High-Frequency Volatility Estimation and the Relative Importance of Market Microstructure Variables

In this paper we examine the relative importance of trading volume, bid-ask spread, order flow, order imbalance, total quote depth, quote depth difference and trading intensity for high-frequency volatility estimation. By using a best subset regression approach, we fi nd that contemporaneous trading intensity and order flow contains the most important information about volatility estimation in general, but the rankings of the importance of the market microstructure (MMS) variables vary between securities. Using a Lognormal Log-Autoregressive Conditional Duration (LL-ACD) model, we show that the inclusion of MMS covariates signi ffcantly improves the goodness-of- fit of the model. Furthermore, we show that the inclusion of MMS covariates in the LL-ACD model leads to substantial improvements in the quality of volatility estimates, both on a daily and an intraday level