The Estimation of Leverage Effect with High Frequency Data

Leverage effect has become an extensively studied phenomenon which describes the negative relation between the stock return and its volatility. Although this characteristic of stock returns is well acknowledged, most studies about it are based on cross-sectional calibration with parametric models. Other than that, most previous work are over daily or longer return horizons and usually do not specify the quantitative measure of it. This paper provides nonparametric estimation of a class of stochastic measures of leverage effect for both cases with and without microstructure noise, and studies the statistical properties of the estimators when the log price process is a quite general continuous semimartingale, in the stochastic volatility context and for high frequency data. The consistency and limit distribution of the estimators are derived, and simulation results present the properties accordingly. This estimator also provides the opportunity to study the empirical relation between skewness and leverage effect, which further leads to the prediction of skewness. Furthermore, adopting similar ideas to these in this paper, it is easy to extend the study to other important aspects of the stock returns, e.g. volatility of volatility