Master Thesis MSc Finance
University of Groningen / Faculty of Economics and Business TKP Investments
An Analysis of ESG Integration by Mainstream Mutual Fund Managers
Abstract
This study analyzes whether and how conventional mutual equity funds integrate environmental, social and governance (ESG) factors into their investment process. This is investigated for 126 funds that are under contract or under research with TKP Investments. I find that the majority of funds have ESG data and staff at their disposal, but they only use it when these factors are likely to have a material influence on financial returns. Usually this is in the case of managing downside risk, translating into avoiding the worst ESG performers. A cross-section regression is used to find out whether high ESG funds are better able to beat their benchmark index than low ESG funds over a 1 year and 3 year period. Indicative evidence is found that high ESG funds performed worse over 2011, but better over the longer 3 year period 2009-2011. So although literature points out that various ESG factors, especially governance, influence returns, there is no unambiguous evidence that this information can be used ex-ante to improve the risk-return profile of an investment portfolio.
JEL Classifications: G11, G12, G23, M14
Keywords: environmental, social and governance (ESG) factors, extra-financial analysis, portfolio management, mutual fund performance
Author: Emiel van Duuren Date: 14-02-2013
Student number: s1684019
First supervisor RUG: Dr. A. Plantinga
Introduction
Over the last decade the responsible investment industry has grown significantly. This is reflected by the growing number of signatories of the UN Principles of Responsible Investment, which has more than doubled over the last 3 years from 500 to more than 1000. One of the principles is ‘the incorporation of ESG (environmental, social and governance) issues into investment analysis and decision-making’. Eurosif, the pan-European network of investors, academic institutes and research associations investigates and stimulates sustainable investing. According to their 2010 SRI Study, ESG integration is the number one strategy in the context of mainstreaming socially responsible investing (SRI), with €3.2 trillion of assets at the end of 2011.
Eurosif describes ESG integration as ‘the explicit inclusion of ESG risk factors into traditional financial analysis’. They acknowledge that the concept of ESG integration remains a challenge to pin down. In this study I argue that ESG integration is fundamentally different from traditional SRI. Traditional SRI relates to funds in a niche that is driven by moral values and mainly comprises exclusionary screening. ESG integration on the other hand, targets mainstream investing. The idea is that the financial performance of a company should not only be predicted by financial indicators, but also by non-financial ESG indicators. Several event studies show abnormal returns from ESG related events or measures, but whether this information really is and can be used in the ex-ante construction of a portfolio remains to be seen. So although many asset managers signed the UN PRI, it remains unclear whether they really share this view and consider ESG factors when a portfolio is constructed, or that they are window-dressing. This is also of academic relevance, because numerous studies have been done on traditional SRI, but the literature on the use of ESG information in mainstream funds is scarce. Is this because mainstream fund managers do not really use information about a company’s social performance, environmental issues and governance structure in their investment decisions?
This lack of insight is also faced by TKP Investments, an asset manager for Dutch pension funds with approximately €16 billion assets under management. The company invests the means of pension funds by selecting external mutual fund managers. The restriction that TKP Investments imposes on these external managers with respect to ESG issues is that they should not invest in companies that do not comply with the UN Global Compact and/or produce controversial weapons. Besides acting in compliance with these minimum standards, the external managers might not consider ESG factors at all or they might have a formal policy and process to integrate ESG factors. It is of particular interest for TKP Investments to know where fund managers are on this continuum. This study addresses this topic by answering the following research question:
I investigate this for 126 mutual equity fund managers that are under contract or under research at TKP Investments. They manage conventional global and region-specific equity funds. Data is collected through a questionnaire in which fund managers are asked about whether, and if so how, ESG factors influence their investment decisions. Based on the answers to the questions, the managers are sorted into 4 categories from ESG-1 to ESG-4. The classification ESG-1 indicates that ESG factors play no role at all for the particular fund, and ESG-4 indicates that ESG factors are systematically integrated into the investment process. It is important to note that this is not a value judgment. A high scoring fund does not mean that the fund is better or more responsible compared to other funds, it only indicates that the fund makes more use of ESG information.
Although corporate social responsibility and reputation is important for TKP Investments, their fiduciary responsibility towards their clients is to maximize the value of their pensions. This also holds for the fund managers that TKP Investments assigns. They have to generate outperformance compared to their benchmark index. Whether the integration of ESG factors into the investment process positively influences this outperformance is not clear. Several companies with a commercial interest in the industry, for example MSCI, like to suggest this relation. However, these claims are often not supported by empirical evidence, because most academic studies focus on traditional SRI and/or the relationship between social responsibility and financial performance on company level. So the second research question that will be addressed is:
Is ESG integration positively related to outperformance?
To answer this question and test whether a high degree of ESG integration is associated with more outperformance, I perform a cross-section ordinary-least-squares (OLS) regression analysis. I regress the ESG-category from the first analysis against 1 and 3 year annualized outperformance. Since the ESG-classification is an ordinal variable, dummies are used. This makes the regression multivariate with 3 (the number of categories minus 1) predicting variables.
So this study is a first attempt to describe the phenomenon of ESG integration from the perspective of mainstream mutual equity funds. It describes policies, practices, characteristics and the consequences for returns. Thereby it should serve as a starting point for further research on the subject.
The remainder of this study is organized as follows. Section I gives an overview of the
Table of Contents I. Literature Review ... 4 A. ESG Integration ... 4 B. Performance ... 7 B.1. Company Level ... 7 B.2. Fund Level ... 9 II. Methodology ... 12 A. Questionnaire ... 12 B. ESG Score ... 13 C. Outperformance ... 15 III. Data ... 17 IV. Results ... 19 A. ESG Integration ... 19 B. Outperformance ... 23
V. Conclusions and Discussion ... 28
VI. Limitations and Future Research ... 30
VII. Recommendations for TKP Investments ... 31
References ... 34
4 I. Literature Review
The literature review consists of two parts. The first part focuses on whether and how
investors make use of ESG information. This gives an indicative answer to my first research question and provides useful input for the questionnaire I construct (see Data Section III). Secondly, I give a brief overview of the existing literature concerning the effect of ESG-integration on stock and fund returns.
A. ESG Integration
Eurosif describes ESG integration as the explicit inclusion of ESG risk factors into traditional financial analysis. Socially Responsible Investment on the other hand is defined as ‘the exercise of ethical and social criteria in the selection and management of investment portfolios’ (Cowton, 1994). SRI relates to funds in a niche that is driven by ethical values and mainly comprises exclusionary screening. ESG integration on the other hand is solely financially oriented and therefore applies to mainstream investing.
According to Modern Portfolio Theory investors are fully rational and stock prices fully reflect available information (Fama, 1970). This suggests that pricing models of investors fully and correctly incorporate all relevant information, including ESG information. However, ESG information is hard to quantify.
In a study of how Swedish investors make investment decisions, Hellman (2000) indeed found that investment decisions are almost exclusively based on financial ratios like return on equity, sales growth and price/earnings ratios. Intangible factors, like ESG issues, are hard to translate into monetary terms, and therefore difficult to incorporate into investment models. Henningsson (2008), who also held in-depth interviews with Swedish investors, found that fund managers did not think that social, environmental and ethical information can explain stock prices or the value of companies. Therefore it is considered as something that has to be ‘ticked-off’, that is to exclude the worst offenders from an ethical point of view, before the financial indicators can be analyzed.
5 Corporate Engagement, confirms these findings. They question 328 investment professionals about their use of extra-financial information in their analyses, valuations and investment decisions. Their findings also indicate that the use of ESG information in mainstream investing is very limited. Here the reasons mentioned are the limited provision of information by the companies they invest in, the lack of a universally accepted methodology to quantify ESG data, and time pressure resulting in skipping the ESG analysis. Furthermore, the majority of the analysts think that ESG factors have more influence on brand and reputation than on financial performance and market value. A global survey of McKinsey (2008) among investment professionals finds the same lack of knowledge about how to value ESG performance. A significant proportion of the investors think that ESG issues are too indirect to measure or too long-term. Several authors with a qualitative approach take the view that fund managers form a social network. Their investment behavior is argued to be determined by industry standards and organizational codes. According to Orléan (2004) investor decision making is influenced by collective beliefs. Collective beliefs are beliefs of which each individual thinks that the majority of the group has this belief. It does not relate to the primary beliefs of individuals. In this light, Louche, Bourghelle and Jemel (2009) state that dominant collective beliefs obstruct the adoption of new practices, like the integration of ESG information. They argue that investors face uncertainty and this leads them to adopt a conventional model. Discounted cash flow models or mean variance portfolio optimizations are such conventional models. In these models there is little room for ESG factors. Fuchs (2001) and Henningsson (2008) found similar results. Both studies consider fund managers as part of a social network. Corporate social information is often taken care of elsewhere in the organization. It only marginally reaches the social networks surrounding fund managers. So these networks are basically a barrier, or an immune system like Fuchs describes it, towards the integration of ESG information.
6 a higher risk premium in their valuation model to companies with ESG risks. On the other hand, companies with a good ESG record do not receive a risk discount. Henningsson (2008) also finds that investors see ESG as a reputation risk. Investors do not care, from a financial perspective, about ESG performance, as long as a company meets a certain minimum level.
A final point of interest is whether there is a difference between the use of environmental (E), social (S) and governance (G) information. According to Jaworski’s (2007) survey, corporate
governance is the most important factor to investors. Respondents claim that this factor has a much more direct effect on the financial performance of a company than other ESG factors. That is because a sound governance policy results in effective managerial decision making and prevents opportunistic behavior. The majority of the respondents in general do not think that environmental and social factors influence financial performance. These factors are only considered important in certain industries. The environment is only relevant for environmentally sensitive industries like automobiles, utilities and mining. Social performance in the form of human rights is only considered to be important for consumer goods. Henningsson’s (2008) interviewees from Sweden confirm these findings. The management and their ability to run a profitable company are considered important, but social aspects not. Eccles, Krzus and Serafeim (2011) take a unique approach when trying to measure the market interest in non-financial information. They gained access to the search results of Bloomberg. By measuring the number of hits for different ESG factors, they assess the relative importance of those factors. This method is flawed, because investors make use of more sources than only Bloomberg for example. However, the paper does deliver some valuable insights when ignoring these flaws. The authors find that the interest in environmental and governance information is much greater than the interest in social information. They think that this is because environmental performance is more measurable than social information, and governance information is known for its influence on financial performance. More specifically, there is major interest in board composition and board activity within the governance category. For the environment category, greenhouse gas and CO2 emissions data receive much interest, except for the US. Another important conclusion from their work is that there is a high level of interest in the degree of transparency around companies’ CSR performance. It is suggested, but not investigated, that fund managers think this is a determinant of risk, where there is less uncertainty about the performance of transparent companies.
In conclusion, ESG integration can be defined as taking environmental, social and governance factors into account in the investment process. There is evidence that these factors affect the
7 Figure 1. Schaltegger and Wagner’s theory on the link between environmental and social performance and economic success.
B. Performance
In this second part of the literature review I focus on the relationship between ESG integration and fund performance. Firstly, I discuss a theoretical framework on the influence of environmental, social and governance performance, which relates to CSR, on financial performance at the company level. Then it is of interest whether investors can benefit from this knowledge. In other words, whether it helps them in generating (abnormal) returns.
B.1. Company Level
A general theoretical framework that describes the relation between environmental and social performance on the one hand and financial performance on the other hand is developed by Schaltegger and Wagner (2006). They distinguish between the ‘traditionalist view’ and the ‘revisionist view’. Both views are displayed below in Figure 1. According to the common traditionalist view, good
environmental and social performance intends to correct for market externalities. Internalizing these externalities leads to an increase in costs for the company. Therefore these factors reduce financial performance, economic success in Figure 1. In contrast, the revisionist view sees environmental and social performance as a potential source of
competitive advantage, because it can improve business processes, exploit market
opportunities, avoid fines and create goodwill with stakeholders. Therefore the economic success goes up, but only to a limited extent, because when a company tries to address all environmental and social factors this will lead to decreasing marginal benefits and eventually to net costs.
8 In contrast to Schaltegger and Wagner’s theoretical framework, there are numerous papers that do test the link between specific environmental and social performance measures and stock returns empirically. A significant amount of research has been done on this subject. Metastudies of Margolis and Walsh (2001, 2003), Orlitzky et al. (2003) and Margolis, Elfenbein and Walsh (2009) document a minor but significant positive effect. This effect is more present for accounting based measures than for stock returns. Margolis, Elfenbein and Walsh (2009) incorporate several contingencies in their regressions and find that revelaed misdeeds have a highly negative effect on stock returns, thereby supporting the notion that doing bad has a more pronounced effect than doing good.
Several papers focus on the influence of one E, S, or G factor on stock returns. For example Klassen and McLaughlin (1996) do this for the environmental factor, Edmans (2011) for the social factor, and Gompers, Ishii and Metrick (2003) for the governance factor. Klassen and McLaughlin perform an event study that finds significant positive stock returns for companies after receiving an environmental award for good environmental management. On the other hand, significant negative returns are reported for weak environmental management, as measured by environmental crises. The limitation of the first result is that it is not a general result with respect to good environmental
management, it only refers to companies that receive an award. The second result, the negative returns for companies with poor environmental performance is in line with the view that companies should meet a minimum standard with respect to environmental performance, otherwise it negatively influences the financial performance. Edmans studies the effect of the social factor employee
satisfaction on long-run stock performance. He shows that a portfolio of 100 stocks of companies that were labeled ‘Best to work for in America’ by Fortune magazine significantly outperformed the broader market from 1998 to 2005 after correcting for the market, size, book-to-market and
9
B.2. Fund Level
It is important to note that it is not correct to derive the effectiveness of ESG integration by comparing the performance of socially responsible funds (SRI) with the performance of conventional funds. According to Hirschberger et al. (2011) socially responsible investing is just negative screening. By excluding certain companies or industries the investment universe is limited, but after this screen the investment analysis does not include the use of ESG information. Some SRI funds even have lower average ESG scores than their conventional counterparts. Companies that are often excluded, like those in the alcohol or tobacco industry may have excellent environmental, social and governance records.
Comparing aggregated ESG best-in-class strategies with conventional investment strategies is not appropriate either. This is because best-in-class ESG funds select the best performing companies on all environmental, social and governance factors. It is unknown which of these factors have what influence on stock returns. The best-in-class principle does not make this distinction. For example, imagine a best-in-class fund that only selects companies that have a high degree of equality among the workforce (social factor) and a high degree of board diversity (governance factor). Suppose that the relationship between a high degree of equality among the workforce and financial performance is negative, and the relationship between a high degree of board diversity and company performance is positive. Then these effects might cancel each other out. Then this fund has the same performance as a similar fund that does not apply this best-in-class strategy. This would lead to conclusion that ESG factors do not influence stock returns, which would be incorrect. So when funds integrate ESG factors into their investment process, it is necessary to analyze the separate effects of every individual factor on stock returns.
Galema, Plantinga and Scholtens (2008) acknowledge this point by stating that the aggregate analysis of ESG may eliminate a relationship if individual ESG factors have opposite effects on performance. They test this hypothesis by performing Fama and MacBeth regressions, but find little evidence. The authors find that individual ESG factors do influence stock returns, but not by
10 Manescu (2011) and Humphrey, Lee and Shen (2012) find similar results concerning the effect of ESG factors on stock returns for the US and UK market, respectively. Manescu uses a long panel of publicly traded US firms from 1992 until 2008 to test the separate effects of seven ESG factors. He finds that six of the seven measures do not impact stock returns significantly. Only community relations has a significant positive effect. Humphrey, Lee and Shen use environmental, social and governance rankings to test whether there is a difference in performance between high and low ranked stocks. They do not find any differences in performance, nor do they find any differences with respect to their market, book-to-market, or momentum exposures in Carhart’s (1997) four-factor model. This is in contrast to the previously mentioned Galema, Plantinga and Scholtens (2008) that find that high ESG scoring stocks have a lower book-to-market ratio.
Scholtens and Zhou (2008) do not find unambiguous evidence either. With a panel fixed effects regression they investigate the separate effects of different ESG strengths and concerns, namely community involvement, corporate governance, employee relations, environmental conduct, diversity of the workforce, human rights policies and products attributes for 289 US companies over 14 years. However, they do not find a positive association between any social strength and stock returns. They actually find a weak negative relation. The authors do find that financial risk is associated with social concerns, leading to the conclusion that financial risk is primarily related to social controversies and not to strengths. This is in line with the previously mentioned findings of Henningsson (2008). Scholtens and Zhou opt for a theory that describes an inverted u-shaped relation between social performance and financial performance. This is basically the relationship that the previously mentioned Schaltegger and Wagner (2006) developed.
The lack of evidence in the above mentioned papers that ESG integration leads to
outperformance seems in contradiction with the meta-analyses of Orlitzky et al. (2003) and Margolis and Elfenbein (2007). These studies both find a minor, but significant positive effect between corporate social performance and corporate financial performance. However, this effect is more present for accounting measures of financial performance than stock returns. Furthermore, they do not provide unambiguous evidence of causality. Margolis and Elfenbein acknowledge that the effect of corporate financial performance on corporate social performance is stronger than vice versa. Bénabou and Tirole describe the view that socially responsible investors are long-term investors. They question whether financial markets are still learning about CSR. There might be a recognition period in which environmental and social factors are gradually becoming recognized as relevant price factors.
In conclusion, the evidence that ESG integration leads to outperformance is weak. Some studies that do find a positive association between ESG and financial performance fail to prove
11 company. This is in agreement with investor views from the first part of the literature review.
12 II. Methodology
A. Questionnaire
In order to assess whether and how much fund managers make use of ESG information in their investment process, an online questionnaire is used. The answers contribute to a qualitative answer to the first research question and inform TKP Investments on the ESG practices of individual managers by giving them an ESG score. In this paragraph I motivate the characteristics and content of the questionnaire. Print-screens of the full questionnaire are in Appendix A.
An online-questionnaire is the fasted, cheapest and most convenient way to address a large group of subjects. Furthermore, results can be analyzed and generalized relatively easily. The most important disadvantage of a questionnaire for this study is that respondents might not give an honest answer. ESG integration is often related to socially responsible investing, and most asset managers would like to associate themselves with it, so window-dressing is a serious issue. This means that the asset managers will always be inclined to give answers that indicate the use of ESG information. However, in the introductory text I make a number of remarks that make an honest answer more likely. The notion that I am a student writing a thesis might create goodwill. Furthermore, from TKP Investments I emphasize that the main goal for fund managers is generating outperformance, not responsible investing per se. The purpose of this survey is to assess whether and how ESG factors have a part in this. To further limit the social desirability bias, in the questions no reference is made to the words ‘ethics’ and ‘social responsibility’. Finally, TKP Investments has detailed insight into the holdings of the funds under contract, so when these fund managers state for example that they exclude poor ESG scoring companies from their portfolio, this can be checked.
I use a combination of open and closed questions, because both have their advantages and disadvantages. Closed questions are more likely to lead to an honest answer (reduces the possibility of window-dressing), are easier to answer and are better comparable, but they can suggest ideas that the respondents might not have otherwise and can force the respondents to give simplistic responses to a complex concept like ESG integration. Open questions on the other hand do offer the possibility to give detailed and unanticipated answers, but these answers are hard to handle statistically and give the respondents the opportunity to give vague and evasive answers. I use both type of questions to capture the advantages of both. The answers on the open questions contribute to a qualitative answer to the first research question. The closed questions are used to give the funds an ESG score, which is also used to answer the second research question.
13 Management. Finally, I consulted experts from TKP Investments and Sustainalytics about how ESG factors can be used in an investment process.
Most important is what fund managers do with respect to the use of ESG information. I first ask them to describe their investment process in general and thereafter whether and how they integrate ESG in it. This way I force the respondents to think about in which phase ESG comes in. Fund
managers make use of several techniques, for example discounted cash flow analysis and financial ratios. If a fund manager claims it integrates ESG factors into their process, than these factors should be part of these techniques. I ask whether it is used to determine the universe, whether it is used in the valuation of companies, or whether it is used to manage downside risks. Also, through a multiple choice question I ask whether they have separate investment guidelines for environmental, social and governance factors in order to find out if one of the ESG factors is analyzed more frequently.
Furthermore, with the idea that what gets measured gets managed, attention is paid to the ESG data, systems and staff that the fund has at its disposal. The more data available, the more likely it is that the data is used in buy and sell decision regarding stocks. The same holds for ESG staff.
Also, there are questions included that ask for concrete examples that indicate that
environmental, social and/or governance issues have affected the position in a stock. If ESG factors are really considered important by fund managers, then it is likely that in a year at least some stocks have been bought or sold (partly) because of ESG related reasons.
B. ESG Score
The questionnaire contains questions on three indicators of ESG integration: ESG methods, ESG data availability and questions about ESG training and staff. The closed questions are
checkboxes. In the process of generating an ESG score, they are treated as dummies, 1 if applicable and 0 if not. For example, one of the questions is whether a fund makes use of ESG ratings. If this is the case, the respondent ticks that box so he scores 1 on this question. The sum of all dummy values leads to an aggregated ESG score. If all three components measure different aspects of ESG
14 Since not all questions are equally important and have the same number of checkboxes, the different topics are weighted. I asked the respondents their opinion about the relative importance of these topics in deciding what signifies ESG integration. However, all issues were considered quite important (they scored around 4 on a scale from 1 to 5), so most items are included. Whether they apply ESG methods is most important, so this topic is weighted more heavily with three-fifth. Data and staff availability are also indications of ESG integration, but to a lesser extent, so one-fifth each. This weighting matches conveniently with the amount of questions asked and is also similar that of the Global Real Estate Sustainability Benchmark (GRESB), which constructed a similar scoring system for real estate funds. Still, this scoring method is subjective and should only be seen as a rough measure. The assignment of points is shown in Figure 2.
Based on their score the funds are sorted into the categories 1, 2, 3 and ESG-4. That is, from a low to a high degree of ESG integration. Four categories because this corresponds to TKP Investments’ fund rating system which also ranges from 1 to 4. Furthermore, by limiting the number of categories, the assignment of categories is less influenced by the subjectivity of the scoring method. A few points do not change in which ESG-category a fund is sorted, so there is less noise.
Figure 2: ESG Scorecard Methodology Table I: Correlationmatrix ESG indicators
Tactics Staff and Training Data
Tactics Correlation 1.000 ***0.418 ***0.466
P-value 0.000 0.000 0.000
Staff and Training Correlation ***0.418 1.000 ***0.641
P-value 0.000 0.000 0.000
Data Correlation ***0.466 ***0.641 1.000
P-value 0.000 0.000 0.000
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
This table shows the correlations (Pearson) between the ESG indicators Tactics, Data, and Staf and Training.
Tactics Staff & Training Data
Points Points Points
Use ESG info in the valuation of companies 5 Specialized staff 5 Ratings 2
Use ESG info to manage risks 5 Continuous training 1 Analysis at company level 2
Red flagging (Monitoring major controversies) 5 Staff meetings 1 Analysis at sector level 2 Detailed instructions on environmental issues 5 News letters 1 Analysis at country level 2
Detailed instructions on social issues 5 Workshop(s) 1 Analysis of specific ESG issues 2
Detailed instructions on governance issues 5 Part of intro program 1 10
15
C. Outperformance
The second research question to be answered relates to the financial performance of the funds. More specifically, whether a high degree of ESG integration is associated with more outperformance. To test this I make use of a cross-section ordinary-least-squares (OLS) regression.
The dependent variable is outperformance, which is the difference between the return of a particular fund (excluding fees) and the return on its benchmark index. This variable is chosen because it is an important financial indicator for TKP Investments in the selection and monitoring of fund managers. TKP Investments determines the asset and region allocation, and thereby the benchmark. The main goal of the appointed third-party managers is to beat this benchmark. I use 1 year and 3 year annualized outperformances, because it is reasonable to assume that ESG policies have changed compared to earlier periods. The assumption that policies haven’t changed over the last 3 years is checked by a question in the questionnaire. In symbols, the outperformance is Ri-Rb, where Ri is the fund return and Rb is the return of the benchmark index. Since return data is not publicly available for all funds, I ask for this in the questionnaire. The respondents might inflate their performance numbers, but I consider this as unlikely, because this might have legal consequences.
The outperformance is regressed against the ESG-category from the first analysis, this is the independent variable. A fund in Category 3 makes more systematically use of ESG information than a fund in Category 2, but it is not possible to say how much more this is. Therefore I consider the ESG-category as an ordinal variable. To deal with the ordinal nature of this variable in a regression, I make use of dummies. ESG-2, ESG-3 and ESG-4 are represented by the dummies D1, D2 and D2. They take the value of 1 if the fund is in the particular category and 0 otherwise. To avoid the dummy trap, there is no dummy for ESG-1. The effect of this category is analyzed when all three dummies are zero. Equation 1 shows the resulting regression.
(Equation 1)
The question of interest is whether ESG integration is positively related to outperformance. Therefore I test the following hypotheses:
(1) H1: The relationship between ESG-category and 1 year outperformance is significantly positive
16 A significant dummy contributes to explaining the variance in outperformance. The
coefficients y1, y2, and y3 measure the difference in outperformance of the categories ESG-2, ESG-3 and ESG-4 compared to the ESG-1 group. The expectation that corresponds with the hypotheses is that the coefficients are of increasing magnitude.
Several factors might affect the regression, so these are added as control variables. These are fund size, manager tenure and fund age. These variables are measured by assets under management in euros, the average age of the two most experienced portfolio managers, and years, respectively. Furthermore Ferreira et al. (2011) argue that risk adjusted fund returns are higher in countries with strong legal institutions and stock markets. All managing companies are located in developed countries, but the financial markets in the UK and US are most liquid and developed, therefore I use an Anglo-Saxon dummy that is 1 when the asset manager is from the UK or US and 0 otherwise. Data on these control variables is from Mercer’s Global Investment Manager Database and Mercer Insight.
If there are no significant relationships found between ESG-category and outperformance, this could mean two things. The obvious conclusion is that this relationship does not exist. The other possibility is that ESG integration does not yield a higher 1 or 3 year outperformance, but it might lead to outperformance over longer terms because then ESG issues become financially material.
17 III. Data
The questionnaire was sent to 251 fund managers and their team of which 14 are under contract and 237 under research at TKP Investments. Initially 83 funds filled out the questionnaire. After a follow-up email this number increased to 126. So the overall response rate was roughly 50%, which is satisfying. More specifically, all 14 funds under contract responded (100%) and 112 funds under research responded (47,3%). Appendix B contains the complete list of respondents and their ESG classification. In total, 35 funds are labeled ESG-1, 35 funds ESG-2, 36 funds ESG-3, and 20 funds are sorted in the highest category ESG-4. XXXXX was unable to fill out the questionnaire on time, but they were willing to provide their view on ESG by phone. This gave me the opportunity to have a more detailed discussion on the subject, see also Results Section IV.
Summary statistics about the sample are presented in Table II. Notice that not all data on the control variables was available for the 126 funds. Although performance numbers were available for all funds, I excluded 35 funds for the regression of the outperformance over 2011, and 45 funds for the regression of the outperformance over 2009. The funds that were excluded, were funds for which the current ESG approach was not yet in place during the periods under attention (2011 and 2009-2011, respectively). Table II shows that on average, the funds in our sample show no substantial under- or outperformance over 2011 compared to their benchmark. Both the mean and median are close to 0. On the other hand, over the longer 3 year period from 2009 until 2011, the funds outperformed their benchmark with an average of 2.3 % and a median of 1.2% per year. The sample contains funds of varies sizes, with assets under management ranging from 10 million to 28.9 billion euro, and ages, ranging from 2 years to 31 years of existence. The average industry experience of the portfolio managers is near 20 years, so this appears to be a condition for managing a mutual fund. I checked for outliers, but these were not found.
Table II: Descriptive statistics
N Min Max Mean Median St dev
Dependent variables Outperformance '11 91 -19.590 17.100 -0.168 0.030 5.686 Outperformance '09-'11 81 -7.420 23.790 2.313 1.200 4.709
Control variables Beta 96 0.137 1.770 0.948 0.934 0.309
AuM 113 0.010 28.900 3.061 1.200 4.915
Fund Age 123 2.000 31.000 14.569 15.000 6.981
Manager Experience 118 6.000 34.000 19.898 20.000 5.920 This table contains the descriptive statistics of the sample. N indicates the number of respondents for which data was acquired. Outperformance numbers (Ri-Rb) are shown in annualized
percentages. Beta is calculated as Cov(ri, rb)/Var(rb). Assets under Management are shown in
18 Table III presents the responses to the closed questions of the online-questionnaire. A first look at the responses indicates that mutual funds use a variety of ESG data and methods. ESG is mostly used to manage risks, but also in the valuation of companies and to monitor for controversies. Funds use different forms of ESG info, but analysis at company level is used the most. Furthermore, about two–third of the funds have specialized ESG staff at their disposal. Concerning the effect on outperformance, the majority thinks the effect is positive or neutral. The next section elaborates further on these results, and also takes the responses to the open questions into account.
Table III: Responses to key questions
Topic Answer Count Percentage
Role of ESG in investment process To manage risks 84 37%
In the valuation of companies 63 27%
To monitor a stock for controversies 56 24%
To specify/limit my investment universe 24 10%
None of the above 3 1%
Red Flagging Yes 37 29%
Buy transactions 36 29%
Sell transactions 24 19%
Type of ESG info used Analysis at company level 102 30%
Ratings 57 17%
Analysis of specific ESG topics 57 17%
Analysis at sector level 49 14%
Raw data 38 11%
Analysis at country level 36 11%
Detailed instructions on Environmental factors 50 40%
Social factors 50 40%
Governance factors 70 56%
Specialized ESG staff Yes 79 63%
Implementing ESG Training on a continuous basis 72 32%
Staff meetings 70 31%
One or multiple workshops 34 15%
Newsletters 28 12%
Part of the program for new employees 24 11%
Positive 54 45%
Other 36 30%
Neutral 28 23%
Negative 2 2%
Able to provide examples of ESG related investment decisions
Opinion effect of ESG integration on outperformance
19 IV. Results
This section presents the results of this study. It is divided into two subsections. In the first subsection I will interpret the results on both the open and closed questions of the questionnaire, thereby answering the first research question on ESG integration. Furthermore, I will provide
additional insights on the relation between manager characteristics and ESG integration. In the second subsection the results of the regression analyses are presented and interpreted. With these results the second research question is answered.
A. ESG Integration
What stands out from the responses is that nearly all funds exhibit some form of ESG integration. Only 11 funds1 (8,7%) filled in that they do not consider ESG information at all in their investment process and/or classified themselves as ESG-1. A closer look revealed that 8 of these funds are quantitative strategies. If one corrects for these funds, then only 3 funds do not use ESG
information at all. However, for the other funds I find a wide variety of ways of integrating this into the investment process. This illustrates the difficulties of ESG factors. Several respondents comment on this. XXXXX Funds and XXXXX Fund for example, think ESG is difficult to incorporate in financial models, due to their longer term and intangible nature. XXXXX thinks that it is hard to quantify such factors in a meaningful and comparable way, because the most important risks are often not found in published data that can be quantified. This implies that for quantitative oriented managers it is even more difficult to incorporate ESG. However, XXXXX and XXXXX, two of those statistical oriented managers, do make attempts to do this. XXXXX actually found that two governance factors, accounting practices and board compensation, add power to their statistical model, so they have incorporated these. The above mentioned difficulties of ESG integration and the resulting dispersion in the answers to the questionnaire do not prevent me from drawing several conclusions that together answer the first research question about whether and how fund managers integrate ESG factors into their investments process.
27% of the respondents use ESG information in the valuation of companies, but a more frequent use is in managing risks (37%), monitoring news for ESG related controversies (24%) and red flagging (29%). This category of ESG integration emphasizes avoiding the worst ESG performers. The XXXXX representative also mentioned that in their view ESG is more about limiting downside risks, then about exploiting opportunities. The XXXXX funds exclude all companies that score G, on a
1
Funds with no or minimal ESG integration: XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX,
20 scale from A to G, on an ESG issue. The XXXXX Fund also states that it eliminates companies with excessive environmental risks, like it would exclude firms with excessive operational or financial risks. In this light, XXXXX explains that in the end, such companies will pay the price for their irresponsible behavior in the form of fines of supervisory bodies or legal damages. XXXXX
summarizes that ESG is one type of risk that has to be balanced with other types of risk and potential returns. This is exemplified in XXXXX Risk Book, which shows style tilts compared to a benchmark. Besides for example leverage and size, it also includes ESG risks, where for example an above
average environmental impact ratio is highlighted. Although it was not a question in the questionnaire, some asset managers2 mentioned that they engage with companies on ESG issues.
A second major finding is the relative importance that is attached to the governance factor (G), compared to the environmental (E) and social (S) factors. From Table III we already saw that 56% has instructions on governance, compared to 40% for environmental and social factors. Checking
environmental and social factors seems to go hand in hand, because every fund that ticked environmental factors, also ticked social factors and vice versa. XXXXX describes these factors as secondary checks, and XXXXX admits that both play a limited role. On the other hand, 19 different asset managers3 emphasized the importance of governance in their comments. XXXXX illustratively explains why: “we believe strong corporate management is the single most important factor behind success in enhancing shareholder value.” Note the emphasis on shareholder value. One may argue that environmental and social factors are more stakeholder related. This might be a direction for future research. XXXXX and XXXXX also stress the importance of governance, and use ESG as an indicator of management quality. XXXXX said that environmental and social factors are also different from
governance in that they are more sector specific. Environmental factors for example are important to companies in the energy sector, whereas these have far less relevance for the financial sector. This point is also mentioned by the XXXXX and XXXXX funds.
Another item that is closely related to the research question is when ESG information is used. Are companies always screened on ESG issues, or only when portfolio managers think it is relevant? From the open questions I conclude that the latter is the case. XXXXX and XXXXX for example, take ESG into account if these issues are ‘financially material’. Several portfolio managers say that they have data available at central ESG platforms or mention sources like RepRisk, RiskMetrics, GES, Bloomberg and GMI Ratings. So if one would describe ESG integration as a push or pull mechanism,
2
Asset managers that mention they have an engagement strategy: XXXXX, XXXXX, XXXXX, XXXXX, XXXXX,
XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX. 3
Governance is most important according to: XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX, XXXXX,
21 then the latter would be the appropriate classification. Only if the team expects ESG to be material, that is influencing future returns, it is taken into account. Given the materiality, XXXXX and XXXXX try to profit from it. When concerns are exaggerated (in the stock price), then it might be an excellent investment. Similarly, if they are underestimated, the company might be a poor investment. This also indicates that, at least for these asset managers, ESG factors are not (only) incorporated for ethical reasons.
Many asset managers that ticked that they have detailed instructions on ESG factors are unable to provide these instructions. It might be that these are confidential, or that they cannot be summarized in a few sentences. However, it can also be an interpreted as an indication of window-dressing. That is, asset managers state that they consider ESG in their investment process, but in reality they do not. Another sign of window-dressing comes from the question that asked for examples of (not) investing in companies, (partly) because of ESG related reasons during last year. Funds usually invest in a wide variety of stocks, so it would be unlikely that if one has detailed ESG instructions, no single stock was bought or sold because of ESG related reasons in a whole year. Nevertheless, only 29% and 19% of the respondents provided examples of stocks they bought or sold. From this I suspect that in general, asset managers tend to present themselves as having integrated ESG in their process, while this is not always the case. The XXXXX representative confirmed this thought. She questioned whether in general mainstream investment professionals really change what they do, and concluded that they merely talk differently about what they do. In this light XXXXX wrote that given the increased focus by many investors on ESG related matters they have now formalized, so not radically changed, their approach.
22 experienced colleagues. One may argue that older fund managers are more traditional, and the younger ones picked up the relatively new concept of ESG. I find no relationship between the degree of ESG integration and fund size or fund age. A final comparison that is of particular interest to TKP Investments is whether the funds that they have under contract exhibit more or less ESG integration, compared to the other funds in their database. Chi-Square could not be used here, because the
expected count would be too low in multiple cells. Although the ESG-category is an ordinal variable, I compare the means of both groups to give an indicative answer. The funds under contract are on average in ESG-category 2,429 (ESG-score: 23,071) , whereas the funds under research are on average in ESG-category 2,313 (ESG-score: 21,000). However, a t-test shows that the difference is not
significant at any level. The details of the above mentioned statistics can be found in Appendix C.
23
B. Outperformance
Before turning to the results of the regressions, I briefly discuss the respondents’ opinion on the effect of ESG integration on outperformance. Table III shows that 45% thinks the effect is positive, 23% neutral, 2% negative and 30% thinks otherwise. A closer look at the ‘other’ answers shows that several managers distinguish between a short-term and a long-term effect. They think ESG integration has no short-term effect on outperformance, but a positive effect in the long run. A second group of ‘other’ respondents thinks that it does not affect outperformance, but lowers the risk profile of the portfolio. Only two funds think there is a negative effect of ESG integration on outperformance. These are XXXXX and XXXXX. Not surprisingly, both funds are in category ESG-1.
Table IV presents the results of the first regression of the 1 year outperformance over 2011 on the ESG-category dummies and the control variables. Note that of the control variables, only Fund Age is significant (p<0.10). Redundant variables F-tests4 confirm that all control variables, except Fund Age, can be excluded from the regression. This is done because including irrelevant variables leads to inefficient coefficients. The final regression (1) becomes the one presented in Table V.
The validity of a linear regression model rests of various assumptions. Statistical tests on whether these hold for regression (1) are addressed briefly here. Tables with detailed statistics can be found in Appendix D. First of all, the mean of the residuals should be equal to 0. If there is a constant in the regression, which is the case, then this always holds. Secondly, the explanatory variables should be independent. Despite several significant correlations, these are all quite modest. Furthermore, the Variance Inflation Factors (VIFs) are all below 2, so multicollinearity is not a major issue. Thirdly, a normal distribution of the residuals is assumed. The distribution is slightly skewed to the left tail and Kurtosis is 1,120. Since the sample is not very large, this moderate degree of non-normality does not raise serious concerns. Finally, I test for heteroskedasticity using White’s test, because it makes few assumptions about the form of the heteroskedasticity. White’s test uses a regression of the residuals on the ESG-categories. The p-value of the accompanying F-statistic is 0.319, which is far from being significant. Therefore I do not consider heteroskedasticity as problematic.
24 Table IV already indicated significant negative effects of Fund Age and ESG-3 on
outperformance. Table V shows that the Fund Age coefficient is still negative and significant (at the 10% level) in the final regression, which indicates that younger funds in our sample generated more outperformance over 2011 than older funds. Regarding the effect of ESG integration, only α, which represents the coefficient on ESG-1, and the coefficient on ESG-3 are significant at the 5% level. However, the sign of the constant is positive, whereas the ESG-3 coefficient is negative. The ESG-3 coefficient shows that funds in this category generated 3.6% less outperformance compared to ESG-1 funds. The fact that low ESG funds performed better than the funds that exhibit a considerable degree of ESG integration indicates that integrating ESG impaired the performance of mutual funds over 2011. Therefore the hypothesis that a higher degree of ESG integration is associated with higher levels
Table IV: Results regression (1) incl. control variables
Coefficient P-value α -3,079 0,465 D ESG-2 -0,792 0,653 D ESG-3 *-3.342 0,054 D ESG-4 -2,936 0,120 Beta 3,836 0,183 AuM 0,045 0,715 Fund Age *-0.169 0,080 Man exp 0,147 0,173 Non Anglo 2,317 0,120
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
This table reports the results of the cross-section regression where I regress the outperformance over 2011 against the ESG-category dummies (ESG-4 is the highest category) and the control variables Beta, Assets under Management (in billions of euro's), Fund Age, Manager Experience (average of the two most senior portfolio managers) and a Non-Anglo Saxon dummy (which is 0 if the Managing Company is located in the US or UK, and 1 otherwise).
Table V: Results regression (1) incl. Fund Age
Coefficient P-value α **3.794 0.033 D ESG-2 -0.264 0.877 D ESG-3 **-3.600 0.033 D ESG-4 -1.924 0.260 Fund Age *-0.165 0.058
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
25 of outperformance in the short-term (1 year, 2011) is rejected. These results relate to ESG integration by mutual funds, not to ESG at stock level, because the returns of the MSCI ESG Best –in-Class and the MSCI World over 2011 did not differ substantially, -7.55% versus -7,62% (MSCI.com). The Dow Jones Sustainability Index, with 8.26%, performed even better than the Dow Jones Global Index, -9,91% (finance.yahoo.com).
The regression (2) of the 3 year annualized outperformance over the period 2009-2011 on the ESG-category dummies and the control variables yields more intuitive results. The results are
presented in Table VI. None of the control variables is close to being significant. The F-statistic5 from the redundant variables test confirms that they can be removed from the regression, without
significantly impairing the results. Table VII presents the results of the final regression (2).
5 F-statistic: (RRSS-URSS)/URSS) * ((T-k)/m). RRSS and URSS denote the residual sum of squares of the restricted and unrestricted regression, respectively. T is the number of observations, k the number of regressors including a constant and m is the number of restrictions. For the restriction that the coefficients of all control variables are 0, F=0.370. Since this is below the critical value of 2.786 (5% significance level), they are all excluded.
Table VI: Results regression (2) incl. control variables
Coefficient P-value α 4.363 0.280 D ESG-2 0.818 0.601 D ESG-3 *3.107 0.052 D ESG-4 0.902 0.590 Beta -1.845 0.511 AuM 0.131 0.318 Fund Age 0.013 0.880 Man exp -0.103 0.303 Non Anglo -0.159 0.908
26 Like for the first regression, the assumptions underlying the linear regression model are checked. Detailed statistics are in Appendix D. Again there is a constant in the regression, so the mean of the residuals is 0 by definition. The Variance Inflation Indicators (VIFs), which are all below 2, do not give raise to multicollinearity concerns either. However, the normality check for the residuals does draw attention. The values of skewness and Kurtosis are 1.449 and 4.505. So there is skewness to the right and excess Kurtosis. The residual plot shows a distribution that looks rather normal. Considering that for this regression the sample is even smaller (n=81) than for the first regression, it is likely that the limited sample size causes this non-normality. So I continue the analysis without adjustments so far. White’s test indicates the presence of heteroskedasticity in the regression. The coefficients are still unbiased, but it might cause the standard errors to be wrong. In order to solve this and make the regression more robust, I use White consistent estimators in regression (2), as described in Table VII.
Table VII shows that again α and the coefficient on ESG-3 are significant. In contrast to the first regression, both have a positive sign now. The interpretation is that funds in the ESG-3 category generated on average 3.49% more annualized outperformance over the 3 period (2009-2011) than the funds in the low ESG-1 category. Compared to the first regression, this is more in line with the hypothesis that coefficients that are increasing in magnitude. The coefficients are increasing until ESG-3. The highest ESG-4 category has a lower, but insignificant effect on outperformance. Therefore I do not reject the hypothesis that the relationship between ESG Integration and 3 year annualized outperformance is positive. However, since only the coefficient on ESG-3 is significant it is not possible to draw strong inferences on this regression. The only valid conclusion is that ESG-3 funds generated significantly more outperformance in the medium term (3y: 2009-2011) than funds that exhibit no or a minor degree of ESG integration.
Note from Table II that not all variables where available for all respondents. If I would exclude all respondents for which one or multiple data items were missing, this would substantially reduce the sample size and thereby the power of the tests. According to Cohen (1988), with 3
Table VII: Results regression (2)
Coefficient P-value
α 0.930 0.070
D ESG-2 0.990 0.392
D ESG-3 **3.493 0.030
D ESG-4 1.020 0.258
27 explanatory variables, a sample size of 80 is needed to find significance for a medium effect. Including one or two control variables would bring the required sample size up to 90 and 100 respectively. Excluding cases would therefore be very costly. Therefore I replaced missing values with the sample mean in the above tests. To check for robustness I also performed the tests where I did exclude missing cases. Besides the significance levels, the signs of the variables remained unchanged.
28 V. Conclusions and Discussion
This study analyzes the integration of environmental, social and governance (ESG) factors into the investment process of conventional mutual equity funds. I investigated this for 126 global and region specific funds that were under contract or under research at TKP Investments.
First I addressed whether and how fund managers integrate ESG factors into their investment process. A questionnaire was used for this purpose. The response showed that funds vary widely in their ESG practices. Based on their ESG methods, data and skills, I sorted the funds into the categories ESG-1 until ESG-4, from no or minor to a high degree of ESG integration. This gives TKP
Investments the possibility to assess quickly whether and how much a fund incorporates ESG in their investment process. Besides the individual classifications, several findings came forward. Except for a small number of mostly quantitative managers, I find that conventional mutual funds do incorporate extra-financial ESG factors into their investment process. However, they only do this when they have indications that these factors are financially material. So although funds have ESG staff and data at their disposal, its use is mostly demand-driven, rather than a systematic approach. Often this is in the case of managing (downside) risk, translating into monitoring and/or avoiding the worst ESG performers. To assess these risks, fund managers and their analysts most often rely on qualitative information at company level, because quantifying ESG measures in a meaningful and comparable way is difficult. Furthermore, asset managers attach more importance to governance than to environmental and social factors, because its link with a company’s financial performance is more pronounced. Besides, environmental and social factors are argued to be more sector-specific. Other insights are that UN PRI signatories, less experienced managers, and managers from outside the US or UK, exhibit a relatively higher degree of ESG integration.
The emphasis on ESG risk and governance is in line with the literature. However, that most funds incorporate ESG information when deemed necessary has important implications for ESG and SRI studies. Often the effect of incorporating extra-financial information on returns is investigated by comparing the performances of specialized SRI or ESG Best-in-Class funds and conventional funds. However, this distinction might be false, because conventional funds also use ESG information. The difference is that specialized funds have a systematic policy with often exclusionary screens that are ethically motivated, whereas conventional funds only consider ESG factors when considered financially material.
29 medium-term (3 year annualized) outperformance against ESG-category dummies and a number of control variables, where high ESG funds were expected to generate more outperformance than low ESG funds. For the short-term the opposite appeared to hold. Funds with a relatively high degree of ESG integration performed significantly worse than their low ESG integration counterparts. On the other hand, for the medium-term, although the evidence was not overwhelming, the results suggested a positive relation between ESG integration and outperformance. This is roughly in line with the
expectations of several respondents to the questionnaire, who thought ESG had no short-term effect, but a positive effect over longer terms.
30 VI. Limitations and Future Research
This study has a number of limitations. Some of them relate to the use of a questionnaire. The response might suffer from a self-selection bias and window-dressing. However, the high response rate and the remarks in the introductory text give me confidence that the social desirability bias is limited. Another disadvantage of a questionnaire is that it is not possible to have a more deep
discussion and ask for motivations behind answers, except for the telephonic interview that I had with
XXXXX. Further, some questions appeared to result in more useful answers than others. Learning and
distinguishing which questions are relevant was helpful for my interview with XXXXX and will be helpful for TKP Investments in their future contacts with external managers and other professionals.
A more severe weakness is the subjectivity of the ESG scoring method. Besides the subjectivity of the weighting and assignment of components, it assumes substitutability among the different aspects of ESG integration. Therefore the ESG classifications should be seen as a rough categorization, rather than an absolute measure.
Other limitations relate to the regression analysis. Cross-section regressions can only indicate association, not causality. So high ESG funds might be associated with a higher degree of medium-term outperformance, but this does not mean that the outperformance is caused by ESG integration. To statistically prove this, a time-series regression is needed. However, ESG integration is not a
continuous variable that can be measured through time. Secondly, I only regressed 1 year and 3 year performance, not performance over longer periods. Several managers changed their ESG practices compared to 5 years ago, so including these in the analysis would lead to invalid inferences.
31 VII. Recommendations for TKP Investments
This study investigated the ESG practices of 126 equity funds, of which 14 are currently under contract. Besides the scores that can be found in Appendix B, I have attached the ESG approach of the funds under contract separately in Appendix E. This research should serve as a first step towards a firm-wide approach to ESG and sustainability that can be implemented internally and can also be communicated to external parties. In this final section I give some further recommendations for this policy.
34 References
Periodicals
Bénabou, R. and Tirole, J., 2010, Individual and Corporate Social Responsibility, Economica, 77, 1-19.
Cowton, C. 1994, The maturing of Socially Responsible Investment: A review of the developing link with Corporate Social Responsibility, Journal of Business Ethics 52, 45-57.
Edmans, A., 2011, Does the stock market fully value intangibles? Employees satisfaction and equity prices, Journal of Financial Economics 101, 621-640.
Fama, E.F., 1970, Efficient capital markets: a review of theory and empirical work, Journal of
Finance 25, 383-417.
Galema, R., Plantinga, A. and Scholtens, B., 2008, The stocks at stake: return and risk in socially responsible investment, Journal of Banking and Finance 32, 2646-2654.
Gompers, P.A., Ishii, J.L. and Metrick, A., 2003, Corporate governance and equity prices, Quarterly
Journal of Economics 118, 107-155.
Henningsson, J., 2008, Does SEE information make a difference to fund managers?, Sustainable
Development 16, 169-179.
Humphrey, J.E., Lee, D.D. and Shen, Y., 2012, The independent effects of environmental, social and governance initiatives on the performance of UK firms, Australian Journal of Management 37, 135-151.
Klassen, R.D. and McLaughlin, C.P., 1996, The impact of environmental management on firm performance, Management Science 42, 1199-1214.
Manescu, C., 2011, Stock returns in relation to environmental, social and governance performance: mispricing or compensation for risk?, Sustainable Development 19, 95-118.
Margolis, J.D. and Walsh J.P., 2003. Misery loves companies: rethinking social initiatives by business,
Administrative Science Quarterly 48. 268–305.
Orlitzky, M., Schmidt, F.L., Rynes S.L., 2003, Corporate social and financial performance: a metaanalysis. Organization Studies 24, 403–441.
35 Scholtens, B. and Zhou, Y., 2008, Stakeholder relations and financial performance, Sustainable
Development 16, 213-232.
Solomon, J.F. and Solomon, A., 2006, Private, social, ethical and environmental disclosure,
Accounting, Auditing & Accountability Journal 19, 564-591.
Institutions
Bonini, S., Brun, N. and Rosenthal, M., 2008, Valuing corporate social responsibility, McKinsey Christensen, M. et al., 2010, European SRI Study 2010 (Revised version), Eurosif.
Hirschberger, M., Steuer, R.E., Utz, S. and Wimmer, M., 2012, Is socially responsible investing just
screening? Evidence from mutual funds, German Research Foundation.
Jaworski, W., 2007, Use of extra-financial information by research analysts and investment managers, European Centre for Corporate Engagement.
Margolis, J.D. and Walsh J.P., 2001, People and Profits? The Search for a Link between a Company’s
Social and Financial Performance. Lawrence Erlbaum and Associates: Mahwah, NJ.
Magazines
Schaltegger, Wagner, 2006, Managing and measuring the business case for sustainability, Managing
the Business Case for Sustainability.
University papers
Ammann, M., Oesch, D. and Schmid, M.M., 2010, Corporate governance and firm value: international evidence, Working paper, University of St. Gallen.
Andries, M., 2008, Social responsibility and asset prices: is there a relationship? Working paper, Mimeo, University of Chicago.
Eccles, R.G., Krzus, M. and Serafeim, G., 2011, Market interest in nonfinancial information, Working paper, Harvard Business School.
36 Hellman, N., 2000, Investor behaviour - An empirical study of how large Swedish institutional
investors make equity investment decisions, PhD thesis, Stockholm School of Economics.
La Porta, R., Lopez-de-Manes, F., Shleifer, A. and Vishny, R., 2001, Investor protection and corporate valuation, Working paper, Harvard University.
Louche, D., Jemel, H. and Bourghelle, C., 2009, The integration of ESG information into investment processes: toward an emerging collective belief, Working paper, European Academy for Business in Society.
Margolis, J.D., Elfenbein, H.A. and Walsh, J.P., 2009. Does it pay to be good … and does it matter? A meta-analysis of the relationship between corporate social and financial performance, Harvard
Business School.
Books
Cohen, J., 1988, Statistical power analysis for the behavioural sciences, second edition (Lawrence Erlbaum Associates Publishers, Hillsdale)
Fuchs, S., 2001, Against Essentialism: a Theory of Culture and Society (Harvard University Press, London).
37 Appendix A
42 Appendix B
This appendix contains a list of the 126 conventional mutual equity funds under attention. The bold funds are currently under contract with TKP Investments.
43
Table C-II: Chi-Square statistics on (non) Anglo Saxon managers and ESG-category
ESG-cat Total
1 2 3 4
Outside UK/US 0 Count 28.0 30.0 31.0 3.0 92.0
Expected 25.6 25.6 26.3 14.6 92.0 1 Count 7.0 5.0 5.0 17.0 34.0 Expected 9.4 9.4 9.7 5.4 34.0 Total Count 35.0 35.0 36.0 20.0 126.0 Expected 35.0 35.0 36.0 20.0 126.0 Pearson Chi-Square ***41.031 Degrees of freedom 3 P-value 0.000
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
This table reports the crosstabulation of whether a manager is from the US or UK (the dummy is 0) or not (dummy is 1) and the ESG-categories. The variables are independent if the counts are close to the expected counts. The Chi-Square statistic tells whether the differences are significant, which is the case here. The minimum expected count here is 5.40. This criterium is met, with 14.6 as smallest expected count.
Table C-I: Chi-Square statistics on UN PRI and ESG-category
ESG-cat Total 1 2 3 4 UN PRI 0 Count 17.0 5.0 3.0 0.0 25.0 Expected 6.9 6.9 7.1 4.0 25.0 1 Count 18.0 30.0 33.0 20.0 101.0 Expected 28.1 28.1 28.9 16.0 101.0 Total Count 35.0 35.0 36.0 20.0 126.0 Expected 35.0 35.0 36.0 20.0 126.0 Pearson Chi-Square ***26.792 Degrees of freedom 3 P-value 0.000
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
This table reports the crosstabulation of UN PRI signatories (the dummy is 1 when the manager is a signatory and 0 otherwise) and the ESG-categories. The variables are independent if the counts are close to the expected counts. The Chi-Square statistic tells whether the differences are significant. The minimum expected count here is 3.97. This criterium is met, with 4 as smallest expected count.
Appendix C
This appendix contains tables and graphs with statistical results of the Chi-Square tests and correlation matrices, which indicate differences in ESG integration between managers with different
44 Table C-III: Correlation matrix of manager characteristics and ESG-categories
ESG
Tactics Staff and Training Data availability Total score
AuM Correlation -0.014 0.059 0.031 0.024
P-value 0.880 0.538 0.747 0.805
Fund age Correlation 0.036 0.074 0.042 0.045
P-value 0.695 0.416 0.647 0.622
Man exp Correlation -0.120 **-0.192 -0.148 *-0.157
P-value 0.196 0.038 0.111 0.090
*,**,*** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
This table presents the Pearson correlation matrix between manager characteristics and (aspects of) ESG integration. AuM is measured in billion euro's, Fund Age in years, and Manager
