Accounting, Organizations and Society 62 (2017) 43e64

Contents lists available at ScienceDirect

Accounting, Organizations and Society

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

The economic consequences associated with integrated report quality:

Capital market and real effects
Mary E. Barth a, Steven F. Cahan b, Li Chen b, Elmar R. Venter c, *
a
Graduate School of Business, Stanford University, USA
b
University of Auckland, Auckland, New Zealand
c
University of Pretoria, Pretoria, South Africa

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

Article history: The International Integrated Reporting Council's Framework identifies two goals for integrated reporting:
Received 4 December 2015 improved information for outside providers of financial capital and better internal decision making. We
Received in revised form extend prior research that finds a positive association between integrated report quality (IRQ) and firm
23 August 2017
value by examining two channels through which this association may ariseda capital market channel
Accepted 24 August 2017
and a real effects channel. To conduct these tests, we disaggregate firm value into three components:
liquidity, cost of capital, and expected future cash flows. Using data from South Africa where integrated
reporting is mandatory and an IRQ measure based on proprietary EY data, we find a positive association
Keywords:
Integrated reporting
between IRQ and liquidity, which supports the capital market channel. We find no evidence of a relation
Corporate social responsibility between IRQ and cost of capital. We also find a positive association between IRQ and expected future
Firm value cash flows. Because this association could reflect better investor cash flow forecastsda capital market
Cost of capital effect, better internal decisionsda real effect, or both, we attempt to distinguish these explanations. We
Expected future cash flows find higher IRQ is (not) associated with higher realized future operating cash flows (greater analyst target
Liquidity price forecast accuracy), and find higher IRQ is associated with higher investment efficiency. These
Investment efficiency findings support the real effects channel. Together, our findings are consistent with integrated reporting
South Africa
achieving its dual objective of improved external information and better internal decisions.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction integrated reporting has the dual objective of providing informa-

tion to external and internal decision makers. We empirically
The question we address is how integrated report quality, IRQ, is evaluate the extent to which, and how, integrated reporting ach-
associated with firm value. In particular, we identify two channels ieves this dual objective.
through which IRQ may be associated with firm valueda capital Since the release of the IIRC Framework in 2013, worldwide in-
market channel that relates to improved information for outside terest in integrated reporting continues to grow. According to the
providers of capital and a real effects channel that relates to IIRC, more than 1500 firms currently prepare an integrated report,
improved internal decision making. Our motivation for focusing on including AES Brasil, Clorox, GE, Mitsubishi, Novo Nordisk, Tata
these effects is that the International Integrated Reporting Council's Steel, and Vivendi. There also has been strong support for integrated
(IIRC) Framework identifies two aims for integrated reporting: (1) reporting from the large accounting firms as well as standard-
improving the quality of information available to outside providers setters and professional accounting bodies in various countries.
of financial capital to enable more efficient capital allocation and Among regulators, the IIRC Framework recently was endorsed by
(2) supporting integrated internal thinking, decision-making, and China's Ministry of Finance, and the Securities and Exchange Board
actions that focus on value creation for the firm. Thus, in contrast to of India (SEBI) is encouraging the top 500 listed firms in India to
accounting standards or other types of disclosure regulation, e.g., adopt voluntarily integrated reporting starting with the 2017e2018
the U.S. Securities and Exchange Commission's (SEC) Form 10-K, financial year. Further, the IIRC has started to engage with other
standard-setters, including the International Accounting Standards
Board (IASB) and the U.S. Financial Accounting Standards Board
(FASB), as part of the Corporate Reporting Dialogue to consider
* Corresponding author. University of Pretoria, Department of Accounting,
Lynnwood Road, Hillcrest, Pretoria, 0002, South Africa. alternative disclosure models, and Richard Howitt, the IIRC Chief
E-mail address: elmar.venter@up.ac.za (E.R. Venter). Executive Officer, has identified 2018 as the start of a “global

http://dx.doi.org/10.1016/j.aos.2017.08.005
0361-3682/© 2017 Elsevier Ltd. All rights reserved.
44 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

adoption phase,” the goal of which is to make integrated reporting performance, accounting quality, firm complexity, overall disclo-
the norm (Howitt, 2017). Against this backdrop, empirical evidence sure quality, and other factors. We also find that firms with larger
on the benefits associated with integrated reporting is sparse. annual increases in IRQ have larger decreases in bid-ask spreads.
As a first step in our evaluation, we determine whether the Relating to expected future cash flows, we find a significant positive
finding in prior research of a positive association between IRQ and association between annual changes in IRQ and changes in analyst
firm value holds for our sample. We then disaggregate firm value estimates of future share prices, i.e., target prices. The relation also
into three componentsdliquidity, cost of capital, and expected is positive for levels of expected cash flows, but not significantly so.
future cash flowsdand determine whether IRQ is associated with Relating to cost of capital, we find no evidence of a relation between
greater liquidity, lower cost of capital, and higher expected future IRQ and the average of four cost of capital proxies commonly used
cash flows. We conduct the analyses of each component using in the literature. Taken together, these findings indicate that IRQ is
levels and changes specifications. Relating to the two channels associated with firm value through increased liquidity and ex-
through which IRQ can affect firm value, we view liquidity and cost pected future cash flows.
of capital as capital market effects. However, a positive relation To provide additional insights into which features of integrated
between IRQ and expected future cash flows could be attributable reports are associated with firm value, we map the proprietary dis-
to capital market participants being able to estimate future cash aggregated EY quality data into 12 categories that reflect the guiding
flows more accurately because they have better informationda principles and content elements in the IIRC's integrated reporting
capital market effectdor managers making better decisions that framework. We find that connectivity, stakeholder relationships,
generate higher future cash flowsda real effectdor both. Thus, we materiality, and conciseness are the most important drivers of our
determine whether analysts' target price forecast accuracy is higher findings for firm value and for its liquidity and expected cash flow
for firms with higher IRQ, which would support the capital market components. The importance of connectivity is particularly pertinent
interpretation, and whether realized future operating cash flow is because connectivity is closely linked to integrated thinking, which is
higher for firms with higher IRQ, which would support the real key to achieving the dual objective of integrated reporting.
effect interpretation of the relation between IRQ and expected Results from our analyses relating to analyst cash flow forecast
future cash flow. We provide additional evidence on the real effect accuracy and realized future operating cash flows indicate that our
channel by determining whether IRQ is associated with firms’ in- finding of a positive relation between IRQ and expected future cash
vestment efficiency by examining a specific type of internal deci- flows is driven by real effects rather than capital market effects.
sion, i.e., investment decisions. First, we do not find an association between IRQ and analysts’ target
Our data come from South Africa where integrated reporting is a price forecast accuracy, which suggests that higher IRQ is not
requirement of the Johannesburg Stock Exchange (JSE). The advan- associated with more accurate cash flow forecasting. Second, we
tage of the South African setting is that we avoid concerns about find a significantly positive association between IRQ and ex post
self-selection that arise when issuance of an integrated report is operating cash flows, which is consistent with improved decision
voluntary. A distinguishing feature of our study is that we obtain making by managers. We also find a significantly positive associa-
proprietary data from EY, which rates the quality of integrated re- tion between IRQ and investment efficiency, which also supports
ports of the top 100 JSE firms each year in terms of market value of the real effects channel. Our investment efficiency evidence is
equity, to measure IRQ. We have access to each firm's report quality consistent with integrated reporting facilitating integrated thinking
category, which is released publicly, and the underlying scores for whereby managers of firms with higher IRQ recognize in-
each quality dimension, which are not. According to the chair of the terdependencies between various types of capital and parts of the
EY panel that rates the reports, the ratings focus on the quality of the firm, which enables them to make better investment choices.
disclosure, specifically whether the integrated report gives readers a Our study contributes to the literature and practice in several
sense of the firm's strategy and value creation process. Thus, our ways. First, we add to the limited empirical research on integrated
measure of IRQ is not simply a disclosure index that captures the reporting by extending Lee and Yeo (2016) and Zhou, Simnett, and
presence or absence of particular items. In addition to the inde- Green (2017), which also focus on South African firms. Whereas Lee
pendent nature of the rating and its focus on quality, there are two and Yeo (2016) examine whether IRQ is associated with firm value,
benefits of using the EY ratings: the EY adjudicators are experts and we investigate the channels by which this association occurs.
the detailed EY score sheet and data allow us to disaggregate the IRQ Whereas Zhou et al. (2017) investigate two capital market effects of
score into its components based on the IIRC Framework.1 integrated reports, i.e., analyst forecast properties and cost of cap-
We find a positive relation between IRQ and firm value, ital, they do not link these effects to firm value and do not consider
measured by Tobin's Q, which is consistent with the findings in Lee real effects, which we do. Second, we answer Leuz and Wysocki’s
and Yeo (2016). We also find that firms with larger annual increases (2016, 530) call for research on disclosure in “novel settings” in
in IRQ have larger increases in firm value. Relating to liquidity, as countries outside the U.S. and to examine “nontraditional disclo-
predicted, we find that IRQ is negatively associated with bid-ask sure” and “the real effects of disclosure mandates.” Third, we
spread, an inverse measure of stock liquidity, after controlling for extend the literature on implications of extra-financial information
corporate governance, Corporate Social Responsibility (CSR) (Dhaliwal, Li, Tsang, & Yang, 2011; Dhaliwal, Radhakrishnan, Tsang,
& Yang, 2012; Lu, Shailer, & Yu, 2017; Plumlee, Brown, Hayes, &
Marshall, 2015). The relations we document include controls for
the issuance of a standalone CSR report as well as accounting
1
Regarding the first benefit, the three adjudicators who evaluate the integrated quality and overall disclosure quality. Thus, our evidence shows
reports on behalf of EY are professors at the University of Cape Town with extensive
that integrated reports are associated with benefits incremental to
experience in reading and evaluating firms' corporate reports. Together they have
more than 50 years of such experience. One adjudicator serves on the board of those associated with existing reports, such as standalone CSR re-
directors of the Global Reporting Initiative and was a member of the Integrated ports. Fourth, we provide evidence for managers, practitioners,
Reporting Working Group of the South African Integrated Reporting Committee, standard-setters, regulators, and investors as they evaluate the
and another offers an elective module in integrated reporting at the University of merits of integrated reporting. For example, the SEC issued a
Cape Town Business School. Regarding the second benefit, we are among the first to
provide evidence on the association between the separate components of IRQ and
Concept Release on proposed changes to business and financial
economic consequences. These benefits come at the expense of a sample restricted disclosures required by Regulation S-K and asked, “How important
to the top 100 JSE firms. to investors is integrated reporting, as opposed to separate financial
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 45

and sustainability reporting?” (SEC, 2016, 214). firms focus on long-term value creation. They contend that inte-
The rest of the paper proceeds as follows. Section 2 provides grated reports allow investors to make better resource allocation
background, discusses theory, and develops hypotheses. Regarding decisions by providing a more comprehensive view of the firm, and
firm value and its components, Section 3 explains the research call for stock exchanges and securities regulators to mandate in-
design relating to liquidity, cost of capital, and expected future cash tegrated reporting to “ensure swift and broad adoption” (Gore &
flows and Section 4 presents the findings. Section 5 explains the Blood, 2011). To date, only one countrydSouth Africadmandates
research design and associated findings relating to cash flow real integrated reporting.
and capital market effects, and Section 6 does so for the investment In South Africa, integrated reporting pre-dates the establish-
efficiency real effect. Section 7 contains additional analyses. Section ment of the IIRC and release of its Framework. To address concerns
8 concludes. about ineffective management in the post-apartheid era, in 1993
the South African Institute of Directors commissioned the King
2. Background, theory, and hypotheses Committee with the mandate to promote the highest standards of
corporate governance in South Africa (West, 2006). Hence, South
2.1. Background Africa's history of apartheid has been influential in shaping social
and environmental governance (De Villiers, Rinaldi, & Unerman,
The IIRC was created in 2010 in the wake of the global financial 2014). Fueled by the belief of the King Committee that conven-
crisis. Supported by the Accounting for Sustainability (A4S) initia- tional financial reporting was no longer meeting the needs of firms'
tive led by HRH The Prince of Wales and the Global Reporting stakeholders, a key recommendation of the third version of the
Initiative (GRI), the IIRC's mandate is to develop a framework for King Code (King III) is that firms be required to issue an integrated
integrated reporting and to promote its use. Issued in 2013, the IIRC report to present the firm's performance in terms of its finance and
Framework states that the primary purpose of an integrated report its sustainability (King, 2013).
is to explain to providers of financial capital how an organization Because the King Code is a JSE listing requirement, the release of
creates value over time. The IIRC Framework is principles-based King III meant that primary listed firms were required to issue an
and does not provide a standard format for integrated reports. integrated report for annual periods ending on or after 1 March
Instead, the framework sets out seven guiding principles and eight 2010, or explain why they did not issue one. This apply-or-explain
content elements for an integrated report. These principles and regulation raises the question of whether integrated reports are
elements relate to six capitals that the firm uses to create value, truly mandatory. However, the apply-or-explain approach recog-
which enables the firm to explain how it creates value.2 nizes that often the issue is not whether to comply, but rather how
Proponents of integrated reports contend that such reports to apply the principles and recommendations (IoD, 2009). Disclo-
address deficiencies of traditional corporate reporting (EY, 2014a). sure regarding how the firm complies with integrated reporting is
Integrated reports are broader in scope in that they include finan- mandatory, but the firm can comply in different ways or apply an
cial and non-financial information. For example, whereas financial equivalent practice.3
statements provide coverage of financial capital, they only partially In May 2010, the Integrated Reporting Committee of South Africa
address manufactured, intellectual, human, social, and natural (IRC) was established to develop guidelines on high quality inte-
capitals (IIRC, 2016a). Thus, the intent of integrated reports is to grated reporting practices. The IRC's discussion paper on a frame-
address all resources and relationships that materially impact the work for an integrated report, which was released on 25 January
value-creation activities of the firm in the short, medium, and long 2011, was the first national initiative on integrated reporting. In
terms (Cohen & Simnett, 2015). March 2014, the IRC endorsed the IIRC Framework for South African
A key feature of integrated reports that distinguishes integrated firms and ceased issuing its own guidance. See De Villiers et al. (2014)
reporting from other reporting formats is the concept of connec- for further background on the development of integrated reporting.
tivity, which means that an “integrated report should show a holistic Outside South Africa, integrated reporting continues to gain
picture of the combination, interrelatedness and dependencies be- momentum, albeit primarily as a voluntary reporting option. For
tween the factors that affect the organization's ability to create value example, recently, China's Ministry of Finance endorsed the IIRC
over time” (IIRC, 2013, 5). Connectivity is related to integrated Framework (Howitt, 2016), India's SEBI asked the top 500 listed
thinking and, according to the IIRC (2016a, 4), integrated thinking is firms in India to adopt integrated reporting (Majmudar & Rana,
an “integral component of integrated reporting.” The IIRC (2016a, 8) 2017), and a new corporate governance code issued by Securities
states that by “breaking down internal silos and increasing the Commission Malaysia encourages Malaysian firms to adopt inte-
shared understanding of how value is created,” integrated thinking grated reporting (Kana, 2017). Further, the U.S. Securities and Ex-
can lead to improved decisions at the management and board levels change Commission (SEC) issued a Concept Release on proposed
as well as more focused reporting and better communication with changes to business and financial disclosures required by Regula-
the firm's stakeholders. Thus, embedded in the IIRC Framework is tion S-K and asked, “How important to investors is integrated
the dual objective of improving information for external users and reporting, as opposed to separate financial and sustainability
improving internal decision making. reporting?” (SEC, 2016, 214). The European Union's recent non-
In a 2011 Wall Street Journal article, former U.S. Vice President Al financial reporting directive is seen as a “stepping stone” to inte-
Gore and David Blood identified mandating integrated reporting as grated reporting (Howitt, 2016).
one of five steps needed to support sustainable capitalism whereby In addition, the IIRC collaborates with other standard-setters to
ensure that integrated reporting plays a role in improving corporate

2
The seven guiding principles are: strategic focus and future orientation, con-
3
nectivity of information, stakeholder relationships, materiality, conciseness, reli- We hand collect data on firms' compliance for all primary listed firms on the JSE
ability and completeness, and consistency and comparability. The eight content in 2014 and find that 97.7% of the firms declare that their report is integrated. Of the
elements are: organizational overview and external environment, governance, six firms that did not so declare, four provided reasons consistent with the apply-
business model, risks and opportunities, strategy and resource allocation, perfor- or-explain principle. Hence, the effective compliance rate is 99.2%. Such a high level
mance, outlook, and basis of preparation. The six capitals are: financial, of compliance is inconsistent with compliance being voluntary. Other researchers
manufacturing, human, intellectual, social and relationship, and natural. See IIRC also classify South Africa's integrated reporting regime as mandatory (e.g., De
(2013). Villiers et al., 2014; Eccles & Krzus, 2015; Serafeim, 2015).
46 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

reporting. For example, in 2014, the Sustainability Accounting market participants, and a real effects channel, which relates to the
Standards Board (SASB) and the IIRC reached a memorandum of quality of the decisions made by managers. Thus, in contrast to Lee
understanding to accelerate the practical implementation of inte- and Yeo (2016) and Zhou et al. (2017), our objective is to determine
grated reporting (SASB, 2014). In April 2017, the Corporate Lead- whether integrated reporting achieves the IIRC's dual objective of
ership Group on Integrated Reporting, a partnership between the enhancing information quality and improving internal decision
IIRC and GRI discussed how GRI standards can be incorporated into making.5
the integrated reporting process (IIRC, 2017). The IIRC also has
joined with other standard-setters including the IASB and FASB to
explore alternative reporting models as part of the Corporate 2.3. Theory and hypotheses
Reporting Dialogue, and in an address to the IIRC, Hans Hoo-
gervorst, chair of the IASB, acknowledged that users need infor- Prior research studies voluntary and mandatory disclosure.
mation, in addition to general purpose financial reports, to make Mandatory disclosure studies can be categorized into those
their decisions (IASB, 2017).4 examining the effects of individual accounting standards and those
examining the adoption of a complete set of accounting standards
such as International Financial Reporting Standards (IFRS) (e.g.,
2.2. Related research
Leuz & Wysocki, 2016). We study mandatory disclosure, but instead
focus on the mandatory adoption of a new reporting framework,
De Villiers et al. (2014) and Velte and Stawinoga (2017) provide
i.e., integrated reporting.
overviews of the emerging literature on integrated reporting. Most
Although IFRS and integrated reporting both seek to improve
studies on integrated reporting are qualitative. Given that our study
information available to investors, there are significant differences.
is quantitative, we limit our discussion to archival studies that use
First, the IFRS Framework is based on principles, but the standards
South African data.
specify how to account for and disclose information about partic-
Baboukardos and Rimmel (2016) and Bernardi and Stark (2017)
ular types of transactions. In contrast, integrated reporting is based
investigate the adoption of mandatory integrated reporting in
solely on the IIRC Framework and is entirely principles based.
South Africa. Baboukardos and Rimmel (2016) find an increase in the
Second, IFRS applies only to financial information whereas inte-
value relevance of earnings after integrated reporting was mandated
grated reporting combines financial and non-financial information
by the JSE, but a decrease in the value relevance of equity book value.
into a single report with a focus on value creation. Third, IFRS
Bernardi and Stark (2017) find a significantly positive association
specifies the content of financial reports, but does not discuss how
between Bloomberg's environmental, social, and governance (ESG)
the reports are communicated to investors. Communication with
scores and analyst earnings forecast accuracy for 40 South African
investors and stakeholders is fundamental to integrated reporting.
firms in the period after integrated reporting became mandatory,
For example, the IIRC Framework states: “The primary purpose of
but not in the pre-mandatory period. Bernardi and Stark (2017)
an integrated report is to explain to providers of financial capital
conclude that the mandatory presentation of integrated reports
how an organization creates value over time” (IIRC, 2013, 4). Fourth,
enhances analysts' understanding of the ESG data. Whereas these
although IFRS may indirectly affect firms' internal decisions by
studies focus on the mandatory integrated reporting adoption event,
increasing monitoring by investors (e.g., Biddle, Callahan, Hong, &
we examine the effects of cross-sectional differences in IRQ after
Knowles, 2015), IFRS is silent on how it may affect managers’ in-
integrated reporting became mandatory in South Africa.
ternal decision making. In contrast, the IIRC (2013, 2016a) expects
Lee and Yeo (2016) examine the association between Tobin's Q
internal decision making to improve as a result of integrated
and a self-constructed proxy for IRQ for a sample of 822 South
thinking. Thus, an examination of mandated integrated reporting is
African firm-year observations from 2010 to 2013. The proxy is a
distinct from that of mandatory IFRS adoption and, given the
rating score based on an assessment of five aspects of each of the
growing interest in integrated reporting around the world, is a topic
eight IIRC Framework content elements. Lee and Yeo (2016) find
worthy of additional research (e.g., De Villiers et al., 2014).
that IRQ is positively associated with Tobin's Q and the association
Our study also is related to the literature on the implications of
is more pronounced for complex firms and firms with external
extra-financial information (e.g., Dhaliwal et al., 2011; Dhaliwal, Li,
financing needs. Using data from 2009 to 2012, Zhou et al. (2017)
Tsang, & Yang, 2014; Lu et al., 2017; Plumlee et al., 2015). Whereas
also develop a self-constructed proxy for IRQ that captures 31
prior studies focus on voluntary CSR reporting, integrated reporting
disclosure components across the eight content elements. Zhou
is mandatory in our setting. Studies examining voluntary reporting
et al. (2017) find a negative association between changes in IRQ
likely suffer from selection bias. Managers who voluntarily disclose
and subsequent changes in analyst earnings forecast errors, but not
information effectively reveal that the benefits of the disclosure
changes in forecast dispersion. They also find a negative association
exceed the cost. Empirical tests of the association between volun-
between changes in IRQ and subsequent changes in cost of equity
tary disclosure quality and economic consequences likely docu-
capital, but only for firms with low analyst following.
ment the different incentives of disclosing and non-disclosing
Lee and Yeo (2016) and Zhou et al. (2017) document economic
firms, especially because correcting for self-selection bias is chal-
benefits of higher IRQ. We examine how those economic benefits
lenging (Lennox, Francis, & Wang, 2012).6 Because mandatory
arise. We investigate two channels through which IRQ may be
associated with firm value, specifically, a capital market channel,
which relates to the quality of information provided to capital 5
There are three additional differences. First, the mandatory integrated reporting
period in our sample, 2011 to 2014, is longer than that in Lee and Yeo (2016), 2011
to 2013, and in Zhou et al. (2017), 2009 to 2012, who combine voluntary and
4
In addition, recent country-level corporate reporting initiatives have incorpo- mandatory sample years. Second, we use an external measure of IRQ, which
rated integrated reporting principles. For example, since October 2013 firms in the mitigates concerns relating to self-constructed indices. Third, the EY sample covers
United Kingdom are required to prepare a standalone strategic report as part of a larger portion of JSE market capitalization than Lee and Yeo (2016), 90% versus
their annual report. Although the “Guidance on the Strategic Report” issued by the 73%.
6
Financial Reporting Council (FRC) includes integrated reporting principles, the Discretion to publish a report voluntarily is different from exercising discretion
objective of this reporting is to provide relevant information to shareholders (FRC, in implementing mandatory reporting requirements. The former gives rise to se-
2014). In contrast, the objectives of integrated reporting include promoting internal lection bias, whereas the latter is the topic of interest in a vast literature on earnings
integrated thinking and decision making (Deloitte, 2015). quality. See Dechow, Ge, and Schrand (2010) for a review.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 47

disclosers may be forced into a second-best reporting alternative, (2004) theory, which suggests that liquidity stimulates the entry
findings based on mandatory settings help establish a lower bound of informed investors into the market thereby making prices higher
for economic consequences of disclosure quality. Further, our and more informative. The higher prices can have a feedback effect
estimation equations include controls for the issuance of a volun- that affects managerial decisions, which leads to better perfor-
tary, standalone CSR report and, thus, we provide evidence on mance and higher firm value. Recognizing this, informed traders
whether mandated integrated reports are associated with eco- trade even more aggressively, which results in a positive cascade
nomic benefits incremental to those associated with CSR reports. whereby good news leads to more good news. Fang et al. (2009) find
We focus on the relation between integrated reporting and firm the relation between liquidity and performance is stronger for firms
value because at its core the IIRC Framework is about the value with high business uncertainty. Because integrated reports deal
creation process. We examine two possible channels through with capitals that are difficult to quantify and contain forward-
which integrated reporting may affect firm value, i.e., the capital looking information that lacks certainty, higher IRQ may affect
market channel and the real effects channel. Following Amihud and liquidity by stimulating additional trading by informed investors.
Mendelson (2012), we investigate three determinants of firm value, Thus, our first hypothesis is:
i.e., liquidity, cost of capital, and expected future cash flows.7 Based
H1. There is a positive association between IRQ and a firm's stock
on Leuz and Wysocki (2016), we view liquidity and cost of capital as
liquidity.
part of the capital market channel. Whether expected future cash
flows reflect a capital market effect, a real effect, or both is an A theoretical link between cost of capital and disclosure exists to
empirical question. the extent that disclosure affects a firm's non-diversifiable risk (Beyer
A theoretical link between liquidity and disclosure quality is et al., 2010). Lambert, Leuz, and Verrecchia (2007) posit that ac-
well established (Leuz & Wysocki, 2016). In a standard agency counting information can influence cost of capital directly through
framework, information asymmetry leads to adverse selection, the market's assessment of the riskiness of future cash flows.
which leads investors with less information to price-protect or exit There are at least three channels through which disclosure can
the market to reduce losses that would arise from trading with affect cost of capital. First, disclosure can reduce information
more informed investors. These actions reduce liquidity, decrease asymmetry. Second, theory suggests disclosure can improve in-
share prices, and increase cost of capital because investors demand vestors' awareness of non-financial aspects of the firm, which re-
a premium as compensation for risk (e.g., Easley & O'Hara, 2004; sults in a larger investor base with increased risk sharing amongst
Francis, Nanda, & Olsson, 2008; Gietzmann & Ireland, 2005). investors (Merton, 1987). Merton’s (1987) capital market equilib-
Disclosure can reduce information asymmetry and increase firm rium model allows for incomplete information. In that setting, in-
value by decreasing investors' monitoring cost and levelling the vestors only purchase shares of firms that they know about because
playing field among investors (Verrecchia, 2001; see Beyer, Cohen, gathering and processing information about a firm is costly. Third,
Lys, & Walther, 2010 for a review). Consistent with this theory, investors do not know the firm's true expected return. Theory
empirical studies find a positive relation between disclosure quality suggests that disclosure reduces parameter uncertainty and esti-
and liquidity. Welker (1995), Healy, Hutton, and Palepu (1999), and mation risk, parts of which are non-diversifiable (Barry & Brown,
Brown and Hillegeist (2007) find a positive relation in the context 1985).
of voluntary disclosure and Daske, Hail, Leuz, and Verdi (2008) find The IIRC (2013) contends that integrated reports can help in-
that liquidity increased after mandatory adoption of IFRS. vestors understand the risks to which the firm is exposed and how
Integrated reporting can reduce information asymmetry about its strategy and business model respond to those risks. Thus, inte-
the capitals that affect value. Because investors have limited grated reports can affect a firm's cost of capital by: (1) improving
attention and processing power, behavioral theory suggests that and expanding the information available to capital market partic-
improving the salience of information affects investors' perceptions ipants, thereby lowering information asymmetry, (2) providing an
(Hirshleifer & Teoh, 2003). As Zhou et al. (2017) explain, integrated inexpensive, but complete, overview of a firm's activities, thereby
reports can provide new value relevant information, and can pre- expanding the firm's investor base, and (3) reducing parameter
sent information, even if previously disclosed, in a more concise uncertainty and estimation risk because it can help providers of
and useful manner. In addition to financial capital, the IIRC financial capital better understand how a firm creates value by
Framework requires that managers of firms report on material as- presenting a holistic picture that relates the six capitals on which a
pects relating to manufactured, intellectual, human, social and firm depends. Further, the strategic focus and future orientation
relational, and natural capitals. Although some aspects of the other guiding principle of integrated reporting and the content elements
capitals may be addressed by voluntary CSR reports, e.g., environ- relating to the business model, risks and opportunities, strategy
mental issues, a criticism of CSR reports is that they are discon- and resource allocation, and outlook can be useful to investors in
nected from the firm's strategy, business model, and financial reducing parameter uncertainty and estimation risk.
performance (e.g., Serafeim, 2015). As such, providers of financial Thus, our second hypothesis is:
capital obtain an incomplete and disjointed picture of the firm's
H2. There is a negative association between IRQ and cost of
ability to create valuedin its broader sensedover the short, me-
capital.
dium, and long terms.
To the extent that disclosure increases liquidity, there also are The relation between IRQ and expected future cash flows could
non-agency based theories that link liquidity to firm value. For reflect a capital market effect, a real effect, or both. A capital market
example, Fang, Noe, and Tice (2009) find that liquidity is positively effect can arise if integrated reports improve investors' ability to
related to firm value. Their results support Khanna and Sonti’s estimate future cash flows by improving the quality, range, and
connectivity of information. Thus, higher quality reports could
improve investors’ forecasting. To the extent that investors bias the
forecasts downward when faced with incomplete information, the
7
In pricing a financial asset, Amihud and Mendelson (1986) posit that investors more detailed and integrated disclosures in the integrated report
consider liquidity costs and their time horizons. The former affects firm value
through the cost of capital whereas the latter creates a concave relation between
could lead to higher forecasted future cash flows and higher valu-
liquidity costs and expected returns. However, liquidity also can affect firm value ations. In addition, disclosure about the six capitals can be
through other channels as in Khanna and Sonti (2004).
48 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

informative to stakeholders, such as customers and employees, expected future cash flows, EFCF, and i and t denote firm and year.
who associate with more socially responsible firms. This can result All variable definitions are in Appendix 1.
in increased financial performance (Plumlee et al., 2015). IRQ_R is the annual percentile rank of IRQ (see Section 3.3). The
A real effect can arise if real decisions made by managers are coefficient on IRQ_R, b1, represents the difference in economic
different from the decisions that would be made in the absence of consequences between the firm with the lowest and highest IRQ
integrated reporting. One of the benefits of integrated reporting each year (Larcker & Rusticus, 2010). Because bid-ask spreads in-
touted by the IIRC is that its use can lead to integrated thinking and crease with illiquidity, if higher quality integrated reports are
integrated decision-makingde.g., by breaking down silos and associated with higher liquidity, we expect b1 is negative in the
focusing on long-term, instead of short-term, strategydthat results Bid_Ask version of equation (1). If higher quality integrated reports
in better real decisions and enhanced firm value. Similarly, Eccles are associated with lower cost of capital, we expect b1 is negative in
and Serafeim (2015) explain that whereas traditional financial re- the COC version. If higher quality integrated reports are associated
ports serve mainly an information function, integrated reports can with higher expected future cash flows, we expect b1 is positive in
affect internal decision making. the EFCF version. To reduce the likelihood of correlated omitted
Thus, our third hypothesis is: variables, we also estimate equation (1) based on first differences of
all variables, i.e., a changes specification, which effectively uses the
H3. There is a positive association between IRQ and expected
firm as its own control.8
future cash flows.
To correct for cross-sectional and time-series dependence in
Although higher expected cash flows increases firm value, the regression residuals from estimating equation (1), we cluster
higher expectations could result from better managerial decision standard errors by firm and by year (Gow, Ormazabal, & Taylor,
making, which should result in higher ex post realized operating 2010). Although cluster-robust methods can over-reject a true
cash flows, i.e., a real effect, or from investors' enhanced ability to null when the number of clusters is small, bootstrapping can
forecast cash flows, which decreases any information-related overcome the small cluster concern (Cameron, Gelbach, & Miller,
downward bias they may assess, i.e., a capital market effect. This 2008). Thus, because we have only four time clusters, we cluster
line of reasoning leads to our fourth and fifth hypotheses, which based on bootstrapping 10,000 iterations.
help distinguish the real and capital market effects explanations for
a relation between IRQ and expected future cash flows. 3.2. Liquidity, cost of capital, and expected future cash flow proxies
H4. There is a positive association between IRQ and ex post real-
ized operating cash flows. Our hypotheses require proxies for three constructs: liquidity,
cost of capital, and expected future cash flows. This section dis-
H5. There is a positive association between IRQ and investors’ cusses our proxies for each construct.
cash flow forecast accuracy. Leuz and Verrecchia (2000) observe that bid-ask spread cap-
To provide additional evidence of a link between IRQ and real tures information asymmetry because when information asym-
effects, we investigate a particular real effect, namely investment metry is small investors are less concerned about adverse selection
efficiency. Conceptually, investment efficiency refers to undertak- and, as a result, are more willing to trade, which results in lower
ing projects with positive net present value (NPV) under the bid-ask spreads (Cheng, Dhaliwal, & Neamtiu, 2011; Leuz &
assumption of no market frictions, such as adverse selection or Verrecchia, 2000; Muller, Riedl, & Sellhorn, 2011). Following
agency costs. As such, under-investment results in passing up in- Bushee, Core, Guay, and Hamm (2010), we use bid-ask spread as a
vestment opportunities with positive NPV, whereas over- proxy for information asymmetry generally and an inverse proxy
investment results in investing in projects with negative NPV for liquidity specifically. Bid_Ask is the natural logarithm of the
(Biddle, Hilary, & Verdi, 2009). Given the dual objective of inte- median of the difference between daily closing bid and ask prices
grated reporting, namely improved information for outside pro- divided by their midpoint (Daske et al., 2008; Lang, Lins, & Maffett,
viders of financial capital and better internal decision making, it is 2012). Our calculation period commences the day after the prior
possible that high quality integrated reports can improve invest- year's integrated report release and ends on the day of the current
ment efficiency by reducing information asymmetry and improving year's integrated report release. We hand collect report release
shareholders’ monitoring ability (Biddle et al., 2009). dates from the JSE Stock Exchange News Service (SENS).9
Thus, our sixth hypothesis is: Proxies for cost of capital are subject to measurement error
because of the assumptions of the models on which they are based
H6. There is a positive association between IRQ and investment and data used to estimate them (Hail & Leuz, 2009). As a result,
efficiency. there is lack of agreement in the literature about the appropriate
proxy (e.g., Dhaliwal et al., 2014). We follow the literature that uses
3. Research design relating to components of firm valuedH1 the average of four cost of capital proxies as the proxy for cost of
through H3 capital (e.g., Daske et al., 2008; Hail & Leuz, 2006, 2009).

3.1. Estimation equations

8
We do not include firm fixed effects in levels specifications because, depending
To examine which components of firm valuedliquidity, cost of on the sample, we have, on average, only approximately 3.5 observations per firm.
capital, and expected future cash flowsdare associated with IRQ, However, the changes specification effectively includes firm fixed effects. Thus, we
we estimate equation (1), base our inferences on the levels and the changes findings taken together.
9
The EY awards for 2011, 2012, 2013, and 2014 were announced on 27 September
2012, 25 July 2013, 6 August 2014, and 7 August 2015. For each firm-year obser-
Econit ¼ b0 þ b1 IRQ Rit þ b2 Govit þ b3 CSRPerfit þ b4 CSR SAit vation, the date on which EY announced the awards was after the date on which
firms released their integrated report for that year. Thus, the market effects in our
þ b5 LowAQit þ b6 Complexit þ b7 Fogit þ b8 Primeit contemporaneous tests cannot be attributed to the announcement of the EY
X
þ bj Controlsit þ εit (1) awards. To determine when the report is publicly available, we search press re-
leases on the JSE SENS. When firms do not announce release of the integrated
report, we use the date they release their annual report or final set of financial
where Econ is bid-ask spread, Bid_Ask, cost of capital, COC, or statements.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 49

Specifically, our cost of capital proxy, COC, is the mean of the cost of based on the 2012 draft IIRC Framework is correlated with the EY
capital proxies in Claus and Thomas (2001), Gebhardt, Lee, and quality categories, which adds validity to our use of the EY scores to
Swaminathan (2001), and Ohlson and Juettner-Nauroth (2005), measure IRQ. Although EY used different score sheets each year to
and the modified price-earnings-growth (PEG) model of Easton reflect integrated reporting guidelines in effect at the time, our
(2004). In calculating COC, we measure share price at the inte- comparison of the guiding principles and content elements
grated report release date for the current year, and use one-year- included in the various score sheets reveals that the fundamental
ahead analyst forecasts issued between the release of the prior principles of what constitutes a high quality integrated report did
and current years' integrated reports. not change substantially over time.
As our proxy for expected future cash flows we use one-year- We have proprietary access to the underlying scores of the three
ahead target share prices forecasted by financial analysts adjudicators that support the publicly announced quality cate-
(Plumlee et al., 2015). Because we only use one-year-ahead fore- gories. Because the underlying scores contain more information
casted target prices, the effect of discounting target prices to the than the quality categories, we construct our proxy for IRQ from the
current year is negligible. Thus, to avoid introducing noise into our detailed information. We identify 12 score components that are
proxy of expected future cash flows by discounting using an esti- consistent across the four annual score sheets. These are strategic
mated firm-specific discount rate, we use the undiscounted target focus and future orientation, SFO; connectivity of information,
price forecasts. Our proxy for expected future cash flows, EFCF, is CONT; stakeholder relationships, STHR; materiality, MAT; concise-
the natural logarithm of the mean of target price per share forecasts ness, CONS; organizational overview and external environment,
issued within 90 days after the release of the integrated report. We OOEE; governance, GOV; business model, BMOD; risks and oppor-
use a 90-day window to obtain a reasonable sample size for our tunities, RIOPP; performance, PERF; basis of preparation and pre-
tests. sentation, BPP; and other, OTH. For each year, we allocate the
questions in the EY score sheet to one of these components and
allocate to OTH questions that do not fit any of the remaining
3.3. Measure of IRQ
components.
Because the number of questions included in the EY score sheet
We construct our proxy for IRQ from the scores underlying the
varies across years, the number of questions we allocate to each
annual EY Excellence in Integrated Reporting Awards. The EY
component also varies. We calculate the raw, i.e., unscaled, score for
evaluations reflect the assessment of three external independent
each question as the mean score of the three adjudicators. We
adjudicators who score each integrated report on multiple criteria;
calculate the score for each of the 12 components as the sum of the
these individuals were the same for all four of our sample years. The
firm's raw score for each question allocated to that component,
chair of the adjudication panel states that the “marking process is
scaled by the sum of the maximum available score for all the
not simply about ‘ticking the box.’ More emphasis is placed on the
questions allocated to that component. Our proxy for integrated
quality of information presenteddthe relevance, understandabil-
report quality, IRQ, is the mean of the 12 component scores.11
ity, accessibility and connectedness of that information, whether
Table 1 provides descriptive statistics relating to IRQ. Panel A
users of the integrated reports would have a reasonable sense of the
presents statistics that compare IRQ across the EY quality cate-
issues that are core to the operations of each of the companies, and
gories. Panel A reveals that in each year, mean IRQ increases
whether companies have dealt with the issues that users would
monotonically from the lowest EY quality categorydprogress to be
have expected” (Graham, 2014, 16).10 EY categorizes the reports
madedto the highest quality categorydtop 10. For example, in
into five quality categoriesdtop 10, excellent, good, average, and
2014, the mean is 25.33% in the progress to be made category and
progress to be madedexcept for 2011 when EY did not use the
82.00% in the top 10 category. Table 1, Panel B, presents distribu-
average category. The quality category for the firm's integrated
tional statistics by year and reveals that mean IRQ is the highest in
report is publicly announced.
2011, 59.57%. Because the EY awards represent a comparison of
Appendix 2 contains a summary of the integrated reporting
report quality across firms in a particular year, higher scores in 2011
guidance that EY incorporated into the adjudicator score sheets for
suggest the EY adjudicators have adjusted their standards over
each year. We evaluated the score sheets used by the adjudicators
time.12 Table 1, Panel C, reports inter-rater reliability statistics for
and found them to be consistent with the IIRC Framework and,
IRQ by year. The intra-class correlation coefficient ranges from 0.91
thus, believe the scores are appropriate to use as a basis for
measuring IRQ. Zhou et al. (2017) find that their measure of IRQ

11
There are at least two alternative approaches to constructing an IRQ measure
10
The scores of the individual adjudicators are inputs to determine the final from the EY data. The first is to sum the raw, i.e., unscaled, scores for each question
ranking. Widely differing adjudicator scores are reviewed to ensure information and divide it by the sum of the maximum score available for each question. This
was not overlooked; the final ranking is based on the mean of the adjudicators' yields a score in which the 12 components are weighted based on the number of
scores, overall perceptions, and discussion among the adjudicators (EY, 2015). EY questions related to that component in EY's score sheet for that year. This yields an
(2014b, 25) provides the following reasons why it only publishes at a high level unadjusted IRQ, UA_IRQ, for which the raw score for each question is, as is IRQ, the
the considerations taken into account in scoring the reports and does not make mean score of the three adjudicators. The second is to calculate the raw score for
public the detailed score sheet: “As a consequence of our intention to change the each question as the median of the adjudicators' scores. We calculate MED_IRQ as
mark plan on an annual basis and the subjectivity involved in its use, the actual the sum of the raw score of each question divided by the sum of the maximum
mark plan is not made public. Furthermore, a mark plan that is publicly available score available for each question. Not surprisingly, untabulated statistics reveal that
could have the disadvantage of encouraging gamesmanship among the partici- IRQ, UA_IRQ, and MED_IRQ are highly correlated. The weakest Pearson correlation in
pants, rather than striving to encourage true excellence in integrated reporting.” the four years is 0.99 between IRQ and MED_IRQ in 2011. Given these strong positive
Thus, we are unable to reveal the detailed score sheets. We interviewed Graham on correlations, using UA_IRQ or MED_IRQ in place of IRQ is unlikely to affect our
21 August 2014 about the adjudication process of scoring the integrated reports. He inferences.
12
indicated that it takes between one-half and 4 hours to evaluate a report, An alternative explanation for higher IRQ scores in 2011 is that quality
depending on the extent to which a firm has implemented integrated reporting decreased over time. A conversation with the chair of the EY adjudication panel on
principles. He also indicated that the adjudicators distinguish between credible 1 December 2016 confirmed that the EY adjudicators were more lenient in the first
information and puffery. Graham stated: “For some firms, it is merely a public re- year of the awards given the lack of guidance on integrated reporting at that time.
lations exercise through pictures, but we would never rate those well. It is hard for a With the release of new integrated reporting guidance, the adjudicators adjusted
firm to be excellent without clear key performance indicators. If a firm has a lot of their expectations of what constitutes a high quality integrated report and became
green washing and narratives are too long, it is not going to be excellent”. stricter. He confirmed that the scores do not reflect deterioration in IRQ over time.
50 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Table 1
Descriptive statistics for IRQ.

Panel A: IRQ compared with EY categories

Top 10 Excellent Good Average Progress to be made

EY Category % % % % %

IRQ (Mean)
2011 84.53 76.31 63.98 N/A 45.03
2012 74.13 65.58 54.98 42.42 25.40
2013 72.78 65.86 50.82 40.27 22.73
2014 82.00 72.78 60.17 45.14 25.33
Number of firms
2011 10 17 29 N/A 44
2012 10 18 30 26 16
2013 10 25 29 20 16
2014 10 21 27 25 17

Panel B: IRQ

Mean Median Std. dev. Min. Max.

% % % % %

2011 59.79 61.88 17.14 24.93 93.13

2012 50.70 52.20 16.30 14.85 80.40
2013 50.17 49.20 16.68 15.77 77.32
2014 55.32 58.63 18.63 17.13 88.63

Panel C: IRQ Inter-rater reliability

2011 2012 2013 2014

Intraclass correlation coefficient 0.91 0.92 0.91 0.94

F-value 10.84 11.74 11.67 15.74
p-value 0.00 0.00 0.00 0.00

(F-statistic ¼ 10.84; p-value ¼ 0.00) in 2011 to 0.94 (F- of capital, and Cho, Lee, and Pfeiffer (2013) find that CSR perfor-
statistic ¼ 15.74; p-value ¼ 0.00) in 2014, which suggests a high mance is negatively associated with bid-ask spread. CSRPerf is the
degree of reliability amongst the adjudicators’ ratings. mean of the environmental and social scores from Asset 4. We also
To the extent that the raw scores change over time, they are not include an indicator variable to capture whether a firm issues a
comparable across years. To address this problem, we convert IRQ standalone CSR report in addition to its integrated report, CSR_SA.
into percentile ranks each year following the literature using As- Cho, Guidry, Hageman, and Patten (2012) and Patten (2002) show
sociation for Investment Management and Research (AIMR) scores that CSR performance and disclosure are separate constructs, and
as a proxy for disclosure quality (e.g., Botosan & Plumlee, 2002; Dhaliwal et al. (2011) find that firms issue a standalone CSR report
Lang & Lundholm, 1993, 1996; Larcker & Rusticus, 2010; in an attempt to lower cost of capital. Following Simnett,
Lundholm & Myers, 2002). Specifically, each year we calculate the Vanstraelen, and Chua (2009), we base CSR_SA on data we hand
rank of IRQ, IRQ_R, as (firm rank ‒ 1)/(number of firms e 1). IRQ_R collect from Corporate Register at www.corporateregister.com and
ranges from zero for the lowest ranked firm to one for the highest firms’ websites.
ranked firm. Consistent with how Lundholm and Myers (2002) Equation (1) includes a control for accounting quality because
treat the change in AIMR rankings, we use the annual change in Francis et al. (2008) show that earnings quality is negatively asso-
IRQ_R to measure a firm's change in relative IRQ. ciated with cost of capital, and Lang et al. (2012) show that liquidity
We also report results from equation (1) after replacing IRQ_R is higher for firms with lower incidence of earnings management.
with the percentile rank score of each of the 12 components.13 Although discretionary accruals is a commonly used proxy for ac-
Because the IRQ components are highly correlated with each counting quality, prior research uses property, plant, and equip-
other, we estimate the equations separately for each component. ment and inventory as key inputs to estimate discretionary
accruals. These inputs are not as pertinent to financial firms, which
3.4. Control variables constitute 31.16% of our sample. Thus, instead, following Bowen,
Rajgopal, and Venkatachalam (2008), we use the number of small
Based on prior literature, we include variables in equation (1) as positive earnings surprises, LowAQ, as our earnings quality mea-
controls for other factors that may be correlated with IRQ_R or sure. LowAQ is the percentage over the prior four years of small
associated with the economic consequences we test (e.g., Dhaliwal earnings surprises, which is a difference between 0 and 0.01 of net
et al., 2011, 2012, 2014; Hail & Leuz, 2006). We include a control for income in year t minus net income in year te1, scaled by total assets
corporate governance, Gov, because Chen, Chen, and Wei (2009) at the end of year te2. To construct LowAQ, we require at least two
find that corporate governance is negatively associated with cost earnings surprises. LowAQ equal to zero (one) represents the
of capital, which also affects firm value. Gov is the mean of the highest (lowest) accounting quality.
board function, board structure, compensation policy, and share- Equation (1) also includes a control for firm complexity, Com-
holder rights scores from Asset 4. We include a control for CSR plex, because it may be more difficult for complex firms to imple-
performance, CSRPerf, because El Ghoul, Guedhami, Kwok, and ment the concepts of integrated reporting. For example, the extent
Mishra (2011) show the importance of controlling for CSR perfor- to which firms comply with the IIRC Framework's guiding princi-
mance in tests of the association between CSR disclosure and cost ples of conciseness and connectivity of information may differ for
firms with multiple segments or divisions and firms with a simple
structure and operations. Complex is the mean of the annual decile
13
We thank a reviewer for suggesting this analysis. rank of earnings volatility, stock return volatility, and number of
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 51

segments (De Franco, Hope, Vyas, & Zhou, 2015). otherwise, Loss, and the equity book-to-market ratio, BTM (Lang
The quality of a firm's integrated report may be a proxy for a et al., 2012).17 The expected future cash flows version includes
firm's overall disclosure quality. Thus, we include in equation (1) a expected growth, ExpGr, the ratio of the mean one-year-ahead
control for overall disclosure quality. Li (2008) uses the Gunning analyst target price to the actual price minus one measured on
Fog index as a measure of the readability of annual reports and the integrated reporting release date, and accruals, Accr, the dif-
shows that firms with a larger Fog index have lower earnings ference between net income before extraordinary items and pref-
persistence and lower future profitability. Lehavy, Li, and Merkley erence dividends and net cash flow from operating activities, scaled
(2011) find that less readable annual reports are associated with by total assets (Barth, Cram, & Nelson, 2001; Bilinski, Lyssimachou,
greater analyst earnings forecast dispersion, lower forecast accu- & Walker, 2013; Doyle, Lundholm, & Soliman, 2003).
racy, and greater forecast uncertainty. Biddle et al. (2009) use the The bid-ask spread and expected future cash flows versions
inverse of the Fog index as a proxy for financial reporting quality include idiosyncratic risk, IRisk, the standard deviation of residuals
and show that the index is positively associated with investment from a firm-specific regression of daily stock returns on market
efficiency. Because we seek to measure disclosure quality separate returns over the past year. The cost of capital version includes an-
from IRQ, we base our measure on the Fog index relating to firms' alyst forecast dispersion, Disp, the natural logarithm of the standard
press releases, which we collect from JSE SENS. JSE listed firms are deviation of one-year-ahead earnings per share forecast divided by
required to announce through SENS news of events, e.g., directors the absolute value of the consensus earnings per share forecast; the
trading in shares or other price sensitive information. Firms also equity book-to-market ratio, BTM; forecasted long-term earnings
can voluntarily announce information through SENS. Our disclo- growth, LtGr, the difference between the two-year-ahead
sure quality proxy, Fog, is the Gunning Fog index of the press re- consensus earnings per share forecast and the one-year-ahead
leases a firm made during the current year, multiplied by minus one consensus earnings per share forecast scaled by the one-year-
so that Fog is increasing in disclosure quality.14 ahead consensus forecast; and leverage, Lev, the ratio of total
Firms with a secondary JSE listing are not subject to the apply- debt to the sum of total debt and book value of common equity
or-explain integrated reporting regulation.15 However, EY's sam- (Dhaliwal et al., 2014; Zhou et al., 2017). All versions include year
ple includes the 100 largest JSE firms based on market capitaliza- and industry fixed effects.18
tion, regardless of whether the firm has a primary or a secondary
JSE listing. EY (2014b, 2015) notes that although many secondary
listed firms do not label their annual reports as integrated, the re- 3.5. Data and sample
ports include many integrated reporting principles and EY has
ranked some of the reports as excellent. Thus, to preserve sample Integrated reporting became effective for firms listed on the JSE
size, we include both primary and secondary listed firms in our for annual periods beginning on or after 1 March 2010. As Section
sample. However, equation (1) includes an indicator variable, 3.3 discusses, we use data compiled by EY South Africa for its
Prime, which equals one for firms with a primary JSE listing and Excellence in Integrated Reporting Awards to construct our IRQ
zero otherwise, as a control for different reporting incentives be- proxy. The first awards were based on integrated reports for 2011.
tween primary and secondary JSE listed firms.16 We use data from 2011 to 2014. EY compiles data on the top 100
We also include in equation (1) controls for additional factors, firms on the JSE based on market capitalization at the end of each
Controls. All versions of equation (1) include the CAPM beta, Beta, calendar year. These firms represent approximately 90% of the
and size, Size, the natural logarithm of beginning-of-year market market capitalization of the JSE in 2011 and 2012 (EY, 2012; EY,
value of equity. The bid-ask spread version includes an indicator 2013) and approximately 94% in 2013 and 2014 (EY, 2014b; EY,
variable that equals one if the firm reports a loss and zero 2015). Thus, even though our sample is small relative to archival
studies based on U.S. or cross-country data, it covers most of the JSE
in terms of market capitalization. Our sample consists of the 80
firms evaluated by EY in all four years. We use these firms to ensure
14
Other control variables such as CSR_SA, LowAQ, and Size also may capture our IRQ measure is not affected by changes in sample composition.
overall disclosure quality. In addition, we repeat our main analyses using a We obtain remaining data from Compustat Global, Datastream,
disclosure quality proxy based on management forecasts, MFDisc. MFDisc measures IBES, Asset4, SENS, and hand-collection. The sample size varies
the frequency and precision of management forecasts in the current and preceding
across analyses because of data availability.
two years (Baginski & Rakow, 2012; Plumlee et al., 2015). We obtain management
forecasts from the Capital IQ Key Developments database. The database covers 164 Table 2 details our sample. Panel A reveals that the sample for
key developments, of which three contain management forecasts: “corporate the bid-ask spread analyses consist of 292 firm-year observations
guidance e lowered” (#26), “corporate guidance e raised” (#27), and “corporate from 79 firms and the sample for the cost of capital (expected
guidance e new/confirmed” (#29). We do not limit the forecasts to earnings only future cash flows) analyses consist of 221 (189) firm-year
because other forecasts such as production and capital expenditure are important
for mining firms, which comprise a large part of our sample. MFDisc is the natural
logarithm of the product of “supplier,” “frequency,” and “precision” plus one.
Supplier equals one if a firm issued at least one management forecast, and zero 17
We do not include analyst following as a control variable in our bid-ask spread
otherwise, frequency is the number of management forecasts issued, and precision equation because doing so would substantially reduce our sample size. Although
is the average of the precision of management forecasts issued, all during the some studies include analyst following as a control variable in bid-ask equations
current and preceding two financial periods. Precision equals one for general (e.g., Bushee et al., 2010), others do not (e.g., Daske et al., 2008). In addition, some
impression forecasts, two for minimum and maximum forecasts, three for range other control variables in the equation, e.g., size, governance, accounting quality,
forecasts, and four for point forecasts. Our inferences are unaffected by using Fog, and the issuance of a standalone CSR report, can be interpreted as proxies for
MFDisc instead of Fog as a proxy for overall disclosure quality. information environment and monitoring.
15 18
See paragraph 18.4 of the JSE Limited Listings Requirements Issue 21 (available Our controls for changes in Loss and CSR_SA are indicator variables; we
online at https://www.jse.co.za/content/JSERulesPoliciesandRegulationItems/JSE% construct three indicators for each. For Loss, the indicators reflect firms that
20Listings%20Requirements.pdf). changed from a loss to a profit, firms that reported losses in both years, and firms
16
Except for 2011, mean IRQ is similar for primary and secondary JSE listed firms. that changed from a profit to a loss. Firms that reported profits in both years are
In 2011, 2012, 2013, and 2014 mean IRQ for primary listed firms is 61.21, 51.26, 50.17, reflected in the intercept. For CSR_SA, the indicators reflect firms that issued a
and 55.82, and for secondary listed firms is 51.10, 46.57, 50.17, and 52.24. These standalone CSR report in the prior year but not in the current year, firms that issued
statistics suggest that even though the integrated reporting regulation is not a standalone CSR report in both years, and firms that did not issue a standalone CSR
mandatory for secondary listed firms, these firms provide integrated reports with report in the prior year, but did so in the current year. Firms that did not issue a
levels of quality similar to those of primary listed firms. standalone CSR report in either year are reflected in the intercept.
52 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Table 2 4.2. Findings

Sample.

Panel A: Sample selection Firms Firm-years We begin by establishing that the association between IRQ and
EY observations (4 years x 80 firms per year) 80 320
firm value documented by Lee and Yeo (2016) is present in our
Firm with missing data (1) (4) sample. As do Lee and Yeo (2016), we follow an extensive literature
Firm-year observations with missing data (24) that uses Tobin's Q as a proxy for firm value.19 Tobin's Q is designed
Sample: Bid-ask 79 292 to reflect the market's valuation of a firm's assets relative to their
Observations with missing data (14) (71) carrying amounts (e.g., Lang & Maffett, 2011). This attribute makes
Sample: Cost of capital 65 221 Tobin's Q a suitable proxy in our setting because intellectual, hu-
Observations with missing data (6) (32) man, social, and natural capitals are only partially, or not, reflected
Sample: Expected future cash flows 59 189
in the carrying amount of assets. Hence, we investigate whether
IRQ is associated with firm value beyond what is contained in the
Panel B: Industry composition
financial statements. Table 3, Panel A, reveals that mean (median)
GICS sector Firm-years % Tobin's Q, TobinQ, is 1.81 (1.47).
Energy 8 2.74 Table 4, Panel A, presents regression summary statistics from
Materials 64 21.92 estimating equation (1) when TobinQ is the dependent variable.
Industrials 23 7.88 Regarding the levels specification, the coefficient on IRQ_R is pos-
Consumer (discretionary) 32 10.96
itive and significantly different from zero (coefficient ¼ 0.350, t-
Consumer (staples) 44 15.07
Health care 15 5.14 statistic ¼ 1.89). This finding suggests that Tobin's Q of the most
Financials 91 31.16 forthcoming IRQ firm is higher by 0.350 than that of the least
Information technology 3 1.03 forthcoming IRQ firm, which is a 23.8% increase over the median
Telecommunication services 12 4.11
Tobin's Q.
Sample: Bid-ask 292 100.00 Regarding the control variables, most notably, Cash, ROA, Beta,
and Prime have positive coefficients, which suggests that firm value
increases with cash levels, profitability, risk, and primary listing
observations from 65 (59) firms. Panel B reveals that the Financial
status. Complex has a negative coefficient, which indicates that
and Materials sectors have the largest number of observations:
more complex firms have lower firm value than less complex firms.
31.16% and 21.92% of the sample. The industry composition of the
AtGr also has a negative coefficient, which is inconsistent with
cost of capital and expected future cash flows samples is similar to
higher growth firms having higher firm value. However, untabu-
that in the bid-ask spread sample.
lated univariate statistics reveal that AtGr is positively correlated
with firm value, which indicates the negative coefficient reported
in Table 4 reflects AtGr's multivariate relation with TobinQ.
4. Results relating to liquidity, cost of capital, and expected The changes specification in Table 4 also provides support for
future cash flows the positive association between IRQ and firm value. The relation
between DIRQ_R and DTobinQ is positive and significant
4.1. Descriptive statistics (coefficient ¼ 0.117, t-statistic ¼ 1.74).
We extend Lee and Yeo (2016) by examining the association
Table 3, Panel A, provides distributional statistics for the vari- between Tobin's Q and the separate components of IRQ. Table 4,
ables we use in the levels analysis. Panel A reveals that mean and Panel B, reveals a significant positive association between TobinQ
median Bid_Ask is 6.07, and mean (median) EFCF is 4.64 (4.66). and materiality, MAT_R (coefficient ¼ 0.362, t-statistic ¼ 3.06),
Mean (median) of COC is 0.11 (0.10), which is comparable to the which is significant at a lower level than is IRQ_R in Table 4, Panel A
mean for South African firms of 0.11 reported in Dhaliwal et al. (coefficient ¼ 0.350, t-statistic ¼ 1.89). In addition, the coefficients
(2014). Mean (median) Gov and CSRPerf are 53.17 (53.94) and on connectivity, CONT_R, stakeholder relationships, STHR_R,
68.71 (77.59). Mean (median) Fog is 16.34 (16.41), which in- conciseness, CONS_R, and governance, GOV_R, are all significantly
dicates that, on average, a reader of average intelligence would need positive (t-statistics range from 1.93 to 2.15). The coefficients on
16 years of formal education to read the text once and understand its strategic focus and future orientation, SFO_R, risks and opportu-
content. Panel A also reveals that 86% of sample firms have a pri- nities, RIOPP_R, performance, PERF_R, and the basis of preparation
mary JSE listing, 45% issue a standalone CSR report in addition to an and presentation, BPP_R, are positive but not significantly so (t-
integrated report, and 6% report losses. Table 3, Panel B, presents statistics range from 1.39 to 1.59). The coefficients on organiza-
distributional statistics for the variables we use in the changes an- tional overview and external environment, OOEE_R, business
alyses, where D denotes annual change. Panel B reveals that the model, BMOD_R, and other, OTH_R, are not significantly different
standard deviation of DIRQ_R is 0.16 and the inter-quartile range is from zero.
0.19. Because IRQ_R varies between 0 and 1, together these statistics Next, we examine whether IRQ is associated with each of the
indicate there is considerable year-to-year change in IRQ_R, which three determinants of firm value. Table 5, Panel A, presents sum-
mitigates concern that IRQ_R is largely unchanged over time. mary statistics from estimating equation (1) when bid-ask spread is
Table 3, Panel C, presents Pearson and Spearman correlations the dependent variable in levels (changes), Bid_Ask (DBid_Ask).
between IRQ_R and the three dependent variables. Panel C reveals Regarding the levels specification, Table 5, Panel A, reveals that,
that IRQ_R is significantly negatively correlated with Bid_Ask in consistent with H1, the coefficient on IRQ_R is negative and
levels and changes; the Pearson (Spearman) correlations are 0.33
(0.33) and 0.13 (0.15). The correlations between IRQ_R and
EFCF are positive and significant in levels, but insignificant in
19
changes. All correlations between COC and IRQ_R are insignificantly We calculate Tobin's Q as total assets minus book value of equity plus market
value of equity, scaled by total assets (Daske et al., 2008; Lang et al., 2012). To
different from zero, except for the Spearman correlation, which is ensure the information in the integrated reports is reflected in firm value, we
significantly positive in levels. Regardless, we base our inferences calculate market value of equity using share price on the date the current year's
on multivariate analyses. integrated report is released.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 53

Table 3
Descriptive statistics.

Panel A: Levels

N Mean Std. dev. Min P25 Median P75 Max

Bid_Ask 292 6.07 0.42 6.99 6.35 6.07 5.80 5.04

EFCF 189 4.64 1.05 1.88 3.88 4.66 5.35 6.95
COC 221 0.11 0.06 0.05 0.09 0.10 0.12 0.42
TobinQ 292 1.81 1.02 0.62 1.08 1.47 2.17 5.90
IRQ_R 292 0.52 0.29 0.00 0.27 0.52 0.77 1.00
Beta 292 1.13 0.46 0.08 0.76 1.17 1.49 2.51
CSRPerf 292 68.71 23.97 8.39 55.23 77.59 86.09 94.50
CSR_SA 292 0.45 0.50 0.00 0.00 0.00 1.00 1.00
Complex 292 5.49 1.90 1.67 4.00 5.33 6.67 9.67
Fog 292 16.34 1.75 20.05 17.59 16.41 15.21 11.78
Gov 292 53.17 16.01 12.26 43.03 53.94 63.26 86.73
LowAQ 292 0.16 0.24 0.00 0.00 0.00 0.25 1.00
Prime 292 0.86 0.34 0.00 1.00 1.00 1.00 1.00
Size 292 10.54 1.16 8.70 9.65 10.31 11.34 14.04
BTM 292 0.56 0.37 0.08 0.28 0.48 0.79 1.96
IRisk 292 1.33 0.33 0.77 1.12 1.28 1.48 2.60
Loss 292 0.06 0.24 0.00 0.00 0.00 0.00 1.00
Accr 189 0.03 0.08 0.34 0.07 0.02 0.01 0.38
ExpGr 189 0.08 0.14 0.47 0.02 0.08 0.16 0.45
Disp 221 2.33 1.32 5.14 3.10 2.48 1.87 2.51
Lev 221 0.28 0.20 0.00 0.12 0.25 0.38 0.83
LtGr 221 0.44 1.65 0.28 0.09 0.14 0.23 13.41

Panel B: Changes

N Mean Std. dev. Min P25 Median P75 Max

DBid_Ask 212 0.08 0.22 0.67 0.22 0.07 0.07 0.49

DEFCF 125 0.08 0.29 0.86 0.03 0.14 0.25 0.56
DCOC 153 0.01 0.03 0.12 0.02 0.01 0.00 0.11
DTobinQ 212 0.03 0.39 1.73 0.13 0.01 0.14 0.91
DIRQ_R 212 0.00 0.16 0.38 0.09 0.01 0.10 0.59
DBeta 212 0.03 0.33 0.76 0.19 0.03 0.25 0.95
DCSRPerf 212 1.30 5.73 12.80 1.00 0.70 2.43 20.60
DComplex 212 0.07 1.10 4.67 0.67 0.00 0.67 3.67
DFog 212 0.01 1.10 2.83 0.65 0.01 0.54 3.64
DGov 212 0.71 9.63 23.12 4.75 0.21 5.71 27.42
DLowAQ 212 0.02 0.11 0.25 0.00 0.00 0.00 0.25
DSize 212 0.15 0.25 0.64 0.02 0.16 0.29 0.88
DBTM 212 0.00 0.18 0.56 0.08 0.01 0.06 0.74
DIRisk 212 0.07 0.25 0.41 0.10 0.05 0.24 0.86
DAccr 125 0.00 0.09 0.32 0.04 0.00 0.03 0.28
DExpGr 125 0.01 0.13 0.36 0.09 0.00 0.05 0.36
DDisp 153 0.05 1.24 4.06 0.61 0.11 0.53 4.00
DLev 153 0.01 0.08 0.23 0.02 0.00 0.04 0.38
DLtGr 153 0.18 2.62 9.12 0.06 0.02 0.02 20.37

Panel C: Correlations

IRQ_R DIRQ_R
Pearson Spearman Pearson Spearman
(p-value) (p-value) (p-value) (p-value)

Bid_Ask 0.33 0.33 DBid_Ask 0.13 0.15

(0.00) (0.00) (0.06) (0.03)
EFCF 0.24 0.23 DEFCF 0.10 0.03
(0.00) (0.00) (0.28) (0.77)
COC 0.00 0.13 DCOC 0.05 0.01
(0.99) (0.05) (0.51) (0.92)

Panels A (Panel B) present descriptive statistics for all the variables used in the levels (changes) regression analyses for the sample period 2011e2014 (2012e2014) with
sufficient data to estimate the least restrictive specification of the regression equation. Panel C presents the Pearson and Spearman correlations between the dependent
variables and IRQ_R, with p-values in parentheses. All variables are defined in Appendix 1. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are
winsorized at the 1 and 99 percentiles.

significantly different from zero (coefficient ¼ 0.192, t- bid-ask spreads. Contrary to Amihud and Mendelson (1989), the
statistic ¼ 2.54), which indicates that firms with higher IRQ have a association between Beta and bid-ask spread is negative.
smaller bid-ask spreads and, thus, higher liquidity. The coefficient The changes specification reveals the same inferences regarding
estimate suggests that bid-ask spread of the most forthcoming IRQ IRQ. In particular, the coefficient on DIRQ_R is significantly negative
firm is 3.2% lower than that of the least forthcoming IRQ firm based (coefficient ¼ 0.184, t-statistic ¼ 4.48). Regarding the control
on median bid-ask spread. Regarding the control variables, Size variables, the coefficients on DBTM, DLowAQ, and DIRisk are posi-
(IRisk and Prime) is negatively (are positively) associated with bid- tive, and the coefficient on DComplex is negative. These findings
ask spread, which indicate that larger firms, firms with lower idio- reveal that firms with increases in the equity book-to-market ratio
syncratic risk, and firms with a secondary JSE listing have smaller and idiosyncratic risk, and decreases in accounting quality and
54 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Table 4
Regression of Tobin's Q on IRQ.

Panel A: Levels and changes

TobinQ DTobinQ
Variable (Pred.) Coefficient t-stat. Variable (Pred.) Coefficient t-stat.

IRQ_R (+) 0.350** 1.89 DIRQ_R (+) 0.117** 1.74

AtGr 0.674** 2.08 DAtGr 0.234 1.33
Cash 1.562** 2.02 DCash 0.498 0.56
ROA 6.578*** 7.99 DROA 0.650 0.80
Lev 0.117 0.38 DLev 1.030 1.24
Beta 0.265** 2.56 DBeta 0.018 0.42
IRisk 0.055 0.43 DIRisk 0.045 0.38
Size 0.083 1.05 DSize 0.040 0.17
Gov 0.001 0.35 DGov 0.003 1.31
CSRPerf 0.001 0.38 DCSRPerf 0.004 0.59
LowAQ 0.196 0.99 DLowAQ 0.126 1.47
Complex 0.069*** 3.10 DComplex 0.017 0.82
Fog 0.036 1.18 DFog 0.018 1.28
Prime 0.411** 2.33
CSR_SA 0.073 0.71 CSR_SA indicators Yes
Div 0.065 0.62 Div indicators Yes
Industry fixed effects Yes Industry fixed effects Yes
Year fixed effects Yes Year fixed effects Yes
Adjusted R2 0.71 Adjusted R2 0.13
N 292 N 212

Panel B: IRQ components

SFO_R CONT_R STHR_R MAT_R CONS_R OOEE_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

TobinQ 0.303* 0.305** 0.223** 0.362*** 0.302** 0.272

(1.57) (1.93) (1.97) (3.06) (2.10) (1.23)

GOV_R BMOD_R RIOPP_R PERF_R BPP_R OTH_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

TobinQ 0.532** 0.134 0.251* 0.296* 0.288* 0.206

(2.15) (0.76) (1.59) (1.50) (1.39) (1.12)

Panel A presents regression results for equation (1) where the dependent variable is TobinQ or DTobinQ. The levels (changes) sample includes 292 (212) firm-year observations
for 79 firms. See Appendix 1 for variable definitions. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are winsorized at the 1 and 99 percentiles.
Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1% level based on a
two-tailed test, except for IRQ_R which is based on a one-tailed test.
Panel B presents the coefficients and t-statistics for each component of IRQ from separate regressions of TobinQ on the 12 categories of IRQ. Because the categories are highly
correlated we estimate the equations separately for each component. For brevity, the coefficients and t-statistics of control variables are not reported. All variables are defined
in Appendix 1. Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1%
level based on a one-tailed test.

complexity have increases in bid-ask spreads. Contrary to expec- PERF_R (coefficient ¼ 0.195), and STHR_R (coefficient ¼ 0.186)
tations, the coefficient on DGov is weakly positive. Taken together, have the largest relation with Bid_Ask. Materiality, MAT_R, organi-
the findings in Table 5, Panel A, indicate that IRQ is positively zational overview and external environment, OOEE_R, and risks and
associated with liquidity, a capital market effect.20 opportunities, RIOPP_R, also are significantly related to Bid-Ask, but
Regarding the bid-ask components analyses, Table 5, Panel B, Governance, GOV_R, is not.
reveals that strategic focus and future orientation, SFO_R, connec- Table 6 presents regression summary statistics from esti-
tivity of information, CONT_R, stakeholder relationships, STHR_R, mating equation (1) when cost of capital, COC or DCOC, is the
conciseness, CONS_R, business model, BMOD_R, performance, dependent variable. The statistics from the levels regression
PERF_R, basis of preparation and presentation, BPP_R, and other reveal that IRQ_R has a negative but insignificant relation with
components, OTH_R, are all significant in explaining Bid_Ask (all p- cost of capital (coefficient ¼ 0.011, t-statistic ¼ 0.57). The
values  1%). In terms of magnitude, SFO_R (coefficient ¼ 0.197), control variable coefficients indicate that higher risk firms have
higher cost of capital, whereas larger firms and primary JSE listed
firms have lower cost of capital. Table 6 also reveals that the co-
20 efficient on DIRQ_R in the changes specification is not significantly
Consistent with Daske et al. (2008) and Lang et al. (2012), our tests are based on
bid-ask spread for the current year, which acknowledges that information in the different from zero (coefficient ¼ 0.010, t-statistic ¼ 1.10). Thus,
integrated report could be revealed and incorporated in equity prices during the the findings in Table 6 provide no evidence of an association be-
year. However, proponents of integrated reporting contend that the connectivity of tween IRQ and cost of capital, which is inconsistent with H2.21
information and the presentation of a holistic view are important features of in-
tegrated reports, which suggests that investors benefit from having access to the
integrated report. As a result, we also conduct a short-window test around the
21
integrated report release date. Specifically, we measure bid-ask spread as the Although this result may appear contrary to Zhou et al. (2017), they find a
natural logarithm of the median of the daily bid-ask spread over days 5 to þ5 relation between a measure of IRQ and cost of capital only for firms with low an-
relative to the release date of the current year's integrated report. The untabulated alyst following. Because our sample includes only the top 100 JSE firms, which
findings reveal inferences similar to those based on Table 5. likely attract the most analyst attention, our results are consistent with theirs.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 55

Table 5
Regression of liquidity on IRQ.

Panel A: Levels and changes

Bid_Ask DBid_Ask
Variable (Pred.) Coefficient t-stat. Variable (Pred.) Coefficient t-stat.

IRQ_R () 0.192*** 2.54 DIRQ_R () 0.184*** 4.48

BTM 0.020 0.21 DBTM 0.196** 2.29
Size 0.222*** 8.21 DSize 0.001 0.01
Gov 0.000 0.27 DGov 0.002* 1.85
CSRPerf 0.000 0.03 DCSRPerf 0.001 0.14
LowAQ 0.015 0.14 DLowAQ 0.236** 2.13
Complex 0.004 0.23 DComplex 0.027* 1.73
Fog 0.003 0.31 DFog 0.016 1.00
Beta 0.147* 1.91 DBeta 0.084 1.41
IRisk 0.176* 1.89 DIRisk 0.081* 1.83
Prime 0.168** 2.04
CSR_SA 0.007 0.18 CSR_SA indicators Yes
Loss 0.016 0.35 Loss indicators Yes
Industry fixed effects Yes Industry fixed effects Yes
Year fixed effects Yes Year fixed effects Yes
Adjusted R2 0.70 Adjusted R2 0.21
N 292 N 212

Panel B: IRQ components

SFO_R CONT_R STHR_R MAT_R CONS_R OOEE_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

Bid_Ask 0.197*** 0.172*** 0.186*** 0.140** 0.166*** 0.135**

(2.72) (2.65) (2.65) (1.78) (2.52) (1.68)

GOV_R BMOD_R RIOPP_R PERF_R BPP_R OTH_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

Bid_Ask 0.118 0.131*** 0.151** 0.195*** 0.180*** 0.156***

(1.27) (2.85) (1.89) (2.86) (2.33) (2.82)

Panel A presents regression results for equation (1) where the dependent variable is Bid_Ask or D Bid_Ask. The levels (changes) sample includes 292 (212) firm-year ob-
servations for 79 firms. See Appendix 1 for variable definitions. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are winsorized at the 1 and 99
percentiles. Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1%
level based on a two-tailed test, except for IRQ_R which is based on a one-tailed test.
Panel B presents the coefficients and t-statistics for each component of IRQ from separate regressions of Bid_Ask on the 12 categories of IRQ. Because the categories are highly
correlated we estimate the equations separately for each component. For brevity, the coefficients and t-statistics of control variables are not reported. All variables are defined
in Appendix 1. Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1%
level based on a one-tailed test.

Table 6
Regression of cost of capital on IRQ.

COC DCOC
Variable (Pred.) Coefficient t-stat. Variable (Pred.) Coefficient t-stat.

IRQ_R () 0.011 0.57 DIRQ_R () 0.010 1.10

Beta 0.034*** 2.56 DBeta 0.008 1.24
Disp 0.003 0.63 DDisp 0.006* 1.85
BTM 0.003 0.21 DBTM 0.013 0.45
Lev 0.023 1.22 DLev 0.029 0.71
LtGr 0.004 0.44 DLtGr 0.001 0.28
Size 0.018* 2.15 DSize 0.029** 2.29
Gov 0.000 0.13 DGov 0.000 0.55
CSRPerf 0.000 1.40 DCSRPerf 0.000 0.57
LowAQ 0.015 0.80 DLowAQ 0.010 0.56
Complex 0.003 1.54 DComplex 0.002* 1.74
Fog 0.001 0.90 DFog 0.002 1.12
Prime 0.078** 2.25
CSR_SA 0.002 0.73 CSR_SA indicators Yes
Industry fixed effects Yes Industry fixed effects Yes
Year fixed effects Yes Year fixed effects Yes
Adjusted R2 0.41 Adjusted R2 0.06
N 221 N 153

This table presents regression results for equation (1) where the dependent variable is COC or DCOC. The levels (changes) sample includes 221 (153) firm-year observations for
91 (60) firms. See Appendix 1 for variable definitions. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are winsorized at the 1 and 99 percentiles.
Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1% level based on a
two-tailed test, except for IRQ_R which is based on a one-tailed test.
56 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Table 7
Regression of expected future cash flows on IRQ.

Panel A: Levels and changes

EFCF DEFCF
Variable (Pred.) Coefficient t-stat. Variable (Pred.) Coefficient t-stat.

IRQ_R (+) 0.386 1.26 DIRQ_R (+) 0.146** 2.02

ExpGr 1.140** 2.85 DExpGr 0.959*** 3.70
Accr 0.846 1.32 DAccr 0.105 1.13
Beta 0.075 0.35 DBeta 0.049 0.60
IRisk 0.361 0.96 DIRisk 0.048 0.40
Size 0.304** 2.35 DSize 0.047 0.58
Gov 0.003 0.50 DGov 0.002 0.75
CSRPerf 0.005 0.82 DCSRPerf 0.004 0.52
LowAQ 0.228 0.59 DLowAQ 0.102 0.61
Complex 0.223*** 4.45 DComplex 0.013 0.60
Fog 0.064 1.37 DFog 0.019 0.80
Prime 0.321 0.54
CSR_SA 0.103 0.63 CSR_SA indicators Yes
Industry fixed effects Yes Industry fixed effects Yes
Year fixed effects Yes Year fixed effects Yes
Adjusted R2 0.54 Adjusted R2 0.31
N 189 N 125

Panel B: IRQ components

SFO_R CONT_R STHR_R MAT_R CONS_R OOEE_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

EFCF 0.233 0.413* 0.508** 0.550** 0.508** 0.302

(0.71) (1.54) (1.76) (2.14) (2.22) (0.96)

GOV_R BMOD_R RIOPP_R PERF_R BPP_R OTH_R

Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient

(t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.) (t-stat.)

EFCF 0.131 0.056 0.246 0.276 0.110 0.441**

(0.42) (0.17) (0.95) (0.91) (0.42) (1.75)

Panel A presents regression results for equation (1) where the dependent variable is EFCF or DEFCF. The levels (changes) sample includes 189 (125) firm-year observations for
59 (54) firms. See Appendix 1 for variable definitions. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are winsorized at the 1 and 99 percentiles.
Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1% level based on a
two-tailed test, except for IRQ_R which is based on a one-tailed test.
Panel B presents the coefficients and t-statistics for each component of IRQ from separate regressions of EFCF on the 12 categories of IRQ. Because the categories are highly
correlated we estimate the equations separately for each component. For brevity, the coefficients and t-statistics of control variables are not reported. All variables are defined
in Appendix 1. Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1%
level based on a one-tailed test.

However, Leuz and Wysocki (2016) note that evidence on the (coefficient ¼ 0.386, t-statistic ¼ 1.26), Table 7, Panel B, reveals
relation between disclosure and cost of capital is mixed, that the coefficients on stakeholder relationships, STHR_R, mate-
and Larcker and Rusticus (2010) find no relation between riality, MAT_R, conciseness, CONS_R, and other, OTH_R, are
disclosure and cost of capital using an instrumental variables significantly positive (t-statistics range from 1.75 to 2.22) and the
approach. coefficient on connectivity, CONT_R, is marginally so (t-
Table 7, Panel A, presents regression summary statistics for statistic ¼ 1.54). The coefficients on the other components are not
equation (1) when expected future cash flows, EFCF or DEFCF, is the significantly different from zero. Materiality, MAT_R, has the
dependent variable. Findings from the levels specification reveal largest coefficient, 0.550, which indicates that the expected future
that the association between IRQ_R and EFCF is positive, but cash flows of the most forthcoming MAT firm are 11.8% higher
insignificant (coefficient ¼ 0.386, t-statistic ¼ 1.26). Regarding the than those of the least forthcoming MAT firm based on median
control variables, the coefficients on ExpGr, Size, and Complex are expected future cash flows.
positive, which suggests that firms with higher expected growth, Together, the findings in Tables 4 through 7 reveal positive
larger firms, and more complex firms have higher expected future economic consequences associated with IRQ. We find that IRQ is
cash flows. The summary statistics from the changes specification positively related to firm value, and that this relation is largely
provide stronger evidence in support of an association between IRQ attributable to liquidity and expected future cash flows, and not
and expected future cash flows in that Table 7 reveals a significant cost of capital.22 The liquidity results support IRQ having a capital
positive relation between DIRQ_R and DEFCF (coefficient ¼ 0.146, t- market effect. Section 5 explores whether the relation be-
statistic ¼ 2.02). Combined, the findings in Table 7, Panel A, provide tween IRQ and expected future cash flows reflects a capital
some support that, consistent with H3, higher quality integrated
reports are positively associated with the expected cash flows
component of firm value.
22
Regarding the expected cash flows and the components Because we find a significant association for liquidity but not cost of capital, our
results are more consistent with a non-agency based reason for the liquidity effect
version of equation (1), although the coefficient on IRQ_R is
(e.g., Khanna & Sonti, 2004). That is, in an agency framework, liquidity affects firm
not significantly different from zero in Table 7, Panel A value by reducing the cost of capital, but our results do not support this link.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 57

Table 8
Future operating cash flows.

CFO1 CFO2 CFOSUM

Variable (Pred.) Coefficient t-stat. Coefficient t-stat. Coefficient t-stat.

IRQ_R (+) 0.047* 1.42 0.049** 1.91 0.124** 1.97

Beta 0.018 1.24 0.030** 1.99 0.083*** 2.72
IRisk 0.020 0.58 0.045* 1.73 0.113* 1.72
Accr 0.338*** 2.65 0.270** 2.44 0.665** 2.18
RevGr 0.038 1.27 0.026 1.13 0.045 0.97
Size 0.014 1.40 0.009 0.85 0.017 0.70
Gov 0.000 0.43 0.000 0.60 0.000 0.37
CSRPerf 0.001 1.32 0.000 1.10 0.001 1.27
LowAQ 0.101*** 4.01 0.093*** 3.39 0.231*** 3.53
Complex 0.010* 1.96 0.013*** 3.01 0.028** 2.35
Fog 0.007 1.13 0.004 0.78 0.014 0.90
Prime 0.024 0.69 0.017 0.50 0.054 0.82
CSR_SA 0.001 0.09 0.004 0.30 0.001 0.05
Ind. fixed effects Yes Yes Yes
Year fixed effects Yes Yes Yes
Adjusted R2 0.32 0.35 0.36
N 285 208 208

This table presents regression results for equation (1) where the dependent variable is CFO1, CFO2, or CFOSUM. The CFO1 (CFO2, CFOSUM) sample includes 285 (208) firm-year
observations for 77 firms. All variables are defined in Appendix 1. All variables, excluding indicator variables, IRQ_R, LowAQ, Complex, and Beta, are winsorized at the 1 and 99
percentiles. Standard errors are clustered by firm and by year based on bootstrapping methods using 10,000 iterations. *, **, and *** denote significance at a 10%, 5%, and 1%
level based on a two-tailed test, except for IRQ_R which is based on a one-tailed test.

market effect, a real effect, or both. Regardless, we find that and CFOSUM), which is consistent with the long-term value crea-
connectivity, stakeholder relationships, materiality, and concise- tion concept of integrated reporting.
ness are the most consistent drivers of the significant associa- Second, we test the capital market effect channel in H5, i.e.,
tions between IRQ and firm value, liquidity, and expected future whether IRQ is associated with accuracy of analyst target price
cash flows. forecasts. If integrated reports improve investors’ ability to forecast
future cash flows more accurately, we expect a negative association
between IRQ and target price forecast errors. Thus, we estimate the
5. Expected future cash flows: real and/or capital market
relation between target price forecast error, TPE, and IRQ_R, the
effect? (H4 and H5)
common control variables in Section 3.4, IRisk, analyst following,
and analyst earnings forecast bias.23 The untabulated findings reveal
Our finding of a positive association between IRQ and expected
no significant association between IRQ_R and target price forecast
future cash flows in H3 is consistent with integrated reporting
errors. This lack of significance suggests that IRQ is not associated
improving decision making by managers, i.e., a real effect, which is
with improved analyst future cash flows forecast accuracy.
consistent with anecdotal and survey evidence (Black Sun, 2014;
Overall, these results are consistent with expected future cash
SAICA, 2015; IIRC, 2016b). However, the finding also is consistent
flows affecting firm value through managers’ real decisions, a real
with investors revising upwards estimates of future cash flows
effect. We find no evidence that this relation is driven by more
because they have a better understanding of the firm's capitals and
accurate forecasting of future cash flows, a capital market effect.
business strategy, i.e., a capital market effect. These two explana-
tions are not mutually exclusive.
First, we test the real effect channel in H4, i.e., whether IRQ is 6. Investment efficiency (H6)
positively associated with future realized operating cash flows. In
particular, we replace EFCF in our expected future cash flows To examine the association between investment efficiency, a real
analysis with ex post net cash flows from operating activities effect, and IRQ, we estimate the following ordered logit equation,
deflated by beginning-of-period total assets. We estimate separate
equations using one-year-ahead, two-year-ahead, and the sum of
Ineffitþ1 ¼b0 þ b1 IRQ Rit þ b2 Govit þ b3 CSRPerfit
one- and two-year-ahead net cash flows from operating activities,
CFO1, CFO2, and CFOSUM. Table 8 reveals a positive and margin- þ b4 CSR SAit þ b5 LowAQit þ b6 Fogit (2)
X
ally significant association between IRQ_R and CFO1 þ b7 Primeit þ bj Controlsit þ εit
(coefficient ¼ 0.047, t-statistic ¼ 1.42). The results are stronger
when we use CFO2 or CFOSUM as the dependent variable
(coefficients ¼ 0.049 and 0.124, t-statistics ¼ 1.91 and 1.97 for CFO2 where Ineff is investment inefficiency. Similar to Biddle et al. (2009)
and Navissi, Sridharan, Khedmati, Lim, and Evdokimov (2017), we
measure investment inefficiency as deviations from the expected
23
Target price forecast error is the natural logarithm of the mean of the absolute level of investment using a model that predicts investment as a
differences between target price forecasts and share price at the end of the 12- function of growth opportunities.24 In particular, Ineff is the annual
month forecast period, scaled by share price at the current year's integrated
report release date. Analyst following is the natural logarithm of the number of
analysts issuing one-year-ahead earnings forecasts for the current year between the
24
release dates of the integrated reports of the prior and current years. Analyst Consistent with prior research, we exclude financial firms from our sample
earnings forecast bias is the one-year-ahead analyst earnings forecast error, i.e., (Biddle et al., 2009; Chen, Young, & Zhuang, 2013; Lara, Osma, & Penalva, 2016).
consensus IBES earnings minus actual IBES earnings per share, divided by the ab- Because the investment efficiency literature does not estimate specifications in
solute value of actual IBES earnings per share. Consensus IBES earnings is the changes, we do not either. For this reason, and to preserve sample size, for this
median IBES one-year-ahead earnings forecasts for the current year issued between analysis, we do not limit our sample to the 80 firms evaluated by EY in all four
the release dates of the integrated reports of the prior and current years. years.
58 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Table 9
Investment efficiency.

Model 1 Model 2

Ineff Under vs. Normal Over vs. Normal

Variable (Pred.) Coefficient z-stat. Coefficient z-stat. Coefficient z-stat.

IRQ_R () 1.290*** 3.00 1.133* 1.58 1.062** 1.95

Sales_Vol 3.771 1.56 3.785*** 3.46 0.208 0.14
CFO_Vol 17.580** 2.09 7.100*** 3.56 6.542*** 2.90
Inv_Vol 0.070 1.55 0.022 0.29 0.125 1.39
OpCycle 0.417 0.79 0.444 0.93 0.979* 1.68
LogAsset 0.124 0.39 0.075 0.25 0.090 0.44
BTM 0.201 0.23 0.525 0.56 3.722*** 4.19
Z_Score 0.124* 1.90 0.186 1.41 0.025 0.14
Tangible 0.005 0.00 0.885 0.48 1.527 1.36
CFO_Sales 0.821 0.59 2.538*** 2.73 8.188*** 6.15
Slack 0.150 0.53 0.677 1.15 1.017 1.31
Age 0.012 0.07 0.004 0.01 0.131 0.68
Cash 0.830 0.28 5.906*** 4.35 0.366 0.15
Lev 1.828** 2.41 1.469 0.95 1.669 1.50
Gov 0.004 0.44 0.011 0.31 0.019 0.88
CSRPerf 0.005 0.36 0.024 1.12 0.019 0.62
LowAQ 0.253 0.25 1.573 0.81 0.894 0.50
Fog 0.062 0.59 0.112 0.66 0.248* 1.88
Prime 0.486 0.55 0.918 0.75 0.083 0.11
CSR_SA 0.131* 1.88 0.403 0.88 0.026 0.04
Loss 1.274 1.14 0.089 0.08 0.985 0.85
Div 0.139 0.17 0.189 0.11 0.363 0.34
Industry fixed effects Yes Yes
Year fixed effects Yes Yes
Pseudo R2 0.10 0.32
N 209 209

This table presents the results of an ordered logit regression (Model 1) and multinomial logit regressions (Model 2) of Ineff on IRQ_R and control variables. The sample includes
209 firm-year observations of 64 firms. See Appendix 1 for variable definitions. All variables, excluding indicator variables, Ineff, IRQ_R, and LowAQ, are winsorized at the 1 and
99 percentiles. Standard errors are clustered by firm and by year and in the case of the multinomial logit regression on bootstrapping methods using 10,000 iterations. *, **, and
*** denote significance at a 10%, 5%, and 1% level based on a two-tailed test, except for IRQ_R which is based on a one-tailed test.

quartile rank of the absolute value of the residuals from equation Table 9 presents the results.26 Model 1 is the ordered
(3), which we estimate by Global Industry Classification Sector and logit regression where the dependent variable is Ineff. The coef-
year and require at least 10 observations in each industry-year. ficient on IRQ_R is significantly negative (coefficient ¼ 1.290, z-
statistic ¼ 3.00). This coefficient indicates that the odds of being
in the top quartile of Ineff is 0.275 times lower for observations with
Investitþ1 ¼ b0 þ b1 SalesGrowthit þ εitþ1 (3)
the highest IRQ than those with the lowest IRQ. This evidence
Invest is the sum of research and development, capital, and suggests that IRQ is associated with improved investment decision-
acquisition expenditures minus proceeds from the sale of fixed making.
assets, multiplied by 100 and scaled by lagged total assets, and Model 2 is the multinomial logit regression, which separately
SalesGrowth is change in sales scaled by lagged sales. The control compares the odds of being in the under- or over-investment
variables in equation (2) are based on the investment efficiency group to the odds of being in groups with normal investment
literature (Biddle et al., 2009; Cheng, Dhaliwal, & Zhang, 2013; levels. The first set of results of Model 2 relates to under-
Navissi et al., 2017).25 investment. The coefficient on IRQ_R is negative, which is consis-
To provide evidence on the association between IRQ and over- tent with less under-investment, but it is not significantly so
and under-investment, we rank the signed residuals from (coefficient ¼ 1.133, z-statistic ¼ 1.58). The second set of results
equation (3) into quartiles each year. The highest quartile of re- of Model 2 relates to over-investment. The coefficient on IRQ_R is
siduals represents the over-investment group, the lowest quar- significantly negative (coefficient ¼ 1.062, z-statistic ¼ 1.95),
tile represents the under-investment group, and the middle which indicates that the odds of being in the over-investment
quartiles represent the benchmark group (Biddle et al., 2009; group rather than the normal investment group is 0.346 times
Navissi et al., 2017). We estimate a multinomial logit version of lower for observations with the highest IRQ than those with the
equation (2) based on this grouping that predicts the likelihood lowest IRQ.
that a firm will be in one of the two extreme quartiles. This allows
us to estimate simultaneously the separate likelihood of over-
and under-investment.

26
Because estimating two-way clustered bootstrapped standard errors for or-
25
The controls are sales volatility, operating cash flow volatility, investment dered logit regressions is complex, Table 9 reports t-statistics based on two-way
volatility, operating cycle, assets, equity book-to-market ratio, Altman's Z-Score, clustered standard errors without bootstrapping. Untabulated statistics reveal
tangible assets, operating cash flow-to-sales ratio, firm age, financing slack, cash, that our inferences are unaffected if we use OLS with two-way clustered boot-
leverage, loss, and dividends. In addition, we include controls for governance, CSR strapped standard errors. Lang, Raedy, and Wilson (2006) and Barth, Landsman,
performance, accounting quality, Fog, listing status, and standalone CSR report. See and Lang (2008) also use OLS instead of logit. Greene (1993) shows that logit
Appendix 1 for variable definitions. models are sensitive to the effects of heteroskedasticity.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 59

7. Additional analyses Specifically, we estimate the relation between changes in one-

year-ahead COC and changes in current year IRQ and changes in
7.1. Endogeneity the control variables. Untabulated findings reveal that the coefficient
on DIRQ_R is not significantly different from zero, which is consistent
The findings in Tables 4 through 7 are based on tests of with the findings in Table 6. We do not claim that this analysis fully
contemporaneous associations between IRQ and the four economic resolves the simultaneity concern because recent evidence suggests
consequences we examinedfirm value, liquidity, cost of capital, that the lead-lag approach is not effective in addressing simultaneity
and expected future cash flows. These association tests are not (Bellemare, Masaki, & Pepinsky, 2015; Reed, 2015). However, a
intended to, and do not, permit us to draw causal inferences. failure to find a significant association in the lead-lag cost of capital
However, descriptive research examining interesting institutional tests also could be attributable to low power because of our small
questions could inform future causal research (Chen & Schipper, sample. Regardless, we are unable to rule out completely endoge-
2016; Gow, Larcker, & Reiss, 2016). Given the growing interest in neity concerns.
integrated reporting, providing descriptive evidence on integrated
reporting seems worthwhile. Nonetheless, we consider the extent 7.2. Further tests
to which endogeneity could affect our inferences.
There are three potential sources of endogeneity that could affect Following prior research (e.g., Lesmond, 2005; Leuz &
our inferences, correlated omitted variable bias, selection bias, and Verrecchia, 2000; Levine & Schmukler, 2006), we view bid-ask
simultaneity (Larcker & Rusticus, 2010; Lennox et al., 2012). To address spread as a more direct proxy for liquidity than share turnover.
correlated omitted variable bias, we estimate changes specifications of Nonetheless, we use share turnover as an alternative proxy in the
the equations underlying the findings in Tables 4 through 7. Given that bid-ask spread version of equation (1). Share turnover is the natural
integrated reporting is mandatory in South Africa for primary JSE logarithm of the median of the daily ratio of the number of shares
listed firms, selection bias is less of a concern than when the issuance traded on the JSE to the total number of shares outstanding,
of an integrated report is voluntary (see footnotes 3, 5, 6 and 15). measured from the release date of the prior year's integrated report
We do not believe simultaneity is a concern for our inferences to the release date of the current year's integrated report. We use
for the following reasons. Regarding bid-ask spread, our hypothesis the same control variables as in the bid-ask spread version of
states that IRQ influences liquidity and we predict a negative equation (1). Untabulated findings reveal that the coefficient on
contemporaneous association between IRQ and bid-ask spread. IRQ_R is positive and significantly different from zero in the levels
However, managers can observe the firm's bid-ask spread and alter specification (coefficient ¼ 0.346, t-statistic ¼ 1.93), which is
the quality of the firm's integrated report. In particular, when a consistent with the bid-ask spread findings in Table 5. However,
firm's bid-ask spread is high, its managers may increase IRQ in the untabulated findings also reveal that in the changes specification
current year to reduce bid-ask spread in the next year. However, the coefficient on DIRQ_R is not significantly different from zero,
this reasoning predicts a positive contemporaneous association which is inconsistent with the findings in Table 5.
between bid-ask spread and IRQ. Thus, if bid-ask spread affects IRQ, As an additional test relating to firm value, we employ a modified
current year IRQ would be higher, which would bias upwards our Ohlson (1995) valuation model in levels, i.e., price, and changes i.e.,
IRQ coefficient estimate. Given that we predict and find a negative returns (see Barth & Clinch, 2009). In the levels specification, we
association, any effect of bid-ask spread on IRQ is unlikely to estimate the relation between share price on the current year's in-
explain fully our findings. Similar reasoning could be applied to tegrated report release date and IRQ_R, book value of equity per
Tobin's Q, expected future cash flows, and cost of capital.27 share, income before extraordinary items per share, and other
However, simultaneity could explain why we find an insignificant control variables. In the changes specification, we estimate the
association between IRQ and cost of capital when we predict a relation between return, which is the percentage change in Data-
negative association (see Table 6). This is because the predicted stream Total Return measured from the release date of the prior
negative contemporaneous association could be offset by the posi- year's integrated report to the release date of the current year's
tive association between IRQ and cost of capital for firms with higher integrated report, and DIRQ_R, earnings before extraordinary items
cost of capital improving their current year IRQ to reduce cost of (IB) scaled by lagged market value of equity, change in IB scaled by
capital in the next year. To address this possibility, we re-estimate lagged market value of equity, and other control variables.29 Unta-
our cost of capital changes specification using a lead-lag approach bulated findings reveal that the coefficient on IRQ_R is not signifi-
similar to that in Dhaliwal et al. (2011) and Zhou et al. (2017).28 cantly different from zero in either specification. Taken together
with our Tobin's Q findings in Table 4, the insignificant findings
associated with share prices and returns suggest that IRQ is more
27
For Tobin's Q and expected future cash flows, the effect of these two economic strongly associated with the excess of market value of assets over
consequences on IRQ likely biases downwards our IRQ coefficient estimates, but we
their carrying amounts than with share prices or returns. This excess
predict and find a positive relation. For cost of capital, Dhaliwal et al. (2011, 63) uses
similar reasoning when stating “endogeneity and self-selection issues can arise if
could reflect some IIRC Framework capitals that are not reflected in
we examine a contemporaneous relation between CSR disclosure and the cost of the financial statements, but are reflected in integrated reports.
equity capital. On the one hand, if CSR disclosure is motivated by a firm's desire to
reduce its high cost of equity capital, then we should find a positive relation be-
8. Conclusion
tween CSR disclosure and the cost of the equity capital. On the other hand, if CSR
disclosure leads to a lower cost of equity capital, then we should find a negative
relation between CSR disclosure and the cost of equity capital.” Although reverse Whereas prior research (Lee & Yeo, 2016) establishes that inte-
causality could be a concern in a lead-lag analysis because managers who expect grated report quality (IRQ) is positively related to firm value, the
the firm to perform well in the future might be more forthcoming and increase purpose of our study is to determine how higher IRQ is associated
current period integrated report quality, our tests are based on contemporaneous,
not lead-lag, associations.
28
Although the lead-lag approach is used in the literature to address simultaneity
29
concerns, it assumes the examined economic outcome is slow to react to changes in The other control variables in the levels (changes) specification are Complex,
disclosure quality. Although this may be plausible for a slowly changing outcome CSRPerf, CSR_SA, Fog, Gov, LowAQ, Prime, AtGr, Lev, (DComplex, DCSRPerf, DFog, DGov,
such as cost of capital, it may not be plausible for other economic outcomes that DLowAQ, DAtGr, DLev, indicator variables for CSR_SA) and industry and year fixed
react more quickly to changes in disclosure quality. effects.
60 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

with higher firm value. Based on the IIRC Framework, we investigate the future of integrated reporting in their jurisdictions. The IIRC may
two possible channels, i.e., a capital market channel and a real effects find our results encouraging as they promote global adoption of
channel. The capital market channel reflects the quality of infor- integrated reporting. Investors, especially institutional investors
mation provided to capital market participants and the real effects with policies for responsible investment, may find our results
channel reflects the quality of internal decision making. informative in evaluating their resource allocation decisions.
Using data from South Africa where integrated reporting is
mandatory for firms listed on the JSE, we disaggregate firm value Acknowledgments
into three determinantsdliquidity, cost of capital, and expected
future cash flowsdand find evidence that supports both channels. The financial support of the Chartered Institute of Management
Specifically, we find a positive relation between IRQ and liquidity, a Accountants (CIMA) is gratefully acknowledged. The authors would
capital market effect, and between IRQ and expected future cash like to thank EY and the College of Accounting at the University of
flows, which could be a capital market effect, i.e., improved infor- Cape Town for making their data available to us. We appreciate
mation enabling capital market participants to make more accurate research assistance from Casey Cahan and coding advice from Ken
cash flow predictions, or a real effect, i.e., improved internal Li and Dan Taylor. We are grateful for comments from two anony-
decision-making that results in higher realized future operating mous reviewers and the editor, Keryn Chalmers, Charl de Villiers,
cash flows. Our findings reveal evidence consistent with the latter Kurt Gee, Lars Hassel, Jenny Hoobler, Jeong-Bon Kim, Wayne
and not the former, which indicates the expected future cash flow Landsman, Shiva Rajgopal, Katherine Schipper, Ann Tarca, and
results are attributable to improved internal decision making, a real seminar participants at Arizona State University, Australian Na-
effect. We also link IRQ to investment efficiency and find that firms tional University, Bristol University, Exeter University, Ludwig
with higher IRQ have higher investment efficiency. Thus, consistent Maximilians University of Munich, University of Pretoria, University
with the IIRC's dual objective of improving information quality and of Waikato, Umeå University, Global Emerging Scholars Workshop
promoting integrated thinking, our findings suggest that the posi- of the American Accounting Association, paper development
tive association between IRQ and firm value is the result of both workshops of the International Association for Accounting Educa-
capital market and real effects. tion and Research in Florence and East-London, European Ac-
Our results may be of interest to managers, practitioners, counting Association Annual Congress, American Accounting
standard-setters, regulators, and investors as they evaluate the Association Annual Meeting, EMAN Conference, and the annual
merits of integrated reporting. For example, the European Commis- Alliance Manchester Business School/London School of Economics
sion, the SEC, and other regulators have expressed interest in inte- Conference. Steven Cahan and Li Chen appreciate funding from the
grated reporting and our results can inform their deliberations on University of Auckland Business School.

Appendix 1. Variable definitions

Variable Definition (data from Compustat, Datastream and Institutional Brokers' Estimate System (IBES) unless stated otherwise)

Dependent variables: H1 through H3

Bid_Ask Bid-ask spread is the natural logarithm of the median of daily (Ask e Bid)/((Ask þ Bid)/2) measured from the day after the release of the prior year's
integrated report to the release date of the current year's integrated report. Ask and Bid are the daily closing ask and bid prices.
TobinQ Market-to-book ratio of total assets calculated as total assets minus total common equity plus common shares outstanding at year-end multiplied by share
price at the release date of the current year integrated report, divided by total assets.
EFCF Natural logarithm of the mean target price forecasts issued up to 90 days after of the release date of the current year's integrated report.
COC Cost of equity capital measured as the average of the four cost of capital proxies of Claus and Thomas (2001), Gebhardt et al. (2001), Ohlson and Juettner-
Nauroth (2005), and the modified price-earnings growth (PEG) ratio model of Easton (2004). Share prices are measured at the release date of the current
year integrated report. One-year-ahead analyst forecasts issued between the release dates of the integrated reports of the prior and current years are used in
the calculation of COC. If any of the cost of capital measures is missing, we use the average of the remaining measures.
Dependent variables: H4 and H5
CFO Ex post net cash flows from operating activities deflated by beginning-of-period total assets where CFO1, CFO2, and CFOSUM represent one-year-ahead, two-
year-ahead, and the sum of one- and two-year-ahead net cash flows from operating activities.
TPE Target price forecast error is the natural logarithm of the mean of the absolute differences between target price forecasts and share price at the end of the 12-
month forecast period, scaled by share price at the current year's integrated report release date.
Dependent variable: H6
Ineff The annual quartile rank of the absolute value of the residuals from a regression of one-year ahead investment levels on sales growth estimated by Global
Industry Classification Sector and year. Investment, Invest, is the sum of research and development expenditure, capital expenditure, and acquisition
expenditure less proceeds from sale of fixed assets, multiplied by 100 and scaled by lagged total assets.
Test variables
IRQ_R Integrated report quality (IRQ) measured annually based on the EY Excellence in Integrated Reporting awards as: (firm rank e 1)/(number of firms e 1). This
gives the annual percentile rank of integrated report quality ranging between 0 and 1 each year.
SFO_R Strategic focus and future orientation component of IRQ ranked as is IRQ_R.
CONT_R Connectivity of information component of IRQ ranked as is IRQ_R.
STHR_R Stakeholder relationships component of IRQ ranked as is IRQ_R.
MAT_R Materiality component of IRQ ranked as is IRQ_R.
CONS_R Conciseness component of IRQ ranked as is IRQ_R.
OOEE_R Organizational overview and external environment component of IRQ ranked as is IRQ_R.
GOV_R Governance component of IRQ ranked as is IRQ_R.
BMOD_R Business model component of IRQ ranked as is IRQ_R.
RIOPP_R Risks and opportunities component of IRQ ranked as is IRQ_R.
PERF_R Performance component of IRQ ranked as is IRQ_R.
BPP_R Basis of preparation and presentation component of IRQ ranked as is IRQ_R.
OTH_R Other component of IRQ ranked as is IRQ_R.
Control variables
Accr Accruals calculated as the difference between net income before extraordinary items and preference dividends and net cash flow from operating activities,
scaled by total assets.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 61

(continued )

Variable Definition (data from Compustat, Datastream and Institutional Brokers' Estimate System (IBES) unless stated otherwise)

Age Natural logarithm of the difference between the reporting date and the earliest year the firm appears in Compustat Global or Worldscope or the JSE listing
date.
Anafol Analyst following is the natural logarithm of the number of analysts issuing one-year-ahead earnings forecasts for the current year between the release dates
of the integrated reports of the prior and current years.
AtGr Asset growth measured as the one-year change in total assets from the prior financial year to the current financial year, scaled by lagged total assets.
Beta Coefficient from a firm-specific regression of daily stock returns on market returns over the past year. Because our Betas are all between 0 and 4 we do not
winsorize them (Chen et al., 2009).
Bias Analyst earnings forecast bias is the one-year-ahead analyst earnings forecast error, i.e., consensus IBES earnings minus actual IBES earnings per share,
divided by the absolute value of actual IBES earnings per share. Consensus IBES earnings is the median IBES one-year-ahead earnings forecasts for the current
year issued between the release dates of the integrated reports of the prior and current years.
BTM Book-to-market ratio of equity calculated as the book value of common shareholders' equity, divided by the number of common shares outstanding
multiplied by end-of-year share price.
Cash Cash and cash equivalents scaled by average total assets.
CFO_Sales Ratio of cash flow from operations to sales.
CFO_Vol Standard deviation of cash flow from operations scaled by average total assets from years te5 to te1.
Complex Firm complexity measured as the mean of the annual deciles of earnings volatility, number of segments, and return volatility. Earnings volatility is the
natural logarithm of the standard deviation of income before extraordinary items per share from years te5 to te1. Number of segments is the number of
product segments. If the number of product segments is not available, we use the number of geographic segments. We hand-collect data when the number
of segments is missing from Datastream. Return volatility is the standard deviation of daily stock returns over te1. If any component of Complex is missing,
we use the average of the remaining components.
CSRPerf Corporate social responsibility performance score calculated as the average of the environmental, ENVSCORE, and the social performance score, SOCSCORE,
from Asset 4.
CSR_SA Indicator variable that equals one if a firm issued a stand-alone CSR report in addition to its integrated report, and zero otherwise. We hand collect this data
item from Corporate Register (www.corporateregister.com) and firms' websites.
Div Indicator variable that equals one if the firm declares or pays a dividend in the current year, and zero otherwise.
Disp Natural logarithm of the standard deviation of one-year-ahead earnings per share forecast divided by the absolute value of the consensus earnings per share
forecast.
ExpGr Expected future growth measured as the ratio of the mean one-year-ahead analyst target price to the actual price minus one measured on the integrated
reporting release date.
Fog Robert Gunning Fog index, multiplied by minus one, of the press releases a firm announced through the JSE Stock Exchange News Service (SENS) during the
financial year, where the fog index is calculated as 0.4 x (average number of words per sentence þ percentage of complex words)
Gov Corporate governance score calculated as the average of board function, CGBF, board structure, CGBS, compensation policy, CGCP, and shareholder rights,
CGSR, scores from Asset 4.
Inv_Vol Standard deviation of Invest from years te5 to te1.
IRisk Idiosyncratic risk measured as the standard deviation of residuals from a firm-specific regression of daily stock returns on market returns over the past year.
Lev Leverage calculated as the ratio of total debt to the sum of total debt and the book value of common shareholders' equity.
LogAsset Natural logarithm of total assets.
Loss Indicator variable that equals one if income before extraordinary items is negative, and zero otherwise.
LowAQ Low accounting quality measured as the percentage over the prior four years of small earnings surprises, where a small earnings surprise is a difference
between 0 and 0.01 of net income in year t minus net income in year te1, scaled by total assets at the end of year te2. To construct LowAQ, we require at least
two earnings surprises.
LtGr Long-term growth measured as the difference between the two-year-ahead consensus earnings per share forecast and the one-year-ahead consensus
earnings per share forecast scaled by the one-year-ahead consensus forecast.
OpCycle Natural logarithm of the sum of receivables to sales and inventory to cost of goods sold multiplied by 360.
Prime Indicator variable that equals one if a firm's primary listing is on the JSE, and zero otherwise.
ROA Return on assets calculated as the ratio of income before extraordinary items to lagged total assets.
Sales_Vol Standard deviation of sales scaled by average total assets from years te5 to te1.
Size Size measured as the natural logarithm of market capitalization at the beginning of the year.
Slack Ratio of cash to property, plant and equipment.
Tangible Ratio of property, plant and equipment to total assets.
Z_Score Bankruptcy score calculated in accordance with Altman (1968).

Appendix 2. Integrated reporting guidance used in the development of the EY score sheet

2013 Dec: IIRC Final 2013 Apr: IIRC 2012 Nov: IIRC 2012 Jul: IIRC Draft 2011 Sept: IIRC 2011 Jan: IRC of SAa
Framework Consultation Draft of Prototype Framework Framework Outline Discussion Paper Discussion Paper
Framework

EY Score Sheet: 2014 EY Score Sheet: 2013 EY Score Sheet: 2012 EY Score Sheet: 2012 EY Score Sheet: 2011, EY Score Sheet: 2011
2012
Guiding principles Guiding principles Guiding principles Guiding principles Guiding principles Reporting principles
Strategic focus and Strategic focus and Strategic focus and Strategic focus Strategic focus Materiality
future orientation future orientation future orientation
Connectivity of Connectivity of Connectivity of Future orientation Future orientation Comparability and
information information information consistency
Stakeholder Stakeholder Stakeholder Connectivity of Connectivity of Faithful representation
relationships responsiveness responsiveness information information
Materiality Materiality and Materiality and Responsiveness and Responsiveness and Scope and boundary of
conciseness conciseness stakeholder stakeholder report
inclusiveness inclusiveness
(continued on next page)
62 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

(continued )

2013 Dec: IIRC Final 2013 Apr: IIRC 2012 Nov: IIRC 2012 Jul: IIRC Draft 2011 Sept: IIRC 2011 Jan: IRC of SAa
Framework Consultation Draft of Prototype Framework Framework Outline Discussion Paper Discussion Paper
Framework

Conciseness Reliability and Reliability Conciseness, reliability Conciseness, reliability

completeness and materiality and materiality
Reliability and Consistency and Comparability and
completeness comparability consistency
Consistency and
comparability
Content elements Content elements Content elements Content elements Content elements Elements to be
addressed in an
integrated report
Organizational Organizational Organizational Organizational Organizational Organizational
overview and overview and external overview and operating overview and business overview and business overview, business
external environment context model model model and governance
environment structure
Business model Business model Business model Operating context, Operating context, Understanding the
including risks and including risks and operating context
opportunities opportunities
Risks and opportunities Opportunities and risks Opportunities and risks Governance and Governance and Remuneration policies
remuneration remuneration
Governance Governance Governance Strategic objectives and Strategic objectives and Strategic objectives,
strategies to achieve strategies to achieve competencies, KPIs and
those objectives those objectives KRIs
Strategy and resource Strategy and resource Strategy and resource Performance Performance Account of the
allocation allocation allocation plans organization's
performance
Performance Performance Performance and Future outlook Future outlook Future performance
outcomes objectives
Outlook Future outlook Future outlook Report profile (scope
and boundary)
Basis of preparation Analytical commentary
and presentation
General reporting
guidance
a
The Integrated Reporting Committee of South Africa.

References Black Sun. (2014). Realizing the benefits: The impact of integrated reporting. Black Sun
Plc.
Botosan, C. A., & Plumlee, M. A. (2002). A re-examination of disclosure level and the
Altman, E. I. (1968). Financial ratios, discriminant analysis and the prediction of
expected cost of equity capital. Journal of Accounting Research, 40, 21e40.
corporate bankruptcy. Journal of Finance, 23, 589e609.
Bowen, R. M., Rajgopal, S., & Venkatachalam, M. (2008). Accounting discretion,
Amihud, Y., & Mendelson, H. (1986). Asset pricing and the bid-ask spread. Journal of
corporate governance and firm performance. Contemporary Accounting
Financial Economics, 17, 223e249.
Research, 25, 351e405.
Amihud, Y., & Mendelson, H. (1989). The effects of beta, bid-ask spread, residual
Brown, S., & Hillegeist, S. A. (2007). How disclosure quality affects the level of in-
risk, and size on stock returns. Journal of Finance, 44, 479e486.
formation asymmetry. Review of Accounting Studies, 12, 443e477.
Amihud, Y., & Mendelson, H. (2012). Liquidity, the value of the firm, and corporate
Bushee, B. J., Core, J. E., Guay, W., & Hamm, S. J. W. (2010). The role of the business
finance. Journal of Applied Corporate Finance, 24, 17e32.
press as an information intermediary. Journal of Accounting Research, 48, 1e19.
Baboukardos, D., & Rimmel, G. (2016). Value relevance of accounting information
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2008). Bootstrap-based improvements
under an integrated reporting approach: A research note. Journal of Accounting
for inference with clustered errors. The Review of Economic and Statistics, 90,
and Public Policy, 35, 437e452.
414e427.
Baginski, S. P., & Rakow, K. C. (2012). Management earnings forecast disclosure
Chen, K. C. W., Chen, Z., & Wei, K. C. J. (2009). Legal protection of investors,
policy and the cost of equity capital. Review of Accounting Studies, 17, 279e321.
corporate governance, and the cost of equity capital. Journal of Corporate
Barry, C., & Brown, S. (1985). Differential information and security market equi-
Finance, 15, 273e289.
librium. Journal of Financial and Quantitative Analysis, 20, 407e422.
Cheng, M., Dhaliwal, D. S., & Neamtiu, M. (2011). Asset securitization, securitization
Barth, M. E., & Clinch, G. (2009). Scale effects in capital markets-based accounting
recourse, and information uncertainty. The Accounting Review, 86, 541e568.
research. Journal of Business Finance and Accounting, 36, 253e288.
Cheng, M., Dhaliwal, D., & Zhang, Y. (2013). Does investment efficiency improve
Barth, M. E., Cram, D. P., & Nelson, K. K. (2001). Accruals and the prediction of future
after the disclosure of material weaknesses in internal control over financial
cash flows. The Accounting Review, 76, 27e58.
reporting? Journal of Accounting and Economics, 56, 1e18.
Barth, M. E., Landsman, W., & Lang, M. (2008). International accounting standards
Chen, Q., & Schipper, K. (2016). Comments and observations regarding the relation
and accounting quality. Journal of Accounting Research, 46, 467e498.
between theory and empirical research in contemporary accounting research.
Bellemare, M. F., Masaki, T., & Pepinsky, T. B. (2015). Lagged explanatory variables and
Foundations and Trends in Accounting, 10, 237e285.
the estimation of causal effects. Working paper. University of Minnesota and
Chen, C., Young, D., & Zhuang, Z. (2013). Externalities of mandatory IFRS adoption:
Cornell University.
Evidence from cross-border spillover effects of financial information on in-
Bernardi, C., & Stark, A. W. (2017). Environmental, social and governance disclosure,
vestment efficiency. The Accounting Review, 88, 881e914.
integrated reporting, and the accuracy of analyst forecasts. forthcoming British
Cho, C. H., Guidry, R. P., Hageman, A. M., & Patten, D. M. (2012). Do actions speak
Accounting Review.
louder than words? An empirical investigation of corporate environmental
Beyer, A., Cohen, D. A., Lys, T. Z., & Walther, B. R. (2010). The financial reporting
reputation. Accounting, Organizations and Society, 37, 14e25.
environment: Review of the recent literature. Journal of Accounting and Eco-
Cho, S. Y., Lee, C., & Pfeiffer, R. J. (2013). Corporate social responsibility performance
nomics, 50, 296e343.
and information asymmetry. Journal of Accounting and Public Policy, 32, 71e83.
Biddle, G. C., Callahan, C. M., Hong, H. A., & Knowles, R. L. (2015). Do adoptions of
Claus, J., & Thomas, J. (2001). Equity premia as low as three percent? Evidence from
International Financial Reporting Standards enhance capital investment efficiency?
analysts' earnings forecasts for domestic and international stock markets. The
Working paper. University of Hong Kong.
Journal of Finance, 56, 1629e1666.
Biddle, G. C., Hilary, G., & Verdi, R. S. (2009). How does financial reporting quality
Cohen, J. R., & Simnett, R. (2015). CSR and assurance services: A research agenda.
relate to investment efficiency? Journal of Accounting and Economics, 48,
Auditing: A Journal of Practice and Theory, 34, 59e74.
112e131.
Daske, H., Hail, L., Leuz, C., & Verdi, R. (2008). Mandatory IFRS reporting around the
Bilinski, P., Lyssimachou, D., & Walker, M. (2013). Target price accuracy: Interna-
world: Early evidence on the economic consequences. Journal of Accounting
tional evidence. The Accounting Review, 88, 825e851.
 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64 63

Research, 46, 1085e1142. 2009. Institute of directors in Southern Africa. Johannesburg: IoD.
De Franco, G., Hope, O., Vyas, D., & Zhou, Y. (2015). Analyst report readability. International Accounting Standards Board (IASB). (2017). The IASB and integrated
Contemporary Accounting Research, 32, 76e104. reporting. IASB speech (26 April). Available at: http://www.ifrs.org/-/media/
De Villiers, C., Rinaldi, L., & Unerman, J. (2014). Integrated reporting: Insights, gaps feature/news/speeches/hans-hoogervorst-integrated-reporting-april-2017.pdf.
and an agenda for future research, Accounting. Auditing and Accountability International Integrated Reporting Council (IIRC). (2013). The international inte-
Journal, 27, 1042e1067. grated reporting framework. London: IIRC.
Dechow, P., Ge, W., & Schrand, C. (2010). Understanding earnings quality: A review International Integrated Reporting Council (IIRC). (2016a). Navigating the corporate
of the proxies, their determinants and their consequences. Journal of Accounting reporting landscape. London: IIRC.
and Economics, 50, 344e401. International Integrated Reporting Council (IIRC). (2016b). Creating value: The
Deloitte. (2015). <IR>: How does it fit into the UK reporting landscape? London: cyclical power of integrated thinking and reporting. London: IIRC.
Deloitte. International Integrated Reporting Council (IIRC). (2017). The GRI/IIRC corporate
Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2011). Voluntary nonfinancial leadership group. IIRC video (30 may). Available at: https://integratedreporting.
disclosure and the cost of equity capital: The initiation of Corporate Social org/news/video-the-gri-iirc-corporate-leadership-group/.
Responsibility Reporting. The Accounting Review, 86, 59e100. Kana, G. (2017). Firms urged to adopt IR standards. The star online (31 March).
Dhaliwal, D. S., Li, O. Z., Tsang, A., & Yang, Y. G. (2014). Corporate social responsibility Available at: http://www.thestar.com.my/business/business-news/2017/03/31/
disclosure and the cost of equity capital: The roles of stakeholder orientation firms-urged-to-adopt-ir-standards/.
and financial transparency. Journal of Accounting and Public Policy, 33, 328e355. Khanna, N., & Sonti, R. (2004). Value creating stock manipulation: Feedback effect of
Dhaliwal, D. S., Radhakrishnan, S., Tsang, A., & Yang, Y. G. (2012). Nonfinancial stock prices on firm value. Journal of Financial Markets, 7, 237e270.
disclosure and analyst forecast accuracy: International evidence on corporate King, M. (2013). Foreword to the PWC publication. The value creation journey: A survey
social responsibility disclosure. The Accounting Review, 87, 723e759. of JSE Top-40 companies' integrated reports. Johannesburg: PWC.
Doyle, J. T., Lundholm, R. J., & Soliman, M. T. (2003). The predictive value of expenses Lambert, R., Leuz, C., & Verrecchia, R. E. (2007). Accounting information, disclosure,
excluded from Forma earnings. Review of Accounting Studies, 8, 145e174. and the cost of capital. Journal of Accounting Research, 45, 385e420.
Easley, D., & O'Hara, M. (2004). Information and the cost of capital. Journal of Lang, M. H., Lins, K. V., & Maffett, M. (2012). Transparency, liquidity, and valuation:
Finance, LIX, 1553e1583. International evidence on when transparency matters most. Journal of Ac-
Easton, P. (2004). PE Ratios, PEG ratios, and estimating the implied expected rate of counting Research, 50, 729e774.
return on equity capital. The Accounting Review, 79, 73e95. Lang, M. H., & Lundholm, R. J. (1993). Cross-sectional determinants of analyst rat-
Eccles, R. G., & Krzus, M. P. (2015). Meaning and momentum in the integrated ings of corporate disclosures. Journal of Accounting Research, 31, 246e271.
reporting movement. Journal of Applied Corporate Finance, 27, 8e17. Lang, M. H., & Lundholm, R. J. (1996). Corporate disclosure policy and analyst
Eccles, R. G., & Serafeim, G. (2015). Corporate and integrated reporting: A functional behavior. The Accounting Review, 71, 467e492.
perspective. In S. Mohrman, J. O'Toole, & E. Lawler (Eds.), Corporate stewardship: Lang, M. H., & Maffett, M. (2011). Transparency and liquidity uncertainty in crisis
Organizing for sustainable effectiveness. Sheffield, UK: Greenleaf Publishing. periods. Journal of Accounting and Economics, 52, 101e125.
El Ghoul, S., Guedhami, O., Kwok, C. C. Y., & Mishra, D. R. (2011). Does corporate Lang, M. H., Raedy, J. S., & Wilson, W. (2006). Earnings management and cross
social responsibility affect the cost of capital? Journal of Banking and Finance, 35, listing: Are reconciled earnings comparable to US earnings? Journal of Ac-
2388e2406. counting and Economics, 42, 255e283.
EY. (2012). Excellence in integrated reporting awards 2012: A survey of integrated Lara, J. M. G., Osma, B. G., & Penalva, F. (2016). Accounting conservatism and firm
reports from South Africa's top 100 companies and top 10 state-owned entities. investment efficiency. Journal of Accounting and Economics, 61, 221e238.
Johannesburg: EY. Larcker, D. F., & Rusticus, T. O. (2010). On the use of instrumental variables in ac-
EY. (2013). Excellence in integrated reporting awards 2013: A survey of integrated counting research. Journal of Accounting and Economic, 49, 186e205.
reports from South Africa's top 100 JSE-listed companies and top 10 state-owned Lee, K. W., & Yeo, G. H. H. (2016). The association between integrated reporting and
companies. Johannesburg: EY. firm valuation. Review of Quantitative Finance and Accounting, 47, 1221e1250.
EY. (2014a). Integrated reporting: Elevating value. London: EY. Lehavy, R., Li, F., & Merkley, K. (2011). The effect of annual report readability on
EY. (2014b). Excellence in integrated reporting awards 2014: A survey of integrated analyst following and the properties of their earnings forecasts. The Accounting
reports from South Africa's top 100 JSE-listed companies and top 10 state-owned Review, 86, 1087e1115.
companies. Johannesburg: EY. Lennox, C. S., Francis, J. R., & Wang, Z. (2012). Selection models in accounting
EY. (2015). Excellence in integrated reporting awards 2015: A survey of the integrated research. The Accounting Review, 87, 589e616.
reports from South Africa's top 100 JSE-listed companies. Johannesburg: EY. Lesmond, D. A. (2005). Liquidity of emerging markets. Journal of Financial Eco-
Fang, V. W., Noe, T. H., & Tice, S. (2009). Stock market liquidity and firm value. nomics, 77, 411e452.
Journal of Financial Economics, 94, 150e169. Leuz, C., & Verrecchia, R. E. (2000). The economic consequences of increased
Financial Reporting Council (FRC). (2014). Guidance on the strategic report. London: disclosure. Journal of Accounting Research, 38, 91e124.
FRC. Leuz, C., & Wysocki, P. D. (2016). The economics of disclosure and financial reporting
Francis, J., Nanda, D., & Olsson, P. (2008). Voluntary disclosure, earnings quality, and regulation: Evidence and suggestions for future research. Journal of Accounting
cost of capital. Journal of Accounting Research, 46, 53e99. Research, 54, 525e622.
Gebhardt, W., Lee, C., & Swaminathan, B. (2001). Toward an implied cost of capital. Levine, R., & Schmukler, S. L. (2006). Internationalization and stock market liquidity.
Journal of Accounting Research, 39, 135e176. Review of Finance, 10, 153e187.
Gietzmann, M., & Ireland, J. (2005). Cost of capital, strategic disclosures and ac- Li, F. (2008). Annual report readability, current earnings, and earnings persistence.
counting choice. Journal of Business Finance and Accounting, 32, 599e634. Journal of Accounting and Economics, 45, 221e247.
Gore, A., & Blood, D. (2011). A manifesto for sustainable capitalism: How business Lundholm, R., & Myers, L. A. (2002). Bringing the future forward: The effect of
can embrace environmental, social and government metrics. The Wall Street disclosure on the returns-earnings relation. Journal of Accounting Research, 40,
Journal. December 14. 809e839.
Gow, I. D., Larcker, D. F., & Reiss, P. C. (2016). Causal inference in accounting Lu, L. Y., Shailer, G., & Yu, Y. (2017). Corporate social responsibility disclosure and the
research. Journal of Accounting Research, 54, 477e523. value of cash holdings. European Accounting Review, 26, 729e753.
Gow, I. D., Ormazabal, G., & Taylor, D. J. (2010). Correcting for cross-sectional and Majmudar, U., & Rana, N. (2017). Richard Howitt talks about the importance of inte-
time-series dependence in accounting research. The Accounting Review, 85, grated reporting in India. The economic times (19 February). Available at: http://
483e512. blogs.economictimes.indiatimes.com/et-commentary/richard-howitt-talks-
Graham, M. (2014). EY's excellence in integrated reporting awards 2013. Johannes- about-the-importance-of-integrated-reporting-for-india/.
burg: EY. Merton, R. C. (1987). A simple model of capital equilibrium with incomplete in-
Greene, W. (1993). Econometric analysis. New York: MacMillan. formation. The Journal of Finance, 42, 483e510.
Hail, L., & Leuz, C. (2006). International differences in the cost of equity capital: Do Muller, K. A., Riedl, E. J., & Sellhorn, T. (2011). Mandatory fair value accounting and
legal institutions and securities regulation matter? Journal of Accounting information asymmetry: Evidence from the European real estate industry.
Research, 44, 485e531. Management Science, 57, 1138e1153.
Hail, L., & Leuz, C. (2009). Cost of capital effects and changes in growth expectations Navissi, F., Sridharan, V. G., Khedmati, M., Lim, E. K. Y., & Evdokimov, E. (2017).
around U.S. cross-listings. Journal of Financial Economics, 93, 428e454. Business strategy, over- (under-) investment and managerial compensation.
Healy, P. M., Hutton, A. P., & Palepu, K. G. (1999). Stock performance and interme- forthcoming Journal of Management Accounting Research.
diation changes surrounding sustained increases in disclosure. Contemporary Ohlson, J. (1995). Earnings, book values and dividends in security valuation.
Accounting Research, 16, 485e520. Contemporary Accounting Research, 11, 661e687.
Hirshleifer, D., & Teoh, S. H. (2003). Limited attention, information disclosure, and Ohlson, J., & Juettner-Nauroth, B. (2005). Expected EPS and EPS growth as de-
financial reporting. Journal of Accounting and Economics, 36, 337e386. terminants of value. Review of Accounting Studies, 10, 349e365.
Howitt, R. (2016). Breakthrough year. IIRC Newsletter. Available at: http://us4. Patten, D. M. (2002). The relation between environmental performance and envi-
campaign-archive1.com/? ronmental disclosure: A research note. Accounting, Organizations and Society, 27,
u¼b36f6aeef75cea67e62812844&id¼ce981ad463&e¼ee7e66415f. 763e773.
Howitt, R. (2017). Interview: Richard Howitt, chief executive of the international in- Plumlee, M., Brown, D., Hayes, R. M., & Marshall, R. S. (2015). Voluntary environ-
tegrated reporting Council. AB magazine (1 may). Available at: http://www. mental disclosure quality and firm value: Further evidence. Journal of Ac-
accaglobal.com/in/en/member/member/accounting-business/2017/05/ counting and Public Policy, 34, 336e361.
interviews/richard-howitt.html. Reed, W. R. (2015). On the practice of lagging variables to avoid simultaneity. Oxford
Institute of Directors (IoD). (2009). King code of governance principles for South Africa Bulletin of Economics and Statistics, 77, 897e905.
64 M.E. Barth et al. / Accounting, Organizations and Society 62 (2017) 43e64

Securities and Exchange Commission (SEC). (2016). Business and financial disclosure of-understanding-240298181.html.
required by Regulation S-K. Available at: https://www.sec.gov/rules/concept/ Velte, P., & Stawinoga, M. (2017). Integrated reporting: The current state of
2016/33-10064.pdf. empirical research, limitations and future research implications. Journal of
Serafeim, G. (2015). Integrated reporting and investor clientele. Journal of Applied Management Control, 28, 275e320.
Corporate Finance, 27, 34e51. Verrecchia, R. E. (2001). Essays on disclosure. Journal of Accounting and Economics,
Simnett, R., Vanstraelen, A., & Chua, W. F. (2009). Assurance on sustainability re- 32, 97e180.
ports: An international comparison. The Accounting Review, 84, 937e967. Welker, M. (1995). Disclosure policy, information asymmetry, and liquidity in equity
South African Institute of Chartered Accountants (SAICA). (2015). Integrated markets. Contemporary Accounting Research, 11, 801e827.
thinking: An exploratory survey. Johannesburg: SAICA. West, A. (2006). Theorising South Africa's corporate governance. Journal of Business
Sustainability Accounting Standards Board (SASB). (2014). SASB and IIRC announce Ethics, 68, 433e448.
memorandum of understanding. PR Newswire (15 January). Available at: http:// Zhou, S., Simnett, R., & Green, W. (2017). Does integrated reporting matter to the
www.prnewswire.com/news-releases/sasb-and-iirc-announce-memorandum- capital market? Abacus, 53, 94e132.