Forecasting volatility using realized stochastic volatility model with time-varying leverage effect

Abstract This paper proposes a realized stochastic volatility model with time-varying leverage effect (hereafter the RSV-TVL model), in which the time-varying leverage effect is modelled based on a linear spline. The model parameters are estimated by using the maximum likelihood method based on a continuous particle filter. Simulation results show that the proposed estimation method works well. An empirical application to S&P 500 index highlights the value of incorporating the realized volatility measure and the time-varying leverage effect into volatility forecasting, and shows that the RSV-TVL model produces more accurate out-of-sample forecasts of volatility than the alternatives.