西南财经大学
成本增长与股票收益关系研究
Cost Growth and Stock Return
学位申请人：单学
学号：2161202Z6019
学科专业：财务管理
研究方向：公司财务理论与实务
指导教师：王新路
定稿时间：2019年3月
摘要传统资本资产定价理论认为股票收益率只取决于系统性风险，但由于市
场上存在信息不对称、市场摩擦、资金成本等限制，越来越多的研究结果拒绝
了资本资产定价模型（Black 和Scholes（1973），Fama 和Macbeth（1973）），
学术界转向对股票定价的多因子的探讨，规模、账面市值比、盈利价格比、资
产回报率等特征变量逐步被发现可以解释股票收益。在诸多特征变量中，盈
余及其组成部分与股票收益间的关系被学术界广泛地讨论，2015年Fama 和
French 将盈利因子纳入五因子模型中，正式确认了会计盈余对股票收益的预
测能力。Ertimur，Livnat 和Martikainen（2003），Jegadeesh 和Livnat（2006），
Huang et al.（2017）等学者还发现作为盈余的组成部分的收入和成本对股票收
益也有显著的预测能力。但在国内很少有人研究盈余组成部分，特别是成本
信息对股票收益的解释能力。
本文选取成本增长率作为成本信息的度量指标，探究在中国股票市场上
成本增长率对股票收益是否有预测能力。本文采用首先使用排序组合法、
Fama-French 五因子模型在投资组合层面验证了成本增长率低（高）的股票组
合有显著更高（低）的股票收益。其次，采用Fama-Macbeth 截面回归法在个
股层面上验证了成本增长率与预期股票收益存在显著的负向关系，且这种负
向关系在控制了规模、账面市值比、动量、资产收益率、净资产回报率、应计
盈余、盈利价格比等14个因素后依然显著，说明了成本增长中含有区别于14
个因素的超额信息。最后，本文尝试对结果进行解释，分别从套利成本角度和
投资者反应角度对研究结果进行解释，研究发现，套利成本导致的错误定价
是该现象可能的解释。在稳健性测试中，本文针对中国股票市场不存在退出
机制等的特殊情况，剔除了市值最小的30%股票样本进行稳健性检验，并采
用三个成本增长替代指标证明了我们结果是稳健的。
成本增长与股票收益关系研究
本文的研究结果具有理论和实践贡献。在理论上，本文的研究弥补了国
内相关研究空白的现状，研究成果发现成本增长率中含有与盈利等14个指标
不同解释信息，为成本增长率作为定价因子的进一步研究奠定了基础；同时
本文使用了Fama 和French 五因子模型进行实证，为该模型在中国股票市场
的适用性提供了新证据。在实践上，本文研究成果可以指导投资者进行投资
决策，成本增长率与股票收益间负向关系为投资者提供了构建投资策略的新
视角、新方法。同时，本文发现套利成本导致的错误定价该现象可能的原因，
这为进一步完善资本市场和套利机制，提高市场有效性提供了依据。
关键词：成本增长率，排序组合法，Fama-French 五因子模型，Fama-
Macbeth 回归
2
Abstract
Abstract
The traditional capital asset pricing theory believes that the stock return rate
depends only on systemic risk.However,due to information asymmetry,market
friction,capital cost and other restrictions in the market,more and more research
results reject the capital asset pricing model (Black and Scholes).(1973),Fama and
Macbeth (1973)),the academic community turned to the discussion of multi-factor
model,and the variables such as size,book-to-market ratio,profit-price ratio,and
return on assets were gradually found to explain stock returns.Among the many
characteristic variables,the relationship between earnings and its components and
stock returns has been extensively discussed.In 2015,Fama and French
incorporated profit into the five-factor model,which officially confirmed the ability
of accounting earnings to predict stock returns.Ertimur,Livnat and Martikainen
(2003),Jegadeesh and Livnat (2006),Huang et al.(2017)and other scholars also
found that income and costs as part of the surplus also have significant predictive
power on stock returns.However,few people in China study the earnings
component,especially the ability of cost information to explain stock returns.
This paper selects the cost growth rate as a measure of cost information and
explores whether the cost information has predictive ability on stock returns in the
Chinese stock market.In this paper,we use the portfolio method and the Fama-
French five-factor model to verify that the stock portfolio with low (high)cost
growth rate has significantly higher (lower)stocks return at the portfolio level.
Secondly,the Fama-Macbeth cross-section regression method is used to verify that
the cost growth rate has a significant negative relationship with the expected stock
returns at the individual stock level,and this negative relationship controls the scale,
book-to-market ratio,momentum,return on assets,After 14factors such as return
on net assets,accrued surplus,and profit-to-price ratio,it is still significant,
indicating that the cost increase contains excess information with 14factors.Thirdly,
in view of the special situation that there is no exit mechanism in China's stock