| Market manipulation is widely believed to affect asset pricing and expected stock returns as it could significantly hurt market efficiency and investor confidence.However,it is very difficult to identify market manipulation behavior in reality,and hence many cases still occurred in recent years given strict regulation policies.Based on the market manipulation punishment cases released by China’s Securities Regulatory Commission between 2013 and 2018,this paper construct an econometric model to quantitatively measure the risk of market manipulation using the information from both high frequency transaction data and low frequency fundamental indicators.Particularly,establish a mixed-frequency Logistic regression to estimate the probability whether an individual stock is manipulated or not,define this probability as MM-score and find that it has predictive power on short-term stock returns.This paper also design a trading strategy based on the MM-score which generates significant abnormal return.The study shows that market manipulation could be effectively captured by the MM-score,and it could be useful for designing investment strategies and establishing an early warning system of market manipulation.Firstly,by using the real cases in the punishment decision for market manipulation published by CSRC,the characteristics of the stock market are divided into five aspects:yield rate,liquidity,trading activity,degree of investors’ trading strategy activism and buying enthusiasm.On the basis of previous literature,manipulation is further divided into warehouse building period,pull-up period and selling period,in order to more effectively analyze the changes of stock characteristics in different stages of manipulation.At the same time,the recovery of various features of stock after manipulation is analyzed.In particular,the decline in yields after manipulation is the direct cause of losses to momentum investors who trade with the trend.Secondly,based on the characteristics of stocks with significant changes,this paper establishes a market manipulation identification model,monitors the market manipulation probability in real time,and achieves the purpose of early warning of manipulation behavior by assigning a certain threshold to the manipulation probability.This paper describes the manipulation behavior by using the characteristic variables of the stock disk,and adds the fundamental static information of the company into the model.The empirical results also confirm that the explanatory ability and recognition ability of the model are greatly improved after the fundamental information of the company is added.In addition,an adjustable threshold setting mechanism of manipulation probability is established to increase the portability and flexibility of model application.Thirdly,this paper introduces liquidity premium theory to explain the relationship between manipulation and expected rate of return.Liquidity in order to solve the multiple dimensions of difficulties,this paper USES principal component analysis（pca）to a variety of liquidity index analysis,based on comprehensive liquidity index,test in the stock market of our country,the result found significant liquidity premium effect in Chinese stock market,liquidity premium after traditional control risk premium factor effect is significant.Fourth,this paper examines the role of manipulation probability in predicting stock returns.After control size and other factors found to manipulate probability still has significant influence on stock expected return,this shows that the manipulation of the probability of expected return beyond the traditional theory explanation ability,and this mechanism is determined jointly by the operator and investor behavior,holding low manipulation of the probability of portfolio and selling high manipulation of the probability of portfolio will produce excess returns.Finally,this paper explores the manipulation of probabilistic risk premium and its source.In this paper,the manipulation factors were established,and then the risk premium ability of the manipulation factors on the basis of Fama and French five-factor model was tested.On this basis,this paper analyzes the attributes of the manipulation factors based on Daniel and Titman（1997）.The results show that the manipulation factors really reflect the characteristics of a class of companies more than the systemic risk,which is closer to the characteristic attributes. |
PDF Full Text Request