Bayesian learning, information diffusion, and the trading behavior in stock markets

I first develop a new approach based on Bayesian learning frame to estimate traders belief parameters. I assume that informed traders place market orders according to their beliefs. Informed traders and market makers make Bayesian inferences when they observe order flows, and they update beliefs to incorporate these inferences. Informed traders new beliefs prompt them to place new orders. This trading and belief-updating process can be summarized in a reduced form equation on order flows and price series, which enables me to estimate belief parameters that determine the order-price relations. I empirically test these parameters on S&P 500 stocks, and show that traders belief uncertainty and informational equality can explain daily bid ask spreads and stock returns. Secondly, I use the estimated belief parameters to investigate how price incorporates information following earnings announcements in the U.S. market. The Bayesian learning framework implies that informed trading drives market makers to gradually adjust the price, until price fully incorporates all the private signals of informed traders and informed trading complete disappears. I analogously call this process an information shockwave. The information shockwave story explains the price dynamics following earnings announcements. I further show that information asymmetry is more severe following earnings announcements because market makers belief uncertainty increases more than that of informed traders. Lastly, I investigate how traders respond to information asymmetry. Prior literature has proposed two views on traders response: (i) rational expectation view, and (ii) agree to disagree view. These two views can be theoretically reconciled under Bayesian learning framework. The empirical question I answer is: which view better captures traders overall behavior? I investigate the order flow relations in Chinese market, whose data enable me to analyze both market and limit orders. The findings show that liquidity traders in one side of the market are more willing to provide liquidity if the other side of the market demands liquidity. It indicates that agree to disagree view provides a better picture of traders behavior