On measuring volatility of diffusion processes with high frequency data

We analyze a recently proposed method to estimate the volatility of a diffusion process with high frequency data. The method is based on Fourier analysis, all observations are included in the computation without any data manipulation. By Monte Carlo experiments, we evaluate its performance in measuring volatility under the assumption that the asset price evolves according to models belonging to the SR-SARV(1) class, which includes GARCH(1, 1) as a particular case. We compare the performance of the method to that associated with the cumulative squared intraday returns. The forecasting capability of the models is also evaluated