Estimation of the Risk Process Based on Moments of Integrated Volatility Using High-frequency Data
This paper uses high-frequency data to model the volatility of asset prices over the period 2007 to 2014 for 28 individual stocks and 19 exchange-traded funds. I use the Heston (1993) model to characterize the evolution of 100-second sampled quadratic variation over a trading day and apply generalized method of moments(GMM) to estimate three parameters of the model: the asymptotic mean, speed of mean reversion and the volatility of volatility. I discover that the Heston model performs well in most cases regardless of the trade volume. I also find common patterns in the estimates: The speed of mean reversion is nearly time invariant, the volatility of volatility is much larger than the mean volatility and the asymptotic mean moves slowly over time but can vary a lot in times of large market uncertainty.