771438

research-article2018
AUM0010.1177/0312896218771438Australian Journal of ManagementBeck et al.

Article

Australian Journal of Management

1­–21
CSR disclosure and financial © The Author(s) 2018
Reprints and permissions:
performance revisited: A cross- sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0312896218771438
https://doi.org/10.1177/0312896218771438
country analysis journals.sagepub.com/home/aum

Cornelia Beck, Geoffrey Frost and Stewart Jones

Discipline of Accounting, The University of Sydney, Darlington, NSW, Australia

Abstract
The relationship between corporate social responsibility (CSR) and corporate financial
performance (CFP) has been the subject of intensive research. However, limitations with
this literature include the use of localised samples, poorly specified control variables and self-
constructed CSR disclosure measures that may not represent a firm’s actual CSR performance.
Answering the call for ‘better’ CSR research in this field, as well as extending research to a
cross-country analysis, this study examines the relationship between corporate CSR engagement
(measured by diversity in voluntary disclosure practices) and financial performance across three
reporting jurisdictions: Australia, Hong Kong and the United Kingdom. We use the Global
Reporting Initiative (GRI) framework to rate companies on their CSR engagement and control
for actual CSR performance using the Vigeo-Eiris CSR sustainability ratings as the proxy measure.
Based on a sample of 116 large public companies, we find evidence that CSR engagement can be
indicative of actual CSR performance. We also find evidence of a significant relationship between
CSR engagement and financial performance, even after controlling for the CSR performance
proxy, firm size, industry-level fixed effects, financial risk and type of assurer. The results appear
to be robust across national reporting jurisdictions and alternative CSR metrics constructed from
the CSR engagement measure.
JEL classification: M41, M14

Keywords
CSR engagement, CSR performance, financial performance, GRI

Corresponding author:
Stewart Jones, Discipline of Accounting, The University of Sydney, Codrington Building, 21-25 Darlington Street,
Darlington, NSW 2006, Australia.
Email: [email protected]

Final transcript accepted 26 March 2018 by Jane Baxter (Editor in Chief).

2 Australian Journal of Management 00(0)

1. Introduction and prior literature

In Milton Friedman’s (in)famous 1970 New York Times magazine article titled ‘The Social
Responsibility of Business Is to Increase Its Profits’, he states,

… the corporate executive would be spending someone else’s money for a general social interest. Insofar
as his actions in accord with his ‘social responsibility’ reduce returns to stockholders, he is spending their
money. Insofar as his actions raise the price to customers, he is spending the customers’ money. Insofar as
his actions lower the wages of some employees, he is spending their money. (p. 1)

Friedman (1970: 3) suggests that corporate social responsibility (CSR) is philosophically mis-
placed if it ‘detracts from the true objectives of corporations: “to make as much money as possible
… without deception and fraud”’ (Friedman, 1982: 133), and can even result in economic costs to
corporate stakeholders. This neo-classical economic position has ignited a long-standing debate on
various philosophical dimensions and economic repercussions of CSR. One thread of this debate
has resulted in more than 30 years of academic research examining the links between CSR and
corporate financial performance (CFP) (see, for example, Al-Tuwaijri et al., 2004; Anderson and
Frankle, 1980; Belkaoui, 1976; Demacarty, 2009; Herbohn et al., 2014; Orlitzky et al., 2003). The
major theme dominating this literature is whether CSR can actually enhance financial performance,
or ‘does it pay to do good’? (Johnson, 2003).
If there is a connection between CSR and CFP, companies might be more motivated to act with
greater social consciousness, redressing their own self-created social or environmental problems
and/or directly engaging in social and environmental programmes that can benefit society gener-
ally (Porter and Kramer, 2011). According to Moser and Martin (2012): ‘if CSR activities improve
financial performance then it is less likely that such investments decrease shareholder value’ (p.
801). Several meta-analyses of the CSR–CFP relationship suggest a positive relationship between
CSR and CFP, although the magnitude of this effect and the causal relationship is still subject to
much debate (Freedman and Jaggi, 1988; Keele and DeHart, 2011; Moser and Martin, 2012). For
instance, in a review of 30 years of empirical research, Orlitzky et al. (2003) found a universally
positive relationship between CSR and CFP across industries and different study contexts, although
this relationship varied from highly positive to modestly positive ‘because of contingencies, such
as reputation effects, market measures of CFP, or CSR disclosures’ (p. 423). By contrast, Margolis
et al. (2007) and Margolis and Elfenbein (2008) found a more limited CSR–CFP relationship,
although they concluded the link might be stronger when financial performance is modelled to
predict future CSR activity, rather than CSR activity used to predict financial performance (see
also Jones and Wright, in press). Roberts (1992) showed a significant relationship between eco-
nomic performance and the level of firms’ CSR disclosure levels. Murray et al.’s (2006) study of
UK companies’ disclosure practices also found a long-term relationship between economic perfor-
mance and high disclosure levels, while De Klerk et al. (2015) examined the relationship between
CSR disclosure and share price performance among the largest UK companies and concluded that
higher levels of disclosure are associated with higher share prices.
There appears to be at least two broad theoretical approaches to CSR measurement, both ema-
nating from a stakeholder perspective (Fassin, 2009; Jamali, 2008). With respect to CSR measure-
ment, one approach envisages CSR as a multi-dimensional construct, which examines firm
performance across multiple objectives and responsibilities, including economic, environmental,
legal, social, ethical and other issues (Boesso and Kumar, 2007; Campbell, 2007; Cormier et al.,
2005; Gamble et al., 1996). Other studies interpret CSR as a tool to manage the expectations of
stakeholders, such as employees, the natural environment, charities or NGOs (Archel et al., 2011;
Banerjee and Bonnefous, 2011; Brammer and Millington, 2003, 2004). As this study draws on the
Beck et al. 3

Global Reporting Initiative’s (GRI) G3.1 guidelines, it fits the first approach because reporting
against the G3 guidelines should result in a report addressing multiple stakeholders’ interest in
business practice covering economic, environmental and social dimensions (GRI, 2011).
Predicting a positive relationship between CSR and CFP suggests that disclosure is driven by a
managerial desire to maintain or manage reputation and impressions (Aerts, 2005; Agle et al.,
1999; Barkemeyer et al., 2014). Many studies use CSR disclosures as proxies for CSR perfor-
mance. However, Al-Tuwaijri et al. (2004) questioned whether CSR disclosures are indicative of
actual CSR performance given their self-referential nature (see also Beck et al., 2010; Deegan and
Rankin, 1996; Wiseman, 1982). Some studies have shown a negative relationship between CSR
disclosure and CSR performance (Cho and Patten, 2007; Guidry and Patten, 2012), supporting the
thesis that CSR reporting is a tool to manage impressions of stakeholders (Aerts, 2005; Duff,
2017). Defining and measuring either types of CSR for the purpose of examining the CSR–CFP
relationship can be challenging with wide variations evident across studies due to the development
of many study-specific constructs (Orlitzky et al., 2003). Furthermore, many CSR–CFP studies do
not use third-party authenticated and quantifiable CSR information (Margolis et al., 2007). Such
information can proxy for transparency and improve construct validity, as CSR constructs must be
open to independent assessment and validation.
This study addresses these concerns in at least two ways. First, we distinguish between CSR
engagement and CSR performance to ascertain whether a relationship between those two variables
can indeed be assumed. CSR engagement reflects a firm’s disclosure across various CSR activities
based on the GRI’s vision that reporting against this index demonstrates engagement (GRI, 2011).
Engagement is, of course, based on CSR disclosure, but is conceptually more wide-ranging than
mere disclosure. In essence, engagement captures the diversity of disclosure types and practices
against the GRI, which requires a range of CSR indicators across several reporting dimensions. A
firm that discloses CSR information across many sustainability dimensions is seen to be engaging
in sustainability actions and practices. ‘Engagement’ is the terminology used by the GRI itself. The
GRI refers to engagement with respect to issues that are relevant to stakeholders, but further men-
tions that the CSR report should be representative of ‘overall engagement’ with sustainability
issues. Accordingly, we do not develop a new coding instrument but apply the GRI’s dominant
reporting guideline, G3.1, to establish our CSR engagement measure. We proxy for CSR perfor-
mance using the CSR performance ratings of the Vigeo-Eiris service, a leading provider of inde-
pendent research into the environmental, social, governance (ESG) and ethical performance of
public companies around the world. These measures are described in more detail below.
Studies also vary widely in terms of the use of localised samples, which are often based on dif-
ferent industry groups as the focus of the study and do not control for possible cross-country
effects. Performance variables, control variables and statistical frameworks also vary across stud-
ies, although there appears to be more consistency in the use of financial and market-based perfor-
mance variables in many studies. For instance, several studies use some type of rate of return
measure as the key financial performance variable; however, a higher proportion of studies use
market returns as the main performance variable (Orlitzky et al., 2003). A particular weakness in
extant research is with respect to the use of control variables (Margolis et al., 2007) that are often
lacking or mis-specified. Margolis et al. (2007) suggested that researchers use a wider set of con-
trol variables, which should include (as a minimum) firm size, industry, risk and other factors.
Thus, the development of the CSR construct provides a significant challenge for any study seeking
to examine the CSR–CFP relationship.
The main objective of this study is to re-examine the CSR–CFP relationship by addressing some
of the specific methodological concerns and challenges raised in the literature. These include first,
the lack of cross-country analysis of reporting practice identified by Cahan et al. (2016); second,
4 Australian Journal of Management 00(0)

the use of a globally accepted sustainability reporting framework to measure CSR engagement in
terms of disclosure, which addresses Swanson and Zhang’s (2012) concern about the definition,
parameters and measurement of CSR in the literature; third, Margolis et al.’s (2007) suggestion to
use quantifiable, third-party authenticated CSR performance information and fourth, Margolis
et al.’s (2007) call for clearly specified control variables. Finally, much previous research on the
CSR–CFP relationship has relied on standard form regression models that can be highly suscepti-
ble to violation of statistical assumptions (such as identical and independently distributed errors or
IID), which can significantly affect the interpretation of empirical results (Freedman, 2010; West
et al., 2015). For example, parameter estimates and significance levels can be impacted by data and
sampling issues (e.g. sample size), rather than anything innately meaningful among the variables
of interest. To deal with this concern, the study uses the linear mixed effects (LMEs) model to
analyse and interpret empirical findings, which is more robust to violation of various statistical
assumptions (see Jones and Wright, in press). Using the LME approach, this study contributes to
the literature by addressing the amorphous nature of CSR disclosure and performance because it
controls for actual CSR performance – rather than building on assumptions of previous studies that
CSR disclosure is a proxy for CSR performance. Furthermore, basing the study on three reporting
jurisdictions improves the external validity or generalisability of our empirical findings.

2. Hypothesis development and variable selection

The theoretical links between CSR and CFP come from two different perspectives. One model
treats CSR as a distinctive resource or means, or a way management can run the organisation to
generate higher revenues or reduce costs to improve financial performance. Improved financial
performance might be driven by more productive and loyal employees (Dobele et al., 2014;
Hillman and Keim, 2001; Rodrigo and Arenas, 2008); the development of more cost-efficient
organisational processes, such as alternative energy sources and recycling techniques (Schaltegger
et al., 2012); more innovative products and services to meet a social need (Szekely and Strebel,
2013; Wagner, 2009) and potential savings that might otherwise result from socially ‘irresponsible’
behaviour, such as penalties for environmental damage or the imposition of costly regulatory over-
sight for an industry (Clarkson et al., 2011). Also, it is argued that the CSR–CFP relationship cre-
ates the appearance of doing ‘good’, even if there is no substantive benefit in the CSR activity itself
(Idowu and Papasolomou, 2007; Johnson, 2003; Rodrigue et al., 2013). Here, CSR confers reputa-
tional value in that it can raise positive perceptions among stakeholder groups, such as consumers,
investors and employees (Berthelot et al., 2012; Brammer and Pavelin, 2004; Cho et al., 2012; Cho
and Roberts, 2010; Weber and Marley, 2012), or indeed obfuscates poor performance as suggested
by Deegan and Rankin (1996), Jones (2011) and Patten and Crampton (2004). Eccles et al. (2011:
3) provided further insight into the CSR–CFP relationship. Developing the idea of a ‘corporate
culture of sustainability’, they distinguish between ‘High’ and ‘Low’ sustainability companies. The
former are characterised by ‘a substantial number of environmental and social policies adopted for
a significant number of years (since the early to mid-1990s)’ (Eccles et al., 2011), which reflect
fundamental strategic choices of these businesses, whereas ‘Low Sustainability’ firms lack such
policies (p. 5). The authors found that ‘High Sustainability’ companies performed in the long-term
significantly better in terms of both stock market and accounting measures of performance relative
to their ‘Low Sustainability’ counterparts. They also found an industry effect for companies
engaged in natural resource extraction and in circumstances where competition was based upon
brand and reputation.
We propose that firms with higher CSR engagement are more likely to address a broader range
of topics in their reporting to signal to stakeholders their information needs are relevant and thus
Beck et al. 5

allowing better management of business risks (GRI, 2011; KPMG, 2013, 2015); This study inves-
tigates whether CSR engagement, as identified by the diversity of GRI disclosures made by a firm,
influences a number of financial and capital market variables and whether CSR engagement itself
is impacted, positively or negatively, by CSR performance. Based on prior literature and existing
theories on the CSR–CFP relationship that assume that better CSR performance will lead to
improved financial performance, the study examines the following hypothesis:

H0. There is a positive relationship between CSR engagement and financial performance, even
after controlling for a CSR performance proxy, industry-level effects, country-level effects, firm
size, financial risk, type of assurer and investment returns.

To test this hypothesis, we utilise the LME model which controls for various fixed and random
effects in the dataset and is robust to violation of the IID condition as well as other statistical
assumptions (see West et al., 2015). Using the LME model, we examine the CSR–CFP relationship
after controlling for a CSR performance proxy, industry-level effects, country-level effects, firm
size, financial risk, type of assurer and firm-level random effects. We provide a brief rationale of
the explanatory and dependent variables tested as follows.

2.1. CSR engagement

This is the dependent variable for the study. The GRI (2011) focuses on firm engagement with
material issues ‘most important to stakeholders, the economy, environment and society’ (p. 3). The
GRI suggests that a balanced CSR report should cover a broad suite of corporate activities identi-
fied by the G3 reporting framework. Consequently, a firm’s engagement with CSR can be estab-
lished through its voluntary disclosure choices and their extensiveness (Adams et al., 1998;
Andrikopoulos and Kriklani, 2013; Cahan et al., 2016). To assist with identifying these disclosure
choices, we draw on the G3.1 indicators as they are the dominant global framework for corporate
sustainability reporting (Adams, 2004; KPMG, 2011, 2013, 2015: 42) with ‘three quarters […] of
the G250 using the GRI framework’.1 The G3.1 provides 123 indicators to guide reporting on CSR
issues by firms. By matching the disclosures in the reports to the G3.1 indicators as defined by the
GRI Technical Protocol, a firm’s CSR engagement is established where the maximum possible
score is 123 (a firm reports against all GRI indicators) and a minimum score of zero (a firm does
not report against any of the indicators).2 For the purposes of this study and in line with the G3.1
guidelines definition, CSR represents how an organisation seeks to engage with its multiple stake-
holders via a company’s reporting practices. The more indicators a company addressed as per defi-
nition of the GRI guidelines and Technical Protocol, the higher the CSR Engagement of the firm.3

2.2. Pre-tax return on equity

Consistent with previous CSR–CFP research, which has focused on earnings and rate of return
measures, our key financial performance variable is pre-tax rate of return on equity (ROE),
extracted from the Thomson Reuters’ ThomsonOne database. Thomson Reuters defines pre-tax
ROE as pre-tax income divided by shareholders equity (common equity + preferred equity). The
ROE measure was selected because it is a ubiquitous performance measure in the accounting and
finance literatures.4 Also ROE and return on assets (ROA) are the most common financial perfor-
mance measures used in prior CSR studies to test the relationship between CSR and CFP (Margolis
et al., 2007; Orlitzky et al., 2003). We use pre-tax ROE to control for differences in tax regimes
6 Australian Journal of Management 00(0)

across Australia, the United Kingdom and Hong Kong. An after-tax measure is not appropriate as
performance can be affected by differences in corporate tax rates.5

2.3. Sustainability performance

Many studies modelling the CSR–CFP relationship use some type of disclosure score/index as a
proxy for CSR performance. This approach has been questioned because disclosure may not be
indicative of actual performance (Al-Tuwaijri et al., 2004; Rodrigue et al., 2013) or be applicable
across different sample sets (Gray, 2006). To address these limitations, we use a CSR performance
rating proxy that is both quantifiable and independently produced. The Vigeo-Eiris CSR perfor-
mance rating is based on ‘the relevance of companies and organisations’ commitments, the effi-
ciency of their managerial systems, their ability to manage risks and their performance on all
environmental, governance, social and societal responsibility factors’.6 The rating is made up of
over 150 qualitative CSR metrics, covering the following specific areas: governance, human rights,
employees, customers and suppliers, and products. Each CSR metric is assigned a score on a
6-point rating scale, ranging from +3 (high positive) to −3 (high negative).7
The methodology involves surveying each company directly based on their sustainability per-
formance and the assessment of each rated company against a set of global criteria. In addition to
the information the firm provides in this survey, Vigeo-Eiris makes use of publicly available infor-
mation provided by the company (e.g. website, annual report (AR), CSR report, etc.), information
from independent regulatory and other industry sources and searches of relevant press and NGO
publications. Companies rated by Vigeo-Eiris can also provide feedback on their rating to clarify
aspects of their CSR performance, which can help ‘fine tune’ the performance rating (Jones and
Wright, in press). Vigeo-Eiris initially assign a negative rating to a firm based on their overall risk
exposure. Companies are then given an opportunity to ‘neutralise’ the risk exposure by taking
appropriate remedial actions. The poorer managements’ response to the risk, the worse the CSR
rating will be.

2.4. Industry groups

The importance of industry as a control variable has been documented in numerous studies
(McWilliams and Siegel, 2001; Margolis et al., 2007). Our study sample includes firms from 10
industry groups based on the GICS (Global Industry Classification Standard) coding system. These
groups are (1) energy, (2) materials, (3) industrials, (4) consumer discretionary, (5) consumer sta-
ples, (6) health care, (7) financials, (8) information technology, (9) telecommunications and (10)
utilities. Many studies examine industry as a control variable as CSR disclosures are expected to
vary systematically across different sectors. For instance, the so-called ‘dirty’ industries, such as
heavy manufacturing, energy and resources companies, typically have higher levels of environ-
mental specific programmes and disclosures relative to other industry groups (Cho et al., 2012;
Cormier and Magnan, 1997; Freedman and Jaggi, 1986; Kolk et al., 2010; Steger et al., 2007).
However, our CSR construct represents a broad, multi-dimensional context for organisations, and
consequently reporting on CSR has increased for many other industry groups such as banks, insur-
ance companies and manufacturing entities (Frost et al., 2013).

2.5. Country-level analysis

This is a control variable to capture differences in institutional, legal and regulatory frameworks
that may affect both the level of CSR disclosure and financial performance across different
Beck et al. 7

reporting jurisdictions. This study focuses on Australia, Hong Kong and the United Kingdom
because these three jurisdictions are argued to be at different stages in their institutionalisation of
CSR reporting practices. In their survey of CSR practices, KPMG (2013: 13) suggested that UK
companies are among the top scoring reporters of CSR in Europe and are providing ‘an example
for other regions’, observing that UK CSR reports were within the top three in terms of reporting
quality. Historically, Australian reporting has been classified as ‘low’ (Frost et al., 2005; Raar,
2007; The Association of Chartered Certified Accountants (ACCA), 2006); however, more recently,
KPMG (2013) commented that Australia’s CSR reporting practices are ‘catching up with Europe’
following a strong increase in CSR reporting since 2011 (p. 22). This increase is attributed to a
significant number of first time CSR reporters in 2013, as well as disclosure recommendations by
the Australian Securities Exchange (ASX) that have increased the importance of CSR disclosures
(GRI, 2014).
Similar to Australia, listing agencies in China have issued guidelines for environmental dis-
closures since 2008 (Lin, 2010). However, CSR reporting among Chinese listed companies
remain generally poor (KPMG, 2013; Liu and Anbumozhi, 2009). Lynn (1992: 109) provided a
number of suggestions to explain the apparently low levels of CSR reporting from Hong Kong
listed companies. These included a low demand for CSR information, the unique political rela-
tionship with China prohibiting ‘lively commentary on social issues’, the industry background
of the main corporate players in Hong Kong, a belief that social and environmental issues belong
‘within the family’ (i.e. as something private) and a limited understanding of the social contract
between companies and society. Gao et al. (2005) updated Lynn’s (1992) study and documented
increased CSR disclosure in Hong Kong by 1997. Weber (2014) noted a link between ownership
status and stock exchange listings and CSR disclosure in Chinese companies. He also found that
CSR reporting influenced both financial and environmental performance dimensions but con-
cluded China still had a lot of ‘catching up’ (p. 303) to do relative to other reporting jurisdictions.
Lin (2010: 99) concurred with this observation citing ‘the early stages of [CSR] development’
when discussing changes in reporting practice and the reporting environment in China. Noronha
et al. (2013: 29) highlighted the perceived ‘very preliminary stage’ of CSR reporting in China
and Hong Kong. With little change in CSR engagement in China, as evidenced by Lin (2010)
and Liu et al. (2010), further empirical research on CSR engagement in China is another motiva-
tion to include this jurisdiction in our sample. Based on KPMG’s analysis, UK companies have
a better history of reporting on CSR than Australia and Hong Kong (which is included under
China in the KMPG report). Indeed, KPMG (2013) refers to the ‘relative maturity’ of the UK
CSR reporting environment ‘compared with countries such as China (including Hong Kong)
where […] reporting is a newer phenomenon’ (p. 33). In short, our sample presents a selection
of three countries, which have been classified in different stages of their voluntary reporting
practices; the United Kingdom (mature), Australia (emerging) and Hong Kong (preliminary
stage).

2.6. Firm size

Firm size (measured as the natural log of market capitalization) is an important control as larger
firms tend to have greater economic resources and possibly excess funds to invest in CSR
(Baumann-Pauly et al., 2013; Davison and Skerratt, 2007). They can also have higher political
costs, putting larger firms under greater public and stakeholder pressure to invest in, and report on,
CSR activity. Prior research on CSR has consistently found larger firms provide higher levels of
CSR disclosure (Andrikopoulos and Kriklani, 2013; Cormier and Gordon, 2001; Cormier and
Magnan, 1999; Holder-Webb et al., 2009).
8 Australian Journal of Management 00(0)

2.7. Financial risk

Prior literature has shown that stable firms with lower risk tend to be more likely to be engaged in
CSR. Brammer and Pavelin (2006) showed that larger, less indebted firms disclosed more environ-
mental and social information. Moreover, CSR has been linked to the risk profiles of firms (Orlitzky
and Benjamin, 2001). Some studies have shown a positive CSR–CFP relationship can actually dis-
appear once risk is formally controlled for in the model (O’Neill et al., 1989). Our proxy for firm
risk is Altman (1968) Z scores, which have been used in much prior literature (see Jones, 2017).

2.8. Type of assurer

Increasingly, organisations are engaging a third party to provide assurance for their sustainability report-
ing (KPMG, 2015). Prior research has highlighted the importance of assurance, stating the degree to
which sustainability information is assured can improve the credibility of the disclosures (Hodge et al.,
2009). While external assurance by an accounting firm is commonly sought on CSR reporting in the
United Kingdom, the practice is less common in Australia and Hong Kong (KPMG, 2011, 2015). With
no mandatory guidelines as to the type of third-party assurance provider engaged, companies employ
both audit firms and/or sustainability consulting firms (Martinov-Bennie et al., 2012). While little prior
research has evaluated the effect of assurance provider on the reporting practices of the firm, Dillard
(2011: 1272) suggested that auditors tend to focus on techniques and output (i.e. numbers in the state-
ments). However, non-audit providers perceive ‘one of their primary objectives to be changing client
organisation’s practices to be more ‘sustainable’’. O’Dwyer (2011) noted that accountants may take a
more cautious, limited approach and offer higher levels of assurance.8 For this study, where a third-party
assurance provider has been engaged, we used a dummy variable to differentiate between a Big4 audit
firm9 and other types of assurance providers (Kim et al., 2012; KPMG, 2013, 2015) We hypothesise that
organisations that engage an audit firm to provide CSR assurance will have a higher level of disclosure,
as the assurance provided may have a greater emphasis on compliance with the GRI guidelines.

2.9. Annual investment returns

Market price data such as investment returns and CSR have been examined in numerous studies
(Al-Tuwaijri et al., 2004; De Klerk et al., 2015; Herbohn et al., 2014; Jones et al., 2007; Marti
et al., 2015).10 Annual investment returns are defined by the Thomson Reuters’ ThomsonOne data-
base as the total unadjusted annual stock price return over the previous 12-month period. We use
this as a control variable as higher stock price returns can be potentially strongly associated with
higher levels of CFP (as measured by ROE) and lower levels of financial risk.

2.10. Leverage
Leverage (measured as total debt to total equity) is controlled for in the LME model because it is
expected to have a strong association with ROE. Companies that employ higher leverage tend to
have higher ROEs because of the higher use of debt financing and vice versa.

3. Methodology
3.1. Sample and data collection
The sample comprises 116 public companies drawn from the top 40 firms (based on market capi-
talization) on the ASX, the Hong Kong Stock Exchange (HKSE) and the London Financial Times
Beck et al. 9

Stock Exchange (FTSE).11 Companies were initially ranked on market capitalization, and the larg-
est companies from the list were selected. From the 116 sampled firms, 40 firms were listed on the
ASX, while 38 firms were listed on the London FTSE and 38 firms were listed on the HKSE. To
collect CSR data, the sampled firms’ websites were accessed and all published English language
reports for 2012 were downloaded.12 The reports were initially coded as an AR or as a stand-alone
sustainability report (SR).13 Most of the CSR disclosures of companies are found in stand-alone
SRs. However, some of the sampled firms disclosed CSR information in the AR and nowhere else.
Hence, it was necessary to examine both sets of reports.
Following the initial screen and analysis, each report’s content was hand collected and then
coded against all 123 indicators of the G3.1.14 Contrary to previous studies, such as Clarkson et al.
(2011), no new index was derived. The actual disclosure guidelines from the GRI (translated into
a spreadsheet) acted as the coding instrument to capture CSR engagement in each report, measured
by breath of disclosure content. Three qualified research assistants (RAs) were trained to collect
information on all GRI indicators as disclosed in the sampled firms’ published AR and/or SR
reports. Each indicator was listed and additional definitions or explanations from the GRI guide-
lines and the accompanying technical protocol were inserted for further guidance to the RAs. GRI
disclosures were coded as a simple dichotomous classification indicating presence/absence of data
fulfilling the requirements of the particular GRI indicator set out in the GRI 3.1 framework. The
coding instrument also required recording the narrative coded as ‘YES’ in a separate cell and refer-
ence of the location in the report. In cases where the reconciliation showed discrepancies, addi-
tional guidance was entered in the coding instrument.15 A random selection of 10 reports was
independently coded against the framework by two RAs and the third RA reconciled their coding.
One of the authors met with the RAs on several occasions to discuss differences and definitions
and relayed inconsistencies/questions to a representative of the GRI’s technical committee for
clarification. Once the inter-coder reliability reached levels above 80%, all three coders were
assigned to code the reports of 40 companies. For each of them, five of those were cross-coded for
consistency by one of the other RAs and agreement levels over 85% were achieved.

4. Empirical results
The following section presents the empirical findings. We first set out the descriptive statistics,
followed by the regression analysis.

4.1. Descriptive statistics

Table 1 provides descriptive statistics for some of the key variables used in this study based on the
whole sample. A breakdown of Table 1 results by country is provided in the Online Appendix.
The mean market capitalization for the sample is US$ 3.14 billion. Among the sampled firms,
the highest industry concentrations were found in the following sectors: financial sector (n = 41),
consumer staples sector (n = 14), materials sector (n = 13), industrials sector (n = 13) and energy
sector (n = 10). The lowest industry concentrations were found in the utilities sector (n = 8), con-
sumer discretionary sector (n = 6), telecommunications sector (n = 5), healthcare sector (n = 4) and
the information technology sector (n = 2). While the distribution of sampled firms across industries
groups is relatively even, there are some noticeable concentrations within particular sectors at the
country level. For instance, there is a strong concentration of energy and materials companies in
the Australian and UK sub-samples, but, as expected, there was very little representation of these
sectors in the Hong Kong sub-sample. By contrast, there is a relatively higher concentration of
financial and information technology firms in the Hong Kong sub-sample, which might be expected
given the economic demographics of this jurisdiction.
10 Australian Journal of Management 00(0)

In terms of financial performance metrics, the median leverage ratio (total debt to total equity)
is 49.28%. As can be seen from Table 1, the mean value of the sample is significantly higher
(103.86%) owing to the effects of banks and other financial institutions being included in the sam-
ple. As shown in the Online Appendix, the UK sub-sample has the highest median leverage ratio
of 70.29% while the Hong Kong sub-sample evidences the lowest median leverage of 39.10% (the
median value was 54.11% for the Australian sub-sample). The median pre-tax ROE across the
sample is 16.47%, with the highest pre-tax ROE displayed in the UK sub-sample (median of 22.84
%) and the lowest pre-tax ROE found in the Hong Kong sample with a median of 14.03% (the
median value for the Australian sub-sample was 14.42%).
Table 1 also shows that the median Vigeo-Eiris sustainability rating across the full sample is
−260. A negative rating means the firm has identified risk exposures that have not been fully
addressed by management. Poorer ratings indicate that firm managements have larger risk expo-
sures to address. A positive rating (greater than zero) means the firm has satisfactorily addressed
risk exposures and the better the rating, the better the job management is doing in redressing identi-
fied risk exposures. The UK sub-sample has the highest overall median rating of −30, followed by
Australian sub-sample with a median rating of −198.5. The Hong Kong sub-sample has a signifi-
cantly lower median rating of −961.
Table 1 shows that the median total CSR engagement score is 44.5 (the mean value is 48.04)
from a theoretical maximum possible value of 123. CSR engagement ranges from 16 to 101 across
the sample. All 116 sampled firms disclosed CSR information within the AR. The median score
was 33 across the sample and 58% of firms (68 in total) produced a discrete SR. On average, CSR
disclosures tended to be higher in the SR relative to the AR, with a median CSR engagement of 39
for the SR.
Of the 22 UK firms that reported CSR information in the SR, 19 were externally assured. Of the
19 externally assured firms, 15 firms used a Big4 accounting firm as the assurance provider. Of the
31 firms in the Australian sub-sample that reported CSR information in the SR, 26 were externally
assured of which three firms used a Big4 accounting firm as the assurance provider. Of the 15 firms
in the Hong Kong sub-sample that reported CSR information in the SR, 8 firms had their SR exter-
nally assured, while 3 firms used a Big4 firm.
As displayed in the Online Appendix, the United Kingdom has the highest total CSR engage-
ment overall (median 51), followed by Australia (median 45.5) and then Hong Kong (median
39.5). The Australian sub-sample has more overall dispersion in CSR engagement, indicating that
sampled firms have the highest and lowest CSR engagements across the sample, while the Hong
Kong sub-sample has low variability in CSR engagement. This finding supports the classification
of the three countries as at different stages in their CSR reporting practices.

4.2. Regression analysis

In this study, we use the LME model to examine the CSR–CFP relationship. As a robustness
check, we also run the LME model on two key components of our total CSR engagement,
namely Total Environmental Indicators and Total Governance Indicators. The LME model
(which combines fixed and random parameters) is an improvement on standard form ordinary
least squares (OLS) regression models because this model form is designed to handle correlated
error structures arising from heteroscedasdicity, multicollinearity and other statistical problems
arising in the dataset.16
Table 2 reports the LME model, which controls for country- and industry-level fixed effects, as
well as size and financial performance on our CSR engagement construct (Model 1).
 Beck et al.

Table 1. Descriptive statistics.

Valid (N) Missing (N) Mean Median Std. Deviation Skewness Kurtosis Minimum Maximum
Market capitalization (US$ million) 116 0 31,419.38 17,875.12 36,789.58 2.601 8.474 3146.97 225,112.56
Leverage (total debt/equity) 116 0 103.8589 49.2800 146.23 2.962 9.933 .00 816.79
Pre-tax ROE 116 0 22.5887 16.4700 22.77 2.636 10.674 −25.89 151.15
Total return (1 year) 116 0 21.1879 18.4212 28.58 0.551 1.101 −52.73 109.43
Vigeo-Eiris rating 116 0 −500.74 −260.00 848.75 −0.785 −0.098 −2902 769
Total CSR engagement 116 0 48.04 44.50 18.885 0.596 −0.140 16 101
Total CSR engagement (AR) 116 0 34.22 33.00 9.752 .477 .281 16 63
Total CSR engagement (SR) 68 48 42.03 39.00 18.362 .643 .408 11 101
Total strategy 116 0 1.0948 1.0000 0.82334 −.179 −1.504 .00 2.00
Total profile 116 0 7.7241 8.0000 1.36787 −1.229 1.371 3.00 9.00
Total report profile 116 0 3.0690 3.0000 0.52294 −.281 2.354 1.00 4.00
Total report boundaries 116 0 4.6207 5.0000 1.93213 −.279 −.821 .00 8.00
Total governance 116 0 7.2759 7.0000 1.65548 −.322 −.165 3.00 10.00
Total external initiatives 116 0 1.6983 2.0000 0.97106 −.229 −.666 .00 4.00
Total stakeholder engagement 116 0 2.1466 2.5000 1.67464 −.191 −1.646 .00 4.00
Total economic Indicators 116 0 2.4914 2.0000 1.19779 0.577 0.619 .00 6.00
Total environmental indicators 116 0 7.1207 5.0000 6.25009 0.924 0.373 .00 29.00
Total social indicators 116 0 3.2155 3.0000 2.50150 0.510 −0.466 .00 10.00
Total human rights 116 0 0.9052 0.0000 1.54345 1.937 3.509 .00 7.00
Total society 116 0 2.0259 2.0000 1.89490 0.852 −0.039 .00 7.00
Total product responsibility 116 0 1.5603 1.0000 1.93498 1.743 2.793 .00 9.00

ROE: return on equity; CSR: corporate social responsibility; AR: annual report; SR: sustainability report.
11
12 Australian Journal of Management 00(0)

The model also controls specifically for actual CSR performance using the Vigeo-Eiris sustain-
ability ratings variable. In addition, we specify a firm-specific random intercept term and random
parameter, which captures random variations in total CSR engagement across firms.
Table 2 displays the same LME analysis where total environmental indicators (Model 2) and
total governance indicators (Model 3) are specified as dependent variables, respectively. As noted
by Kim et al. (2012), corporate governance is often perceived as a distinct construct from CSR and
its impacts on financial reporting have been widely examined in previous literature (Eccles et al.,
2011). Prior research has also shown high correlations between CSR disclosure behaviour and
industry membership (Adams et al., 1998; Galani et al., 2012), with resource intensive sectors,
whose activities are considered especially high in environmental pollution, such as extractive
industries, traditionally discharging environmental accountability through targeted disclosures
(Cormier and Magnan, 1999; Deegan et al., 2002; Galani et al., 2012).
For Model 1, a number of significant parameters (fixed effects) indicate that higher CSR
engagement scores are statistically associated with greater firm size (t = 4.964, p = .000), higher
pre-tax ROE (t = 2.098, p = .039), lower leverage (t = −1.757, p = .082) and with the type of assurer
(t = 3.733, p = .000). This suggests that larger firms that use an accounting firm as the primary
assurer for sustainability information are statistically associated with higher overall CSR engage-
ment. It is noteworthy that pre-tax ROE is positive and significant after controlling for all other
effects, including the actual CSR performance proxy.17 CSR performance ratings are significant
and positively associated with total CSR engagement (t = 1.66, p = .10) although only at the 10%
level. This suggests that firms with higher CSR engagement also have higher actual CSR perfor-
mance, as proxied by the Vigeo-Eiris performance ratings, which provides support to some studies
which envisage CSR disclosures to be an useful proxy for actual CSR performance.
Table 2 indicates that higher total CSR engagement is negatively associated with financial risk
(lower risk is associated with higher CSR engagement) but the parameter estimate is not statisti-
cally significant. Investment returns also show a positive relationship with CSR engagement, but
the parameter is again not significant.
Table 2 also shows some quite strong industry-level fixed effects in Model 1. Several industry
groups have significant negative parameter estimates, including Consumer Discretionary
(t =−2.035, p = .046), Financials (t = −2.32, p = .023) and the Telecommunication services sector
(t = −1.985, p = .05). These industry sectors appear to be associated with significantly lower total
CSR engagement relative to other industries. However, the Materials sector has a significantly
positive parameter (t = 2.541, p = .013), while the Energy sector has a positive parameter which is
not statistically significant. This suggests that these sectors are associated with higher CSR engage-
ment and may reflect a stronger tradition of environmental disclosure under various regulations
and standards of best practice within these industries. Panel B of Table 2 displays the results for
random parameters. This is a test of whether the intercepts of the firm-level CSR engagement are
significantly different. A statistically significant parameter could suggest that there is a lot of ran-
dom variation in CSR engagement that is not accounted for in the fixed effects (suggesting poten-
tial error clustering).
As a robustness check, we also disaggregated the total CSR engagement into specific indicators
– namely, total environmental and total governance – also shown in Table 2. In these particular
areas, we expected more diversity in reporting across national boundaries. For instance, Australia
has a much higher proportion of companies concentrated in the Materials and Energy sectors,
which, as previously mentioned, has a stronger tradition of environmental disclosure than other
industries (Eccles et al., 2011). Corporate governance is often viewed as a distinct construct within
CSR (Kim et al., 2012). Both Australia and the United Kingdom have established corporate gov-
ernance practices whereas Hong Kong evidenced the lowest CSR engagement in this area.
Table 2. Linear mixed effects models.
Panel A: Fixed parameters Model 1 Model 2 Model 3
Beck et al.

Total CSR engagement Total environmental Total governance

Estimate t value p value Estimate t value p value Estimate t value p value

Market capitalization 8.457 4.964*** .000 3.417 5.407*** .000 0.502 2.81** .006
ROE 0.1516 2.098** .039 0.0480 1.784* .078 0.0198 2.623** .010
Financial risk −0.0170 −1.136 .259 −0.004974 −0.892 .375 −0.0001 −0.106 .916
Type of assurer 14.074 3.733*** .000 4.977 3.557** .001 −0.2387 −0.605 .547
Annual investment returns 0.01462 0.241 .810 −0.0099 −0.440 .661 −0.0063 −0.317 .317
Leverage −0.0204 −1.757* .082 −0.0032 −0.749 .455 −0.0029 −2.446** .016
Vigeo-Eiris CSR performance rating 0.0036 1.66* .100 0.0018 2.285** .024 −0.00019 −0.876 .383
Country-level fixed effects
Australia −27.491 −1.666 .099 −20.131 −3.292 .001 4.065 2.353 .021
Hong Kong −25.705 −1.501 .137 −20.171 −3.175 .002 1.938 1.081 .282
United Kingdom −38.884 −2.192 .031 −25.655 −3.896 .000 2.735 1.473 .144
Industry-level fixed effects
Energy 0.72516 0.096 .923 −0.0402 −0.014 .989 −1.147 −1.458 .148
Materials 18.159 2.541 .013 6.837 2.578 .011 −0.4645 −0.621 .536
Industrials −9.678 −1.576 .118 −4.774 −2.095 .039 −0.1267 −0.197 .844
Consumer discretionary −14.682 −2.035 .046 −5.821 −2.156 .034 −1.589 −2.112 .037
Consumer staples −2.746 −0.421 .675 0.1922 0.079 .937 −1.231 −1.803 .074
Health care −4.739 −0.540 .591 −3.082 −0.945 .347 −0.729 −0.793 .430
Financials −12.677 −2.320 .023 −5.043 −2.485 .015 −0.499 −0.873 .385
Info technology 12.876 0.895 .373 11.6196 2.176 .032 0.186125 0.124 .902
Telecommunication Services −16.178 −1.985 .050 −9.488 −3.135 .002 −0.3839 −0.450 .654
Utilities – – –
Panel B: Random Parameters
Residual 97.057 0.463 15.448 0.9281 –
Intercept (firms) 78.252 0.370 2.800 5.923*** 0.9943 3.674***

This table shows linear mixed effect models for three dependent variables: (1) Total CSR engagement, (2) total environmental and (3) total governance. The utilities sector dummy is redundant in the
model as it is perfectly correlated with the intercept term (i.e. the dummy variable trap problem). This table displays fixed parameter estimates, t values, significance levels for parameter estimates for
market capitalization, pre-tax ROE, financial risk, type of assurer, annual investment returns, leverage, Vigeo-Eiris CSR performance ratings (a proxy for sustainability performance), country-level fixed effects
and industry-level fixed effects. Panel B shows the same information for the random parameter estimate (variance in CSR engagement at the firm level) which is found to be statistically significant for total
environmental and the total governance. The table shows that for each of the dependent variables, the CSR–CFP relationship is significant and positive after controlling for all other factors, including CSR
13

performance, and is robust across national reporting boundaries.

14 Australian Journal of Management 00(0)

When the LME model was estimated using total environmental indicators as the dependent
variable (Model 2), the results were quite similar to Model 1. Model 2 shows that market capitali-
zation and type of assurer are both positive and strongly significant. However, the pre-tax ROE
variable is positive and significant but only at the 10% level (t = 1.784, p = .078) and the financial
risk parameter is negative but not significant. It is also noteworthy that the CSR performance rating
variable has a stronger statistical impact in Model 2 (t = 2.28, p = .024) suggesting that higher envi-
ronmental disclosure is more strongly correlated with actual CSR performance as proxied by the
Vigeo-Eiris ratings. Furthermore, industry-level fixed effects are more significant overall than the
Model 1 results. The Materials sector again has a strong positive parameter (t = 2.57, p = .011) indi-
cating that this industry is more strongly associated with higher environmental disclosures. It is
also interesting to observe that the random parameter intercept is significant suggesting that there
is evidence of random variation in total environmental scores at the firm level not accounted for by
the fixed effects of the model.
Model 3 in Table 2 also displays the results for the total governance indicator dependent varia-
ble. We find that our financial performance metric, pre-tax ROE, is again positive and significant,
including other parameters such as firm size, which has a positive parameter (t = 2.81, p = .006), and
leverage, which has a negative parameter (t = −2.446, p = .016). The CSR performance variable,
financial risk and type of assurer are all statistically insignificant.
Total governance is not strongly associated with any industry-level fixed effects, as only one
industry-level parameter estimate is significant at the 5% level. The CSR performance rating vari-
able has a negative parameter but is not significant. Panel B of Table 2 indicates that the random
parameters intercept is highly significant, indicating that total governance score varies signifi-
cantly across individual firms. In other words, total governance disclosures can be lower or higher
for individual firms which appear to be random because they arise from explanatory variables
which are unmeasured or omitted from the fixed effects.
Overall, these results suggest that the CSR–CFP relationship is not only robust across national
boundaries but also holds up reasonably well to different constructions of our CSR metric which
are likely to heighten the differences in CSR engagement across national boundaries.

5. Concluding comment
The debate surrounding the CSR–CFP relationship has been confounded by the lack of consistent
empirical evidence or indeed evidence supporting causality in either direction.18 While this field of
research has been subjected to extensive investigation in recent years, there remain fundamental
discrepancies in research designs and empirical findings that undermine our understanding of the
CSR–CFP relationship. In this article, we have sought to redress some of these concerns through
(1) the utilisation of a more robust statistical framework for the analysis, and (2) establishing the
validity of our broad-based CSR engagement variable. We achieve this by using the only globally
adopted voluntary reporting guidelines as our analytical lens; setting more clearly specified control
variables, including a CSR performance measure; and, our capacity to interpret the CSR–CFP
relationship across national boundaries. We demonstrate a positive association between CSR and
CFP after controlling for CSR performance, industry-level effects, country-level effects, firm size,
financial risk, type of assurer and investment returns.
We document CSR practices of 116 large public companies on a wide range of GRI sustainability
indicators from the GRI Framework (G3.1 Guidelines). Each firm’s CSR disclosures were hand col-
lected and converted to a CSR engagement score by assessing them against 123 specific G3.1 indica-
tors. While there is significant variation in the overall CSR reporting practices across national
boundaries, industry groups and organisational sizes, the total CSR engagement is found to positively
Beck et al. 15

and significantly associate with financial performance, even after controlling for our CSR perfor-
mance proxy, industry-level fixed effects, country-level fixed effects, firm size, financial risk and
type of assurance provider (i.e. whether a Big4 accounting firm or some other provider). The results
suggest that larger and more profitable firms with better actual CSR performance, lower financial
risk, and who use a Big4 firm to assure CSR disclosures demonstrate higher CSR engagement over-
all. The results from our sample also support the assumption that the level of disclosure, indicated
through our CSR engagement variable, appears to be broadly indicative of CSR performance.
This study makes several contributions to the extant literature. First, it addresses the problem of
inconsistency in, and the replicability of, studies relating to disclosure content by using an existing
CSR reporting guideline as the data-coding instrument. Second, we respond to criticisms of models
used to test relationships by introducing a more sophisticated LME model, which overcomes some
of the statistical limitations of standard models. The third contribution stems from the sample
selection. While much of the previous research is confined to a single country analysis, we draw
on three reporting jurisdictions to improve the generalisability of our empirical findings. Finally,
by establishing a positive relationship between CSR disclosures and (externally verified) CSR
performance, we provide evidence that CSR disclosures can be a useful proxy for actual CSR
performance.
The study’s findings come with some caveats. Unlike KLD sustainability data and Viego-Eiris
sustainability ratings, GRI information is not available in commercial databases so it must be hand
collected by the researcher. As this process is time intensive, our CSR engagement variable is
based on a relatively small sample size which can potentially limit generalisability. However, our
sample includes the largest public companies in Australia, the United Kingdom and Hong Kong,
representing most of total market capitalization on each jurisdiction’s local stock exchange. As the
literature has established that larger companies tend to disclose more CSR information, concerns
with empirical generalizability should not be overstated.
Another limitation relates to the use of Vigeo-Eiris ratings as the CSR performance control vari-
able. The Vigeo-Eiris ratings methodology is proprietary and while the ratings seem to go beyond
reliance on voluntary disclosure of companies, it is not clear how much of the rating is weighted
on disclosure or other forms of independent assessment of actual CSR performance. Finally, this
study does not use multiple time frames, which could be useful for assessing the direction of the
CSR–CFP relationship. For instance, Jones and Wright (in press) used multiple time frames and
find that financial performance appears to drive their ‘creating shared value’ construct, rather than
shared value itself causing companies to become more profitable. Extending this study to multiple
time frames is a useful direction for future research.

Acknowledgements
The authors thank the editors and associate editors of Australian Journal of Management for their guidance
in improving this article and Zoe Baker, Nemone Goonasekera and Nichole Orth for their assistance with data
collection.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publica-
tion of this article: The authors acknowledge the financial support for this project from CPA Australia.

Notes
1. Until the launch of the latest guidelines (G4) in May 2013, G3.1 was the most recent set of reporting
guidelines that companies could adopt for their corporate social responsibility (CSR) disclosures. KPMG
16 Australian Journal of Management 00(0)

(2015: 42) observed a drop in Global Reporting Initiative (GRI) referencing for CSR reporting following
the introduction of G4.
2. The use of the G3.1 as the CSR engagement measure is consistent with a number of prior studies (see
Clarkson et al., 2008, 2011; Galani et al., 2012; Legendre and Coderre, 2013).
3. We acknowledge that using the GRI framework as our CSR disclosure measure does have certain limita-
tions. Our data collection, by necessity, is limited to self-reported CSR disclosures as the GRI is not a
mandatory framework. This can lead firms to ‘cherry-pick’ their sustainability disclosures (Boiral, 2013;
Clarkson et al., 2011; Frost et al., 2005; Knudsen, 2005) as well as positive self-selection bias in CSR
reporting (Aerts, 2005; Deegan et al., 2002; Deegan and Rankin, 1996; Laufer, 2003; Neu et al., 1998).
We deal with this concern by introducing an independent CSR performance rating as a control variable
in the analysis.
4. For instance, it is a key measure used in residual income valuation models, and for calculating a firm’s
sustainable growth rate (Palepu et al., 2015).
5. Another measure that could be used is EBIT-to-total assets, which is basically pre-tax ROA. We get
similar results whether a pre-tax version of ROA or return on equity (ROE) is used. This should not be
surprising as ROA and ROE tend to be highly correlated measures.
6. See http://www.vigeo-eiris.com/vigeo-eiris/methodology-quality-assurance/ (accessed February 2017).
7. Other possible scores are +2 (medium positive), −2 (medium negative), +1 (low positive) and −1 (low
negative).
8. Audit firms are constrained by standards such as ISAE3000 ‘Assurance Engagements Other than Audits
or Reviews of Historical Financial Information’. In addition, practitioners can refer to the AA1000
Assurance Standard developed by SustainAbility and a multitude of additional assurance frameworks to
inform the assurance they provide.
9. KPMG (2013) observed 70% of G250 firms sought assurance from accounting firms. We find that no
audit firm other than a Big4 was engaged by the sampled firms.
10. See also the meta-analysis studies of Orlitzky et al. (2003) and Margolis et al. (2007), which cite numer-
ous studies that use market price data as a performance variable to model the CSR–CFP (corporate
financial performance) relationship.
11. While the sample size is relatively small, this is quite typical of studies in this field, particularly for
cross-country studies (Cahan et al., 2016). Despite the small sample size, it is expected to have good
external validity within the particular reporting jurisdictions examined. We have selected 116 of the larg-
est companies in United Kingdom, Australia and Hong Kong, representing a very significant proportion
of the market capitalization of public companies within these jurisdictions. Furthermore, much previ-
ous research shows that sustainability disclosures tend to be concentrated in the largest companies (e.g.
Orlitzky et al., 2003).
12. For the Hong Kong sample, not all companies within the top 40 provide annual reports in English and as
such the sample is slightly smaller.
13. All stand-alone reports were coded to the same name for ease of data management.
14. Not all sample companies applied the GRI for their reporting, so indicator 3.12 was excluded from this
exercise. The exclusion is further warranted as most GRI adopters tend to produce the GRI content grid
required by this indicator in a separate document referring back to the respective pages in the reports.
15. For example, following the guidance from the Technical Protocol Committee, indicator LA1 – accord-
ing to the G3.1 guidelines – requires information on total workforce by employment type, employment
contract and region, broken down by gender – but was recorded as YES when presented by gender, even
if not all components were addressed.
16. The vast majority of CSR–CFP studies use standard ordinary least squares (OLS) regression models.
Standard regression models assume the identical and independently distributed errors (IID) condition.
Violation of this assumption can significantly bias parameter estimates and significance tests (Freedman,
2010). Linear mixed effects (LMEs) estimated on cross-sectional data (as in the current study) are some-
times referred to as ‘clustered’ data models (see West et al., 2015). Clustered data are defined where the
dependent variable is measured once for each subject or unit of analysis (i.e. for each firm in our study).
They are also referred to as hierarchical mixed effect models, because the unit of analysis can be grouped
Beck et al. 17

into or nested within clusters of units. In the context of our own study, the hierarchies make good sense.
Level 1 is the firm level (the lowest level of observation), Level 2 is the industry level (as firms can be
grouped or nested within industries) and Level 3 would be the country level (as industries can be grouped
or nested within countries).
17. We also tested for random effects of (1) firms nested within their industry groups and (2) firms nested
within their industry groups and industries nested at the country level. Neither of these random effect
parameters are found to be significant.
18. For the purposes of this study, we could not find evidence of an endogeneity problem in the CSR–CFP
relationship (suggesting that CFP leads to CSR and not the reverse). Using one and two period lags of
the financial variables as instrument variables, applying the Hausman–Wu endogeneity test produced
a p value greater than .1, leading us to accept the null hypothesis that IV estimates do not significantly
improve on OLS model estimates.

References
Adams CA (2004) The ethical, social and environmental reporting-performance portrayal gap. Accounting,
Auditing & Accountability Journal 17: 731–757.
Adams CA, Hill W and Roberts CB (1998) Corporate social reporting practices in Western Europe:
Legitimating corporate behaviour? The British Accounting Review 30: 1–21.
Aerts W (2005) Picking up the pieces: Impression management in the retrospective attributional framing of
accounting outcomes. Accounting, Organizations and Society 30: 493–517.
Agle BR, Mitchell RK and Sonnenfeld JA (1999) Who matters to CEOS? An investigation of stakeholder
attributes and salience, corporate performance and CEO values. Academy of Management Journal 42:
507–525.
Al-Tuwaijri S, Christensen TE and Hughes KE (2004) The relations among environmental disclosure, envi-
ronmental performance, and economic performance: A simultaneous equations approach. Accounting,
Organizations and Society 29: 447–471.
Altman EI (1968) Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The
Journal of Finance 23: 589–609.
Anderson JC and Frankle AW (1980) Voluntary social reporting: An ISO-beta portfolio analysis. The
Accounting Review 55: 468–479.
Andrikopoulos A and Kriklani N (2013) Environmental disclosure and financial characteristics of the firm:
The case of Denmark. Corporate Social Responsibility and Environmental Management 20: 55–64.
Archel P, Husillos J and Spence C (2011) The institutionalisation of unaccountability: Loading the dice of
Corporate Social Responsibility discourse. Accounting, Organizations and Society 36: 327–343.
Banerjee SB and Bonnefous A (2011) Stakeholder management and sustainability strategies in the French
nuclear industry. Business Strategy and the Environment 20: 124–140.
Barkemeyer R, Comyns B, Figge F, et al. (2014) CEO statements in sustainability reports: Substantive infor-
mation or background noise? Accounting Forum 38: 241–257.
Baumann-Pauly D, Wickert C, Spence LJ, et al. (2013) Organizing corporate social responsibility in small
and large firms: Size matters. Journal of Business Ethics 115: 693–705.
Beck AC, Campbell D and Shrives PJ (2010) Content analysis in environmental reporting research: Enrichment
and rehearsal of the method in a British-German context. The British Accounting Review 42: 207–222.
Belkaoui A (1976) The impact of the disclosure of environmental effects of organizational behaviour on the
market. Financial Management 5: 26–31.
Berthelot S, Coulmont M and Serret V (2012) Do investors value sustainability reports? A Canadian study.
Corporate Social Responsibility and Environmental Management 19: 355–363.
Boesso G and Kumar K (2007) Drivers of corporate voluntary disclosure: A framework and empiri-
cal evidence from Italy and the United States. Accounting, Auditing & Accountability Journal 20:
269–296.
Boiral O (2013) Sustainability reports as simulacra? A counter-account of A and A 1 GRI reports. Accounting,
Auditing & Accountability Journal 26: 1036–1071.
18 Australian Journal of Management 00(0)

Brammer S and Millington A (2003) The evolution of corporate charitable contributions in the UK between
1989 and 1999: Industry structure and stakeholder influences. Business Ethics: A European Review 12:
216–228.
Brammer S and Millington A (2004) Stakeholder pressure, organizational size, and the allocation of depart-
mental responsible for the management of corporate charitable giving. Business & Society 43: 268–295.
Brammer S and Pavelin S (2004) Building a good reputation. European Management Journal 46: 704–713.
Brammer S and Pavelin S (2006) Voluntary environmental disclosures by large UK companies. Journal of
Business Finance & Accounting 33: 1168–1188.
Cahan SF, De Villiers C, Jeter DC, et al. (2016) Are CSR disclosures value relevant? Cross-country evidence.
European Accounting Review 25: 579–611.
Campbell JL (2007) Why would corporations behave in socially responsible ways? An institutional theory of
corporate social responsibility. Academy of Management Review 32: 946–967.
Cho CH and Patten DM (2007) The role of environmental disclosures as tools of legitimacy: A research note.
Accounting, Organizations and Society 32: 639–647.
Cho CH and Roberts RW (2010) Environmental reporting on the internet by America’s Toxic 100: Legitimacy
and self-presentation. International Journal of Accounting Information Systems 11: 1–16.
Cho CH, Guidry RP, Hageman AM, et al. (2012) Do actions speak louder than words? An empirical investi-
gation of corporate environmental reputation. Accounting, Organizations and Society 37: 14–25.
Clarkson PM, Li Y, Peter M, et al. (2008) Revisiting the relation between environmental performance and
environmental disclosure: An empirical analysis. Accounting, Organizations and Society 33: 303–327.
Clarkson PM, Overell MB and Chapple L (2011) Environmental reporting and its relation to corporate envi-
ronmental performance. Abacus 47: 27–60.
Cormier D and Gordon IM (2001) An examination of social and environmental reporting strategies.
Accounting, Auditing & Accountability Journal 14: 587–616.
Cormier D and Magnan M (1997) Investors’ assessment of implicit environmental liabilities: An empirical
investigation. Journal of Accounting and Public Policy 16: 215–241.
Cormier D and Magnan M (1999) Corporate environmental disclosure strategies: Determinants, costs and
benefits. Journal of Accounting, Auditing & Finance 14: 429–451.
Cormier D, Magnan M and Van Velthoven B (2005) Environmental disclosure quality in large German com-
panies: Economic incentives, public pressures or institutional conditions? European Accounting Review
14: 3–39.
Davison J and Skerratt L (2007) Words, pictures and intangibles in the corporate report. EAN 9781904574211.
Available at: https://www.icas.com/technical-resources/publication-words-pictures-and-intangibles-in-
the-corporate-report
De Klerk M, de Villiers C and van Staden C (2015) The influence of corporate social responsibility disclosure
on share prices: Evidence from the United Kingdom. Pacific Accounting Review 27: 208–228.
Deegan C and Rankin M (1996) Do Australian companies report environmental news objectively? An anal-
ysis of environmental disclosures by firms prosecuted successfully by the environmental protection
authority. Accounting, Auditing & Accountability Journal 9: 50–67.
Deegan C, Rankin M and Tobin J (2002) An examination of the corporate social and environmental dis-
closures of BHP from 1983–1997: A test of legitimacy theory. Accounting, Auditing & Accountability
Journal 15: 312–343.
Demacarty P (2009) Financial returns of corporate social responsibility, and the moral freedom and responsi-
bility of business leaders. Business and Society Review 114: 393–433.
Dillard J (2011) Double loop learning; or, just another service to sell: A comment on ‘the case of sustainability
assurance: Constructing a new assurance service’. Contemporary Accounting Research 28: 1267–1277.
Dobele AR, Westberg K, Steel M, et al. (2014) An examination of corporate social responsibility implementa-
tion and stakeholder engagement: A case study in the Australian mining industry. Business Strategy and
the Environment 23: 145–159.
Duff A (2017) Corporate social responsibility as a legitimacy maintenance strategy in the professional
accountancy firm. The British Accounting Review 49: 513–531.
Beck et al. 19

Eccles RG, Ioannou I and Serafeim S (2011) The impact of a corporate culture of sustainability on corpo-
rate behavior and performance. Working Paper no. 12-035. Harvard Business School. Available at:
http://trippel.sdg.no/wp-content/uploads/2014/09/Eccles-HBR_The-Impact-of-a-Corporate-Culture-of-
Sustainability1.pdf
Fassin Y (2009) The stakeholder model refined. Journal of Business Ethics 84: 113–135.
Freedman DA (2010) Statistical Methods and Causal Inference: A Dialogue with the Social Sciences (eds D
Collier, JS Sekhon and PB Stark). Cambridge: Cambridge University Press.
Freedman M and Jaggi B (1986) An analysis of the impact of corporate pollution disclosures included in
annual financial statements on investors’ decisions. Advances in Public Interest Accounting 1: 193–212.
Freedman M and Jaggi B (1988) An analysis of the association between pollution disclosure and economic
performance. Accounting, Auditing & Accountability Journal 1: 43–58.
Friedman M (1970) The social responsibility of business is to increase its profits. New York Times Magazine,
13 September: 122–126.
Friedman M (1982) Capitalism and Freedom. Chicago, IL: University of Chicago Press (originally published
in 1962).
Frost GR, Jones S and Lee P (2013) The Reality and the Rhetoric: Organisational Sustainability Reporting.
Sydney, NSW, Australia: Sydney University Press.
Frost GR, Jones S, Loftus L, et al. (2005) A survey of sustainability reporting practices of Australian reporting
entities. Australian Accounting Review 15: 89–96.
Galani D, Gravas D and Stavropoulos A (2012) Company characteristics and environmental policy. Business
Strategy and the Environment 21: 236–247.
Gamble GO, Hsu K, Jackson C, et al. (1996) Environmental disclosures in annual reports: An international
perspective. The International Journal of Accounting 31: 293–331.
Gao SS, Heravi S and Xiao JS (2005) Determinants of corporate social and environmental reporting in Hong
Kong: A research note. Accounting Forum 29: 233–242.
Gray R (2006) Social, environmental and sustainability reporting and organisational value creation?
Accounting, Auditing & Accountability Journal 19: 793–819.
Global Reporting Initiative (GRI) (2011) RG: Sustainability Reporting Guidelines (Version 3.1: Including
Technical Protocol). Amsterdam: GRI.
Global Reporting Initiative (GRI) (2014) State of Sustainability Disclosure: Australia. Sydney, Australia:
GRI.
Guidry RP and Patten DM (2012) Voluntary disclosure theory and financial control variables: An assessment
of recent environmental disclosure research. Accounting Forum 36: 81–90.
Herbohn K, Walker J and Loo HYM (2014) Corporate social responsibility: The link between sustainability
disclosure and sustainability performance. Abacus 50: 422–459.
Hillman AJ and Keim GD (2001) Shareholder value, stakeholder management, and social issues: What’s the
bottom line? Strategic Management Journal 22: 125–139.
Hodge K, Subramaniam N and Stewart J (2009) Assurance of sustainability reports: Impact on report
users’ confidence and perceptions of information credibility. Australian Accounting Review 19:
178–194.
Holder-Webb L, Cohen JR, Nath L, et al. (2009) The supply of corporate social responsibility disclosures
among U.S. firms. Journal of Business Ethics 84: 497–527.
Idowu SO and Papasolomou I (2007) Are the corporate social responsibility matters based on good inten-
tions or false pretences? An empirical study of the motivations behind the issuing of CSR reports by UK
companies. Corporate Governance 7: 136–147.
Jamali D (2008) Stakeholder approach to corporate social responsibility: A fresh perspective into theory and
practice. Journal of Business Ethics 82: 213–231.
Johnson HH (2003) Does it pay to be good? Social responsibility and financial performance. Business
Horizons 46: 34–40.
Jones MJ (2011) The nature, use and impression management of graphs in social and environmental account-
ing. Accounting Forum 35: 75–89.
20 Australian Journal of Management 00(0)

Jones S (2017) Corporate bankruptcy prediction: A high dimensional analysis. Review of Accounting Studies
2: 1366–1422.
Jones S and Wright C (in press) Fashion or future: Does creating shared value pay?’ Accounting & Finance.
Jones S, Frost G, Loftus J, et al. (2007) An empirical examination of the market returns and financial perfor-
mance of entities engaged in sustainability reporting. Australian Accounting Review 17: 78–87.
Keele DM and DeHart S (2011) Partners of USEPA climate leaders: An event study on stock performance.
Business Strategy and the Environment 20: 485–497.
Kim Y, Park MS and Wier B (2012) Is earnings quality associated with corporate social responsibility? The
Accounting Review 87: 761–796.
Knudsen JS (2005) The Global Reporting Initiative in Denmark: Emperor’s new clothes or useful reporting
tool? Available at: http://www.copenhagencentre.org/ (accessed 20 January 2009).
Kolk A, Hong P and van Dolen W (2010) Corporate social responsibility in China: An analysis of domestic
and foreign retailers’ sustainability dimensions. Business Strategy and the Environment 19: 289–303.
Kolk A, Walhain S and van de Wateringen S (2001) Environmental reporting by the Fortune Global 250:
Exploring the influence of nationality and sector. Business Strategy and the Environment 10: 15–28.
KPMG (2011) KPMG International Survey of Corporate Responsibility Reporting 2011. Amstelveen: KPMG
International.
KPMG (2013) The KPMG Survey of Corporate Responsibility Reporting. Amstelveen: KPMG International.
KPMG (2015) Currents of Change: The KPMG Survey of Corporate Responsibility Reporting. Amstelveen:
KPMG International.
Laufer WS (2003) Social accountability and corporate green washing. Journal of Business Ethics 43: 253–261.
Legendre S and Coderre F (2013) Determinants of GRI G3 application levels: The case of the Fortune Global
500. Corporate Social: Responsibility and Environmental Management 20: 182–192.
Lin L-W (2010) Corporate social responsibility in China: Window dressing or structural change. Berkeley
Journal of International Law 28: 64–100.
Liu X and Anbumozhi V (2009) Determinant factors of corporate environmental information disclosure: An
empirical study of Chinese listed companies. Journal of Cleaner Production 17: 593–600.
Liu X, Yu Q, Fujitsuka T, et al. (2010) Functional mechanisms of mandatory corporate environmental disclo-
sure: An empirical study in China. Journal of Cleaner Production 18: 823–832.
Lynn M (1992) A note on corporate social disclosure in Hong Kong. The British Accounting Review 24:
105–110.
McWilliams A and Siegel D (2001) Corporate social responsibility: A theory of the firm perspective. Academy
of Management Review 26: 117–127.
Margolis JD and Elfenbein HA (2008) Do well by doing good? Don’t count on it. Harvard Business Review
86: 19–20.
Margolis JD, Elfenbein HA and Walsh JP (2007) Does it pay to be good? A meta-analysis and redirection
of research on the relationship between corporate social and financial performance. In Proceedings of
the Academy of Management annual meeting, Philadelphia, PA. Available at: https://sites.hks.harvard.
edu/m-rcbg/papers/seminars/margolis_november_07.pdf
Marti CP, Rovira-Val MR and Drescher LGJ (2015) Are firms that contribute to sustainable development bet-
ter financially? Corporate Social Responsibility and Environmental Management 22: 305–319.
Martinov-Bennie N, Frost G and Soh DSB (2012) Assurance on sustainability reporting: State of play and
future directions. In: Jones S and Ratnatunga J (eds) Contemporary Issues in Sustainability Accounting,
Assurance and Reporting. Bingley: Emerald Publishing, pp. 267–283.
Moser DV and Martin PR (2012) A broader perspective on corporate social responsibility research in account-
ing. The Accounting Review 87: 797–806.
Murray A, Sinclair CD, Power DM, et al. (2006) Do financial markets care about social and environmental
disclosure? Accounting, Auditing and Accountability Journal 19: 228–255.
Neu D, Warsame H and Pedwell K (1998) Managing public impressions: Environmental disclosures in annual
reports. Accounting, Organizations and Society 23: 265–282.
Noronha C, Tou S, Cynthia MI, et al. (2013) Corporate social responsibility reporting in china: An overview and
comparison with major trends. Corporate Social Responsibility and Environmental Management 20: 29–42.
Beck et al. 21

O’Dwyer B (2011) The case of sustainability assurance: Constructing a new assurance service. Contemporary
Accounting Research 28: 1230–1266.
O’Neill HM, Saunders CB and McCarthy AD (1989) Board members, corporate social responsiveness and
profitability: Are tradeoffs necessary? Journal of Business Ethics 8: 353–357.
Orlitzky M and Benjamin JD (2001) Corporate social performance and firm risk: A meta-analytic review.
Business & Society 40: 369–396.
Orlitzky M, Schmidt FL and Rynes SL (2003) Corporate social and financial performance: A meta-analysis.
Organization Studies 24: 403–441.
Palepu K, Healy P, Wright S, et al. (2015) Business Analysis and Valuation: Using Financial Statements. 2nd
ed. Boston, MA: Cengage Learning.
Patten DM and Crampton W (2004) Legitimacy and the internet: An examination of corporate webpage envi-
ronmental disclosure. Advances in Environmental Accounting and Management 2: 31–57.
Porter ME and Kramer MR (2011) Creating shared value. Harvard Business Review 89: 62–77.
Raar J (2007) Reported social and environmental taxonomies: A longer-term glimpse. Managerial Auditing
Journal 22: 840–860.
Roberts RW (1992) Determinants of corporate social responsibility disclosure: An application of stakeholder
theory. Accounting, Organizations and Society 17: 595–612.
Rodrigo P and Arenas D (2008) Do employees care about CSR programs? A typology of employees accord-
ing to their attitudes. Journal of Business Ethics 83: 265–283.
Rodrigue M, Magnan M and Cho CH (2013) Is environmental governance substantive or symbolic? An
empirical investigation. Journal of Business Ethics 114: 107–129.
Schaltegger S, Viere T and Zvezdov D (2012) Tapping environmental accounting potentials of beer brewing:
Information needs for successful cleaner production. Journal of Cleaner Production 29–30: 1–10.
Steger U, Ionescu-Somers A and Salzmann O (2007) The economic foundations of corporate sustainability.
Corporate Governance 7: 162–177.
Swanson LA and Zhang DD (2012) Perspectives on corporate responsibility and sustainable development.
Management of Environmental Quality 23: 630–639.
Szekely F and Strebel H (2013) Incremental, radical and game-changing: Strategic innovation for sustain-
ability. Corporate Governance 13: 467–481.
The Association of Chartered Certified Accountants (ACCA) (2006) Getting the Message: A Snapshot of
Company Reporting in Australia Part 1: Sustainability Reporting. London: ACCA.
Wagner M (2009) Innovation and competitive advantages from the integration of strategic aspects with
social and environmental management in European firms. Business Strategy and the Environment 18:
291–306.
Weber J and Marley KA (2012) In search of stakeholder salience: Exploring corporate social and sustainabil-
ity reports. Business & Society 51: 626–649.
Weber O (2014) Environmental, social and governance reporting in China. Business Strategy and the
Environment 23: 303–317.
West BT, Welch K and Galecki AT (2015) Linear Mixed Models. New York: CRC Press.
Wiseman J (1982) An evaluation of environmental disclosures made in corporate annual reports. Accounting,
Organizations and Society 7: 53–63.