Content uploaded by Roland Gillet
Author content
All content in this area was uploaded by Roland Gillet on Jan 22, 2018
Content may be subject to copyright.
Original Article
Leading or lagging indicators
of risk? The informational
content of extra-financial
performance scores
Received (in revised form): 18th December 2016
Amos Sodjahin
is a Professor of Finance at Universite
´de Moncton (Canada) and also a researcher for the Research Group in Applied Finance
(GReFA) at Universite
´de Sherbrooke. He studied at the Paris 1 Panthe
´on-Sorbonne University and defended his thesis (PhD) at the
Paris-Dauphine University, France. His research areas include financial risk assessment and management as well as responsible
finance. Some of his studies were done in close collaboration with the financial industry, especially Banque de France (France) and
Caisse de de
´po
ˆt de placement du Que
´bec (Canada).
Claudia Champagne
is a Professor of Finance at Universite
´de Sherbrooke in Canada. She is the main researcher of the Desjardins Chair in Responsible
Finance and a member of the Research Group in Applied Finance (GReFA) where she is responsible for the financial institutions
management research theme. She has published her works in academic journals, such as the Journal of Banking and Finance and
Financial Management, and in professional journals such as Canadian Investment Review. She regularly presents her research at
international scientific conferences. She collaborates on research projects with different financial institutions in Canada and acts as
expert evaluator for the SSHRC (Canada). Her current research interests include responsible finance, risk management, and
financial institutions management.
Frank Coggins
is a Full Professor of Finance at Universite
´de Sherbrooke in Canada. He is the Chairholder of the Desjardins Chair in Responsible
Finance and Director of the Research Group in Responsible Finance (GReFA). He has been a visiting professor at Laval University
(Canada) and University Paris 1 Panthe
´on-Sorbonne (France). He has published in academic journals, such as Review of Finance,
Journal of Banking and Finance, Journal of Financial Research, and International Review of Financial Analysis. His works are also
published in professional journals such as the Journal of Risk Management in Financial Institutions, an official publication of the
Professional Risk Managers’ International Association (PRMIA). He collaborates on research projects with different financial
institutions in Canada and acts as expert evaluator for two scientific research funds, the FRQSC (Quebec, Canada) and the F.R.S.-
FNRS (Belgium). His current research interests include responsible finance, portfolio management, and market risk.
Roland Gillet
is full Professor of Finance at the Sorbonne in France, where he is a member of the PRISM laboratory and Director of the ‘‘Efficiency
of financial markets’’ section of the ‘‘Financial Regulation’’ (‘‘Re
´gulation Financie
`re’’—ReFi) Labex. He is also a Professor at the
Solvay Brussels School of Economics and Management at the Universite
´Libre de Bruxelles. In addition, he is—or was—a professor
and/or a visiting fellow/researcher at various universities worldwide: for instance, the University of Warsaw in Poland; the Universite
´
de Sherbrooke in Canada; Fudan University in Shanghai, China; and Harvard University and M.I.T. in the United States. He has
been the author of several books/works and numerous articles in leading scientific journals (for example, Journal of Banking and
Finance, Journal of Business, Finance and Accounting, Finance, International Journal of Business, European Financial
Management). He is an academic representative of the Euronext index committee as well as a scientific adviser for various
public and private institutions.
Correspondence: Claudia Champagne, Department of Finance, Business School, Universite
´de Sherbrooke and GReFA,
Sherbrooke, Canada
E-mail: Claudia.Champagne@Usherbrooke.ca
ABSTRACT This study investigates the informational content of extra-financial agency
scoring by examining the relationship between firm beta and extra-financial performance
score upgrades and downgrades. Specifically, we study the variations in the extra-financial
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
www.palgrave.com/journals
score of 266 Canadian corporations between 2007 and 2012 with a conditional model. We
find no evidence that changes in firm beta precede changes in extra-financial scores.
Rather, our results suggest that a firm’s systematic risk increases following a downgrade of
its extra-financial performance. In terms of score upgrades, the overall effect is not signifi-
cant. However, score upgrades for firms with already-high scores predict higher systematic
risk, while score upgrades for firms with low scores predict lower systematic risk. These
results suggest that extra-financial scores are informational and can be useful to portfolio
managers, notably for their risk management strategies.
Journal of Asset Management (2017) 18, 347–370. doi:10.1057/s41260-016-0039-y;
published online 11 January 2017
Keywords: corporate social responsibility; extra-financial performance; informational
content; systematic risk; conditional model
JEL Classifications: G10; G14; M14
INTRODUCTION
Firm spending in corporate social
responsibility (CSR) activities has
substantially increased in recent years.
According to Hong et al (2012), CSR
spending amounts to hundreds of millions of
dollars annually. One reason for this renewed
interest for socially responsible investments
(SRI) is that investors, including institutional
investors, are becoming more concerned
with the extra-financial consequences of
corporate decisions. Firms that neglect this
aspect may therefore face greater financial
risk due to possible actions by regulators and
activists that will affect their profitability
(e.g., Baron and Diermeier, 2007; Lyon and
Maxwell, 2011).
Growing enthusiasm for SRI has led to a
surge in rating agencies that specialize in social
and environmental rating and scoring (e.g.,
MSCI ESG STATS
1
in the US; EIRIS
2
in the
UK; Thomson Reuter’s ASSET4
3
and
Sustainalytics
4
which operate globally). In
addition to their main role of providing
investors with information on CSR strategies,
some agencies also publish extra-financial
performance scores.
5
Our study examines
whether variations in extra-financial scores
affect financial markets, similar to what
changes in credit ratings do (e.g., Jorion and
Zhang, 2007; Holthausen and Leftwic, 1986;
Weinstein, 1977). Specifically, we investigate
the information content of extra-financial
scores by examining their relationships to
firms’ systematic risk (beta). We address the
following two questions: (1) is beta related to
extra-financial performance scores and, (2) if
so, do extra-financial performance score
changes (upgrades and downgrades) lead or lag
indicators of beta? Knowing whether firm
beta varies before or following the extra-
financial performance changes is an important
practical question. Specifically, if extra-
financial performance changes predict
systematic risk changes, then extra-financial
rating agencies’ scores can be an excellent risk
management tool, particularly for institutional
investors. In theory, extra-financial
performance scores can be leading indicators
of corporate systematic risk if extra-financial
rating agencies are able, through their analysis
of a company’s environment, social or
governance (ESG) criteria, to predict future
losses or risk events such as operational or
reputational losses. If, however, rating
agencies are mostly reacting to corporate
events that are related to extra-financial
performance, then extra-financial
performance score changes will lag beta
variations. In this case, extra-financial
agencies’ scores would be less useful as
predicting tools.
Our study contributes to the literature in
several ways. Firstly, while previous studies
Sodjahin et al
348 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
investigate the impact of extra-financial
performance score levels on financial risk
(e.g., Kim et al, 2014; Bouslah et al, 2013;
Oikonomou et al, 2012), our research focuses
on the impacts of extra-financial performance
score changes on systematic risk and by
distinguishing between the effects of score
upgrades and downgrades. From a risk
manager’s point of view, score changes are
fundamentally different from score levels, as
they are related to new information about a
firm’s risk. For instance, a firm can
experience a score downgrade and still
maintain a high score, or experience a
positive change in its extra-financial score
and still have a low score.
Secondly, while most studies agree on the
existence of a relationship between extra-
financial performance and financial risk, as
evidenced by Orlitzky and Benjamin’s (2001)
meta-analysis, the question of the direction of
the relationship between the two variables is
not yet settled. Some authors argue that
systematic risk is a determinant of CSR, as
managers in lower-risk companies have
access to more stable cash flows, allowing
them to improve their extra-financial
performance (Hasseldine et al, 2005; Roberts,
1992; McGuire et al, 1988). Furthermore,
Kru
¨ger (2015) shows that the occurrence of
firm-specific events, related to
environmental, social or governance risks,
has an important influence on KLD0s
scorings. Specifically, he shows that KLD’s
scores are updated to account for information
on events that have already occurred. Others
believe that an improvement in a firm’s
extra-financial performance is likely to be
rewarded by the market in terms of improved
risk perception, and thus by a lower beta
(e.g., Oikonomou et al, 2012; Salama et al,
2011; Sharfman and Fernando, 2008).
Further, the previous studies do not address
the basic question: whether a firm’s financial
risk is low (high) because of its high (low)
CSR or whether its CSR is high (low)
because of its low (high) financial risk? The
mere observation of a negative correlation
between some annual CSR measures and
financial risk is consistent with at least two
different interpretations: either more
responsible firms tend to be less risky or,
alternatively, less-risky firms tend to channel
more resources into projects that increase
their CSR. Our extra-financial performance
data allow us to test whether beta variations
occur before or following changes in firms’
extra-financial performance.
Thirdly, our study addresses the
omnipresent issue of over-investment and
managerial opportunism (e.g., McWilliams
and Siegel, 2001; Preston and O’Bannon,
1997), which suggests that, under certain
circumstances, extra-financial performance can
be a potential source of risk, for instance,
because of overinvestment (McWilliams and
Siegel, 2001). To do so, we examine whether
extra-financial performance score upgrades for
firms with already high scores predict higher
systematic risk. Finally, in order to reflect the
qualitative differences across the dimensions of
extra-financial performance, we separately
analyze each of the three components of extra-
financial performance (i.e., environment,
social and governance) as well as in an
aggregate measure of performance.
Our results show no significant evidence
that extra-financial score changes lag beta
variations. Rather, we observe that
systematic risk increases following extra-
financial score downgrades. The overall
predictive power of score upgrades is not
significant. Therefore, extra-financial
performance scores are not simply reacting to
market information but are, particularly
downgrades, leading indicators of firm
systematic risk variations. Further, we show
that extra-financial score upgrades for firms
with already-high scores predict higher
systematic risk while they predict lower
systematic risk for firms with low extra-
financial scores. This result suggests that, for
firms with already-high extra-financial
scores, further improvements can be
counterproductive and lead to an increase in
long-term risk, possibly because of costs that
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 349
investors feel are too high and inopportune.
By contrast, systematic risk decreases when
firms with low extra-financial scores make an
effort to improve their social image.
Regarding score downgrades, as for our
general results, we find that they predict
higher systematic risk for firms with low
extra-financial scores.
The rest of the study is structured as
follows. ‘‘Prior Research on the Impact of
Extra-Financial Performance on Financial
Risk’’ section presents a summary of the
literature on the impact of extra-financial
performance on shareholder wealth (return
and risk). ‘‘Theoretical Framework and
Research Hypotheses’’ section presents the
theoretical framework and research
hypotheses. ‘‘Data and Methodology’’ section
describes the data and the methodology used
in order to test our hypotheses. ‘‘Empirical
Results and Discussion’’ section presents and
discusses our empirical results, and finally,
‘‘ Conclusion’’ section concludes the paper.
PRIOR RESEARCH
ON THE IMPACT OF EXTRA-
FINANCIAL PERFORMANCE
ON FINANCIAL RISK
Unlike the abundant literature on the impact of
extra-financial performance on firm financial
performance, there are few studies that
examine the relationship between financial risk
and extra-financial performance. These few
studies analyze different measures of financial
risk, such as variance and its components
(idiosyncratic risk, and systematic risk). Some
studies suggest that extra-financial performance
affects only idiosyncratic risk because extra-
financial performance is firm specific. For
example, using data between 1995 and 1999
from the ‘‘Canadian Social Investment
database’’, Boutin-Dufresne and Savaria (2004)
find a negative relationship between CSR and
firm idiosyncratic risk. This observation is
confirmed by Lee and Faff (2009), who study
the impact of CSR on financial risk for firms
listed in the Dow Jones Sustainability Index.
The authors demonstrate that socially
responsible firms are less risky than their socially
irresponsible counterparts. Using Fortune’s
MAC data between 2002 and 2003 as a
measure of CSR, Luo and Bhattacharya (2009)
show that CSR decreases a firm’s idiosyncratic
risk and provides insurance against the volatility
of the firm’s future cash-flows. Mishra and
Modi (2012) and Bouslah et al (2013)confirm
this result by using KLD data as a principal
proxy for extra-financial performance. Mishra
and Modi (2012) observe that CSR has a
significant effect on idiosyncratic risk over the
period spanning from 2000 to 2009, with
positive CSR scores reducing risk and negative
CSR scores increasing it. Bouslah et al (2013)
focus their analysis on the individual
components of extra-financial performance.
They find that financial risk (measured by
idiosyncratic risk as well as stock return
volatility) is negatively related to two CSR
components, namely employee relations and
human rights, while other CSR components
do not affect financial risk.
Several studies argue that investigating the
effects of CSR on systematic risk is more
relevant because, in the absence of market
imperfections, only systematic risk is priced;
idiosyncratic risk can be eliminated through
diversification. McGuire et al (1988) find that
CSR, proxied by firm ranking in Fortune’s
list of America’s most admired companies
(MAC), is negatively related to market risk
loadings over the period 1983–1985. In this
study, beta is a lagged independent variable,
so that low financial risk is theorised to create
the planning certainty that facilitates
investment in CSR. Unlike McGuire et al
(1988) and Luo and Bhattacharya (2009)
consider systematic risk as the dependent
variable and simultaneously control for small
cap and book-to-value effects in their
systematic and idiosyncratic risk estimations.
The authors conclude that a firm’s extra-
financial performance, as evaluated by
Fortune magazine, is negatively correlated
with beta measures. Salama et al (2011)
Sodjahin et al
350 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
address this issue in the UK by examining
firm activity from 1994 to 2006. They also
consider systematic risk as the dependent
variable and predominantly focus on
environmental responsibility. Their results do
not stray too far from those related to CSR in
the American context and attest that the
environmental performance of UK firms is
inversely related to systematic risk. Jo and Na
(2012) find that a firm’ overall CSR
engagement alleviates not only total risk, but
also systematic risk and sensitivities to market
fluctuations, particularly for controversial
industries in the US Also in the US context,
Oikonomou et al (2012) present a
longitudinal study that analyzes the
relationship between corporate extra-
financial performance and systematic risk
between 1992 and 2009 using the KLD
database and find that CSR is negatively
related to systematic risk. The authors find a
negative (positive) relation between
systematic risk and a measure of aggregate
extra-financial strengths (concerns).
However, they also find that only
community, employment, and
environmental concerns are significantly and
positively related to systematic risk. The
authors also note that the impact of ESG
criteria on firm risk varies according to the
economic context measured with market
volatility. Lastly, Kim et al (2014) show that if
socially responsible firms commit to a high
standard of transparency they would have
lower crash risk. However, if managers
engage in CSR to cover up bad news and
divert shareholder scrutiny, CSR would be
associated with higher crash risk.
THEORETICAL FRAMEWORK
AND RESEARCH HYPOTHESES
There are two major theoretical arguments
that link corporate extra-financial
performance to financial risk. The first posits
that a high extra-financial performance brings
about extra operating costs and potential
sacrifices and, hence, puts firms with high
extra-financial performance scores into a risk
disadvantage. The second argument, based
on the stakeholder theory, contends that
although increasing its extra-financial
performance can be costly for a firm, it can
reduce other costs and/or improve revenues
and thereby decrease financial risk.
Stakeholder theory
Stakeholder theory states that every modern
firm has explicit and implicit relationships
with a variety of stakeholders who have the
power to determine its success or failure (e.g.,
Jones, 1995; Wijnberg, 2000).
The advantages of adopting a CSR
approach that takes into account stakehold-
ers’ interest are multiple and go with the
principles of a risk management system main
objective of which is to prevent or avoid the
disruption, loss or damage to business oper-
ations. For example, the fact that all stake-
holders (including shareholders) feel more
involved in the decision-making process
reduces information asymmetry (see, for e.g.,
Waddock and Graves, 1997) and uncertainty
about future cash flows (e.g., McGuire et al,
1988). Sharfman and Fernando (2008) argue
that risk management of social or environ-
mental issues is theoretically synonymous
with strategic risk management because it
reduces potential risks (e.g., accidents, labor
disputes, consumer boycotts, damage to
brand image, and reputation), lowers favor-
able investor recognition and, consequently,
reduces the number of potential claimants on
a firm’s cash flows (e.g., potential fines,
compliance cost, etc.). Sharfman and Fer-
nando (2008) conclude that, when potential
litigations are reduced, cash flows are more
stable, and a firm’s resources can be dedicated
to strategic decisions and investments that
contribute to reducing the financial risk
perceived by the market (i.e., systematic
risk).
However, the failure of firms to meet the
claims of implicit stakeholders can result in
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 351
costly explicit claims (e.g., lawsuits, regula-
tory intervention etc.) to force their hands.
Investors can anticipate this situation and
consider investment in these firms as risky
(e.g., Stern, 2006; Porter and Kramer, 2006).
Assuming that stakeholder claims are of
similar nature across all firms, these collective
claims might lead to a systematic event, such
as a downturn in the economic cycle or a
change in the legislative framework and can
have systematic effects on all firms or com-
mon groups of firms.
Our study attempts to test the informa-
tional content of extra-financial rating agency
scorings by examining the relationship
between firm betas and extra-financial per-
formance score changes. Following the above
discussion, we expect that downgrades (up-
grades) in extra-financial performance scores
are related to increases (decreases) in firms’
systematic risk. This is summarized in our
first research hypothesis:
H1 Extra-financial performance score chan-
ges are related to firms’ systematic risk variations.
As mentioned previously, while there
seems to be consensus on the relationship
between extra-financial performance and
financial risk, as evidenced by Orlitzky and
Benjamin (2001), the direction of the causal
link between the two variables is still an
empirical issue. Roberts (1992) and Has-
seldine et al (2005) argue that systematic risk
is a determinant of CSR, as managers in
lower-risk companies have access to more
stable cash flows, allowing them to improve
their extra-financial performance. Others
believe that the improvement of extra-fi-
nancial performance is likely to be rewarded
by the market in terms of improved risk
perception, and thus by a lower beta (e.g.,
Oikonomou et al, 2012; Salama et al, 2011;
Sharfman and Fernando, 2008). To address
this dual link, we decompose H1 into two
testable subhypotheses (H1a and H1b). If
extra-financial rating agencies are able,
through their analysis of a company’s ESG
criteria, to predict future losses or risk events
(such as operational or reputational losses),
then extra-financial scores should be leading
indicators of systematic risk. This is high-
lighted in subhypothesis H1a:
H1a Extra-financial performance scores are
leading indicators of systematic risk.
If, however, rating agencies are mostly
reacting to corporate events that are related
to extra-financial performance, then extra-
financial performance scores will be lagging
indicators of systematic risk. This is summa-
rized in subhypothesis H1b:
H1b Extra-financial performance scores
are lagging indicators of systematic risk.
We anticipate the relationship between
extra-financial performance and financial risk
to be asymmetrical. As argued by some
authors (e.g., Mattingly and Berman, 2006;
or Oikonomou et al, 2012), it is unreasonable
to assume that stakeholders will react to
responsible and irresponsible behaviors in
opposite yet symmetrical manners. Further-
more, there is recent empirical evidence that
CSR and corporate social irresponsibility affect
a firm’s bottom line to differing magnitudes
(Kru
¨ger, 2015). This is summarized in the
second research hypothesis:
H2 The informational content of extra-fi-
nancial performance score downgrades is
higher than score upgrades.
Over-investment
and managerial opportunism
theories
Over-investment and managerial oppor-
tunism theories (e.g., McWilliams and Siegel,
2001; Preston and O’Bannon, 1997) support
a positive relationship between extra-finan-
cial performance and financial risk. Accord-
ing to the proponents of these theories,
managers may choose to improve their firm’s
extra-financial performance score at the
expense of shareholders by over-investing in
CSR activities in order to build their own
personal reputation as good social citizens
Sodjahin et al
352 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
(Barnea and Rubin, 2010) or to generate
support from social and environmental acti-
vists, local communities, politicians, NGOs,
etc., in order to reduce the probability of
their replacement in a future period (Cespa
and Cestone, 2007) or even to hide bad
management (Hemingway and Maclagan,
2004). This strategy, if known, would be
sanctioned by a higher financial risk. Lastly,
McWilliams and Siegel (2001) believe that
there is an optimal level of extra-financial
performance, beyond which it is less likely to
shield the firm against the uncertainty and
vulnerability of future cash flows.
At very high levels of extra-financial per-
formance, the disadvantages of CSR in the
context of a firm’s economic purposes may
outweigh its benefits, thus likely inducing
more unstable future profits and less insur-
ance-like protection against stock return risk.
This is summarized in our risk-related third
research hypothesis:
H3 The relationship between extra-finan-
cial performance and financial risk is stronger
for firms with already-high extra-financial
performance scores.
DATA AND METHODOLOGY
Data
We use corporate social ratings data from the
Sustainalytics database. Sustainalytics spe-
cialises in the measurement of corporate
extra-financial performance against a prede-
termined set of criteria, as shown in Appen-
dix, and is principally used by institutional
investors. Unlike MSCI ESG, which evalu-
ates CSR based on seven qualitative criteria,
Sustainalytics scores firms on over 100 pro-
prietary indicators for the three ESG criteria.
Furthermore, unlike MSCI ESG, which
assigns positive and negative ratings (i.e.,
strengths and concerns), Sustainalytics’ extra-
financial performance scores range from 0
(worst) to 10 (best). Daily returns and
macroeconomic variables used herein are
collected from the Canadian Financial Mar-
kets Research Center (CFMRC) and
Bloomberg databases.
Our final sample consists of 266 publicly
traded firms listed on the Toronto Stock
Exchange for which there are at least two
observations from January 2007 to December
2012 in the Sustainalytics database. In total,
2213 extra-financial score changes are stud-
ied, consisting of 1312 upgrades and 901
downgrades. Table 1presents descriptive
statistics on extra-financial score levels and
changes for the firms in our sample. Score
changes and levels are presented for the
aggregate score and by dimension (environ-
mental, social, and governance). Extra-fi-
nancial performance changes appear to be
asymmetric, as evidenced by the skewness
coefficients reported in Panel A of Table 1.
Specifically, score changes are negatively
skewed for the environment and social
dimensions and positively skewed for the
governance dimension as well as for the
aggregate score. The Jarque–Bera statistics are
significant for all series and confirm that an
assumption of normality is not verified. In
addition, statistical tests (ttests for the mean
and Wilcoxon signed-rank test for the med-
ian) do not reject the null hypotheses for
both the mean and median of extra-financial
performance changes at the 99 per cent level
of significance.
From Panel B, which shows score change
statistics for upgrades and downgrades, we
note that score upgrades are more frequent
than score downgrades, as evidenced by the
higher number of observations in the first
case.
Methodology
Both theoretical developments and empirical
evidence suggest that systematic risk is not
constant, but changes over time (e.g., Ferson
and Schadt, 1996; Christopherson et al, 1998;
Champagne et al, 2015). These changes are
related to predetermined information vari-
ables. Our empirical model extends the
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 353
Table 1: Descriptive statistics on extra-financial performance scores
Panel A: descriptive statistics for extra-financial performance score levels and changes
Extra-financial performance
(score/10)
Level Change Obs.
Mean Std Min Max Med. Skew. Kurt. Mean Std Min Max Med. Skew. Kurt. Jarque–
Berra
Aggregate 5.467* 0.837 2.700 7.954 5.400* -0.258 0.300 0.022 1.282 -6.708 6.780 0.023 0.065 3.010 837.13* 2213
Environment 5.387* 1.000 2.929 8.900 5.301* 0.313 0.013 0.023 1.317 -7.657 5.932 0.024 -0.016 3.012 709.34* 1876
Social 5.458* 1.479 1.000 9.586 5.510* -0.280 -0.013 0.031 1.406 -6.277 7.602 0.030 -0.075 2.981 538.80* 1451
Governance 6.493* 1.065 3.500 9.686 6.500* 0.050 -0.629 -0.005 1.421 -6.577 6.612 -0.005 0.088 2.978 640.33* 1727
Panel B: descriptive statistics for extra-financial performance score upgrades and downgrades
Extra-financial performance
(score/10)
Score upgrades Score downgrades
Mean Std Min Max Med. Observations Mean Std Min Max Med. Observations
N # of firms N # of firms
Change
Aggregate 0.206 0.210 0.002 2.074 0.534 1312 236 -0.141 0.159 -2.400 -0.002 -0.530 901 244
Environment 0.565 0.598 0.002 3.547 0.581 1104 254 -0.457 0.573 -5.031 -0.014 -0.430 772 235
Social 0.713 0.680 0.000 5.172 0.735 930 252 -0.498 0.493 -3.093 -0.004 -0.696 521 221
Governance 0.572 0.552 0.004 4.360 0.524 867 232 -0.693 0.667 -4.360 -0.088 -0.523 860 254
Note: This table presents descriptive statistics for the extra-financial performance scores for a sample of 266 Canadian firms between 2007 and 2012. Statistics are presents for
the aggregate score and for each extra-financial score dimension: Environment, Social, and Governance. Panel A presents statistics for score levels and changes, while Panel B
shows statistics for score upgrades and downgrades. Nis the number of observations used to estimate the statistics. ‘‘*’’ indicate rejection of the null hypotheses for the mean,
median, and normality at the 99 per cent significance level for statistical tests (respectively ttest for the mean, Wilcoxon signed-rank test for the median, and Jarque–Bera test for
normality).
Sodjahin et al
354 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
conditional modeling approach suggested by
Christopherson et al (1998), by adding extra-
financial performance score changes.
6
For-
mally, similar to the methodology used by
Champagne et al (2015), our conditional
model incorporates changes in extra-financial
scores as well as publicly available economic
instruments into financial performance esti-
mation to account for the possibility of time
variation in betas and abnormal performance
(alpha). After integrating Fama–French’s
(1993) factors, our empirical model is of the
form:
Rit Rft ¼aitðdEt ;Zn;t1Þ
þbitðdEt ;Zn;t1ÞðRmt Rft Þ
þb1iSMBtþb2iHMLtþb3iJant
þb4iMontþeit;ð1Þ
where Rit Rft is the excess daily return of
firm ion day t.Rit and Rft respectively,
designate the return for firm iand the risk-free
rate (i.e., the daily yield on a 90-day maturity
government bond) on day t. Market portfolio
return Rmt is the value-weighted stock return
of the S&P-TSX index. Risk factors HMLt
and SMBtrespectively represent the book-to-
market ratio effect and the size effect (Fama
and French, 1993). Jantand Montare binary
variables that control for the January and
Monday effects, respectively, and equal 1 for
the control period and 0 otherwise. dEt is a
dummy variable that equals 1 if day tis
included in the event window (i.e., starting
when we observe a change in firm i’s extra-
financial score), and 0 otherwise. Vector
Zn;t1with n¼1;2;...;5, includes the five
7
macroeconomic information variables that
condition beta, bit ðdEt;Zn;t1Þ.eit is the error
term for firm iand eit Nð0;rÞ:
Conditional beta for firm iis defined as
follows:
bitðdEt ;Zn;t1Þ¼b0iþbEi dEt þX
5
n¼1
bnizn;t1;
ð2Þ
where z
n,t-1
=Z
n,t-1
-E(Z
n
)isavector
of the deviations of Z
n,t-1
from the
unconditional means. b0imeasures the
average conditional beta unrelated to score
changes and macroeconomic information
variables. Parameters bni (for n=1,…,5)
measure conditional beta’s sensitivity
to the five macroeconomic information
variables, Zn;t1.bEi, represents beta varia-
tions associated with extra-financial per-
formance score changes (upgrades or
downgrades). More specifically, bEi mea-
sures the difference between beta estimated
with the model that takes into account
the changes in extra-financial firms
scores and beta estimated without taking
these changes into account. Formally, since:
for dEt ¼1;bitð1;Zn;t1Þ¼b0iþbEi:1þ
P
5
n¼1
bnizn;t1and for dEt ¼0;bit ð0;Zn;t1Þ¼
b0iþbEi:0þP
5
n¼1
bnizn;t1,
we obtain:
bEi ¼bitð1;Zn;t1Þbit ð0;Zn;t1Þ
¼ðb0iþbEi:1þX
5
n¼1
bnizn;t1Þ
ðb0iþbEi:0þX
5
n¼1
bnizn;t1Þ:ð3Þ
We estimate t-stats for the models
using the heteroskedasticity-consistent esti-
mation techniques of Newey and West
(1987).
To test H1a and H1b, we estimate the
average coefficient bEi around changes in
extra-financial performance scores. More
specifically, we estimate this average coeffi-
cient on analysis periods of 60 and 120 days
before changes in extra-financial scores (i.e.,
[-60; 0], [-120; 0]), and on analysis periods
of 60, 120 and 250 days following changes in
extra-financial scores (i.e., [0; +60], [0; 120],
and [0; +250]).
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 355
EMPIRICAL RESULTS
AND DISCUSSION
Preliminary results
We first examine whether the conditional
model framework commonly used in the US
context is appropriate in this study which
focuses on the Canadian context. To do so,
we estimate models (1) and (2) without the
term bEidEt over the [-500; 250] window,
which corresponds to the 250 days following
changes in extra-financial scores. Table 2
shows the results of these regressions. We
note that our augmented conditional Fama–
French model is relevant. Specifically, the
coefficients for both HML and SMB are
positive and significant. Further, while the
January (Jan) effect is not clear, the Monday
(Mon) effect is relevant with a positive and
significant coefficient. Finally, coefficients for
macroeconomic information variables
(Z
n,t-1
) are highly significant, which indicate
that systematic risk (beta) is a function of the
economic context. The use of a conditional
model is therefore justified to disentangle the
impact of the economic context from the
impact of extra-financial score changes on
corporate financial risk. Because extra-fi-
nancial performance and macroeconomic
Table 2: Conditional model and time-varying beta
Variable Score upgrades Score downgrades
Agg. Envir. Social Gov. Agg. Envir. Social Gov.
SMB 0.1824 0.1980 0.2052 0.1999 0.1799 0.2236 0.2190 0.1671
(7.32) (9.66) (6.23) (8.81) (8.70) (6.72) (9.51) (8.63)
HML 0.1358 0.1240 0.1147 0.1260 0.1092 0.1318 0.1192 0.1399
(5.85) (5.26) (4.17) (4.19) (5.27) (5.38) (5.35) (5.14)
Jan -0.0001 0.0001 -0.0002 -0.0002 0.0000 0.0000 -0.0001 -0.0001
(-1.26) (0.52) (-1.77) (-1.61) (0.02) (0.17) (-0.31) (-0.40)
Mon -0.0004 -0.0004 -0.0004 -0.0005 -0.0003 -0.0003 -0.0004 -0.0003
(-6.36) (-4.39) (-4.49) (-5.88) (-3.29) (-3.03) (-2.96) (-2.86)
b
0i
0.0003 0.0005 0.0002 0.0000 0.0002 0.0003 0.0004 0.0003
(1.46) (1.85) (0.58) (-0.03) (0.77) (1.00) (1.01) (0.972)
Short-term
rate
-0.0372 -0.0290 -0.0430 -0.0329 -0.0315 -0.0253 -0.0768 -0.0082
(-4.64) (-3.01) (-3.76) (-3.07) (-2.95) (-2.14) (-4.97) (-2.01)
Term
structure
slope
0.0272 0.0233 0.0272 0.0264 0.0271 0.0290 0.0279 0.0233
(5.43) (5.68) (2.63) (2.87) (4.20) (6.85) (4.82) (5.80)
Stock
market
return
0.0368 0.0385 0.0318 0.0273 0.0321 0.0340 0.0299 0.0249
(4.22) (4.90) (3.11) (2.47) (3.07) (3.16) (2.98) (1.98)
Stock
market
implied
volatility
0.0209 0.0263 0.0254 0.0264 0.0215 0.0239 0.0247 0.0251
(2.42) (2.68) (2.78) (3.87) (3.20) (4.53) (2.63) (1.91)
Credit
spread
0.0447 0.0551 0.0593 0.0468 0.0573 0.0732 0.0354 0.0535
(10.34) (10.40) (10.40) (8.61) (10.22) (11.86) (4.79) (8.90)
Adj-R
2
0.228 0.236 0.238 0.229 0.231 0.237 0.237 0.224
F-stat 7092.45 5426.44 4266.14 4281.57 4649.01 4122.91 2513.37 3402.86
Obs. 1312 1104 930 867 901 772 521 860
Note: This table presents the regression results from the estimation of a different specification of model (1) in which
the score change components have been removed. The coefficients for the remaining four variables: SMB, HML,
Jan, and Mon are presented in the top part of the table (variables are defined in section ‘‘Data and Methodology’’).
The bottom part of the table shows the results for the specification of model (2) in which the score change
component has been removed. The coefficients for b
0
i, which represents the average conditional beta, and for each
of the five macroeconomic variables (Z
t
) (i.e., the short-term interest rates, the term structure slope, the stock market
return, the stock market (implied) volatility, and the credit spread) are presented in the bottom part of the table. The
analysis period is [-500; 250] days. Models are estimated on a subsample of score upgrade observations and a
subsample of score downgrades and, in each case, for the four types of score dimensions [aggregate (agg.),
environment (envir.), social, and governance (gov.)]. Our sample includes 266 Canadian corporations from January
2007 to December 2012. The estimated coefficients’ mean values are presented, with tstatistics in parentheses.
Numbers in bold indicate significance at the 10 per cent level.
Sodjahin et al
356 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
factors are known to co-vary (e.g., Albu-
querque et al, 2014; Oikonomou et al, 2012;
Chen et al, 2010), the ability to separate the
two effects is crucial. The fact that our sample
period covers the period from 2007 to 2012,
which is characterized by financial turmoil,
provides an even stronger case for the use of a
conditional model.
Beta variations around changes
in extra-financial performance
Table 3presents the mean values for coef-
ficients bEi that measure beta variations
around changes in firms’ extra-financial
performance scores [see model (2)] for both
the unconditional model (estimated without
the information variables, Z
n,t-1
) and con-
ditional model.
8
Results show extra-finan-
cial score changes are negatively related to
systematic risk, irrespective of the model
used. Specifically, score upgrades are related
to beta decreases and score downgrades are
related to beta increases. Significant rela-
tionships are observed almost exclusively for
the post-score-changes periods, supporting
the hypothesis that extra-financial perfor-
mance scores are leading indicators of sys-
tematic risk (H1a). Further, the leading
effect is mostly associated with score
downgrades, as opposed to upgrades (with
the exception of score upgrades for the
governance dimension), that are leading
indicators of beta decreases. Specifically, we
can observe that downgrades for almost all
extra-financial scores are followed by sig-
nificant increases in betas. For example,
under the conditional model, governance
performance score downgrades are followed
by beta increases of 0.014 (t=2.38), 0.021
(t=3.74) and 0.017 (t=1.94) for the [0;
+60], [0; +120], and [0; +250] periods fol-
lowing score changes, respectively. The
pattern is similar for downgrades in the
social and environmental dimensions, as
well as for the aggregate score. These results
are consistent with an asymmetrical
relationship between extra-financial perfor-
mance and systematic risk (H2).
Oikonomou et al (2012) examine the
association between corporate social perfor-
mance and financial risk for S&P 500 com-
panies between the years 1992 and 2009 and
also find that CSR is negatively but weakly
related to systematic firm risk and that cor-
porate social irresponsibility is positively and
strongly related to financial risk. Their
results, as well as ours, are consistent with the
stylized fact according to which financial
markets react more strongly to bad news. For
example, De Bondt and Thaler (1985) show
that investors ‘‘overreact’’ to unexpected and
dramatic news events. Avouyi-Dovi and
Neto (2004) observe that the asymmetric
reaction to the signs of shocks can be
explained by market participants’ long posi-
tions on equity markets that would make
them more sensitive to negative shocks.
The observed leading effect is the stron-
gest for downgrades in the environment
dimension. For example, under the condi-
tional model, environment score downgrades
are related to beta increases of 0.049
(t=7.09) for the 120-day period following
score downgrades, and related to increases of
0.016 (t=1.93), 0.021 (t=3.74) and 0.011
(t=1.90) when downgrades involve social,
governance, and aggregate scores, respec-
tively. Increasing betas following extra-fi-
nancial performance score downgrades (i.e.,
negative stakeholder information) is consis-
tent with the view that there is a substantial
and non negligible cost associated with
environmental irresponsibility.
Overall, we find that beta variations fol-
lowing extra-financial performance score
changes are economically and statistically
significant. These results validate hypothesis
H1a, which postulates that extra-financial
performance scores are leading indicators of
systematic risk. Further, the leading effect is
asymmetrical, as most significant relationships
are observed following score downgrades,
which is consistent with hypothesis H2.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 357
Table 3: Impact of extra-financial performance changes on systematic risk
Analysis period Score upgrades Score downgrades
Unconditional model Conditional model Unconditional model Conditional model
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
[-120; 0] -0.002 0.004 -0.001 -0.001 -0.003 0.017 -0.002 -0.042 0.012 0.035 0.004 0.001 0.010 0.022 0.002 0.004
(-0.29) (0.13) (-0.00) (-0.32) (-0.21) (0.48) (-0.26) (-0.09) (0.48) (0.91) (0.11) (0.06) (0.65) (1.06) (1.01) (0.35)
[-60; 0] -0.006 0.008 -0.001 -0.003 -0.004 -0.003 -0.002 -0.013 0.009 0.042 0.006 0.012 0.008 0.039 0.003 0.009
(-0.15) (0.63) (-0.07) (-0.01) (-0.46) (-0.33) (-0.17) (-0.38) (0.82) (1.59) (0.34) (1.89) (1.11) (1.47) (1.04) (1.72)
[0; +60] -0.006 0.001 -0.014 -0.016 -0.007 -0.011 -0.011 -0.009 0.013 0.038 0.002 0.018 0.010 0.035 0.006 0.014
(-1.23) (0.09) (-1.47) (-1.88) (-1.19) (-1.42) (-1.01) (-1.71) (1.92) (3.83) (0.48) (2.61) (1.97) (2.38) (0.84) (2.38)
[0; +120] -0.004 -0.023 -0.018 -0.015 -0.005 -0.002 -0.021 -0.013 0.018 0.062 0.020 0.023 0.011 0.049 0.016 0.021
(-0.26) (-1.60) (-1.50) (-4.71) (-1.55) (-0.26) (-1.38) (-4.11) (2.33) (8.41) (2.74) (4.11) (1.90) (7.09) (1.93) (3.74)
[0; +250] -0.003 -0.005 -0.028 -0.011 -0.004 -0.002 -0.023 -0.009 0.011 0.041 0.027 0.015 0.008 0.036 0.023 0.017
(-1.43) (-1.11) (-1.33) (-2.46) (-1.29) (-1.14) (-1.33) (-2.58) (2.49) (5.92) (5.62) (1.82) (1.72) (4.29) (5.02) (1.94)
Obs. 1312 1104 930 867 1312 1104 930 867 901 772 521 860 901 772 521 860
Note: This table presents the results for the estimation of model (2). Mean values for b
Ei
are shown, with t-statistics in parentheses. Other coefficients in model (2) are estimated
but not shown to save valuable space. Model (2) is estimated on a subsample of score upgrades and a subsample of score downgrades and, in each case, for the four types of
score dimensions [aggregate (agg.), environment (envir.), social, and governance (gov.)]. Five analysis periods are considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0;
+250]), for which the estimation periods are, respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and [-500; +250]. The overall sample includes 266 Canadian
firms from January 2007 to December 2012. Numbers in bold indicate significance at the 10 per cent level.
Sodjahin et al
358 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
Conditioning on the current
extra-financial performance
of the firm
One of the problems with focusing on
average beta variations estimated over the
full sample is that it can conceal cases or
special circumstances for which extra-fi-
nancial score upgrades (i.e., positive stake-
holder information) are related to beta
increases. For instance, as previously out-
lined, we could observe that, dependent on
certain conditions (e.g., overinvestment),
CSR projects can increase systematic risk.
To explore this conditionality further, we
separate the firms in our sample into quin-
tiles according to their extra-financial scores.
We then estimate model (2) on a subsample
of highly scored firms, defined as firms in
the highest quintile in terms of extra-fi-
nancial performance, and on a subsample of
lowly scored firms, defined as firms in the
lowest quintile in terms of extra-financial
performance.
Table 4provides results for highly scored
firms, which show that extra-financial score
upgrades for firms with already-high score
predict higher systematic risk, particularly
for the environmental and social dimen-
sions, as well as for the aggregate score. In
contrast, score downgrades have no signifi-
cant impact on beta, except for the gover-
nance dimension for which score
downgrades are related to beta increases in
the 250-day period following score changes.
These results support hypothesis H3 and
suggest that, for firms with already-high
extra-financial performance, further
improvements can be counterproductive
andleadtoanincreaseinsystematicrisk,
possibly because of costs that investors feel
are too high and inopportune. Our results
also suggest that extra-financial irresponsi-
bility (with the exception of the governance
dimension) is not related to any future sys-
tematic risk increase if the firm already has a
high extra-financial score.
Our results corroborate those from pre-
vious studies, including McWilliams and
Siegel (2001) who determine that an optimal
level of social performance exists beyond
which it is less likely to shield the firm against
the uncertainty and vulnerability of future
cash flows. At extremely high levels of social
performance, the drawbacks (costs) of CSR
programs may outweigh the advantages
(Handelman and Arnold, 1999; Smith,
2003), eventually leading to an increase in
systematic risk for those firms. Other studies
take an agency-cost perspective and express a
negative view on the managerial motivations
for pursuing CSR (e.g., Jensen and Meck-
ling, 1976; Friedman, 1970; McWilliams
et al, 2006; Kru
¨ger, 2015). These studies
argue that managers may opportunistically
use CSR to advance their careers or other
personal agenda. Hemingway and Maclagan
(2004) argue that one motivation for com-
panies to adopt CSR is to cover up corporate
misbehavior. The infamous firm Enron, for
example, was widely viewed as a model of
CSR and won several national awards for its
environmental and community programs
while at the same time engaging in massive
accounting frauds that lead to its collapse in
2001 (Bradley, 2009). If firms use CSR as a
tool to disguise bad news and divert share-
holder scrutiny, CSR may then be associated
with higher financial risk.
By contrast, according to Table 5, which
provides results for firms with low extra-fi-
nancial performance, we see that score
upgrades are associated with lower systematic
risk while score downgrades predict higher
systematic risk. The leading effect of score
upgrades for low-score firms is also greater in
terms of magnitude and significance than for
the full sample of firms. These results suggest
that the efforts of low-score firms to improve
their social image are related to future
decreases in their systematic risk, while their
social irresponsibility is related to increases in
their systematic risk.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 359
Table 4: Beta changes around changes in extra-financial performance for highly scored firms
Analysis period Score upgrades Score downgrades
Unconditional model Conditional model Unconditional model Conditional model
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
[-120; 0] 0.008 0.022 0.036 -0.012 0.002 0.017 0.030 -0.010 -0.011 -0.001 -0.053 0.024 0.000 -0.003 -0.046 0.020
(0.93) (0.33) (0.25) (-1.36) (0.68) (0.68) (1.02) (-1.24) (-0.71) (-0.00) (-0.21) (0.40) (-0.00) (-0.01) (-0.05) (0.24)
[-60; 0] 0.003 0.031 0.054 -0.022 0.004 0.039 0.042 -0.022 -0.009 -0.002 -0.084 0.029 -0.006 -0.001 -0.070 0.023
(0.71) (1.11) (1.02) (-1.62) (0.24) (1.06) (0.55) (-1.27) (-0.35) (-0.55) (-0.20) (1.01) (-0.16) (-0.41) (-0.02) (0.58)
[0; +60] 0.017 0.036 0.139 -0.023 0.016 0.022 0.133 -0.021 -0.004 -0.008 -0.078 0.055 -0.001 -0.005 -0.063 0.066
(1.32) (1.08) (8.65) (-1.62) (1.51) (1.05) (8.02) (-1.32) (-0.12) (-0.37) (-0.99) (0.37) (-0.04) (-0.31) (-0.79) (0.41)
[0; +120] 0.023 0.044 0.176 -0.014 0.020 0.043 0.165 -0.011 -0.005 -0.009 -0.062 0.066 -0.008 -0.006 -0.068 0.062
(2.53) (3.72) (13.70) (-1.64) (2.29) (3.48) (12.59) (-1.40) (-0.40) (-0.56) (-0.80) (1.64) (-0.69) (-0.06) (-0.63) (1.17)
[0; +250] 0.019 0.030 0.130 -0.017 0.021 0.029 0.127 -0.016 -0.034 -0.005 -0.054 0.046 -0.024 -0.001 -0.041 0.045
(1.96) (2.98) (9.95) (1.39) (2.01) (2.82) (9.53) (-1.07) (-0.22) (-0.34) (-0.11) (4.72) (-0.19) (-0.17) (-0.71) (4.63)
Obs. 261 220 185 173 261 220 185 173 180 154 103 171 180 154 103 171
Note: This table presents the results for the estimation of model (2) on a subsample of highly scored firms. Highly-score firms are defined as firms in the highest quintile in terms of
extra-financial score. Mean values for b
Ei
are shown, with tstatistics in parentheses. Other coefficients in model (2) are estimated but not shown to save valuable space. Model (2)
is estimated on a subsample of score upgrades and a subsample of score downgrades and, in each case, for the four types of score dimensions [aggregate (agg.), environment
(envir.), social, and governance (gov.)]. Five analysis periods are considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0; +250]), for which the estimation periods are,
respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and [-500; +250]. The overall sample includes 266 Canadian firms from January 2007 to December 2012.
Numbers in bold indicate significance at the 10 per cent level.
Sodjahin et al
360 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
Table 5: Beta changes around changes in extra-financial performance for firms with low extra-financial scores
Analysis period Score upgrades Score downgrades
Unconditional model Conditional model Unconditional model Conditional model
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
[-120; 0] 0,002 0,000 0,001 -0,002 0,003 0,002 0,002 -0,001 0,002 0,005 0,001 0,003 0,001 0,002 0,002 0,002
(0.14) (0.00) (0.02) (-0.22) (0.23) (0.07) (0.09) (-0.04) (0.13) (1.07) (0.05) (0.12) (0.06) (0.83) (0.01) (0.03)
[-60; 0] -0,013 -0,034 -0,004 -0,003 -0,020 -0,029 -0,002 -0,001 0,011 0,035 0,011 0,018 0,010 0,029 0,008 0,012
(-0.15) (-1.12) (-0.01) (-0.01) (-0.99) (-1.11) (-0.17) (-0.01) (1.01) (1.03) (0.78) (1.69) (0.33) (1.03) (0.30) (1.61)
[0; +60] -0,009 -0,030 -0,018 -0,019 -0,009 -0,023 -0,016 -0,013 0,019 0,041 0,020 0,022 0,015 0,039 0,014 0,020
(-2.07) (-1.18) (-2.11) (-2.77) (-1.95) (-1.51) (-1.91) (-2.60) (2.34) (4.28) (1.94) (3.04) (2.62) (3.02) (1.81) (2.77)
[0; +120] -0,008 -0,027 -0,020 -0,016 -0,006 -0,021 -0,017 -0,014 0,021 0,065 0,023 0,026 0,013 0,054 0,019 0,023
(-2.76) (-2.41) (-2.74) (-4.98) (-2.29) (-2.09) (-2.41) (-4.52) (2.99) (7.32) (2.74) (4.89) (2.03) (7.11) (2.78) (3.90)
[0; +250] -0,006 -0,021 -0,019 -0,014 -0,005 -0,019 -0,015 -0,011 0,015 0,045 0,032 0,021 0,011 0,039 0,027 0,019
(-2.23) (-1.91) (-2.41) (-2.68) (-1.97) (-1.85) (-2.27) (-2.61) (3.14) (6.03) (6.18) (2.28) (2.65) (5.11) (5.82) (2.04)
Obs. 263 222 187 175 263 222 187 175 181 156 105 169 181 156 105 169
Note: This table presents the results for the estimation of model (2) on a subsample of firms with low extra-financial scores. Lowly scored firms are defined as firms in the lowest
quintile in terms of extra-financial score. Mean values for b
Ei
are shown, with t-statistics in parentheses. Other coefficients in model (2) are estimated but not shown to save
valuable space. Model (2) is estimated on a subsample of score upgrades and a subsample of score downgrades and, in each case, for the four types of score dimensions
[aggregate (agg.), environment (envir.), social, and governance (gov.)]. Five analysis periods are considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0; +250]), for which the
estimation periods are, respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and [-500; +250]. The overall sample includes 266 Canadian firms from January
2007 to December 2012. Numbers in bold indicate significance at the 10 per cent level.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 361
Robustness tests
Three robustness tests are conducted to
ensure the validity of our results in different
circumstances. The first test examines the
impact of extra-financial performance
changes on risk on a subsample of firms that
are not contaminated by any firm-specific
event. The second test examines the
potentially different impact of extra-finan-
cial performance score upgrades and
downgrades on risk according to the nature
of the firm’s business. The third test
investigates the potentially different impact
of extra-financial performance score
upgrades and downgrades on risk depend-
ing on market conditions.
Impact of extra-financial performance
score changes on risk
for a subsample of uncontaminated
firms
Long-term studies are sensitive to the
presence of confounding effects because
other value-relevant events, which are not
necessarily related to CSR, can occur
throughout a given year. To ensure that
our results are not attributable to other
firm-specific events, we re-estimate model
(2) on a subsample of firms for which no
important event takes place during the
analysis period. To do so, we use Bloom-
berg data to estimate differences between
quarterly earnings announcements for firms
in the TSX and market expectations prior
to these announcements. Following the
literature (e.g., Mendenhall, 2004), we
normalize these differences by their stan-
dard deviation over the period under study
and select events with the largest absolute
valueasproxiesforpotentiallycontami-
nating events. This approach removes
approximately 15 per cent of our observa-
tions. Results for the estimation of model
(2) on the remaining uncontaminated
observations are available in Table 6and
are very similar, even stronger, than results
obtained previously.
Impact of extra-financial performance
score upgrades and downgrades
on risk conditional on the firm’s
business sector
In this section, we consider the possibility that
the relationship between extra-financial per-
formance and systematic risk is heterogeneous
across industries.
9
Specifically, there is
empirical evidence suggesting that firm risk
varies by industry (Fama and French, 1997;
Gebhardt et al, 2001). In addition, some
studies show that extra-financial performance
varies significantly across sectors (e.g., Carroll,
1979; Griffin and Mahon, 1997;Brammer
et al, 2006; Godfrey et al, 2008). To verify if
the impact of extra-financial performance
depends on the business sector of the firm, we
examine the impacts of score upgrades and
downgrades on systematic risk for each of the
ten industrial sectors in our sample, based on
the Global Industry Classification Standard
(GICS) of each firm. Specifically, we estimate
model (2) separately for each of the ten
industries. Table 7reports the estimates for bEi
for each industry for the [0; +250] period.
Results for are available in Table 7. We first
note that score downgrades are strongly rela-
ted to increases in systematic risk (beta) for all
industries. Secondly, there is no clear evidence
that the impact of extra-financial performance
differs according to the firm’s business sector.
For instance, for the aggregate score, condi-
tional model coefficients range from 0.011 to
0.027. We nevertheless note that the effect of
environmental-dimension score downgrades
on beta is the highest for the following
industries: (i) Energy (conditional model
coefficient of 0.059) which includes the oil
and gas sectors; (ii) Materials (0.057) which
include the metals and mining sector, and (iii)
Industrials (0.055) which include the airline,
marine, road, and rail sector. The fact that
these industries are particularly exposed to
environmental issues may explain the rela-
tively stronger effect for the environmental
dimension of extra-financial performance. In a
similar matter, we observe that the effect of
Sodjahin et al
362 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
Table 6: Changes in firm beta around changes in extra-financial performance scores on a subsample of uncontaminated firms
Analysis period Score upgrades Score downgrades
Unconditional model Conditional model Unconditional model Conditional model
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
[-120; 0] -0.007 -0.002 -0.002 -0.009 -0.004 -0.001 -0.003 -0.002 0.017 0.036 0.008 0.002 0.021 0.025 0.003 0.001
(-0.83) (-0.47) (-0.11) (-0.25) (-0.21) (-0.83) (-0.26) (-0.03) (0.73) (1.01) (0.14) (0.01) (0.99) (1.19) (1.13) (0.04)
[-60; 0] -0.001 -0.008 -0.003 -0.009 -0.008 -0.003 -0.002 -0.011 0.005 0.047 0.004 0.004 0.011 0.042 0.093 0.003
(-0.99) (-0.89) (-0.07) (-0.14) (-0.73) (-0.93) (-0.28) (-0.67) (0.73) (1.61) (0.22) (1.19) (1.29) (1.59) (1.20) (1.04)
[0; +60] -0.006 -0.002 -0.023 -0.011 -0.007 -0.015 -0.019 -0.012 0.014 0.041 0.008 0.021 0.013 0.037 0.005 0.017
(-1.37) (-1.22) (-1.49) (-1.99) (-1.31) (-1.26) (-1.36) (-2.23) (2.48) (4.07) (0.79) (3.14) (3.02) (2.73) (0.24) (2.97)
[0; +120] -0.044 -0.030 -0.022 -0.018 -0.005 -0.029 -0.021 -0.015 0.019 0.065 0.022 0.026 0.013 0.053 0.018 0.023
(-1.32) (-1.60) (-1.61) (-5.02) (-1.43) (-1.21) (-1.42) (-4.87) (2.76) (8.66) (2.90) (4.69) (2.33) (7.73) (2.56) (4.03)
[0; +250] -0.004 -0.027 -0.031 -0.014 -0.004 -0.027 -0.033 -0.011 0.015 0.044 0.029 0.019 0.011 0.039 0.023 0.018
(-1.53) (-1.29) (-1.39) (-2.93) (-1.36) (-0.89) (-1.36) (-2.81) (2.81) (6.03) (6.13) (2.63) (2.21) (4.93) (5.41) (2.16)
Obs. 1105 970 866 783 1105 970 866 783 769 688 483 807 769 688 483 807
Note: This table presents the results for the estimation of model (2) on a subsample of firms for which there are no surprises in quarterly earnings announcements during the
analysis period. Surprises are defined as differences between quarterly earnings announcements for firms in the TSX and market expectations prior to these announcements as
estimated by Bloomberg. Mean values for b
Ei
are shown, with tstatistics in parentheses. Other coefficients in model (2) are estimated but not shown to save valuable space.
Model (2) is estimated on a subsample of score upgrades and a subsample of score downgrades and, in each case, for the four types of score dimensions [aggregate (agg.),
environment (envir.), social, and governance (gov.)]. Five analysis periods are considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0; +250]), for which the estimation periods
are, respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and [-500; +250]. The overall sample includes 266 Canadian firms from January 2007 to December
2012. Numbers in bold indicate significance at the 10 per cent level.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 363
Table 7: Changes in firm beta around changes in extra-financial performance scores conditional on the business sector
Sector Score upgrades Score downgrades
Unconditional model Conditional model Unconditional model Conditional model
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
Consumer
discretionary
-0.006 -0.002 -0.028 -0.005 -0.004 -0.001 -0.025 -0.003 0.016 0.040 0.037 0.011 0.015 0.037 0.035 0.010
(-1.53) (-0.11) (-2.09) (-1.26) (-1.39) (-0.05) (-1.98) (-0.97) (3.21) (3.91) (6.22) (3.02) (2.98) (3.42 (6.03) (2.96)
Consumer staple -0.005 -0.001 -0.029 -0.003 -0.004 -0.001 -0.028 -0.002 0.017 0.041 0.039 0.010 0.016 0.039 0.038 0.009
(-1.46) (-0.07) (-2.33) (-1.02) (-1.30) (-0.03) (-2.03) (-0.61) (3.50) (3.96) (6.53) (3.29) (3.39) (3.59) (6.23) (2.95)
Energy -0.008 -0.015 -0.005 -0.012 -0.006 -0.012 -0.004 -0.010 0.029 0.062 0.030 0.017 0.027 0.059 0.029 0.015
(-1.66) (-1.79) (-1.13) (-3..19) (-1.49) (-1.62) (-1.02) (-3.01) (5.22) (6.93) (5.69) (5.81) (5.01) (6.23) (5.32) (5.17)
Financials -0.007 -0.003 -0.019 -0.017 -0.006 -0.002 -0.017 -0.015 0.023 0.037 0.027 0.020 0.021 0.036 0.024 0.018
(-1.68) (-0.11) (-1.68) (-4.63) (-1.37) (-0.09) (-1.53) (-4.12) (3.94) (2.92) (3.91) (6.83) (3.63) (2.29) (3.72) (6.14)
Health care -0.001 -0.010 -0.003 -0.005 -0.001 -0.008 -0.002 -0.003 0.013 0.043 0.025 0.010 0.011 0.041 0.023 0.009
(-0.13) (-1.69) (-1.04) (-1.14) (-0.09) (-1.62) (-0.96) (-1.01) (2.19) (5.97) (3.71) (3.15) (2.09) (3.78) (3.53) (2.93)
Industrials -0.005 -0.010 -0.008 -0.011 -0.004 -0.008 -0.006 -0.009 0.019 0.057 0.026 0.015 0.018 0.055 0.025 0.014
(-1.51) (-1.77) (-1.63) (-2.10) (-1.41) (-1.60) (-1.51) (-1.99) (3.44) (6.73) (3.82) (4.73) (3.39) (5.61) (3.35) (4.12)
Information
technology
-0.003 -0.004 -0.006 -0.002 -0.002 -0.003 -0.004 -0.001 0.013 0.038 0.023 0.011 0.012 0.036 0.021 0.009
(-1.03) (-1.09) (-1.11) (-0.66) (-0.23) (-0.95) (-1.09) (-0.31) (2.47) (3.35) (3.22) (3.07) (2.16) (3.09) (3.10) (2.89)
Materials -0.006 -0.014 -0.009 -0.012 -0.005 -0.011 -0.006 -0.011 0.022 0.058 0.031 0.016 0.020 0.057 0.030 0.015
(-1.61) (-1.83) (-1.41) (-2.67) (-1.29) (-1.61) (-1.53) (-2.13) (3.74) (6.83) (5.82) (4.96) (3.59) (5.69) (5.65) (4.20)
Telecommunication
services
-0.003 -0.003 -0.003 -0.005 -0.002 -0.002 -0.002 -0.003 0.015 0.036 0.027 0.009 0.012 0.036 0.025 0.008
(-1.00) (-0.89) (-1.03) (-1.09) (-0.29) (-0.10) (-0.93) (-0.78) (2.52) (3.14) (3.89) (2.92) (2.41) (2.97) (3.74) (2.63)
Utilities -0.003 -0.005 -0.005 -0.003 -0.003 -0.003 -0.004 -0.001 0.014 0.035 0.026 0.009 0.011 0.034 0.024 0.007
(-1.08) (-1.13) (-1.10) (-0.40) (-1.06) (-1.11) (-1.00) (-0.13) (2.33) (2.97) (3.81) (2.81) (2.06) (2.37) (3.69) (2.63)
Note: This table presents the results for the estimation of model (2) conditional on the business sector of the firm. Mean values for b
Ei
are shown, with tstatistics in parentheses.
Other coefficients in model (2) are estimated but not shown to save valuable space. Model (2) is estimated on a subsample of score upgrades and a subsample of score
downgrades and, in each case, for the four types of score dimensions [aggregate (agg.), environment (envir.), social, and governance (gov.)]. Five analysis periods are
considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0; +250]), for which the estimation periods are, respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and
[-500; +250]. The overall sample includes 266 Canadian firms from January 2007 to December 2012. Numbers in bold indicate significance at the 10 per cent level.
Sodjahin et al
364 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
Table 8: Changes in firm beta around changes in extra-financial performance scores conditional on market conditions
Analysis period Score upgrades (Conditional model) Score downgrades (Conditional model)
2007–2009 2010–2012 2007–2009 2010–2012
Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov. Agg. Envir. Social Gov.
[-120; 0] -0.004 -0.001 -0.001 -0.005 -0.003 0.003 0.000 -0.004 0.009 0.027 0.004 0.002 0.010 0.023 0.001 0.001
(-0.98) (-0.11) (-0.13) (-0.03) (-0.09) (0.48) (-0.01) (-0.14) (0.43) (1.11) (1.08) (0.01) (0.99) (1.00) (0.52) (0.89)
[-60; 0] -0.006 -0.003 -0.004 -0.012 -0.004 -0.004 -0.001 -0.009 0.008 0.040 0.008 0.018 0.008 0.039 0.002 0.007
(-1.23) (-0.74) (-0.77) (-1.50) (-0.29) (-0.14) (-0.22) (-1.22) (1.45) (1.63) (1.13) (2.66) (1.59) (1.35) (0.98) (1.04)
[0; +60] -0.008 -0.008 -0.013 -0.013 -0.006 -0.010 -0.011 -0.012 0.011 0.037 0.0015 0.025 0.010 0.035 0.008 0.012
(-1.47) (-1.51) (-1.24) (-3.68) (-1.33) (-139) (-1.30) (-1.90) (2.97) (3.07) (1.53) (4.55) (2.03) (2.44) (0.83) (2.63)
[0; +120] -0.006 -0.005 -0.024 -0.016 -0.005 -0.004 -0.022 -0.014 0.012 0.050 0.018 0.029 0.011 0.049 0.016 0.011
(-1.67) (-1.27) (-1.41) (-4.11) (-1.27) (-1.03) (-1.21) (-2.19) (2.83) (7.74) (2.49) (4.98) (2.58) (7.17) (2.06) (2.37)
[0; +250] -0.008 -0.004 -0.025 -0.012 -0.006 -0.003 -0.020 -0.010 0.009 0.037 0.025 0.024 0.008 0.035 0.022 0.010
(-1.63) (-1.52) (-1.63) (-4.86) (-1.01) (-1.23) (-1.06) (-2.97) (2.89) (4.89) (5.49) (3.28) (2.43) (4.08) (4.96) (2.01)
Obs. 583 282 291 401 729 822 639 466 340 156 157 255 561 616 364 605
Note: This table presents the results for the estimation of model (2) conditional on market conditions. The sample is split into two subperiods: 2007–2009 and 2010–2012 and
model (2) is estimated for each subperiod. Mean values for b
Ei
are shown, with tstatistics in parentheses. Other coefficients in model (2) are estimated but not shown to save
valuable space. Model (2) is estimated on a subsample of score upgrades and a subsample of score downgrades and, in each case, for the four types of score dimensions
[aggregate (agg.), environment (envir.), social, and governance (gov.)]. Five analysis periods are considered ([-120; 0], [-60; 0], [0; +60], [0; +120], and [0; +250]), for which the
estimation periods are, respectively: [-500; -120], [-500; -60], [-500; +60], [-500; +120], and [-500; +250]. The overall sample includes 266 Canadian firms from January
2007 to December 2012. Numbers in bold indicate significance at the 10 per cent level.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 365
social-dimension score downgrades seems
slightly stronger for industries related to retail
and light manufacturing such as consumer
discretionary (conditional model coefficient of
0.035) and consumer staple (0.038). One
potential explanation is that employee rela-
tions and human rights issues are very
important for these labor-intensive industries.
In terms of score upgrades, there is weak
evidence, mostly for the unconditional model,
that score upgrades are associated with beta
changes. However, some industries do appear
to be more affected than others by extra-fi-
nancial performance changes. For instance,
only the two retail industries, namely con-
sumer discretionary and consumer staple,
experience a significant reduction in system-
atic risk following social-dimension score
upgrades (for the conditional model). Simi-
larly, only four industries are affected by
governance-dimension extra-financial perfor-
mance changes: financials, energy, industrials
and materials. In the four cases, systemic risk is
lower following score upgrades.
Impact of extra-financial performance
score upgrades and downgrades
on risk conditional on the economic
context
With our third robustness test, we wish to
investigate whether market conditions can
mitigate or amplify theimpact of extra-financial
performance score upgrades and downgrades
on risk. Recent studies (e.g., Albuquerque et al,
2014; Oikonomou, 2012; Chen et al, 2010)
show that extra-financial performance and
macroeconomic factors can co-vary.
10
Further,
our earlier results suggest that systematic risk is a
function of the economic context. It is there-
fore possible that extra-financial performance
changes have a different impact on risk
depending on market conditions. This analysis
is particularly interesting since the period cov-
ered by our study,from 2007 to 2012, is marked
by the subprime financial crisis.
To test whether the impact of extra-financial
performance score changes on risk depends on
market conditions, we split our sample into two
subperiods: (i) the 2007–2009 subperiod,
which corresponds to the financial crisis, and (ii)
the 2010–2012 subperiod, which corresponds
to the post-crisis relatively stable period.
11
We
estimate model (2) for each subperiod. Results
in Table 8show that the estimates of coefficient
bEi are remarkably stable over time, suggesting
that there is very little significant evidence that
the impact of extra-financial performance
varies according to market conditions. We
nevertheless note that the effect of governance
score changes, especially score downgrades,
are greater for the 2007–2009 subperiod than
for the 2010–2012 subperiod. This suggests
that, during times of economic uncertainty,
governance-dimension-downgraded firms
experience higher systematic risk, but gover-
nance-dimension upgraded firms are not
rewarded, at least in terms of systematic risk,
by the market. Corporate governance there-
fore appears to be a greater concern for
investors during a financial crisis.
CONCLUSION
This study investigates the informational
content of extra-financial performance scores
by examining the relationship between extra-
financial score changes (upgrades and
downgrades) and systematic risk (beta)
variations. Our work is based on changes in
Sustainalytics’ extra-financial performance
scores for a sample of 266 Canadian
corporations between 2007 and 2012 and
provides important empirical findings. First,
we find no significant evidence that changes
in extra-financial performance scores lag beta
variations. Rather, we find that systematic risk
increases follow extra-financial score
downgrades. Extra-financial scores therefore
do not appear to be established a posteriori on
the basis of stock market information but
rather appear to be leading indicators of
systematic risk variations.
Our results also show that score upgrades for
firms with already-high scores predict higher
Sodjahin et al
366 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
systematic risk, while score upgrades predict
lower systematic risk for lower-scored firms.
This result suggests that, for firms with already-
high extra-financial scores, improvements can
be counterproductive and lead to an increase in
systematic risk, possibly because of costs that
investors feel are too high and inopportune.
However, systematic risk decreases when firms
with low extra-financial scores make an effort
to improve their social image. Finally, while
score downgrades are not related to beta
variations for firms with high extra-financial
scores, they are related to higher systematic risk
for firms with low extra-financial scores.
Overall,thisstudyprovidesevidenceofthe
usefulness of extra-financial agencies’ scorings
for managers in the development of their risk
management strategies. Investors may limit
their exposure to systematic risk by following
changes in firms’ ESG ratings. They can also
build investment strategies based on the changes
in corporate extra-financial performance score.
Our results also imply that it may be appropriate
to include an irresponsibility risk factor in a
general asset pricing model. We leave this
question and tests to future research.
One of limitations of this study is that
Sustainalytics from which the scores are
collected, may be one of the largest providers
of corporate social responsibility intelligence
in Canada, but is nonetheless only one
agency among many.
ACKNOWLEDGEMENTS
The authors thank the Caisse de De
´po
ˆtet
Placement du Que
´bec (CDPQ), the GReFA, the
Desjardins Chair in Responsible Finance and
the CIBC Research Chair in Financial Integrity
for their financial support. Special thanks also go
out to Stephen Kibsey, Joanne Pichette and
Ginette Depelteau of the CDPQ for their
invaluable comments and contributions.
NOTES
1. Formerly KLD Research & Analytics, Inc.
2. Ethical Investment Research and Information Service
(http://www.eiris.org/).
3. ASSET4 provides investment research information on
economic, environmental, social, and governance (ESG)
aspects of corporate performance.
4. Sustainalytics was formed from the merger between the Dutch
firm ‘‘Sustainalytics’’ and the Canadian firm ‘‘Jantzi Research
Inc’’ in August 2009 (http://www.sustainalytics.com/).
5. We use the term ‘‘extra-financial’’ performance to include
all types of non financial performance that are deemed
‘‘responsible’’. These include the social, governance and
environmental performances of firms.
6. Christopherson et al (1998) show that a conditional
approach, using time-varying measures of risk (beta) and
abnormal performance (alpha), is better able to predict
future performance than conditional beta models that
consider only time-varying measures of risk (Ferson and
Schadt, 1996).
7. We follow a two-step methodology to select the variables
that reflect the Canadian economy. In the first step, based
on the literature (e.g., Ferson and Qian, 2004), which
mainly focuses on the U.S. economy, we identify seven
macrofinancial information variables intended to reflect
the state of the economy: (i) short-term interest rates, (ii)
interest rate volatility, (iii) term structure of interest rates,
(iv) term structure concavity, (v) stock market perfor-
mance, (vi) stock market (implied) volatility, and (vii)
credit spread. In the second step, we use stepwise
regression techniques to identify those information vari-
ables that have a predictive power on financial perfor-
mance and beta. In the end, we identify five information
variables, Z
n,t
, that are used throughout: (i) short-term
interest rates, (ii) term structure of interest rates, (iii) stock
market performance, (iv) stock market (implied) volatility,
and (v) credit spread.
8. The other parameters in model (2) are estimated and
included in the regressions, but are not reported.
9. The authors would like to thank an anonymous referee for
this suggestion.
10. Given the cost of fulfilling ESG criteria during difficult
economic times, firms may reduce their investments in
CSR initiatives (Albuquerque et al, 2014) and thereby
appear to be less observant of CSR criteria which may
negatively affect their extra-financial performance score.
On the other hand, some authors (see, for e.g.,
Oikonomou et al, 2012) believe that a higher extra-
financial performance score should be expected during
periods of economic uncertainty when firms may be more
inclined to implement good practices, including socially
responsible ones, to reduce risk (Chen et al, 2010).
11. As in Aloui et al (2011), we choose the subperiod of 2007–2009
as representative of the financial subprime crisis. As high-
lighted by Longstaff (2010), the subprime crisis actually
began in early 2007. Moreover, this period includes the
period of contraction from December 2007 to June 2009,
as identified by NBER.
REFERENCES
Aloui, R., Aı
¨ssa, M.S.B., Nguyen, D.K. (2011) Global
financial crisis, extreme interdependences, and contagion
effects: The role of economic structure? Journal of Banking &
Finance 35: 130–141.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 367
Albuquerque, R.A., Durnev, A. and Koskinen, K. (2014)
Corporate social responsibility and firm risk: Theory and
empirical evidence. SSRN working paper.
Avouyi-Dovi, S.D. and Neto, D. (2004) Equity market
interdependence: the relationship between European and
US stock markets. Banque de France, Financial Stability
Review 4: 108–126.
Barnea, A. and Rubin, A. (2010) Corporate social responsi-
bility as a conflict between shareholders. Journal of Business
Ethics 97: 71–86.
Baron, D. and Diermeier, D. (2007) Strategic activism and
nonmarket strategy. Journal of Economics and Management
Strategy 16: 599–634.
Bouslah, K., Kryzanowski, L. and M’Zali, B. (2013) The
impact of the dimensions of social performance on firm
risk. Journal of Banking & Finance 37: 1258–1273.
Boutin-Dufresne, F. and Savaria, P. (2004) Corporate social
responsibility and financial risk. Journal of Investing 13:
57–66.
Bradley Jr., R.L. (2009) Corporate social responsibility and
energy. In Horowitz, I.L. (Ed.) Culture and Civilization.
Transaction Publishers, New Jersey, 181-197.
Brammer, S., Brooks, C., Pavelin, S. (2006) Corporate social
performance and stock returns: UK evidence from
disaggregate measures. Financial Management 35: 97–116.
Carroll, A.B. (1979) Three-dimensional conceptual model of
corporate performance. The Academy of Management Review
4: 497–505.
Cespa, G. and Cestone, G. (2007) Corporate social responsi-
bility and managerial entrenchment. Journal of Economic
Management and Strategy 16: 741–771.
Chen, C., Guo, W. and Tay, N. (2010) Are member firms of
corporate groups less risky? Financial Management 39:
59–82.
Christopherson, J., Ferson, W. Glassman, D. (1998). Condi-
tioning manager alphas on economic information: another
look at the persistence of performance. Review of Financial
Studies 11: 111–142.
Champagne C., Chre
´tien S. and Coggins F. (2015) Effects of
pension fund freezing on firm performance and risk.
Canadian Journal of Administrative Sciences.
De Bondt W.-M. and Thaler R. (1985) Does the stock market
overreact? Journal of Finance 40: 793–805.
Fama, E. and French, K. (1993) Common risk factors in the
returns on stocks and bonds. Journal of Financial Economics
33: 3–56.
Fama, E.F., French, K.R. (1997) Industry cost of equity.
Journal of Financial Economics 43: 153–193.
Ferson, W.E. and Qian, M. (2004) Conditional Performance
Evaluation Revisited, Research Foundation Monograph of
the CFA Institute, ISBN 0-943205-69-7.
Ferson, W. and Schadt, R. (1996) Measuring fund strategy and
performance in changing economic conditions. Journal of
Finance 51: 425–462.
Friedman, M., (1970) The social responsibility of business is to
increase its profits. New York Times, September, 13,
122–126.
Gebhardt, W., Lee, C., Swaminathan, B. (2001) Toward an
implied cost of capital. Journal of Accounting Research 39:
135–176.
Godfrey, P.C., Hatch, N.W., Hansen, J.M. (2008) Toward a
general theory of CSRs: The roles of beneficence,
profitability, insurance, and industry heterogeneity. Business
and Society, 49: 316–344.
Griffin, J.J., Mahon, J.F. (1997) The corporate social perfor-
mance and corporate financial performance debate:
Twenty-five years of incomparable research, Business and
Society 36: 5–31.
Handelman, J.M. and Arnold S.J. (1999) The role of marketing
actions with a social dimension: Appeals to the institutional
environment. Journal of Marketing 63: 33–48.
Hasseldine, J., Salama, A. and Toms, J.S. (2005) Quantity
versus quality: the impact of environmental disclosures on
the reputations of UK Plcs. British Accounting Review 2:
231–248.
Hemingway, C.A. and Maclagan, P.W. (2004) Managers’
personal values as drivers of corporate social responsibility.
Journal of Business Ethics 50: 33–44.
Holthausen R.W. and Leftwic R.W. (1986) The effect of
bond rating changes on common stock prices. Journal of
Financial Economics 17(1): 57–89.
Hong, H.G., Kubik, J.K., Scheinkman, J.A. (2012) Financial
constraints on corporate goodness. Unpublished working
paper. Princeton University and Syracuse University.
Jensen, M., Meckling, W. (1976) Theory of the firm:
Managerial behavior, agency cost and capital structure.
Journal of Financial Economics 3: 305–360.
Jo, H. and Na, H. (2012) Does CSR reduce firm risk?
Evidence from controversial industry sectors. Journal of
Business Ethics 110: 441–457.
Jones, T.M. (1995) Instrumental stakeholder theory: A
synthesis of ethics and economics’. Academy of Management
Review 2: 404–437.
Jorion, P., Zhang, G. (2007) Information effects of bond
rating changes: The role of the rating prior to the
announcement. The Journal of Fixed Income 16(4): 45–59.
Kim, Y., Li, H. and Li, S. (2014) Corporate social respon-
sibility and stock price crash risk. Journal of Banking &
Finance 43: 1–13.
Kru
¨ger, P. (2015) Corporate goodness and shareholder wealth.
Journal of Financial Economics 115: 304–329.
Lee, D.D. and Faff, R.W. (2009) Corporate sustainability
performance and idiosyncratic risk: A global perspective.
Financial Review 2: 213–237.
Longstaff, F.A. (2010) The subprime credit crisis and conta-
gion in financial markets. Journal of Financial Economics 97:
436–450.
Luo, X. and Bhattacharya, C.B. (2009) The debate over doing
good: Corporate social performance, strategic marketing
levers, and firm-idiosyncratic risk. Journal of Marketing 73:
198–213.
Lyon, T.P. and Maxwell, J.W. (2011) Greenwash: Corporate
environmental disclosure under threat of audit. Journal of
Economics & Management Strategy 20: 3–41.
Mattingly, J.E. and Berman, S.L. (2006) Measurement of
corporate social action: Discovering taxonomy in the
Kinder Lydenburg domini ratings data. Business & Society
45: 20–46.
McGuire, J.B., Schneeweis, T. and Sundgren, A. (1988)
Corporate social responsibility and firm financial perfor-
mance. Academy of Management Journal 3: 854–872.
McWilliams, A. and Siegel, D. (2001) Corporate social
responsibility: A theory of the firm perspective. Academy of
Management Review 26: 117–127.
Sodjahin et al
368 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370
McWilliams, A., Siegel, D., Wright, P. (2006) Guest editors’
introduction corporate social responsibility: Strategic
implications. Journal of Management Studies 43: 1–18.
Mendenhall, R.R. (2004) Arbitrage risk and post-earnings-
announcement drift. The Journal of Business 77: 875–894.
Mishra, S. and Modi, S.B. (2012) Positive and negative
corporate social responsibility, financial leverage, and
idiosyncratic risk. Journal of Business Ethics 2: 431–448.
Newey, W. and West, K.D. (1987) A simple, positive semi-
definite heteroskedasticity and autocorrelation-consistent
covariance matrix. Econometrica 55: 703–708.
Oikonomou, I., Brooks, C. and Pavelin, S. (2012) The impact
of corporate social performance on financial risk and utility:
A longitudinal analysis. Financial Management 41: 483–515.
Orlitzky, M. and Benjamin, J.D. (2001) Corporate social
performance and firm risk: A meta-analytic review. Business
and Society 4: 369–396.
Porter, M.E. and Kramer, M.R. (2006) Strategy and society:
the link between competitive advantage and corporate
social responsibility. Harvard Business Review 12: 78–92.
Preston, L.E. and O’Bannon, D.P. (1997) The corporate
social-financial performance relationship. A typology and
analysis. Business and Society 4: 109–125.
Roberts, R.W. (1992) Determinants of corporate social
responsibility disclosure: An application of stakeholder
theory. Accounting, Organizations and Society 6:
595–612.
Salama, A., Anderson, K.P. and Toms, S. (2011) Does
community and environmental responsibility affect firm
risk? Evidence from UK panel data 1994–2006. Business
Ethics 20: 192–204.
Sharfman, M.P. and Fernando, C.S. (2008) Environmental
risk management and the cost of capital. Strategic Manage-
ment Journal 6: 569–592.
Smith, C.N. (2003) Corporate social responsibility: Whether
or how? California Management Review 45: 52–76.
Stern, N. (2006) Stern Review:The Economics of Climate Change.
Cambridge: New Economics Foundation.
Waddock, S.A. and Graves, S. B. (1997) The corporate social
performance-financial performance link. Strategic Manage-
ment Journal 18: 303–319.
Weinstein, M.I. (1977) The effect of a rating change
announcement on bond price. Journal of Financial Economics
5(3): 329–350.
Wijnberg, N.M. (2000) Normative stakeholder theory and
aristotle: the link between ethics and politics. Journal of
Business Ethics 4: 329–342.
APPENDIX
See Table 9.
Leading or lagging indicators of risk?
2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370 369
Table 9: Sustainalytics’s ESG metrics
Environment Social Governance
Operations Formal environmental policy Employees Policy on freedom of association Business
ethics
Policy on bribery and
corruption
Environmental management system Formal policy on the elimination of
discrimination
Signatory to UN global
compact
External certification of EMS Programs to increase workforce diversity Tax transparency
Environmental fines and non-
monetary sanctions
Percentage of employees covered by
collective bargaining agreements
Business ethics related
controversies or incidents
Participation in carbon disclosure
project (CDP)
Employee turnover rate Reporting,
transparency
and oversight
CSR reporting quality
Scope of corporate reporting on
GHG emissions
Top employer recognition External verification of CSR
reporting
Programs and targets to reduce
direct GHG emissions
Employee related controversies or
incidents
Disclosure of directors’
remuneration
Programs and targets to increase
renewable energy use
Supply chain Scope of social supply chain standards Oversight of ESG issues
Carbon intensity Contractors & supply chain related
controversies or incidents
Executive compensation tied
to ESG performance
Carbon intensity trend Tenants Public position statement on responsible
marketing
Board diversity
per cent Primary energy use from
renewables
Customer related controversies or
incidents
Separation of board chair and
CEO roles
Operations related controversies or
incidents
Community and
philanthropy
Activities in sensitive countries Board independence
Supply chain Formal policy or programme on
green procurement
Policy on human rights Audit committee
independence
contractors & supply chain related
controversies or incidents
Society & community related controversies
or incidents
Governance related
controversies or incidents
Products & services
and sustainability
Sustainability related products &
services
Guidelines for philanthropic activities and
primary areas of support
Public policy Transparency on payments to
host governments
Products & services related
controversies or incidents
Corporate foundation Public policy related
controversies or incidents
Sodjahin et al
370 2017 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 18, 5, 347–370