基于公司间的经济关联而衍生出的动量溢出效应是实证资产定价领域关注的话题之一。本文利用A股市场特有的概念板块数据，构建了基于概念板块的关联网络，并定义了核心变量——利用共同覆盖数对收益率加权求和得到的概念关联动量CWMom。以往的文献大多数聚焦于行业、分析师、供应链等常见的经济关联，而本文创新性地使用概念板块的数据来构造公司间关联，丰富了动量溢出效应领域的研究。从数据特点来看，概念板块数据具备覆盖面广、针对性强、自由度高等特点，且近年来A股投资者概念投资的热度不减、概念板块的轮动速度加快，因此本研究具备较强的现实意义与实证价值。本文通过Fama-Macbeth回归与单变量排序构建投资组合的方法开展实证研究，结果表明，A股市场存在显著的概念关联动量溢出效应，即通过概念相关联公司的过去一段时间的收益能够正向预测本公司的未来收益，最佳预测周期为周度，基于概念关联动量单变量排序的多空组合具有显著的收益（费前年化收益15.184%），且无法被现有因子模型解释。将现有文献提到的行业、分析师、新闻等溢出效应考虑在内仍然具备显著的超额收益。基于A股市场存在短期反转的现象，本文还提出了一个改进的残差关联动量因子ResCWMom，相较于关联动量因子，投资组合的收益得到了提升且最大回撤减小。在机制检验部分，本文先通过构造基本面关联指标ConceptSUE，证实了关联企业之间存在基本面的溢出。为了进一步探索因子的收益来源，本文计算了因子溢价与因子承载并进行了回归分析，结果表明，概念关联动量的溢出效应来源于错误定价而不是风险补偿，随着预测周度的延长正向预测力发生反转，说明是由于投资者过度反应导致的错误定价。此外，双变量排序检验表明，公司市值越小、分析师覆盖数目越低的情况下，概念关联动量的预测力越强，这可以被有限注意力理论以及散户的占比来解释。在稳健性检验部分，当改变核心变量的计算方式、改变计算窗口的长度、改变样本池范围后仍然保持显著，证明了结论的稳健性。
Momentum spillover effects based on economic links among firms are one of the topics of interest in the field of empirical asset pricing. In this paper, we utilize the conceptual sector data specific to the A-share market to construct networks, and define the core variable, the concept weighted momentum CWMom, obtained by weighting and summing the returns using the common coverage number.While most of the previous literatures focus on the economic links such as industries, analysts, supply chains, etc., this paper innovatively uses data from conceptual sectors to construct inter-firm links, enriching the research in the area of momentum spillovers. The conceptual sector data has a wide coverage, strong relevance, high degree of freedom, and in recent years, the conceptual investment of A-share investors has been unabated, and the rotation speed of conceptual sectors has been accelerated, so this study has a strong practical significance and empirical value.This paper carries out empirical research through Fama-Macbeth regression and univariate sorted portfolios, and the results show that there is a significant concept-associated momentum spillover effect in the A-share market, i.e., the concept associated company‘s returns can positively predict the company‘s future returns, and the optimal prediction period is weekly, and the concept weighted momentum based on the univariate sorting of the long/short portfolio has significant returns (15.184% annualized returns before fees) and cannot be explained by the existing factor models. Taking into account the spillover effects of industry, analysts, news, etc. mentioned in the existing literature, there is still a significant excess return. Based on the phenomenon of short-term reversal in the A-share market, this paper also proposes an improved residual weighted momentum factor, ResCWMom, which improves the portfolio return and reduces the maximum drawdown.To test the mechanism, this paper first confirms the existence of fundamental spillovers among affiliated firms by constructing the fundamental indicator ConceptSUE. In order to further explore the source of returns, this paper calculates the factor premium and factor carrying and conducts a regression analysis. The results show that the spillover effect of the concept weighted momentum originates from mispricing rather than risk compensation, and the positive predictive power reverses as the forecasting period extends, suggesting that it is due to mispricing caused by investor overreaction. In addition, bivariate sorting tests show that the predictive power of concept weighted momentum is stronger the smaller the firm‘s market capitalization and the lower the number of analysts‘ coverage, which is consistent with the mechanism of limited attention, where information is transmitted slower and lagged price movements are more likely to occur when the firm receives less attention.In the robustness test section, the robustness of the findings is demonstrated by the fact that the core variables remain significant when the calculation of the core variables is changed, the length of the lookback window is changed, and the range of the sample pool is changed.
