投资者异质信念、预期演化与股票市场流动性

投资者异质信念不但是股票市场的现实，也会引起股票市场流动性的变化。本文建立从两期到多期的交易模型，利用马尔科夫决策过程模拟投资者的预期演化，旨在分析由于信息不对称和认知偏差导致的多种异质信念共存的情况下，投资者异质信念与股票市场流动性的内在关系。研究发现，投资者异质信念的波动程度、投资者风险厌恶程度和信息不对称程度均与股票市场流动性呈现负相关关系；而投资者规模与股票市场流动性正相关。利用MATLAB软件数值模拟多期交易过程，验证了在投资者预期演化的情况下，上述规律仍然成立。研究结论在一定程度上拓展了行为金融和市场微观结构的研究范畴，也对投资者优化决策行为及监管部门加强流动性管理有一定的参考价值。

Investor Heterogeneous Beliefs,Expect Evolution and Stock Market Liquidity

The paper attempts to explore the internal relationship between heterogeneous beliefs and market liquidity, from the perspective of investor cognition and information asymmetry. Describing the investor's prior's beliefs with cognitive bias and making information asymmetry as the motivation for the renewal of posterior heterogeneous beliefs, investors adjust their beliefs according to the Bayesian learning process by using the information set they currently possess, and generate new posterior beliefs to trigger different decision-making behaviors. It has caused market shocks and changed the liquidity of stocks.The relationship between heterogeneous beliefs and liquidity in two-phase transactions is studied. There are n types of investors. Heterogeneous beliefs are defined as different expectations of the intrinsic value of stocks. The current information set owned by the investor can be divided into the real intrinsic value information V^* and the noise information ε_i,t of the stock. All investors will receive the same information about the true intrinsic value of the stock, but investors will judge it according to their own cognitive bias as V.So in fact each type of investor judges the information structure forI_i,t=V+ε_i,t.According to Bayes rule, the posterior belief of investors can be constructed. In addition using Samuelson's Over Lapping Generation Models (OLG) considers the relationship between investor heterogeneous beliefs and stock liquidity in the two-period trading process.It is found that the degree of volatility of investor heterogeneous beliefs, investor risk aversion and information asymmetry are negatively correlated with stock market liquidity; while investor size is positively correlated with stock market liquidity. Further, it is assumed that the process of investor expectation evolution is a Markov Decision Process (MDP). Using MATLAB to numerically simulate the multi-period transaction process is verified that the above rules are still valid in the case of investors' expected evolution. The conclusion of this paper extends research scope of behavioral finance and market microstructures, and has certain reference value for investors to optimize decision-making behavior.