Volatility Estimates, Based On The Logarithmic Rate Of Return On The Assets Of The Intraday Data

The fast development in finance industry coupled with various mushrooming financial derivatives are making our financial markets more and more risky. The Asian financial crisis in 1998 and the recent subprime mortgage default have beaten the financal world hard. Consequently, many institutes without enough management on risks get bankrupted one after another. Risk management plays a more important role than ever before. One of the key risk managment jobs is to measure and monitor on the market risks. For many risk managemnet methods—VaR, Stress Test, RiskMetrics, the estimation of asset price daily volatility, be it variance or standard error, is indispensable. The popular models for daily asset price volatility estimation can be categorized into two groups: The first group of models mainly for forecasting volatility are generally based on interday data, such as daily log asset returns. Among these models GARCH model is good representative. While, the second group of models are built on intraday data, for example the highest price and lowest price of ever trading day. Obviously, intraday data models using more information can catch more risk details of ever trading day than those on interday data. Therefore, they are mostly used to monitor daily asset price volatility. In this paper, we first introduce several representative intraday data models—Simple Volatility Estimator, Highest/Lowest Estimator, Best Analytic Scale-invariant Estimator, Realized Bipower Estimator and VARHAC Estimator. Sencond, we make estimation with these models on simulated data. Third , we will use these models to carry a empirical study in Chinese stock market. At last, we analyze both simulation and empirical results to compare the pros and cons of each estimators.