ESSAYS ON FINANCIAL ECONOMETRICS: INTRADAY DYNAMICS IN FINANCIAL MARKETS

This dissertation consists of three essays on financial econometrics, all focusing on intraday dynamics in financial markets. The first essay investigates the intraday dynamics of equity market correlation and the impacts of some macro-announcements on the dynamics. The average correlation between assets in the stock market measures stock price comovement. Analyzing the intraday dynamics of correlations among high-frequency returns is challenging due to market microstructure noise (e.g. discretization, asynchronous trading). The task becomes particularly difficult for illiquid periods, such as pre-open and post-close trading sessions. This chapter proposes a novel quadrant-based correlation measure that is robust to market microstructure noise and more computationally efficient than the conventional Pearson correlation. I apply this measure to 1-second NASDAQ trade-price data for S&P 500 constituents to examine the intraday dynamics of market correlation, including the pre-open and post-close trading sessions. I also analyze how major macroeconomic announcements influence these intraday correlation patterns. The results show that correlation is lower in the pre-open and post-close sessions compared to active trading hours and correlation increases over the course of the trading day. I also find that correlation responds differently to different macroeconomic announcements. The second essay designs and implements score-driven models to study the intraday dynamics of equity market correlation. I find that intraday correlation estimates have a dominant factor and each time slot contributes to the factor approximately equally. The third essay studies the intraday dynamics of the price discovery process in Bitcoin markets. I analyze the intraday pattern of the Bitcoin price discovery process between Binance and Coinbase. I use two price discovery models, information share and component share, to analyze minute-level Bitcoin prices in the two exchanges for a 2-year period. I find that Binance dominates the price discovery process during nighttime in the U.S., while Coinbase slightly dominates during daytime in the U.S.