Advances in Accounting 53 (2021) 100531

Contents lists available at ScienceDirect

Advances in Accounting

journal homepage: www.elsevier.com/locate/adiac

Internal control effectiveness, textual risk disclosure, and their

usefulness: U.S. evidence☆
Mohamed Elsayed a,b,⁎, Tamer Elshandidy c,d
a
Alliance Manchester Business School, The University of Manchester, United Kingdom
b
Accounting Department, Mansoura University, Egypt
c
Department of Accounting, Ajman University, United Arab Emirates
d
Department of Accounting, Helwan University, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the impact of internal control effectiveness (ICE) on the level of textual risk disclosure
Received 18 April 2020 (TRD; including aggregate risk disclosure and its tone of good news and bad news about risk). Our findings sug-
Received in revised form 25 April 2021 gest that firms with an ineffective internal control system exhibit significantly lower levels of TRD than firms with
Accepted 26 April 2021
effective internal controls. Besides, we show a significant change in TRD behavior provided by managers of firms
Available online 7 May 2021
with recurrent ineffective internal controls. Pursuant to agency theory, this behavior change is prompted to re-
Editor: Dennis Caplan duce the expected public uncertainty and agency problems. We also investigate the usefulness of ICE reporting
and TRD to the market. Results suggest that firms reporting ineffective internal controls are likely to have higher
investor-perceived risk than firms reporting effective internal controls. Furthermore, TRD improves firms' market
Keywords: liquidity, and such improvement is principally driven by good news rather than bad news about risk. Collectively,
SOX our results fill an apparent gap in the literature on the importance of ICE, as well as the usefulness of the external
Internal control auditor's attestation on a firm's internal controls and management TRD.
Textual risk disclosure © 2021 Published by Elsevier Ltd.
Risk-related tones
Market liquidity
Investor-perceived risk
JEL classification:
M41
M42
M48

1. Introduction reliability of financial reporting and ultimately whether this regulation

Longstanding research (e.g., Ashbaugh-Skaife, Collins, LaFond, &
Kinney, 2008; Clinton, Pinello, & Skaife, 2014; Iliev, 2010; Schneider,
Gramling, Hermanson, & Ye, 2009) investigates the benefits of
Section 404(b) of the Sarbanes–Oxley Act of 2002 (SOX) (which obli-
gates the external auditor of accelerated filers to report on the effective-
ness of internal control over financial reporting) in improving the
☆ We are grateful to Dennis Caplan (The Editor) and the anonymous referees for helpful
comments and suggestions. This paper has benefited from comments and suggestions
from participants at the British Accounting and Finance Association SWAG Annual
Conference (University of South Wales, 2019) and research seminar at Ajman University
(2020). We thank Kirak Kim, Wim A Van der Stede, Gilad Livne, Catherine Shakespeare,
Svetlana Mira, Christopher Godfrey, and Jia Cao for their helpful suggestions. This
research did not receive any specific grant from funding agencies in the public,
commercial, or not-for-profit sectors.
⁎ Corresponding author at: Accounting and Finance Division, Alliance Manchester
Business School, The University of Manchester, Booth Street West, Manchester M15 6PB,
UK.
E-mail addresses: mohamed.elsayed@manchester.ac.uk (M. Elsayed),
t.elshandidy@ajman.ac.ae (T. Elshandidy).
is useful to the market.1 However, whether internal control effective-
ness drives firms to efficiently disclose their risks externally, an impor-
tant component for reliable information (AICPA, 1987; PCAOB, 2004;
SEC, 2003), is still unexplored (Elshandidy, Shrives, Bamber, &
Abraham, 2018).2 Additionally, investigation of the usefulness of con-
veying such internal knowledge (i.e., TRD and SOX 404(b) reporting)
to the market is inconclusive (e.g., Elshandidy et al., 2018; Gupta,
Sami, & Zhou, 2018; Schneider et al., 2009). To address this issue, we
raise two research questions: first, whether ICE influences TRD, and sec-
ond, whether ICE attestation by the external auditor and management's
TRD are useful to the market.
1
Accelerated filers are issuers that have a public float of at least $75 million as of six
months before fiscal year-end (SEC, 2005).
2
On average, 80% of a typical annual report consists of textual disclosures with a great
focus on risk-related information that is crucial for grasping and interpreting the numbers
and representations of quantitative data (Dyer et al., 2017; Kravet & Muslu, 2013; Lo,
Ramos, & Rogo, 2017).

https://doi.org/10.1016/j.adiac.2021.100531
0882-6110/© 2021 Published by Elsevier Ltd.
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

RQ1: Does internal control effectiveness (ICE) influence textual risk disclosure (TRD)?

TRD

Risk Factors Detection & Repotting System

Firm’s ICE Firm’s Managers

Remediation/Recurrence

External Auditor Reporting on ICW

RQ2: Are external auditor’s opinion under SOX 404 (b) and management’s TRD useful to the market?

Market Assessment

Market Liquidity Investor-Perceived Risk

Fig. 1. Research design. This figure illustrates the relationship between factors and variables associated with the paper's first and second research questions. Solid boxes show the
observable variables and thus, solid arrows present the direct/observable relations that are the main interest of our study. Dashed boxes show unobservable bodies and variables and
thus, dashed arrows present the indirect/unobservable relations that our study infers.

According to PCAOB (2004) Auditing Standard No. 2 (AS2), internal
control material weaknesses (ICW) are considered the most severe
form of internal control deficiencies (ICD) as compared to significant
deficiencies and control deficiencies. An ICW implies that there is an in-
formational problem in the firm's detecting and reporting system and is
an indicator that the firm has ineffective (i.e., low quality) internal con-
trols and, hence, its information is less reliable (Ashbaugh-Skaife,
Collins, & LaFond, 2009; Cheng, Dhaliwal, & Zhang, 2013; Feng, Li, &
McVay, 2009).3 Prior TRD literature largely focuses on two aspects: ex-
ploring the underlying drivers of risk disclosure and/or studying the
usefulness of such disclosure (for a recent review, see Elshandidy
et al., 2018). The literature, however, leaves a considerable gap around
the incentives of TRD because it remains largely voluntary, despite the
strictness of regulations on risk disclosure, due to subjectivity in
assessing firms' risks and uncertainties (e.g., Hope, Hu, & Lu, 2016;
SEC, 1997, 2010). Moreover, the informativeness of such disclosure is
substantially unknown (Kravet & Muslu, 2013). Consistent with these
premises, we examine the relationship between ICE as a driver or an in-
centive and the level of TRD, interpreted as aggregate risk disclosure, its
tone of good news and bad news about risk. We also investigate the
benefit of those two types of internal information that are carried to
the market by the external auditor and management as shown in Fig. 1.
Consistent with PCAOB (2004), 2007), prior research (e.g., Ashbaugh-
Skaife et al., 2008; Ashbaugh-Skaife et al., 2009; Donelson, Ege, & McInnis,
2017; Doyle, Ge, & McVay, 2007a) posits that in the presence of ICW,
there is more than a remote likelihood that the internal control system
is ineffective in detecting and disclosing risk factors and contingencies
on a timely basis. Consistent with this conjecture, we hypothesize that
3
Consistent with prior research (e.g., Donelson et al., 2017), we use ICW reported by
the external auditor under SOX 404 (b) as it ensures credible and rigorous evaluation of
internal control, as discussed in Section 3.1. Accordingly, we use ICE and ICW interchange-
ably throughout the paper.
an effective internal control system reveals more risk factors and contin-
gencies, and thereby enhances managers' ability to disclose a higher level
of risk information. In our further analyses, we test whether managers re-
spond to the external auditor's public report on their firms' ICW. If man-
agers respond, we expect mainly to observe a significant increase in
disclosure of bad news about risk because it becomes costly or difficult
for managers to delay releasing such news after receiving an external
auditor's adverse opinion on their firms' internal controls (Bao, Kim,
Mian, & Su, 2019; Skinner, 1994). Specifically, consistent with prior liter-
ature (e.g., Beneish, Billings, & Hodder, 2008; Cheng et al., 2013; Deumes
& Knechel, 2008; Feng et al., 2009), we expect managers of firms
reporting recurrent ineffective internal controls to be the most likely to
change their TRD behavior, moving from a relatively lower to a relatively
higher level of TRD. There are two possible reasons for this. First, these
managers would become aware from the external auditor about their
firms' internal control risks. Second, since an external auditor's adverse
opinion on a firm's internal controls entails an adverse public signal,
managers are expected to address this problem positively by providing
more disclosure to guide their firms' stakeholders.
Our second question, which concerns the benefit of internal control
and risk information provided under SEC requirements (e.g., SEC, 1997,
2005), adds to internal control literature (e.g., Elshandidy et al., 2018;
Schneider et al., 2009) by providing combined evidence about the useful-
ness of both ICW reporting and TRD. Prior research suggests that ICW
reporting would negatively impact investors' assessments of firm risk
(e.g., Ashbaugh-Skaife et al., 2009) but fails to find an impact on market
liquidity (e.g., Gupta et al., 2018). Such research also fails to offer an em-
pirical explanation for underlying reasons that could prevent detection
of the expected negative effect of ICW reporting on market liquidity,
leaving it as an open question. Prior research suggests that textual disclo-
sures (driven by good news) convey decision-relevant information that
is incorporated and utilized in addressing the adverse selection problem
(e.g., Elshandidy & Shrives, 2016; Yekini, Wisniewski, & Millo, 2016).

2
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

Since the process of interpreting textual disclosures is an individual cog- related disclosures (focusing on risk) and SOX 404(b) by investigating
nitive process, managers employ their disclosures strategically to favor- their informativeness to market participants and how TRD would re-
ably impact the decisions of investors (e.g., Schleicher & Walker, 2010). shape market assessment around ICW reporting.
Therefore, our paper argues that both ICW reporting and TRD are infor- The remainder of the paper is organized as follows. Section 2 reviews
mative to the market. If our conjecture is true, we expect to find a nega- related literature and develops the hypotheses. Section 3 describes the
tive impact of ICW reporting on the market (namely, investor-perceived methodology and data. Section 4 discusses the empirical results, and
risk). Besides, we expect to find a positive impact of TRD on the market provides further and robustness tests. Section 5 concludes, states limita-
(namely, market liquidity). This would imply that TRD (driven by good tions and suggests avenues for future research.
news about risk) is the conduit utilized by management to remove infor-
mation asymmetry around ICW reporting and thus, ensure steady trad-
2. Literature review and hypothesis development
ing, decreasing the difference between bid-ask orders, and increasing
market liquidity.
2.1. The effectiveness of internal controls and textual risk disclosure
Our findings suggest that firms with an ineffective internal control
system exhibit significantly lower levels of TRD than firms with effective
According to AS2, internal control over financial reporting is a pro-
internal controls. In our further analysis, we also document that recur-
cess designed by the company's principal executive and officers to en-
rently identified and publicly reported ICW prompts managers to signif-
sure the reliability of financial reporting for external purposes in
icantly change their TRD behavior by providing higher levels of risk
accordance with generally accepted accounting principles. Empirical re-
disclosures relative to other firms. In terms of the informativeness of
search (e.g., Cheng et al., 2013; Doyle et al., 2007a) indicates that finan-
ICW reporting and TRD, we find that ICW reporting does not signifi-
cial reporting reliability is driven by (is a function of) the firm's ICE. The
cantly impact market liquidity. However, it leads to a significant and
effectiveness of a firm's internal controls is determined by, among other
positive increase in investor-perceived risk. Furthermore, our finding
factors, risk assessments and information and communication
suggests that the level of aggregate risk disclosure positively and signif-
(Ashbaugh-Skaife et al., 2009). Consequently, an ineffective internal
icantly increases firms' market liquidity. When we distinguish the tone
control system (i.e., where ICW exist) is likely to adversely affect the
of the aggregate risk disclosure, results show that the positive impact on
ability of managers or their employees, in the normal course of perfor-
market liquidity is driven by good news about risk. This implies that
mance, to detect and report risk factors and contingencies on a timely
managers can affect investors' response to ICW reported by the external
basis (Ashbaugh-Skaife et al., 2009; Doyle et al., 2007a; Feng et al.,
auditor by reducing information asymmetries (i.e., increasing market li-
2009).
quidity) through providing greater amount of aggregate risk disclosure
Consistent with signaling theory, managers are motivated to provide
(particularly with a tone of good news about the risk), indicating their
a high level of information on how they effectively identify and manage
realization of their firms' risks.
their risks in order to distinguish their firms from other firms that do not
Our findings have several theoretical and practical implications.
manage risks or do so less effectively (Elshandidy & Shrives, 2016). In
First, they suggest that ICW affects not only the accuracy and quality
addition, providing a high level of risk information is a key mechanism
of information used by managers and disclosed to investors as was
that managers use to establish or change investors' risk expectations, re-
argued by Feng et al. (2009) and Ashbaugh-Skaife et al. (2008), but
duce litigation risk, and improve the firm's reputation for transparent
also the quantity of TRD. This, in turn, suggests that internal control
and credible disclosure (Feng et al., 2009). Therefore, managers of
effectiveness has broader implications than those previously docu-
firms with effective internal controls would be more capable and willing
mented in the literature and provides further support in favor of
to disclose more TRD.
the benefit of SOX 404(b) in improving financial reporting reliability.
Prior research on the relationship between the firm's internal control
Second, the ways in which the internal control system or external
and information reliability has reported various results. One strand
auditor's opinion under SOX 404(b) may help managers to increase
(e.g., Ashbaugh-Skaife et al., 2008; Chan, Farrell, & Lee, 2008; Doyle
the level and improve usefulness of their TRD should be of particular
et al., 2007a; Feng et al., 2009) indicates a positive association between
interest to regulators (e.g., AICPA; SEC; PCAOB) when assessing the
ICE and the accuracy and quality of disclosed information. Another strand
related regulations. Third, our results extend the evidence provided
finds negative implications of a firm's ICW for its financial reporting, in-
by Clinton et al. (2014) regarding the negative implications of ICW
cluding managers committing fraud (Donelson et al., 2017), adverse ef-
for financial analysts and provide external validity and generalization
fects on earnings forecasts (Chen, Martin, Roychowdhury, Wang, &
to the experimental evidence on the informativeness of ICW
Billett, 2017), and inefficient investment (Cheng et al., 2013).
reporting (Church & Schneider, 2016; Lopez, Vandervelde, & Wu,
This discussion suggests that effectiveness of internal control over fi-
2009). Fourth, the results also manifest the importance of identifying
nancial reporting is likely to have a causal association with firms' effi-
the tone of risk disclosure (Elshandidy & Shrives, 2016) to reach an
ciency in revealing and disclosing more risk factors and contingencies
insightful understanding of the drivers for and usefulness of TRD.
and thus, increases managers' ability to provide a high level of TRD. Al-
Fifth, the evidence from our textual analysis draws investors' atten-
ternatively, an ICW implies an information problem in the firm's finan-
tion to the reliability and usefulness of information conveyed by
cial reporting system that results in inefficiency in risk detection and
the 10-K narratives.
reporting. Therefore, we formulate the following hypothesis:
The insights provided by our paper contribute to both the internal
controls and risk disclosure literatures. We add to the prior research H1. Ceteris paribus, firms with an effective internal control system tend
(e.g., Campbell, Chen, Dhaliwal, Lu, & Steele, 2014) calling for the in- to exhibit a significantly higher level of TRD relative to firms with an in-
vestigation of risk disclosure incentives by hypothesizing that the effective internal control system.
nexus between ICE and the level of TRD is two-fold, concerning first,
the role of an effective internal control system in risk revelation; second,
the change in managers' risk disclosure behavior as a response to the 2.2. The informativeness of the internal control reporting and textual risk
identification and public reporting of ineffective internal controls. disclosure
This rationalizes the debate around the importance of SOX 404
(b) reports by accelerated filers in improving financial reporting reli- Ashbaugh-Skaife et al. (2009) posit that the indirect real effect of a
ability. We also contribute to the ongoing discussion among aca- firm's ineffective internal control over financial reporting translates
demics (e.g., Elshandidy et al., 2018; Gupta et al., 2018) and into investors' negative assessments of firm risk. They also suggest
regulators (e.g., SEC, 2009) about the usefulness of both narrative- that market participants' assessments of a firm's accounting signals

3
M. Elsayed and T. Elshandidy
are impaired when ICW is present. As an alignment with agency theory,
firms can reduce information asymmetries either between the firm and
market participants or between informed and non-informed investors
by increasing the levels of disclosure between these participants
(e.g., Beyer, Cohen, Lys, & Walther, 2010). Empirically, El-Mahdy and
Park (2014) conclude that some firm-specific characteristics (e.g., loan
characteristics in the secondary loan market) help to mitigate the
market's negative assessment of the disclosed internal control deficien-
cies by reducing information asymmetry.
Similarly, when an external auditor provides an adverse opinion on a
firm's internal controls, which indicates the reduction in the value of ac-
counting information provided to the public in the financial statements,
investors are expected to search for alternate sources of information
(e.g., corporate disclosures) to assess firm risk. In such corporate disclo-
sures, managers reveal information that reduces the market's informa-
tion asymmetry and, thus, motivate more active buyers and sellers of
the security and tighter bid-ask spreads (e.g., Diamond & Verrecchia,
1991; Leuz & Verrecchia, 2000). Specifically, prior research indicates
that annual report narratives are an important medium of communica-
tion that is utilized by managers to signal to investors their understand-
ing of the firm's activities and status (e.g., Beyer et al., 2010; Loughran &
McDonald, 2016). Previous studies (e.g., Schleicher & Walker, 2010)
indicate that textual disclosures (driven by good news) convey
decision-relevant information that can influence the information gap
between informed and uninformed investors. Some prior research
(e.g., Elshandidy & Shrives, 2016) emphasizes the importance of
employing the tone of disclosures in observing the impact of corporate
disclosures on market indicators.
Consistent with this notion, previous studies on the link between
market reaction and assessment of internal controls reporting docu-
ment a higher cost of equity (Ashbaugh-Skaife et al., 2009) and negative
stock price reactions (e.g., Beneish et al., 2008). Beneish et al. (2008)
suggest that investors' response, in terms of abnormal returns and eq-
uity cost of capital, to ICW reporting depends on recognizing the uncer-
tainty information disclosed by firms. Further, Kim and Park (2009)
suggest that a firm's voluntary disclosure that reduces market uncer-
tainty can mitigate the adverse influence of internal control deficiencies
reporting. Yekini et al. (2016) indicate that the level of good news can
affect investors' response to disclosed information. Schleicher and
Walker (2010) find that firms with large impending performance de-
cline, loss firms, risky firms and firms under monitoring provide more
good news. Campbell et al. (2014) conclude that the level of risk disclo-
sure reduces information asymmetry. Additionally, Elshandidy and
Shrives (2016) find that the tone of risk disclosure is directionally asso-
ciated with the market's information asymmetry, where good news
about risk, compared to bad, is found to mitigate information asymme-
try and thus improve market liquidity.
Thus, for testing the usefulness of ICW reporting and TRD, we em-
ploy bid-ask spread and stock return volatility, since they are well-
established proxies for market liquidity and investor-perceived risk in
the accounting literature (e.g., Elshandidy et al., 2018; Elshandidy &
Shrives, 2016). Specifically, while the bid-ask spread has long been con-
sidered a proxy for market liquidity (e.g., Bagehot, 1971), research links
investor-perceived risk (related to uncertainty about the firm's cash
flows) and stock return volatility (e.g., Shalen, 1993). While some
prior studies indicate that these two proxies are positively correlated
(e.g., Brennan & Subrahmanyam, 1996), stock return volatility is not a
reliable measure for information asymmetry, implying that these two
proxies capture different market behaviors (e.g., Campbell et al., 2014;
Elshandidy & Shrives, 2016; Garfinkel, 2009; Gupta et al., 2018; Kravet
& Muslu, 2013; Leuz & Verrecchia, 2000). The general intuition is that
stock return volatility denotes investors' risk perceptions since it cap-
tures the range and confidence level in their predictions of future per-
formance. The bid-ask spread is commonly thought to explicitly
measure information asymmetry since it addresses the adverse selec-
tion problem that arises from transacting in firm shares in the presence
4
Advances in Accounting 53 (2021) 100531
of asymmetrically informed investors (e.g., Leuz & Verrecchia;
Garfinkel, 2009; Bao & Datta, 2014).
Collectively, the above-mentioned literature suggests that while
ICW reporting would negatively impact investors' assessments of firm
risk, it is unlikely to impact market liquidity, without offering an empir-
ical explanation for underlying reasons that could prevent detection of
the expected negative effect of ICW reporting on market liquidity
(e.g., Ashbaugh-Skaife et al., 2009; Gupta et al., 2018). Consistent with
Bertomeu, Beyer, and Dye's (2011) theoretical argument that disclosing
more information can increase firms' market liquidity, the contempora-
neous TRD is likely the medium that positively affects liquidity. Specifi-
cally, the theoretical and empirical link between risk disclosure
(particularly good news about risk) and market liquidity is strong
(e.g., Diamond & Verrecchia, 1991; Elshandidy & Neri, 2015;
Elshandidy & Shrives, 2016; Leuz & Verrecchia, 2000; Yekini et al.,
2016). However, evidence on the impact of such disclosure on
investor-perceived risk is debatable (probably due to the lack of
company-specific risk information) (e.g., Abraham & Shrives, 2014;
Bao & Datta, 2014; Kaplan, 2011; Kravet & Muslu, 2013; Linsley &
Shrives, 2006; Schrand & Elliott, 1998). This discussion, therefore,
leads to formulating the following hypothesis:
H2. Ceteris paribus, ICW reporting and TRD are informative to the mar-
ket (as proxied by stock return volatility and bid-ask spread).
3. Methodology
3.1. Sample selection and data collection
Our sample compiles data related to 10-K filings of listed firms on
the SEC EDGAR database for fiscal years 2004–2006 ending on Decem-
ber 31 (reasons are given below). We require sample firms to have a
SOX 404(b) auditor opinion available on Compustat (i.e., our sample
contains accelerated filers), and no missing data from Compustat,
Datastream, and CRSP databases. Various identifiers, involving CIK of
the Edgar filings and the above-mentioned databases' codes such as
TICKER, GVKEY and PERMNO, are used to merge our data. We exclude
1669 firm-years with Compustat equity market capitalization less
than $75 million (or missing) at the fiscal year ends in order to keep
the accelerated filers and minimize the possibility that a firm no longer
exists.4 We also exclude 5939 firm-years of foreign firms because they
were not subject to SOX404 until July 15, 2006 (July 15, 2007, for foreign
firms with a public float of under $700 million) and 8042 firm-years of
financial firms (SIC 6000–6999) because of their distinct regulations
and accounting practices (Chan et al., 2008; Iliev, 2010).
Our sample retains only firms with a fiscal year end on December 31
to synchronize the time period of firms with ICW and without ICW, as
well as exploring the impact of internal control and risk information on
the market assessment in a precise and timely fashion (e.g., Ashbaugh-
Skaife et al., 2009; Chan et al., 2008; Elshandidy & Shrives, 2016). Our
final sample is composed of 3043 firm-year observations, 222 of which
have ineffective internal controls (i.e., at least one ICW exists at year-
end). Sample construction is outlined in Table 1.
In our analysis, consistent with prior research (e.g., Clinton et al.,
2014; Donelson et al., 2017; Feng et al., 2009), we depend on ICW re-
ported by the external auditor under SOX 404(b) instead of that
disclosed by management under SOX 404(a) and SOX 302 because the
latter is more ambiguous and less rigorous. Additionally, previous stud-
ies (Schneider et al., 2009) indicate that the external auditor, rather
than management, is more effective in detecting and publicly disclosing
ICW. For example, Donelson et al. (2017) suggest that using the external
4
Following prior research (e.g., Ashbaugh-Skaife et al., 2008), we define accelerated
filers by initially employing market capitalization as a proxy for public float because they
are very similar for most firms (see footnote 1). Then, we manually assure that our final
sample is comprised of accelerated filers.
M. Elsayed and T. Elshandidy
Table 1
Sample details.
Sample selection
All Compustat firms for fiscal years 2004, 2005, and 2006
Excluding:
Foreign firms
Financial firms (SIC 6000–6999)
Firms with a fiscal year end other than December 31
Auditor disclaimer of opinion on internal control or delayed filing
Companies with Compustat equity market capitalization less than
$75 million (or missing) at the end of their fiscal year
Missing data of audit, financial, ownership structure or market
information
Final firm-year observations
This table summarizes our sample construction.
auditor's opinion as opposed to management disclosure is more accu-
rate, especially if managers are fraudulent. Consequently, we consider
that the internal control system is ineffective if the external auditor's
opinion is adverse (i.e., there is an ICW), while it is considered as effec-
tive if the auditor's opinion is clean (see Appendix B, Panel B).
December 2004 is the starting point of our sample because SOX is ef-
fective for accelerated filers for fiscal years ending on or after November
15, 2004. By extending our sample to December 2006, we address the
potential criticism of using a sample period limited to the first year of
SOX 404 for studying the benefits of ICE and the impact on market
assessment.5 We use this longer sample period to establish
intertemporal analyses for the association between ICW and TRD, as
well as the impact of ICW reporting and TRD on market indicators, to
test their usefulness. We end our sample in December 2006 for two
main reasons. First, to avoid measurement error endogeneity bias
resulting from the application of AS5 (which is related to a significant
decline in the accuracy of identifying and reporting ICW; see, SEC
(2009); Rice and Weber (2012); Chasan (2013); PCAOB (2013);
Schroeder and Shepardson (2016)) and the financial crisis in 2007
(e.g., Dedman, 2016). Second, to avoid results bias due to the smaller
variability of ICW reporting. Prior studies (Dowdell, Kim, Klamm, &
Watson, 2013; Feng et al., 2009; Schroeder & Shepardson, 2016) indi-
cate that the proportion of firms reporting ineffective internal control
decreases significantly over time, especially from the application of
AS5 in 2007 onwards. For a sample period from 2004 to 2013, Chen,
Eshleman, and Soileau (2016), p. 15) are compelled to delete 14,326 ob-
servations from their main sample of 18,593 firm-year observations,
leading to a final sample of 4267 firm-year observations (which is
roughly similar to our final sample), “because these firms do not exhibit
variation in the ICW variable during the sample period.”6 Like prior re-
search (e.g., Rose-Green, Huang, & Lee, 2011), we employ the most ap-
propriate setting for our study.
3.2. Textual risk disclosure analysis7
The clean textual content of our sample 10-K filings, after eliminat-
ing HTML, ASCII-encoded graphics, and tables, is used because we
focus on the narrative sections. Dyer, Lang, and Stice-Lawrence (2017)
document that textual disclosure on both risk factors and internal con-
trol is not confined to a single section of the 10-K but spreads and
5
Arguably, it is less suitable to identify cross-sectional variation in ICE in the first year of
SOX 404, and, thus, contemporaneous audited reporting measures are not ideal to identify
ICE enhancements (Schroeder & Shepardson, 2016).
6
Attempting to address these limitations on ICW as a proxy for ICE, Buslepp, Legoria,
Rosa, and Shaw (2019) suggest the misclassification of audit-related fees as an alternative
proxy for ICE. This proxy is, however, limited to the M&A setting and developed using the
unaudited disclosures of management.
7
We acknowledge Bill McDonald for providing access to his data repository.
Advances in Accounting 53 (2021) 100531
interweaves across all the sections. Thus, employing TRD analysis that
encompasses the entire 10-K is reasonable and consistent with our
Firm-year study's purpose. In order to capture the TRD in the 10-K narratives, we
observations employ Elshandidy and Shrives' (2016) complete risk wordlist, which
33,233 is consistent with that of Kravet and Muslu (2013), as both relied on
searching the entire sections of annual reports or 10-K fillings. Follow-
5939
ing Elshandidy and Shrives (2016), we classify the aggregate risk
8042 words in terms of tone into good or bad news about risk. After excluding
7065 neutral words that reflect neither the up nor the downside
6341 (e.g., significant, probable, and differ) from aggregate risk words, a
1669
word that reflects the positive side of the risk (i.e., potential gains and
1134 opportunities) is classified as good news about risk, while a word that
reflects the negative side of the risk (i.e., potential losses/threats) is clas-
3043 sified as bad news about risk. Accordingly, we identify each filing's ag-
gregate risk disclosure, good news and bad news about risk using the
terms shown in Appendix B, Panel A.
Following textual analysis literature (see the review of Loughran and
McDonald (2016)), we employ automated textual content analysis
using Diction 7 software to measure each filing's aggregate risk disclo-
sure, its tone of good news or bad news about risk. The aggregate risk
disclosure score is calculated by the percentage of words that are
contained in the complete risk wordlist (i.e., the number of words indi-
cating the aggregate risk scaled by the total number of words in the 10-
K). Similarly, each filing is further assessed based on its tone of good or
bad news about risk. The score of good news about risk is calculated by
the percentage of words that are classified as having a positive side,
while the percentage of words that are classified as having a negative
side is calculated to measure the score of bad news about risk.
Following prior research (e.g., Abraham & Cox, 2007; Elshandidy &
Shrives, 2016), we check the reliability and validity of the TRD scores
generated by the complete risk wordlist as follows. The reliability of
the aggregate risk disclosure scores and tone of risk (good news and
bad news about risk) is tested using Cronbach's alpha. This statistical
test enables judgment of the extent to which a dataset captures a partic-
ular underlying construct. The Cronbach's alpha of 82.99% for the com-
puted scores of the TRD implies that the internal consistency between
the aggregate risk disclosure and its tone is higher than the generally ac-
cepted value in the social sciences of 70% (Abraham & Cox, 2007). For
the validity check, we test the correlation between TRD scores that are
generated by the complete risk wordlist of Elshandidy and Shrives
(2016) and that of Kravet and Muslu (2013).8 Our full sample's results
show that both risk wordlists are highly correlated (r = 0.85, significant
at the 1% level), which implies that each of the two risk wordlists cap-
tures a large proportion of risk disclosure from the 10-K narratives.
The evidence suggests that the computed TRD scores are both valid
and reliable.
3.3. Empirical model
To test H1 and H2, we employ a fixed effects model using all internal
control audit data for our sample period. Using a fixed effects model en-
ables us to account for changes in TRD as a result of the effectiveness of
internal controls during the period of the study (Eq. (1)), plus changes
in market assessment or benefit as a result of the observed risk informa-
tion and ICW (Eq. (2)). The model also accounts for bias that would arise
in the dependent variable due to firm and/or industry-specific effects; it
also excludes the effects of time-invariant covariates.
nj
TRDit ¼ Bo þ B1 ICW it þ ∑ δj Control Variablesjit þ εit ð1Þ
j¼1
8
The risk wordlist of Kravet and Muslu (2013) comprises 20 risk-related keywords
(where * implies that suffixes are allowed): can/cannot, could, may, might, risk*, uncer-
tain*, likely to, subject to, potential*, vary*/varies, depend*, expos*, fluctuat*, possibl*, sus-
ceptible, affect, influenc*, and hedg*.
5
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

In Eq. (1), TRD, in separate tests, equals the score of aggregate risk dis- Table 2
closure (AGG_RISK), and the scores of the tone of risk as good news Descriptive statistics.

(GOOD_RISK), or bad news about risk (BAD_RISK). ICW, our independent Variable Obs. Mean Median Std. Dev. Q1 Q3
variable of interest, is a dummy variable that takes a value of one if the ex- Textual risk disclosure (TRD):
ternal auditor issued an adverse opinion on the firm's internal control AGG_RISK 3043 1.294 1.288 0.281 1.102 1.481
system (ICW exists), and zero if the opinion is clean (ICW does not BAD_RISK 3043 0.476 0.463 0.125 0.386 0.549
exist). Following prior literature on TRD (e.g., Campbell et al., 2014; GOOD_RISK 3043 0.346 0.335 0.138 0.242 0.432
Elshandidy & Shrives, 2016) and internal control (e.g., Ashbaugh-Skaife Market indicators (usefulness):
et al., 2008; Deumes & Knechel, 2008; Feng et al., 2009), our set of control SD 3043 0.022 0.020 0.011 0.015 0.026
variables includes inside ownership concentration and capital structure, SPREAD 3043 0.192 0.126 0.194 0.078 0.225
which we employ as surrogates for agency problems in addition to two Explanatory and control variables:
dummy variables: big four auditors as a surrogate for external audit qual- ICW 3043 0.073 0.000 0.260 0.000 0.000
ity and auditor opinion on the financial statements. In addition, we con- INSIDE_OWN 3043 0.205 0.157 0.193 0.034 0.302
DEBT_EQU 3043 0.702 0.371 1.563 0.026 0.868
trol for firm characteristics including size, profitability, liquidity,
AUD_OPIN 3043 0.467 0.000 0.499 0.000 1.000
performance, and growth, as well as market beta. The definitions and BIG_4 3043 0.917 1.000 0.276 1.000 1.000
measures of these control variables are provided in detail in Appendix A. LN_TA 3043 6.911 6.726 1.733 5.635 8.022
ROE 3043 0.069 0.116 0.331 0.039 0.189
nj CR 3043 2.796 1.960 2.471 1.310 3.230
INFOitþ1 ¼ μ 0 þ μ 1 ICW it þ μ 2 TRDit þ ∑ Ωj Control Variablesjit þ eit ð2Þ FFO 3043 0.171 0.167 0.557 0.071 0.349
j¼1 GROWTH 3043 0.202 0.129 0.341 0.052 0.261
BETA 3043 1.329 1.231 0.683 0.846 1.734
BM 3043 0.414 0.373 0.227 0.248 0.551
In Eq. (2), INFO denotes informativeness which, in separate tests, is TRAD_VOL 3043 0.849 0.680 0.643 0.418 1.079
DIVIDENDS 3043 0.426 0.000 0.495 0.000 1.000
proxied by investor-perceived risk as represented by the volatility of
market returns (SD); and market liquidity as represented by bid-ask This table presents summary statistics for the variables used in our analyses. All continu-
spread (SPREAD). These dependent variables are measured, based on ous variables are winsorized at 1% on both tails. Variable definitions, measures, and
sources are provided in Appendix A.
daily data, as the average over 60 trading days period beginning two
trading days after the 10-K filing (e.g., Campbell et al., 2014; Kravet &
Muslu, 2013). Consistent with prior research (e.g., Elshandidy &
Shrives, 2016; Leuz & Verrecchia, 2000), we make our examination pe- non-ICW firms, the univariate tests indicate that ICW firms are more
riod long enough for investors to assess TRD and ICW reporting, but likely to disclose relatively less good news about risk and more bad
short enough to limit the influence of confounding events. Regarding news about risk (|t|-statistics of 3.085, significant at the 1% level, and
our independent variables of interest, TRD and ICW are as defined in 2.332, significant at the 5% level, respectively). This initially supports
Eq. (1). Control variables are common to those present in Eq. (1), but, our hypothesis that managers of firms with effective internal controls
consistent with prior research on the informativeness of general disclo- would be in a better position to signal more good news about the risk,
sure and TRD (e.g., Campbell et al., 2014; Elshandidy & Shrives, 2016; whereas, the cost of the publicly-reported ICW may prompt managers
Leuz & Verrecchia, 2000) and internal control (e.g., Dowdell et al., to respond by increasing the disclosure of bad news about the risk.
2013; El-Mahdy & Park, 2014; Gupta et al., 2018), we further control The t-tests show a statistically insignificant difference between aggre-
for dividends payout (dichotomous) and other market factors of the gate risk disclosure levels for ICW and non-ICW firms. This result,
book-to-market ratio and trading volume. All independent and control though unexpected, accords with Elshandidy and Shrives' (2016) find-
variables are measured at fiscal year-end t, and detailed variable defini- ing that aggregate risk disclosure is less likely to be associated with
tions and measures are provided in Appendix A. firms' environmental incentives. The univariate tests additionally reveal
greater investor-risk perceptions for ICW firms compared to non-ICW
4. Empirical results firms (|t|-statistic of 4.427, significant at the 1% level). This provides ini-
tial support for our second hypothesis that ICW reporting increases
4.1. Descriptive statistics investor-perceived risk. In terms of the control variables, we observe
that ICW firms are more likely to receive an unqualified audit opinion
Table 2 reports summary statistics for the explanatory, control, and on their financial statements, and are more likely to have higher market
dependent variables. These descriptive statistics are shown for the en- beta, book to market ratio, and trading volume than non-ICW firms (|t|-
tire dataset, which consists of 3043 firm-year observations, of which statistics of 2.283, 4.703, 2.475, and 2.379, significant at the 5% level or
222 firm-year observations have ineffective internal controls (i.e., at better). We also observe that ICW firms are smaller, have lower profit-
least one ICW was identified on the audit). Continuous variables are ability and are less likely to report dividends payout than non-ICW
winsorized at 1% on both tails to mitigate the effect of outliers. firms (|t|-statistics of 5.644, 2.826, and 5.879, significant at the 1%
Over the sample period, on average, the percentage of words that re- level). Overall, these differences are consistent with results obtained
flect aggregate risk information represents about 1.30% relative to the from studies on the determinants of ICW (e.g., Doyle, Ge, & McVay,
total number of words disclosed by U.S. non-financial firms in their 2007b). These results also illustrate the importance of controlling for
10-Ks. On average, about 0.48% of the 10-Ks' words suggest bad news these innate firm characteristics in our analyses.
related to risk, while 0.35% of words suggest good news about risk. Table 3 reports the pair-wise correlations (Pearson product moment
This implies that about 0.47% (1.30% - 0.48% - 0.35%) of the 10-Ks' correlations are exhibited on the upper-right-hand portion, and Spear-
words represent a neutral tone associated with risk disclosure. With man rank-order correlations are exhibited on the lower-left-hand por-
net tone about risk (calculated as the difference between good news tion). We discuss the Pearson correlations but note that the Spearman
and bad news about risk, i.e., optimistic residual; untabulated for brev- rank-order correlations are generally consistent with the Pearson corre-
ity) averaging about −0.13%, the U.S. non-financial firms' sentiment in lations. The aggregate risk disclosure is positively associated with both
their 10-Ks is marginally pessimistic in disclosing risk-related bad news (0.81) and good news (0.73) about risk, which implies that
information. U.S. non-financial firms significantly employ the tone of risk informa-
We examine the difference in means between ICW and non-ICW tion to communicate signals about risks involved in their aggregate
firms using a t-statistics test (unreported for brevity). Compared to risk disclosure. Consistent with our descriptive statistics, the ICW

6
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

This table reports the correlation coefficients of the variables used in our analyses. Bold numbers indicate significance based on two-tailed t-tests, at the 0.05 level or better. All continuous variables are winsorized at 1% on both tails. Variable def-
variable is negatively (positively) correlated with good news (−0.06)

−0.121

−0.310
−0.169
−0.106
−0.076

−0.283

−0.146
−0.295

−0.283
−0.007

0.292

0.081
0.112
0.084
0.465
0.277

0.121

0.064
and (bad news about risk; 0.04). The ICW variable also reveals a positive
19 correlation with the volatility of stock return (0.08). As expected, mar-
ket liquidity and risk perception proxies exhibit relatively large positive

−0.063

−0.258

−0.130

−0.058
−0.072

−0.156

−0.323
−0.020
−0.002
0.082
0.082

0.182

0.043

0.044

0.178

0.202
0.395
0.004
correlations (0.43). Consistent with Elshandidy and Shrives (2016),
18

good news about risk is positively (negatively) correlated with market

−0.094 liquidity (investor-perceived risk) proxies, while bad news about the

−0.048

−0.050

−0.122
−0.077

−0.176
−0.005

−0.007

−0.016
risk has opposite directions; |r| ranging from about 0.07 to about 0.17.9
0.051

0.160
0.045

0.096

0.106

0.085
0.016

0.030

0.001
17

4.2. Testing H1: The influence of ICE on TRD

−0.088

−0.121

−0.244
−0.186

−0.062

−0.081

−0.287
−0.002

−0.022
0.127
0.189

0.314

0.085

0.259

0.111

0.415
0.028

0.022
16

Table 4 shows results related to H1, which addresses whether ICE in-
fluences the level of TRD. Across each of the three models' estimations,
−0.050

−0.042

−0.094
−0.100

−0.042

−0.174

−0.159
−0.008

−0.033

−0.035
−0.014
−0.018
−0.026

−0.023

ICW is negatively associated with the level of aggregate risk disclosure,

0.112

0.079

0.149

0.235
its tone of bad news and good news of risk. Despite the statistically triv-
15

ial result with respect to bad news about risk, this finding collectively in-
−0.052

−0.181
−0.232

−0.053

−0.195

−0.036
−0.165
−0.002
−0.018

−0.001
−0.033

dicates that firms with an ineffective internal control system exhibit

0.076

0.237

0.147
0.503

0.108

0.061
0.062
significantly lower levels of risk information, which are observed in
14

terms of aggregate risk disclosure (t-statistic −2.009, significant at the

5% level) and its tone of good news about risk (t-statistic −2.720, signif-
−0.226

−0.172
−0.152
−0.118
−0.435
−0.241

−0.067

−0.335
−0.011
0.086
0.175

0.219
0.156

0.075

0.064
0.091
0.337

0.216

icant at the 1% level). That is, all else being equal, the economic signifi-
13

cance of having effective internal controls is related to higher TRD of

3.09% (0.040/1.294) of the mean of aggregate risk disclosure, 1.68%
−0.108

−0.268
−0.278
−0.051
−0.046

−0.249

−0.165
−0.295
0.055

0.319

0.081

0.043
0.290

0.474
0.135

0.336
0.024

0.008

(0.008/0.476) of the mean of bad news and 6.36% (0.022/0.346) of the

12

mean of good news about risk. This evidence supports H1 that, com-
pared to having effective internal controls, the existence of an ICW im-
−0.039
−0.075

−0.427
−0.495
−0.102
−0.246

−0.519

−0.083
−0.234

−0.018

plies an informational problem in the firm's reporting system.

0.229

0.232
0.246
0.287

0.347

0.126

0.463
0.006
11

Consequently, managers of firms with ICW are less likely to disclose

high levels of TRD, possibly because their internal control system is inef-
−0.097
−0.209

−0.102

−0.109
−0.001

−0.035

−0.030
−0.017

fective in revealing the unknown and inherent risk factors and

0.081

0.079
0.079

0.303
0.059

0.045
0.062
0.084
0.012

0.019

uncertainties.
10

Our results extend the conclusions drawn by prior research by sug-

−0.154
−0.095

−0.092

−0.189
−0.038
−0.062
−0.105

gesting that ICW affects not only the accuracy (e.g., Feng et al., 2009)
0.037

0.095

0.041

0.084

0.079
0.240

0.103

0.112
0.023

0.020

0.018

and quality (e.g., Chan et al., 2008) of information or decisions by man-

9

agers (e.g., Cheng et al., 2013) but also the amount of their TRD in the
10-K filings. They also accord with textual analysis research showing
−0.048
−0.036

−0.092
−0.053

−0.521
−0.361
−0.089
−0.220

−0.126
0.037

0.168
0.166
0.541
0.137

0.095

0.288
0.010
0.028

that good news is more reliable and representative of the contextual na-
8

ture of textual disclosures (which is confirmed further when testing H2)

(e.g., Schleicher & Walker, 2010; Yekini et al., 2016). Turning to the con-
−0.055

−0.102
−0.109
−0.124
−0.310
−0.111

−0.112
−0.144
−0.017
−0.011

−0.019
−0.022

−0.022
0.134
0.190

0.168

0.096
0.029

trol variables, firm size and market beta appear to be the most influen-
7

tial factors on the level of TRD, either for the aggregate risk disclosure or
its tone of bad news and good news about risk. These results are consis-
−0.056

−0.040

−0.107
−0.119

−0.055

−0.106
−0.003

−0.035

−0.007

−0.010
0.042

0.080
0.072

0.047

0.041

0.076
0.047
0.041

tent with prior research on risk disclosure (e.g., Abraham & Cox, 2007;
Elshandidy & Shrives, 2016), where risky firms are motivated to detail
6

their risk exposure and procedures followed to mitigate it. Large firms
−0.165

−0.172
−0.106
−0.231
−0.656
−0.410

−0.262
−0.095

−0.317
−0.244
−0.024

are more likely to own the resources required for establishing a strong
0.072

0.427

0.126
0.300

0.227

0.074
0.229

risk management system that are able to detect and convey risk infor-
5

mation efficiently.
−0.142

−0.306
−0.191
−0.135
−0.560
−0.280

−0.121

−0.093

−0.430
0.044
0.089

0.314
0.122
0.226

0.368

0.178
0.403

0.239
Pearson (top) and Spearman (bottom) correlation coefficients.

initions, measures, and sources are provided in Appendix A.

4.3. Managers' TRD behavior as a response to the publicly reported ICW

4

−0.190
−0.179
−0.055
−0.076

−0.195

−0.071

−0.069

In the previous analyses, we posited and found that the level of TRD
0.728
0.331

0.112
0.092
0.076
0.240
0.317

0.197
0.049

0.041

0.293

(either aggregate risk, or its tone of good news and bad news about risk)
3

is negatively associated with a firm's ineffective internal controls. How-

−0.138

−0.105
−0.182

−0.079
−0.103

−0.127
−0.020

−0.005

ever, as we discussed in Section 1, the external auditor's public report of

0.806

0.341
0.129
0.079
0.040

0.180

0.197

0.094
0.028

0.032

an ICW would prompt managers to provide more information about

2

risk to alleviate investors' uncertainties and agency problems (Deumes

& Knechel, 2008; Leuz & Verrecchia, 2000). This change in TRD behavior
−0.116

−0.047
−0.018

−0.017

−0.002

−0.005
0.796
0.726
0.055

0.038

0.103
0.078

0.135

0.085
0.002

0.008

0.027

0.015

would occur especially if the ICW is recurrent due to managers' aware-

1

ness of ICW and the adverse public signal that the reported ICW implies.
According to Feng et al. (2009), managers would change their disclosure
7 INSIDE_OWN

19 DIVIDENDS
18 TRAD_VOL
3 GOOD_RISK

15 GROWTH
8 DEBT_EQU
9 AUD_OPIN
1 AGG_RISK
2 BAD_RISK

5 SPREAD

11 LN_TA
10 BIG_4

9
In addition, the correlation coefficients are also used to diagnose multicollinearity. The
16 BETA
Variable

12 ROE

14 FFO

17 BM
6 ICW

13 CR

unreported variance inflation factors (VIFs) are less than 10, ranging from 1.03 to 1.62,
Table 3

4 SD

which implies that multicollinearity is not inherent in our multivariate regressions

(Ashbaugh-Skaife et al., 2008).

7
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

Table 4 Table 5
Fixed effect panel regressions of TRD on ICW. Results for the change of internal control effectiveness and management behavior.

Variables AGG_RISK BAD_RISK GOOD_RISK Panel A: Fixed effects panel regressions of TRD on recurrence of ICW

Model (1) Model (2) Model (3) Variables AGG_RISK BAD_RISK GOOD_RISK

ICW −0.040** −0.008 −0.022*** Model (1) Model (2) Model (3)
(−2.009) (−1.127) (−2.720)
ICW −0.049** −0.011 −0.025***
INSIDE_OWN 0.013 0.004 0.007
(−2.343) (−1.410) (−2.933)
(0.235) (0.174) (0.319)
ICW_ RECUR 0.116*** 0.033** 0.039***
DEBT_EQU −0.003 −0.000 −0.000
(2.848) (2.248) (2.637)
(−0.626) (−0.223) (−0.261)
Control variables Included Included Included
AUD_OPIN 0.013 0.008** 0.003
Constant 0.777*** 0.483*** 0.066
(1.416) (2.370) (0.647)
(5.226) (8.630) (1.040)
BIG_4 0.014 0.009 −0.005
Observations 3043 3043 3043
(0.374) (0.585) (−0.452)
R-squared 0.018 0.018 0.024
LN_TA 0.066*** −0.006 0.039***
F-value 2.939*** 2.716*** 4.172***
(3.063) (−0.693) (4.254)
ROE 0.006 −0.021* 0.019*
Panel B: Within-firm changes in TRD and changes in ICE (remediation versus
(0.170) (−1.732) (1.696)
recurrence)
CR 0.003 0.003 −0.000
(0.574) (1.317) (−0.111) Variables Δ AGG_RISK Δ BAD_RISK Δ GOOD_RISK
FFO 0.017 0.000 0.015**
Model (1) Model (2) Model (3)
(0.947) (0.010) (2.493)
GROWTH −0.015 −0.013** 0.001 ADV_ADV 0.096** 0.029* 0.038**
(−0.970) (−2.085) (0.206) (2.203) (1.895) (2.260)
BETA 0.029*** 0.013*** 0.007* UNQ_ADV −0.022 −0.009 −0.009
(2.772) (3.306) (1.767) (−0.711) (−0.662) (−0.786)
Constant 0.772*** 0.482*** 0.065 ADV_UNQ 0.006 0.001 0.009
(5.159) (8.552) (1.011) (0.240) (0.138) (0.885)
Observations 3043 3043 3043 Control variables Included Included Included
R-squared 0.015 0.016 0.022 Constant 0.069*** 0.019*** 0.023***
F-value 2.616*** 2.580*** 4.151*** (10.092) (7.214) (8.226)
Observations 1937 1937 1937
This table reports the coefficients on the explanatory variables of the fixed effects panel re-
R-squared 0.019 0.020 0.016
gression models. It examines H1 to answer the first research question about whether ICE
F-value 2.668*** 2.897*** 2.803***
(proxied by ICW) influences TRD. Robust standard errors adjusted for clustering at the
firm level. T-statistics in parentheses. Significance level: *** p < 0.01, ** p < 0.05 and * Panel A of this table shows our level analysis and reports the coefficients on the explana-
p < 0.1, using two-tailed tests. Variable definitions, measures, and sources are provided tory variables and the summary of the fixed effects panel regression models. ICW_RECUR
in Appendix A. is a dummy variable equal to 1 when there is an adverse internal control opinion in two
successive years. Panel B of this table shows our change analysis, where the drop in sample
size due to the requirement of data on the difference between successive years). Δ denotes
behavior because the publicly-disclosed ICW could, inter alia, lead the the change with respect to each of the mentioned variables. The change is also applied for
the control variables. The change (Δ) indicates the differences between a firm's score in
market to discount their normal disclosures. Following Ashbaugh-
year t + 1 and its score in year t. ADV_ADV, UNQ_ADV, and ADV_UNQ are three dummy
Skaife et al. (2008), Feng et al. (2009) and Donelson et al. (2017), we ad- indicators of the change in internal control effectiveness where they imply a status
dress this issue by conditioning this behavior on managers' awareness where a firm receives two successive adverse internal control opinions, receives an ad-
about ICW identified by the external auditor, providing further and ro- verse internal control opinion after an unqualified opinion, or receives an unqualified in-
bust insights into how managers respond to the auditors' attestation ternal control opinion after an adverse opinion. These three distinct groups are
introduced relative to a status where a firm has an effective internal control system
on their firms' ICE. (UNQ_UNQ). The two analyses of Panel A and Panel B allow us to answer our question
In Panel A of Table 5, we estimate Eq. (1) and include a dummy var- of whether managers respond to the external auditors' attestation on internal control ef-
iable (ICW_RECUR) that takes a value of 1 when there is an adverse fectiveness. Control variables presented in Table 4 are included. Robust standard errors ad-
audit opinion on the firm's internal controls in two successive years justed for clustering at the firm level. T-statistics in parentheses. Significance level: *** p <
0.01, ** p < 0.05 and * p < 0.1, using two-tailed tests. Variable definitions, measures, and
and zero otherwise. The level analysis fixed effect regressions of aggre-
sources are provided in Appendix A.
gate risk disclosure and its tone on ICW remain consistent with our pre-
vious results in Table 4. Meanwhile, the coefficients on ICW_RECUR,
i.e., recurrence of ICW, indicate that managers of firms that in two suc- Employing Eq. (1), but with the three defined dummy variables instead
cessive years received an adverse opinion from the external auditor on of the ICW variable, we find a significant positive coefficient only on
their internal controls significantly respond by providing high levels of ADV_ADV in terms of its effect on within-firm changes in aggregate
aggregate risk disclosure and its tone of good news and bad news risk disclosure, and its tone of good news and bad news about risk (t-
about risk (t-statistics 2.848, 2.637 and 2.248, significant at the 1%, 1% statistics 2.203, 2.260 and 1.895, significant at the 5%, 5% and 10% levels,
and 5% levels, respectively). respectively).
Panel B of Table 5 shows our within-firm change analysis, where The fact that managers respond to publicly-reported ICW by increas-
changes in the TRD indicate the differences between a firm's score in ing the level of TRD (particularly, bad news about risk) is not ipso facto
year t + 1 and its score in year t. We define four distinct cases of the surprising to us. Kothari, Shu, and Wysocki (2009), consistent with
change in ICW and test their effects on within-firm changes in TRD. agency theory, indicate that managers typically prefer to keep bad
For these cases, we employ three dummy indicators of the change in in- news undisclosed up to a threshold level where it becomes too costly
ternal control effectiveness, where ADV_ADV implies a status of a firm or difficult for managers to withhold. Literature fundamentally links
receiving two successive adverse internal control audit opinions this threshold to incentives that managers face and that affect their will-
(i.e., recurrence of ICW), UNQ_ADV implies a status of a firm receiving ingness to accelerate the disclosure of bad news. For example, Skinner
an adverse internal control audit opinion after an unqualified opinion, (1994) and Bao et al. (2019) illustrate that litigation risk and reputation
and ADV_UNQ implies a status of a firm receiving an unqualified inter- concerns prompt managers to quickly release bad news. Thus, the ad-
nal control audit opinion after an adverse opinion (i.e., remediation of verse public signal conveyed by a credible audit report on a firm's inter-
ICW). These three distinct statuses are introduced relative to a status nal control represents a certain point at which it is seen as too costly and
in which a firm has an effective internal control system (UNQ_UNQ). difficult for managers to not reveal bad news about their firm's risk.

8
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

Consistent with Feng et al.'s (2009) quality finding, our evidence and 4 suggest that the effect of ICW reporting on information asymme-
suggests that whether managers' awareness about their firms' ICW pre- try is not statistically significant, wherein observing a positive sign is ra-
cedes or follows ICW identification by the external auditor, the observed tional. Furthermore, Model 3 also shows that the level of aggregate risk
change (increased quantity) of their TRD is prompted by the identified disclosure negatively and significantly impacts information asymmetry
and publicly-reported ICW. That is, the external auditor's adverse opin- by reducing the bid-ask spread (t-statistic of −2.207, significant at the
ion on a firm's internal controls prompts managers to increase their 5% level).
level of TRD to indicate their grasp of their firm's risks (i.e., bad news When we distinguish good news and bad news about risk
about risk) and their effort to manage it (i.e., good news about risk), (i.e., aggregate risk disclosure tone), shown under Model 4, the results
and so, in turn reduce uncertainties and agency problems. suggest that market participants are more likely to appreciate good
news compared to bad news about risk. While bad news information at-
tracts trivial interest from the market, good news about risk entails a
4.4. Testing H2: The usefulness of ICW reporting and TRD significant positive market liquidity. Specifically, good news about risk
is associated with a lower SPREAD (t-statistic −2.899, significant at
Table 6 presents results related to H2 — whether ICW reporting by the 1% level).
the external auditor and TRD by management are useful to the market. Overall, these results support H2 that: a) firms reporting ICW are
Across regression estimates for investor-perceived risk (SD) and market likely to have higher investor-perceived risk than firms without ICW,
liquidity (SPREAD), results suggest that ICW reporting leads to a signif- and b) the level of contemporaneous TRD (driven by good news about
icant and positive increase in SD (Models 1 and 2; t-statistics of 2.059 risk) is likely to reduce market information asymmetry. To put this in
and 2.076 respectively, significant at the 5% level). However, Models 3 an economic perspective, all else being equal, receiving an adverse
audit opinion under SOX 404(b) is likely to result in an increase in
investor-perceived risk at 9.09% (0.002/0.022) of the mean of SD. A
one-standard-deviation increase in aggregate risk disclosure is associ-
Table 6 ated with a 0.79% (−0.028 * 0.281) lower bid-ask spread; i.e., lower in-
Fixed effect panel regressions of investor-perceived risk and market liquidity on ICW
formation asymmetry and thus, higher market liquidity. Similarly, a
reporting and TRD.
one-standard-deviation increase in good news about risk results in a
Variables Investor-perceived risk (t + 1) Market liquidity (t + 1) 1.20% (−0.087 * 0.138) decrease in SPREAD. These results extend previ-
SD SD SPREAD SPREAD ous research (e.g., Dowdell et al., 2013; Gupta et al., 2018) by providing
Model (1) Model (2) Model (3) Model (4) empirical evidence showing TRD as the conduit by which managers af-
fect information asymmetry around ICW reporting.
ICW 0.002** 0.002** 0.007 0.006
(2.059) (2.076) (0.675) (0.616)
This evidence adds to previous research on the usefulness of ICW re-
AGG_RISK −0.000 −0.028** ported by the external auditor to the capital markets (e.g., Dowdell et al.,
(−0.504) (−2.207) 2013; Gupta et al., 2018). It also accords with the experimental results of
BAD_RISK −0.002 0.025 Lopez et al. (2009) and Church and Schneider (2016) about the value-
(−0.616) (0.510)
relevance of ICW reporting and complements the findings of Clinton
GOOD_RISK 0.001 −0.087***
(0.253) (−2.899) et al. (2014) that illustrate that analysts' coverage declines following
INSIDE_OWN 0.002 0.002 0.039 0.040 ICW reporting. Market liquidity results are in line with the findings of
(0.764) (0.758) (1.098) (1.115) Campbell et al. (2014) about the impact of aggregate risk disclosure
DEBT_EQU −0.000* −0.000* 0.003 0.003 on bid-ask spread. Results about the good news about risk in a firm's
(−1.681) (−1.680) (1.395) (1.436)
AUD_OPIN −0.002*** −0.002*** 0.000 −0.000
10-Ks, which suggest a decrease in the firm's information asymmetry
(−5.247) (−5.227) (0.060) (−0.017) and thus, more active buyers and sellers of the security and tighter
BIG_4 0.003** 0.003** 0.072** 0.071** bid-ask spreads, accord with that of Elshandidy and Shrives (2016).
(2.318) (2.329) (2.368) (2.349) This evidence is also consistent with the theoretical literature
LN_TA −0.004*** −0.004*** −0.096*** −0.094***
(e.g., Diamond & Verrecchia, 1991; Leuz & Verrecchia, 2000) on the in-
(−3.534) (−3.594) (−7.115) (−7.216)
ROE −0.000 −0.000 −0.074*** −0.072*** fluence of disclosure on reducing information asymmetry, thereby
(−0.261) (−0.305) (−4.166) (−4.074) prompting higher confidence in the fairness of stock transactions.
CR −0.000* −0.000* −0.008*** −0.008***
(−1.873) (−1.857) (−2.861) (−2.927)
FFO 0.000 0.000 −0.007 −0.006
4.5. Robustness tests
(0.105) (0.091) (−0.712) (−0.642)
GROWTH 0.000 0.000 −0.005 −0.004
(0.640) (0.615) (−0.589) (−0.503) Prior research (e.g., Elshandidy & Shrives, 2016; Kravet & Muslu,
BETA −0.001* −0.001* 0.011 0.010 2013) proposes a change analysis technique in order to mitigate
(−1.716) (−1.683) (1.574) (1.489) endogeneity concerns related to correlated omitted covariates and re-
TRAD_VOL −0.001** −0.001** −0.042*** −0.042***
verse causality, as well as to establish a strong cause-effect relationship
(−2.085) (−2.094) (−3.456) (−3.441)
BM −0.001 −0.001 0.195*** 0.190*** between explanatory and dependent variables. Accordingly, we define
(−0.711) (−0.642) (6.161) (6.120) changes as the differences between a firm's observed value and the me-
DIVIDENDS 0.001 0.001 0.001 0.001 dian value for other firms in the same industry over the years of study.
(0.420) (0.415) (0.070) (0.055)
We additionally address any potential endogeneity concern by
Constant 0.053*** 0.053*** 0.785*** 0.756***
(6.626) (6.657) (8.317) (8.102)
employing the two-stage least squares (2SLS) statistical technique. Fol-
Observations 3043 3043 3043 3043 lowing prior research (see Elamer et al. (2020) for detailed discussion
R-squared 0.054 0.054 0.159 0.159 and multiple references) each model's control (exogenous) variables
F-value 6.422*** 6.021*** 10.34*** 9.767*** are instrumented in the first stage so as to estimate the predicted
This table reports the coefficients on the explanatory variables of the fixed effects panel re- value of our explanatory variables in the second stage.10 This turns to
gression models. It examines H2 to answer the second research question about whether c in Eq. (1); μc and μc in Eq. (2).
β
ICE attestation by the external auditor (proxied by ICW) and management's TRD are useful 1 1 2
to the market. Robust standard errors adjusted for clustering at the firm level. T-statistics
10
in parentheses. Significance level: *** p < 0.01, ** p < 0.05 and * p < 0.1, using two-tailed Previous studies (e.g., Doyle et al., 2007b) provide further support for the validity of
tests. Variable definitions, measures, and sources are provided in Appendix A. this estimation.

9
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

Table 7 indicators on ICW and TRD). Δ denotes the change with respect to each of the mentioned
Robustness tests. variables. The change (Δ) indicates the differences between a firm's observed value and
the median value for other firms in the same industry over the years. Robust standard er-
Panel A: Fixed effect panel regressions of changes in TRD on ICW rors adjusted for clustering at the firm level. T(Z)-statistics in parentheses. Significance
Variables ΔAGG_RISK ΔBAD_RISK ΔGOOD_RISK level: *** p < 0.01, ** p < 0.05 and * p < 0.1, using two-tailed tests. Variable definitions,
measures, and sources are provided in Appendix A.
Model (1) Model (2) Model (3)

ICW −0.040** −0.008 −0.022***

(−1.971) (−1.097) (−2.595) Table 8
Control variables Included Included Included Fixed effect panel regressions of investor-perceived risk and market liquidity on ICW
Constant −0.515*** 0.016 −0.269*** reporting and TRD.
(−3.346) (0.284) (−4.233)
Observations 3043 3043 3043 Variables Investor-perceived risk Market liquidity (t + 1)
R-squared 0.015 0.016 0.022 (t + 1)
F-value 2.691*** 2.825*** 3.925*** SD SD SPREAD SPREAD

Panel B: Two-stage least squares (2SLS) second stage regressions of TRD on ICW Model (1) Model (2) Model (3) Model (4)

Variables AGG_RISK BAD_RISK GOOD_RISK ICW 0.002** 0.002** 0.006 0.005

(2.038) (2.058) (0.601) (0.540)
Model (1) Model (2) Model (3) AGG_RISK −0.000 −0.028**
ICW −0.032* −0.001 −0.021*** (−0.373) (−2.187)
(−1.806) (−0.113) (−2.823) BAD_RISK −0.002 0.026
Control variables Included Included Included (−0.644) (0.532)
Constant 1.246*** 0.430*** 0.279*** GOOD_RISK 0.001 −0.088***
(28.752) (21.872) (13.491) (0.361) (−2.915)
Observations 3043 3043 3043 SPREAD 0.004 0.004
R-squared 0.009 0.014 0.011 (1.515) (1.535)
Wald-chi2 71.86*** 114.23*** 248*** SD 0.459 0.465
(1.540) (1.560)
Panel C: Fixed effect panel regressions of market indicators on changes in ICW Control variables Included Included Included Included
reporting and TRD Constant 0.049*** 0.050*** 0.760*** 0.731***
(5.922) (6.017) (7.906) (7.720)
Variables Investor-perceived risk (t Market liquidity (t + 1) Observations 3043 3043 3043 3043
+ 1) R-squared 0.055 0.056 0.160 0.161
SD SD SPREAD SPREAD F-value 6.049*** 5.695*** 10.230*** 9.635***

Model (1) Model (2) Model (3) Model (4) This table reports the coefficients on the explanatory variables of the fixed effects panel re-
gression models. It further examines H2 to answer the second research question about
ΔICW 0.002* 0.002* 0.007 0.006 whether ICE attestation by the external auditor (proxied by ICW) and management's
(1.895) (1.913) (0.717) (0.659) TRD are useful to the market, while interchangeably controlling for SPREAD and SD.
ΔAGG_RISK −0.000 −0.028*** Models 1and 2 control for SPREAD and Models 3 and 4 control for SD. Control variables
(−0.464) (−2.719) presented in Table 6 are included. Robust standard errors adjusted for clustering at the
ΔBAD_RISK −0.002 0.025 firm level. T-statistics in parentheses. Significance level: *** p < 0.01, ** p < 0.05 and * p
(−0.618) (0.665) < 0.1, using two-tailed tests. Variable definitions, measures, and sources are provided in
ΔGOOD_RISK 0.001 −0.087*** Appendix A.
(0.264) (−2.621)
Control variables Included Included Included Included
Constant 0.052*** 0.053*** 0.748*** 0.738*** In Table 7, as expected and qualitatively consistent with prior find-
(7.597) (7.622) (10.435) (10.229) ings, results from the change analyses (Panel A) and 2SLS (Panel
Observations 3043 3043 3043 3043 B) suggest that a lower level of TRD is associated with firms with ICW.
R-squared 0.054 0.054 0.159 0.159
Turning to Panels C (change analyses) and D (2SLS) testing the robust-
F-value 7.243*** 6.800*** 24.15*** 22.73***
ness of our market findings, results are also qualitatively similar to that
Panel D: Two-stage least squares (2SLS) second stage regressions of market obtained by previous analyses, suggesting: a) firms reporting ICW are
indicators on ICW reporting and TRD likely to have higher investor-perceived risk than firms without ICW,
Variables Investor-perceived risk (t Market liquidity (t + 1) and b) the level of contemporaneous TRD (driven by good news about
+ 1) risk) is likely to reduce market information asymmetry. Notably, the ob-
SD SD SPREAD SPREAD served increase in some coefficient estimates is in line with prior re-
search suggesting possible stronger estimates from analysis using
Model (1) Model (2) Model (3) Model (4)
instrumental variables (e.g., Elamer et al., 2020).
ICW 0.001* 0.001* 0.012 0.010 Next, we rerun our models using the net tone of risk (obtained by
(1.730) (1.696) (1.330) (1.191)
the adjusted score of good news about risk after deducting the score
AGG_RISK 0.001 −0.019**
(1.279) (−2.130) of bad news about risk). Qualitatively similar inferences to those ob-
BAD_RISK 0.002 0.080*** tained from our previous tests of H1 and H2 are derived from the net
(1.019) (2.878) tone of risk analyses (results unreported for brevity). Furthermore, con-
GOOD_RISK 0.000 −0.096***
sistent with prior research (e.g., Elshandidy & Shrives, 2016; Gupta et al.,
(0.137) (−3.798)
Control variables Included Included Included Included
2018), SD and SPREAD show a relatively large positive correlation of
Constant 0.032*** 0.032*** 0.630*** 0.577*** 31%. Therefore, it could be argued that controlling for these two proxies
(20.255) (21.518) (24.758) (20.829) interchangeably in our market tests is necessary to promote our argu-
Observations 3043 3043 3043 3043 ments. Accordingly, after interchangeably controlling for these two
R-squared 0.006 0.006 0.139 0.140
proxies in our market tests, Table 8 shows qualitatively similar results
Wald-chi2 883*** 881.3*** 1158*** 1096***
to that of our main analyses.11
This table reports the coefficients on the explanatory variables and the summary of the
fixed effects panel regression models (Panels A and C) and two-stage least squares
(2SLS) second stage regression models (Panels B and D). Control variables presented in
Table 4 are included in the models of Panels A and C (regressing TRD on ICW). Control var-
11
iables presented in Table 6 are included in the models of Panels B and D (regressing market We thank the anonymous referee for suggesting this point to us.

10
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

Additionally, we rerun our market main analyses using alternative Declaration of interest
market indicators, namely: a) Amihud's illiquidity ratio measured as
the mean of the daily ratio of absolute value of stock return divided by None.
dollar trading volume, multiplied by ten million; b) Trading volume
measured as the percentage of the mean of the daily trading volume
by the number of outstanding shares. Both market indicators are also Appendix A. Variable definitions
measured (in t + 1) over a 60 trading days period beginning two trad-
ing days after the 10-K filing, using data obtained from CRSP. The
Variable Definition, measures and sources (Data source is Unit
untabulated results are qualitatively consistent with our previous find- Compustat unless otherwise mentioned)
ings from the market analyses. Collectively, our sensitivity tests illus-
Aggregate risk disclosure All risk information that is exhibited in the %
trate that our inferences are robust and not subject to possible (AGG_RISK) narrative sections of the 10-K. The score is the
endogeneity problems. percentage of the number of words indicating risk
in the narrative sections of the10-K divided by the
total number of words in the 10-K. Textual analy-
sis is processed using Diction 7 software
5. Conclusion employing the risk wordlist of Elshandidy and
Shrives (2016) and SEC EDGAR 10-K form filings
This paper explores the impact of internal control effectiveness on accessed through Bill McDonald data repository.
the level of textual risk disclosure (where we define aggregate risk dis- The textual analysis is used to further identify the
tone of aggregate risk disclosure as introduced
closure and its tone of good news and bad news about risk). The paper
below.
also examines the usefulness of ICW reporting and TRD. After control- Bad news about risk All feasible information about risk that exhibits %
ling for a variety of innate firm characteristics that previous research (BAD_RISK) bad news in the narrative sections of the 10-K. The
proposes to be related to incentivizing TRD, and market assessment of score is the percentage of the number of words
risk information conveyed by management and internal control attesta- indicating bad news in the narrative sections of
the10-K divided by the total number of 10-K
tion by external auditors, our study offers four major results.
words.
First, firms with an ineffective internal control system exhibit signif- Good news about risk All feasible information about risk that exhibits %
icantly lower levels of TRD relative to firms with an effective internal (GOOD_RISK) good news in the narrative sections of the 10-K.
control system. Second, consistent with agency theory, recurrently The score is the percentage of the number of
words indicating good news in the narrative sec-
identified and publicly reported ICW, which represents an adverse pub-
tions of the10-K divided by the total number of
lic signal, prompts managers to significantly change their TRD behavior 10-K words.
by providing higher levels of TRD relative to other firms. Third, firms ICW A dummy variable that takes a value of 1 if the 0,1
reporting ineffective internal controls are likely to have higher opinion of the external auditor on the
investor-perceived risk than firms reporting effective internal controls, effectiveness of internal control is adverse (ICW
exists) and 0 otherwise.
but are unlikely to have market liquidity affected. Fourth, TRD increases
Inside ownership Captured by the percentage of closely held shares /
firms' market liquidity, which is specifically associated with good news concentration (i.e., shares owned by firm insiders) divided by
about risk. This implies that despite the negative impact of ICW (INSIDE_OWN) common shares outstanding. Data source is
reporting on investor-perceived risk, it is unlikely to affect the firms' Datastream.
Capital structure Measured by the ratio of total debt to total equity. /
market liquidity because of the positive impact of TRD (particularly its
(DEBT_EQU)
tone of good news about risk) in removing information asymmetry. Beta (BETA) Reflects systematic risk and is measured as CAPM #
Overall, our results suggest the importance of ICE in addition to the use- beta estimated using weekly returns requiring a
fulness of ICW reporting and TRD. minimum of 30 and maximum of 50 observations.
Our results make several contributions to both the internal control Data source is CRSP.
Firm size (LN_TA) Measured as the natural logarithm of total assets. LN#
and risk disclosure literatures and suggest broader implications for
Growth (GROWTH) Measured as the ratio of change in net sales, /
reporting on internal control effectiveness and risk factors than previ- [(salest/salest-1) − 1].
ously documented. However, we acknowledge the following limitations. Profitability (ROE) Measured by the return on equity ratio as net /
First, our market analyses could be affected by contemporaneous news income before extraordinary items dividing by the
year-end common equity.
such as internal control certificates by management under SOX 302
Liquidity (CR) Measured by the current ratio as total current /
and/or SOX 404(a) or risk information released by other outlets of cor- assets dividing by total current liabilities.
porate communication, including conference calls, financial analysts' re- Performance (FFO) Captured by the ratio of net funds from operations /
ports and/or online resources. Second, for purposes of accuracy, to total liabilities.
comparability and data availability, we test SOX 404(b) during the pe- Dividend payout A dummy variable that takes a value of 1 if the 0,1
(DIVIDENDS) firms pays dividends, and 0 otherwise.
riod in which AS2 was in effect. In November 15, 2007, AS5 replaced
Audit quality (BIG_4) A dummy value that equals 1 if the auditor is a big 0,1
AS2, effectively reducing internal control auditing requirements and 4 and 0 otherwise.
thereby potentially negatively impacting both the effectiveness of Auditor opinion A dummy value that equals 1 if the auditor issued 0,1
firms' internal control systems and the precision of the external auditor's (AUD_OPIN) a qualified opinion on financial statements and 0
opinion on the internal control (Schroeder & Shepardson, 2016). Al- otherwise.
Book to market (BM) Represents book to market ratio, measured by /
though this does not limit our internal control inferences, generalizing book value of equity to market value of equity.
them to the less rigorous AS5 regime is a subject for future research. Trading volume Measured by the percentage of the mean of the %
Also, future research can investigate the moderating role of external au- (TRAD_VOL) daily trading volume by the number of
ditor characteristics like expertise and/or internal corporate governance outstanding shares. Data source is CRSP.
Investor-perceived risk The mean of the volatility (standard deviation) of #
mechanisms (e.g., board of directors and managerial incentives) for the
(t + 1) the daily market returns. It is measured over a 60
associations among or between internal control, narrative-related dis- Standard deviation (SD) trading days period beginning two trading days
closures, and market reactions. A fruitful expansion of the present after the 10-K filing. Data source is CRSP.
study would be to investigate whether and how internal control may in- Market liquidity (t + 1) It is the mean of the relative percentage of spread, %
centivize TRD and the informativeness of TRD in the debt market. Bid-ask spread (SPREAD) which is calculated by dividing the difference
between the daily ask and bid prices by the
Inspecting the effects of company-level versus process-level weaknesses
is another potential extension to our study. (continued on next page)

11
M. Elsayed and T. Elshandidy Advances in Accounting 53 (2021) 100531

(continued) Framework issued by the Committee of Sponsoring Organizations of the Treadway

Variable Definition, measures and sources (Data source is
Compustat unless otherwise mentioned)
average of the daily ask and bid prices. It is
measured (in t + 1) over a 60 trading days period
beginning two trading days after the 10-K filing.
Data source is CRSP.
Appendix B. The complete risk wordlist, and examples of risk-
related disclosures and external auditor's opinion
Commission (COSO) and our report expressed an unqualified opinion on
Unit management's assessment of the Company's internal control over financial
reporting, and an adverse opinion on the effectiveness of the Company's internal
control over financial reporting.
Panel A of this appendix presents the risk-related keywords, risk tone classification,
and examples of the extracted TRD (aggregate and tone). Risk-related keywords are
bold italicized. Panel B of this appendix presents an example in which the external
auditor's opinion is adverse, i.e., there are one or more internal control weaknesses,
implying that the company's internal control system is ineffective.
References

Panel A: Risk-related keywords
Aggregate risk
Risk, Risks, Risky, Riskiness, Loss, Losses,
Decline, Declined, Decrease, Decreases,
Decreased, Less, Low, Lower, Fail, Fails,
Failed, Failure, Threat, Reverse, Reversed,
Against, Catastrophe, Catastrophy,
Catastrophic, Shortage, Expose, Exposes,
Exposed, Exposure, Unable, Challenge,
Challenged, Challenges, Uncertain,
Uncertainty, Uncertainties, Gain, Gains,
Chance, Chances, Increase, Increases,
Increased, Peak, Peaked, High, Highest,
Higher, Hedge, Hedging, Diversify,
Diversified, Diversification,
Diversifications, Fluctuate, Fluctuated,
Fluctuation, Fluctuations, Differ, Differed,
Difference, Differences, Different,
Differential, Differentiate, Differentiated,
Differentiation, Probable, Probably,
Probability, Probabilities, Possible,
Possibly, Significant, Significantly,
Significance, Against, Subject, Affect,
Affects, Affected, Affecting, Potential,
Potentially, Depend, Depends, Depended,
Depending, Vary, Varies, Likely, Might,
Influence, Influenced, Susceptible, Viable
Good news about risk
Gain, Gains, Chance, Chances, Increase,
Increases, Increased, Peak, Peaked, High,
Highest, Higher, Hedge, Hedging,
Diversify, Diversified, Diversification,
Diversifications
Bad news about risk
Risk, Risks, Risky, Riskiness, Loss, Losses,
Decline, Declined, Decrease, Decreases,
Decreased, Less, Low, Lower, Fail, Fails,
Failed, Failure, Threat, Reverse, Reversed,
Against, Catastrophe, Catastrophy,
Catastrophic, Shortage, Expose, Exposes,
Exposed, Exposure, Unable, Challenge,
Challenged, Challenges, Uncertain,
Uncertainty, Uncertainties
Panel B: Example of external auditor's adverse opinion on the firm's internal
controls
We have audited, in accordance with the standards of the Public Company Accounting
Oversight Board (United States), the effectiveness of the Company's internal control
over financial reporting, based on criteria established in Internal Control-Integrated
Examples
0000003673
The commodity price risk exposure
results from market fluctuations in…
The operation of power generation
facilities involves many risks, including
the risk of breakdown or failure of
equipment, fuel interruption and
performance below expected levels of
output or efficiency.
Management is unable to provide
assurance that the ultimate cost of
decommissioning the Cook Plant will not
be significantly different than current
projections.
0000003673
The sale resulted in a gain of $94.8
million…
AE Supply entered into two treasury lock
agreements to hedge its exposure to
changing United States Treasury interest
rates on the forecasted issuance of
long-term, fixed-rate debt in April 2002.
Retail electric revenue increased due to
higher kWh sales resulting from increases
in the average number of customers
served and customer usage.
0000003673
International operations are subject to
certain additional risks inherent in
conducting business outside the United
States, including…
Market liquidity has significantly declined
over the past three years.
Among other things, significant price
volatility… and overall declines in
electricity demand and in the economy,
generally, have contributed to this
slowdown.
Catastrophic events may exceed reserves
or insurance, if any, for repairs, which may
adversely impact Allegheny's results of
operations and financial condition.
The suit alleges that the Board and senior
management breached fiduciary duties to
AE that have exposed AE to the securities
class action lawsuits.
Abraham, S., & Cox, P. (2007). Analysing the determinants of narrative risk information in
UK FTSE 100 annual reports. The British Accounting Review, 39(3), 227–248.
Abraham, S., & Shrives, P. J. (2014). Improving the relevance of risk factor disclosure in
corporate annual reports. The British Accounting Review, 46(1), 91–107.
American Institute of Certified Public Accountants (AICPA) (1987). Report on the task force
on risks and uncertainties. New York.
Ashbaugh-Skaife, H., Collins, D., LaFond, R., & Kinney, W. (2008). The effect of internal
control deficiencies and their remediation on accrual quality. The Accounting
Review, 83(1), 217–250.
Ashbaugh-Skaife, H., Collins, D. W., & Lafond, R. (2009). The effect of SOX internal control
deficiencies on firm risk and cost of equity. Journal of Accounting Research, 47(1),
1–43.
Bagehot, W. (1971). The only game in town. Financial Analysts Journal, 27(2), 12–14.
Bao, D., Kim, Y., Mian, G. M., & Su, L. (2019). Do managers disclose or withhold bad news?
Evidence from short interest. The Accounting Review, 94(3), 1–26.
Bao, Y., & Datta, A. (2014). Simultaneously discovering and quantifying risk types from
textual risk disclosures. Management Science, 60(6), 1371–1391.
Beneish, M. D., Billings, M. B., & Hodder, L. D. (2008). Internal control weaknesses and in-
formation uncertainty. The Accounting Review, 83(3), 665–703.
Bertomeu, J., Beyer, A., & Dye, R. A. (2011). Capital structure, cost of capital, and voluntary
disclosures. The Accounting Review, 86(3), 857–886.
Beyer, A., Cohen, D. A., Lys, T. Z., & Walther, B. R. (2010). The financial reporting environ-
ment: Review of the recent literature. Journal of Accounting and Economics, 50(2),
296–343.
Brennan, M. J., & Subrahmanyam, A. (1996). Market microstructure and asset pricing: On
the compensation for illiquidity in stock returns. Journal of Financial Economics, 41(3),
441–464.
Buslepp, W., Legoria, J., Rosa, R., & Shaw, D. (2019). Misclassification of audit-related fees
as a measure of internal control quality. Advances in Accounting, 46, 100425.
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. Review of
Accounting Studies, 19(1), 396–455.
Chan, K. C., Farrell, B., & Lee, P. (2008). Earnings management of firms reporting material
internal control weaknesses under section 404 of the Sarbanes-Oxley act. Auditing: A
Journal of Practice & Theory, 27(2), 161–179.
Chasan, E. (2013). SEC official sounds alarm on decline in “material weaknesses.”. Wall
Street Journal (December 5). Available at: http://on.wsj.com/1kgdrTQ.
Chen, C., Martin, X., Roychowdhury, S., Wang, X., & Billett, M. T. (2017). Clarity begins at
home: Internal information asymmetry and external communication quality. The
Accounting Review, 93(1), 71–101.
Chen, Y., Eshleman, J. D., & Soileau, J. S. (2016). Board gender diversity and internal con-
trol weaknesses. Advances in Accounting, 33, 11–19.
Cheng, M., Dhaliwal, D., & Zhang, Y. (2013). Does investment efficiency improve after the
disclosure of material weaknesses in internal control over financial reporting? Journal
of Accounting and Economics, 56(1), 1–18.
Church, B. K., & Schneider, A. (2016). The impact of section 302 and 404 (b) internal con-
trol disclosures on prospective investors’ judgments and decisions: An experimental
study. International Journal of Auditing, 20(2), 175–185.
Clinton, S. B., Pinello, A. S., & Skaife, H. A. (2014). The implications of ineffective internal
control and SOX 404 reporting for financial analysts. Journal of Accounting and
Public Policy, 33(4), 303–327.
Dedman, E. (2016). CEO succession in the UK: An analysis of the effect of censuring the
CEO-to-chair move in the combined code on corporate governance 2003. The
British Accounting Review, 48(3), 359–378.
Deumes, R., & Knechel, W. R. (2008). Economic incentives for voluntary reporting on in-
ternal risk management and control systems. Auditing: A Journal of Practice & Theory,
27(1), 35–66.
Diamond, D. W., & Verrecchia, R. E. (1991). Disclosure, liquidity, and the cost of capital.
The Journal of Finance, 46(4), 1325–1359.
Donelson, D. C., Ege, M. S., & McInnis, J. M. (2017). Internal control weaknesses and finan-
cial reporting fraud. Auditing: A Journal of Practice & Theory, 36(3), 45–69.
Dowdell, T. D., Kim, J. C., Klamm, B. K., & Watson, M. W. (2013). Internal control reporting
and market liquidity. Research in Accounting Regulation, 25(1), 30–40.
Doyle, J., Ge, W., & McVay, S. (2007b). Determinants of weaknesses in internal control
over financial reporting. Journal of Accounting and Economics, 44(1–2), 193–223.
Doyle, J. T., Ge, W., & McVay, S. (2007a). Accruals quality and internal control over finan-
cial reporting. The Accounting Review, 82(5), 1141–1170.
Dyer, T., Lang, M., & Stice-Lawrence, L. (2017). The evolution of 10-K textual disclosure:
Evidence from latent Dirichlet allocation. Journal of Accounting and Economics, 64
(2–3), 221–245.

12
M. Elsayed and T. Elshandidy
Elamer, A. A., Ntim, C. G., Abdou, H. A., Owusu, A., Elmagrhi, M., & Ibrahim, A. E. A. (2020).
Are bank risk disclosures informative? Evidence from debt markets. International
Journal of Finance and Economics. https://doi.org/10.1002/ijfe.1849 In press.
El-Mahdy, D. F., & Park, M. S. (2014). Internal control quality and information asymmetry
in the secondary loan market. Review of Quantitative Finance and Accounting, 43(4),
683–720.
Elshandidy, T., & Neri, L. (2015). Corporate governance, risk reporting practices, and mar-
ket liquidity: Comparative evidence from the UK and Italy. Corporate Governance: An
International Review, 23(4), 331–356.
Elshandidy, T., Shrives, P., Bamber, M., & Abraham, S. (2018). Risk reporting: A review of
the literature and implications for future research. Journal of Accounting Literature, 40,
54–82.
Elshandidy, T., & Shrives, P. J. (2016). Environmental incentives for and usefulness of tex-
tual risk reporting: Evidence from Germany. The International Journal of Accounting,
51(4), 464–486.
Feng, M., Li, C., & McVay, S. (2009). Internal control and management guidance. Journal of
Accounting and Economics, 48(2), 190–209.
Garfinkel, J. A. (2009). Measuring investors’ opinion divergence. Journal of Accounting
Research, 47(5), 1317–1348.
Gupta, P. P., Sami, H., & Zhou, H. (2018). Do companies with effective internal controls
over financial reporting benefit from Sarbanes–Oxley sections 302 and 404? Journal
of Accounting, Auditing and Finance, 33(2), 200–227.
Hope, O. K., Hu, D., & Lu, H. (2016). The benefits of specific risk-factor disclosures. Review
of Accounting Studies, 21(4), 1005–1045.
Iliev, P. (2010). The effect of SOX section 404: Costs, earnings quality, and stock prices. The
Journal of Finance, 65(3), 1163–1196.
Kaplan, R. S. (2011). Accounting scholarship that advances professional knowledge and
practice. The Accounting Review, 86(2), 367–383.
Kim, Y., & Park, M. S. (2009). Market uncertainty and disclosure of internal control defi-
ciencies under the Sarbanes–Oxley act. Journal of Accounting and Public Policy, 28
(5), 419–445.
Kothari, S. P., Shu, S., & Wysocki, P. D. (2009). Do managers withhold bad news? Journal of
Accounting Research, 47(1), 241–276.
Kravet, T., & Muslu, V. (2013). Textual risk disclosures and investors’ risk perceptions.
Review of Accounting Studies, 18(4), 1088–1122.
Leuz, C., & Verrecchia, R. (2000). The economic consequences of increased disclosure.
Journal of Accounting Research, 38, 91–124.
Linsley, P., & Shrives, P. (2006). Risk reporting: A study of risk disclosure in the annual re-
ports of UK companies. The British Accounting Review, 38(4), 387–404.
Lo, K., Ramos, F., & Rogo, R. (2017). Earnings management and annual report readability.
Journal of Accounting and Economics, 63(1), 1–25.
Lopez, T. J., Vandervelde, S. D., & Wu, Y. J. (2009). Investor perceptions of an auditor’s ad-
verse internal control opinion. Journal of Accounting and Public Policy, 28(3), 231–250.
Loughran, T., & McDonald, B. (2016). Textual analysis in accounting and finance: A survey.
Journal of Accounting Research, 54(4), 1187–1230.
Public Company Accounting Oversight Board (PCAOB) (2004). An audit of internal control
over financial reporting performed in conjunction with an audit of financial statements.
Auditing Standard No. 2. Washington, D.C: PCAOB.
13
Advances in Accounting 53 (2021) 100531
Public Company Accounting Oversight Board (PCAOB) (2007). An audit of internal control
over financial reporting that is integrated with an audit of financial statements. Auditing
standard no.5. Washington, D.C.: PCAOB.
Public Company Accounting Oversight Board (PCAOB) (2013). Considerations for audits of
internal control over financial reporting. Staff practice alert no. 11. (October 24).
Washington, DC: PCAOB.
Rice, S. C., & Weber, D. P. (2012). How effective is internal control reporting under SOX
404? Determinants of the (non-) disclosure of existing material weaknesses.
Journal of Accounting Research, 50(3), 811–843.
Rose-Green, E., Huang, H. W., & Lee, C. C. (2011). The association between auditor indus-
try specialization and firms’ disclosure of internal control weaknesses. International
Journal of Auditing, 15(2), 204–216.
Schleicher, T., & Walker, M. (2010). Bias in the tone of forward-looking narratives.
Accounting and Business Research, 40(4), 371–390.
Schneider, A., Gramling, A. A., Hermanson, D. R., & Ye, Z. S. (2009). A review of academic
literature on internal control reporting under SOX. Journal of Accounting Literature, 28,
1–46.
Schrand, C., & Elliott, J. (1998). Risk and financial reporting: A summary of the discussion
at the 1997 AAA/FASB conference. Accounting Horizons, 12(3), 271–282.
Schroeder, J. H., & Shepardson, M. L. (2016). Do SOX 404 control audits and management
assessments improve overall internal control system quality? The Accounting Review,
91(5), 1513–1541.
Securities and Exchange Commission (SEC) (1997). Disclosure of accounting policies for de-
rivative financial instruments and derivative commodity instruments and disclosure of
qualitative and quantitative information about market risk inherent in derivative finan-
cial instruments, other financial instruments, and derivative commodity instruments.
Washington, D.C.
Securities and Exchange Commission (SEC) (2003). Financial reporting release no. 72.
Commission guidance regarding management’s discussion and analysis for financial con-
dition and results of operations. Securities act release no. 33–8350. Washington, D.C.
Securities and Exchange Commission (SEC) (2005). Revisions to accelerated filer definition
and accelerated deadlines for filing periodic reports. Release no. 33–8644. Washington, D.
C.
Securities and Exchange Commission (SEC) (2009). Remarks before the 2009 AICPA na-
tional conference on current SEC and PCAOB developments (Besch, D.) (December 7).
Washington, D.C.
Securities and Exchange Commission (SEC) (2010). 17 CFR parts 211, 231 and 241 com-
mission guidance regarding disclosure related to climate change. Washington, DC.
Shalen, C. T. (1993). Volume, volatility, and the dispersion of beliefs. The Review of
Financial Studies, 6(2), 405–434.
Skinner, D. J. (1994). Why firms voluntarily disclose bad news. Journal of Accounting
Research, 32(1), 38–60.
Yekini, L. S., Wisniewski, T. P., & Millo, Y. (2016). Market reaction to the positiveness of
annual report narratives. The British Accounting Review, 48(4), 415–430.