Econometric analysis of high-frequency market microstructure

This thesis introduces new econometric tools to analyse high-frequency financial data emerged from high-frequency trading. The analysis is based on the consensus that asset prices at high-frequencies have a permanent component that reflects the fundamental value, and a transitory microstructure noise induced by market imperfection. While the classic economic theory predicts that the fundamental value follows a semimartingale, the microstructure noise, however, exhibits rich dynamics.
Chapter 2 develops econometric tools to analyse the integrated volatility of the fundamental value and the dynamic properties of the microstructure noise in high-frequency data under dependent noise. Specifically, a finite sample analysis reveals the essential roles played by the finite sample bias in applications. A two-step approach is proposed accordingly to refine the finite sample performance. Chapter 3 introduces a simple and intuitive measure of the microstructure noise under a general nonparametric setting. The new econometric techniques provide two liquidity measures that gauge the instantaneous and average bid-ask spread with potentially autocorrelated order flows. While being flexible with respect to the autocorrelation structures, the new estimators only employ the transaction prices, thus do not require any knowledge of the order flows. Chapter 4 further extends the method introduced in Chapter 3 to the joint estimation of arbitrary finite moments of microstructure noise using high-frequency data, under a general setting that allows for irregular observation schemes and nonstationary, serially dependent noise.