Hummel Et Al. (2019)

J Bus Ethics (2019) 154:733–757
https://doi.org/10.1007/s10551-016-3410-5
ORIGINAL PAPER
The Role of Sustainability Performance and Accounting Assurors

in Sustainability Assurance Engagements
Katrin Hummel1 • Christian Schlick2 • Matthias Fifka3
Received: 2 June 2016 / Accepted: 3 December 2016 / Published online: 6 January 2017
Springer Science+Business Media Dordrecht 2017
Abstract Research on sustainability assurance is still in its sustainability-related processes and systems. Assurance
beginnings. One of the key questions in this field that also providers that do not belong to the accounting profession in
is of the highest practical relevance is concerned with the turn are associated with broader assurance statements.
quality of the assurance process. However, a common
understanding of assurance quality and how it should be Keywords Sustainability assurance Sustainability
measured is still missing. We try to close this gap by reporting Assurance quality Content analysis
building on the financial audit literature. We introduce a Sustainability performance Accounting assuror
definition of assurance quality that comprises two key
aspects: the depth of the assurance process and the breadth
of the assurance statement. Based on prior research, we
hypothesize that a firm’s sustainability performance is Introduction
related to the depth of the assurance process, while the type
of the assurance provider, more precisely the affiliation to In the past two decades, sustainability reporting has con-
the accounting profession, is related to the breadth of the stantly increased and is now a common business practice
assurance statement. Results for a sample of 122 European with 92% of the largest 250 companies worldwide pub-
firms reveal a negative relationship between sustainability lishing stand-alone or integrated sustainability reports in
performance and assurance process depth as well as 2015 (KPMG 2015, p. 22). Likewise, the proportion of
between the affiliation to the accounting profession and firms that seek external assurance of these reports has
assurance statement breadth. Thus, we find evidence that grown from 29% in 2002 to 63% in 2015 among the largest
poor sustainability performers ask for in-depth assurance 250 companies worldwide (KPMG 2015). The question
services, most likely as a means to enhance their internal arises why companies undertake increasing efforts to have
their reports assured, although a financial burden is inevi-
tably connected to it. Potential explanations could be pro-
& Matthias Fifka vided by legitimacy and signaling theory, since companies
matthias.fifka@fau.de
are likely to strive for more legitimacy by signaling to their
Katrin Hummel stakeholders that they can trust the content disclosed in
katrin.hummel@business.uzh.ch
their reports because it has been attested by a third party.
Christian Schlick This theoretical explanation is supported by the fact that
christian.schlick@glemco.de
assurance is voluntary. The voluntary character, however,
1
Department of Business Administration, University of raises questions on the quality of the assurance, as com-
Zurich, Affolternstrasse 56, 8050 Zurich, Switzerland panies might seek to keep costs for it low and still be able
2
Glemco GmbH, Kammergasse 9a, 85354 Freising, Germany to signal that assurance has taken place. Accordingly,
3 Cohen and Simnett (2015, p. 68) note that ‘‘there are many
Institute of Economics, Friedrich-Alexander University
Erlangen-Nuernberg, Kochstrasse 4 (17), 91054 Erlangen, research opportunities as we try to assess the quality of
Germany CSR report assurance engagements.’’
123
734 K. Hummel et al.
For instance, researchers have not yet agreed on a client’s sustainability reporting system. To comprise these
common definition, let alone measurement approach of two aspects of assurance quality—discovering and report-
assurance quality. Moreover, prior research has shown that ing—we investigate the depth of the assurance process and
firm characteristics are related to the adoption of sustain- the breadth of the assurance statement. We concentrate on
ability assurance (Simnett et al. 2009; Kolk and Perego a firm’s sustainability performance and the affiliation to the
2010; Cho et al. 2014; Casey and Grenier 2015), and accounting profession as main firm and provider charac-
provider characteristics are related to the content of teristics that are related to assurance quality. While there is
assurance statements (e.g., O’Dwyer and Owen 2005; only scarce research on the role of sustainability perfor-
Manetti and Toccafondi 2012). Thus, there are strong mance in assurance engagements, the literature on sus-
indications that the client firm and the assurance provider tainability disclosure provides extensive evidence for the
jointly determine the terms of the assurance engagement in important role that sustainability performance plays in
its entirety. Therefore, it is necessary to investigate the role shaping sustainability disclosure (Cho and Patten 2007;
of firm and provider characteristics at the same time to Clarkson et al. 2008). Due to this gap, Kolk and Perego
come to conclusions on their impact. Results from this (2010, p. 195) call for more empirical research on the
investigation can also enlighten the on-going discussion in ‘‘diffusion of assurance services in relation to a firm’s
the literature on the existence of managerial capture in social and environmental performance.’’ According to our
sustainability assurance engagements. Some researchers research, there is currently only one study (Casey and
argue that sustainability assurance is ‘‘captured’’ by the Grenier 2015) that investigates the role of sustainability
management to ‘‘advance the corporate image’’ at the performance for the adoption of external assurance. The
expense of transparency and accountability (Owen et al. results reveal a positive relationship between sustainability
2000; O’Dwyer and Owen 2005). Such capture has severe concerns—indicating a firm’s weaknesses in sustainabil-
ethical implications because it would contradict the ethical ity—and assurance adoption. Two explanations are possi-
standards to which professional accounts are bound when ble: firms with sustainability concerns may ask for external
providing assurance based on the International Standard on assurance to improve on these concerns or for impression
Assurance Engagements 3000 (ISAE 3000). Assurance management purposes. To further disentangle these two
providers in this case must also comply with the IESBA explanations, we investigate the relationship between sus-
Code of Ethics for Professional Accountants, which tainability performance and assurance quality. More pre-
demands professional competence, independence, and cisely, we concentrate on assurance process depth rather
objectivity to ensure assurance quality. In the case of than assurance statement breadth, since firms can particu-
managerial capture, assuror independence would have to be larly influence the depth of the assurance processes during
questioned in particular. their negotiations on the terms of the assurance.
Moreover, accountants are not only held to ethical In order to examine the influence of firm and assurance
conduct, they also evaluate the ethical performance of the provider characteristics, we investigate a sample of 122
companies they audit. It is commonly agreed that aside European firms that obtain external assurance. Besides
from social and environmental aspects, ethics is another firm’s sustainability performance and affiliation to the
vital dimension of sustainability assurance. In theory, this accounting profession, we control for a number of addi-
would mean that assurance could lead to a better ethical tional variables that previous research has identified to
performance of companies, but such assumption does determine assurance adoption and assurance quality,
heavily depend on the motivation of the company that respectively (Simnett et al. 2009; Kolk and Perego 2010;
seeks assurance and the way in which the assurance pro- Zorio et al. 2013; Cho et al. 2014; Casey and Grenier
cess is conducted. Inevitably linked to this is the question 2015). Our analysis is based on multivariate regression
how well ethical performance can actually be measured analyses, and we perform additional statistical checks on
because ‘‘it is impractical to actually observe unethical the robustness of our findings.
behavior for research purposes’’ (Oz 2001, p. 138). Due to In so doing, we contribute to the literature in several
this difficulty, assurance could be regarded as ineffective in ways. First, prior content-based studies on assurance
terms of ethical behavior. statements investigated and discussed the content and dis-
Against this background of the current state of research, closure of each assurance statement element separately, but
we investigate the relationship between firm and provider lack an aggregate measure of assurance quality. We pro-
characteristics and assurance quality empirically. For this vide a definition of assurance quality along with mea-
purpose, we draw on the financial audit literature (DeAn- surement approaches for assurance process depth and
gelo 1981) to provide a definition of assurance quality. assurance statement breadth, which other researchers could
Assurance quality is defined as the joint probability that the use to determine content-based assurance quality. This
assuror will discover and report problematic issues in the measurement approach is central to our paper, as it
123
The Role of Sustainability Performance and Accounting Assurors in Sustainability Assurance… 735
addresses the prevailing difficulty to assess the determi- on results from these studies to identify firm characteristics
nation of assurance quality. Second, there is currently no that may be related to assurance quality.
study that investigates the relationship between sustain- The second group of studies investigates assurance
ability performance and assurance quality, although liter- statements in detail (Ball et al. 2000; Owen et al. 2000;
ature on sustainability disclosure has demonstrated the O’Dwyer and Owen 2005; Deegan et al. 2006; Cooper and
importance of this variable. This lack of research might be Owen 2007; Mock et al. 2007; Manetti and Becatti 2009;
attributable to the general difficulty of measuring sustain- Manetti and Toccafondi 2012; Perego and Kolk 2012;
ability performance. Thus, we rely on an innovative mea- Zorio et al. 2013). These studies are among the first to
surement concept that has been recently introduced by investigate the phenomenon of sustainability assurance
Hummel and Schlick (2016). Third, in the absence of more thoroughly and mainly provide a descriptive over-
legally binding regulations on the assurance process, view of the content and disclosure of certain assurance
results from our study enlighten the discussion on the statement elements (Ball et al. 2000; O’Dwyer and Owen
existence of managerial capture in the literature which has 2005). Some of these studies derive a very critical picture
also ethical implications. In particular, our results reveal of assurance engagements highlighting the existence of
that poor sustainability performers ask for in-depth assur- managerial capture as a ‘‘concept that sees management
ance services, probably as a means to enhance their internal take control of the whole process […] by strategically
sustainability-related systems and process. This finding collecting and disseminating only the information it deems
does not support the previously critical notion of man- appropriate to advance the corporate image, rather than
agerial capture. On the other hand, our results show that being truly transparent and accountable to the society it
accounting assurance providers are associated with lower serves’’ (Owen et al. 2000, p. 85). This would render
assurance statement breadth although they are more likely assurance under managerial capture to be highly unethical,
to apply assurance standards. This finding may be partially because it may lead to a heavily distorted perception of the
related to the use of the ISAE 3000, which has not been company in question.
specifically designed for the sustainability context. Besides Other studies include more complex investigations of
the use of more specific assurance standards, implicit eth- associations between the content and disclosure of
ical and moral standards might also be an effective way to assurance statement elements and firm or provider char-
ensure high assurance quality. acteristics. For instance, Deegan et al. (2006) split up the
In order to pursue our objectives, we proceed as follows. results on assurance statement elements by country and
In the next section, we develop the hypotheses necessary type of assurance provider. Mock et al. (2007) conduct a
for our empirical investigation based on existing literature logistic regression of the type of assurance provider on
and a definition of assurance quality. The subsequent sec- various assurance statement elements. While these studies
tion outlines our research design. In particular, the data have substantially improved our understanding of sus-
sample and empirical model are presented and the variables tainability assurance statements, it is difficult to draw
are introduced. In section four, descriptive statistics are conclusions on the factors that are most closely related to
presented and our empirical findings are discussed. We sustainability assurance statements due to the descriptive
conclude by drawing practical and theoretical implications nature of most studies and omitted variable bias. More-
and make recommendations for future research. over, they investigate the content and disclosure of each
assurance statement element separately, which provides
insights into the dissemination of each assurance state-
Literature Review and Hypotheses Development ment element but at the same time hampers the use of
multivariate statistical analyses. Only Zorio et al. (2013)
Literature Review and Perego and Kolk (2012) base their analyses on an
overall measure of assurance (statement) quality, which is
Research on sustainability assurance includes three pri- calculated as the sum of the coding of various assurance
mary areas of interest.1 Studies on the adoption of assur- statement elements. Our study draws from these two
ance predominantly focus on firm-, industry-, and country- studies. While the measure provided by Zorio et al.
specific determinants (Simnett et al. 2009; Kolk and Perego (2013) refers to the ‘‘quality of the assurance report’’ and
2010; Cho et al. 2014; Casey and Grenier 2015). We draw is based on an assessment of the disclosure versus non-
disclosure of assurance statement elements, the measure
provided by Perego and Kolk (2012) refers to ‘‘assurance
1 quality’’ in general and accounts not only for the dis-
Moreover, experimental studies have examined report users’
perceptions of the reliability of sustainability disclosure and assurance closure but also partly for the content of assurance
(Hodge et al. 2009; Pflugrath et al. 2011). statement elements. We argue that this distinction is
123
important as only the latter allows conclusions that go Assurance Process Depth and Assurance Statement
beyond the assurance statement and include the assurance Breadth
process. We come back to this important distinction in
the next section, when we define assurance quality and Assurance engagements are defined by the International
introduce our measures. Auditing and Assurance Standards Board (International
Apart from that, both measures facilitate the compar- Auditing and Assurance Standard Board (IAASB) 2013,
ison of assurance (statement) quality on an aggregated p. 7) as engagements ‘‘[…] in which a practitioner aims to
level (e.g., countries, industries, assurance providers) and obtain sufficient appropriate evidence in order to express a
the use of multivariate analyses, which are appropriate to conclusion designed to enhance the degree of confidence of
overcome the problem of omitted variable bias. Perego the intended users other than the responsible party about the
and Kolk (2012), for instance, show that assurance quality subject matter information (that is, the outcome of the
on average is highest among firms in Belgium and Swe- measurement of evaluation of an underlying subject matter
den and that accountant assurors provide slightly higher against criteria).’’ Since research on sustainability assurance
assurance quality than certification bodies and specialists. is still in its beginnings, a common understanding, let alone a
To the best of our knowledge, only Zorio et al. (2013) run definition, of assurance quality is still missing. Accordingly,
multivariate regression analyses to investigate the rela- in a recent proposal for future research on sustainability
tionship between firm and provider characteristics and assurance Cohen and Simnett (2015, p. 68) note that ‘‘there
assurance statement quality. Results from their analyses are many research opportunities as we try to assess the
reveal that audit firms (compared to consultants) provide quality of CSR report assurance engagements.’’ Following
higher assurance statement quality and that size is posi- the financial accounting literature (DeAngelo 1981), we
tively and leverage is negatively related to assurance define assurance quality as the joint probability that the
statement quality. assuror will discover and report problematic issues in the
This development towards more complex models to client’s sustainability reporting system. These two aspects
explain assurance (statement) quality is also reflected in of assurance quality—discovering and reporting—are
an increase in qualitative studies that examine the reflected in the depth of the assurance process—referring to
demand and the supply side of sustainability assurance the discovering of issues—and the breadth of the assurance
more thoroughly (Park and Brorson 2005; O’Dwyer statement—referring to the reporting of issues. We derive
2011; O’Dwyer et al. 2011). This third group of studies both variables based on content analysis of the publicly
provides more detailed insights into the process of disclosed assurance statements.
assurance engagements based on qualitative methods Assurance process depth is defined as the intensity of
such as interviews and case studies. Results from these the assurance process. A more intensive assurance process
studies reveal that assurance providers exhibit a strong is more likely to reveal problematic issues. We rely on
commitment to accountability towards external stake- insights from prior research (Park and Brorson 2005;
holders (O’Dwyer 2011; O’Dwyer et al. 2011). Some of Edgley et al. 2010; O’Dwyer et al. 2011) to identify those
these studies also examine companies’ motives for sus- assurance statement elements that describe the intensity
tainability assurance (Park and Brorson 2005; Simnett and scope of the assurance process. We measure assurance
et al. 2009; Edgley et al. 2010), identifying enhanced process depth based on the content of these elements. Since
credibility, refined internal systems or processes, and more intensive assurance services are associated with
improved sustainability performance as primary reasons higher costs, these elements are typically negotiated
for adopting assurance. Similar reasons are mentioned by between the client who bears the extra costs and the
Huggins et al. (2011). By contrast, companies may avoid assurance provider. For instance, Park and Brorson (2005,
adopting assurance because of the additional costs, the p. 1101) report: ‘‘The assurance provider and the reporting
lack of perceived value added, and a lack of stakeholder organization agree on the scope of the assurance, the cri-
demand (Park and Brorson 2005). Taken together, results teria for the assessment, number of site visits and other
from these studies emphasize that both the supply side— practical details.’’ Assurance process depth captures the
i.e., the assurance provider—and the demand side of scope of the assurance, the extent to which key sustain-
assurance engagement—i.e., the client firm—need to be ability-related performance indicators are covered by the
accounted for when analyzing sustainability assurance assurance, the level of the assurance, the variety of the
quality. The rather abstract concept of assurance quality applied methods, whether the materiality of the report is
can be expressed through the two components of assessed, and whether recommendations are provided.
assurance process depth and assurance statement breadth Assurance statement breadth, on the other hand, is
which are described in more detail in the next section. defined as the extent of the assurance statement based on a
123
minimum list of assurance statement elements. A more services. In pure economic terms, the company thus would
comprehensive assurance statement—reflected by assur- have to calculate if the expenses for the assurance process
ance statement breadth—enables the reader to better assess are covered by the degree of legitimacy gained. Although
the assurance process and the results of the assurance. We such calculation is hardly feasible, cost–benefit consider-
rely on the two most important assurance standards, ations are an important factor for a company when it
namely the ISAE 3000 and the AccountAbility 1000 decides on obtaining external assurance and the quality of
Assurance Standard (AA1000AS), and the Global Report- assurance services sought (Park and Brorson 2005; Casey
ing Initiative (GRI) recommendations as well as prior and Grenier 2015; Cohen and Simnett 2015). Certainly, the
research on assurance statements (Ball et al. 2000; Owen question on who is to provide the assurance is crucial in
et al. 2000; O’Dwyer and Owen 2005; Deegan et al. 2006; this regard and companies can chose from a large range of
Cooper and Owen 2007; Mock et al. 2007; Manetti and providers. In this context, Zorio et al. (2013) have found
Becatti 2009; Manetti and Toccafondi 2012; Perego and that the provision of assurance services by an auditor
Kolk 2012; Zorio et al. 2013) to determine the breadth of instead of a consultant increases the assurance quality.
the assurance statement. It is measured based on the dis- Therefore, the nature of the provider is a factor that needs
closure versus non-disclosure of a list of assurance state- to be considered. Likewise, the company’s sustainability
ment elements. Thus, we follow Zorio et al. (2013) by performance as such seems to be a factor that influences the
assessing the disclosure of certain assurance statement assurance process (Park and Brorson 2005). From the
elements in the assurance statement. The measurements of perspective of legitimacy and signaling theory, one must
both assurance process depth and assurance statement assume that companies with a good performance are more
breadth are further elaborated in the research design likely to obtain assurance since they simply have more or
section. more positive information to convey. Nevertheless, one
could also argue that companies with poor performance
Theoretical Background and Hypotheses seek assurance in order to improve their performance and
thus increase their legitimacy in the long run. These con-
From a theoretical perspective, multiple theoretical tradicting assumptions call for a more extensive discussion
approaches have been employed for the research on sus- on the impact of sustainability performance.
tainability assurance (Cohen and Simnett 2015). In par-
ticular, legitimacy theory and stakeholder theory contribute The Role of Sustainability Performance
to the understanding of previous findings on the adoption,
outcome, and process of sustainability assurance engage- While there are currently no empirical studies investigating
ments (Darnall et al. 2009; O’Dwyer et al. 2011). As the relationship between a firm’s corporate sustainability
indicated by the term assurance itself, companies try to performance and assurance quality, the literature on sus-
assure the readers of their reports, who in turn are stake- tainability disclosure provides ample evidence for the
holders of various kinds, that the contents disclosed are true important role that sustainability performance plays in
and have been verified by a third party. This verified dis- determining a firm’s sustainability disclosure (Cho and
closure seeks to obtain the stakeholders’ appreciation of the Patten 2007; Clarkson et al. 2008). With respect to sus-
company’s transparency efforts and thus maintain or tainability assurance, Park and Brorson (2005) have iden-
increase its legitimacy (Weaver et al. 1999). tified the improvement of a company’s actual sustainability
The key mechanism in this process is signaling. Sig- performance as major reason why companies seek external
naling theory (Spence 1973, 2002) implies that sustain- assurance. As a consequence, Kolk and Perego (2010,
ability reporting serves as a signal to the stakeholders about p. 195) call for an empirical investigation of the ‘‘diffusion
the company’s behavior and performance and attempts to of assurance services in relation to a firm’s social and
convince readers of its social responsibility (Elitzur and environmental performance.’’ To the best of our knowl-
Gavious 2003; Jain 2016). The question of credibility is edge, there is only one empirical study (Casey and Grenier
crucial in this regard, as the reader has no possibility— 2015) that investigates the role of sustainability perfor-
unless the report has been examined by non-governmental mance in sustainability assurance engagements, yet this
organizations or journalists—to judge if the contents dis- study is limited to the adoption of sustainability assurance
closed are actually true. Therefore, assurance itself serves and does not investigate assurance quality in more detail.
as a signal easily to detect by the reader to convey the Based on a U.S. sample of 2659 firm-year observations for
company’s truthfulness. a period of 18 years, Casey and Grenier (2015) find posi-
However, the possibility to signal comes at a certain tive relationships for both a firm’s sustainability strengths
cost, as the assurance provider, who is elementary in the and a firm’s sustainability concerns, i.e., weaknesses, and
process, charges a considerable amount of money for his the probability of obtaining external assurance. While the
123
positive relationship between sustainability strengths and obtaining assurance, even if only of insignificant depth, is
assurance adoption indicates that there is no need to gain sufficient as a signal to be sent in order to maintain legit-
extra credibility through assurance in case of low sustain- imacy. In this case, a negative relationship between sus-
ability performance, two explanations could be offered for tainability performance and assurance process depth exist,
the positive relationship between sustainability concerns which is why Hypothesis 1b is stated as follows:
and assurance adoption. Firms may ask for additional
H1b There is a negative relationship between corporate
assurance to enhance credibility and improve their internal
sustainability performance and assurance process depth.
sustainability-related systems or they may use sustain-
ability assurance for impression management purposes. We frame hypotheses H1a and H1b for the dependent
Since assurance adoption does not account for the quality variable assurance process depth (as against assurance
of the assurance engagement, it is difficult to determine statement breadth) because we expect firms’ influence to be
which of these explanations apply. most prevalent during negotiations with the assurance
Examining assurance quality, namely assurance process provider on the terms of the assurance engagement.
depth, instead of assurance adoption thus enables us to Assurance process depth is thus primarily determined by
investigate these explanations more thoroughly. On the one the firm based on their motives for assurance adoption and
hand, a positive relationship between a firm’s sustainability cost–benefit considerations, whereas assurance statement
performance and assurance process depth could be found if breadth is primarily determined by the assurance provider
sustainability assurance is obtained to enhance credibility because the assurance statement is written and delivered by
among stakeholders or comply with common industry the assurance provider.
practices of adopting at least some type of assurance. In
this case, superior sustainability performers obtain in-depth The Role of the Assurance Provider
assurance services, whereas poor sustainability performers
ask for superficial assurance services as in-depth assurance We identify the affiliation to an accounting versus non-
might disclose the weaknesses, and a rather superficial accounting assurance provider as provider characteristic
assurance can generate some legitimacy. that has most commonly been studied in the literature, yet
Moreover, signaling theory also suggests that good with differing results. The question whether there are
sustainability performers are more likely to obtain in-depth fundamental differences in the assurance statements pro-
assurance service because this allows them to send stronger vided by different types of assurance providers has been at
signals. As Spence (1973) argued, the more one can signal, the forefront of studies that investigate assurance state-
the more convincing signaling is. As good sustainability ments more thoroughly, i.e., the second group of studies in
performance leads to profound assurance in this case, and the literature review. Most of these studies differentiate
poor performance to low quality assurance, we develop between accounting versus other types of assurance pro-
Hypotheses 1a as follows: viders (such as consultants, non-profit organizations, and
certification bodies) and report summary statistics for each
H1a There is a positive relationship between corporate
assurance statement element for the different groups of
sustainability performance and assurance process depth.
assurance providers (Ball et al. 2000; O’Dwyer and Owen
However, the relationship between sustainability per- 2005; Deegan et al. 2006; Mock et al. 2007; Manetti and
formance and assurance process depth can also be dia- Toccafondi 2012). Some of the reported differences might
metric, as existing literature has suggested. If firms with a relate to institutional differences between accountant and
poor sustainability record are interested in improving their non-accountant companies, for instance, the application of
internal systems, processes, and sustainability perfor- a certain assurance standard that is mandatory for profes-
mance, they can obtain assurance services of high quality sional accountants. Others might reflect cultural differ-
to achieve these goals and to strengthen their credibility ences. For instance, O’Dwyer and Owen (2005, p. 214)
(Park and Brorson 2005). Such a relationship would sub- reckon that accountant assurors are less likely to compre-
stantially differ from the critical notion of sustainability hensively report on their competence in the assurance
assurance as a process that is ‘‘captured’’ by the manage- statement because these assurors believe that the name of
ment to merely enhance the corporate image. Superior the firm is sufficient ‘‘to indicate competence to undertake
sustainability performers in turn might also seek less pro- the assurance process.’’ Overall, results from prior studies
found assurance services because of the conviction that are ambiguous with respect to the direction of these dif-
their efforts are convincing by themselves and that in-depth ferences. Recently, Huggins et al. (2011) comprehensively
assurance does not create additional value for improvement discuss the pros and cons of accounting versus non-ac-
or reputation against the background of cost–benefit con- counting assurance providers with respect to the assurance
siderations. Thus, the underlying assumption is that of greenhouse gas information. They argue that accounting
123
assurance providers have high levels of quality controls Hypothesis H2b therefore relates to a negative relationship
and stringent assurance procedures, whereas non-account- between the affiliation to an accounting assurance provider
ing assurors have technical expertise and specific knowl- and assurance statement breadth.
edge of the subject matter (i.e., greenhouse gas reporting).
H2b There is a negative relationship between affiliation
The authors conclude that multidisciplinary teams are
to an accounting assurance provider and assurance state-
particularly beneficial with respect to the assurance of
ment breadth.
greenhouse gas information.
Hypothesis H2a draws on studies that reveal more rig- Note that hypotheses H2a and H2b are aimed at assur-
orous disclosure of assurance statement elements by ance statement breadth as dependent variable (and not
accounting versus non-accounting assurance providers and assurance process depth) because we expect providers’
thus posits a positive relationship. For instance, Manetti influence to become particularly evident with respect to the
and Toccafondi (2012) reveal that 29% of the non-ac- breadth of the assurance statement. After all, it is the
counting assurance providers include no reference to the assurance provider who authors and delivers the assurance
applied assurance standard compared to 0% of the statement, whereas assurance process depth is determined
accounting assurance providers and 41% of the non-ac- by both the client firm and the assurance provider. We
counting assurance providers do not state the assurance further investigate this distinction in the section ‘‘Results
level compared to only 9% of the accounting assurors. from regression analysis.’’
With respect to overall measures of assurance quality, both
Perego and Kolk (2012) and Zorio et al. (2013) report
higher assurance quality by accounting versus non-ac- Research Design
counting assurance providers. Such a positive relationship
could stem from a higher expertise of accounting assurance Sampling and Empirical Model
providers in formal aspects of the assurance including the
complete disclosure of a minimum list of assurance state- Our initial sample consists of 195 companies located in
ment elements. We formally state this hypothesis as France, Germany, Italy, Spain, Sweden, Switzerland, or
follows: the United Kingdom. To be included, a company must be
listed in the Bloomberg European 500 index as of January
H2a There is a positive relationship between affiliation to
2013, provide sufficient information on its sustainability
an accounting assurance provider and assurance statement
performance for the reporting year 2011, and belong to an
breadth.
industry group that contains at least five companies in
However, there are also studies that find the opposite total. Both prerequisites stem from the measurement
results. Accounting assurance providers may provide approach of corporate sustainability performance that we
assurance on a broad range of topics including sustainability, apply in this study and which has recently been intro-
whereas non-accounting assurance providers are often spe- duced by Hummel and Schlick (2016). Corporate sus-
cialized in sustainability assurance. As a consequence, non- tainability performance is measured based on
accounting assurance providers may have a higher expertise environmental and social key performance indicators
in sustainability assurance and thus may provide more which a firm is required to report in order to be included
comprehensive assurance statements including important in the sample. Moreover, the rescaling of the original
sustainability-specific statement items, such as the materi- values requires an industry group of at least five compa-
ality of the report. Perego and Kolk (2012), for instance, nies per industry to define reasonable peer groups. Of
provide some descriptive evidence indicating that non-ac- these 195 companies, 68% seek external assurance. Fur-
counting assurance providers are more elaborate and infor- ther, of these 133 companies, eleven are excluded because
mative when it comes to formulating recommendations and the corresponding assurance statements could not be
conclusions. Moreover, O’Dwyer and Owen (2005) report retrieved from corporate websites, a web search engine, or
that consultant assurors more frequently rely on assurance by direct request. The sample selection is summarized in
standards in the assurance process, include references on Panel A of Table 1. Our final sample includes 122
their independence in the assurance statement, report on the European companies across 18 industry groups. The
performed procedures of the assurance process, and com- sample composition is depicted in Panel B of Table 1.
ment on the completeness of the assured report. Mixed Our sample thus includes only firms that obtain external
findings are provided by Mock et al. (2013, p. 291) who show assurance because we are interested in the relationships
negative and significant correlations between the provision between firm and provider characteristics and assurance
of positive assurance and recommendations and Big4 firms quality, respectively. While this procedure is common in
which constitute a large portion of the accounting assurors. research on sustainability assurance (e.g., O’Dwyer and
123
Table 1 Sample selection and sample composition

Panel A: Sample selection
Initial Bloomberg European 500 sample 500
Less: firms not domiciled in France, Germany, Italy, Spain, Sweden, Switzerland, or the United Kingdom -112
388
Less: firms with insufficient data on sustainability performance -151
237
Less: firms that belong to an industry group with less than five companies -42
Sample size (Hummel/Schlick 2016) 195
Less: firms that do not obtain external assurance -62
133
Less: firms of which the assurance statement cannot be obtained -11
Final sample size 122
By country n By industry n
Panel B: Sample composition by country and industry

1 France 26 1 Chemicals 6
2 Germany 18 2 Building materials, paper, steel 10
3 Italy 12 3 Mining 8
4 Spain 19 4 Advertising, entertainment, media 4
5 Sweden 5 5 Telecommunication 9
6 Switzerland 11 6 Computers and software 3
7 United Kingdom 31 7 Car manufacturers 4
8 Retail 4
9 Food and beverages 5
10 Pharmaceuticals 6
11 Oil and gas 8
12 Banking and financial services 18
13 Insurance 4
14 REITS 4
15 Aerospace and defense 5
16 Engineering and construction 9
17 Electricity 7
18 Gas, water, and electricity distribution 8
Total 122 Total 122
Owen 2005; Deegan et al. 2006; Manetti and Toccafondi breadth ¼ b0 þ b1 sustainability performance
2012), we acknowledge that it could induce selection bias þ b2 accounting provider
which could drive our main findings and we address this X
i¼6
concern in our robustness section. þ bi controls
We use the following econometric models to test the i¼3
hypotheses H1a, H1b, H2a, and H2b: X
j¼23 kX
¼29
þ bj industry þ bk country þ e: ð2Þ
depth ¼ b0 þ b1 sustainability performance j¼7 k¼24
þ accounting provider
The variables included in our empirical model are sum-
X
i¼6 X
j¼23
marized in Table 2 along with descriptions and data sour-
þ bi controls þ bj industry
i¼3 j¼7
ces and described in greater detail in the next sections. The
kX
¼29 control variables include firm financial performance, firm
þ bk country þ e: ð1Þ size, firm leverage, provider independence, and both
k¼24 industry and country dummies. The data sources comprise
123
Table 2 Variables overview

Variable Description Source of data
Dependent variables
depth Assurance process depth, measured as an index variable as described in Table 4 and Assurance statement,
in the ‘‘Measurement of assurance statement breadth and assurance process sustainability disclosure
depth’’ section
breadth Assurance statement breadth, measured as an index variable as described in Table 3 Assurance statement
and in the ‘‘Measurement of assurance statement breadth and assurance process
depth’’ section
Main variables of interest
sustainability_performance Firm’s sustainability performance, measured as described in Hummel and Schlick Sustainability disclosure
(2016) and in the ‘‘Variables measurement’’ section
accounting_provider Dummy variable indicating whether the assurance provider is an accounting Assurance statement
company (1) or not (0)
Control variables
financial_performance Financial performance of the client, measured as the free Cash flow (cash flow from Bloomberg database
operating activities - total capital expenditures) in millions of euros per
employee at the end of the fiscal year
size Firm size of the client, measured as the log of total employees at the end of the fiscal Annual report,
year sustainability disclosure
leverage Financial leverage of the client measured as (average total assets/average total Bloomberg database
common equity) during the fiscal year
provider_independence Independence of the assuror, measured as a binary index variable indicating Assurance statement,
potentially threatened independence (0) or not (1) annual report
industry Industry group dummy variables as reported in Panel B of Table 1, and the Bloomberg database,
reference category is banking and financial services annual report
country Country of domicile dummy variables as reported in Panel B of Table 1, and the Bloomberg database
reference category is the UK
assurance statements for the measurement of our dependent AccountAbility 1000 Assurance Standard (AA1000AS),
variables, firm’s sustainability disclosure (i.e., corporate and the GRI recommendations, we determine thirteen
sustainability reports, annual reports, and integrated assurance statement elements along four categories which
reports) for the measurement of firm’s sustainability per- are coded on a binary basis and summed up to the overall
formance, assurance statements and firm’s annual reports measure assurance statement breadth (breadth). Table 3
to determine provider independence and firm size, and the provides an overview of these thirteen assurance statement
Bloomberg database to obtain data on the financial per- elements.
formance, leverage, and industry and country affiliation of General information covers three assurance statement
the firm. elements, namely whether the assurance statement is
addressed towards all stakeholders (B1), whether the
Measurement of the Dependent Variables assurance statement references an assurance standard (B2),
and whether miscellaneous information is completely
Assurance Statement Breadth mentioned in the assurance statement (B3). Three assur-
ance statement elements refer to the disclosure of provider
Based on previous research on assurance statement quality information, such as the description of the provider’s
(O’Dwyer and Owen 2005; Manetti and Toccafondi 2012; competence (B4) and independence (B5), and information
Perego and Kolk 2012; Mock et al. 2013; Zorio et al. 2013) on both the firm’s and the provider’s area of responsibility
and assurance standards, namely the International Standard (B6). Three assurance statement elements cover the breadth
on Assurance Engagements 3000 (ISAE 3000)2 and the of information that is provided on the assurance process,
namely on the scope (B7), the methods (B8), and the level
2
Our sample covers the reporting year 2011 and we thus refer to the
ISAE 3000 as of 2004 (International Auditing and Assurance Footnote 2 continued
Standard Board (IAASB) 2004). Note that the standard was revised (IAASB) 2013) and the revised standard became effective for assur-
in 2013 (International Auditing and Assurance Standard Board ance reports dated on or after December 15, 2015.
123
Table 3 Measurement of assurance statement breadth

Item Measurement AA ISAE GRI
1000 3000
AS
General information
B1 Addressees Whether the assurance statement is addressed towards all stakeholders (1) or not (0) (X)a (X)a (X)a
B2 Assurance Whether at least one assurance standard is mentioned as framework for the assurance (1) or not X X X
standard (0)
B3 Miscellaneous Whether miscellaneous information (provider’s firm name, provider’s representative, date, and Xb Xc Xd
information place) are completely mentioned in the assurance statement (1) or not (0)
Provider information
B4 Provider Whether the assurance statement contains information on the provider’s competence (1) or not X (X) –e
competence (0)
B5 Provider Whether the assurance statement contains information on the provider’s independence (1) or X X –e
independence not (0)
B6 Area of Whether the assurance statement contains information on both the firm’s and the provider’s X X X
responsibility area of responsibility (1) or not (0)
Assurance process
B7 Scope Whether the assurance statement contains information on the scope of the assurance process X X X
(1) or not (0)
B8 Methods Whether the assurance statement contains information on the methods and procedures of the X X X
assurance (1) or not (0)
B9 Level Whether the assurance statement specifies the level of the assurance (1) or not (0) X X X
Conclusions
B10 Veracity Whether the assurance statement indicates the conclusion/opinion of the assurance provider (1) X X X
or not (0)
B11 Materiality Whether the assurance statement contains information on the materiality of the report (1) or Xf –g X
not (0)
B12 Limitations Whether the assurance statement contains limitations with respect to the sustainability report X X X
(1) or not (0)
B13 Recommendations Whether the assurance statement contains recommendations (1) or not (0) Xh (X) X
Total 0–13
a
Only the addressing of the intended users is mandatory
b
Provider’s representative not contained
c
Provider’s firm name or provider’s representative
d
Only provider’s representative and date
e
Only mentioned as important characteristic of assurance providers
f
Represented by the principles of inclusivity, materiality, and responsiveness
g
ISAE 3000 is not a specialized sustainability assurance standard and requires a conclusion with respect to the specific assurance engagement
h
Observations and/or recommendations
() Brackets indicate non-mandatory elements
(B9) of the assurance. The category conclusions include information) is coded based on whether miscellaneous
four assurance statement elements, namely whether a information is completely provided in the assurance state-
conclusion/opinion (B10), information on the materiality of ment (1) or not (0). Taken together, assurance statement
the report (B11), limitations (B12), and recommendations breadth ranges in the interval between zero (lowest assur-
(B13) are disclosed. Of the thirteen assurance statement ance statement breadth) and thirteen (highest assurance
elements, eleven elements are coded based on disclosure statement breadth) and measures the extent to which an
(1) versus non-disclosure (0). One assurance statement assurance statement enables its readers to assess the
element (addressees) is coded based on whether the assurance process. Nine assurance statement elements are
assurance statement is addressed towards all stakeholders mandatory assurance statement elements according to
(1) or not (0). Another statement element (miscellaneous ISAE 3000 and AA1000AS. Addressing the intended users
123
Table 4 Measurement of assurance process depth

Item Measurement
Assurance process
D1 Scope The assurance covers
(3) The full report
(2) Numerical values and qualitative statements
(1) Numerical values or qualitative statements
D2 Coverage Proportion of the number of assured sustainability performance indicators relative to the number of reported
sustainability performance indicators
D3 Methods The sum of the following methodologies that are applied in the assurance process: documentation checks, interviews,
data analyses, on-site visits, and public media search
D4 Level The assurance level indicated
(3) High/reasonable
(2) Mixed
(1) Moderate/limited
(0) None
Conclusions
D5 Materiality Whether the assurance statement contains information on the materiality of the report (1) or not (0)
D6 Recommendations Whether the assurance statement contains information indicating that recommendations are provided to the client (1)
or not (0)
Total 0–6
of the assurance statement is mandatory according to both assurance statement breadth, because they provide impor-
ISAE 3000 and AA1000AS and recommended by the GRI, tant information to the reader and have also been included
yet the standards do not specify the intended users. We in prior content-based studies on assurance statements
nevertheless include addressing the assurance statement to (O’Dwyer and Owen 2005; Perego and Kolk 2012).
all stakeholders as an element of assurance statement
breadth, since the divergence between the addressees of the Assurance Process Depth
sustainability report and of the corresponding assurance
statement has repeatedly been observed and criticized in Our measurement of assurance process depth primarily
the literature (Ball et al. 2000; O’Dwyer and Owen 2005). derives from qualitative research on the assurance process,
Two assurance statement elements (provider’s competence i.e., the third group of studies in the literature review. We
and recommendations) are suggested, although they do not measure assurance process depth (depth) based on the con-
constitute mandatory elements of an assurance statement, tent of the assurance statement elements scope, methods,
according to ISAE 3000. Moreover, one assurance state- level, materiality, and recommendations. In addition, we
ment element (materiality) is not mentioned in the ISAE include coverage which measures the proportion of assured
3000. sustainability performance indicators relative to the number
Our measure aims at capturing the breadth of the of reported sustainability performance indicators. Table 4
assurance statement as one aspect of assurance quality. presents an overview of the measurement of each item of
Nevertheless, it could be argued that this measure is biased assurance process depth. Assurance process depth is mea-
towards the use of the AA1000AS standard, since it com- sured as the sum of the six items. If necessary, the elements
prises assurance statement elements that are not required are proportionally rescaled on a [0, 1] interval. Assurance
by ISAE 3000, in particular the assessment of materiality process depth thus ranges in the interval between zero
and the provision of recommendations. The importance of (lowest assurance process depth) and six (highest assurance
the principle of materiality with respect to sustainability process depth).
disclosure quality is reflected in the release of the GRI The first element scope refers to the question of which
sustainability disclosure standard G4 and has also been disclosures are covered by the assurance engagement.
highlighted by researchers (e.g., Hummel and Festl-Pell Scope indicates whether the assurance covers numerical
2015). The disclosure of limitations and recommendations values or qualitative statements (equals ‘1’), numerical
in the assurance statement enables the readers to compre- values and qualitative statements (equals ‘2’), or the full
hensively assess the current state of sustainability disclo- report (equals ‘3’). Similar to scope, the element coverage
sure. We thus include these elements in our measure of captures the number of core sustainability performance
123
indicators that are covered by the assurance. Drawing on highlights the advisory nature of this statement element.
our measurement scheme for sustainability performance The interviewed assurance providers in his case study
outlined in the ‘Measurement of the independent variables’ emphasize: ‘‘That is very important, because that is really
section, we measure coverage as the proportion of the what the clients pay for.’’ (O’Dwyer 2011, p. 1247) It is
number of assured core sustainability performance indica- crucial to note that recommendations of the assurance
tors relative to the number of reported core indicators. We process depth variable slightly differs from the assurance
argue that a more comprehensive assurance scope and a statement breadth element as the latter includes only rec-
higher coverage of core sustainability indicators reflect a ommendations that are explicitly described in the assurance
deeper and more intensive assurance process. The third statement, while the former includes all sorts of
assurance intensity element methods captures the presence/ recommendations.
absence of five assurance methods, namely documentation Similar to potential criticism regarding our assurance
checks, interviews, data analyses, on-site visits, and public statement breadth measure, it might be argued that our
media search, and thus ranges in the interval between zero measure of assurance process depth is positively biased
(no methods are mentioned) and five (all categories are towards the use of AA1000AS with subsequent effects on
mentioned). Previous research highlights that in addition to the relationship between the accounting assurance provider
the traditional methods of interviews, documentation and assurance process depth. We in turn argue that both the
checks and data analyses, both on-site visits as well as assessment of the materiality of the sustainability report
public media search are important for ensuring the quality and the provision of recommendations are important
of sustainability assurance (Park and Brorson 2005; Edgley aspects of an in-depth assurance engagement and thus
et al. 2010; O’Dwyer 2011). For instance, on-site visits reflect assurance quality. Moreover, in our robustness
help assurance providers to evaluate the rigor and quality section we provide additional analyses addressing these
of the data collection processes (Park and Brorson 2005, concerns.
p. 1102) as well as the implementation of codes of conduct In order to determine assurance statement breadth and
(O’Dwyer 2011, p. 1246). Public media search on the other assurance process depth of the 122 companies in the final
hand enables assurors to critically reflect on the content of sample, we conducted a directed content analysis of the
the firm’s sustainability disclosure against stakeholder assurance statements in question (Weber 1990; Hsieh and
criticism and potentially concealed sustainability-related Shannon 2005). To guarantee the validity and reliability of
risks (Edgley et al. 2010, p. 546). We therefore argue that the results obtained, two assumptions hold: (1) all assur-
assurance based on a broader spectrum of different ance statements are independent of each other4; and (2) the
methodologies correspond with a deeper assurance elements can be coded into a nominal scale and are
processes. mutually exclusive and exhaustive (Stemler 2001). To rule
The fourth element level corresponds to the assurance out researcher bias, the assurance statements were coded
engagement risk of having not detected any circumstances independently by two researchers using the binary scoring
that may call into question the assuror’s opinion on the method described above (Campbell and Swinscow 2009).
objects being examined. According to AA1000AS, ‘‘the Finally, we moderated the results of the coders and
assurance provider achieves high assurance where suffi- aggregated the results.
cient evidence has been obtained to support their statement
such that the risk of their conclusion being in error is very Measurement of the Independent Variables
low but not zero.’’3 A higher level of assurance requires
more intensive assurance procedures to sufficiently reduce Corporate Sustainability Performance
assurance engagement risk. Materiality refers to an assur-
ance engagement in which the assuror is requested to To measure corporate sustainability performance (sustain-
assess whether the sustainability report reflects ‘‘the orga- ability_performance) we follow the approach developed by
nization’s significant economic, environmental and social Hummel and Schlick (2016). Based on the GRI reporting
impacts’’ (Global Reporting Initiative (GRI) 2013, p. 17). guidelines, the authors derive eight core sustainability
Finally, recommendations refer to suggested improvements performance indicators that can be applied to firms in
to a company’s sustainability reporting, systems, and pro- different industry groups. The scheme comprises four
cesses. The preparation of recommendations requires environmental and four social performance indicators that
additional work by the assuror and thus indicates a more
intensive assurance engagement. O’Dwyer (2011) 4
One may argue that assurance statements provided by the same
assurance provider are not independent from each other and we
3
Instead of ‘‘high’’ and ‘‘moderate,’’ the ISAE 3000 proposes the address this concern in the robustness section by including provider
terms ‘‘reasonable’’ and ‘‘limited’’. fixed effects in our models.
123
are evaluated based on hand-recorded data. The environ- Prior research has shown a positive relationship between
mental performance indicators include energy consump- the firm’s financial performance and the voluntary adoption
tion, water withdrawal, greenhouse gas emissions, and total of sustainability assurance (Simnett et al. 2009; Ruhnke
weight of waste. The social performance indicators include and Gabriel 2013). On the other hand, Zorio et al. (2013)
employee turnover, lost time incident rate, employee do not obtain a significant relationship between firm’s
training, and share of women in the highest corporate financial performance and assurance statement quality.
bodies.5 After the raw indicator data are winsorized at a This insignificant result could be attributed to the focus on
10% level and proportionally [0, 1] normalized on an assurance statement in contrast to assurance process, which
industry group basis, the indicators are aggregated into the we argue is more influenced by the assurance provider than
total sustainability performance score by calculating their the assurance seeking firm. We measure financial perfor-
mean. Thus, sustainability_performance is a continuous mance (financial_performance) as the free cash flow in
variable that ranges in the interval between zero and one millions of euros per employee at the end of the fiscal year.
and higher values indicate higher sustainability perfor- Since higher assurance quality is associated with higher
mance. An overview of the measurement approach is costs, we expect a positive relationship between a firm’s
provided in Table 5. Note, that we calculate the sustain- financial performance and assurance quality, in particular
ability performance variable for the sample of 195 Euro- assurance process depth.
pean firms which includes both firms that obtain external We also control for firm size, which researchers argue
assurance and firms that do not obtain external assurance. serves as a proxy for public pressure (e.g., Patten 1991) and
This approach to measuring sustainability_performance firms may respond to increased pressure through the
is especially useful for our research setting because sus- adoption of sustainability assurance which has been found
tainability reports typically refer to both environmental and by Casey and Grenier (2015), Darnall et al. (2009), Ruhnke
social dimensions. Firms seek to protect their perceived and Gabriel (2013), and Simnett et al. (2009). Firm size
legitimacy on both dimensions by increasing their sus- (size) is measured as the decimal logarithm of the number
tainability report’s credibility through external assurance. of employees at the end of the fiscal year and we expect to
Thus, the applied measurement scheme directly addresses find a positive relationship between firm size and assurance
the two dimensions and comprehensively covers sustain- quality, in particular assurance process depth.
ability performance within them. Next, we include the financial leverage of a company,
which refers to the informational needs of a company’s
Accounting Assurance Provider creditors. Because creditors of a highly leveraged firm may
demand higher security regarding sustainability-related
accounting_provider is a dummy variable that equals 1 if risks to assess the long-term success of a company, lever-
the assurance provider belongs to an accounting firm and 0 age may be positively related to assurance quality. How-
if not. Accounting firms typically comprise the Big4 ever, high leverage may also indicate financial difficulties
companies, i.e., Deloitte, Ernst & Young, KPMG, or or a shareholder structure dominated by myopic investors
PricewaterhouseCoopers, and other small accounting firms. who demand cost cutting. Such a negative relationship
Non-accounting assurance providers are service companies between leverage and assurance adoption has been found
that are typically specialized in consulting, reporting, and by Casey and Grenier (2015). We measure leverage
assurance for non-financial areas, for example, SGS, (leverage) as the average total assets divided by average
Bureau Veritas, and Corporate Citizenship. total common equity. Due to these contradictory theoretical
considerations, we do not predict the sign of the corre-
Control Variables sponding regression coefficient.
In addition, we include the independence of the
In addition to our main variables of interest, we include the assurance provider as additional control variable since
following control variables: firm financial performance, findings from the financial audit literature suggest that
firm size, firm leverage, independence of the assurance limited auditor independence corresponds to lower audit
provider, country of origin, and industry affiliation. quality (DeAngelo 1981; Blay and Geiger 2013). This
finding is qualified by studies that report no relationship
or even a negative relationship depending on market
conditions and audit firm rotation (Daniels and Booker
2011; Jamal and Sunder 2011). We argue that an inde-
pendent assurance provider can more easily stipulate
5
A detailed description of the performance indicators is provided by more comprehensive assurance services for the self-
Hummel and Schlick (2016). serving purpose of increasing revenues. Consequently, we
123
Table 5 Measurement of corporate sustainability performance (Hummel and Schlick 2016)

Performance indicator Measurement Unit
Environmental dimension
Energy consumption (Direct ? indirect energy consumption)/number of employeesa MW/
employee
Water withdrawal (Total water withdrawal - cooling water)/number of employeesa m3/
employee
Greenhouse gas emissions (GHG protocol scope 1 ? scope 2 emissions)/number of employeesa t/employee
Total weight of waste Total weight of waste/number of employeesa t/employee
Social dimension
Employee turnover Total number of employeesa who leave/number of employeesa 9 100 Percent
Lost time incident rate Number of incidents resulting in lost time from work/(total hours worked/200,000) Incidents/h
Employee training Total training time/number of employeesa h/
employee
Share of women in the highest Total number of women in the highest corporate bodiesb/total number of members of the Percent
corporate bodiesb highest corporate bodiesb 9 100
a
If available, the number reflects full-time equivalents and, otherwise, headcount
b
Management board and supervisory board for two-tier system countries, board of directors for one-tier system countries
expect to find a positive relationship between assuror Results

independence and assurance process depth. Common
proxies for auditor independence in the financial audit Descriptive Statistics
literature are often derived from audit fees (Blay and
Geiger 2013; Markelevich and Rosner 2013). However, a Panel A of Table 6 presents descriptive statistics on the
similar approach is not feasible with respect to the disclosure of the thirteen assurance statement elements.
measurement of assuror independence because companies Overall, assurance statement breadth is rather high, since
do not separately disclose their expenses for sustainabil- eight assurance statement elements, namely assurance
ity assurance services. We therefore rely on four condi- standard, miscellaneous information, provider indepen-
tions that might indicate constraints on assuror dence, area of responsibility, scope, methods, level, and
independence. Our variable provider_independence indi- veracity, exhibit mean values above 0.75. On the other
cates potentially threatened provider independence if at hand, in line with previous studies (O’Dwyer and Owen
least one of these constraints is present (equals ‘0’) or 2005; Deegan et al. 2006; Cooper and Owen 2007; Manetti
none is present (equals ‘1’). These conditions include and Toccafondi 2012) only 13% of the assurance state-
additional sustainability-related consulting or assurance ments are addressed towards all stakeholders and only 57%
services, congruence between assuror and financial describe the provider’s competence. Moreover, only 37%
statement auditor, the characterization of the engagement of the assurances cover the materiality of the report, only
as an iterative process, and a negative or missing dec- 17% of the assurance statements identify limitations on the
laration of independence (AccountAbility 2008; IFAC sustainability report and only 44% delineate recommen-
2013). dations. Taken together, assurance statements on average
Moreover, many studies have demonstrated the impact provide sufficient information on general issues, the
of institutional country-level factors on sustainability assurance provider, and the assurance process, but lack
practices (Fifka and Drabble 2012; Fifka and Pobizhan information regarding the conclusion of the assurance.
2014) in general, but also on the probability of assurance Panel B of Table 6 presents descriptive statistics on
adoption in specific (Kolk and Perego 2010; Cho et al. the six assurance process depth elements scope, coverage,
2014). Thus, we include country dummy variables to methods, level, materiality, and recommendations. While
account for potential differences arising from different the data in Panel A depict the disclosure of the assurance
regulatory, cultural, and societal environments. Finally, statement elements, the data in Panel B refer to the
sustainability practices as well as accountability and content of the assurance statement elements and thus
transparency toward external stakeholders also vary across capture the depth and intensity of the assurance
industries (Reverte 2009; Kolk 2010; Matisoff et al. 2013; engagement. The descriptive statistics on scope reveal
Zorio et al. 2013). We account for this variability by also that only 7% of the sample firms seek assurance of the
including industry group dummies. full report, while most companies ask for assurance only
123
Table 6 Descriptive statistics

n Min Max Mean SD
for model variables
Panel A: Descriptive statistics for the assurance statement breadth elements
B1 Addressees 122 0.00 1.00 0.13 0.34
B2 Assurance standard 122 0.00 1.00 0.93 0.26
B3 Miscellaneous information 122 0.00 1.00 0.79 0.41
B4 Provider competence 122 0.00 1.00 0.57 0.50
B5 Provider independence 122 0.00 1.00 0.80 0.40
B6 Area of responsibility 122 0.00 1.00 0.97 0.18
B7 Scope 122 0.00 1.00 1.00 0.00
B8 Methods 122 0.00 1.00 0.98 0.13
B9 Level 122 0.00 1.00 0.89 0.31
B10 Veracity 122 1.00 1.00 1.00 0.00
B11 Materiality 122 0.00 1.00 0.37 0.48
B12 Limitations 122 0.00 1.00 0.17 0.38
B13 Recommendations 122 0.00 1.00 0.44 0.50
0 1 2 3 4 5 Mean SD
Panel B: Descriptive statistics for the assurance process depth elements (original values before rescaling)
D1 Scope – 31% 61% 7% – – 1.76 0.58
D2 Coverage – – – – – – 0.80 0.26
D3 Methods 2% 0% 9% 37% 44% 8% 3.47 0.89
D4 Level 11% 69% 17% 3% – – 1.13 0.63
D5 Materiality 63% 37% – – – – 0.37 0.48
D6 Recommendations 45% 55% – – – – 0.55 0.50
n Min Max Mean SD
Panel C: Descriptive statistics for the regression variables

(1) breadth 122 3.00 12.00 9.04 1.58
(2) depth 122 1.43 5.80 3.04 1.06
(3) sustainability_performance 122 0.06 0.93 0.55 0.18
(4) accounting_provider 122 0.00 1.00 0.82 0.39
(4) financial_performance 122 -1.97 1.17 0.02 0.31
(7) size 122 1.93 5.46 4.56 0.68
(8) leverage 122 -72.03 62.48 6.04 10.98
(6) provider_independence 122 0.00 1.00 0.26 0.44
Panel A of the table descriptive statistics for the elements included in the dependent variable breadth. Panel
B of the table descriptive statistics for the elements included in the dependent variable depth. Note that the
categories ‘0’ to ‘5’ do not apply to coverage which is a continuous variable. Panel C shows breadth and
depth on an aggregated level along with the other variables included in the regression analysis
on selected numerical values and/or qualitative state- sustainability reports. A more positive picture is revealed
ments. Similarly, most firms (69%) request a moderate by the results for coverage, methods, and recommenda-
level of assurance, 17% rely on mixed levels, and only tions. Of the core sustainability indicators covered by our
3% seek a high assurance level. The remaining 11% of measurement scheme for sustainability_performance, on
assurance statements do not provide an assurance level, average, 80% are verified by an external assuror. More-
indicating a superficial assurance engagement of low over, more than three different classes of evidence-
depth. In addition, only 37% of the firms mandate their gathering methods are applied on average in the assur-
assurance providers to assess the materiality of their ance process by the assurors. Finally, 55% of the firms
123
are also provided with recommendations on future provider independence, and area of responsibility. These
improvements. assurance statement elements are more often disclosed in
Panel C of Table 6 presents the descriptive statistics for the group of poor sustainability performers compared to
the regression variables. The positive findings on the high sustainability performers. Not in line with our
assurance statement breadth elements directly translate into expectations, there are no significant differences in the
a mean of 9.04 and a standard deviation of 1.58 for the assurance process depth elements between poor and high
dependent variable breadth. The mixed findings of the sustainability performers.
assurance process depth elements result in a mean of 3.04 With respect to the assurance provider, the results reveal
and standard deviation of 1.06 for the dependent variable that accounting assurance providers more often give
depth. The variable sustainability_performance ranges information on the assurance standard, area of responsi-
between a minimum value of 0.06 and a maximum value of bility, the methods, and the level of the assurance and
0.93, with a mean of 0.55. These results are similar to those complete information on miscellaneous issues, such as
of Hummel and Schlick (2016) and do not indicate selec- provider name and representative, date, and place. The
tion bias in our sample with respect to sustainability per- higher disclosure of the assurance level by accounting
formance. With respect to our second variable of interest— assurors is also reflected in the assurance process depth
accounting_provider—we find 82% of the assurance pro- element level. On the other hand, non-accounting assurors
viders are accounting firms and all of them belong to the more often address all stakeholders, describe the provider
Big4 accounting firms, namely Deloitte, Ernst & Young, competence, provide information on the materiality of the
KPMG, or PricewaterhouseCoopers. Non-accounting report, and disclose recommendations in the assurance
assurance providers in our sample include Bureau Veritas, statement. Again, these differences are also reflected in the
DNV, Two Tomorrows,6 ERM, LRQA, AENOR, Corpo- assurance process depth elements materiality and recom-
rate Citizenship, and Sustainability and Excellence. With mendations. These findings are primarily in line with
respect to the control variables, firms’ financial perfor- findings provided by Perego and Kolk (2012, p. 184) who
mance exhibit substantial heterogeneity, which results from describe that accounting assurors are more likely to dis-
the variety of industry groups contained in the sample. close formal assurance statement elements, such as the
Descriptive statistics on size (original values, untabulated) applied assurance standard, information on the provider
show that 73,430 employees work for our sample firms on firm name, and area of responsibility, but are more hesitant
average during 2011 and we also find substantial hetero- in disclosing information on the materiality examination
geneity with respect to the capital structure of our sample and recommendations. Some of the differences reported in
firms. Finally, provider_independence is rather low with a Table 7 might stem from the preferred use of ISAE 3000
mean of 0.26. Low levels of provider independence reflect by accounting assurors and AA1000AS by non-accounting
the nature of sustainability assurance engagements, i.e., assurors, in particular differences in the disclosure of
providers are often engaged in additional consulting or materiality assessments and we address this concern in the
auditing services and iterative interaction with their clients. robustness section. Taken together, while there are a
Table 10 in the appendix presents the correlation matrix for number of significant differences in the assurance state-
the regression model variables which indicates no multi- ment elements between accounting and non-accounting
collinearity problems. assurance providers, we find only limited support for our
assumption that the firm’s sustainability performance also
Results from Univariate Analysis plays an important role with respect to the assurance
engagement. In the next section, we will test our
We use a t test of means to identify significant differences hypotheses based on multiple linear regression models.
in the assurance process depth elements and the assurance
statement breadth elements between poor and high sus-
tainability performers and non-accounting and accounting Results from Regression Analysis
assurance providers. For this purpose, we divide our sam-
ple into poor and high sustainability performers based on The results from multiple regression analyses with robust
the median sustainability performance. The results are standard errors (White 1980) are shown in Table 8. We
presented in Table 7. present stepwise regression models, gradually introducing
The results reveal significant differences between poor country and industry fixed effects. The results presented in
and high sustainability performers with respect to the dis- column III and column VI correspond to our full models,
closure of the assurance standard, the disclosure on i.e., Eqs. (1) and (2), respectively. Columns I–III refer to
assurance process depth, and columns IV–VI refer to
6
In May 2012, Two Tomorrows was acquired by DNV. assurance statement breadth.
123
Table 7 Results of t tests of mean assurance statement breadth elements and assurance process depth elements
Poor High t value Non-accounting Accounting t value
sustainability sustainability assurors assurors
performers performers (n = 22) (n = 100)
(n = 63) (n = 59)
Mean SD Mean SD Mean SD Mean SD
Panel A: Assurance process depth elements

D1 Scope 1.7619 0.4988 1.7627 0.6523 -0.0077 1.7727 0.6853 1.7600 0.5527 0.0935
D2 Coverage 0.8298 0.2426 0.7600 0.2662 1.5154 0.7976 0.2727 0.7957 0.2532 0.0323
D3 Methods 3.4603 0.9302 3.4746 0.8581 -0.0878 3.2273 1.3428 3.5200 0.7585 -1.3984
D4 Level 1.1587 0.5738 1.1017 0.6872 0.4988 0.8636 0.8335 1.1900 0.5631 -2.2389*
D5 Materiality 0.3968 0.4932 0.3390 0.4771 0.6574 0.5455 0.5096 0.3300 0.4726 1.9090*
D6 Recommendations 0.5873 0.4963 0.5085 0.5042 0.8700 0.9091 0.2942 0.4700 0.5016 3.9509***
Panel B: Assurance statement breadth elements
B1 Addressees 0.1429 0.3527 0.1186 0.3261 0.3929 0.4545 0.5096 0.0600 0.2387 5.5102***
B2 Assurance standard 0.9683 0.1767 0.8814 0.3261 1.8455* 0.8184 0.3948 0.9500 0.2190 -2.1649**
B3 Miscellaneous information 0.8095 0.3958 0.7671 0.4291 0.6268 0.5000 0.5118 0.8500 0.3589 -3.8112***
B4 Provider competence 0.5556 0.5009 0.5763 0.4984 -0.2288 0.7273 0.4558 0.5300 0.5016 1.6961*
B5 Provider independence 0.9048 0.2959 0.6949 0.4644 2.9959*** 0.9091 0.2942 0.7800 0.4163 1.3784
B6 Area of responsibility 1.0000 0.0000 0.9322 0.2536 2.1229** 0.8636 0.3513 0.9900 0.1000 -3.1063***
B7 Scope 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
B8 Methods 0.9841 0.1260 0.9831 0.1302 0.0464 0.9091 0.2942 1.0000 0.0000 -3.1363***
B9 Level 0.9365 0.2458 0.8475 0.3626 1.5966 0.6818 0.4767 0.9400 0.2387 -3.7219***
B10 Veracity 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
B11 Materiality 0.3968 0.4932 0.3390 0.7743 0.6574 0.5455 0.5096 0.3300 0.4726 1.9090*
B12 Limitations 0.2063 0.4079 0.1356 0.3453 1.0306 0.2727 0.4558 0.1500 0.3589 1.3800
B13 Recommendations 0.4444 0.5009 0.4407 0.5007 0.0415 0.8182 0.3848 0.3600 0.4824 4.1551***
The table presents descriptive statistics on all assurance process depth elements (Panel A) and assurance statement breadth elements (Panel B) for
poor versus high sustainability performers and non-accounting versus accounting assurors. Poor sustainability performers are firms with a
sustainability performance score equal to or below the median. Accounting assurors are those assurance providers that are affiliated to the
accounting profession
***, **, and * Significance at the 1, 5, and 10% level, respectively (two-sided tests)
With respect to our first set of hypotheses, we find a the contract is signed, but not the assurance statement
negative relationship between sustainability performance which is ultimately delivered by the assurance provider.
and assurance process depth, which supports H1b. On Regarding our second set of hypotheses, the results
average, poor sustainability performers seek more in-depth reveal a negative and significant relationship between the
assurance services to enhance their internal systems, pro- affiliation to an accounting assurance provider and assur-
cesses, and sustainability performance. These poor sus- ance statement breadth, which supports hypothesis H2b.
tainability performers benefit the most from an intensive Assurance statements delivered by non-accounting assur-
assurance engagement, because they are more likely to face ance providers are broader compared to assurance state-
legitimacy threats and need to demonstrate the seriousness ments delivered by accounting assurors which is in line
of their endeavors. Accordingly, superior sustainability with the findings provided by O’Dwyer and Owen (2005).
performers on average demand less in-depth assurance Except for one model specification (column II), we obtain
services owing to cost–benefit considerations. As expected, this result only for assurance statement breadth but not for
we find this relationship only for assurance process depth, assurance process depth. This finding again supports our
but not for assurance statement breadth. This finding sup- reasoning that assurance providers particularly influence
ports our reasoning that client firms primarily influence the the breadth of the assurance statement as they ultimately
depth of the assurance engagement which is determined deliver the assurance statement, whereas their influence on
based on negotiations with the assurance provider before the depth of the assurance process is limited. This finding is
123
Table 8 Regression results: main model

(I) (II) (III) (IV) (V) (VI)
depth depth depth breadth breadth breadth
sustainability_performance H1a: ?; H1b: - -1.0115* -0.8681* -1.1945*** -1.2233 -0.4678 -0.5105

(-1.8685) (-1.7829) (-2.8152) (-1.6502) (-0.7219) (-0.7172)
accounting_provider H2a: ?; H2b: - -0.4246 -0.6078** 0.1563 -1.2440*** -1.2692*** -1.5422***
(-1.5181) (-2.1790) (0.6149) (-3.2479) (-3.4184) (-3.6168)
financial_performance ? 0.4749 0.4823 1.1026*** 0.0596 0.1695 0.2900
(1.3465) (1.5378) (3.5167) (0.1233) (0.4061) (0.5514)
size ? 0.1566 0.0488 0.3269* 0.4842** 0.4016** 0.6206**
(0.9483) (0.3230) (1.9419) (2.1408) (1.9989) (2.1970)
leverage ? -0.0153 -0.0202** -0.0288*** 0.0073 0.0082 0.0251
(-1.6147) (-2.4499) (-3.1284) (0.5634) (0.7493) (1.6256)
provider_independence ? 0.5287** 0.2336 0.4943** 0.5112 0.0093 -0.0858
(2.3286) (1.1253) (2.5973) (1.6440) (0.0335) (-0.2687)
intercept 3.4765*** 3.7776*** 2.9367*** 8.3785*** 8.4325*** 7.1247***
(4.3623) (5.3002) (3.4903) (7.6774) (8.8891) (5.0476)
industry dummies No No Yes No No Yes
country dummies No Yes Yes No Yes Yes
observations 122 122 122 122 122 122
2
adjusted R 0.0908 0.3228 0.5285 0.0988 0.3702 0.3147
F statistic 3.0137*** 5.8055*** 5.6765*** 3.2099*** 6.9275*** 2.9163***
The table reports ordinary least squares coefficient estimates and t statistics (in parentheses) based on Huber–White robust standard errors. ***,
**, and * indicate statistical significance at the 1, 5, and 10% level, respectively (two-sided tests)
more in line with the results obtained by O’Dwyer and Nevertheless, higher assurance statement breadth is not
Owen (2005), who observe a higher disclosure rate of necessarily associated with higher costs and is thus not
assurance statement items (including the assurance stan- related to the firm’s financial performance. We also find a
dard applied, provider’s competence, and provider’s inde- positive relationship between size and both assurance
pendence) by consultant versus accounting assurance statement breadth and assurance process depth. In this
providers. In this regard, more recent investigations tend to context, firm size serves as a proxy for public pressure.
reveal a higher assurance statement quality by accounting Larger firms may deal with this pressure through broader
versus consultant assurors (Perego and Kolk 2012; Zorio assurance statements, providing additional credibility
et al. 2013). We caution our readers to interpret these among their stakeholders.
findings against the background of the measure used for Leverage is negatively related to assurance process
assurance statement quality and assurance statement depth, which is consistent with the results provided by
breadth, respectively. In particular, our measure of assur- Casey and Grenier (2015). One potential reason for this
ance statement breadth might be positively biased towards negative relationship could be stronger bank monitoring
the use of AA1000AS, which might affect the relationship that restrains highly leveraged firms from seeking in-depth
between assurance provider and assurance statement (and expensive) assurance services. With respect to
breadth. We address this concern in the robustness section provider_independence, the results show a positive and
of the paper. significant relationship with assurance process depth. This
With respect to our control variables, there is a positive finding may indicate that independent assurors presumably
relationship between firms’ financial performance and possess more negotiating power and thus are more likely to
assurance process depth, yet only for the full model convince the client to purchase more in-depth assurance
including industry and country fixed effects. This finding is services (O’Dwyer 2011; O’Dwyer et al. 2011). However,
in accordance with results obtained by Simnett et al. (2009) there is no significant relationship between independence
and Ruhnke and Gabriel (2013) on assurance adoption. As and assurance statement breadth, which is contrary to our
expected, more in-depth assurance services are more reasoning that provider characteristics are more strongly
costly, and firms’ financial performance seems to deter- related to assurance statement breadth than to assurance
mine the capacity to bear these additional expenses. process depth. Finally, consistent with previous studies, we
123
find significant differences (not tabulated) depending on external assurance (n = 122) and those firms that are
industry (Simnett et al. 2009; Cho et al. 2014) and country excluded from our sample and do not obtain external
(Simnett et al. 2009; Kolk and Perego 2010; Ruhnke and assurance (n = 73).7 The results indicate no significant
Gabriel 2013) dummies. differences between these two groups with respect to the
firm characteristics except for firm size. Next, we assign
Analyses of Robustness the dependent variables breadth and depth as well as
accounting_provider and provider_independence the value
We perform a number of supplemental analyses to check of zero for our subsample of firms that do not obtain
the robustness of our results. First, we re-run the regression external assurance and re-run the regression analyses for
analyses with alternative measures for sustainability per- the full sample of 195 firms. The results from this analysis
formance, financial performance, size, and leverage. More (untabulated) remain similar to our main findings except
precisely, we winsorize the sustainability performance data for the relationship between accounting_provider and
at the 5 and 1% levels instead of 10%, and we use return on breadth and depth which is now positive. However, this
assets and return on equity as alternative proxies for positive relationship can be explained by the scoring since
financial performance, a log-transformation of the market these variables are scored zero (instead of missing value)
capitalization and total assets as alternative proxies for firm for the subsample of firms that do not obtain external
size, and total debt to total assets as alternative proxy for assurance. Therefore, the results from this second set of
leverage. Moreover, we run specifications with financial robustness tests alleviate our concerns that our results are
performance and leverage being winsorized at the 1, 5, and biased due to the sampling procedure.
10% level to minimize the effects of potential outliers. Third, we investigate whether our main results are driven
None of these variations substantial alter our main findings. by the applied assurance standard. One may argue that the
We also disaggregate our provider independence measure relationship between affiliation to the accounting assurance
into four binary variables that correspond with the condi- provider and assurance statement breadth may be caused by
tions of our original variable. This procedure provides spurious correlations due to the omission of the applied
valuable insights into how the various aspects of inde- assurance standard. In particular, our measurement of
pendence relate to assurance statement breadth and assur- assurance statement breadth might be positively biased to the
ance process depth, respectively. use of the assurance standard AA1000AS since all assurance
While our main results on sustainability performance statement elements are required by AA1000AS, but not by
and accounting assuror do not change, the results reveal ISAE 3000 and professional accountants affiliated with
that the declaration of independence and additional con- national accountants’ associations that are members of the
sulting activities are both negatively related to breadth and International Federation of Accountants (IFAC) are obli-
there are no significant relationships between the other gated to consider ISAE 3000 in their work. Therefore,
binary disaggregated variables of independence and our accounting assurors are more likely to apply ISAE 3000,
dependent variables. We also experiment with different whereas non-accounting assurors are more likely to apply
clusterings, namely by country and accounting_provider. AA1000AS which might cause the negative relationship
The results tend to be robust, yet the relationship between between accounting provider and assurance statement
accounting_provider and assurance statement breadth breadth. We run a number of tests to investigate this concern.
becomes insignificant when clustering by country. We also First, we investigate differences in the use of the two stan-
re-run the regression analyses including provider fixed dards for accounting and non-accounting assurors based on
effects instead of the distinction between accounting versus contingency tables and the corresponding Chi square
non-accounting assurors. The results reveal negative and statistic. The results reveal indeed that accounting assurors
significant relationships with assurance statement breadth use ISAE 3000 more and AA1000AS less frequently than
for some (KPMG, Ernst & Young, and Pricewater- non-accounting assurors (Panel A of Table 9). Next, we re-
houseCoopers) but not all accounting assurance providers. run the regression analyses including additional dummy
Taken together, this first set of robustness tests suggests variables indicating the use of the ISAE 3000 and
that our main results are robust to the measurement of our AA1000AS standards.8 The results from this analysis reveal
independent variables.
7
Second, we investigate whether the sample selection More precisely, this subsample of 73 firms includes both firms that
might bias our findings since we exclude firms from our do not obtain external assurance and firms that obtain assurance but
do not provide the corresponding assurance statement.
analysis that do not obtain external assurance. First, based 8
More precisely, ISAE 3000 equals ‘‘1,’’ if the ISAE 3000 standard
on a t test of means we investigate whether there are fun-
is applied and ‘‘0’’ otherwise. Accordingly, AA1000AS equals ‘‘1,’’ if
damental differences in the firm characteristics between the AA1000AS standard is applied and ‘‘0’’ otherwise. Please note
firms that are included in our sample and thus obtain that this model specification has some multicollinearity concerns due
123
positive relationships between the use of the ISAE 3000 and We also investigate whether our assurance process depth
breadth for all model specifications and positive relationship measure is influenced by the assurance level since ISAE
between the use of the AA1000AS and both depth and 3000 No. 49 states: ‘‘The degree to which the practitioner
breadth for all model specifications (columns I–VI of Panel B considers each of these components is affected by the
of Table 9). Nevertheless, our main result on the relationship engagement circumstances, in particular […] whether a
between accounting provider and assurance statement reasonable assurance or a limited assurance engagement is
breadth still holds. Next, we specifically address the concern performed.’’ (International Auditing and Assurance Stan-
that our measures for assurance statement breadth does not dard Board (IAASB) 2004) For this purpose, we investigate
score highest for assurance statements that are completely in differences in the assurance process depth elements between
line with the reporting requirements of the applied assurance those firms with a limited assurance level and those firms
standard, in particular ISAE 3000. Therefore, we exclude the with a reasonable assurance level. Results from a t test of
assurance statement elements addressees, materiality, and means do not indicate significant differences between the
recommendations from the measurement of assurance two groups, except for coverage. Overall, the results from
statement breadth. We also exclude materiality and recom- this third set of robustness checks attenuate potential con-
mendations from the measurement of assurance process cerns regarding the influence of the applied assurance stan-
depth, as these elements are not required by the assurance dards in our models.
standard ISAE 3000. We re-run the analyses with the alter- Finally, we address potential endogeneity concerns
native proxies breadth1 (excluding B1), breadth2 (excluding resulting from the possibility that firms might choose their
B4), breadth3 (excluding B11), breadth4 (excluding B13), assurance provider based on the provider’s reputation. In
and breadth5 (excluding B4, B11, and B13) as well as depth1 this case, the relationship between sustainability perfor-
(excluding D5), depth2 (excluding D6), and depth3 (ex- mance and assurance process is mediated by the accounting
cluding D5 and D6) instead of assurance statement breadth provider which might influence our results. We therefore
and assurance process depth, respectively. The full results build a structural model in which sustainability perfor-
are displayed in Panel C of Table 9. mance influences assurance process depth both directly and
Except for depth2 and depth3, the results from our base- via the accounting provider. We include all control vari-
line models still hold. This robustness analysis provides two ables except for industry and country dummies. Figure 1
important insights. First, the negative relationship between displays our results which we obtain from covariance-
accounting assurance provider and assurance statement based structural modeling (Bentler and Chou 1987).
breadth persists, even if we exclude those assurance state- We caution the readers to carefully interpret the results
ment elements that are not mandatory according to ISAE since the overall model fit appears to be rather poor
3000. This finding demonstrates that the negative correlation (RMSEA = 0.193 and CFI = 0.339, respectively). Never-
between accounting assurance provider and assurance theless, it is important to note that both the direct and the total
statement breadth is not primarily driven by the applied effects between sustainability performance and assurance
measurement of assurance statement breadth, in particular process depth are negative and significant. This additional
by the inclusion of elements that are only mandatory analysis alleviates potential concerns regarding the rela-
according to AA1000AS, but also holds for a less compre- tionship between sustainability performance and assurance
hensive measurement of assurance statement breadth. Sec- process depth, but endogeneity problems cannot be ruled out
ond, the negative relationship between sustainability completely. In order to do so, panel data or an experimental
performance and assurance process depth does not hold for study would be necessary. Thus, it is necessary to
an alternative measure of assurance process depth that does acknowledge that potential endogeneity concerns might
not account for the provision of recommendations. This remain. Nevertheless, the overall results from the additional
finding indicates that the provision of recommendations analyses largely support the robustness of our findings.
seems to be critical for the relationship between sustain-
ability performance and assurance process depth, which
further corroborates our reasoning that poor sustainability
performers obtain in-depth assurance services to enhance Conclusions
their internal sustainability-related processes and systems.
For such enhancements, the provision of recommendations is This paper empirically investigates the relationships between
of great importance. firm sustainability performance and the affiliation of the
assurance provider to the accounting profession and assurance
quality. We define assurance quality as the joint probability
Footnote 8 continued
to the strong correlation between the assurance provider and the use that the assuror will discover and report problematic issues in
of the ISAE 3000 standard. the client’s sustainability reporting system and distinguish
123
Table 9 Results of the robustness tests addressing potential bias due to the applied assurance standard
Non-accounting assurors (n = 22) Accounting assurors (n = 100) Total
Panel A: Contingency tables

ISAE 3000 = 0 17 (77%) 12 (12%) 29 (24%)
ISAE 3000 = 1 5 (23%) 88 (88%) 93 (73%)
Subtotal 22 (100%) 100 (100%) 122 (100%)
v2 = 42.40 p = 0.000
AA1000AS = 0 13 (59%) 79 (79%) 92 (75%)
AA1000AS = 1 9 (41%) 21 (21%) 30 (25%)
Subtotal 22 (100%) 100 (100%) 122 (100%)
v2 = 3.85 p = 0.050
(I) (II) (III) (IV) (V) (VI)
depth breadth depth breadth depth breadth
Panel B: Regression results with ISAE 3000 and AA1000AS as additional control variables
sustainability_performance H1a: ?; H1b: - -1.2077*** -0.7065 -0.7254* -0.0468 -0.7268* -0.0668
(-2.7962) (-0.9990) (-1.8218) (-0.0803) (-1.8023) (-0.1251)
accounting_provider H2a: ?; H2b: - 0.1160 -2.1391*** -0.4040* -1.3147*** -0.4177 -2.0372***
(0.3389) (-3.8184) (-1.7021) (-3.7816) (-1.3103) (-4.8275)
ISAE 3000 ? 0.0498 0.9770* 0.0181 1.3840***
(0.1655) (1.9840) (0.0646) (3.7228)
AA1000AS ? 1.6252*** 1.9407*** 1.6261*** 2.0855***
(7.4731) (6.0931) (7.3858) (7.1562)
controls Yes Yes Yes Yes Yes Yes
industry dummies Yes Yes Yes Yes Yes Yes
country dummies Yes Yes Yes Yes Yes Yes
observations 122 122 122 122 122 122
adjusted R2 0.5162 0.3245 0.5708 0.5042 0.5645 0.5743
F statistic 5.3039*** 2.9374*** 6.3632*** 5.1022*** 6.0593*** 6.2655***
(I) (II) (III) (IV) (V) (VI) (VII) (VIII)

depth1 depth2 depth3 breadth1 breadth2 breadth3 breadth4 breadth5
Panel C: Regression results for variations of the dependent variables

sustainability_performance -0.9096 *** -0.4380 -0.0499 -0.2414 -0.5376 -0.4214 -0.3250 -0.1646
(-2.6796) (-1.0584) (-0.1960) (-0.3700) (-0.8005) (-0.6867) (-0.5896) (-0.4152)
accounting_provider -0.1300 -0.0071 0.1283 -1.0907*** -1.3758*** -1.4171*** -1.0919*** -0.7443***
(-0.6357) (-0.0284) (0.8442) (-2.7750) (-3.4005) (-3.8334) (-3.2883) (-3.1163)
controls Yes Yes Yes Yes Yes Yes Yes Yes
industry dummies Yes Yes Yes Yes Yes Yes Yes Yes
country dummies Yes Yes Yes Yes Yes Yes Yes Yes
observations 122 122 122 122 122 122 122 122
adjusted R2 0.3344 0.2604 0.1543 0.3979 0.2917 0.2109 0.3339 0.2937
F statistic 3.0959*** 2.4691*** 1.7550*** 3.7575*** 2.7184*** 2.1151*** 3.0911*** 2.7352***
Panel A of the table contingency tables and the Pearson Chi square statistics and corresponding p values. Panel B of the table reports ordinary
least squares coefficient estimates and t statistics (in parentheses) based on Huber–White robust standard errors. Columns (I)–(VI) refer to model
specifications additionally controlling for the applied assurance standard. Panel C reports ordinary least squares coefficient estimates and
t statistics (in parentheses) based on Huber–White robust standard errors. Columns (I)–(VIII) present results based on an alternative measure for
depth and breadth, namely depth1 excluding D5, depth2 excluding D6, depth3 excluding D5 and D6, and breadth1 excluding B1, breadth2
excluding B4, breadth3 excluding B11, breadth4 excluding B13, and breadth5 excluding B4, B11, and B13
***, **, and *Statistical significance at the 1, 5, and 10% level, respectively (two-sided tests)
123
sustainability systems and obtain less profound assurance

accounting
services due to cost–benefit considerations.
provider
These significant findings are only obtained for assur-
ance process depth, but not for assurance statement
breadth, which further supports our reasoning. We find
sustainability -0.9280*
further evidence for the importance of cost–benefit con-
(-1.8723)
depth
performance
siderations with respect to assurance engagements, as firm
financial performance is positively related to assurance
process depth. Moreover, the results do not hold for an
control alternative measure of assurance process depth that
variables excludes the provision of recommendations. Once again,
this indicates that enhancements of the internal processes
Fig. 1 Results from structural equation modeling. This figure displays and systems are a major reason for poor sustainability
the path coefficients and corresponding z statistics (in parentheses) for performers to obtain in-depth assurance services. The
the structural equation model. ***, **, and *statistical significance at results from our study thus do not provide indications that
the 1, 5, and 10% level, respectively (two-sided tests)
the assurance process has been ‘‘captured’’ by the man-
agement. This is to be seen positively from an ethical
assurance quality into assurance process depth—referring to perspective, as there are no indications for a managerial
the discovering of issues—and assurance statement breadth— attempt to positively distort the image of the company by
referring to the reporting of issues. Assurance statement hiding critical information from the stakeholders.
breadth thus describes the extent of the assurance statement With respect to our second hypothesis, we find a negative
and is measured based on the disclosure versus non-disclosure relationship between the affiliation to the accounting provi-
of a list of minimum assurance statement elements which we der and assurance statement breadth. Our sample data thus
determine based on the assurance standards AA1000AS and indicate that non-accounting assurance providers are asso-
ISAE 3000. Assurance process depth in turn describes the ciated with higher assurance statement breadth, which might
intensity of the assurance process and is measured based on result from a more specialized knowledge in non-financial
the content of those assurance statement elements that refer to audit services. We perform a number of robustness tests to
the process and conclusion of the assurance. investigate whether this finding is driven by the applied
We hypothesize that firm sustainability performance is assurance standard, since non-accounting assurors predom-
related to assurance process depth, because firms negotiate inantly apply the assurance standard ISAE 3000, which has a
on the terms of the assurance engagement, which is reflected generic focus and does not specifically address sustainability
by the depth of the assurance process. Affiliation to the reporting. While the results from these additional tests
accounting profession is argued to be related to assurance indeed indicate that the applied assurance standard is cor-
statement breadth, as it is the assurance provider that ulti- related with assurance statement breadth, the negative rela-
mately writes and delivers the assurance statement. Since tionship between accounting assuror and assurance
existing theory and prior literature provide ambiguous statement breadth still holds. Moreover, the negative rela-
results, we have no clear expectations regarding the sign of tionship also persists for alternative measures of assurance
the relationships. Results from a sample of 122 European statement breadth that exclude those disclosure items that are
companies reveal a negative relationship between firm sus- only mandatory according to AA1000AS. While we
tainability performance and assurance process depth. One acknowledge that our measure of assurance statement
potential explanation for this finding stems from prior stud- breadth can be criticized for this bias towards the use of
ies’ findings on the reasons for and against obtaining sus- AA1000AS, we provide a comprehensive measure for
tainability assurance (Park and Brorson 2005; Edgley et al. assurance statement breadth. This measurement of assurance
2010). Firms obtain external assurance to enhance their statement breadth might thus come along with some quality
credibility and improve their internal sustainability-related judgments on the governing assurance standards. However,
systems and processes. Particularly poor sustainability per- assurance providers in principle are free to apply both ISAE
formers are in need of these improvements, and in-depth 3000 and AA1000AS to achieve assurance processes and
assurance processes are more likely to result in these statements of higher quality.
improvements, which partly explains the negative relation- Our findings and conclusions are moderated by certain
ship. Superior sustainability performers, on the contrary, limitations. First, the sample is restricted to large listed
obtain external assurance primarily to maintain their credi- companies. Thus, we cannot account for other companies
bility and legitimacy among stakeholders. They are not in that may receive less public attention and pursue different
need of in-depth assurance processes to enhance their legitimization strategies. In addition, companies that do not
123
obtain external assurance are excluded from the sample, also be worthwhile to examine differences in assurance
which might bias our results. However, results from addi- quality across countries and over a longer reporting period,
tional robustness tests provide no indication for such which would generate valuable insights into the potential
sample selection bias. Furthermore, our sample is restricted impact of the institutional environment and time trends.
to countries from highly industrialized European countries, Moreover, potential endogeneity concerns regarding the
and assurance practices may be different in other regions. relationship between sustainability performance and assur-
Second, data collection on both assurance process depth ance process depth could also be mitigated by using lagged
and assurance statement breadth relies on information variables and panel data. Second, a closer investigation of
contained in sustainability assurance statements, and this the negotiations between the client firm and the assurance
information is assumed to be true and complete. Although provider on the terms of the assurance engagement is a vital
common assurance standards provide guidance on assur- avenue for future research. To the best of our knowledge,
ance statement elements, latitude for interpretation and there are no studies on this phase yet, although the results
discretionary decisions by the assurors and their clients from our study provide some indication that firms try to
remains. Moreover, since there is currently no common influence the assurance engagement during these negotia-
approach for measuring assurance quality, our measures of tions. Therefore, future research could provide deeper
assurance statement breadth and assurance process depth insights into this stage of the assurance engagement, pos-
can be criticized for including/excluding certain assurance sibly based on interviews. Such a research design also
statement elements from our measures. We try to mitigate allows to investigate the composition of the assurance teams
this criticism by focusing on a minimum list of assurance more thoroughly and the process of how the auditor is
statement elements which are—except for four elements— chosen and which criteria are used in the selection phase.
specified by the two leading sustainability assurance stan- Interviews and case studies could also serve to investigate
dards. Besides, our measure of assurance process depth is the motives that drive companies to seek assurance and the
closely related to insights provided by prior case-based goals they pursue by obtaining it, and thus deliver further
research. In addition, we perform several robustness tests to insights into the existence of managerial capture and the
investigate these concerns and provide further insights into negative ethical implications connected to it. From an eth-
the data. Third, the affiliation to the accounting profession ical perspective, it would also be interesting to examine how
is a binary variable that is determined based on the assur- well the assurance process works with regard to assessing
ance company. Thus, we cannot account for the possibility the ethical performance of the companies audited. Due to
of multidisciplinary teams, which (Huggins et al. 2011) the complexity of measuring and observing ethical behavior
argue to be most beneficial in certain assurance contexts. (Oz 2001), this provides a particular challenge. Overall,
Fourth, our data are limited to a single reporting year and there are manifold opportunities for research on sustain-
we thus cannot completely rule out the possibility that the ability assurance, as it still is a rather unexplored field.
relationship between sustainability performance and
Compliance with ethical standards
assurance quality may be endogenous.
The limitations of our study indicate the potential for Conflict of interest The authors declare that they have no conflict of
future research. First, we strongly encourage researchers to interests.
apply our measures for assurance process depth and
assurance statement breadth to firms domiciled in other
countries as well as to non-listed companies. In particular, Appendix
the assurance practices of small- and medium-sized enter-
prises are still a blank spot on the research map. It could See Table 10.
Table 10 Correlation statistics

(1) (2) (3) (4) (5) (6) (7) (8)
(1) 1.0000
(2) 0.6974 1.0000
(0.0000)
(3) -0.1672 -0.1286 1.0000
(0.0657) (0.1580)
(4) -0.1371 -0.1793 -0.0561 1.0000
(0.1321) (0.0482) (0.5393)
123
Table 10 continued
(1) (2) (3) (4) (5) (6) (7) (8)
(5) 0.0759 0.1333 0.0981 -0.0754 1.0000

(0.4061) (0.1432) (0.2825) (0.4093)
(6) 0.1217 0.0507 0.0372 0.2799 0.3357 1.0000
(0.1818) (0.5794) (0.6841) (0.0018) (0.0002)
(7) -0.0323 -0.0996 -0.1165 0.1802 0.1838 0.0000 1.0000
(0.7239) (0.2753) (0.2013) (0.0471) (0.0427) (0.9998)
(8) 0.1743 0.2152 -0.0218 -0.1565 -0.0239 -0.1417 -0.0571 1.0000
(0.0548) (0.0173) (0.8116) (0.0851) (0.7938) (0.1194) (0.5324)
The table presents bivariate Pearson correlation coefficients and p values (in parentheses) for a two-tailed test of statistical significance for the
variables used in the regression analysis. (1) refers to breadth, (2) refers to depth, (3) refers to sustainability_performance, (4) refers to
accounting_provider, (5) refers to financial_performance, (6) refers to size, (7) refers to leverage, and (8) refers to provider_independence.
Statistics are presented for the full sample of 122 firms
Deegan, C., Cooper, B. J., & Shelly, M. (2006). An investigation of

References TBL report assurance statements. UK and European evidence.
Managerial Auditing Journal, 21(4), 329–371.
AccountAbility. (2008). AA1000 Assurance Standard 2008. London: Edgley, C. R., Jones, M. J., & Solomon, J. F. (2010). Stakeholder
AccountAbility. inclusivity in social and environmental report assurance. Ac-
Ball, A., Owen, D. L., & Gray, R. (2000). External transparency or counting, Auditing & Accountability Journal, 23(4), 532–557.
internal capture? The role of third-party statements in adding Elitzur, R., & Gavious, A. (2003). Contracting, signaling, and moral
value to corporate environmental reports. Business Strategy and hazard: A model of entrepreneurs, ‘‘angels’’, and venture
the Environment, 9(1), 1–23. capitalists. Journal of Business Venturing, 18, 709–725.
Bentler, P. M., & Chou, C. P. (1987). Practical issues in structural Fifka, M., & Drabble, M. (2012). Focus and standardization of
modelling. Sociological Methods and Research, 16(1), 78–117. sustainability reporting. A comparative study of the United
Blay, A. D., & Geiger, M. A. (2013). Auditor fees and auditor Kingdom and Finland. Business Strategy and the Environment,
independence: Evidence from going concern reporting decisions. 21(7), 455–474.
Contemporary Accounting Research, 30(2), 579–606. Fifka, M., & Pobizhan, M. (2014). An institutional approach to
Campbell, M. J., & Swinscow, T. D. V. (2009). Statistics at square corporate social responsibility in Russia. Journal of Cleaner
one (11th ed.). Chicester: Wiley-Blackwell Publishing. Production, 82, 192–201.
Casey, R. J., & Grenier, J. H. (2015). Understanding and contributing Global Reporting Initiative (GRI). (2013). Reporting principles and
to the enigma of corporate social responsibility (CSR) assurance standard disclosures. Amsterdam: Global Reporting Initiative.
in the United States. Auditing: A Journal of Practice & Theory, Hodge, K., Subramaniam, N., & Stewart, J. (2009). Assurance of
34(1), 97–130. sustainability reports. Impact on report users’ confidence and
Cho, C. H., Michelon, G., Patten, D. M., & Roberts, R. W. (2014). perceptions of information credibility. Australian Accounting
CSR report assurance in the USA. An empirical investigation of Review, 19(3), 178–194.
determinants and effects. Sustainability Accounting, Manage- Hsieh, H. F., & Shannon, S. E. (2005). Three approaches to
ment and Policy Journal, 5(2), 130–148. qualitative content analysis. Qualitative Health Research,
Cho, C. H., & Patten, D. M. (2007). The role of environmental 15(9), 1277–1288.
disclosures as tools of legitimacy: A research note. Accounting, Huggins, A., Green, W. J., & Simnett, R. (2011). The competitive
Organizations and Society, 32(7), 639–647. market for assurance engagements on greenhouse gas state-
Clarkson, P. M., Overell, M. B., & Chapple, L. (2008). Revisiting the ments: Is there a role for assurers from the accounting
relation between environmental performance and environmental profession? Current Issues in Auditing, 5(2), A1–A12.
disclosure: An empirical analysis. Accounting, Organizations Hummel, K., & Festl-Pell, D. (2015). Much Ado About Nothing?
and Society, 33(4), 303–327. Sustainability disclosure in the banking industry. Zeitschrift für
Cohen, J. R., & Simnett, R. (2015). CSR and assurance services: A Wirtschafts- und Unternehmensethik (Journal for Business,
research agenda. Auditing: A Journal of Practice & Theory, Economics & Ethics), 16(3), 369–393.
34(1), 59–74. Hummel, K., & Schlick, C. (2016). The relationship between
Cooper, S. M., & Owen, D. L. (2007). Corporate social reporting and sustainability performance and sustainability disclosure quality.
stakeholder accountability: The missing link. Accounting, Orga- Reconciling voluntary disclosure theory and legitimacy theory.
nizations and Society, 32(7–8), 649–667. Journal of Accounting and Public Policy, 35(5), 455–476.
Daniels, B. W., & Booker, Q. (2011). The effects of audit firm IFAC. (2013). Handbook of the code of ethics for professional
rotation on perceived auditor independence and audit quality. accountants (2013th ed.). New York: IFAC.
Research in Accounting Regulation, 23(1), 78–82. International Auditing and Assurance Standard Board (IAASB).
Darnall, N., Seol, I., & Sarkis, J. (2009). Perceived stakeholder (2004). International standard on assurance engagement 3000.
influences and organizations’ use of environmental audits. Assurance engagement other than audits or reviews of historical
Accounting, Organizations and Society, 34(2), 170–187. information. http://www.ifac.org/system/files/downloads/b012-
DeAngelo, L. E. (1981). Auditor independence, ‘low balling’, and 2010-iaasb-handbook-isae-3000.pdf.
disclosure regulation. Journal of Accounting and Economics, International Auditing and Assurance Standard Board (IAASB).
3(2), 113–127. (2013). ISAE 3000 (revised), assurance engagements other than
123
audits or reviews of historical financial information. Interna- agenda of social champions? The European Accounting Review,
tional framework for assurance engagements and related 9(1), 81–98.
conforming amendments. http://www.ifac.org/sites/default/files/ Oz, E. (2001). Organizational commitment and ethical behavior: An
publications/files/ISAE%203000%20Revised%20-%20for%20 empirical study of information system professionals. Journal of
IAASB.pdf. Business Ethics, 34, 137–142.
Jain, T. (2016). Decoupling corporate social orientations: A cross- Park, J., & Brorson, T. (2005). Experiences of and views on third-
national analysis. Business & Society. doi:10.1177/000765 party assurance of corporate environmental and sustainability
0315610609. reports. Journal of Cleaner Production, 13(10–11), 1095–1106.
Jamal, K., & Sunder, S. (2011). Is mandated independence necessary Patten, D. M. (1991). Exposure, legitimacy, and social disclosure.
for audit quality? Accounting, Organizations and Society, Journal of Accounting and Public Policy, 10(4), 297–308.
36(4–5), 284–292. Perego, P., & Kolk, A. (2012). Multinationals’ accountability on
Kolk, A. (2010). Trajectories of sustainability reporting by MNCs. sustainability. The evolution of third-party assurance of sustain-
Journal of World Business, 45(4), 367–374. ability reports. Journal of Business Ethics, 110(2), 173–190.
Kolk, A., & Perego, P. (2010). Determinants of the adoption of Pflugrath, G., Roebuck, P., & Simnett, R. (2011). Impact of assurance
sustainability assurance statements. An international investiga- and assurer’s professional affiliation on financial analysts’ assess-
tion. Business Strategy and the Environment, 19(3), 182–198. ment of credibility of corporate social responsibility information.
KPMG. (2015). The KPMG survey of corporate responsibility Auditing: A Journal of Practice & Theory, 30(3), 239–254.
reporting 2015. Switzerland: KPMG International. Reverte, C. (2009). Determinants of corporate social responsibility
Manetti, G., & Becatti, L. (2009). Assurance services for sustain- disclosure ratings by Spanish listed firms. Journal of Business
ability reports. Standards and empirical evidence. Journal of Ethics, 88(2), 351–366.
Business Ethics, 87, 289–298. Ruhnke, K., & Gabriel, A. (2013). Determinants of voluntary
Manetti, G., & Toccafondi, S. (2012). The role of stakeholders in assurance on sustainability reports. An empirical analysis.
sustainability reporting assurance. Journal of Business Ethics, Journal of Business Economics, 83(9), 1063–1091.
107(3), 363–377. Simnett, R., Vanstraelen, A., & Chua, W. F. (2009). Assurance on
Markelevich, A., & Rosner, R. L. (2013). Auditor fees and fraud sustainability reports. An international comparison. The
firms. Contemporary Accounting Research, 30(4), 1590–1625. Accounting Review, 84(3), 937–967.
Matisoff, D. C., Noonan, D. S., & O’Brien, J. J. (2013). Convergence Spence, M. (1973). Job market signaling. Quarterly Journal of
in environmental reporting. Assessing the carbon disclosure Economics, 87(3), 355–374.
project. Business Strategy and the Environment, 22(5), 285–305. Spence, M. (2002). Signaling in retrospect and the information
Mock, T. J., Rao, S. S., & Srivastava, R. P. (2013). The development structure of markets. American Economic Review, 92, 434–459.
of worldwide sustainability reporting assurance. Australian Stemler, S. (2001). An overview of content analysis. Practical
Accounting Review, 23(4), 280–294. Assessment, Research & Evaluation, 7(17), 137–146.
Mock, T. J., Strohm, C., & Swartz, K. M. (2007). An examination of Weaver, G. R., Treviño, L. K., & Cochran, P. L. (1999). Integrated
worldwide assured sustainability reporting. Australian Account- and decoupled corporate social performance: Management
ing Review, 17(1), 67–77. commitments, external pressures, and corporate ethics practices.
O’Dwyer, B. (2011). The case of sustainability assurance. Construct- The Academy of Management Journal, 42(5), 539–552.
ing a new assurance service. Contemporary Accounting Weber, R. P. (1990). Basic content analysis. Newbury Park, CA: Sage
Research, 28(4), 1230–1266. Publications.
O’Dwyer, B., & Owen, D. L. (2005). Assurance statement practice in White, H. (1980). A heteroskedastic-consistent covariance matrix
environmental, social and sustainability reporting. A critical estimator and a direct test of heteroskedasticity. Econometrica,
evaluation. The British Accounting Review, 37(2), 205–229. 48(4), 817–838.
O’Dwyer, B., Owen, D. L., & Unerman, J. (2011). Seeking legitimacy Zorio, A., Garcı́a-Benau, M. A., & Sierra, L. (2013). Sustainability
for new assurance forms. The case of assurance on sustainability development and the quality of assurance reports Empirical
reporting. Accounting, Organizations and Society, 36(1), 31–52. evidence. Business Strategy and the Environment, 22, 484–500.
Owen, D. L., Swift, T. A., Humphrey, C., & Bowerman, M. (2000).
The new social audits. Accountability, managerial capture or the
123

Hummel Et Al. (2019)

Uploaded by

Copyright:

Available Formats

Hummel Et Al. (2019)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hummel Et Al. (2019)

Uploaded by

Copyright:

Available Formats

J Bus Ethics (2019) 154:733–757

The Role of Sustainability Performance and Accounting Assurors

Table 1 Sample selection and sample composition

Panel B: Sample composition by country and industry

Table 2 Variables overview

Table 3 Measurement of assurance statement breadth

Table 4 Measurement of assurance process depth

Table 5 Measurement of corporate sustainability performance (Hummel and Schlick 2016)

expect to find a positive relationship between assuror Results

Table 6 Descriptive statistics

Panel C: Descriptive statistics for the regression variables

Panel A: Assurance process depth elements

Table 8 Regression results: main model

sustainability_performance H1a: ?; H1b: - -1.0115* -0.8681* -1.1945*** -1.2233 -0.4678 -0.5105

Panel A: Contingency tables

(I) (II) (III) (IV) (V) (VI) (VII) (VIII)

Panel C: Regression results for variations of the dependent variables

sustainability systems and obtain less profound assurance

Table 10 Correlation statistics

(5) 0.0759 0.1333 0.0981 -0.0754 1.0000

Deegan, C., Cooper, B. J., & Shelly, M. (2006). An investigation of

You might also like