THESIS
2011
xi, 89 p. : ill. ; 30 cm
Abstract
It is a usual practice in the financial industry to estimate the volatility from the sum of frequently sampled squared returns. However, tremendous empirical evidences have been found showing that the price process can not be sampled too often, due to a broad collection of issues known as microstructure noises. In this thesis, we consider the price model where both additive and rounding errors are present. We establish asymptotic results for the realized volatility, and propose two intercept estimators based on regular and generalized least square regressions. We compare our proposed estimators with other vastly adopted volatility estimators. Simulation and empirical studies show that our estimators outperform the existing popular ones in terms of consistency, efficiency, robustness wit...[
Read more ]
It is a usual practice in the financial industry to estimate the volatility from the sum of frequently sampled squared returns. However, tremendous empirical evidences have been found showing that the price process can not be sampled too often, due to a broad collection of issues known as microstructure noises. In this thesis, we consider the price model where both additive and rounding errors are present. We establish asymptotic results for the realized volatility, and propose two intercept estimators based on regular and generalized least square regressions. We compare our proposed estimators with other vastly adopted volatility estimators. Simulation and empirical studies show that our estimators outperform the existing popular ones in terms of consistency, efficiency, robustness with respect to sample sizes and robustness with respect to different sources of market microstructure noises.
Post a Comment