Empirical Study On Three Anomalies: Turnover Anomaly,IVOL Anomaly And Time Series Momentum Anomaly

Nowadays,the capital market in China is developing in a very high speed,and the stock market’s contribution to the real economy is also growing accordingly.One of the most important tasks for financial development is to keep the stock market stable and healthy.This requires a good understanding of the market which can be earned by doing asset pricing research.This thesis focuses on three topics in asset pricing area,each of which has attracted a lot of attention in last ten years.These three topics cover three anomalies which are the trading volume anomaly,the idiosyncratic volatility puzzle and the time series momentum anomaly.This thesis is aimed to reconcile disputed conclusions on the three topics and has found several striking findings,both of which will be introduced later.After reviewing the literature on the three anomalies,this thesis puts the three anomalies into a unified behavioral finance framework and proposes underreaction to explain them.Then the research on the three anomalies is introduced in detail.Each sector of the three anomalies follows a sequence of explaining research motivation,introducing the research questions,reporting empirical results and making research conclusions.In the sector of trading volume anomaly,I first explain the research motivation by introducing the conflicting empirical results from Hou et al.(2018)and Datar et al.(1998).Then I study the relation between trading volume and future stock returns.The relation between raw trading volume and future stock returns is proved to be insignificant.This finding supports the research conclusion of Hou et al.(2018).But A significant negative relation between average trading volume and future stock returns and a significant positive relation between detrended trading volume and future stock returns are also found.These two findings inspire me to explore the relations deeply.I find that the negative relation between average trading volume and future stock returns is determined by the relation between average trading volume during the period from the last sixth month to the last twenty-fourth month.And the positive relation between detrended trading volume and future stock returns is determined by the short term trading volume shocks.I also find that the predictive power of trading volume comes from the factors affect both trading volume and stock returns rather than from the systematic risks carried by trading volume.This finding leads me to study the relation between trading volume and stock characteristics and to build multiple factor models on trading volume series.The trading volume five-factor model I built has a much greater power in explaining trading volume comparing to the one-factor model which was used in previous research.And the trading volume five-factor not only has a good performance in the US stock market but also in Chinese stock market.Besides the findings above,this sector has two minor findings.One is on the relation between length of reference periods and the performance of detrended trading volume strategies.Six months is found to be the best choice.The other one is on the comparison among three methods of identifying abnormal trading volume.I find the moving average method has the supreme power of identifying abnormal trading volume.In the sector of idiosyncratic volatility(IVOL)puzzle,I first introduce the theoretical foundation connecting IVOL puzzle to trading behavior and explain the relation between overreaction and IVOL puzzle and the relation between underreaction and IVOL puzzle.By analyzing the flaws in Huang et al.(2010),I state the research question of this sector.This sector is to find out which leads to the IVOL puzzle,overreaction or underreaction.With portfolio level analysis and firm level analysis,the IVOL puzzle is found to be a result of underreaction,not a result of overreaction.This finding contradicts the research conclusion in Huang et al.(2010).And I reconcile my finding to theirs by analyzing the empirical results.After then,several robustness tests are introduced.All the results support the conclusion.At the end of this sector,relying on the conclusion of this sector,I attempt to improve the IVOL trading strategy performance through enhancing the ability of market timing and stock picking.In the sector of time series momentum,I first introduce the research motivation by analyzing the flaws in Goyal and Jegadeesh(2018).This sector is aims to take part into the disputation of the existence of time series momentum between Goyal and Jegadeesh(2018)and Moskowitz et al.(2012).In the comparison between time series momentum and cross section momentum,I propose to build zero-cost time series momentum trading strategy to overcome the flaws in Goyal and Jegadeesh(2018).The elements in the trading portfolios,the trading strategy performance and the risk behind time series momentum and cross section momentum are examined.And the findings support the conclusion in Moskowitz et al.(2012)which is that the time series momentum is distinct to cross section momentum.After this,I propose a new method to identify the time series momentum effect in stock markets.Comparing to the method in Moskowitz et al.(2012),the new method has a greater ability in capturing the time series momentum effect.And this sector has two additional minor contributions.One is that a new time series trading strategy is formed and the strategy performance is fairly well.The other one is that I proved that D’Souza et al.(2018)should not use double sort to construct double momentum strategy.