Journal of Corporate Finance: Hatem Rjiba, Samir Saadi, Sabri Boubaker, Xiaoya (Sara) Ding

Journal of Corporate Finance 67 (2021) 101902
Contents lists available at ScienceDirect
Journal of Corporate Finance

journal homepage: www.elsevier.com/locate/jcorpfin
Annual report readability and the cost of equity capital☆

Hatem Rjiba a, *, Samir Saadi b, e, Sabri Boubaker c, f, Xiaoya (Sara) Ding d
a
Paris School of Business, Paris, France
b
Telfer School of Management, University of Ottawa, Ottawa, Ontario, Canada
c
EM Normandie Business School, Métis Lab, France
d
School of Management, University of San Francisco, San Francisco, California, USA
e
IPAG Business School, Paris, France
f
International School, Vietnam National University, Hanoi, Vietnam
A R T I C L E I N F O A B S T R A C T
JEL classification: Using a large panel of U.S. public firms, we examine the relation between annual report read
G12 ability and cost of equity capital. We hypothesize that complex textual reporting deters investors’
G14 ability to process and interpret annual reports, leading to higher information risk, and thus higher
G32
cost of equity financing. Consistent with our prediction, we find that greater textual complexity is
M41
associated with higher cost of equity capital. Our results are robust to a battery of sensitivity
Keywords:
checks, including use of multiple estimation methods, alternative proxies of annual report
Annual report readability
readability and cost of equity capital measures, and potential endogeneity concerns. In addition,
Bog index
Disclosure tone we hypothesize and test whether the nature of the relation between readability and cost of capital
Information risk depends on the tone of 10-K filings. Our results show that the effect of annual report complexity
Cost of equity capital on cost of equity is greater when disclosure tone is more negative or more ambiguous. We also
find that the effect of annual report readability on cost of equity capital depends on the degree of
stock market competition, level of institutional investors’ ownership, and analyst coverage.
“In a financial world driven largely by mathematical formulas and computers trading thousands of times a second, a young investor is
searching for investments in the most old-fashioned way possible: by reading”1 – Jason Zweig. The Wall Street Journal (April 1st,
2016).
☆
We are very grateful to two anonymous reviewers and William Megginson (Editor in Chief) for helpful comments and suggestions. We also
benefited from comments offered by Ali Akyol, Imed Chkir, Lamia Chourou, Alfred Davis, Vijay Jog, Dev Gandhi, Kose John, Janelle Mann, Fabio
Moneta, Koray Sayili, Ligang Zhong, and seminar participants at the University of Ottawa, University of San Francisco, Carleton University, Ontario
Tech University, Paris School of Business, the 2014 International Governance Conference, and the 2016 Telfer Annual Conference in Accounting and
Finance on earlier versions of this paper. Samir Saadi gratefully acknowledges the financial support from the Social Sciences and Humanities
Research Council of Canada (grant number 430–2014-0045) and the University of Ottawa. We thank Rui Duan and Sahar Shabani for excellent
research assistance.
* Corresponding author.
E-mail addresses: [email protected] (H. Rjiba), [email protected] (S. Saadi), [email protected] (S. Boubaker),
[email protected] (X.(S. Ding).
1
Available at: http://blogs.wsj.com/moneybeat/2016/04/01/its-time-for-investors-to-re-learn-the-lost-art-of-reading/
https://doi.org/10.1016/j.jcorpfin.2021.101902
Received 26 January 2019; Received in revised form 16 December 2020; Accepted 21 January 2021
Available online 26 January 2021
0929-1199/© 2021 Elsevier B.V. All rights reserved.
H. Rjiba et al. Journal of Corporate Finance 67 (2021) 101902
1. Introduction
Is information risk priced? This deceptively simple question has become the subject of intense debate at both theoretical and
empirical levels.2 In theory, the traditional asset-pricing models show that investors are compensated only for holding market risks.
Thus, information risk should not be priced as all relevant information is assumed to be reflected in the stock price at all times.3
However, this static view of market efficiency is not consistent with the dynamic of new information arrival (Easley et al. (2002)). In
fact, commenting on the exclusion of information quality as a determinant of required rate of returns, Easley and O’Hara (2004) argue
that “[t]his exclusion is particularly puzzling given the presumed importance of market efficiency in asset pricing. If information matters for the
market, why then should it not also matter for the firms that are in it?”
Intriguingly, Easley and O’Hara’s statement is in line with those of regulators and standard setters. For instance, Neel Foster, a
former member of the Financial Accounting Standards Board (FASB), notes that “[m]ore information always equates to less uncertainty,
and it is clear that people pay more for certainty. Less uncertainty results in less risk and a consequent lower premium being demanded. In the
context of financial information, the end result is that better disclosure results in a lower cost of capital.” (June 2003).4 More recently, Mary
Schapiro, former chairman of the U.S. Securities and Exchange Commission (SEC), states that the primary focus of the agency is to
“ensure that investors have the information that they need in a form that is helpful to them to make decisions about the allocation of their capital”
(Schapiro (2011)).5 Thus, from the regulator and standard-setter viewpoint, information is important with respect to required returns.
Yet, and similar to the theoretical literature, the empirical research linking information quality to cost of equity capital is unable to
produce conclusive evidence on this issue.6
The lack of consensus in the literature therefore suggests that the effect of information quality on firm cost of equity capital remains
an empirical question. The present study contributes to this debate using measures of information quality rooted in the growing
literature on textual disclosure. A common feature that characterizes prior empirical studies is that they are centered mainly on
quantitative information in their appraisal of disclosure quality, and overlook a key component of corporate reporting, namely,
narrative disclosure. Beyond financial statement numbers, narrative disclosure is a ubiquitous feature in the management reporting
environment and is considered a rich repository of value-relevant information that could be harnessed to enhance decision making by
investors, creditors, and other capital providers. For instance, more than 80% of financial analysts surveyed in AIMR (2000) report that
they rely on textual data when valuing firms. Moreover, thanks to advancements in computational linguistics, content analysis, and
natural language processing, textual disclosure complements our understanding of management disclosure choices and the subsequent
economic outcomes. Such narratives can provide a useful setting for understanding the generation process of numerical financial data
(Li (2010)), and they offer a promising way to gauge firms’ financial reporting quality (e.g., Core (2001); Beyer et al. (2010); Berger
(2011)).
Echoing Core’s (2001) call for more research that advances the use of natural language research in accounting and finance, a
number of papers have examined the economic consequences of narrative disclosure on, among others, capital investment efficiency
(e.g., Biddle et al. (2009)), earnings quality (e.g., Li (2008)), stock market efficiency (e.g., Lawrence (2013)), litigation risk (e.g.,
Nelson and Pritchard (2016)), and bond ratings (e.g., Bonsall and Miller (2017)). To contribute to this growing stream of literature, we
examine the association between annual report readability and cost of equity capital. In particular, we hypothesize that if low
readability increases estimation risk and information asymmetry, then firms with hard-to-read annual reports should have higher cost
of equity. While the primary focus of this study is on the effect of disclosure readability on cost of equity capital, using disclosure tone
of corporate filings as proxy for information content, we also examine the view that the interaction between the quality of information
disclosure and cost of capital depends on the content of the information releases from a firm. Precisely, we hypothesize that the
potential adverse effect of the level of annual reports complexity on cost of equity capital is more pronounced when disclosure tone is
more negative or more ambiguous.
Using a large panel of U.S. public firms over the 1995–2017 period, we find that greater textual complexity is associated with higher
cost of equity financing, even after controlling for firm-specific factors identified in prior studies as systematically affecting cost of
equity capital. Our estimates indicate that the effect of readability is statistically and economically significant. In fact, an increase in
annual reports’ textual complexity from the first to the third quartile results in about 54 basis point increase in cost of equity capital,
ceteris paribus. To carefully examine the reliability of our findings, we perform a battery of sensitivity checks, including the use of
alternative model estimation procedures and various proxies of textual complexity and cost of equity measures. Moreover, we
implement several approaches to rule out alternative explanations and alleviate endogeneity concerns. For instance, to mitigate the
concern that our results are driven by business complexity, we regress our readability proxies against a large set of factors deemed to
2
See, for example, Botosan (1997), Botosan and Plumlee (2002), Hughes et al. (2007), Lambert et al. (2007), Core et al. (2008), El Ghoul et al.
(2013), Johnstone (2016), and Larson and Resutek (2017).
3
Following Easley and O’Hara (2004), among others, we use the terms “information risk” and “information quality” interchangeably. Similarly,
the terms “cost of equity capital,” “required rate of return,” and “equity risk premium” are used as substitutes throughout the paper.
4
Available at: http://www.fasb.org/articles&reports/Foster_FASBReport.pdf
5
In the same vein, Arthur Levitt, also a former chairman of the SEC, states that “[q]uality information is the lifeblood of strong, vibrant markets.
Without it, liquidity dries up. Fair and efficient markets cease to exist.” (Levitt (1999)).
6
See, for instance, Botosan (1997), Easley et al. (2002), Francis et al. (2005a), Ecker et al. (2006), Cohen (2008), Core et al. (2008), Duarte and
Young (2009), Mohanram and Rajgopal (2009), Easley et al. (2010), Ng (2011), Aslan et al. (2011), Ogneva (2012), El Ghoul et al. (2013), and Liu
and Wysocki (2016).
2
capture firm complexity, and then use the residuals as our proxy of annual report readability. We also show that the effect of textual
complexity on cost of equity capital is more pronounced in ambiguous annual reports. In particular, we find that the adverse effect of
filing complexity on cost of equity capital increases with how negative or ambiguous (i.e., uncertain/weak model) the annual report
disclosure tone is.
We augment our paper with several important additional analyses. First, we investigate the moderating role of sophisticated in
formation users (i.e., institutional investors and security analysts) on the relation between annual report readability and cost of equity
capital and find that the effect of textual complexity decreases with the level of institutional ownership and analyst coverage. Second,
we examine the role of imperfect stock market competition and show that the effect of annual readability on the cost of equity depends
on the level of competition in equity markets. Specifically, we find that the lack of annual report readability affects cost of equity
capital only in situation of imperfect competition. Third, we consider the role of financial statement comparability on the relation
between annual report readability and cost of equity capital. We document that the effect of textual complexity is less pronounced for
firms with more comparable financial statements.7 This result suggests that comparability reduces the cost of information acquisition
associated with low annual report readability and enhances investor confidence in firm disclosure (De Franco et al. (2011)).8 Finally,
we examine the role of the plain English disclosure rule mandated by the SEC in 1998 and find that the magnitude of the effect of
annual reports’ textual complexity on cost of equity is less pronounced during the post-1998 period.
This study contributes to the long-standing stream of research examining the implications of corporate disclosure on cost of equity
capital (e.g., Francis et al. (2005b); Kim and Shi (2011); Cao et al. (2017)). Existing studies focus mainly on quantitative disclosure in
their appraisal of disclosure quality. With respect to non-quantitative disclosure, a growing trend of research examines the economic
outcomes of textual information (e.g., Li (2008); Loughran and McDonald (2009); Miller (2010); Lee (2012); Lawrence (2013); Bonsall
and Miller (2017)). Our study contributes to this literature by focusing on the effect of annual report readability on cost of equity
financing. Our paper adds also to the studies examining whether information environment matters to cost of equity capital in the
context of imperfect market competition (Armstrong et al. (2011); Lambert et al. (2012)).
This paper contributes also to the growing literature on the determinants of cost of equity capital (e.g., Gode and Mohanram (2003);
Daouk et al. (2006); Hail and Leuz (2009); Chava and Purnanandam (2010); Hann et al. (2013); Dhaliwal et al. (2014)). Specifically, it
expands the scope of research focusing on firm-level factors of cost of equity capital, such as information disclosure (e.g., Botosan and
Plumlee (2002); Francis et al. (2005b)), firm-level governance (e.g., Cheng et al. (2006); Chen et al. (2009)), firm geographic location
(El Ghoul et al. (2013)), corporate diversification (e.g., Hann et al. (2013)), and corporate social responsibility (El Ghoul et al. (2011);
Dhaliwal et al. (2014)).
Finally, besides its contributions to several streams of literature, our paper has practical implications not only for public firms but
also for policy makers. Easley and O’Hara (2004) conclude that if “the quality of information affects asset pricing, then how information is
provided to the markets is clearly important.” (Page 1578). By uncovering an association between annual reports’ readability and cost of
equity capital, our work provides empirical support to regulators’ claim on the importance of firms providing clear and concise
disclosure to their stakeholders.
The rest of the paper proceeds as follows. Section 2 reviews the related literature and develops our hypotheses. Section 3 describes
data sources and the variables used in this study. Section 4 reports summary statistics and correlations among variables. We present our
empirical findings in Section 5. Sections 6 and 7 carry out additional analyses and perform a battery of robustness checks, respectively.
Section 8 concludes the paper.
2. Review of related literature and hypothesis development
2.1. Issues with the choice of information quality measure
The link between information quality and cost of equity capital is the subject of an ongoing debate in accounting and finance
research. The influential Easley and O’Hara (2004) put forth a rational expectations equilibrium model where uninformed investors
require a higher premium to compensate for the risk that they may be trading against their informed peers, which brings a higher cost
of equity capital. However, Hughes et al. (2007) argue that the Easley and O’Hara’s pricing effect stems from under-diversification
given the finite number of assets in their model. Yet, Lambert et al. (2007) find that disclosure quality affects investor assessment
of the covariance between firm and market cash flows, and that this effect cannot be diversified in large economies.
Similar to the theoretical literature, there is no consensus in the empirical literature on whether information risk is priced. For
instance, using a measure of private information (PIN) as a proxy for information asymmetry, Easley et al. (2002) and Easley et al.
(2010) report results supporting the Easley and O’Hara (2004) theoretical model. Casting doubt on Easley et al. (2002) and Easley et al.
(2010) findings, Aktas et al. (2007) and Duarte and Young (2009) report evidence against the use of PIN as a proxy for information
7
De Franco et al. (2011) define financial statement comparability as how similarly two firms’ accounting systems provide similar mapping of
economic events in financial statements.
8
Moreover, estimation risk is likely to be lower for firms with comparable peers, since investors can incorporate information from those peers as
an additional input in their cash flow forecasts.
3
quality. Employing other proxies of information asymmetry, several empirical papers examine whether information quality matters to
firm cost of equity capital.9 While reporting evidence in support of the theoretical findings of Easley and O’Hara (2004), these studies
faced similar criticism as those using PIN (e.g., Core et al. (2008)).
One way to address the above issues is to separate the factors affecting cost of equity capital from those related to information risk.
However, this task is far from trivial, which may explain the mixed empirical results. Motivated by the growing literature on annual
report readability, we identify such a construct, generating an opportunity to efficiently examine whether information quality affects
firm equity financing costs.
2.2. Annual report readability and cost of equity
In this paper, we argue that linguistic features of annual reports may have an effect on firms’ cost of equity capital. This effect comes
about through two possible channels. First, firms disclosing more complex (i.e., longer and difficult to read) annual reports are more
likely to suffer from severe information asymmetry problems. The premise underlying this argument is that managers of these firms
tend to strategically structure annual reports (e.g., use of sesquipedalian prose) to obfuscate their poor performance or to conceal bad
news from investors (e.g., Li (2008); Lo et al. (2017)). The seminal Li (2008) finds that 10-K SEC filings of poor performing firms are
longer and harder to read. Li (2008) also reports a negative association between annual report readability and earnings persistence.
More recently, Lo et al. (2017) show that firms that manage their earnings upward to meet (or just beat) earnings targets tend to
disguise their opportunistic activities by making their disclosure less readable and harder to decipher. Ertugrul et al. (2017) argue that
in the very situation where managers attempt to obfuscate mandated earnings-relevant information, they tend to bury the results in
less readable and more ambiguous disclosure. Chakrabarty et al. (2018) find that CEOs strategically obfuscate disclosures by issuing
less readable annual reports when given greater options compensation (i.e., higher risk incentives).10
Second, more complex narratives are expected to alter users’ understanding of future firm performance and skew their judgments,
which may generate greater uncertainty about the underlying parameters of the firm’s stock return distribution and increase
disagreement among investors on how to interpret public signals. Analytically, Bloomfield and Fischer (2011) demonstrate that
opinion divergence decreases the precision of investor beliefs about future stock returns, leading to higher cost of equity capital.
Rennekamp (2012) provides experimental evidence that less readable disclosure reduces investor belief that they can rely on infor
mation in a disclosure, which affects their subsequent judgments and decisions. In the same vein, Elliott et al. (2015) find that concrete
language, which use is recommended by the SEC’s plain English disclosure rule, increases investors’ feelings of comfort in their ability
to evaluate an investment.
The effect of readability on capital market outcome variables has been well documented in prior literature. For instance, Courtis
(1986) is among the first studies examining the relationship between a firm’s annual report readability and its level of risk and return.
The author does not find, however, a strong relationship between annual report readability and the level of risk and return.11 Loughran
and McDonald (2009) and Miller (2010) show that harder-to-read annual reports discourage investors from processing information,
thereby decreasing stock trading activity. Using detailed shareholding data from a U.S. brokerage house, Lawrence (2013) documents
that retail investors are more willing to hold securities of firms with more readable financial reports. Lee (2012) and Callen et al.
(2013) examine the effect of readability on the information efficiency of stock prices. Lee (2012) finds that more textually complex
quarterly filings hinder market’s ability to incorporate earnings-related information in a timely manner. Similarly, Callen et al. (2013)
document that lower 10-K readability is associated with less timely price adjustment and higher future stock returns. More recently,
Bai et al. (2019) show that less readable financial reports decrease the amount of firm-specific information available to the market,
which increases reliance on market-wide information and increases the synchronicity of stock price movements. Additionally, Kim
et al. (2019) report that information opacity resulting from more complex 10-K narratives leads to large negative return skewness (i.e.,
stock price crash risk).
Our paper is closely related to two recent papers that have also examined the effect of annual report textual disclosure on the cost of
equity capital. Garel et al. (2019) show that U.S. firms issuing less readable 10-K filings face a higher cost of equity financing. Focusing
on the U.K. context, Athanasakou et al. (2020) find a U-shaped relationship between cost of equity and level of disclosure in annual
report narratives. There are several key differences between these two papers and our study. First, Garel et al. (2019) rely on the
Gunning-Fog index in measuring report readability. We, however, employ the Bog index, a measure proposed by Bonsall et al. (2017)
that draws on the SEC’s plain English disclosure rule. Second, our study differs from Athanasakou et al. (2020) in that they focus on the
volume of U.K. annual report narratives using a word-count-based index, while our focus is on the readability of 10-K filings. Third, our
paper differs from Garel et al. (2019) and Athanasakou et al. (2020) in that: i) we examine an important alternative hypothesis stating
that better information disclosure can increase a firm’s cost of equity capital under certain conditions (Table 5); ii) we examine the
moderating role of institutional ownership and analyst coverage (Table 6); and iii) we test Lambert et al.’s (2012) theoretical
9
These proxies include abnormal accruals in accounting earnings, dispersion in analyst earnings forecasts, and level of information disclosure (e.
g., Bhattacharya et al. (2003); Botosan et al. (2004); Francis et al. (2004); Aboody et al. (2005); Francis et al. (2005a); Ecker et al. (2006); Core et al.
(2008)).
10
Apart from managers’ purposeful engagement in information obfuscation, increased linguistic complexity per se requires more time and effort
on the part of readers to process filings and extract useful information. Indeed, several papers report evidence of investors underreacting to in
formation from less readable filings (e.g., You and Zhang (2009); Rennekamp (2012); Tan et al. (2015); Koonce et al. (2018)).
11
The findings of Courtis (1986) are constrained by the very small sample size (Li (2008)).
4
prediction that information quality affects cost of equity capital only in situations of imperfect market competition (Table 7).
On the basis of the previous discussion, we seek to extend the extent literature by examining the effect of 10-K readability on the
cost of equity financing. Formally, we propose the following testable hypothesis:
H1. : Firm cost of equity capital, ceteris paribus, increases with the level of complexity of annual reports.
2.3. The role of annual report disclosure tone
Finding support for H1 would corroborate the widely held view that better information disclosure reduces a firm’s cost of financing.
There is, however, a number of papers that challenge the view that better information disclosure leads unambiguously to a lower cost
of equity financing (e.g., Botosan and Plumlee (2002); Gao (2010); Dempsey et al. (2015); Johnstone (2015); Johnstone (2016); Larson
and Resutek (2017)). For instance, Johnstone (2016) argues that the interaction between the quality of information disclosure and cost
of capital depends on the content of the information releases from a firm (i.e., good or bad news). Specifically, Johnstone (2016)
demonstrates that if information changes investor’s assessment about expected payoff (i.e., the mean effect), bad (good) news in
creases (decreases) the uncertainty of future expected payoff and thus drives cost of equity capital upward (downward). To test
Johnstone’s (2016) proposition in our context, we use negative disclosure tone as a proxy for bad news and contend that the degree of
filing complexity (i.e., less readable filings) is negatively associated with cost of equity capital when disclosure tone is negative.
Similar to negative words, uncertain words may also change the relation between information quality and cost of capital. Empirical
evidence from, among others, Li (2006), Kravet and Muslu (2013) and Campbell et al. (2014), suggests that increases in textual risk
disclosures increase users’ risk perceptions. For instance, Li (2006) shows that an increase in textual risk disclosures, measured by
count of words risk’ and ‘uncertainty’, is associated with lower future earnings and poor future stock returns. Using a broad list of word
roots indicating risk and uncertainty to gauge risk disclosure, Kravet and Muslu (2013) show that changes in textual risk disclosure are
positively associated with changes in stock return volatility, changes in trading volume, as well as changes in volatility of analysts’
forecast revisions. Employing Loughran and McDonald’s (2011) textual analysis dictionary, Bonsall and Miller (2017) document that
firms’ use of uncertain language in business disclosure is associated with higher cost of corporate bond financing, less favorable bond
ratings, and higher disagreement among bond rating agencies. In a contemporary study, Ertugrul et al. (2017) use Loughran and
McDonald’s (2011) word lists to examine the effect of a firm’s annual report ambiguous tone on its cost of debt. They find that
management’s use of uncertain tone (reflecting imprecision) and weak modal tone (reflecting lack of confidence) in corporate filings is
associated with more stringent contractual terms in bank loans and higher cost of borrowing. Consistent with the view that ambiguous
(i.e., uncertain/weak modal) tone could increase a firm’s perceived information risk, Loughran and McDonald (2011) show that a
larger fraction of uncertain and weak modal words in 10-K filings leads to higher stock return volatility. In the same vein, Loughran and
McDonald (2013) examine the effect of language sentiment in S-1 filings on the valuation of initial public offerings, and find that the
use of negative, uncertain, and weak modal words is positively associated with first-day returns, offer price revisions, and subsequent
volatility. More recently, Irtisam et al. (2020) show that firm’s use of negative, uncertain and weak modal tone in 10-K and 10-Q
reduces information asymmetry between sophisticated and unsophisticated traders. Irtisam et al.’s findings suggest that the use of
ambiguous tone diminishes the ability of sophisticated traders to extract useful information from corporate filings, which reduces their
information advantage over unsophisticated traders.
Taken together, the literature suggests that firm’s use of negative, uncertain, or weak modal words in corporate disclosure increases
the range of investors’ predictions of future performance and decreases investors’ confidence in their predictions. It follows that
corporate disclosure reflecting negative or ambiguous (uncertain/weak modal) tone could increase a firm’s perceived information risk,
which is then priced by investors. Accordingly, we hypothesize that the potential adverse effect of filing complexity on a firm’s cost of
equity capital increases with the fraction of negative, uncertain, or weak modal words in 10-K filings. Hence our second hypothesis:
H2. : The adverse effect of the level of complexity of annual reports on cost of equity capital is more pronounced when disclosure tone is more
negative or more ambiguous.
3. Data and variable construction
This section describes our data sources and construction of the variables used in the empirical analysis. Definitions and data sources
for the regression variables are found in the Appendix.
3.1. Data sources
To examine the relation between annual report readability and cost of equity capital, we begin by merging three databases: (i)
Compustat, which provides financial data; (ii) the Center for Research in Security Prices (CRSP), which provides information on stock
prices; and (iii) I/B/E/S, which contains data on analyst forecasts, dispersion of forecasts, and growth estimates. We require firms to
have one-year-ahead and two-year-ahead earnings forecasts, a long-term growth estimate (if not available, a three-year-ahead
earnings forecast), book values, and stock prices to compute implied cost of equity estimates. Firms that do not have a valid
5
construct of their cost of equity capital and have missing accounting and financial data are dropped. We also exclude foreign firms
listed on U.S. stock exchanges. We obtain our main annual report readability measure (i.e., Bog readability index) from Brian Miller’s
website.12 Our final sample is composed of 39,181 firm–year observations over the 1995–2017 period.
3.2. Regression variables
3.2.1. Cost of equity capital

There is extensive evidence implying that realized returns are exceedingly noisy. Elton (1999) for example, concludes that “realized
returns are a very poor measure of expected returns” even over a long period; a statement that is further confirmed by Easton and
Monahan (2005) and Lundblad (2007). Fama and French (1997) further stress that uncertainty in the factor premiums and imprecision
in the factor loading estimates lead to inaccurate estimation of expected returns by ex-post returns and asset-pricing models. Moreover,
Bekaert and Harvey (2000) and Hail and Leuz (2006) argue that ex-post stock returns are likely to capture external shocks to the
growth opportunities of companies. More recently, Lee et al. (2009) add that the use of implied cost of equity capital offers clear
evidence of economic relations that would otherwise be concealed by the noise in realized returns. In addition, an increasing number of
studies comparing the ex-ante and ex-post approaches provide support to the former as a better approach to proxy for expected returns
(e.g., Pástor et al. (2008); Li et al. (2013); Lee et al. (2020)). Li et al. (2013), for example, document that the implied cost of equity
approach is substantially better in predicting future market returns than traditional valuation measures, such as book-to-market ratio,
dividend-to-price-ratio, earnings-to-price ratio, and payout yield. More recently, Lee et al. (2020), show that factor-based estimates
perform poorly relative to implied cost of equity measures in terms of cross-sectional and time-series measurement-error variance.
Accordingly, to examine the effect of annual report readability on equity risk premium, we estimate the latter using the ex-ante cost
of equity implied in current stock prices and forecast earnings per share (minus the risk-free rate). Precisely, we start by employing a
widely used approach based on four popular implied cost of equity models developed by Claus and Thomas (2001), Gebhardt et al.
(2001), Easton (2004), and Ohlson and Juettner-Nauroth (2005).13 Afterwards, we calculate the implied equity risk premium by
subtracting the 10-year Treasury bond yield (as of June of year t) from each of the aforementioned cost of equity model estimates,
which we label rCT, rGLS, rMPEG, and rOJN, respectively. Same with all continuous variables used in this study, in order to mitigate the
potential impact of outliers, we winsorize each of those cost of equity premium measures at the 1st and 99th percentile levels.
All four models consist of valuation equations based on the idea that analyst forecasts represent investor expectations. However,
each model relies on its own set of assumptions regarding the explicit forecast horizon and growth expectations. Following the recent
empirical literature, to avoid spurious results that might be driven by a specific cost of equity premium measure, we calculate the
arithmetic average of the four different estimates, denoted rAVG.14 Although averaging across the four models in our primary analysis
ensures that the distinctive characteristics of any single model are not spuriously behind our evidence, we show that our core results
are robust to re-estimating our regressions using each individual cost of equity metric, the median, or the first principal component of
the four models.
The evidence supporting the ex-ante approach relative to the ex-post approach (as a proxy for future returns) does not mean that
implied cost of equity measures are “perfect” measures. In fact, implied cost of equity proxies may suffer from measurement errors as a
result of assumptions and inputs used to empirically implement the models (e.g., Easton (2009); Lambert (2009)). One of the main
sources of measurement error arises from analyst forecast bias and sluggishness. A number of studies, for example, show that consensus
analyst earnings per share and long-term growth forecasts are likely to be more optimistic for growth firms, which can induce sys
tematic upward bias in implied cost of equity for growth firms relative to value firms (e.g., Dechow and Sloan (1997); Frankel and Lee
(1998); Guay et al. (2011)).
Another source of measurement error in implied cost of equity estimates arises when expected returns are stochastic (Hughes et al.
(2009)). Because implied cost of equity implicitly assumes constant expected returns, it will differ from expected returns and result in
measurement errors that can be correlated with variables that are traditionally not related to systemic risk exposure. Hughes et al.
(2009) stress the following variables, which are documented in the literature to be correlated with implied cost of equity: leverage,
expected earnings growth, and forecast dispersion.
We carry out a battery of robustness checks to address concern over measurement errors induced by assumptions embedded in the
implied cost of equity models. For instance, we follow the literature and address analyst bias and sluggishness in several ways. We also
control for leverage, expected earnings growth, and forecast dispersion to lessen concerns over measurement errors resulting from
assuming constant expected returns when they are stochastic.
To further reinforce the reliability of our findings, we repeat our analyses using realized returns as a proxy for future returns. In fact,
Wang (2015) suggests that to make robust inferences on expected returns, empirical studies should complement implied cost of equity
12
We are grateful to Brian Miller for kindly sharing the Bog index data (available at: https://kelley.iu.edu/bpm/activities/bogindex.html). Details
on calculating the Bog index are available at http://www.stylewriter-usa.com/stylewriter-editing-readability.php
13
See, among others, Hail and Leuz (2006), Chen et al. (2009), El Ghoul et al. (2012), Attig et al. (2013), Boubakri et al. (2014), Boubakri et al.
(2016), Dhaliwal et al. (2016), Ferris et al. (2017), and Breuer et al. (2018). This approach is also consistent with growing evidence reinforcing the
importance of avoiding specifying a single implied cost of equity capital estimate when examining the determinants of equity pricing (e.g., Botosan
and Plumlee (2002); Dhaliwal et al. (2006); Guay et al. (2011)).
14
See, among others, Hail and Leuz (2006), Chen et al. (2009), El Ghoul et al. (2012), Attig et al. (2013), Chen et al. (2016), Dhaliwal et al. (2016),
and Ferris et al. (2017).
6
measures through a realized-returns measure. To do so, we follow Hann et al. (2013) and complement our main implied cost of equity
measure by realized equity returns computed as buy-and-hold returns accumulated over 12 months starting in June of year t + 1 minus
the rate on a 10-year treasury note. We denote this measure as REARET.
To alleviate concern over the noisy nature of realized returns, Hann et al. (2013) consider another cost of equity measure that
combines information from ex-post and ex-ante approaches. Precisely, we regress ex-post realized returns on a set of six ex-ante
measures of cost of equity premium and use the fitted value from the regression as the proxy for expected returns. The ex-ante
measures that we consider at the first-stage regression are: a) implied cost of equity premium proposed by Claus and Thomas
(2001); b) implied cost of equity premium derived from Gebhardt et al. (2001); c) implied cost of equity premium designed by Easton
(2004); d) expected returns from the Fama–French three-factor model; e) earnings yield; and f) earnings yield adjusted for growth. We
denote this measure of “instrumented” returns as INSRET.
3.2.2. Annual report readability

A great deal of empirical accounting research employs the Fog index to assess financial disclosure readability (e.g., Biddle et al.
(2009); Miller (2010); Ramanna and Watts (2012); Lawrence (2013); Merkley (2014); Frankel et al. (2016)). However, Loughran and
McDonald (2014) raise concerns regarding the appropriateness of the Fog index in capturing complexity in corporate disclosure. For
example, the authors note that words like operations and telecommunications are mechanically classified as complex according to the
Fog index, while in fact they are easy for investors to comprehend. Loughran and McDonald (2014) propose hence using the size of the
filing (in megabytes) as a proxy for financial readability. More recently, Bonsall et al. (2017) argue that the file size measure proposed
by Loughran and McDonald (2014) is a rather noisy proxy for readability, since large document size may be due to inclusion of el
ements unrelated to the underlying text in the disclosure (e.g., embedded images, PDF files, HTML, and XML tags). Given these
limitations, the current paper employs the Bog index (BOG) validated and recommended by Bonsall et al. (2017) as primary measure of
annual report readability. The Bog index is based on plain English writing attributes (e.g., passive and hidden verbs, complex, abstract,
and legal words, among others) and captures most of the SEC’s plain English writing guidelines (see, SEC (1998)).15 Higher BOG values
indicate more complex annual filings.16
3.2.3. Control variables

To single out the incremental role that textual complexity plays in equity pricing, we include a set of control variables shown to
systematically affect cost of equity capital (e.g., Dhaliwal et al. (2006); Hail and Leuz (2006); Chen et al. (2011)). These variables
include market beta (BETA), book-to-market ratio (BTM), firm size (SIZE), financial leverage (LEV), long-term earnings growth rate
(LTG), analysts’ forecast bias (FBIAS), and dispersion of analyst forecasts (DISP). We control for industry fixed effects using the 48-in
dustry classification scheme of Fama and French (1997). In addition, year dummies are included to address potential time-series
differences in cost of equity capital, and to control for changing macroeconomic conditions through time.
Market beta (BETA)

We include market beta in the regressions to control for a firm’s systematic risk. It is obtained from estimating the market model
based on CRSP monthly stock returns over the preceding 60 months and ending in June of each year.
Book-to-market ratio (BTM)

Following Hail and Leuz (2006), we include book-to-market ratio to control for differences in firms’ growth opportunities. Firms
with high growth prospects are expected to have higher prices and to generate higher long-term growth in cash flows, which translates
to a lower cost of equity capital. We proxy for growth opportunities using the ratio of book value of equity to market value of equity.
Firm size (SIZE)

Diamond and Verrecchia (1991) demonstrate that greater information disclosure can lead to higher prices, which in turn lowers
risk premium. More importantly, these authors show that the upward price movement resulting from greater information availability
is more pronounced for larger firms. Further, Gebhardt et al. (2001) and Gode and Mohanram (2003) document a negative association
between firm size and implied cost of equity capital. We proxy for firm size via natural logarithm of total assets.
Financial leverage (LEV)

Gode and Mohanram (2003) and Botosan and Plumlee (2005) argue that levered firms are more exposed to financial distress and
hence face higher risk. Both papers provide evidence that leverage is positively associated with cost of equity capital. We measure firm
leverage as the ratio of long-term debt to total assets.
15
Bonsall et al. (2017) show that using the Fog index or file size to capture the multidimensional attributes of the SEC’s plain English rule is nearly
impossible.
16
We test the robustness of our results to the use of alternative readability proxies, namely, total document length (LENGTH), the natural logarithm
of file size (GROSSFSIZE), and net file size (NETFSIZE). The results remain qualitatively the same. The Appendix contains detailed definitions of
these variables.
7
Long-term earnings growth (LTG)

Gebhardt et al. (2001) argue that firms with higher long-term growth face greater downward pressure on their equity risk premium.
This argument is based on the findings of La Porta (1996), who shows that analyst forecasts are, on average, more optimistic for firms
with higher long-term growth, which leads to higher stock prices. In contrast, Gode and Mohanram (2003) contend that high-growth
firms are perceived as risky, since any errors in the estimation of growth can have a significant impact on prices. We proxy for long-
term growth using the I/B/E/S long-term earnings growth rate.
Analysts’ forecast bias (FBIAS)

Easton and Sommers (2007) argue that analyst earnings forecasts are optimistically biased, which introduces an upward bias in the
implied cost of equity estimates. In our regressions, we control for forecast optimism bias (FBIAS), defined as the difference between
the one-year ahead consensus earnings forecast and realized earnings deflated by end-of-June stock price.
Dispersion of analyst forecasts (DISP)

Previous research suggests that firms with wider analyst dispersion are assigned a higher equity risk premium (e.g., Botosan et al.
(2004)). We control for forecast dispersion using the coefficient of variation of one-year-ahead analyst earnings forecast per share as of
June in year t.
4. Descriptive statistics
Table 1 provides summary statistics for the sample of 39,181 firm–year observations between 1995 and 2017. The mean and the
median cost of equity premium for our sample are 6.32% and 5.81%, respectively, with an interquartile range from 4.43% to 7.55%.
Overall, these descriptive statistics are comparable to other studies using the same approach of computing cost of equity capital (i.e.,
rAVG). For example, Dhaliwal et al. (2006), Chen et al. (2011), El Ghoul et al. (2013), and Mishra and O’Brien (2019) report an average
equity risk premium of 4.95%, 4.94%, 5.64%, and 7.56%, respectively. BOG has a mean of 83.59 and a median value of 84.00,
suggesting that firms’ annual reports are, on average, difficult to read. Moreover, BOG displays a substantial cross-sectional variation,
as evidenced in the standard deviation and the interquartile range of 7.54 and 10.00, respectively.
Table 2 presents Pearson correlations among all regression variables. Consistent with our theoretical prediction, BOG displays a
positive and significant (at the 1% level) correlation with rAVG, which lends preliminary support to the hypothesized relation between
annual report readability and cost of equity capital. In addition, we do not find high correlations between the independent variables,
which assures that multicollinearity is less likely to be a concern.17
5. Results
5.1. Univariate tests
Univariate tests, reported in Table 3 , compare the mean and median cost of equity premiums for firms with low and high read
ability scores based on the median readability value. The mean (median) equity premium of firms with low BOG score (i.e., higher
readability) is 5.131% (4.209%), while it is 5.698% (5.061%) for firms with high BOG score (i.e., lower readability). These results
suggest that the mean (median) cost of equity capital for firms with high BOG score is 56.7 (85.2) basis points higher than that of firms
with low BOG score. These differences are significant at the 1% level, providing preliminary support for the prediction that firms with
more readable annual filings enjoy cheaper cost of equity financing.
5.2. Main regression analysis
This subsection investigates the relation between annual report readability and cost of equity capital by estimating the following
panel regression model:
rAVGit = β0 + β1 BOGit + β2 BETAit + β3 BTM it + β4 SIZEit + β5 LEV it + β6 LTGit + β7 FBIASit + β8 DISPit + IndustryFE + YearFE + εit (1)
where rAVG is our measure of cost of equity premium defined as the arithmetic average of cost of equity premium estimates obtained
from the models of Claus and Thomas (2001), Gebhardt et al. (2001), Easton (2004), and Ohlson and Juettner-Nauroth (2005). BOG is
the Bog readability index obtained from Brian Miller’s website, with higher values indicating more difficult to read filings.
Table 4 portrays the results from regression analysis of the relation between annual report readability and cost of equity capital,
after controlling for other potential determinants of cost of equity capital, but also for industry- and time-fixed effects. We estimate Eq.
(1) using a pooled cross-sectional time-series regression with robust standard errors clustered at the firm-level. Column 1 of Table 4
shows that the coefficient on BOG is positive and statistically significant at less than the 1% threshold, suggesting that firms with less
readable annual reports are penalized with a higher cost of equity capital.
Our main results from Column 1 rely on robust standard errors clustered at the firm-level, and hence correct for time-series
17
For all regressions, we verify that multicollinearity is not a factor in that we calculate variance inflation factors (VIF) and confirm they are within
reasonable ranges.
8
Table 1
Full sample summary statistics.
Variables Number Mean Standard deviation First quartile Median Third quartile
rAVG (%) 39,181 6.32 2.89 4.43 5.81 7.55

BOG 39,181 83.59 7.54 79.00 84.00 89.00
BETA 39,181 1.10 0.80 0.56 0.95 1.46
BTM 39,181 0.55 0.43 0.28 0.47 0.71
LEV 39,181 0.18 0.18 0.01 0.13 0.28
SIZE 39,181 7.16 1.84 5.85 7.06 8.33
LTG 39,181 15.32 8.49 10.00 13.75 19.00
FBIAS 39,181 0.20 1.99 − 0.33 − 0.01 0.32
DISP 39,181 0.10 0.25 0.01 0.03 0.07
This table reports descriptive statistics for 39,181 firm-year observations from 1995 to 2017. rAVG is the average cost of equity capital measure
obtained from models developed by Easton (2004), Ohlson and Juettner-Nauroth (2005), Claus and Thomas (2001), and Gebhardt et al. (2001),
which we denote rCT, rGLS, rMPEG, and rOJN, respectively. BOG is a proprietary measure of readability created by Editor Software’s plain English
software, StyleWriter. The formula is based on several plain English factors such as sentence length, passive voice, weak verbs, overused words,
complex words, and jargon. BETA is market beta obtained from regressions of firms’ monthly excess stock returns on the corresponding CRSP value-
weighted index excess returns, using at least 24 (and up to 60) months and ending in June of each year. BTM is the ratio of the book value to the
market value of equity. SIZE is the natural logarithm of total assets in $ millions. LEV is the leverage ratio defined as the ratio of long-term debt to total
assets. LTG is the average long-term growth in forecasted earnings. FBIAS is the signed forecast error, which is defined as the difference between the
one-year-ahead consensus earnings forecast and realized earnings deflated by end of June stock price. DISP is the dispersion of analyst forecasts
defined as the coefficient of variation of one-year-ahead analyst forecasts of earnings per share in June in year t.
Table 2
Correlation matrix.
rAVG BOG BETA BTM LEV SIZE LTG FBIAS DISP
rAVG 1
BOG 0.102*** 1
BETA 0.217*** 0.174*** 1
BTM 0.157*** − 0.048*** − 0.019*** 1
LEV 0.033*** 0.050*** -0.003 − 0.061*** 1
SIZE − 0.290*** − 0.014*** − 0.147*** 0.091*** 0.217*** 1
LTG 0.179*** 0.062*** 0.282*** − 0.230*** − 0.173*** − 0.443*** 1
FBIAS 0.190*** − 0.008 0.037*** 0.005 − 0.006 − 0.128*** 0.069*** 1
DISP 0.151*** 0.061*** 0.114*** 0.091*** 0.047*** − 0.054*** 0.061*** 0.110*** 1
This table reports Pearson correlations among the main variables used in this paper. rAVG is our cost of equity capital measure, defined as the average
of rCT, rGLS, rMPEG, and rOJN. BOG is a proprietary measure of readability created by Editor Software’s plain English software, StyleWriter. The formula
is based on several plain English factors such as sentence length, passive voice, weak verbs, overused words, complex words, and jargon. BETA is
market beta obtained from regressions of firms’ monthly excess stock returns on the corresponding CRSP value-weighted index excess returns, using
at least 24 (and up to 60) months and ending in June of each year. BTM is the ratio of the book value to the market value of equity. SIZE is the natural
logarithm of total assets in $ millions. LEV is the leverage ratio defined as the ratio of long-term debt to total assets. LTG is the average long-term
growth in forecasted earnings. FBIAS is the signed forecast error, which is defined as the difference between the one-year-ahead consensus earn
ings forecast and realized earnings deflated by end of June stock price. DISP is the dispersion of analyst forecasts defined as the coefficient of variation
of one-year-ahead analyst forecasts of earnings per share in June in year t. *** denotes statistical significance at the 1% level.
Table 3
Univariate tests.
Means Medians
(1) (2) (1)–(2) (3) (4) (3)–(4)
Low BOG High BOG Difference Low BOG High BOG Difference
(Obs. = 21,139) (Obs. = 18,042) (t − stat) (Obs. = 19,590) (Obs. = 19,591) (Z − stat)
rAVG 5.131 5.698 − 0.567 *** 4.209 5.061 − 0.852 ***

BETA 1.003 1.220 − 0.217 *** 0.751 1.111 − 0.360 ***
BTM 0.541 0.594 − 0.053 *** 0.474 0.507 − 0.033 ***
LEV 0.167 0.191 − 0.024 *** 0.114 0.124 − 0.010 ***
SIZE 7.267 7.0755 0.192 *** 6.102 5.709 0.393 ***
LTG 14.954 15.739 − 0.785 *** 13.430 14.270 − 0.840 ***
FBIAS 0.167 0.223 − 0.056 ** 0.008 0.032 − 0.024 ***
DISP 0.087 0.105 − 0.018 *** 0.034 0.059 − 0.025 ***
This table reports mean and median comparison tests for average cost of equity capital measure (rAVG) across subsamples of high (above the mean and
the median) and low (below the mean and the median) BOG scores. The sample contains 39,181 firm-year observations from 1995 to 2017. rAVG is our
cost of equity capital measure, defined as the average of rCT, rGLS, rMPEG, and rOJN. *** denotes statistical significance at the 1% level.
9
Table 4
The impact of financial readability on implied cost of equity capital.
Variable Clustering by Firm Fama–MacBeth Prais–Winsten Firm Fixed Effects Economic impact
(1) (2) (3) (4) (5)
BOG 0.054*** 0.050*** 0.065*** 0.032*** 0.54

0.01 0.01 0.01 0.01
BETA 1.629*** 1.630*** 1.514*** 0.506*** 1.46
0.11 0.19 0.12 0.14
BTM 3.987*** 3.892*** 3.362*** 3.437*** 1.71
0.22 0.33 0.24 0.29
LEV 4.811*** 4.843*** 4.573*** 5.180*** 1.29
0.42 0.45 0.43 0.59
SIZE − 0.586*** − 0.588*** − 0.749*** − 0.254* − 1.45
0.05 0.07 0.05 0.15
LTG 0.089*** 0.071*** 0.080*** 0.043*** 0.80
0.01 0.01 0.01 0.01
FBIAS 0.582*** 0.517*** 0.388*** 0.347*** 0.37
0.05 0.04 0.04 0.05
DISP 3.904*** 3.831*** 2.553*** 2.379*** 0.23
0.29 0.38 0.29 0.32
INTERCEPT 5.356*** 5.393*** 4.451*** 8.027***
0.89 0.62 1.03 1.49
Industry effects Yes Yes Yes No
Year effects Yes No Yes Yes
Cluster by firm Yes No Yes Yes
N 39,181 39,181 39,181 39,181
Adj. R2 0.211 0.253 0.256 0.091
This table shows the effect of annual report readability on the cost of equity capital using alternative econometric methodologies. The sample contains
39,181 firm-year observations from 1995 to 2017. The dependent variable rAVG is the average of rCT, rGLS, rMPEG, and rOJN. BOG denotes the variable
used to measure readability. Control variables include BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP. BETA is market beta obtained from regressions of
firms’ monthly excess stock returns on the corresponding CRSP value-weighted index excess returns, using at least 24 (and up to 60) months and
ending in June of each year. BTM is the ratio of the book value to the market value of equity. SIZE is the natural logarithm of total assets in $ millions.
LEV is the leverage ratio defined as the ratio of long-term debt to total assets. LTG is the average long-term growth in forecasted earnings. FBIAS is the
signed forecast error, which is defined as the difference between the one-year-ahead consensus earnings forecast and realized earnings deflated by end
of June stock price. DISP is the dispersion of analyst forecasts defined as the coefficient of variation of one-year-ahead analyst forecasts of earnings per
share in June in year t. Industry (based on the Fama and French (1997) 48-indsutry classification) and year dummies are also included in the re
gressions. Robust standard errors adjusted for clustering by firm are reported in parentheses. The superscript asterisks ***, **, and * denote two–tailed
statistical significance at the 1%, 5%, and 10% levels, respectively
dependence (i.e., residuals for a given firm are correlated across years). Cross-sectional dependence (i.e., residuals in a given year are
correlated across firms), however, remains a potential concern. In Column 2, we re-estimate our regressions using the Fama and
MacBeth (1973) two-step procedure, including industry dummies based on the Fama and French (1997) 48-industry classification. We
adjust the standard errors for heteroskedasticity and autocorrelation using the Newey and West (1987) correction. As suggested by
Cochrane (2001), the serial correlation in the estimated coefficients is captured using a first-order autoregressive process. Our evidence
regarding the role of readability in equity pricing continues to hold when we account for correlation of errors across observations using
the Prais-Winsten methodology in Column 3.18 Finally, we exploit the panel structure of the sample by running a fixed-effects spec
ification in Column 4. In each of these specifications, BOG always loads positive and is statistically significant at the 1% threshold
level.19
Our results are not only statistically significant but are also economically sizable. In particular, on average, an interquartile change
in BOG from the 25th to the 75th percentile of the sample distribution yields a 54 basis point increase in cost of equity premium. To put
this figure into perspective, the economic impact of higher systematic risk (BETA) results in a 146 basis point increase in cost of equity
18
In unreported analyses, available upon request, we find that our results continue to hold if we use Newey–West methodology and when we apply
the weighted least-squares methodology, using the inverse of the number of firm–year observations per industry as weights.
19
As discussed earlier in subsection 3.2.1, defining the cost of equity capital (rAVG) as the average of the four individual implied cost of equity
premium estimates (i.e., rCT, rGLS rMPEG, and rOJN) provides some assurance that our findings are not spuriously driven by assumptions particular to
each model. In unreported analyses, to shore up our inferences, we rely on alternative aggregation techniques by using the median and the first
principal component of four risk premium models. We find that our results are robust to the median and the first principal component approach
alternative specifications. In particular, consistent with the evidence presented in Column 1 of Table 4, the coefficient on BOG continues to be
positive and statistically significant at the 1% threshold level. To further probe the influence of readability on cost of equity, we re-estimate our
baseline regression, after replacing rAVG with the individual risk premiums rCT, rGLS, rMPEG, and rOJN, respectively. We do so to dispel concern that our
inferences are influenced by the particular assumptions of each individual model. We find that the coefficient on BOG is positive and statistically
significant at the 1% level, lending further support to our prediction that firms with less readable filings face a higher cost of equity capital. All of the
above results are available upon request from the authors.
10
premium.
Turning to our control variables, consistent with previous research, we find that firms with higher systematic risk, firms with lower
growth prospects, firms with higher leverage, firms with higher long-term earnings growth, and firms with wider analyst dispersion
face higher equity financing costs. Moreover, we document that higher information availability, as proxied by firm size, decreases
equity risk premium. This result is consistent with Fama and French (1992) and Gebhardt et al. (2001). We also find a positive and
statistically significant coefficient on FBIAS, confirming that analyst optimism inflates equity risk premium estimates.
The results from Table 4 are in line with the common view that better information disclosure decreases investor uncertainty about
future cash flows, and hence lowers cost of equity capital. However, as stated above, there is a growing body of literature suggesting
that better information disclosure has a rather ambiguous effect on cost of capital and that is because the effect of information on cash
flow uncertainty depends on the content of the disclosure (e.g., Gao (2010); Dempsey et al. (2015); Johnstone (2016); Larson and
Resutek (2017)). For instance, Johnstone (2016) points out that precise information that signal bad news increases uncertainty, and
thereby increases cost of equity capital. Several empirical studies have shown that the greater the fraction of negative, uncertain, or
weak modal words in corporate filings the higher is a firm’s perceived information risk (e.g., Li (2006); Kravet and Muslu (2013);
Loughran and McDonald (2013); Bonsall and Miller (2017); Irtisam et al. (2020)). In this paper, we hypothesize that if the firm’s
perceived information risk is priced by investors, then the adverse effect of filing complexity on the cost of equity capital increases with
the fraction of negative, uncertain, or weak modal words in 10-K filings. To test this hypothesis (H2), we use three sentiment word lists
constructed by Loughran and McDonald (2011) (i.e., negative, uncertain, and weak modal) on the basis of their likely appearance in
financial documents. One advantage of using these word lists is that they were built for the specific context of financial reporting.20 We
define NEGATIVE as the percentage of negative words (e.g., “loss,” “failure,” “bankruptcy”). UNCERATAIN is the percentage of un
certain words (e.g., “arbitrary,” “confusing,” “likelihood”). WEAK_ MODAL is measured as the percentage of weak modal words (e.g.,
“maybe,” “sometimes,” “suggest”).
Panels A, B and C of Table 5 present the results for negative tone, uncertain tone and weak modal tone, respectively. To conserve
space and highlight the main results from the three panels, we do not report the control variables or constant terms for the regressions.
Columns 1 and 2 of Panel A present our findings where we split our sample based on the median level of negative tone disclosure (i.e.,
lower versus higher negative tone). Our results suggest that the positive effect of filing complexity on cost of equity capital, docu
mented in Table 4, increases with how negative the disclosure tone is. Our findings continue to hold when we split our sample based on
the lowest vs. highest tercile (Columns 3 and 4) or lowest vs. highest quintile (Columns 5 and 6). Thus, our results from Table 5 do not
support Johnstone’s (2016) theoretical prediction. It is noteworthy, however, that our examination of Johnstone’s proposition de
pends on the assumption that negative disclosure tone is a good proxy for bad information. We find similar results in Panels B and C. In
fact, the adverse effect of filing complexity on cost of equity capital, increases with how uncertain or weak modal the disclosure tone is.
Taken together, our findings from Table 5 lend support to our hypothesis H2.
6. Additional analysis
We carry out four additional analyses. First, we investigate the moderating role of elite information processors (i.e., institutional
investors and financial analysts) on the relation between annual report readability and cost of equity capital. Second, we examine the
role of imperfect market competition. Third, we consider the effect of financial statement comparability on the relationship between
annual report readability and cost of equity capital. Finally, we examine the effect of the SEC’s plain English rule of 1998.
6.1. Role of institutional investors and financial analysts
This subsection examines the moderating role of institutional investors and analyst coverage on the relation between annual report
readability and cost of equity capital. Presumably, 10-K readability mainly concerns lay investors’ information processing costs and
would have thus a weaker impact on experienced users’ cost of information acquisition. By means of their role as financial in
termediaries, institutional investors and security analysts can attenuate information asymmetry arising from poor 10-K readability
(Lehavy et al. (2011); Lee (2012)). Therefore, we posit that the adverse effect of lack of readability on cost of equity financing is likely
to be lessened among firms with greater institutional ownership or higher analyst coverage. To test this prediction, we first rank and
assign firms into three terciles of low, medium, and high institutional ownership (analyst coverage). The comparisons are then made
between low (bottom tercile) and high (top tercile) subgroups. Columns 1 and 2 of Table 6 show that the adverse effect of the lack of
readability on cost of equity capital decreases with the level of institutional ownership (IO), suggesting that sophisticated investors are
less affected by difficult-to-read filings. To gain a deeper insight on the role of elite information processors, we have also considered
institutional investors’ investment horizon, and find similar results with greater ownership by long-term (LONG_IO) as well as short-
term (SHORT_IO) institutional investors (Columns 3–6 of Table 6).21 Similarly, the last two columns of Table 6 show that the positive
relation between lack of readability and cost of equity capital is less pronounced for firms with greater analyst coverage. This finding is
consistent with Lehavy et al. (2011) and Lee (2012) who document that the informativeness of financial analysts’ reports is greater for
20
Loughran and McDonald (2011) find that roughly three-fourths of negative terms in 10-K filings using the Harvard dictionary are not completely
negative in a financial context.
21
Following Gaspar et al. (2005) and Chen et al. (2007), we classify institutional investors as long-term and short-term investors using the churn
rate and portfolio turnover over the past four quarters.
11
Table 5
Financial Readability, Disclosure Tone, and Implied Cost of Equity Capital
Panel A: Negative Tone
Below Median Above Median Lowest Tercile Highest Tercile Lowest Quintile Highest Quintile
(1) (2) (3) (4) (5) (6)
BOG 0.032*** 0.059*** 0.022*** 0.086** 0.025*** 0.095***

0.01 0.01 0.01 0.02 0.01 0.02
Chi-square = 4.21 Chi-square = 11.55 Chi-square = 11.83
p-value = 0.04 p-value = 0.001 p-value = 0.006
Panel B: Uncertain Tone

BOG 0.055*** 0.063*** 0.054*** 0.073*** 0.047*** 0.089***
0.01 0.01 0.01 0.01 0.01 0.02
Panel C: Weak Modal Tone

BOG 0.040*** 0.060*** 0.037*** 0.074*** 0.035*** 0.091***
0.01 0.01 0.01 0.01 0.01 0.02
This table shows the effect of annual report readability on the cost of equity capital depending on the level of negative (Panel A), uncertain (Panel B),
and weak modal (Panel C) disclosure tone. The sample contains 39,181 firm-year observations from 1995 to 2017. The dependent variable rAVG is the
average of rCT, rGLS, rMPEG, and rOJN. BOG denotes the variable used to measure readability. Negative, uncertain, and weak modal tones are measured
as the proportion of words that are classified as negative, uncertain, and weak modal using the Loughran and McDonald (2011) word dictionaries,
respectively. In all the regressions, we include the control variables BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP. BETA is market beta obtained from
regressions of firms’ monthly excess stock returns on the corresponding CRSP value-weighted index excess returns, using at least 24 (and up to 60)
months and ending in June of each year. BTM is the ratio of the book value to the market value of equity. SIZE is the natural logarithm of total assets in
$ millions. LEV is the leverage ratio defined as the ratio of long-term debt to total assets. LTG is the average long-term growth in forecasted earnings.
FBIAS is the signed forecast error, which is defined as the difference between the one-year-ahead consensus earnings forecast and realized earnings
deflated by end of June stock price. DISP is the dispersion of analyst forecasts defined as the coefficient of variation of one-year-ahead analyst
forecasts of earnings per share in June in year t. We also control for industry fixed effects (based on the Fama and French (1997) 48-indsutry
classification) and year fixed effects. Robust standard errors adjusted for clustering by firm are reported in parentheses. The superscript asterisks
***, **, and * denote two–tailed statistical significance at the 1%, 5%, and 10% levels, respectively.
firms with less readable filings.
6.2. Role of imperfect market competition
In a theoretical work, Lambert et al. (2012) show that the effect of information asymmetry on cost of equity capital is more
pronounced if firms’ shares trade in markets that are characterized by imperfect competition. Consistent with Lambert et al.’s theo
retical prediction, Armstrong et al. (2011) find that information asymmetry influences cost of equity capital only when markets are
imperfect. Following Armstrong et al. (2011), we construct a market competition proxy that takes into account the number of
shareholders as well as the level of trading activity in firms’ shares.22 In particular, we start by dividing our sample into terciles based
on the number of shareholders of record as of the fiscal year end. Subsequently, we consider the upper and the lower terciles (which we
label A and B, respectively) and divide each one of them into terciles based on the level of trading activity in firms’ shares, computed as
annual volume in year t divided by average shares outstanding. We then consider the upper (lower) tercile within A (B) as the perfect
(imperfect) competition tercile. We re-estimate our main regression for each of the two subsamples and present the results in Table 7.
Our analyses show that the BOG coefficient continues to load positively on cost of equity capital when we use the imperfect
competition tercile (Column 1). However, it becomes statistically insignificantly different from zero when we use the perfect
competition tercile (Column 2). In the same vein, we find that the coefficient on dispersion of analyst forecasts (DISP), which is another
measure of information asymmetry, affects cost of capital only in imperfect market competition tercile. Thus, consistent with Arm
strong et al. (2011) and Lambert et al. (2012), our results suggest that the effect of information environment on cost of equity capital
depends on the degree of market competition.
6.3. Role of financial statement comparability
The conceptual framework of the FASB considers comparability a necessary qualitative attribute of financial information that
enhances its usefulness by enabling users to identify similarities and differences between two sets of economic events (FASB (2010)).
Recent empirical studies provide evidence that comparability, defined as the closeness between two firms’ accounting systems in
22
As indicated in Armstrong et al. (2011), the number of shareholders can be a noisy measure of the degree of market competition.
12
Table 6
The role of institutional investors and analyst coverage.
IO LONG_IO SHORT_IO ANALYST
Variable Low High Low High Low High Low High
(1) (2) (3) (4) (5) (6) (7) (8)
BOG 0.088*** 0.016* 0.097*** 0.071*** 0.038*** 0.020** 0.097*** 0.038***

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
BETA 2.002*** 1.364*** 1.932*** 1.639*** 1.649*** 1.213*** 2.187*** 1.253***
0.11 0.08 0.11 0.11 0.08 0.07 0.12 0.07
BTM 4.161*** 3.985*** 4.421*** 3.765*** 4.140*** 3.791*** 4.392*** 2.776***
0.19 0.14 0.2 0.18 0.15 0.13 0.19 0.15
LEV 5.439*** 4.252*** 6.375*** 4.788*** 4.584*** 3.945*** 6.235*** 5.445***
0.52 0.31 0.54 0.47 0.35 0.29 0.54 0.32
SIZE − 0.777*** − 0.239*** − 1.056*** − 0.673*** − 0.413*** − 0.208*** − 1.227*** − 0.041
0.05 0.04 0.06 0.04 0.04 0.03 0.09 0.03
LTG 0.101*** 0.076*** 0.087*** 0.138*** 0.083*** 0.078*** 0.135*** 0.032***
0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
FBIAS 0.506*** 0.615*** 0.518*** 0.556*** 0.561*** 0.639*** 0.553*** 0.328***
0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.04
DISP 2.750*** 4.814*** 2.619*** 2.632*** 4.503*** 4.801*** 3.007*** 4.688***
0.31 0.23 0.32 0.29 0.24 0.22 0.31 0.24
INTERCEPT 1.855 6.404*** 2.513 2.440* 5.456*** 5.067*** 4.771*** 2.510***
1.5 1.05 1.68 1.32 1.23 0.97 1.61 0.89
Industry effects Yes Yes Yes Yes Yes Yes Yes Yes
Year effects Yes Yes Yes Yes Yes Yes Yes Yes
Cluster by firm Yes Yes Yes Yes Yes Yes Yes Yes
N 16,107 11,537 16,107 11,537 16,107 11,537 11,437 12,864
Adj. R2 0.212 0.238 0.216 0.221 0.197 0.24 0.225 0.198
Chi-square = 21.15 Chi-square = 19.89 Chi-square = 9.88 Chi-square = 12.63
p-value = 0.000 p-value = 0.000 p-value = 0.001 p-value = 0.000
This table examines the moderating role of institutional investors and analyst coverage on the relationship between annual report readability and the
cost of equity capital. The sample contains 39,181 firm-year observations from 1995 to 2017. The dependent variable rAVG is the average of rCT, rGLS,
rMPEG, and rOJN. BOG denotes the variable used to measure readability. IO is the proportion of firm shares owned by institutional investors. LONG_IO
(SHORT_IO) is defined as the proportion of long-term (short-term) institutional ownership. ANALYST is the natural logarithm of the number of
financial analysts providing a one-year-ahead forecast of earnings per share. Control variables include BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP.
BETA is market beta obtained from regressions of firms’ monthly excess stock returns on the corresponding CRSP value-weighted index excess returns,
using at least 24 (and up to 60) months and ending in June of each year. BTM is the ratio of the book value to the market value of equity. SIZE is the
natural logarithm of total assets in $ millions. LEV is the leverage ratio defined as the ratio of long-term debt to total assets. LTG is the average long-
term growth in forecasted earnings. FBIAS is the signed forecast error, which is defined as the difference between the one-year-ahead consensus
earnings forecast and realized earnings deflated by end of June stock price. DISP is the dispersion of analyst forecasts defined as the coefficient of
variation of one-year-ahead analyst forecasts of earnings per share in June in year t. Industry (based on the Fama and French (1997) 48-indsutry
classification) and year dummies are also included in the regressions. Robust standard errors adjusted for clustering by firm are reported in pa
rentheses. The superscript asterisks ***, **, and * denote two–tailed statistical significance at the 1%, 5%, and 10% levels, respectively
mapping economic events in financial statements, reduces user information gathering and processing costs, and improves the quality of
financial information (e.g., De Franco et al. (2011); Kim et al. (2013); Kim et al. (2016); Chen et al. (2018), among others). For
example, De Franco et al. (2011) contend that analysts are more likely to choose benchmark peers with higher comparability levels,
since doing so enables them to draw sharper inferences about similarities and differences across comparable firms and to gain a better
understanding of how economic events manifest in financial performance. In addition, these authors argue that because comparable
firms constitute good benchmarks for each other, information transfer among them may reduce the effort that users need to exert to
understand and analyze their financial statements. However, in environments where a firm lacks a sufficient set of comparable peers,
lower information quality is expected to hinder investor ability to evaluate firm performance and draw accurate forecasts about future
prospects. In our context, we argue that the effect of less readable annual reports is more pronounced for firms with lower financial
statement comparability.
To investigate the moderating effect of financial statement comparability on the relation between readability and cost of equity, we
follow the methodology outlined in De Franco et al. (2011) and estimate comparability scores (COMPACCTijt) for all firm i-j pairs in the
same two-digit SIC code industry.23 We then compute firm-specific comparability score (COMPACCTit) as the mean of COMPACCTijt in
each year.24 We finally rerun our regressions of Table 4 (Column 1) after interacting BOG with the financial statement comparability
measure. The results, reported in Table 8, reveal that the coefficients of BOG are positive and highly significant. Moreover, the
23
We thank Rodrigo Verdi for kindly sharing his SAS code to compute the financial statement comparability measures (https://mitmgmtfaculty.
mit.edu/rverdi/).
24
We also use COMPACCT4it, defined as the mean COMPACCTijt of the four firms with the highest comparability to firm i during period t, as an
alternative comparability measure. The results remain qualitatively similar to those using COMPACCTit (Columns 3 and 4 of Table 8).
13
Table 7
The Role of Imperfect Stock Market Competition.
Variable Level of Stock Market Competition
Imperfect Competition Tercile Perfect Competition Tercile
(1) (2)
BOG 0.100*** 0.039

0.03 0.03
BETA 1.373*** 1.059***
0.18 0.37
BTM 4.874*** 4.064***
0.40 0.43
LEV 7.821*** 6.664***
0.99 1.21
SIZE − 0.809*** − 1.090***
0.12 0.19
LTG 0.114*** 0.132***
0.02 0.03
FBIAS 0.495*** 0.416***
0.06 0.09
DISP 4.114*** 0.499
0.53 0.89
INTERCEPT 3.162** 7.836**
1.54 3.13
Industry effects Yes Yes
Year effects Yes Yes
Cluster by firm Yes Yes
N 3893 1370
Adj. R2 0.188 0.221
Chi-square = 6.08
p-value = 0.0137
This table investigates the effect of annual report readability on the cost of equity capital depending on the degree of stock
market competition. The dependent variable rAVG is the average of rCT, rGLS, rMPEG, and rOJN. The main independent variable is
the Bog readability index. Control variables include BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP. BETA is market beta
obtained from regressions of firms’ monthly excess stock returns on the corresponding CRSP value-weighted index excess
returns, using at least 24 (and up to 60) months and ending in June of each year. BTM is the ratio of the book value to the
market value of equity. SIZE is the natural logarithm of total assets in $ millions. LEV is the leverage ratio defined as the ratio
of long-term debt to total assets. LTG is the average long-term growth in forecasted earnings. FBIAS is the signed forecast
error, which is defined as the difference between the one-year-ahead consensus earnings forecast and realized earnings
deflated by end of June stock price. DISP is the dispersion of analyst forecasts defined as the coefficient of variation of one-
year-ahead analyst forecasts of earnings per share in June in year t. Industry (based on the Fama and French (1997) 48 −
indsutry classification) and year dummies are also included in the regressions. Robust standard errors adjusted for clustering
by firm are reported in parentheses. The superscript asterisks ***, **, and * denote two tailed statistical significance at the
1%, 5%, and 10% levels, respectively.
coefficients of the interaction variables load negatively and significantly, suggesting that the effect of textual complexity on cost of
equity capital is less pronounced for firms with highly comparable financial statements. This finding is consistent with De Franco et al.
(2011) who provide evidence that financial statement comparability reduces cost of information acquisition.
6.4. The effect of SEC’s, 1998 plain English rule
The SEC has set in motion a continuous effort to streamline and simplify firms’ reporting to promote better information disclosure
for the benefit of general investors. In this regard, the SEC published in 1998 their Plain English Handbook that provides a toolkit for
creating clearer, more concise, and more informative disclosure documents (SEC (1998)). If the adoption of SEC’s plain English rule
improved the readability of firms’ 10-K filings, we expect the effect of textual complexity on cost of equity capital to be less pronounced
during the post-1998 period. Consistent with this prediction, Columns 1 and 2 of Table 9 show that the coefficient of BOG is signif
icantly lower in the post-1998 years compared to the pre-1998 period.25 In Columns 3–6 we further split our post-1998 subsample
years into four subperiods to examine how the relation between annual report readability and cost of equity capital has evolved over
time. Overall, our results indicate that the effect of lack of readability on cost of equity financing is decreasing over time. This result
provides empirical support for the SEC’s view on the importance of firms providing more readable disclosure to their investors.
25
Chi-square test indicates that the difference in the coefficients between the pre- and post-plain English rule adoption subsamples is significant at
the 1% level.
14
Table 8
The impact of financial statement comparability
COMPACCT COMPACCT4
Variable (1) (2) (3) (4)
BOG 0.051*** 0.035*** 0.051*** 0.039***

0.01 0.01 0.01 0.01
COMPACCT − 0.791*** 1.494
0.1 1.18
BOG * COMPACCT − 0.027**
0.01
COMPACCT4 − 0.928*** 1.735
0.14 1.61
BOG * COMPACCT4 − 0.041**
0.02
BETA 1.090*** 1.072*** 1.127*** 1.111***
0.12 0.12 0.12 0.12
BTM 3.674*** 3.707*** 3.692*** 3.722***
0.26 0.26 0.26 0.26
LEV 3.869*** 3.882*** 3.934*** 3.941***
0.48 0.48 0.49 0.49
SIZE − 0.480*** − 0.480*** − 0.484*** − 0.484***
0.05 0.05 0.05 0.05
LTG 0.074*** 0.074*** 0.075*** 0.075***
0.01 0.01 0.01 0.01
FBIAS 0.562*** 0.562*** 0.563*** 0.563***
0.06 0.06 0.06 0.06
DISP 3.797*** 3.802*** 3.812*** 3.816***
0.32 0.32 0.32 0.32
INTERCEPT 4.891*** 6.099*** 4.921*** 5.891***
0.93 0.99 0.94 0.98
Industry effects Yes Yes Yes Yes
Year effects Yes Yes Yes Yes
Cluster by firm Yes Yes Yes Yes
N 24,433 24,433 24,433 24,433
Adj. R2 0.223 0.224 0.222 0.222
This table summarizes the results of the effect of financial statement comparability on the readability-cost of equity capital link. To measure financial
statement comparability, we first compute all pair-wise comparability scores, denoted as COMPACCTijt, for firms in the same 2 − digit SIC code
industry following the methodology in De Franco et al. (2011). We then compute firm-wise comparability scores, which we denote COMPACCTit, as
the average of COMPACCTijt in each year, where high values of COMPACCTi denote more comparable financial statements. COMPACCT4 is the mean
COMPACCTijt of the four firms with the highest comparability to firm i during period t. The dependent variable rAVG is the average of rCT, rGLS, rMPEG,
and rOJN. The main independent variable is the Bog readability index. Control variables include BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP. BETA is
of one-year-ahead analyst forecasts of earnings per share in June in year t. Industry (based on the Fama and French (1997) 48 − indsutry classifi
cation) and year dummies are also included in the regressions. Robust standard errors adjusted for clustering by firm are reported in parentheses. The
superscript asterisks ***, **, and * denote two tailed statistical significance at the 1%, 5%, and 10% levels, respectively.
7. Robustness checks
This section performs multiple sensitivity tests to check the robustness of our results. These tests consist in correcting for mea
surement errors in analyst forecasts, using alternative proxies of main dependent and independent variables and alternative estimation
techniques, including various additional control variables, and addressing endogeneity issues, among other measures.
7.1. Noise in analyst forecasts
A potential limitation of our cost of equity estimates is measurement errors arising from bias in analyst forecasts. Prior research
argues that analyst forecasts tend to be pervasively optimistic for growth stocks compared to value stocks, which introduces a sys
tematic upward bias in cost of equity estimates (e.g., Easton and Sommers (2007); Guay et al., 2011). We address this concern in two
different ways. First, we recursively exclude the top 5, 10, 25, and 50% of firm–year observations in the FBIAS distribution. The results
reported in Columns 1–4 of Table 10 show a positive loading on BOG at the 1% threshold level. Our second approach to address
measurement errors in analyst forecasts consists of successively removing the top 5, 10, 25, and 50% of firm–year observations with
extremely high long-term growth forecasts. Columns 5–8 of Table 10 corroborate our earlier findings.
A second source of measurement error is sluggishness in analysts’ forecasts, that is, the tendency of analysts’ short- and long-term
15
Table 9
The effect of SEC’s, 1998 plain english rule
1995–1998 Post-1998 Post-1998
Variable 1999–2003 2004–2007 2008–2011 2012–2017
(1) (2) (3) (4) (5) (6)
BOG 0.107*** 0.041*** 0.059*** 0.042*** 0.021** 0.023**

0.01 0.01 0.01 0.01 0.01 0.01
BETA 0.723*** 1.764*** 1.676*** 2.390*** 1.225*** 2.097***
0.16 0.06 0.12 0.12 0.1 0.15
BTM 3.648*** 3.994*** 3.752*** 4.210*** 4.610*** 3.580***
0.26 0.1 0.18 0.22 0.25 0.20
LEV 6.229*** 4.565*** 3.451*** 6.448*** 5.313*** 3.752***
0.59 0.25 0.44 0.59 0.45 0.54
SIZE − 0.801*** − 0.544*** − 0.412*** − 0.775*** − 0.467*** − 0.617***
0.06 0.03 0.05 0.06 0.05 0.06
LTG 0.044*** 0.098*** 0.077*** 0.107*** 0.112*** 0.079***
0.01 0.01 0.01 0.01 0.01 0.01
FBIAS 0.618*** 0.568*** 0.507*** 0.593*** 0.614*** 0.508***
0.04 0.02 0.04 0.05 0.03 0.05
DISP 5.098*** 3.621*** 4.193*** 2.270*** 4.153*** 3.328***
0.35 0.17 0.31 0.42 0.31 0.34
INTERCEPT 4.029*** 5.946*** 1.706 9.110*** 9.480*** 6.248***
1.45 0.72 1.38 1.55 1.36 1.56
Industry effects Yes Yes Yes Yes Yes Yes
Year effects Yes Yes Yes Yes Yes Yes
Cluster by firm Yes Yes Yes Yes Yes Yes
N 5160 33,165 7353 7970 8439 10,259
Adj. R2 0.258 0.204 0.175 0.235 0.241 0.198
Chi-square = 21.69
p-value = 0.0000
This table examines the effect of the implementation of SEC’s (1998) plain English rule on the relationship between annual report readability and the
cost of equity capital. The sample contains 39,181 firm-year observations from 1995 to 2017. The dependent variable rAVG is the average of rCT, rGLS,
rMPEG, and rOJN. BOG denotes the variable used to measure readability. Control variables include BETA, BTM, LEV, SIZE, LTG, FBIAS, and DISP. BETA is
of one-year-ahead analyst forecasts of earnings per share in June in year t. Industry (based on the Fama and French (1997) 48-indsutry classification)
and year dummies are also included in the regressions. Robust standard errors adjusted for clustering by firm are reported in parentheses. The su
perscript asterisks ***, **, and * denote two–tailed statistical significance at the 1%, 5%, and 10% levels, respectively.
forecasts to incorporate new information about earnings more slowly than stock returns (Guay et al. (2011)). We address this concern
in three ways. First, we follow Hail and Leuz (2006) and use the accuracy-weighted average cost of equity estimates as dependent
variables in our regressions, where analyst forecast accuracy is measured by the absolute one-year-ahead forecast error scaled by assets
per share. This approach gives more weight to observations with higher forecast accuracy and assigns less weight to estimates that are
likely to be noisy. Second, we use end-of-January instead of end-of-June stock prices, thereby allowing analysts extra time to resolve
the sluggishness in their forecasts with respect to information that is embedded in January stock prices. Finally, we follow Chen et al.
(2009) and Guay et al. (2011) and control for price momentum measured as compounded stock returns over the previous 6 (Column
11) and 12 months (Column 12). In all the specifications reported in Table 10, we continue to find a positive and statistically significant
coefficient on BOG, lending further support to our conjecture that firms with difficult-to-read annual filings face higher equity
financing costs.
7.2. Alternative cost of equity estimates
To check whether our findings are robust to the proxy of cost of equity capital, we re-estimate our baseline model (Table 4, Column
1) using three additional cost of equity measures. The results are reported in online Appendix A. The cost of equity capital in Column 1
is estimated using the Easton (2004) price-earnings-growth model (rPEG), which assumes zero growth in abnormal earnings beyond the
forecast horizon and no dividend payments. Column 2 applies the forward earnings yield (rFEYD), defined as the one-year-ahead
forecast of earnings per share scaled by stock price minus the rate on a 10-year treasury note. Column 3 uses the trailing earnings-
to-price ratio, denoted as rTEYD, defined as current earnings per share scaled by stock price minus the rate on a 10-year treasury
note. In all three specifications, BOG has a positive and statistically significant loading on cost of equity measure.
Despite the widespread popularity of implied cost of equity models in accounting and finance research, ex-post realized returns may
present some advantages over ex-ante returns. Wang (2015) argues that researchers are often confronted with a tradeoff between bias
16
H. Rjiba et al.
Table 10
Robustness to Noise in Analyst Forecasts
Panel A. Forecast Optimism Bias Panel B. Analyst Forecasts Sluggishness
Variable FBIAS less than jth percentile LTG less than jth percentile Accuracy Weighted Regression January RET6 RET12
Stock Price
j ¼ 95% j ¼ 90% j ¼ 75% j ¼ 50% j ¼ 95% j ¼ 90% j ¼ 75% j ¼ 50%
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
BOG 0.050*** 0.044*** 0.038*** 0.037*** 0.053*** 0.048*** 0.043*** 0.036*** 0.072*** 0.046*** 0.054*** 0.055***
0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
BETA 1.582*** 1.519*** 1.453*** 1.536*** 1.633*** 1.631*** 1.629*** 1.822*** 1.998*** 1.532*** 1.718*** 1.601***
0.11 0.11 0.11 0.13 0.11 0.12 0.13 0.17 0.06 0.11 0.11 0.11
BTM 3.901*** 3.781*** 3.311*** 3.314*** 3.853*** 3.765*** 3.591*** 3.415*** 3.761*** 4.263*** 4.446*** 4.130***
0.22 0.22 0.21 0.25 0.23 0.23 0.24 0.3 0.11 0.24 0.23 0.22
LEV 4.876*** 4.751*** 4.335*** 4.436*** 4.686*** 4.437*** 4.097*** 3.675*** 5.782*** 4.712*** 5.344*** 5.216***
0.42 0.4 0.4 0.47 0.43 0.43 0.45 0.53 0.26 0.42 0.42 0.42
SIZE − 0.506*** − 0.405*** − 0.272*** − 0.279*** − 0.576*** − 0.533*** − 0.443*** − 0.186*** − 0.762*** − 0.583*** − 0.629*** − 0.634***
17
0.04 0.04 0.04 0.05 0.05 0.05 0.04 0.05 0.03 0.04 0.05 0.05
LTG 0.083*** 0.071*** 0.056*** 0.044*** 0.068*** 0.039*** 0.034* − 0.054** 0.036*** 0.076*** 0.101*** 0.123***
0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.03 0.01 0.01 0.01 0.01
FBIAS 0.028 − 0.468*** − 1.061*** − 1.302*** 0.570*** 0.613*** 0.629*** 0.607*** − 0.497*** 0.378*** 0.450*** 0.416***
0.07 0.08 0.11 0.12 0.05 0.05 0.07 0.09 0.02 0.04 0.05 0.05
DISP 4.233*** 4.370*** 4.276*** 3.830*** 4.294*** 4.418*** 4.811*** 5.178*** 3.409*** 3.116*** 3.427*** 3.073***
0.3 0.32 0.39 0.47 0.3 0.31 0.35 0.43 0.17 0.26 0.28 0.28
RET − 4.997*** − 4.367***
0.2 0.13
INTERCEPT 5.394*** 5.360*** 5.077*** 3.993*** 5.631*** 6.046*** 6.109*** 5.227*** 5.750*** 5.943*** 5.826*** 5.845***
0.87 0.82 0.83 1.03 0.87 0.86 0.87 0.91 0.75 0.84 0.87 0.86
Industry effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Journal of Corporate Finance 67 (2021) 101902

Year effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Cluster by firm Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
N 37,883 36,212 30,582 19,986 38,081 36,543 31,650 21,769 39,181 39,264 39,172 39,092
Adj. R2 0.188 0.19 0.195 0.201 0.205 0.205 0.209 0.217 0.201 0.182 0.249 0.269
and efficiency. While ex-ante models are more precise, they are biased because of measurement errors. On the other hand, realized
returns yield consistent inferences in large samples but are noisy and imprecise. To obtain reliable inferences, Wang (2015) recom
mends complementing ex-ante cost of equity proxies with measures based on realized equity returns. Following Hann et al., 2013, we
estimate realized equity returns as buy-and-hold returns accumulated over 12 months starting in June of year t + 1 minus the rate on a
10-year treasury note, which we label as REARET. To mitigate concern about the noisy nature of realized returns, we consider an
alternative measure that combines information from the ex-post and ex-ante approaches. Specifically, we regress ex-post returns on six
ex-ante cost of equity measures and then use the fitted value from the regression as our proxy of expected returns. The six ex-ante cost
of equity capital measures included in the first-stage regression are: 1) implied cost of equity premium proposed by Claus and Thomas
(2001); 2) implied cost of equity premium derived from Gebhardt et al. (2001); 3) implied cost of equity premium designed by Easton
(2004); 4) expected returns from the Fama–French three-factor model; 5) earnings yield; and 6) earnings yield adjusted for growth.
This measure of “instrumented” returns, which we denote INSRET, is likely to filter out the information shocks in realized returns and
hence reduce their noisiness (Hann et al. (2013)). Columns 4 and 5 report the regression results using REARET and INSRET, respec
tively. Our core findings that firms with hard-to-read filings face a higher cost of equity capital remain qualitatively unchanged.
7.3. The role of earnings quality
In an early paper, Francis et al. (2008) document a negative association between the level of voluntary disclosure and cost of equity
capital. The authors show that when the quality of a firm’s earnings is taken into account, the effect of disclosure on cost of equity is
substantially reduced or even disappears completely. To test for this possibility in our context, we re-estimate our main regression
(Table 4, Column 1) after controlling for earnings quality. Following Francis et al. (2008) and He et al. (2019), we employ four earnings
quality proxies namely, the accruals quality measure of Dechow and Dichev (2002) as modified by McNichols (2002), the standard
deviation of the firm’s earnings over a ten-year rolling window, the absolute value of the abnormal accruals obtained from the
modified Jones (1991) model, and the principal component of the three previous measures.26 Results are reported in Table 11.
Consistent with prior evidence, the coefficients on all four proxies of earnings quality are significantly positive, suggesting that lower
earnings quality increases a firm’s cost of equity financing. More importantly, the effect of annual report readability on the implied cost
of equity continues to hold even after controlling for earnings quality derived from financial statements, suggesting that both read
ability and earnings quality capture different dimensions of the firm’s information environment.
7.4. Additional control variables
Although regressions reported in Table 4 include a large set of control variables motivated by prior studies, the concern that our
results may suffer from omitted variable bias remains valid. To mitigate this issue, we sequentially include various additional control
variables that might affect cost of equity capital. The results are reported in online Appendix B. Specifically, we control for firm
proprietary costs (FLUIDITY) in Column 1. We control for institutional ownership (IO) in Column 2. In Column 3, we account for the
number of financial analysts providing a one-year-ahead forecast of earnings per share (ANALYST). Column 4 employs a dummy
variable that indicates whether a firm hires a Big N auditor (BIG_N). In addition, we control for the ratio of stock-based compensation to
total compensation (STOCKCOMP) in Column 5. Columns 6 and 7 account for business complexity using the number of business
segments (NBUSSEG) and the number of geographic segments (NGEOSEG), respectively. In column 8, we use a dummy variable set to 1
if the firm reports negative earnings (LOSS), and zero otherwise. The last column includes all the aforementioned additional controls.
In all the specifications of online Appendix B, we continue to find a positive and statistically significant coefficient on BOG, indicating
that omitted variable bias is not a serious concern.
7.5. Endogeneity
The endogenous nature of corporate disclosure policy is well recognized in empirical accounting research (e.g., Nikolaev and Van
Lent (2005); Larcker and Rusticus (2010)). Thus far, we have tackled this concern in Table 4 by running a fixed-effects regression, and
in online Appendix B by including a large set of additional control variables. In this subsection, we conduct several additional tests to
further address this issue.
Our first approach adopts a propensity score matching (PSM) methodology to mitigate the concern that the observable firm
characteristics associated with annual reports’ readability level cause differences in the relationship between readability and cost of
equity capital. To implement the PSM methodology, we follow the procedure outlined in Shipman et al. (2017). First, we create a
dummy variable, BOG_DUMMY, that equals one if BOG is higher than the sample median; and zero otherwise, and then classify ob
servations into treatment (BOG_DUMMY = 1) and control (BOG_Dummy = 0) groups. Next, we estimate propensity scores using a
logistic model that regresses the BOG_DUMMY on the set of control variables used in Table 4. We then perform a one-to-one match
26
These variables are denoted EQ_McNichols, EQ_abs_ModifiedJones, EQ_SDEarnings, and EQ_pc, respectively, where higher values indicate poorer
earnings quality.
18
Table 11
Controlling for Earnings Quality
Variable EQ_McNichols EQ_SDEarnings EQ_abs_ModifiedJones EQ_pc
(1) (2) (3) (4)
BOG 0.050*** 0.055*** 0.052*** 0.052***

0.01 0.01 0.01 0.01
EQ_Measure 3.199*** 1.649*** 1.263*** 1.428***
0.48 0.61 0.29 0.21
BETA 1.655*** 1.569*** 1.663*** 1.681***
0.12 0.11 0.12 0.12
BTM 3.807*** 3.958*** 3.899*** 3.901***
0.24 0.23 0.24 0.24
LEV 5.615*** 4.870*** 5.671*** 5.721***
0.46 0.42 0.46 0.46
SIZE − 0.656*** − 0.560*** − 0.680*** − 0.672***
0.05 0.05 0.05 0.05
LTG 0.079*** 0.082*** 0.080*** 0.080***
0.01 0.01 0.01 0.01
FBIAS 0.546*** 0.558*** 0.552*** 0.563***
0.05 0.05 0.05 0.05
DISP 3.739*** 3.864*** 3.777*** 3.775***
0.3 0.28 0.3 0.3
INTERCEPT 6.086*** 4.658*** 5.323*** 6.141***
0.93 0.89 0.94 0.94
Industry effects Yes Yes Yes Yes
Year effects Yes Yes Yes Yes
Cluster by firm Yes Yes Yes Yes
N 33,382 37,974 33,855 33,382
Adj. R2 0.215 0.214 0.211 0.211
This table reports the results from regressions that control for the quality of reported earnings. Measures of earnings quality are EQ_McNichols,
EQ_SDEarnings, EQ_abs_ModifiedJones, and EQ_pc. EQ_McNichols is based on McNichols’ (2002) modification of Dechow and Dichev’s (2002) accruals
model. EQ_SDEarnings is computed as the standard deviation of the firm’s earnings over the prior ten-year period. EQ_abs_ModifiedJones is the absolute
value of abnormal accruals obtained using the modified Jones (1991) model. EQ_pc is the principal component of EQ_McNichols, EQ_SDEarnings, and
EQ_abs_ModifiedJones. The dependent variable rAVG is the average of rCT, rGLS, rMPEG, and rOJN. The Appendix contains definitions of the regression
variables. Industry dummies are based on the Fama and French (1997) 48 − indsutry classification. Robust standard errors adjusted for clustering by
firm are reported in parentheses. The superscript asterisks ***, **, and * denote two–tailed statistical significance at the 1%, 5%, and 10% levels,
respectively.
without replacement of observations on propensity scores and retain matches that are within a caliper distance of 0.0001.27 Finally, we
estimate average treatment effect on the matched sample after including all control variables of Table 4 as well as year and industry
fixed effects. Results reported in Table 12 (Panel A, Column 1) document a significantly positive coefficient on BOG, suggesting that
more complex annual reports are associated with a higher cost of equity capital after matching for several observed firm differences.
Our second endogeneity test aims to assuage the concern that potential omitted variables drive our results by estimating a change
regression in Column 2 of Panel A in Table 12. Specifically, we regress the change in rAVG (ΔrAVG) against the change in BOG (ΔBOG),
controlling for the change in all control variables, and including industry and year fixed effects. The first differencing approach
considers how changes in annual report readability over time affect changes in cost of equity capital over the same time period, thus
reducing concerns about omitted time-invariant variables in our level specification of Table 4, Column 1. As shown in Column 2 (Panel
A, Table 12), the change in BOG is associated with a positive and significant change in rAVG.
Our third endogeneity check aims to rule out the potential reverse causality concern by performing a granger causality test
following Dyck et al. (2019). Columns 1 and 2 (Panel B, Table 12) present the estimation results. We find that past changes in annual
report readability positively influence the current change in the implied cost of equity capital but document no relationship between
past cost of equity capital changes and the current change in readability. These results confirm that more textually complex annual
reports lead to more costly equity financing and there appears to be no reverse causality from the cost of equity capital to annual report
readability. Collectively, the results reported in Table 12 rule out potential endogeneity concerns and provide support for our
conjecture that equity investors charge a higher risk premium for firms with more textually complex (i.e., less readable) annual filings.
7.6. Alternative financial readability measures
This section aims to assess the robustness of our results to the use of alternative readability proxies. The results are reported in
27
Admittedly, the choice of the caliper width is arbitrary (Shipman et al. (2017)). Thus, to assess the sensitivity of our results to the choice of the
PSM design, we employ several additional caliper distances commonly used in the empirical accounting literature (i.e., 0.01, 0.03, and 0.10). In
unreported results, our core finding continues to hold.
19
Table 12
Robustness to Endogeneity.
Panel A: Propensity Score Matching and Change Regression
Variable Propensity Score Matched Variable Change regression
(1) (2)
BOG 0.047*** ΔBOG 0.038***

0.01 0.01
CONTROLS Yes ΔCONTROLS Yes
INTERCEPT 5.007*** INTERCEPT − 1.068

1.21 1.5
Industry effects Yes Industry effects Yes

Year effects Yes Year effects Yes
Cluster by firm Yes Cluster by firm Yes
N 18,894 N 9835
Adj. R2 0.213 Adj. R2 0.144
Panel B: Granger Causality.
Variable rAVG Variable BOG
(1) (2)
Lag (BOG) 0.029** Lag (BOG) 0.224***

0.01 0.02
Lag (rAVG) 0.043*** Lag (rAVG) − 0.008
0.01 0.01
Lag (CONTROLS) Yes Lag (CONTROLS) Yes
INTERCEPT 9.653*** INTERCEPT 58.229***

1.75 1.64
Firm fixed effects Yes Firm fixed effects Yes
Year effects No Year effects No
Cluster by firm Yes Cluster by firm Yes
N 13,503 N 13,897
Adj. R2 0.005 Adj. R2 0.120
This table examines the robustness of our main results (Table 4, Column 1) to endogeneity concerns. We report the results of the propensity score
matching approach in Column 1 of Panel A. Column 2 of Panel A regresses changes of the cost of equity capital measure (∆rAVG) on changes of annual
report readability (∆BOG). Columns 1 and 2 of Panel B report the results of the granger causality test. The Appendix contains definitions of the
regression variables. Industry dummies are based on the Fama and French (1997) 48 − indsutry classification. Robust standard errors adjusted for
clustering by firm are reported in parentheses. The superscript asterisks ***, **, and * denote two–tailed statistical significance at the 1%, 5%, and
10% levels, respectively.
online Appendix C. Column 1 employs total document length (LENGTH), measured as the natural logarithm of total word count of the
10-K filing. Columns 2 and 3 heed the Loughran and McDonald (2014) recommendation and use simple file size as a proxy for business
document readability. More specifically, Column 2 reports results using gross file size (GROSSFSIZE) defined as the natural logarithm
of file size in megabytes of the SEC EDGAR “complete submission text file” for the 10-K filing. Column 3 uses net file size, denoted
NETFSIZE, where only text content is included. All readability variables have a significantly positive loading on rAVG, suggesting that
our conclusions are immune to the choice of annual report readability proxy.
7.7. Abnormal (unexplained) financial readability
Recent anecdotal and academic evidence points out the growing complexity of corporate business filings (e.g., Li, 2008; Monga and
Chasan (2015); Cazier and Pfeiffer (2016); Dyer et al. (2017)). Li, 2008 attributes this increased complexity to the nature of business
operations, which are likely difficult to present and hence require more complex language. More recently, Dyer et al. (2017) argue that
this trend is due primarily to increasing compliance requirements from regulators and standards setters. Accordingly, it is unclear
whether higher textual complexity (i.e., lower readability) reflects managers’ willingness to obfuscate information or is simply driven
by increased regulation and inherent business complexity. To dispel this concern, we isolate the obfuscation component of textual
complexity by regressing our readability proxies against a large set of factors that explain cross-sectional differences in business
complexity in the absence of obfuscation incentives (Li (2008); Bushee et al. (2018)). These factors are taken from Li (2008) and
Bushee et al. (2018), and include firm size (MVE), financial leverage (LEV), book-to-market (BTM), historical stock performance
(RETURNS), acquisitions (ACQUISITIONS), two dummy variables, M&A and SEO, to capture firm-year specific merger-and-acquisition
and seasoned equity offering events, respectively, capital intensity (CAPINTENSITY), capital expenditure (CAPEX), research and
development (R&D), debt and equity issuance (FINANCING), cash flow volatility (σCFO), goodwill impairment (GOODWILL),
restructuring charges (RESTRUCTURING), the logarithm of the number of business segments (NBUSSEG), and the logarithm of the
20
number of geographic segments (NGEOSEG). More specifically, we define unexplained readability (RES_READ) as the residual from the
following regression model:
READABILITY = α + β1 MVE + β2 LEV + β3 BTM + β4 RETURNS + β5 ACQUISITIONS + β6 CAPINTENSITY + β7 M&A + β8 SEO
+ β9 CAPEX + β10 R&D + β11 FINANCING + β12 σ CFO + β13 GOODWILL + β14 RESTRUCTURING + β15 NBUSSEG
+ β16 NGEOSEG + INDUSTRY FE + YEAR FE + ε (2)
where READABILITY denotes our main and alternative financial readability proxies (i.e., BOG, LENGTH, GROSSFSIZE, and NETFSIZE)
The results portrayed in online Appendix D are broadly consistent with our primary findings regarding the positive relation be
tween readability and cost of equity capital. Overall, we continue to find strong support for the conjecture that equity investors charge
a higher risk premium for firms with less readable annual filings.
8. Conclusion
Corporate narrative disclosure constitutes an important channel through which managers analyze past performance and
communicate future firm prospects to market participants. Prior literature documents that more complex financial reporting influences
investor judgment and increases disagreement among market participants (e.g., Miller (2010); Lawrence (2013); Tan et al. (2015)).
The present study examines the effect of annual report readability on cost of equity capital for a large sample of U.S. public firms over
the period 1995–2017. After controlling for many risk factors, we find that firms with hard-to-read annual reports face higher cost of
equity financing. These findings are consistent with the conjecture that less readability is likely to undermine investor ability to draw
accurate forecasts of future firm performance and increases their risk perception. Investors thus require a higher risk premium, which
translates into higher cost of equity capital.
To further validate our results, we conduct a host of robustness checks. These checks include using alternative proxies for cost of
equity capital and annual report readability and addressing potential problems arising from residual correlation at both the cross-
sectional and time-series dimension. Our conclusions also hold when we correct for potential measurement error in cost of equity
estimates. Our findings also show that the level of competition among equity investors plays a major role in shaping the relation
between information asymmetry and cost of equity capital. We also find that the level of institutional investors’ ownership and analyst
coverage play a moderating role in the relation between annual report readability and cost of equity capital.
While the main focus of this study is on the effect of annual report readability on cost of equity capital, we have also examined the
role of disclosure tone. Our results show that the adverse effect of filing complexity on cost of capital increases with the percentage of
negative, uncertain, and weak modal tone in corporate disclosure. Thus, in addition to readability, negative and ambiguous disclosure
tones result in higher cost of external financing.
Our focus on textual properties of firm disclosures supplements the well-trodden field of research on the determinants of cost of
equity capital with novel evidence and furthers our understanding of how non-quantitative disclosure influences equity market
participants. Moreover, our results provide strong support for the SEC’s view concerning the importance of firms providing clear and
concise disclosure to their stakeholders.
Appendix
Variables definition and data sources.
Variable Definition Data source
Implied cost of equity

rCT Implied cost of equity estimate derived from the Claus and Thomas (2001) model minus Authors’ calculation based on I/B/E/S and
the rate on a 10-year treasury note. Compustat data
rGLS Implied cost of equity estimate derived from the Gebhardt et al. (2001) model minus As above
the rate on a 10-year treasury note.
rMPEG Implied cost of equity estimate derived from the Easton (2004) model minus the rate on As above
a 10-year treasury note.
rOJN Implied cost of equity estimate derived from the Ohlson and Juettner-Nauroth (2005) As above
model minus the rate on a 10-year treasury note.
rAVG Average of rCT, rGLS, rMPEG, and rOJN. As above
Financial readability
BOG A proprietary measure of readability created by Editor Software’s plain English Brian Miller’s website
software, StyleWriter. The formula is based on several plain English factors such as
sentence length, passive voice, weak verbs, overused words, complex words, and
jargon. Higher values of the index imply lower readability.
(continued on next page)
21
(continued )
Controls
BETA Market beta obtained from regressions of firms’ monthly excess stock returns on the Authors’ calculation based on CRSP data
corresponding CRSP value-weighted index
excess returns, using at least 24 (and up to 60) months and ending in June of each year.
Excess returns are monthly returns minus the one-month Treasury bill rate.
BTM The ratio of the book value to the market value of equity. Authors’ calculation based on Compustat data
SIZE Natural logarithm of total assets in $ millions. As above
LEV Leverage ratio defined as the ratio of long-term debt to total assets. As above
LTG Average long-term growth in forecasted earnings. I/B/E/S
FBIAS Signed forecast error, which is defined as the difference between the one-year-ahead Authors’ calculation based on I/B/E/S data
consensus earnings forecast and realized earnings deflated by end of June stock price.
DISP Dispersion of analyst forecasts defined as the coefficient of variation of one-year-ahead As above
analyst forecasts of earnings per share in June in year t.
Alternative cost of equity estimates

rPEG Implied cost of equity estimate derived from the Easton (2004) model minus the rate on Authors’ calculation based on I/B/E/S and
a 10-year treasury note. Compustat data
rFEYD Forward earnings-to-price ratio computed as one-year-ahead forecast of earnings per As above
share scaled by stock price minus the rate on a 10-year treasury note.
rTEYD Trailing earnings yield defined as current earnings per share divided by stock price As above
minus the rate on a 10-year treasury note.
REARET Realized equity returns, computed as buy-and-hold returns accumulated over 12 Authors’ calculation based on Compustat data
months starting in June of year t + 1 minus the rate on a 10-year treasury note.
INSRET Instrumented returns, computed as the fitted value from the regression of ex-post Authors’ calculation
realized returns on a set of six ex-ante cost of equity capital measures: implied cost of
equity premium proposed by Claus and Thomas (2001), implied cost of equity premium
derived from Gebhardt et al. (2001), implied cost of equity premium designed by
Easton (2004), expected returns from the Fama-French three-factor model, earnings
yield, and earnings yield adjusted for growth.
Variables used in additional analysis

COMPACCTijt Comparability measure between pairs of firms i-j belonging to the same 2-digit SIC code Authors’ calculation using the methodology in
industry during year t. De Franco et al. (2011)
COMPACCT The mean of COMPACCTijt in each year. As above
COMPACCT4 The mean COMPACCTijt of the four firms with the highest comparability to firm i during As above
period t.
NEGATIVE % of words that are classified as negative using the Loughran and McDonald (2011) Loughran and McDonald data
word list.
UNCERTAIN % of words that are classified as uncertain using the Loughran and McDonald (2011) As above
word list.
WEAK_MODAL % of weak modal words using the Loughran and McDonald (2011) word list. As above
Alternative readability measures

LENGTH Natural logarithm of total number of words. WRDS SEC readability and sentiment database
GROSSFSIZE The natural logarithm of the file size in megabytes of the SEC EDGAR “complete Loughran and MacDonald data
submission text file” for the 10-K filing.
NETFSIZE The natural logarithm of the file size in megabytes of the SEC EDGAR “complete As above
submission text file” for the 10-K filing, where only text content is included.
RES_READ The residuals from regressing readability measures against firm size (MVE), firm Authors’ calculation
leverage (LEV), book-to-market ratio (BTM), historical stock performance (RETURNS),
acquisitions (ACQUISITIONS), capital intensity (CAPINTENSITY), firm events (M&A
and SEO), capital expenditure (CAPEX), research and development (R&D), debt and
equity issuance (FINANCING), cash flow volatility (σCFO), goodwill impairment
(GOODWILL), restructuring charges (RESTRUCTURING), number of business segments
(NBUSSEG), and the number of geographic segments (NGEOSEG).
Additional control variables

FLUIDITY The Herfindahl-Index calculated as the sum of the squared market shares using firm Hoberg and Phillips (2010) Data Library
sales, based on the text-based network industry classification of Hoberg and Phillips.
IO The proportion of firm shares owned by institutional investors. Authors’ calculation based on Thomson
Reuters Institutional (13F) Holdings database
LONG_IO The proportion of firm shares owned by long-term institutional investors. As above
SHORT_IO The proportion of firm shares owned by short-term institutional investors. As above
ANALYST The natural logarithm of the number of analysts providing one-year-ahead earnings I/B/E/S
forecasts.
BIG_N A dummy variable set to 1 if the firm hires a Big N auditor; and 0 otherwise. Authors’ calculation based on Compustat data.
STKMIX Ratio of executives’ stock-based compensation to total compensation.
(continued on next page)
22
(continued )
Authors’ calculation based on ExecuComp

data.
LOSS Indicator variable set to 1 if net income before extraordinary items is negative in the As above
current and previous year; and 0 otherwise.
Business complexity variables

MVE The logarithm of the market value of equity at the end of the fiscal year. Authors’ calculation
LEV Long term debt plus short term debt, scaled by total assets. As above
BTM Book value of equity scaled by market value of equity at the beginning of the year. As above
RETURNS The buy-and-hold return over the twelve months prior to the10-K filing date. Authors’ calculation based on CRSP stock
returns
ACQUISITIONS Acquisitions scaled by total assets. Authors’ calculation based on Compustat data.
CAPINTENSITY Net plant, property, and equipment scaled, scaled by total assets at the beginning of the As above
year.
M&A A dummy variable equal to 1 for a year in which a company appears in the SDC SDC Platinum database
Platinum M&A database as an acquirer; and 0 otherwise.
SEO A dummy variable equal to 1 for a year in which a company has a common equity As above
offering in the secondary market according to the SDC Global New Issues database and
0 otherwise.
CAPEX Amount of capital expenditures scaled by total assets at the beginning of the year. Authors’ calculation based on Compustat data.
R&D Ratio of research and development expense to sales. As above
FINANCING Amount raised from stock and debt issuances during the year scaled by total assets. As above
σCFO Standard deviation of cash flows from operations over the prior five years scaled by As above
total assets.
GOODWILL Indicator variable for whether the firm had a goodwill impairment charge that year. As above
RESTRUCTURING Indicator variable for whether the firm had a restructuring charge that year. As above
NBUSSEG The number of business segments. As above
NGEOSEG The number of geographic segments. As above
Earnings quality measures

EQ_McNichols The accruals quality measure based on McNichols’ (2002) modification of Dechow and Authors’ calculation based on Compustat data.
Dichev’s (2002) model.
EQ_SDEarnings The standard deviation of the firm’s earnings over the prior ten-year period, where As above
earnings are defined as earnings before extraordinary items scaled by total assets.
EQ_abs_ModifiedJones The absolute value of abnormal accruals obtained using the modified Jones (1991) As above
model.
EQ_pc The principal component of EQ_McNichols, EQ_SDEarnings, and EQ_abs_ModifiedJones. As above
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jcorpfin.2021.101902.
References
Aboody, D., Hughes, J., Liu, J., 2005. Earnings quality, insider trading, and cost of capital. J. Account. Res. 43, 651–673.
Aktas, N., de Bodt, E., Declerck, F., Van Oppens, H., 2007. The PIN anomaly around M&a announcements. J. Financ. Mark. 10, 169–191.
Armstrong, C.S., Core, J.E., Taylor, D.J., Verrecchia, R.E., 2011. When does information asymmetry affect the cost of capital? J. Account. Res. 49, 1–40.
Aslan, H., Easley, D., Hvidkjaer, S., O’Hara, M., 2011. The characteristics of informed trading: implications for asset pricing. J. Empir. Financ. 18, 782–801.
Athanasakou, V., Eugster, F., Schleicher, T., Walker, M., 2020. Annual report narratives and the cost of equity capital: U.K. evidence of a U-shaped relation. Eur.
Account. Rev. 29, 27–54.
Attig, N., Cleary, S., El Ghoul, S., Guedhami, O., 2013. Institutional investment horizons and the cost of equity capital. Financ. Manag. 42, 441–477.
Bai, X., Dong, Y., Hu, N., 2019. Financial report readability and stock return synchronicity. Appl. Econ. 51, 346–363.
Bekaert, G., Harvey, C.R., 2000. Foreign speculators and emerging equity markets. J. Financ. 55, 565–613.
Berger, P., 2011. Challenges and opportunities in disclosure research: a discussion of ‘the financial reporting environment: review of the recent literature’. J. Account.
Econ. 51, 204–218.
Beyer, A., Cohen, D.A., Lys, T.Z., Walther, B.R., 2010. The financial reporting environment: review of the recent literature. J. Account. Econ. 50, 296–343.
Bhattacharya, U., Daouk, H., Welker, M., 2003. The world price of earnings opacity. Account. Rev. 78, 641–678.
Biddle, G., Hilary, G., Verdi, R., 2009. How does financial reporting quality relate to investment efficiency? J. Account. Econ. 48, 112–131.
Bloomfield, R., Fischer, P.E., 2011. Disagreement and the cost of capital. J. Account. Res. 49, 41–68.
Bonsall, S.B., Miller, B.P., 2017. The impact of narrative disclosure readability on bond ratings and the cost of debt capital. Rev. Acc. Stud. 22, 608–643.
Bonsall, S.B., Leone, A.J., Miller, B.P., Rennekamp, K.M., 2017. A plain English measure of financial reporting readability. J. Account. Econ. 63, 329–357.
Botosan, C.A., 1997. Disclosure level and the cost of equity capital. Account. Rev. 72, 323–349.
Botosan, C.A., Plumlee, M.A., 2002. A re-examination of disclosure level and the expected cost of equity capital. J. Account. Res. 40, 21–40.
Botosan, C.A., Plumlee, M.A., 2005. Assessing alternative proxies for the expected risk premium. Account. Rev. 80, 21–53.
Botosan, C.A., Plumlee, M.A., Xie, Y., 2004. The role of information precision in determining the cost of equity capital. Rev. Acc. Stud. 9, 233–259.
Boubakri, N., El Ghoul, S., Saffar, W., 2014. Political rights and equity pricing. J. Corp. Finan. 27, 326–344.
23
Boubakri, N., Guedhami, O., Saffar, W., 2016. Geographic location, foreign ownership, and cost of equity capital: evidence from privatization. J. Corp. Finan. 38,
363–381.
Breuer, W., Müller, T., Rosenbach, D., Salzmann, A., 2018. Corporate social responsibility, investor protection, and cost of equity: a cross-country comparison. J. Bank.
Financ. 96, 34–55.
Bushee, B.J., Gow, I.D., Taylor, D.J., 2018. Linguistic complexity in firm disclosures: obfuscation or information? J. Account. Res. 56, 85–121.
Callen, J.L., Khan, M., Lu, H., 2013. Accounting quality, stock price delay, and future stock returns. Contemp. Account. Res. 30, 269–295.
Campbell, J.L., Chen, H., Dhaliwal, D.S., Lu, H.M., Steele, L.B., 2014. The information content of mandatory risk factor disclosures in corporate filings. Rev. Acc. Stud.
19, 396–455.
Cao, Y., Myers, L.A., Tsang, A., Yang, Y.G., 2017. Management forecasts and the cost of equity capital: international evidence. Rev. Acc. Stud. 22, 791–838.
Cazier, R.A., Pfeiffer, R.J., 2016. Why are 10-K filings So long? Account. Horiz. 30, 1–21.
Chakrabarty, B., Seetharaman, A., Swanson, Z., Wang, X., 2018. Management risk incentives and the readability of corporate disclosures. Financ. Manag. 47,
583–616.
Chava, S., Purnanandam, A., 2010. Is default risk negatively related to stock returns? Rev. Financ. Stud. 23, 2523–2559.
Chen, C.-W., Collins, D.W., Kravet, T.D., Mergenthaler, R., 2018. Financial statement comparability and the efficiency of acquisition decisions. Contemp. Account.
Res. 35, 164–202.
Chen, K.C., Chen, Z., Wei, K., 2011. Agency costs of free cash flow and the effect of shareholder rights on the implied cost of equity capital. J. Financ. Quant. Anal. 46,
171–207.
Chen, K.C.W., Chen, Z., Wei, K., 2009. Legal protection of investors, corporate governance, and the cost of equity capital. J. Corp. Finan. 15, 273–289.
Chen, X., Harford, J., Li, K., 2007. Monitoring: which institutions matter? J. Financ. Econ. 86, 279–305.
Chen, Z., Li, O.Z., Zou, H., 2016. Directors’ and officers’ liability insurance and the cost of equity. J. Account. Econ. 61, 100–120.
Cheng, C.S.A., Collins, D., Huang, H.H., 2006. Shareholder rights, financial disclosure and the cost of equity capital. Rev. Quant. Finan. Acc. 27, 175–204.
Claus, J., Thomas, J., 2001. Equity premia as low as three percent? Evidence from analysts’ earnings forecasts for domestic and international stock markets. J. Financ.
56, 1629–1666.
Cochrane, J., 2001. Asset Pricing. Princeton University Press.
Cohen, D., 2008. Does information risk really matter? An analysis of the determinants and economic consequences of financial reporting quality. Asia-Pacific J.
Accounting and Economics 15, 69–90.
Core, J.E., 2001. A review of the empirical disclosure literature: discussion. J. Account. Econ. 31, 441–456.
Core, J.E., Guay, W.R., Verdi, R., 2008. Is accruals quality a priced risk factor? J. Account. Econ. 46, 2–22.
Courtis, J., 1986. An investigation into annual report readability and corporate risk-return relationships. Account. Bus. Res. 16, 285–294.
Daouk, H., Lee, C.M., Ng, D., 2006. Capital market governance: how do security laws affect market performance? J. Corp. Finan. 12, 560–593.
De Franco, G., Kothari, S.P., Verdi, R.S., 2011. The benefits of financial statement comparability. J. Account. Res. 49, 895–931.
Dechow, P., Dichev, I., 2002. The quality of accruals and earnings: the role of accrual estimation errors. Account. Rev. 77, 35–59.
Dechow, P.M., Sloan, R.G., 1997. Returns to contrarian investment strategies: tests of naive expectations hypotheses. J. Financ. Econ. 43, 3–27.
Dempsey, S.J., Harrison, D.M., Sheng, H., 2015. Core earnings uncertainty, dividend change announcements and the reduction of covariance component risks. J. Bus.
Financ. Acc. 42, 1075–1120.
Dhaliwal, D., Li, O.Z., Tsang, A., Yang, Y.G., 2014. Corporate social responsibility disclosure and the cost of equity capital: the roles of stakeholder orientation and
financial transparency. J. Account. Public Policy 33, 328–355.
Dhaliwal, D., Judd, J.S., Serfling, M., Shaikh, S., 2016. Customer concentration risk and the cost of equity capital. J. Account. Econ. 61, 23–48.
Dhaliwal, D.A.N., Heitzman, S., Zhen Li, O., 2006. Taxes, leverage, and the cost of equity capital. J. Account. Res. 44, 691–723.
Diamond, D.W., Verrecchia, R.E., 1991. Disclosure, liquidity, and the cost of capital. J. Financ. 46, 1325–1359.
Duarte, J., Young, L., 2009. Why is PIN priced? J. Financ. Econ. 91, 119–138.
Dyck, A., Lins, K.V., Roth, L., Wagner, H., 2019. Do institutional investors drive corporate social responsibility? International evidence. J. Financ. Econ. 131, 693–714.
Dyer, T., Lang, M.H., Stice-Lawrence, L., 2017. The evolution of 10-K textual disclosure: evidence from latent Dirichlet allocation. J. Account. Econ. 64, 221–245.
Easley, D., O’Hara, M., 2004. Information and the cost of capital. J. Financ. 59, 1553–1583.
Easley, D., Hvidkjaer, S., O’Hara, M., 2002. Is information risk a determinant of asset returns? J. Financ. 57, 2185–2221.
Easley, D., Hvidkjaer, S., O’Hara, M., 2010. Factoring information into returns. J. Financ. Quant. Anal. 45, 293–309.
Easton, P., 2009. Estimating the cost of capital implied by market prices and accounting data. Found. Trends Acc. 2, 241–364.
Easton, P.D., 2004. PE ratios, PEG ratios, and estimating the implied expected rate of return on equity capital. Account. Rev. 79, 73–95.
Easton, P.D., Monahan, S.J., 2005. An evaluation of accounting-based measures of expected returns. Account. Rev. 80, 501–538.
Easton, P.D., Sommers, G.A., 2007. Effect of Analysts’ optimism on estimates of the expected rate of return implied by earnings forecasts. J. Account. Res. 45,
983–1015.
Ecker, F., Francis, J., Kim, I., Olsson, P.M., Schipper, K., 2006. A returns-based representation of earnings quality. Account. Rev. 81, 749–780.
El Ghoul, S., Guedhami, O., Kwok, C.C., Mishra, D.R., 2011. Does corporate social responsibility affect the cost of capital? J. Bank. Financ. 35, 2388–2406.
El Ghoul, S., Guedhami, O., Ni, Y., Pittman, J., Saadi, S., 2012. Does religion matter to equity pricing? J. Bus. Ethics 111, 491–518.
El Ghoul, S., Guedhami, O., Ni, Y., Pittman, J., Saadi, S., 2013. Does information asymmetry matter to equity pricing? Evidence from Firms’ geographic location.
Contemp. Account. Res. 30, 140–181.
Elliott, W.B., Rennekamp, K.M., White, B.J., 2015. Does concrete language in disclosures increase willingness to invest? Rev. Acc. Stud. 20, 839–865.
Elton, E.J., 1999. Expected return, realized return, and asset pricing tests. J. Financ. 54, 1199–1220.
Ertugrul, M., Lei, J., Qiu, J., Wan, C., 2017. Annual report readability, tone ambiguity, and the cost of borrowing. J. Financ. Quant. Anal. 52, 811–836.
Fama, E., French, K., 1992. The cross section of expected stock returns. J. Financ. 47, 427–465.
Fama, E., French, K., 1997. Industry costs of equity. J. Financ. Econ. 43, 153–193.
Fama, E., MacBeth, J., 1973. Risk, return, and equilibrium: empirical tests. J. Polit. Econ. 81, 607–636.
FASB, 2010. Statement of financial accounting concepts no.8. Available at. http://www.fasb.org/jsp/FASB/Page/PreCodSectionPage&cid=1176156317989>.
Ferris, S.P., Javakhadze, D., Rajkovic, T., 2017. The international effect of managerial social capital on the cost of equity. J. Bank. Financ. 74, 69–84.
Francis, J., LaFond, R., Olsson, P., Schipper, K., 2004. Costs of equity and earnings attributes. Account. Rev. 79, 967–1010.
Francis, J., LaFond, R., Olsson, P., Schipper, K., 2005a. The market pricing of accruals quality. J. Account. Econ. 39, 295–327.
Francis, J., Nanda, D., Olsson, P., 2008. Voluntary disclosure, earnings quality, and cost of capital. J. Account. Res. 46, 53–99.
Francis, J.R., Khurana, I.K., Pereira, R., 2005b. Disclosure incentives and effects on cost of capital around the world. Account. Rev. 80, 1125–1162.
Frankel, R., Lee, C.M.C., 1998. Accounting valuation, market expectation, and cross-sectional stock returns. J. Account. Econ. 25, 283–319.
Frankel, R., Jennings, J., Lee, J., 2016. Using unstructured and qualitative disclosures to explain accruals. J. Account. Econ. 62, 209–227.
Gao, P., 2010. Disclosure quality, cost of capital, and investor welfare. Account. Rev. 85, 1–29.
Garel, A., Gilbert, A.B., Scott, A., 2019. Linguistic Complexity and Cost of Equity Capital. Working Paper. Available at SSRN. https://ssrn.com/abstract=3240292.
Gaspar, J.-M., Massa, M., Matos, P., 2005. Shareholder investment horizons and the market for corporate control. J. Financ. Econ. 76, 135–165.
Gebhardt, W.R., Lee, C., Swaminathan, B., 2001. Toward an implied cost of capital. J. Account. Res. 39, 135–176.
Gode, D., Mohanram, P., 2003. Inferring the cost of capital using the Ohlson–Juettner model. Rev. Acc. Stud. 8, 399–431.
Guay, W., Kothari, S., Shu, S., 2011. Properties of implied cost of capital using analysts’ forecasts. Aust. J. Manag. 36, 125–149.
Hail, L., Leuz, C., 2006. International differences in the cost of equity capital: do legal institutions and securities regulation matter? J. Account. Res. 44, 485–531.
Hail, L., Leuz, C., 2009. Cost of capital effects and changes in growth expectations around US cross-listings. J. Financ. Econ. 93, 428–454.
Hann, R.N., Ogneva, M., Ozbas, O., 2013. Corporate diversification and the cost of capital. J. Financ. 68, 1961–1999.
24
He, J., Plumlee, M.A., Wen, H., 2019. Voluntary disclosure, mandatory disclosure and the cost of capital. J. Bus. Financ. Acc. 46, 307–335.
Hoberg, G., Phillips, G., 2010. Product market synergies and competition in mergers and acquisitions: a text-based analysis. Rev. Financ. Stud. 23, 3773–3811.
Hughes, J., Liu, J., Liu, J., 2009. On the relation between expected returns and implied cost of capital. Rev. Acc. Stud. 14, 246–259.
Hughes, J.S., Liu, J., Liu, J., 2007. Information asymmetry, diversification, and cost of capital. Account. Rev. 82, 705–729.
Irtisam, R., Black, J., Jain, P.K., 2020. Management Tone and Information Asymmetry. University of Memphis, Working paper.
Johnstone, D., 2015. Information and the cost of capital in a mean-variance efficient market. J. Bus. Financ. Acc. 42, 79–100.
Johnstone, D., 2016. The effect of information on uncertainty and the cost of capital. Contemp. Account. Res. 33, 752–774.
Jones, J.J., 1991. Earnings management during import relief investigations. J. Account. Res. 29, 193–228.
Kim, C., Wang, K., Zhang, L., 2019. Readability of 10-K reports and stock Price crash risk. Contemp. Account. Res. 36, 1184–1216.
Kim, J.-B., Li, L., Lu, L.Y., Yu, Y., 2016. Financial statement comparability and expected crash risk. J. Account. Econ. 61, 294–312.
Kim, J.W., Shi, Y., 2011. Voluntary disclosure and the cost of equity capital: evidence from management earnings forecasts. J. Account. Public Policy 30, 348–366.
Kim, S., Kraft, P., Ryan, S.G., 2013. Financial statement comparability and credit risk. Rev. Acc. Stud. 18, 783–823.
Koonce, L., Leitter, Z., White, B.J., 2018. The Effect of a Cautionary Notice on Investors’ Reactions to More and less Readable Disclosures. University of Texas Working
Paper.
Kravet, T., Muslu, V., 2013. Textual risk disclosures and investors’ risk perceptions. Rev. Acc. Stud. 18, 1088–1122.
La Porta, R., 1996. Expectations and the cross-section of stock returns. J. Financ. 51, 1715–1742.
Lambert, R., Leuz, C., Verrecchia, R., 2007. Accounting information, disclosure, and the cost of capital. J. Account. Res. 45, 385–420.
Lambert, R.A., 2009. Discussion of “on the relation between expected returns and implied cost of capital”. Rev. Acc. Stud. 14, 260–268.
Lambert, R.A., Leuz, C., Verrecchia, R., 2012. Information asymmetry, information precision, and the cost of capital. Review of Finance 16, 1–29.
Larcker, D.F., Rusticus, T.O., 2010. On the use of instrumental variables in accounting research. J. Account. Econ. 49, 186–205.
Larson, C.R., Resutek, R.J., 2017. Types of investor uncertainty and the cost of equity capital. J. Bus. Financ. Acc. 44, 1169–1193.
Lawrence, A., 2013. Individual investors and financial disclosure. J. Account. Econ. 56, 130–147.
Lee, C., Ng, D., Swaminathan, B., 2009. Testing international asset pricing models using implied costs of capital. J. Financ. Quant. Anal. 44, 307–335.
Lee, C.M., So, E., Wang, C., 2020. Evaluating firm-level expected-return proxies: implications for estimating treatment effects. The review of financial studies
hhaa066.
Lee, Y.J., 2012. The effect of quarterly report readability on information efficiency of stock prices. Contemp. Account. Res. 29, 1137–1170.
Lehavy, R., Li, F., Merkley, K., 2011. The effect of annual report readability on analyst following and the properties of their earnings forecasts. Account. Rev. 86,
1087–1115.
Levitt, A., 1999. Quality Information: The Lifeblood of Our Markets. Available at: https://www.sec.gov/news/speech/speecharchive/1999/spch304.htm.
Li, F., 2006. Do Stock Market Investors Understand the Risk Sentiment of Corporate Annual Reports. University of Michigan, Working Paper.
Li, F., 2008. Annual report readability, current earnings, and earnings persistence. J. Account. Econ. 45, 221–247.
Li, F., 2010. Textual analysis of corporate disclosures: a survey of the literature. J. Account. Lit. 29, 143.
Li, Y., Ng, D.T., Swaminathan, B., 2013. Predicting market returns using aggregate implied cost of capital. J. Financ. Econ. 110, 419–436.
Liu, M., Wysocki, P., 2016. Cross-sectional determinants of information quality proxies and cost of capital measures. Q. J. Financ. 7, 1650016.
Lo, K., Ramos, F., Rogo, R., 2017. Earnings management and annual report readability. J. Account. Econ. 63, 1–25.
Loughran, T., McDonald, B., 2009. Plain English, Readability, and 10-K Filings. Unpublished Working Paper–Notre Dame.
Loughran, T., McDonald, B., 2011. When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. J. Financ. 66, 35–65.
Loughran, T., McDonald, B., 2013. IPO first-day returns, offer price revisions, volatility, and form S-1 language. J. Financ. Econ. 109, 307–326.
Loughran, T., McDonald, B., 2014. Measuring readability in financial disclosures. J. Financ. 69, 1643–1671.
Lundblad, C., 2007. The risk return tradeoff in the long run: 1836–2003. J. Financ. Econ. 85, 123–150.
McNichols, M.F., 2002. The quality of accruals and earnings: the role of accrual estimation errors: discussion. Account. Rev. 77, 61–69.
Merkley, K.J., 2014. Narrative disclosure and earnings performance: evidence from R&D disclosures. Account. Rev. 89, 725–757.
Miller, B.P., 2010. The effects of reporting complexity on small and large investor trading. Account. Rev. 85, 2107–2143.
Mishra, D.R., O’Brien, T., 2019. Fama-French, CAPM, and implied cost of equity. J. Econ. Bus. 101, 73–85.
Mohanram, P., Rajgopal, S., 2009. Is PIN priced risk? J. Account. Econ. 47, 226–243.
Monga, V., Chasan, E., 2015. The 109,894-Word Annual Report (The Wall Street Journal).
Nelson, K.K., Pritchard, A.C., 2016. Carrot or stick? The shift from voluntary to mandatory disclosure of risk factors. J. Empir. Leg. Stud. 13, 266–297.
Newey, W., West, K., 1987. A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55, 703–708.
Ng, J., 2011. The effect of information quality on liquidity risk. J. Account. Econ. 52, 126–143.
Nikolaev, V., Van Lent, L., 2005. The endogeneity bias in the relation between cost-of-debt capital and corporate disclosure policy. Eur. Account. Rev. 14, 677–724.
Ogneva, M., 2012. Accrual quality, realized returns, and expected returns: the importance of controlling for cash flow shocks. Account. Rev. 87, 1415–1444.
Ohlson, J.A., Juettner-Nauroth, B.E., 2005. Expected EPS and EPS growth as determinantsof value. Rev. Acc. Stud. 10, 349–365.
Pástor, Ľ., Sinha, M., Swaminathan, B., 2008. Estimating the intertemporal risk–return tradeoff using the implied cost of capital. J. Financ. 63, 2859–2897.
Ramanna, K., Watts, R.L., 2012. Evidence on the use of unverifiable estimates in required goodwill impairment. Rev. Acc. Stud. 17, 749–780.
Rennekamp, K., 2012. Processing fluency and investors’ reactions to disclosure readability. J. Account. Res. 50, 1319–1354.
Schapiro, M., 2011. Roundtable on International Financial Reporting Standards in the United States, Washington, D.C., July 07.
SEC, 1998. A Plain English Handbook: How to Create Clear SEC Disclosure. Available at: https://www.sec.gov/pdf/handbook.pdf.
Shipman, J.E., Swanquist, Q.T., Whited, R.L., 2017. Propensity score matching in accounting research. Account. Rev. 92, 213–244.
Tan, H.-T., Wang, E.Y., Zhou, B., 2015. How does readability influence Investors’ judgments? Consistency of benchmark performance matters. Account. Rev. 90,
371–393.
Wang, C.C., 2015. Measurement Errors of Expected-Return Proxies and the Implied Cost of Capital. Harvard University, Working paper.
You, H., Zhang, X.-J., 2009. Financial reporting complexity and investor underreaction to 10-K information. Rev. Acc. Stud. 14, 559–586.
25

Journal of Corporate Finance: Hatem Rjiba, Samir Saadi, Sabri Boubaker, Xiaoya (Sara) Ding

Uploaded by

Copyright:

Available Formats

Journal of Corporate Finance: Hatem Rjiba, Samir Saadi, Sabri Boubaker, Xiaoya (Sara) Ding

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Journal of Corporate Finance: Hatem Rjiba, Samir Saadi, Sabri Boubaker, Xiaoya (Sara) Ding

Uploaded by

Copyright:

Available Formats

Journal of Corporate Finance 67 (2021) 101902

Contents lists available at ScienceDirect

Journal of Corporate Finance

Annual report readability and the cost of equity capital☆

2. Review of related literature and hypothesis development

2.1. Issues with the choice of information quality measure

2.2. Annual report readability and cost of equity

2.3. The role of annual report disclosure tone

3. Data and variable construction

3.1. Data sources

3.2. Regression variables

3.2.1. Cost of equity capital

3.2.2. Annual report readability

3.2.3. Control variables

Market beta (BETA)

Book-to-market ratio (BTM)

Firm size (SIZE)

Financial leverage (LEV)

Long-term earnings growth (LTG)

Analysts’ forecast bias (FBIAS)

Dispersion of analyst forecasts (DISP)

5.1. Univariate tests

5.2. Main regression analysis

rAVG (%) 39,181 6.32 2.89 4.43 5.81 7.55

(1) (2) (1)–(2) (3) (4) (3)–(4)

rAVG 5.131 5.698 − 0.567 *** 4.209 5.061 − 0.852 ***

(1) (2) (3) (4) (5)

BOG 0.054*** 0.050*** 0.065*** 0.032*** 0.54

6.1. Role of institutional investors and financial analysts

(1) (2) (3) (4) (5) (6)

BOG 0.032*** 0.059*** 0.022*** 0.086** 0.025*** 0.095***

Panel B: Uncertain Tone

Panel C: Weak Modal Tone

firms with less readable filings.

6.2. Role of imperfect market competition

6.3. Role of financial statement comparability

Variable Low High Low High Low High Low High

(1) (2) (3) (4) (5) (6) (7) (8)

BOG 0.088*** 0.016* 0.097*** 0.071*** 0.038*** 0.020** 0.097*** 0.038***

Imperfect Competition Tercile Perfect Competition Tercile

BOG 0.100*** 0.039

6.4. The effect of SEC’s, 1998 plain English rule

Variable (1) (2) (3) (4)

BOG 0.051*** 0.035*** 0.051*** 0.039***

7.1. Noise in analyst forecasts

Variable 1999–2003 2004–2007 2008–2011 2012–2017

(1) (2) (3) (4) (5) (6)

BOG 0.107*** 0.041*** 0.059*** 0.042*** 0.021** 0.023**

7.2. Alternative cost of equity estimates

Journal of Corporate Finance 67 (2021) 101902

7.3. The role of earnings quality

7.4. Additional control variables

(1) (2) (3) (4)

BOG 0.050*** 0.055*** 0.052*** 0.052***

7.6. Alternative financial readability measures

Variable Propensity Score Matched Variable Change regression

BOG 0.047*** ΔBOG 0.038***

INTERCEPT 5.007*** INTERCEPT − 1.068

Industry effects Yes Industry effects Yes

rAVG 5.131 5.698 − 0.567 * 4.209 5.061 − 0.852 *

BOG 0.054* 0.050* 0.065* 0.032* 0.54

BOG 0.032* 0.059* 0.022* 0.086 0.025* 0.095*

BOG 0.088*** 0.016* 0.097* 0.071* 0.038* 0.020 0.097* 0.038*

BOG 0.051* 0.035* 0.051* 0.039*

BOG 0.107* 0.041* 0.059* 0.042* 0.021 0.023

BOG 0.050* 0.055* 0.052* 0.052*

BOG 0.047* ΔBOG 0.038*

Lag (BOG) 0.029 Lag (BOG) 0.224*

INTERCEPT 9.653* INTERCEPT 58.229*