ArticlePDF Available

ESG and financial performance: Aggregated evidence from more than 2000 empirical studies


Abstract and Figures

The search for a relation between environmental, social, and governance (ESG) criteria and corporate financial performance (CFP) can be traced back to the beginning of the 1970s. Scholars and investors have published more than 2000 empirical studies and several review studies on this relation since then. The largest previous review study analyzes just a fraction of existing primary studies, making findings difficult to generalize. Thus, knowledge on the financial effects of ESG criteria remains fragmented. To overcome this shortcoming, this study extracts all provided primary and secondary data of previous academic review studies. Through doing this, the study combines the findings of about 2200 individual studies. Hence, this study is by far the most exhaustive overview of academic research on this topic and allows for generalizable statements. The results show that the business case for ESG investing is empirically very well founded. Roughly 90% of studies find a nonnegative ESG–CFP relation. More importantly, the large majority of studies reports positive findings. We highlight that the positive ESG impact on CFP appears stable over time. Promising results are obtained when differentiating for portfolio and nonportfolio studies, regions, and young asset classes for ESG investing such as emerging markets, corporate bonds, and green real estate.
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
Download by: [] Date: 30 January 2016, At: 20:09
Journal of Sustainable Finance & Investment
ISSN: 2043-0795 (Print) 2043-0809 (Online) Journal homepage:
ESG and financial performance: aggregated
evidence from more than 2000 empirical studies
Gunnar Friede, Timo Busch & Alexander Bassen
To cite this article: Gunnar Friede, Timo Busch & Alexander Bassen (2015) ESG and financial
performance: aggregated evidence from more than 2000 empirical studies, Journal of
Sustainable Finance & Investment, 5:4, 210-233, DOI: 10.1080/20430795.2015.1118917
To link to this article:
© 2015 The Author(s). Published by Taylor &
Published online: 15 Dec 2015.
Submit your article to this journal
Article views: 2511
View related articles
View Crossmark data
ESG and nancial performance: aggregated evidence from more than 2000
empirical studies
Gunnar Friede
, Timo Busch
*and Alexander Bassen
Deutsche Asset & Wealth Management Investment, Frankfurt am Main, Germany;
School of Business,
Economics and Social Science, University of Hamburg, Hamburg, Germany
(Received 22 October 2015; accepted 9 November 2015)
The search for a relation between environmental, social, and governance (ESG) criteria and
corporate nancial performance (CFP) can be traced back to the beginning of the 1970s.
Scholars and investors have published more than 2000 empirical studies and several review
studies on this relation since then. The largest previous review study analyzes just a fraction
of existing primary studies, making ndings difcult to generalize. Thus, knowledge on the
nancial effects of ESG criteria remains fragmented. To overcome this shortcoming, this
study extracts all provided primary and secondary data of previous academic review studies.
Through doing this, the study combines the ndings of about 2200 individual studies.
Hence, this study is by far the most exhaustive overview of academic research on this topic
and allows for generalizable statements. The results show that the business case for ESG
investing is empirically very well founded. Roughly 90% of studies nd a nonnegative
ESGCFP relation. More importantly, the large majority of studies reports positive ndings.
We highlight that the positive ESG impact on CFP appears stable over time. Promising
results are obtained when differentiating for portfolio and nonportfolio studies, regions, and
young asset classes for ESG investing such as emerging markets, corporate bonds, and
green real estate.
Keywords: second-order meta-analysis; vote-count studies; nancial performance; ESG
criteria; business case
Close to 60 trillion US Dollars in assets under management or 50% of the total global insti-
tutional assets base are currently managed by Principles for Responsible Investment (PRI) sig-
natories (PRI 2015a). On the one hand, this development clearly demonstrates the commitment of
nancial markets toward environmental, social, and governance (ESG) criteria within investment
decisions. However, on the other hand, far-reaching shifts of mainstream investors toward embra-
cing sustainable investment practices remain rather slow (Reynolds 2014; Busch, Bauer, and
© 2015 The Author(s). Published by Taylor & Francis.
*Corresponding author. Email:
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License
(, which permits non-commercial re-use, distribution, and reprod uction in any medium,
provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
Journal of Sustainable Finance & Investment, 2015
Vol. 5, No. 4, 210233,
Downloaded by [] at 20:09 30 January 2016
Orlitzky 2015;PRI2015b). Less than a quarter of investment professionals consider extra-nan-
cial information frequently in their investment decisions (EY 2015) and just about 10% of global
professionals receive formal training on how to consider ESG criteria in investment analysis (CFA
Institute 2015). For many, the business case for responsible investing seems not obvious (Feri
2009; Cohen et al. 2011; Riedl and Smeets 2015). Still, the question of how compatible ESG cri-
teria are with corporate nancial performance (CFP) has remained a central debate for prac-
titioners and academics alike for more than 40 years.
Though there are many positive examples for the ESGCFP relation, researchers often claim
that results are ambiguous, inconclusive, or contradictory (Aupperle, Carroll, and Hateld 1985;
Grifn and Mahon 1997; Rowley and Berman 2000; van Beurden and Gössling 2008; Hoepner
and McMillan 2009; Revelli and Viviani 2015). Scholars and practitioners are, in particular, unde-
cided about the general effect including its measurement and durability (Barnett 2007; Devinney
2009; Wood 2010; Orlitzky 2011; Borgers et al. 2013; Orlitzky 2013; Reynolds 2014; Authers
2015). Thus, there is an ongoing debate about the role and the impact of the nancial sector
on the natural environment and society (Weber 2014). In order to derive a more comprehensive
picture, several review studies summarize primary ESGCFP studies. Yet, all these rst-level
review studies provide an incomplete picture. This study is the rst effort to provide aggregated
evidence based on more than 2000 empirical studies that have been released since the 1970s (see
Figure 1).
We chose a two-step research method to analyze existing review and primary studies. First, we
include ndings from so-called vote-count studies. Vote-count studies count the number of studies
with signicant positive, negative, and nonsignicant results and votesthe category with the
highest share as winner (Light and Smith 1971). These studies provide interesting insights, but
are less sophisticated from a methodological point of view. The shortcomings are well documen-
ted in the literature.
Second, we aggregate the ndings of econometric review studies so-called
meta-analyses to derive a second-order meta-analysis.
In total, 60 review studies both vote-count studies and meta-analyses with a gross number of
3718 underlying studies on the empiric relation between ESG criteria and CFP provide the starting
point for our second-level review study.
When adjusted for overlaps, this gure reduces to a net
number of more than 2200 unique studies. This still represents a dataset, which is 35 times
Figure 1. Estimated number of empirical studies on the ESGCFP relation over time.
Journal of Sustainable Finance & Investment 211
Downloaded by [] at 20:09 30 January 2016
larger than the average of analyzed primary studies in prior review studies. In this study, we explain
both systematic methods of summarizing extant research and present a research symbiosis of vote-
count studies and meta-analyses in the spirit of a best-evidence synthesis (Slavin 1986).
Through analyzing what is by far the most comprehensive dataset on existing ESGCFP
research to date, we nd that the business case for ESG investing is empirically well founded.
Investing in ESG pays nancially. Furthermore, we highlight that the positive ESG impact on
CFP is stable over time. Based on the data, we are able to derive conclusions for portfolio and
nonportfolio studies,
different asset classes, regions, and categories of E, S, and
G. Particularly promising results are obtained when we differentiate between regions, nonportfo-
lio studies, and asset classes other than equities.
Since the earliest review of vote-count ESGCFP studies (Aldag and Bartol 1978), the studies
providing secondary analysis of this relation has risen considerably, including both academic
and numerous additional practitioner papers. The growth in number of ESGCFP research pub-
lications has been particularly tremendous since the beginning of the 1990s. Based on our sample,
we nd that at least 2200 empirical ESGCFP studies exist.
For our analysis, only academic studies regardless if they are working papers, published journal
papers, or written for a commercial audience were considered. Review papers that did not provide
quantitative summaries of their ndings were not included in our sample. Besides ancestry research
and expert opinion, all relevant scholar databases and publisher sites were searched: Academy of
Management Journals, ABI/Inform, Ebsco, Emerald, Google Scholar, Oxford Journals, Sage,
Science Direct, Sprinker Link, and Web of Science. We also searched for nonpublished material
on Econbiz, NBER, Repec, and SSRN. The keyword search combinations included the three com-
ponents of E, S, and G and its abbreviations. In particular, we used the search terms environment(al)
(performance), social (performance), responsib(le/ility), sustainab(le/ility), human capital, (corpor-
ate) governance allinrelationto(corporate)nancial performance.
The rst 100 hits of each single database and key word query, sorted by relevance, were
further processed. Within this pre-ltered results we then searched for the terms meta, review, lit-
erature, overview, analysis, study/ies, and examination. Together with the expert opinion studies,
this yielded a narrower sample of 149 studies, which were analyzed in more detail by abstract or
full paper. Single study designs, narrative reviews without clear tables/explicit summary results,
and review studies without relevant ESGCFP categorization were excluded. We applied a de-
nition of ESG that reects the exemplary list of variables of Clarkson (1995), Wood (2010), and
the investment approaches in GSIA (2013). We did not differentiate whether the motives for ESG
performance of the rm are for altruistic or strategic reasons (McGuire 1969; Baron 2001; McWil-
liams, Siegel, and Wright 2006). CFP measures were dened as accounting-based performance,
market-based performance, operational performance, perceptual performance, growth metrics,
risk measures, and the performance of ESG portfolios (Cochran and Wood 1984; Orlitzky and
Benjamin 2001; Orlitzky, Schmidt, and Rynes 2003; Peloza 2009). We also considered specic
parts of a study (Viviers and Eccles 2012; Mayer-Haug et al. 2013; Stam, Arzlanian, and
Elfring 2014) when its focus was not entirely on the ESGCFP relation provided a vote-
count estimate or effect size calculation was possible. In case of different versions of a review
study, the latest version or ideally, the published version remained in our sample. All
studies were required to be available in electronic format. The cut-off date for study inclusion
was online availability until December 2014.
212 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
Table 1. Overview of studies on the ESGCFP relation (vote-count studies sample).
Share of ndings
Study Focus
Number of
studies (N) Positive Neutral Negative Mixed
Arlow and Gannon (1982) S 7 42.9% 42.9% 14.3%
Cochran and Wood (1984) S, E 13 69.2% 23.1% 7.7%
Aupperle, Carroll, and Hateld
S, E 9 55.6% 22.2% 11.1% 11.1%
Ullmann (1985) S, E 24 54.2% 20.8% 12.5% 12.5%
Capon, Farley, and Hoenig (1990) S, E 14 75.9% 19.5% 4.6%
Wood and Jones (1995) S, E 51 49.0% 21.6% 13.7% 15.7%
Pava and Krausz (1996) S, E 21 57.1% 38.1% 4.8%
Grifn and Mahon (1997) S, E 50 44.0% 12.0% 22.0% 22.0%
Roman, Hayibor, and Agle (1999) S, E 45 60.0% 24.4% 4.4% 11.1%
Richardson, Welker, and
Hutchinson (1999)
E, S 22 50.0% 45.5% 4.5%
Margolis and Walsh (2003) S, E 126 42.9% 22.2% 5.6% 29.4%
Salzmann, Ionescu-Somers, and
Steger (2005)
S, E 12 50.0% 25.0% 25.0%
McWilliams, Siegel, and Wright
S, E 12 33.3% 25.0% 16.7% 25.0%
Gillan and Starks (2007) G 39 35.9% 43.6% 5.1% 15.4%
Ambec and Lanoie (2007) E 41 68.3% 22.0% 4.9% 4.9%
van Beurden and Gössling (2008) E, S 34 67.6% 26.5% 5.9%
Peloza (2009) S, E 130 63.0% 22.0% 15.0%
Blanco, Rey-Maquieira, and
Lozano (2009)
E 32 71.9% 21.9% 6.3%
Molina-Azorín et al. (2009) E 32 62.5% 12.5% 12.5% 12.5%
Horváthová (2010) E 44 54.7% 29.7% 15.6% 0%
Westlund and Adam (2010) S 21 85.7% 14.3%
Love (2010) G 45 77.8% 0% 22.2%
Derwall, Koedijk, and Horst (2011) Funds 18 16.7% 33.3% 22.2% 27.8%
Günther, Hoppe, and Endrikat
E 274 44.5% 11.8% 43.7%
Sjöström (2011) E, S 21 23.8% 33.3% 14.3% 28.6%
Boaventura, Santos da Silva, and
Bandeira-de-Mello (2012)
S, E 58 55.2% 27.6% 10.3% 6.9%
Rathner (2013) Funds 25 13.2% 72.0% 14.9% 0%
Schultze and Trommer (2012) E 36 50.0% 19.4% 5.6% 25.0%
Viviers and Eccles (2012) Funds 59 23.4% 56.2% 20.3%
Fifka (2013) Reporting 45 53.3% 42.2% 4.4%
Kleine, Krautbauer, and Weller
E, S, G 182 30.8% 31.9% 7.7% 29.7%
Revelli and Viviani (2013) Funds 75 24.0% 48.0% 14.7% 13.3%
Capelle-Blancard and Monjon
Funds 61 3.3% 47.5% 16.4% 32.8%
Clark, Feiner, and Viehs (2015) E, S, G 110 85.5% 5.1% 0.9% 8.5%
Schröder (2014) E, S 28 57.1% 7.1% 10.7% 25.0%
Total/n-weighted average 1.816 48.2% 23.0% 10.7% 18.0%
This table displays all considered vote-count studies for the analysis. Meta-analytical studies with nontransferable or
nontransparent effect sizes that nonetheless allow a vote-count analysis were included in the vote-count studies sample as
well. Focus Sand Edenote a Social (S) or Environmental (E) focus. For studies with combined E and S focus, the
order of S and E indicates the relative weight of S vs. E. The labeling E, S, Gindicates no relative weight within groups.
The number of primary studies in each vote-count analysis is denoted N. For vote-count studies with transparent vote-
count on primary study, the share of ndings is calculated based on this primary information. For all other cases, the
reported summary results of the vote-count reviewers are used.
Journal of Sustainable Finance & Investment 213
Downloaded by [] at 20:09 30 January 2016
In total, we identied 35 vote-count studies (Table 1) and 25 meta-analyses (Table 2) which
combine results from 3718 (gross) primary studies of which 1816 studies stem from vote-
count studies and 1902 from meta-analyses. All studies were scaled with a unique identier in
the format author 1, author 2, , author i(year). Different review author citations formats, cita-
tions years of study versions, and author typing errors were normalized. All available statistical
Table 2. Overview of studies on the ESGCFP relation (meta-analyses sample).
Authors Focus
Number of
studies (N)
Number of
observations (N)
Average correlation r
Frooman (1997) E, S 22 2.161 0.312(b)
Orlitzky and Benjamin (2001) S, E 18 6.186 0.149
Orlitzky (2001) S, E 20 6.889 0.061
Orlitzky, Schmidt, and Rynes
S, E 62 33.878 0.184
Allouche and Laroche (2005) S, E 79 57.409 0.143
Combs et al. (2006) S 90 19.319 0.150
Wu (2006) S, E 120 21.933 0.166
Rosenbusch, Bausch, and
Galander (2007)
E 62 21.742 0.190
Darnall and Sides (2008) E 9 30.000 0.077
Pavie and Filho (2008) S, E 112 170.737 0.083
van Wijk, Jansen, and Lyles
S 28 4.627 0.190
Margolis, Elfenbein, and
Walsh (2009)
S, E 214 38.483 0.133
Vishwanathan (2010) E, S 189 n.a. 0.070
Crook et al. (2011) S 66 12.163 0.170
Rosenbusch, Brinckmann,
and Bausch (2011)
S 46 21.270 0.133
Unger et al. (2011) S 70 24.733 0.076
Rubera and Kirca (2012) S 153 33.544 0.146
Albertini (2013) E 52 62.943 0.090
del Mar Miras-Rodríguez
et al. (2015)
E, S 91 31.878 0.067
Dixon-Fowler et al. (2013) E 39 22.869 0.062
Golicic and Smith (2013) E 31 15.160 0.305
Mayer-Haug et al. (2013) S 58 50.045 0.044(b)
Endrikat, Guenther, and
Hoppe (2014)
E 148 201.511 0.082(b)
Stam, Arzlanian, and Elfring
S 43 13.263 0.157
Revelli and Viviani (2015) Funds 80 89.496 0.003(a)
Total/n-weighted average 1.902 992.239 0.118
This table displays all considered meta-analyses for the analysis and includes the number of primary studies in each meta-
analysis and the corresponding number of observations. Amplications on the original reported number of included
primary studies have been made, if not the entire sample was used. For four meta-analyses not all originally reported
number of studies could be veried through data in the provided appendix and for three studies a condensed study set was
used. The gross number of studies therefore decreases from 2091 to 1902 studies. Focus Sand Edenote a Social (S) or
Environmental (E) focus. For studies with integrated S and E focus, the order of S and E indicates the relative weight of S
vs. E studies. The indices (a) and (b) for the uncorrected effect size rindicate the source of the effect size in case, it was
modied from the originally stated results: (a) transformed from d/g in rand (b) derived from stated corrected study results
with either a meta-analysis provided individual attenuation factor or a calculated artifact attenuation factor of 0.72.
214 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
summary information of review studies and all information reported on primary study level were
imported and normalized for further statistical analysis.
Not all primary studies were made transparent by the review authors. Eight review studies
containing 929 primary studies (25.0% of the sample) were not identiable on primary level.
This meant that the results were included in the summary effects, but no further analysis on
primary study level was possible. Within the remaining uniquely identiable 2789 (gross)
primary studies, the overlap within review studies was subsequently accounted for. The resulting
net number of identiable unique primary studies was n= 723 for the vote-count studies and
n= 1214 for the meta-analyses. Of these, 259 studies overlap within the two review approaches
which brought the nal number of unique identiable primary studies in the sample to n= 1678.
Those 259 overlapping studies remained within the vote-count studies and meta-analyses sample
as separation to one or the other review approach was not possible without losing data granularity.
Based on our sample of unique identiable studies (n= 1678) and the number of nontransparent
studies (n= 929), we estimated that at least 550 studies need to be added for a more complete estimate
of the overall number of existing empirical studies on ESGCFP published since the 1970s. This esti-
mate was adjusted for the various overlaps within the vote-count studies and meta-analyses sample.
Two different ways for aggregation of the primary and secondary study results are applied, each
with different calculation methods depending on the context. For comparability between results in
the vote-count studies and meta-analyses, we compute distributions of outcomes and correlation
effect sizes. Besides aggregated summary effects, we provide further ne-grained analysis on sub-
group level. Depending on data availability in vote-count studies and meta-analyses, an analysis
for different asset classes, regions, categories of E, S, and G as well the relation over time is con-
ducted. When both vote-count studies and meta-analyses offer this information, the more compre-
hensive primary study sample is chosen. When the sub-sample stems from vote-count studies, the
analysis focuses on the distribution of outcomes; when the sub-sample stems from meta-analyses,
the focus is on effect sizes.
Raw correlations, corrected correlations, sample sizes as well as corresponding variances, stan-
dard errors, condence interval (CI), and credibility interval (CrI) have been extracted from the orig-
inal meta-analyses as far as possible for further calculations. If necessary, some of the effect sizes
and variances were transformed or derived for the calculation of a second-order meta-analysis.
Calculation of distributions
Vote-count studies
Distributions of positive, negative, neutral, and mixed outcomes are calculated for vote-count
studies based on the results of the gross study sample and the net study sample. Within the
gross study sample, it is possible that the same primary study is analyzed multiple times by differ-
ent review study authors who may interpret each study differently. These interpretations are
treated as independent study outcomes no further adjustments are made. When a primary
study is analyzed by more than one review author, the net study sample is adjusted for this con-
straint and different review authorsinterpretations are harmonized. On average, every unique
primary study in the vote-count sample is analyzed by 1.8 review authors. To decide on the
overall interpretation per unique study, a binomial test with three equally probable outcomes is
applied (positive, neutral, and negative). A probability of greater than .95 served as cut-off
point to determine the nal interpretation for the study. If no clear positive or negative assignment
was possible, the study is classied as neutral and/or mixed.
Journal of Sustainable Finance & Investment 215
Downloaded by [] at 20:09 30 January 2016
Vote-count reviewers provide an assessment of the extent to which an observed relation in a
primary study is a signicant outcome. When undertaking a meta-analysis of primary studies,
this assessment is performed by the second-level reviewer. In order to adjust for signicance
of the results, we employ a 95% CI and 95% CrI based on the determined meta-analytical variance
in the n= 25 meta-analyses. We calculate the 95% CI via the determined standard error (SE) for
attenuated (r) and disattenuated (p) results.
and SEp=
CIL=0+1.96(SEr/p)and CIU=0+1.96(SEr/p) (2)
The 95% CrI is then calculated via the standard deviation of the attenuated and disattenuated
and SDp=
CrIL=0+1.96(SDr/p) and CrIU=0+1.96(SDr/p) (4)
The true variance for r
ˆ=i(the meta-analytical mean of attenuated correlations) is ˆ
iand the
corresponding variance for p
ˆ=i(the meta-analytical mean of disattenuated correlations) is ˆ
we are interested in the degree of signicant positive and negative results, we place the intervals
around zero instead of the meta-analytical mean. The resulting distributions may appear unusual
on rst glance as they dene results signicantly different from zero. The calculation is also con-
ducted for the uncorrected and corrected correlations in the 551 primary studies, which possess
transparent effect size data. We apply the same corresponding intervals that are utilized for the
set of meta-analyses. Finally, we determine the number of studies that are above and below the
intervals, categorize them as positive or negative and put them in relation with the sample size.
All studies within the interval are classied as neutral.
Calculation of effect sizes
Vote-count studies
Even though vote-count studies usually do not report effects sizes like standardized mean differences
(d/g) or correlations, it is possible to approximate them with the provided data. Methods have been
introduced during the time when broader application of meta-analytical techniques were being devel-
oped (Hedges and Olkin 1980; Hedges and Olkin 1985) and were further rened in the 1990s
(Bushman 1994; Bushman and Wang 1995).
The effect size ris determined by calculating the
ratio of p0(
), which divides the number of positive studies by the sum of positive and negative
studies, and by putting it in relation with the corresponding number of studies n. The estimation
for the correlation coefcient rof a single vote-count study is subsequently determined through
linear extrapolation based on the correlations coefcients provided (Hedges and Olkin 1985, 63ff).
In a nal step, the estimated correlation factors per vote-count study are sample-size weighted and
aggregated to an overall estimation of the average correlation rfor the vote-count sample.
216 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
First-order meta-analytical results for the sample of primary studies are calculated with the Hunter
Schmidt approach (Hunter, Schmidt, and Jackson 1982; Hunter and Schmidt 2004). The approach is
used by more than 80% of meta-analyses in management research (Aguinis et al. 2011). It is similar
to the second-order meta-analytical methodology which applies a fully random effect model.
All other average effect sizes and summary statistics of the 25 meta-analyses are determined
with Schmidt and Ohs method for second-order meta-analysis (Schmidt and Oh 2013). A second-
order meta-analysis combines a number of methodologically comparable and independent rst-
order meta-analyses. It allows knowledge aggregation across a tremendous set of primary
studies. Such a meta-analysis sample is potentially closer to a complete set of studies in certain
research elds and allows for robust generalizations. Apart from the efcient aggregation of
huge datasets, the method is statistically superior to other approaches for summarizing rst-
order meta-analyses. Conventional approaches will most likely provide inaccurate estimates of
the true mean effect size and are prone to second-order sampling errors in the variances across
all meta-analyses. The approach chosen considerably reduces the remaining sampling error var-
iance of rst-order meta-analyses and allows a better estimation of the true (nonartifactual) var-
iance across these mean effect sizes (Schmidt and Oh 2013). Because of the considerable number
of rst-order meta-analyses in our sample which make use of artifact distribution correction, we
calculate our results with the artifact distribution approach of Schmidt and Oh.
In order to differentiate whether correlations and corresponding variance are rst-order (based
on extracted primary studies) or second-order (aggregated vote-count studies or meta-analyses), we
add one or two lines above letters for attenuated correlations rand disattenuated correlations p.The
applied circumex accent indicates that the values in the meaning of psychometric meta-analysis are
estimates of the parameters, not the parameters themselves (Schmidt and Oh 2013;Schmidtand
Hunter 2015, 229). Correlations containing lines above the letter but not marked with a circumex
are meta-analytical averages but are not determined using a psychometric meta-analysis.
Figure 2 displays our summary of ndings: approximately 90% of studies nd a nonnegative
ESGCFP relation, of which 47.9% in vote-count studies and 62.6% in meta-analyses yield posi-
tive ndings with a central average correlation level in studies of around 0.15. The following para-
graphs discuss the ndings in more detail.
Figure 2. Overall summary results.
Journal of Sustainable Finance & Investment 217
Downloaded by [] at 20:09 30 January 2016
Summary effects: distributions
Vote-count studies
In a rst step for the analysis of distribution results, all 1816 vote-count studies in the gross
sample are treated as unique studies without adjusting for overlap among the vote-count
studies. The overall weighted share of positive ndings in the sample is calculated at 48.2%.
In 41.0% of all results, the ndings lead to neutral (23.0%) or mixed ndings (18.0%). Just
10.7% of all analyzed studies exhibit a negative ESGCFP relation.
In a second step, we account for the amount of nondisclosed and overlapping studies among
the gross number of 1816 studies. The transparent studies are netted and in case that the rst-level
reviewer assessments differ, the ndings are synthesized with a binomial test. The additional
check does not meaningfully change the distribution of the positive and neutral ndings
(47.9% and 22.5%, respectively). However, a proportion of the negative ndings cannot be con-
sidered statistically signicant anymore if two or more reviewer interpretations are synthesized
with a binomial test. The share of negative ndings in the sample decreases to 6.9% of
studies. Instead, the share of mixed results increases to 22.7%.
Either way, depending on which of the two approaches (unadjusted gross studies/net studies
adjusted with binomial test) is applied, close to 50% of all analyzed studies in the vote-count
sample nd a positive relation and around 10% a negative one. The small distribution difference
in the results is explained by a slightly more comprehensive overall sample and the net approach
for study interpretation when more than one reviewer analyzed the same primary study (Figure 3).
Out of the 25 meta-analyses in the sample, just one study displays a summary effect size that has a
negative ESGCFP correlation albeit very close to zero (Revelli and Viviani 2015). The sample
size adjusted share of absolute positive correlation ndings in meta-analytical summary effect for
the 1902 studies stands with 95.8% considerable higher than in vote-count studies. However, this
number is not adjusted for statistical signicance. If we apply the 95% CI and 95% CrI, for the
meta-analytical summary effects and the number of transparent primary studies, the gures
change accordingly (Table 3). For the 25 meta-analyses, the share of positive ndings is
Figure 3. ESGCFP relation in vote-count studies.
218 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
reduced to 74.9% (95% CrI, attenuated results). However, the share of negative results remains at
0%, as the lowest effect in the 25 meta-analyses is 0.003. A quarter (25.1%) of the sample effect
sizes is within the CrI and is correspondingly classied as neutral results.
To eliminate potential positive biases in these meta-analytical summary effects, we also drill
down to the primary study level and the sample of 551 studies. The share of studies with signi-
cant positive correlations is reduced to a minimum of 62.6% (95% CrI, attenuated results) with a
maximum percentage of negatives as high 14.5% (95% CI, disattenuated results). An attenuated
correlation level (interval) above 0.141 would be needed to bring down the percentage of positive
correlations to the level in the vote-count studies of 47.9%. This cut-off is close to the population
unweighted average correlation of 0.159.
Summary effects: correlations
Vote-count studies
Next, an approximation of the correlation effect size in vote-count studies based on the vote-count
method of Hedges and Olkin (1985, 47ff) is conducted. The weighted average correlation rv
=in all
vote-count studies is calculated at 0.146. The corresponding p-value of <.001 indicates a corre-
lation factor highly signicant and different from zero. The additional check for statistical power
(Cohen 1988; Faul et al. 2007) reveals that for the determined rv
=and the corresponding number of
n, the chance of a Type II error is close to zero.
For reasons of comparability with the vote-count effect size estimate rv
=, we compute the attenu-
ated sample-size weighted average correlation for the 25 meta-analyses. The calculated corre-
lation rm
=is 0.118. The p-value of similarity of rm
=and rv
=is notably high at 0.638. This means
vote-count studies and meta-analyses determine statistically comparable results for the ESG
CFP relation. However, generalizing this nding for both methods universally may not be appro-
priate due to the almost independent samples containing few overlaps and very different variance
Next, we calculate the rst-order meta-analytical averages as both uncorrected and corrected
parameters for the transparent sub-sample of 551 primary studies. The correlation
iis determined
at 0.119 and
iat 0.169. The meta-analytical second-order effect size for the 25 meta-analyses
Table 3. Distribution results in dependency of correlation intervals.
Percentage of studies
Share in Interval type
Correlation interval
(r) Positive Negative Neutral
n = 25 meta-
adj. 95% CI, attenuated ±0.0147 95.8 0 4.2
adj. 95% CrI, attenuated ±0.0733 74.9 0 25.1
adj. 95% CI, disattenuated ±0.0185 95.8 0 4.2
adj. 95% CrI, disattenuated ±0.0924 90.7 0 9.3
n = 551 primary
adj. 95% CI, attenuated ±0.0147 80.9 14.2 4.9
adj. 95% CrI, attenuated ±0.0733 62.6 8.0 29.4
adj. 95% CI, disattenuated ±0.0185 80.9 14.5 4.5
adj. 95% CrI, disattenuated ±0.0924 63.9 8.5 27.6
This table outlines the results of the distribution analysis of positive, negative, and neutral results in meta-analyses and in
dependency of different correlation intervals. Whereas CI indicates the condence interval and CrI the credibility interval.
Journal of Sustainable Finance & Investment 219
Downloaded by [] at 20:09 30 January 2016
combining 1902 gross studies reveals a correlation of r
ˆ=i= 0.108 and for the corrected effect size
ˆ=i= 0.150. Worth mentioning is that the value for the vote-count effect size rv
=is statistically not
different from the rst- and second-order meta-analytical results (minimum p-value .351 for the
difference to
i). Even though the vote-count technique is a rough estimate based on simplied
assumptions, it nonetheless yields surprisingly comparable estimations of the ESGCFP relation
compared to the sample of meta-analyses aggregated with the method for second-order meta-
analysis at least for our setup.
The p-values for all of our meta-analytical means are below .01 and indicate a statistical highly
signicant positive deviation from zero. In a similar manner, the 95% CrI of 0.0580.242 for
another indicator of the positive nature of the ESGCFP relation. Moreover, the control for stat-
istical power of these values reveals very robust results, with a Cohens power for all gures
above 0.8 and in four cases close to 1 (Table 4).
Portfolio studies and nonportfolio studies
All previous vote-count and meta-analysis effects contain a blend of nonportfolio and portfolio
studies. Making this differentiation is important as aggregated rm performance in virtual port-
folios and nancial products such as mutual funds or indices may deviate from primary rm
data. Several primary portfolio studies and corresponding reviews report an abnormally low
level of positive ndings. And vice versa, a high level of mixed ndings, compared to nonport-
folio studies (Bauer, Koedijk, and Otten 2005; Peloza 2009; Kleine, Krautbauer, and Weller 2013;
Revelli and Viviani 2015). In order to differentiate between portfolio and nonportfolio effects, we
deconstruct all distributions and summary effect sizes with sufcient sample size in both blocks of
study groups.
The relevance of this distinction becomes apparent when looking at the vote-count studies.
The share of positive results in the n= 155 identiable portfolio-related studies shrinks
Table 4. Effect size in dependence of aggregation approach and sample.
Number of
analyses N
size value power
α= 0.05
35 1.816 rv
=0.146*** 0.999
25 1.902 rm
=0.118*** 0.999
551 r
i0.119*** 0.804
i0.169*** 0.991
25 1.902
i0.108*** 0.997 0.0014 0.094 0.123 0.035 0.182 717
25 1.902 p
ˆ=i0.150*** 0.999 0.0022 0.132 0.169 0.058 0.242 987
This table depicts rst-order and second-order meta-analytical results. Attenuated correlations are marked as rand
disattenuated correlations as p. The number of lines above letters indicates the rst- or second-level nature of the effect
size. The additional circumex accent indicates psychometric meta-analytically effect sizes. Correlations not marked with
a circumex are meta-analytical averages but not determined with psychometric meta-analysis. rv
=is the sample-size
weighted second-level effect size in vote-count analyses and rm
=the sample-size weighted second-level average effect size
in meta-analyses.
iare the attenuated and disattenuated rst-level effects sizes in transparent primary studies, and
and p
ˆ=iare the second-order meta-analytical attenuated and disattenuated averages. The deviation of all effect sizes from
zero is tested for signicance. ˆ
2is the estimated true variance for r
ˆ=iand p
ˆ=i. Power 0.05 is Cohens power with α= 0.05
and for the corresponding sample and effect size. For the second-order meta-analytical correlation means r
ˆ=iand p
symbolize the 95% CI and CrI
the 95% CrI. L indicates the lower and U the upper bound of the interval. Fail-Safe N
is Rosenthals(1979,1991) statistic to detect potential publication bias in meta-analyses. Fail-Safe Nstates the number of
(future) studies with null results, until the effect size loses its signicance level. A level above 5*n+ 10 for Fail-Safe Nis
considered unlikely to assume publication bias within studies. This can be ruled out. The signicance thresholds for
p-values are ***p< .01.
220 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
considerably (15.5%) in comparison to nonportfolio-based studies (56.7%). Studies with neutral
or mixed ndings increase proportionately in portfolio-based studies and constitute nearly three
quarters. The share of negative studies increases marginally compared to nonportfolio studies
(11.0% vs. 5.8%) (Figure 4).
Comparable results are found when all portfolio-focused vote-count studies are separately
analyzed based on estimated effect sizes. The ve primarily portfolio-focused vote-count
studies exhibit a negative correlation rv
=(p)of 0.061 in comparison to the 30 primarily nonport-
folio-focused vote-count studies with rv
=(non-p)of 0.177. The difference between both groups is
highly signicant (Table 5). These conclusions are supported when the rst-level meta-analytical
results in the transparent primary studies are deconstructed. The differences in correlations are not
so pronounced, nonetheless signicant (Table 5). This distinct deviation of portfolio and nonport-
folio ndings is examined in more detail in the Discussionsection.
Figure 4. ESGCFP relation in vote-count studies in dependence of portfolio- and nonportfolio sample.
Table 5. Effect size in dependence of portfolio and nonportfolio samples.
Number of
review analyses NEffect size Effect size value power
α= 0.05
Non-pand pdifference
30 1.578 rv(nonp)
=0.177*** 0.999 .238***
5 238 rv(p)
=0.061* 0.154
i(nonp)0.131*** 0.815 .076**
80 r
i(p)0.055** 0.078
i(nonp)0.183*** 0.981 .089**
i(p)0.094*** 0.132
This table deconstructs the results in Table 4 for rv
i, and
iin nonportfolio studies effect sizes (non-p) and portfolio
studies effect sizes (p). Power0.05 is Cohens power with α= 0.05 and for the corresponding sample and effect size. The
difference of portfolio and nonportfolios studies in effect sizes is tested for signicance. The signicance thresholds for p-
values are *p< .10, **p< .05, and ***p< .01.
Journal of Sustainable Finance & Investment 221
Downloaded by [] at 20:09 30 January 2016
Sub-effects in asset classes
Aside from the overall distribution of results and correlation factors in vote-count studies and
meta-analyses, the data allow for examinations of differences in asset classes (DAntonio,
Johnsen, and Hutton 1997; Bhojraj and Sengupta 2003; Eichholtz, Kok, and Quigley 2010)
although with limited availability of nonequity classes. In a sub-sample consisting of 751
gross and 334 net studies within the vote-count sample 87.1% analyze equity-linked relations.
In contrast, nonequity asset classes both for bonds and real estate display a considerably
higher share of positive ndings over equities. More than two-thirds of studies uncover signicant
positive performance relations to ESG criteria. The share of positive votes for the 36 analyzed
bond studies stands at 63.9% with 13 neutral or mixed ndings (36.1%). The relatively
young research eld of green real estate studies is reected with seven studies in the total
sample. Five studies (71.4%) nd a positive and the other two a neutral relation (Figure 5).
Sub-effects in ESG categories
A key question is whether any of the three ESG letters may have a dominating effect on CFP.
Some meta-analyses nd signicant positive relations for corporate environmental performance
and CFP (Albertini 2013; Dixon-Fowler et al. 2013; Endrikat, Guenther, and Hoppe 2014).
Human capital-focused meta-analyses (Combs et al. 2006; Crook et al. 2011; Rosenbusch,
Brinckmann, and Bausch 2011) also nd highly signicant positive correlations. Various
review studies on multifaceted corporate governance aspects and its relation to CFP exist and
also support a positive relation (Dalton et al. 1999; Gillan and Starks 2007; Love 2010).
However, not all of the E-, S-, and G-specicndings are free from ambiguity and no large-
scale comparison between the subgroups has been undertaken yet.
For our sample of vote-count studies with identiable ESG categories in 644 studies, we
determine a relatively even positive relation for E, S, and G. The highest proportion is found
in G with 62.3% of all cases. Governance-related aspects, on the other hand, demonstrate also
the highest percentage of negative correlations with 9.2%. If the share of negative ndings is
deducted from positive ones, environmental studies offer the most favorable relation (58.7
4.3%). Studies with a social focus show 55.1% (5.1%) positive (negative) outcomes, hence the
weakest relation.
Figure 5. ESGCFP relation in main asset classes (vote-count studies sample), n= 334 net studies.
222 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
When reviewing studies with various combinations of ESG criteria, 35.3% report positive
(respectively 7.1% negative) ndings. The downside bias primarily arises from a high proportion
of portfolio-based studies in this section (39.1%). If all these studies were excluded, the positive
(negative) rate stands at 51.7% (4.8%) which is nonetheless lower than pure E, S, and G
approaches (Figure 6).
Sub-effects in regions
Some studies have analyzed potential differences in the ESGCFP relation across regions.
Though ndings are far from consistent, some hypothesized that the ESGCFP relation across
countries is particularly affected by a higher humane orientation (del Mar Miras-Rodríguez, Car-
rasco-Gallego, and Escobar-Pérez 2015). Others nd that the ESGCFP relationship for US assets
is signicantly higher compared to non-US assets (Allouche and Laroche 2005; Dixon-Fowler
et al. 2013). In contrast, a few researchers also discover signicantly higher effects for studies
conducted in the rest of the world (Albertini 2013; Golicic and Smith 2013).
We detect two main patterns in the data based on 843 gross studies with disclosed regional
identier that are netted for a nal sample of 402 studies. First, developed markets excluding
North America exhibit a smaller share of positive results. This contrast is most apparent
between North America (42.7% positive) and developed Europe (26.1% positive). Developed
Asia/Australia possess a positive share of 33.3%, though with the largest share of negatives as
well at 14.3%. The total sample excluding North American stands at 27.8% positive share. A
check of the underlying studies reveals a larger share of portfolio-based studies within the
European and Asian/Australian sample that potentially biases the data. However, when omitting
all portfolio studies for the developed market samples, the positive ratio for North America
increases to 51.5%, and for Europe and Asia/Australia combined to 45.6%. This implies that
the previous gap between the two samples shrinks considerably from 14.9 to 5.9 percentage
Second, the Emerging Markets sample shows, with 65.4%, a considerable higher share of
positive outcomes over developed markets. Excluding the proportion of portfolio studies, the
ratio increases further to 70.8%. Based on 52 single studies in Emerging Markets solely
focused on equity-linked studies, the spread to developed markets is considerable (Figure 7).
Figure 6. E, S, and G categories and their relation to CFP (vote-count studies sample), n= 644 net studies.
Journal of Sustainable Finance & Investment 223
Downloaded by [] at 20:09 30 January 2016
ESG effect over time
The question has been raised of whether the ESGCFP relation is stable over time (Grifn and
Mahon 1997; Borgers et al. 2013). Theoretically, the increasing amount of UN PRI signatories
and, presuming an increasing ESG awareness within investment strategies, a decreasing ESG
alpha (shrinking correlations over time) would be expected due to learning effects in capital
markets. Empiric ndings of meta-analyses investigating if investorsincreased focus on
Figure 7. ESGCFP relation in various regions (vote-count studies sample), n= 402 net studies.
Figure 8. ESGCFP correlation factors in primary studies in dependency of study publishing dates (meta-
analyses sample), n= 551 net studies.
224 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
stakeholder issues also lead to changing ESGCFP patterns over time present a fuzzy picture
(Pavie and Filho 2008; Rubera and Kirca 2012; Albertini 2013; Endrikat, Guenther, and
Hoppe 2014)(Table 2).
Out of the 1214 primary studies in meta-analyses, 551 studies possess transparent correlation
coefcients and publication years. For this sub-sample, we do not nd indications to support the
learning hypothesis. Although the dispersion of effects, positive and negative alike, increase since
the beginning of the 1990s, the aggregated picture stays unchanged. Besides simple observations
of the regression line, the time-invariant relation is supported by various trend tests which all fail
to detect a time-dependent change of the correlation factors for every year since the mid of the
1990s (Figure 8).
Both vote-count and meta-analytic studies yield comparable results. This is a surprising outcome
since the underlying studies are comprised of nearly independent samples (12.9% overlap) and
apply different methods. Both methods yield robust results which reinforces the claim that
there is a business case for ESG investing. On the one hand, the effect size-transformed vote-
count results do not overestimate effect sizes for our sample and lead to comparable results
measured as correlation rin comparison to the meta-analytical studies (rv
== 0.146 vs. mean cor-
relation in meta-analyses between 0.108 and 0.169). Vote-count studies produce, on the other
hand, more modest estimates for determining the proportion of positive and negative ndings
compared to meta-analyses. The share of neutral/mixed results is potentially overestimated for
vote-count studies. Vote-count reviews determine whether the effect per study is signicant by
narrowly focusing on the underlying primary study sample size. Meta-analyses, by contrast,
average effects across the entire sample of underlying studies which reduces the meta-analytical
mean variance. Smaller variances mean lower thresholds (lower correlations) for signicance in
While overall correlation averages between 0.108 (r
ˆ=i) and 0.169 (
i) could be considered
rather small(Cohen 1988,1992), they reect common effect sizes in social sciences
(Richard, Bond, and Stokes-Zoota 2003; Tamim et al. 2011; Lipsey et al. 2012) and, notably,
might have relatively high relevance for competitive global securities markets. Based on corre-
lation factors and the distribution analysis of more than 2000 empirical studies, we feel condent
in generalizing that ESG criteria and CFP are, on average, positively correlated.
The distinct positive empiric result is found across various approaches, regions, and asset
classes except for portfolio-related studies. This outlier is potentially the source of the wide-
spread misperception on the ESGCFP relation. Institutional and private investors typically con-
clude that the ESGCFP relation is, at best, neutral consistent with the neoclassical
understanding of capital markets (Markowitz 1952; Fama 1970; Friedman 1970; Fama 1991).
Such an assumption about the ESGCFP relation can be a key barrier for the broad uptake of sus-
tainable investing among investors and investment advisors (Paetzold and Busch 2014; Reynolds
2014; CFA Institute 2015; Paetzold, Busch, and Chesney 2015).
The realized performance in portfolios depends on the overlapping effects of systematic and
idiosyncratic risks (Campbell et al. 2001; Luo and Bhattacharya 2009), on construction con-
straints (Clarke, de Silva, and Thorley 2002), and on costs for portfolio implementation
(Carhart 1997; Khorana, Servaes, and Tufano 2009) which may distorts pure ESG performance.
Indeed, we nd a signicant difference in correlation levels of portfolio and nonportfolio studies.
We argue that ESG portfolios should be expected to exhibit lower correlations to CFP and less
positive ndings for the following three reasons: (1) following the drowned out by noiseargu-
ment (Peloza 2009), various overlapping market and nonmarket factors in a portfolio tend to
Journal of Sustainable Finance & Investment 225
Downloaded by [] at 20:09 30 January 2016
cover potentially existing ESG alpha. (2) Most ESG funds constitute a mixture of so-called nega-
tive and positive ESG-screened funds, which could result in distortion and cancellation of any
remaining effects (Derwall, Koedijk, and Ter Horst 2011). (3) Only studies on portfolios (in par-
ticular mutual funds) embed management fees and other costs such as performance fees and
trading costs. Observed effects in rm-specic study designs are typically calculated without
such fees and costs. As roughly 2.5% per annum in various fees are carried by the average
mutual fund (Carhart 1997; Barber, Odean, and Zheng 2005; Khorana, Servaes, and Tufano
2009), real correlation patterns in portfolio studies are most likely distorted. We conclude that
portfolio-study ndings have to be treated as a specic outcome of a subgroup within the
entire ESGCFP discussion. Investors, on average, are unlikely to harvest the existing ESG
alpha after implementation costs. However, it can be argued, sophisticated investors are more
likely to do so (Grossman and Stiglitz 1980; Hoepner 2013; Nagy, Kassam, and Lee 2015).
Thus, our results underpin previous ndings: at the worst case, investors in ESG mutual funds
can expect to lose nothing compared to conventional fund investments (Hamilton, Jo, and
Statman 1993; Humphrey and Tan 2014; Revelli and Viviani 2015).
Regional ndings reveal that within developed markets, there is a higher share of positive
results from North America compared to Europe and Asia/Australia. This can partially be
explained by the lower share of portfolio studies within the sub-sample for North America.
Within the individual E, S, and G categories, E and G exhibit a slightly more positive relation
than S-focused studies. However, the difference between E and S studies with positive and nega-
tive outcomes is marginal (maximum 4.3% percentage points). Meta-analyses focusing on social
aspects (van Wijk, Jansen, and Lyles 2008; Crook et al. 2011; Stam, Arzlanian, and Elfring 2014)
usually nd higher correlations, in contrast to environmental-focused meta-analyses (Albertini
2013; Dixon-Fowler et al. 2013; Endrikat, Guenther, and Hoppe 2014). We conclude that no
single E, S, and G category demonstrates a meaningful superior positive relation to CFP.
The strength of our analysis is the aggregation of a large number of studies through secondary
research on review studies, but it is also uncovers the inherit limitations of the underlying studies.
One of them is the lengthy academic publication period of primary and likewise secondary
research. Although our second-level review study includes all relevant review studies published
until the end of 2014, it loses some representativeness for primary studies with a publication date
of 2012 and younger.
Through a second-level review of 60 review studies including both, vote-count studies and
meta-analyses on the ESGCFP relation, we are able to combine more than 3700 study
results from more than 2200 unique primary studies. Based on this sample, we clearly nd evi-
dence for the business case for ESG investing. This nding contrasts with the common perception
among investors. The contrary perception of investors may be biased due to ndings of portfolio
studies, which exhibit, on average, a neutral/mixed ESGCFP performance relation. It is impor-
tant to be aware that the results of these (to date about 150 studies) are overlaid by various sys-
tematic and idiosyncratic risks in portfolios and, in the case of mutual funds, by implementation
costs. Still more than 2100 other in particular company-focused empiric studies suggest a
positive ESG relation.
ESG outperformance opportunities exist in many areas of the market. In particular, we nd
that this holds true for North America, Emerging Markets, and in nonequity asset classes. Our
results propose that capital markets so far demonstrate no consistent learning effects regarding
the ESGCFP relation: Since the mid-1990s, the positive correlation patterns in primary
studies have been stable over time (Table 1).
226 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
Based on this exhaustive review effort, our main conclusion is: the orientation toward long-
term responsible investing should be important for all kinds of rational investors in order to fulll
their duciary duties and may better align investorsinterests with the broader objectives of
society. This requires a detailed and profound understanding of how to integrate ESG criteria
into investment processes in order to harvest the full potential of value-enhancing ESG factors.
A key area for future research is to better understand the interaction of different ESG criteria
in portfolios and the relevance of specic ESG sub-criteria for CFP. These insights will shed
further light on the ESG determinants for long-term positive performance impacts.
The authors like to thank The PRI, Deutsche Asset & Wealth Management (Deutsche AWM) and two anon-
ymous reviewers for their support. The views expressed in this paper are not necessarily shared by PRI or
Deutsche AWM.
Disclosure statement
No potential conict of interest was reported by the authors.
1. The statistical explanatory power in studies is usually low and the primary study might come to the con-
clusion, based on its calculated signicance values and sample sizes, that a certain effect is nonsigni-
cant. Vote-count reviews may also come to biased conclusions by simply concentrating on signicant
statistics of primary studies to decide if an effect across studies is positive or negative. Potentially they
overestimate nonsignicant results. Besides, the explanatory power of vote-count studies shrinks with
the increasing number of (contradictory) studies. Meta-studies directly import effect sizes and samples
sizes to compute a summary effect across all primary studies. This aggregation method of data could
better detect existing correlation patterns in combined samples (Hedges and Olkin 1980; Hunter
et al. 1982).
2. The term second-level review studydescribes our aggregation of rst-level review studies, regardless
if they are vote-count studies or meta-analyses. Second-order meta-analysisis the psychometric
aggregation technique for rst-level meta-analyses as introduced by Schmidt and Oh (2013). This tech-
nique is used for the statistical aggregation of the 25 meta-analyses in our sample to compute summary
effect sizes.
3. Portfolio studies comprise of studies on long-short ESG portfolios and in particular studies on ESG
mutual funds and indices.
4. Two of the typical ways to treat missing data are model-based distribution estimation and the replace-
ment of missing data (imputation) with estimated ones (Schafer and Graham 2002; Tsikriktsis 2005).
The latter is applied due to the nonparametric nature of the data. We estimate the total number of
missing net studies based on the subgroup means of overlaps in transparent vote-count studies, meta-
analyses, and among both. The determined subgroup overlap means are applied to each subgroup of
nontransparent studies.
5. The method assumes simplistically comparable sample sizes for the underlying primary studies, which
is rather the exception in research synthesis. It is also constructed as xed effect model, which assumes
that studies draw samples from a population with the same standardized mean difference (Hedges and
Olkin 1980). The calculated effect size for the vote-count sample should therefore be seen as quick
approximate estimate instead of a nal analysis (Hedges and Olkin 1985).
6. Please refer to table 1, p. 210 in Schmidt and Oh (2013) for a technical summary of the approach.
7. The applied tests are Pettitt, SNHT, Buishand, and von Neumann. The null hypothesis of the tests veri-
es if a time series is homogenous between two randomly selected times within the time series. The
different tests allow conclusions not only for an assumed normal distribution but also for nonparametric
distributions. Only data previous to 1997 are assessed as nonhomogenous to later observations in 2 of 4
tests. The SNHT test detects signicantly higher results before 1985. The Buishand test detects signi-
cantly higher results previous to 1997 at the .05 level.
Journal of Sustainable Finance & Investment 227
Downloaded by [] at 20:09 30 January 2016
Aguinis, Herman, Dan. R. Dalton, Frank A. Bosco, Charles A. Pierce, and Catherine M. Dalton. 2011.
Meta-Analytic Choices and Judgment Calls: Implications for Theory Building and Testing, Obtained
Effect Sizes, and Scholarly Impact.Journal of Management 37 (1): 538. doi:10.1177/
Albertini, Elisabeth. 2013.Does Environmental Management Improve Financial Performance? A Meta-
Analytical Review.Organization & Environment 26 (4): 431457. doi:10.1177/1086026613510301.
Aldag, Ramon J., and Kathryn M. Bartol. 1978.Empirical Studies of Corporate Performance and Policy: A
Survey of Problems and Results.In Research in Corporate Social Performance and Policy, edited by L.
E. Preston, 165199. Greenwich: JAI Press.
Allouche, José, and Patrice Laroche. 2005.A Meta-Analytical Investigation of the Relationship Between
Corporate Social and Financial Performance.Revue de Gestion Des Ressources Humaines 57: 118.
Ambec, Stefan, and Paul Lanoie. 2007.When and Why Does It Pay To Be Green?20. Série Scientique
Scientic Series Montréal.
Arlow, Peter, and Martin J. Gannon. 1982.Social Responsiveness, Corporate Structure, and Economic
Performance.Academy of Management Review 7 (2): 235241. doi:10.5465/AMR.1982.4285580.
Aupperle, Kenneth E., Archie B. Carroll, and John D. Hateld. 1985.An Empirical Examination of the
Relationship Between Corporate Social Responsibility and Protability.Academy of Management
Journal 28 (2): 446463. doi:10.2307/256210.
Authers, John. 2015.Vice versus Nice.Financial Times, Europe, June 25, 7.
Barber, Brad M., Terrance Odean, and Lu Zheng. 2005.Out of Sight, Out of Mind: The Effects of Expenses
on Mutual Fund Flows.The Journal of Business 78 (6): 20952120. doi:10.1086/497042.
Barnett, Michael L. 2007.Stakeholder Inuence Capacity and the Variability of Financial Returns to
Corporate Social Responsibility.Academy of Management Review 32 (3): 794816. doi:10.5465/
Baron, David P. 2001.Private Politics, Corporate Social Responsibility, and Integrated Strategy.Journal of
Economics & Management Strategy 10 (1): 745. doi:10.1162/105864001300122548.
Bauer, Rob, Kees Koedijk, and Rogér Otten. 2005.International Evidence on Ethical Mutual Fund
Performance and Investment Style.Journal of Banking and Finance 29 (7): 17511767. doi:10.
van Beurden, Pieter, and Tobias Gössling. 2008.The Worth of Values a Literature Review on the Relation
Between Corporate Social and Financial Performance.Journal of Business Ethics 82 (2): 407424.
Bhojraj, Sanjeev, and Partha Sengupta. 2003.Effect of Corporate Governance on Bond Ratings and Yields:
The Role of Institutional Investors and Outside Directors.The Journal of Business 76 (3): 455475.
Blanco, Esther, Javier Rey-Maquieira, and Javier Lozano. 2009.The Economic Impacts of Voluntary
Environmental Performance of Firms: A Critical Review.Journal of Economic Surveys 23 (3): 462
502. doi:10.1111/j.14676419.2008.00569.x.
Boaventura, João Maurício Gama, Ralph Santos da Silva, and Rodrigo Bandeira-de-Mello. 2012.Corporate
Financial Performance and Corporate Social Performance: Methodological Development and the
Theoretical Contribution of Empirical Studies.Revista Contabilidade & Finanças 23 (60): 232245.
Borgers, Arian, Jeroen Derwall, Kees Koedijk, and Jenke ter Horst. 2013.Stakeholder Relations and Stock
Returns: On Errors in InvestorsExpectations and Learning." Journal of Empirical Finance 22 (June):
159175. doi:10.1016/j.jempn.2013.04.003.
Busch, Timo, Rob Bauer, and Marc Orlitzky. 2015.Sustainable Development and Financial Markets: Old
Paths and New Avenues.Business & Society 127. doi:10.1177/0007650315570701.
Bushman, Brad J. 1994.Vote-Counting Procedures in Meta-Analysis.In The Handbook of Research
Synthesis and Meta-Analysis, edited by Harris Cooper and Larry V. Hedges, 193213. New York:
Russel Sage Foundation.
Bushman, Brad J., and Morgan C. Wang. 1995.A Procedure for Combining Sample Correlation
Coefcients and Vote Counts to Obtain an Estimate and a Condence Interval for the Population
Correlation Coefcient.Psychological Bulletin 117 (3): 530546. doi:10.1037/0033-2909.117.3.530.
Campbell, John Y., Martin Lettau, Burton G. Malkiel, and Yexian Xu. 2001.Have Individual Stocks
Become More Volatile? An Empirical Exploration of Idiosyncratic Risk.The Journal of Finance 56
(1): 143. doi:10.1111/0022-1082.00318.
228 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
Capelle-Blancard, Gunther, and Stéphanie Monjon. 2014.The Performance of Socially Responsible Funds:
Does the Screening Process Matter?European Financial Management 20 (3): 494520. doi:10.1111/j.
Capon, Noel, John U. Farley, and Scott Hoenig. 1990.Determinants of Financial Performance: A Meta-
Analysis.Management Science 36 (10): 11431159.
Carhart, Mark M. 1997.On Persistence in Mutual Fund Performance.Journal of Finance 52 (1): 5782.
CFA Institute. 2015.Environmental, Social and Governance (ESG) Survey.
Clark, Gordon L., Andreas Feiner, and Michael Viehs. 2015.From the Stockholder to the Stakeholder: How
Sustainability Can Drive Financial Outperformance.doi:10.2139/ssrn.2508281.
Clarke, Roger, Harindra de Silva, and Steven Thorley. 2002.Portfolio Constraints and the Fundamental
Law of Active Management.Financial Analysts Journal 58 (5): 4866. doi:10.2469/faj.v58.n5.2468.
Clarkson, Max B. E. 1995.A Stakeholder Framework for Analyzing and Evaluating Corporate Social
Performance.Academy of Management Review 20 (1): 92117. doi:10.5465/AMR.1995.9503271994.
Cochran, Philip L., and Robert A. Wood. 1984.Corporate Social Responsibility and Financial
Performance.Academy of Management Journal 27 (1): 4256. doi:10.2307/255956.
Cohen, Jacob. 1988.Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Academic
Cohen, Jacob. 1992.A Power Primer.Psychological Bulletin 112 (1): 155159. doi:10.1037/0033-2909.
Cohen, Jeffrey, Lori Holder-Webb, Leda Nath, and David Wood. 2011.Retail InvestorsPerceptions of the
Decision-Usefulness of Economic Performance, Governance, and Corporate Social Responsibility
Disclosures.Behavioral Research in Accounting 23 (1): 109129. doi:10.2308/bria.2011.23.1.109.
Combs, James, Yongmei Liu, Angela Hall, and David Ketchen. 2006.How Much Do High-Performance
Work Practices Matter? A Meta-Analysis of Their Effects on Organizational Performance.Personnel
Psychology 59 (3): 501528. doi:10.1111/j.1744-6570.2006.00045.x.
Crook, T. Russell, Samuel Y. Todd, James G. Combs, David J. Woehr, and David J. Ketchen. 2011.Does
Human Capital Matter? A Meta-Analysis of the Relationship Between Human Capital and Firm
Performance.Journal of Applied Psychology 96 (3): 443456. doi:10.1037/a0022147.
Dalton, Dan R., Catherine M. Daily, Jonathan L. Johnson, and Alan E. Ellstrand. 1999.Number of Directors
and Financial Performance: A Meta-Analysis.Academy of Management Journal 42 (6): 674686.
DAntonio, Louis, Tommi Johnsen, and R. Bruce Hutton. 1997.Expanding Socially Screened Portfolios.
The Journal of Investing 6 (4): 7986. doi:10.3905/joi.1997.408434.
Darnall, Nicole, and Stephen Sides. 2008.Assessing the Performance of Voluntary Environmental
Programs: Does Certication Matter?The Policy Studies Journal 36 (1): 95117. doi:10.1111/j.
Derwall, Jeroen, Kees Koedijk, and Jenke Ter Horst. 2011.A Tale of Values-Driven and Prot-Seeking
Social Investors.Journal of Banking & Finance 35 (8): 21372147. doi:10.1016/j.jbankn.2011.01.
Devinney, Timothy M. 2009.Is the Socially Responsible Corporation a Myth? The Good, the Bad, and the
Ugly of Corporate Social Responsibility.Academy of Management Perspectives 23 (2): 4456. doi:10.
Dixon-Fowler, Heather R., Daniel J. Slater, Jonathan L. Johnson, Alan E. Ellstrand, and Andrea M. Romi.
2013.Beyond Does It Pay to Be Green?A Meta-Analysis of Moderators of the CEPCFP
Relationship.Journal of Business Ethics 112 (2): 353366. doi:10.1007/s10551-012-1268-8.
Eichholtz, Piet, Nils Kok, and John M. Quigley. 2010.Doing Well by Doing Good? Green Ofce
Buildings.American Economic Review 100 (5): 24922509. doi:10.1257/aer.100.5.2492.
Endrikat, Jan, Edeltraud Guenther, and Holger Hoppe. 2014.Making Sense of Conicting Empirical
Findings: A Meta-Analytic Review of the Relationship Between Corporate Environmental and
Financial Performance.European Management Journal 32 (5): 735751. doi:10.1016/j.emj.2013.12.
EY. 2015.Tomorrows Investment Rules 2.0. Emerging Risk and Stranded Assets Have Investors Looking for
More from Nonnancial Reporting (EY Climate Change and Sustainability Services).
Fama, Eugene F. 1970.Efcient Capital Markets: A Review of Theory and Empirical Work.The Journal
of Finance 25 (2): 383417. doi:10.1111/j.1540-6261.1970.tb00518.x.
Fama, Eugene F. 1991.Efcient Capital Markets: II.The Journal of Finance 46 (5): 15751617. doi:10.
Journal of Sustainable Finance & Investment 229
Downloaded by [] at 20:09 30 January 2016
Faul, Franz, Edgar Erdfelder, Albert-Georg Lang, and Axel Buchner. 2007.G*Power 3: A Flexible
Statistical Power Analysis Program for the Social, Behavioral, and Biomedical Sciences.Behavior
Research Methods 39 (2): 175191. doi:10.3758/bf03193146.
Feri. 2009.Nachhaltige Investments Und Nachhaltigkeitsfonds Aus Anlegerperspektive Studie 2009 (Feri
Euro Rating Services). Bad Homburg.
Fifka, Matthias S. 2013.Corporate Responsibility Reporting and Its Determinants in Comparative
Perspective a Review of the Empirical Literature and a Meta-Analysis.Business Strategy and the
Environment 22 (1): 135. doi:10.1002/bse.729.
Friedman, Milton. 1970.The Social Responsibility of Business is to Increase Its Prots.The New York
Times Magazine.doi:10.1007/978-3-540-70818-6_14.
Frooman, Jeff. 1997.Socially Irresponsible and Illegal Behavior and Shareholder Wealth: A Meta-Analysis
of Event Studies.Business & Society 36 (3): 221249. doi:10.1177/000765039703600302.
Gillan, Stuart L., and Laura T. Starks. 2007.The Evolution of Shareholder Activism in the United States.
Journal of Applied Corporate Finance 19 (1): 5573. doi:10.1111/j.1745-6622.2007.00125.x.
Golicic, Susan L., and Carlo D. Smith. 2013.A Meta-Analysis of Environmentally Sustainable Supply
Chain Management Practices and Firm Performance.Journal of Supply Chain Management 49 (2):
7895. doi:10.1111/jscm.12006.
Grifn, Jennifer J., and John F. Mahon. 1997.The Corporate Social Performance and Corporate Financial
Performance Debate: Twenty-Five Years of Incomparable Research.Business & Society 36 (1): 531.
Grossman, Sanford J., and Joseph E. Stiglitz. 1980.On the Impossibility of Informationally Efcient
Markets.The American Economic Review 70 (3): 393408.
GSIA. 2013.Global Sustainable Investment Review 2012.146. Accessed August 3 http://
Günther, Edeltraud, Holger Hoppe, and Jan Endrikat. 2011.Corporate Financial Performance and
Corporate Environmental Performance: A Perfect Match?Zeitschrift Für Umweltpolitik Und
Umweltrecht 34 (3): 279296.
Hamilton, Sally, Hoje Jo, and Meir Statman. 1993.Doing Well While Doing Good? The Investment
Performance of Socially Responsible Mutual Funds.Financial Analysts Journal 49 (6): 6266. doi:
Hedges, Larry V., and Ingram Olkin. 1980.Vote-Counting Methods in Research Synthesis.Psychological
Bulletin 88 (2): 359369. doi:10.1037/0033-2909.88.2.359.
Hedges, Larry V., and Ingram Olkin. 1985.Statistical Methods for Meta-Analysis. San Diego: Academic
Hoepner, Andreas G. F. 2013.Environmental, Social, and Governance (ESG) Data: Can It Enhance Returns
and Reduce Risks? Deutsche AWM Global Financial Institute.
Hoepner, Andreas G. F., and David G. McMillan. 2009.Research on Responsible Investment:An
Inuential Literature Analysis Comprising a Rating, Characterisation, Categorisation and Investigation.
SSRN Electronic Journal:184. doi:10.2139/ssrn.1454793.
Horváthová, Eva. 2010.Does Environmental Performance Affect Financial Performance? A Meta-
Analysis.Ecological Economics 70 (1): 5259. doi:10.1016/j.ecolecon.2010.04.004
Humphrey, Jacquelyn E., and David T. Tan. 2014.Does It Really Hurt to Be Responsible?Journal of
Business Ethics 122 (3): 375386. doi:10.1007/s10551-013-1741-z.
Hunter, John E., and Frank L. Schmidt. 2004.Methods of Meta-Analysis: Correcting Error and Bias in
Research Findings. 2nd ed. Newbury Park: Sage.
Hunter, John E., Frank L. Schmidt, and Gregg B. Jackson. 1982.Meta-Analysis: Cumulating Research
Findings Across Studies. Beverly Hills: Sage.
Khorana, Ajay, Henri Servaes, and Peter Tufano. 2009.Mutual Funds Fees Around the World.Review of
Financial Studies 22 (3): 12791310. doi:10.1093/rfs/hhn042.
Kleine, Jens, Matthias Krautbauer, and Tim Weller. 2013.Nachhaltige Investments Aus Dem Blick Der
Wissenschaft: Leistungsversprechen Und Realität. Research Center for Financial Services Steinbeis-
Hochschule Berlin.
Light, R. J., and P. V. Smith. 1971.Accumulating Evidence: Procedures for Resolving Contradictions
among Different Research Studies.Harvard Educational Review 41 (4): 429471. doi:10.17763/
Lipsey, Mark W., Kelly Puzio, Cathy Yun, Michael A. Hebert, Kasia Steinka-Fry, Mikel W. Cole, Megan
Roberts, Karen S. Anthony, and Matthew D. Busick. 2012.Translating the Statistical Representation
230 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
of the Effects of Education Interventions into More Readily Interpretable Forms.National Center for
Special Education Research (NSCER 2013-3000).
Love, Inessa. 2010.Corporate Governance and Performance Around the World: What We Know and What
We Dont.The World Bank Research Observer 26 (1): 4270. doi:10.1093/wbro/lkp030.
Luo, Xueming, and C. B. Bhattacharya. 2009.The Debate over Doing Good: Corporate Social
Performance, Strategic Marketing Levers, and Firm-Idiosyncratic Risk.Journal of Marketing 73 (6):
198213. doi:10.1509/jmkg.73.6.198.
Margolis, Joshua D., Hillary A. Elfenbein, and James P. Walsh. 2009.Does It Pay to Be Good and Does
It Matter? A Meta-Analysis of the Relationship Between Corporate Social and Financial Performance.
SSRN Electronic Journal:168. doi:10.2139/ssrn.1866371.
Margolis, Joshua D., and James P. Walsh. 2003.Misery Loves Companies: Rethinking Social Initiatives by
Business.Administrative Science Quarterly 48 (2): 268305. doi:10.2307/3556659.
Markowitz, Harry. 1952.Portfolio Selection.Journal of Finance 7 (1): 7791. doi:10.2307/2329297.
del Mar Miras-Rodríguez, María, Amalia Carrasco-Gallego, and Bernabé Escobar-Pérez. 2015.Are
Socially Responsible Behaviors Paid Off Equally? A Cross-Cultural Analysis.Corporate Social
Responsibility and Environmental Management 22 (4): 237256. doi:10.1002/csr.1344.
Mayer-Haug, Katrin, Stuart Read, Jan Brinckmann, Nicholas Dew, and Dietmar Grichnik. 2013.
Entrepreneurial Talent and Venture Performance: A Meta-Analytic Investigation of SMEs.
Research Policy 42 (67): 12511273. doi:10.1016/j.respol.2013.03.001.
McGuire, Joseph W. 1969.The Changing Nature of Business Responsibilities.The Journal of Risk and
Insurance 36 (1): 5561. doi:10.2307/251140.
McWilliams, Abagail, Donald S. Siegel, and Patrick M. Wright. 2006.Corporate Social Responsibility:
Strategic Implications.Journal of Management Studies 43 (1): 118. doi:10.1111/j.1467-6486.2006.
Molina-Azorín, José F., Enrique Claver-Cortés, Maria D. López-Gamero, and Juan J. Tarí. 2009.Green
Management and Financial Performance: A Literature Review.Management Decision 47 (7): 1080
1100. doi:10.1108/00251740910978313.
Nagy, Zoltán, Altaf Kassam, and Linda-Eling Lee. 2015.Can ESG Add Alpha? An Analysis of ESG Tilt
and Momentum Strategies (MSCI ESG Research Inc.).
Orlitzky, Marc. 2001.Does Firm Size Confound the Relationship Between Corporate Social Performance
and Firm Financial Performance?Journal of Business Ethics 33 (2): 167180. doi:10.1023/
Orlitzky, Marc. 2011.Institutional Logics in the Study of Organizations: The Social Construction of the
Relationship Between Corporate Social and Financial Performance.Business Ethics Quarterly 21
(3): 409444. doi:10.5840/beq201121325.
Orlitzky, Marc. 2013.Corporate Social Responsibility, Noise, and Stock Market Volatility.Academy of
Management Perspectives 27 (3): 238254. doi:10.5465/amp.2012.0097.
Orlitzky, Marc, Frank L. Schmidt, and Sarah L. Rynes. 2003.Corporate Social and Financial Performance:
A Meta-Analysis.Organization Studies 24 (3): 403441. doi:10.1177/0170840603024003910.
Orlitzky, Marc, and John D. Benjamin. 2001.Corporate Social Performance and Firm Risk: A Meta-
Analytic Review.Business & Society 40 (4): 369396. doi:10.1177/000765030104000402.
Paetzold, Falko, and Timo Busch. 2014.Unleashing the Powerful Few: Sustainable Investing Behaviour of
Wealthy Private Investors.Organization & Environment 27 (4): 347367. doi:10.1177/1086026614555991.
Paetzold, Falko, Timo Busch, and Marc Chesney. 2015.More than Money: Exploring the Role of
Investment Advisors for Sustainable Investing.Annals in Social Responsibility 1 (1): 195223.
Pava, Moses L., and Joshua Krausz. 1996.The Association Between Corporate Social-Responsibility and
Financial Performance: The Paradox of Social Cost.Journal of Business Ethics 15 (3): 321357.
Pavie, Juliana Junqueira Esmeraldo, and Luiz Alberto Nascimento Campos Filho. 2008.Corporate Social
Responsibility and Financial Performance: A Meta-Analysis.Paper presented at the 34th EIBA annual
conference paper, Tallinn, December 1113.
Peloza, John. 2009.The Challenge of Measuring Financial Impacts from Investments in Corporate Social
Performance.Journal of Management 35 (6): 15181541. doi:10.1177/0149206309335188.
PRI (Principles for Responsible Investment). 2015a.Signatory Base AUM Hits $59 Trillion.http://www.
PRI (Principles for Responsible Investment). 2015b.The Principles for Responsible Investment Report on
Progress 2015.
Journal of Sustainable Finance & Investment 231
Downloaded by [] at 20:09 30 January 2016
Rathner, Sebastian. 2013.The Inuence of Primary Study Characteristics on the Performance Differential
Between Socially Responsible and Conventional Investment Funds: A Meta-Analysis.Journal of
Business Ethics 118 (2): 349363. doi:10.1007/s10551-012-1584-z.
Revelli, Christophe, and Jean-Laurent Viviani. 2013.The Link Between SRI and Financial Performance:
Effects and Moderators.Management International 17 (2): 105122. doi:10.7202/1015403ar.
Revelli, Christophe, and Jean-Laurent Viviani. 2015.Financial Performance of Socially Responsible
Investing (SRI): What Have We Learned? A Meta-Analysis.Business Ethics: A European Review
24 (2): 158185. doi:10.1111/beer.12076.
Reynolds, Fiona. 2014.Mainstream Slow to Accept Benets of Responsible Investment.Financial Times,
Europe, November 17, 22.
Richard, F. D., Charles F. Bond, and Juli J. Stokes-Zoota. 2003.One Hundred Years of Social Psychology
Quantitatively Described.Review of General Psychology 7 (4): 331363. doi:10.1037/1089-2680.7.4.
Richardson, Alan J., Michael Welker, and Ian R. Hutchinson. 1999.Managing Capital Market Reactions to
Corporate Social Responsibility.International Journal of Management Reviews 1 (1): 1743. doi:10.
Riedl, Arno, and Paul Smeets. 2015.Why Do Investors Hold Socially Responsible Mutual Funds?SSRN
Working Paper Series. doi:10.2139/ssrn.2354905.
Roman, Ronald M., Sefa Hayibor, and Bradley R. Agle. 1999.The Relationship Between Social and
Financial Performance: Repainting a Portrait.Business & Society 38 (1): 109125. doi:10.1177/
Rosenthal, Robert. 1979. The File Drawer Problem and Tolerance for Null Results.Psychological Bulletin
86 (3): 638641. doi:10.1037/0033-2909.86.3.638.
Rosenthal, Robert. 1991. Meta-Analytic Procedures for Social Research. 2nd ed. Newbury Park: Sage.
Rosenbusch, Nina, Andreas Bausch, and Anne Galander. 2007.The Impact of Environmental
Characteristics on Firm Performance: A Meta-Analysis.Academy of Management Annual Meeting
Proceedings:16. doi:10.5465/AMBPP.2007.26530718.
Rosenbusch, Nina, Jan Brinckmann, and Andreas Bausch. 2011.Is Innovation Always Benecial? A Meta-
Analysis of the Relationship Between Innovation and Performance in SMEs.Journal of Business
Venturing 26 (4): 441457. doi:10.1016/j.jbusvent.2009.12.002.
Rowley, Tim, and Shawn Berman. 2000.A Brand New Brand of Corporate Social Performance.Business
& Society 39 (4): 397418. doi:10.1177/000765030003900404.
Rubera, Gaia, and Ahmet H. Kirca. 2012.Firm Innovativeness and Its Performance Outcomes: A Meta-
Analytic Review and Theoretical Integration.Journal of Marketing 76: 130147. doi:10.1509/jm.10.
Salzmann, Oliver, Aileen Ionescu-Somers, and Ulrich Steger. 2005.The Business Case for Corporate
Sustainability: Literature Review and Research Options.European Management Journal 23 (1): 27
36. doi:10.1016/j.emj.2004.12.007.
Schafer, Joseph L., and John W. Graham. 2002.Missing Data : Our View of the State of the Art.
Psychological Bulletin 7 (2): 147177. doi:10.1037//1082-989X.7.2.147.
Schmidt, Frank L., and In-Sue Oh. 2013.Methods for Second Order Meta-Analysis and Illustrative
Applications.Organizational Behavior and Human Decision Processes 121 (2): 204218. doi:10.
Schmidt, Frank L., and John E. Hunter. 2015.Methods of Meta-Analysis: Correcting Error and Bias in
Research Findings. 3rd ed. Newbury Park: Sage.
Schröder, Michael. 2014.Financial Effects of Corporate Social Responsibility: A Literature Review.
Journal of Sustainable Finance & Investment 4 (4): 337350. doi:10.1080/20430795.2014.971096.
Schultze, Wolfgang, and Ramona Trommer. 2012.The Concept of Environmental Performance and Its
Measurement in Empirical Studies.Journal of Management Control 22 (4): 375412. doi:10.1007/
Sjöström, Emma. 2011.The Performance of Socially Responsible Investment A Review of Scholarly
Studies Published 20082010.SSRN Electronic Journal:119. doi:10.2139/ssrn.1948169.
Slavin, Robert E. 1986.Best-Evidence Synthesis: An Alternative to Meta-Analytic and Traditional
Reviews.Educational Researcher 15 (9): 511. doi:10.3102/0013189X015009005.
Stam, Wouter, Souren Arzlanian, and Tom Elfring. 2014.Social Capital of Entrepreneurs and Small Firm
Performance: A Meta-Analysis of Contextual and Methodological Moderators.Journal of Business
Venturing 29 (1): 152173. doi:10.1016/j.jbusvent.2013.01.002.
232 G. Friede et al.
Downloaded by [] at 20:09 30 January 2016
Tamim, Rana M., Robert M. Bernard, Eugene Borokhovski, Philip C. Abrami, and Richard F. Schmid. 2011.
What Forty Years of Research Says About the Impact of Technology on Learning: A Second-Order
Meta-Analysis and Validation Study.Review of Educational Research 81 (1): 428. doi:10.3102/
Tsikriktsis, Nikos. 2005.A Review of Techniques for Treating Missing Data in OM Survey Research.
Journal of Operations Management 24 (1): 5362. doi:10.1016/j.jom.2005.03.001.
Ullmann, Arieh A. 1985.Data in Search of a Theory: A Critical Examination of the Relationships among
Social Performance, Social Disclosure, and Economic Performance of U.S. Firms.The Academy of
Management Review 10 (3): 540557. doi:10.2307/258135.
Unger, Jens M., Andreas Rauch, Michael Frese, and Nina Rosenbusch. 2011.Human Capital and
Entrepreneurial Success: A Meta-Analytical Review.Journal of Business Venturing 26 (3): 341
358. doi:10.1016/j.jbusvent.2009.09.004.
Vishwanathan, Pushpika. 2010.The Elusive Relationship Between Corporate Social and Financial
Performance: Meta-Analyzing Four Decades of Misguided Evidence.Academy of Management
Proceedings 2010 (1): 17. doi:10.5465/AMBPP.2010.54493634.
Viviers, S., and N. S. Eccles. 2012.35 Years of Socially Responsible Investing (SRI) Research General
Trends over Time.South African Journal of Business Management 43 (4): 116. doi:http://hdl.handle.
Weber, Olaf. 2014.The Financial Sectors Impact on Sustainable Development.Journal of Sustainable
Finance & Investment 4 (1): 18. doi:10.1080/20430795.2014.887345.
Westlund, Hans, and Frane Adam. 2010.Social Capital and Economic Performance: A Meta-Analysis of 65
Studies.European Planning Studies 18 (6): 893919. doi:10.1080/09654311003701431.
van Wijk, Raymond, Justin J. P. Jansen, and Marjorie A. Lyles. 2008.Inter- and Intra-Organizational
Knowledge Transfer: A Meta-Analytic Review and Assessment of Its Antecedents and
Consequences.Journal of Management Studies 45 (4): 830853. doi:10.1111/j.1467-6486.2008.
Wood, Donna J. 2010.Measuring Corporate Social Performance: A Review.International Journal of
Management Reviews 12 (1): 5084. doi:10.1111/j.1468-2370.2009.00274.x.
Wood, Donna J., and Raymond E. Jones. 1995.Stakeholder Mismatching: A Theoretical Problem in
Empirical Research on Corporate Social Performance.The International Journal of Organizational
Analysis 3 (3): 229267. doi:10.1108/eb028831.
Wu, Meng-Ling. 2006.Corporate Social Performance, Corporate Financial Performance, and Firm Size: A
Meta-Analysis.The Journal of American Academy of Business 8 (1): 163171.
Journal of Sustainable Finance & Investment 233
Downloaded by [] at 20:09 30 January 2016
... Some scholars believe that the ESG practice of enterprises violates the principle of profit maximization, and the company may miss investment opportunities, resulting in inefficient investment portfolios and destroying enterprise value (Fabozzi et al., 2008;Revelli and Viviani, 2015). However, most scholars have found that ESG practices can improve corporate competitive advantage, bring returns to shareholders (Porter and Kramer, 2006), and improve the performance of enterprises (Friede et al., 2015;Mervelskemper and Streit, 2017;Yu et al., 2018). ...
Full-text available
According to information asymmetry theory and stakeholder theory, this article explores the impact and mechanism of environmental, social, and governance (ESG) information disclosure on the company’s future stock price crash risk based on the A-share listed companies from 2010 to 2019. We find that ESG information disclosure significantly reduces the company’s future stock price crash risk. This conclusion remains robust after a series of robustness tests, such as PSM-DID. The heterogeneity analysis shows that the negative relationship between ESG disclosure and stock price crash risk is more significant in state-owned enterprises, companies with higher agency costs, and when companies in the bull market. The mechanism is that companies choose to disclose ESG information to alleviate information asymmetry problems and enhance corporate reputation capital, thus reducing the future stock price crash risk. This article shows that strengthening ESG construction will help improve the efficiency of China’s resource allocation and promote the capital market development.
... A mobilização do sistema financeiro e questões éticas, associadas à ideia de sustentabilidade remonta a 1970 (Friede et al., 2015). No entanto, é recente e crescente a argumentação que a temática de finanças é um vetor central para as economias se transformarem mais sustentáveis (NaIdoo, 2020). ...
Full-text available
Palavras-chave: ESG, Finanças Sustentáveis; ODS; Sustentabilidade; Inovação; Eletrobras; Setor Elétrico; 1. Introdução Em novembro de 2021 foi realizada a Conferência das Partes (COP 26) em Glasgow (Escócia), encontro promovido pelas Nações Unidas para discussão do Clima. Além de mobilizar participantes de quase 200 países, promoveu mais uma vez a discussão sobre como a humanidade se posicionará nos próximos anos para mitigar os potenciais efeitos decorrentes das mudanças climáticas. À exceção da pandemia (COVID 19), nenhum outro assunto tem concorrido mais pelo protagonismo do que a discussão global sobre a urgência de se pensar no futuro e nas relações humanas com o meio ambiente. Como desenvolver um ambiente sustentável de forma que a vida humana se desenvolva economicamente consistente e duradoura e sem que para isso o planeta e a espécie humana fiquem sob ameaça é uma questão fundamental. O debate global possui várias motivações, sendo a perenidade da espécie humana a maior delas. O assunto não é novo, nem de urgência recente, os líderes mundiais não podem mais negar o intervalo cada vez mais curto da frequência de desastres naturais que assolam o mundo. Sejam queimadas na Austrália, secas no Pantanal, furacões nos Estados Unidos ou tempestades na Itália. Segundo o Atlas da Mortalidade e das Perdas Econômicas por Extremos Climáticos, Hídricos e do Tempo, publicado pela Organização Meteorológica Mundial (WMO, 2021), entre 1970 e 2019, foram registrados mais de 11 mil desastres do tipo pelo mundo, que causaram mais de 2 milhões de mortes e geraram perdas de US$ 3,64 trilhões. O documento é o mais abrangente já feito e segundo a WMO, 91% dessas mortes por eventos climáticos ocorreram em países em desenvolvimento. Até aqueles que defendiam prioritariamente uma visão economicista, viram nos últimos anos seus dogmas estremecidos com os custos econômicos que esses eventos climáticos trouxeram. As mudanças climáticas tendem a causar impactos financeiros e uma pressão sobre o modelo de negócio carbono-intensivo, e a discussão da busca pela economia de baixo carbono não poupa riscos a serem enfrentados.
... A finanszírozók és befektetők irányából érkező nyomás a környezeti, szociális és irányítási (environmental, social and governance, ESG) tényezők figyelembevételére jelentős hatással van a vállalati pénzáramokra, és jelentősen megváltozó működési-finanszírozási környezetet eredményez. (Friede-Busch-Bassen, 2015). ...
Az 1990-es évek rendszerváltó nemzedéke napjainkra elérte a nyugdíjas kort, aktuálissá vált a generációváltás. A családi vállalkozások utódlásával foglalkozó kutatások a vállalati belső faktorok – professzionalizálódás, külső menedzser bevonása, utódok szocializációja, a vagyonérték problémája – vizsgálatára fókuszálnak elsősorban. Azonban az utódlási lehetőségeket meghatározó intézményi környezetnek – az öröklési szabályoknak, a formális közvetítőrendszernek, az utódláshoz kapcsolódó adózási szabályoknak – is jelentős hatása van a családi vállalkozások hosszú távú fennmaradására és a generációk közötti vagyonátadásra. Jelen tanulmány célja, hogy rávilágítson a hazai családi vállalkozások sajátosságaira, az utódlási alternatívákat meghatározó belső vállalati tényezőkre. Bemutatja a formális intézményrendszert, valamint az intézményrendszer és a transzgenerációs vagyonátadás, az öröklési stratégia kapcsolatát. A vegyes kilépési alternatívák esetében a családi vállalkozás tulajdonosai és a menedzserei közé ékelődő „vagyonkezelői ékek” közül a bizalmi vagyonkezelés alkalmazásának a lehetőségeit, előnyeit és kockázatait elemezzük.
... Big data, fast access to the market and information and technological developments simplify and shorten market penetration channels and times. Pressure by financiers and investors to consider environmental, social and governance (ESG) factors has a major impact on company cashflows and results in an operational-financing environment significantly changed (Friede-Busch-Bassen, 2015). ...
The 1990s generation of the fall of communism has reached retirement age by now, so the time has come for a generational transition. Studies on the succession of family enterprises mainly focus on internal factors such as professionalisation, involvement of external managers, successors’ socialisation, or the issue of asset values. However, the institutional environment shaping succession options including inheritance law, formal intermediaries, succession-related tax rules also have a major impact on the long-term survival of family businesses and the transfer of wealth from one generation to the next. The objective of this paper is to shed light on the peculiar features of family enterprises and their internal factors defining succession alternatives in this country. It intends to describe the formal set of institutions and the relationship between those institutions, transgenerational wealth transfer, and succession strategies. In the case of mixed exit strategies, it analyses the options, advantages, and risks of the application of fiduciary asset management as one of the “wedges of asset management” between the principals of family enterprises and the managers.
... Future research could also consider the outcomes of employee pro-ESG behaviors. A first area to explore is whether they are related to organization-level ESG ratings and financial performance (Friede et al., 2015). Other areas worth examining include employees' perceptions of their firm's external prestige (Kim, Lee, Lee, and Kim, 2010) and employee engagement (Khan, 1990). ...
Purpose. With the increasing importance of environmental, social and governance (ESG) reporting, organizations are seeking ways to improve ESG performance. One of these ways is encouraging employee behaviors that support ESG. While some research has examined employee behaviors that support specific components of ESG, such as pro-environment behavior, this research is fragmented. Thus, the purpose of this paper is to review the various ESG outcomes which employees can influence, and then propose a single scale to measure employee support for ESG, along with recommendations for future research. Methodology. The authors reviewed ESG reports from several large, multinational companies, as well as previous research on ESG and corporate social responsibility. The authors sought to identify ESG outcomes which employees can influence, and then develop a scale based upon these areas. Findings. The authors identify several areas within ESG which employees can influence. These include areas such as support for volunteering, employee giving campaigns, diversity and inclusion, environmental initiatives, data governance and others. Based upon these findings, the authors developed the above-mentioned scale. Originality. While previous studies have examined employee support for specific components of ESG, this paper offers a comprehensive framework. In particular, there has been little research focused on employee support for governance behaviors, such as data stewardship. Study implications. This paper provides a scale which can be used to measure employee support for ESG behaviors. This scale can help employers understand their employees’ level of support for ESG. In addition, it can help researchers understand whether employee support for ESG is related to organization-level outcomes, such as ESG ratings.
This study examines the role of corporate social responsibility (CSR) in the investment preferences of tour operators through a discrete choice experiment conducted among tour operator managers worldwide. Stakeholder theory is used as a theoretical platform for explaining the role of CSR within the tour operators’ investment preferences. The findings indicate that, when making investment decisions, tour operators generally tend to balance the interests of the local community, employees and businesses, and to consider the effects of their investments on the local economy and the environment. However, empirical evidence indicates that tour operator’s investment preferences are moderated by three factors, namely: local government pressure, size of the investment, and tour operator profile. In particular, greater attention should be paid to high-scale investments, and to investments made by generalist tour operators if destinations want to preserve their distinctive sociocultural and natural assets and provide well-being to local communities.
Full-text available
With the growing number of environmental, social, and governance (ESG) problems, many companies have begun to implement more sustainable business practices. In the midst of this change, institutional shareholders declare and adopt socially responsible investment procedures, which is a way of engaging in investor activism. Despite the growing interest in investor activism following the introduction of the stewardship code, little attention has been paid to how socially responsible investment practices of institutional investors affect the non-financial value of the pillars of environmental, social, and governance as well as financial performance, including short-term accounting (ROE, ROA) and long-term market performance (Tobin q). The current study examines whether the national pension fund (NPF), the world’s third-largest Korean pension fund, can increase the ESG performance of investee firms in addition to accounting and market performance through institutional investors’ shareholding. This study, by applying path analysis, attempts to explore the relationship between the NPF’s socially responsible investing, ESG, and the financial performance of the investee firms. This research offers evidence that ESG performance acts as a moderator or a mediator between NPF’s shareholding and financial performance.
Full-text available
One of the roles of accounting is to provide information on business performance, either through financial accounting indicators or otherwise. Theoretical-empirical studies on the relationship between Corporate Financial Performance (CFP) and Corporate Social Performance (CSP) have increased in recent years, indicating the development of this research field. However, the contribution to the theory by empirical studies is made in an incremental manner, given that each study normally focuses on a particular aspect of the theory. Therefore, it is periodically necessary to conduct an analysis to evaluate how the aggregation of empirical studies has contributed to the evolution of the theory. Designing such an analysis was the objective of the present study. The theoretical framework covered the following: stakeholder theory, the relationship between CSP and CFP, good management theory, and slack resource theory. This research covered a 15-year period (1996 to 2010), and the data collection employed a search tool for the following databases: Ebsco, Proquest, and ISI. The sampling process obtained a set of 58 exclusively theoretical-empirical and quantitative articles that test the CSP-CFP relationship. The main results in the theoretical field reinforce the proposed positive relationship between CSP and CFP and good management theory and demonstrate a deficiency in the explanation of the temporal lag in the causal relationship between CSP and CFP as well as deficiencies in the description of the CSP construct. These results suggest future studies to research the temporal lag in the causal relationship between CSP and CFP and the possible reasons that the positive association between CSP and CFP has not been assumed in some empirical studies.
Full-text available
This article describes 35 years of academic research into investment practices that in some way integrate a consideration of environmental, social and corporate governance issues. A review of 190 academic papers was undertaken to identify trends in five domains, namely 'Primary Name', 'Research Themes', 'Ethical Foundations', 'Research Approach' and 'SRI Strategies'. The evidence reveals that more than half the researchers refer to such investment practices as Socially Responsible Investing (SRI) and for this reason the name is used in this review as a generic term for the genre. A myriad of other names were also identified. In terms of research themes, one particularly dominant theme was that of financial performance, which was often discussed in relation to fiduciary responsibility and legal aspects. Although the primary ethical foundation was not always directly observable, the majority of papers implied utilitarianism or 'the greatest good for the greatest number'. Increased mention of ethical egoism (self-interest) is observed in later periods. An equal split between qualitative and quantitative research methodologies was noted, with a qualitative approach being more favoured in recent years. Three SRI strategies have dominated academic discussions over the past 35 years, namely negative screening, positive screening and shareholder activism. Gaps in the literature have been identified and suggestions for future research made.
Full-text available
As the number of scientific studies continues to grow, it becomes increasingly important to integrate the results from these studies. One simple approach involves counting votes. In the conventional vote-counting procedure, one simply divides studies into three categories: those with significant positive results, those with significant negative results, and those with nonsignificant results. The category containing the most studies is declared the winner. For example, if the majority of studies examining a treatment found significant positive results, then the treatment is considered to have a positive effect. Many authors consider the conventional vote-counting procedure to be crude, flawed, and worthless (see Friedman 2001; Jewell and McCourt 2000; Lee and Bryk 1989; Mann 1994; Rafaeli-Mor and Steinberg 2002; Saroglou 2002; Warner 2001). Take, for example, the title of one article: "Why Vote-Count Reviews Don't Count" (Friedman 2001). We agree that the conventional votecounting procedure can be described in these ways. But all vote-counting procedures are not created equal. The vote-counting procedures described in this chapter are far more sophisticated than the conventional procedure. These more sophisticated procedures can have an important place in the meta-analyst's toolbox. When authors use both vote-counting procedures and effect size procedures with the same data set, they quickly discover that vote-counting procedures are less powerful (see Dochy et al. 2003; Jewell and McCourt 2000; Saroglou 2002). However, vote-counting procedures should never be used as a substitute for effect size procedures. Research synthesists generally have access to four types of information from studies: • a reported effect size, • information that can be used to compute an effect size estimate (for example, raw data, means and standard deviations, test statistic values), • information about whether the hypothesis test found a statistically significant relationship between the independent and dependent variables, and the direction of that relationship (for example, a significant positive mean difference), and • information about only the direction of the relationship between the independent and dependent variables (for example, a positive1 mean difference). These types are rank ordered from most to least in terms of the amount of information they contain (Hedges 1986).2 Effect size procedures should be used for the studies that contain enough information to compute an effect size estimate (see chapters 12 and 13, this volume). Vote-counting procedures should be used for the studies that do not contain enough information to compute an effect size estimate but do contain information about the direction and the statistical significance of results, or that contain just the direction of results. We recommend that vote-counting procedures never be used alone unless none of the studies contain enough information to compute an effect size estimate. Rather, vote-counting procedures should be used in conjunction with effect size procedures. As we describe in section 11.4, effect size estimates and votecount estimates can be combined to obtain a more precise overall estimate of the population effect size.
This paper provides a theory of private politics in which an activist seeks to change the production practices of a firm for the purpose of redistribution to those whose interests it supports. The source of the activists's influence is the possibility of support for its cause by the public. The paper also addresses the issue of corporate social responsibility by distinguishing among corporate redistribution as motivated by profit maximization, altruism, and threats by the activist. Private politics and corporate social responsibility not only have a direct effect on the costs of the firm, but also have a strategic effect by altering the competitive positions of firms in an industry. From an integrated-strategy perspective the paper investigates the strategic implications of private politics and corporate social responsibility for the strategies of rival firms when one or both are targets of an activist campaign. Implications for empirical analysis are derived from the theory.