Ball合作论文被JFE接受发表，探讨会计信息与资产定价，还是50年前那个Ball...

原创： Analytical 分析式会计研究 今天

本期向大家推荐一篇国际顶级期刊 JFE 最新接受发表的一篇论文，作者之一为大名鼎鼎的Ball（球神/球爷爷），就是 Ball and Brown (1968) 一文（实证会计研究开山之作）的作者，50多年后继续发表论文，让人由衷敬佩！

此文基于会计信息的视角，探讨了资产定价因子。经典的Fama-French因子模型其中一个因子为权益帐面-市值比（book-to-market ratio），前段时间的推送“JFE最新论文: Fama认可的中国三因子模型来！”认为盈余-价格比（earnings-price ratio）因子在中国更有效。与此不同，Ball et al. (2019) 这篇论文认为，所有者权益由 retained earnings 和 contributed capital 两个部分构成，前者才是影响定价因素的核心，它是企业累积与平均后的盈余，可以避免单期盈余的计量误差或实体经营冲击。Retained earnings-to-market 能够预测股票回报，因为其反应了企业的基础收益率，而不是公司内在价值。

所以，会计信息对资产定价还是非常重要的？earnings 与 book value of equity 到底谁对资产定价更重要？

论文具体如下：

Journal of Financial Economics

Available online 18 May 2019

In Press, Corrected Proof

Earnings, retained earnings, and book-to-market in the cross section of expected returns

Ray Ball

University of Chicago

Joseph Gerakos 

Dartmouth College

Juhani T. Linnainmaa

University of Southern California & NBER

Valeri Nikolaev

University of Chicago

Abstract

Book value of equity consists of two economically different components: retained earnings and contributed capital. We predict that book-to-market strategies work because the retained earnings component of the book value of equity includes the accumulation and, hence, the averaging of past earnings. Retained earnings-to-market predicts the cross section of average returns in U.S. and international data and subsumes book-to-market. Contributed capital-to-market has no predictive power. We show that retained earnings-to-market, and, by extension, book-to-market, predicts returns because it is a good proxy for underlying earnings yield (Ball, 1978; Berk, 1995) and not because book value represents intrinsic value.

Keywords

Book-to-market

Contributed capital

Earnings yield

Mispricing

Retained earnings

Value premium

https://doi.org/10.1016/j.jfineco.2019.05.013

附文中几张图，点击左下角“阅读原文”，可下载品读全文。

Fig. 1. Mean reversion in earnings-to-price. We assign firms into quintiles each year by earnings-to-price and compute the median earnings-to-price for each quintile-year pair from ten years before the portfolio sort to ten years after. This figure reports the time series averages of these median earnings-to-price ratios. Earnings-to-price in year t is defined as income before extraordinary items in the fiscal year that ends in calendar year t deflated by the market value of equity in December of year t. The thickest line represents firms with the highest earnings-to-price at the time of portfolio formation; the thinnest line, those with the lowest earnings-to-price. The sample, at the time of portfolio formation at time 0, consists of all but microcap firms with non-missing income before extraordinary items and non-missing market value of equity. All but microcap firms are stocks with market values of equity at or above the 20th percentile of the NYSE market capitalization distribution.

Fig. 2. Comparison of the predictive power of lagged BE/ME and lagged RE/ME. This figure plots t-values associated with the Fama and MacBeth (1973) regression slopes for log (BE/ME) and log (RE/ME) from cross-sectional regressions that predict monthly returns. The regressions are estimated using data from July 1969 through December 2017 for stocks with a market value of equity above the 20th percentile of the NYSE market capitalization distribution (all but microcaps), non-missing and positive book value of equity, and non-missing market value of equity. The regressions are estimated separately using book-to-market or retained earnings-to-market as the main regressor. The other regressors are prior one-month return, prior one-year return skipping a month, and log-size. The regressions with retained earnings-to-market also include an indicator variable that identifies observations with RE ≤ 0. log (RE/ME) is set to zero for these observations. The control variables are updated by month, and the book-to-market and retained earnings-to-market variables are lagged by the value indicated on the x-axis. The estimates at x=2, for example, explain cross-sectional variation in returns using the values of log (BE/ME) and log (RE/ME) recorded two years earlier. The dashed line indicates the threshold for statistical significance at the 5% level. ME = market value of equity; BE = book value of equity; RE = retained earnings.

x=2,

Fig. 3. Annual cross-sectional means and correlations of book-to-market and its components. Panel A plots the cross-sectional means of book-to-market, retained earnings-to-market, and contributed capital-to-market from 1964 through 2017. Panel B plots the cross-sectional means of book-to-market for firms classified by cohort. A firm’s cohort is determined by its first appearance on the CRSP database. Panel C plots the cross-sectional means of retained earnings-to-market by cohort. Panel D plots cross-sectional correlations between book-to-market and retained earnings-to-market and between book-to-market and contributed capital-to-market. ME = market value of equity; BE = book value of equity; RE = retained earnings; CC = contributed capital.

Fig. 4. Predictive power of current earnings with and without controlling for retained earnings-to-market. This figure plots t-values associated with the Fama and MacBeth (1973) regression slopes from cross-sectional regressions that predict monthly returns. The regressions are estimated using data from July 1964 through December 2017 for stocks with a market value of equity above the 20th percentile of the NYSE market capitalization distribution (all but microcaps), non-missing and positive book value of equity, non-missing market value of equity, and non-missing net income. The main regressor is E/BE, defined as net income-to-book value of equity. The dotted line presents t-values from regressions that do not include log(RE/ME) as an additional control; the thick solid line, regressions that include log(RE/ME). Both sets of regressions also control for prior one-month return, prior one-year return skipping a month, and log-size. The regressions with retained earnings-to-market also include an indicator variable that identifies observations with RE ≤ 0. log (RE/ME) is set to zero for these observations. We predict returns at the horizon indicated on the x-axis. The estimates at x=2, for example, are from regressions that predict the cross section of monthly returns two years after the regressors are recorded. The dashed line indicates the threshold for statistical significance at the 5% level. ME = market value of equity; BE = book value of equity; RE = retained earnings.

推送内容仅用于学术交流，版权归原作者和出版机构所有，转载请与我们联系！

分析式会计研究｜品读经典 聚焦前沿

微信ID：Analytical_AR

  长按二维码关注，欢迎评论分享！