAYESIAN STATE-SPACE ESTIMATION OF STOCHASTIC VOLATILITY FOR STORABLE COMMODITIES

Improving volatility modeling has important implications for option pricing, risk management, and forecasting, among other uses. Bayesian state-space framework is used because classical likelihood-based estimation of SV models using the Kalman filter tends to be unreliable due to the nonlinearity of the model. An increasingly used, Bayesian analog to the Kalman filter is Bayesian Markov chain Monte Carlo (MCMC). This is a simulation based approach for dependent data that is particularly well suited to the estimation of general, possibly nonlinear, state-space models. When the level of stocks is high relative to a given level of demand, supply and demand shocks could be absorbed by stocks, implying smaller price movements caused by the shock. Inversely, low levels of stocks relative to a given demand will put upward pressure on prices when a shock to supply or demand occurs, since borrowing from the future is not possible.

BAYESIAN STATE-SPACE ESTIMATION OF STOCHASTIC VOLATILITY FOR STORABLE COMMODITIES Academic Article