The Influence of Ethics on Investment
Performance
A comparative study of Swedish national pension funds and Swedish private investment
funds
Paper within: Master Thesis Business Administration (Civilekonom exa-mensarbete), specialization in International/Financial Analysis Author: Nokelainen, Mattias
Tutor: Stephan, Andreas
Abstract
Previous research found Swedish national pension funds to perform worse than the mar-ket average. This study finds that, contradictory to previous research, the AP-funds per-form better than the market. This study also finds that ethical fund management increases the stability of a fund’s performance over time but can also limit the fund’s success in shorter time frames. The study utilizes the sustainability rating released by Morningstar Inc., and negative and positive screening practises of the sampled funds, as well as the risk adjusted performance measures Sharpe, and Treynor ratios, in order to investigate the relation between ethics and performance, in a comparison of Swedish national pension funds and Swedish private investment funds.
Key words: International and Swedish Stock Market, Pension Funds, Ethics, SRI, CSR,
ESG, Performance Measures
Acknowledgements
The author would like to express his gratitude to the board of GES Investment Services for their support and for introducing him to his interviewee John Howchin, to whom he is also grateful.
The author also wants to thank his supervisor Andreas Stephan for being supportive and understanding throughout the thesis writing.
Table of Contents
Glossary ... iv
1
Introduction ... 1
1.1 Purpose ... 2 1.2 Perspective ... 3 1.3 Delimitations ... 32
Theoretical Framework and Previous Empirical
Studies ... 4
2.1 Classifying Performance Measurement ... 4
2.1.1 Investor Specific Performance Measurement ... 4
2.1.2 Sharpe Ratio ... 5
2.1.3 Treynor Ratio ... 6
2.2 Corporate Ethics ... 6
2.2.1 Corporate Social and Environmental Responsibility ... 7
2.2.2 Socially Responsible Investing ... 7
2.3 Modern Portfolio Theory ... 8
2.4 Previous Empirical Studies ... 9
3
Empirical Framework ... 10
3.1 The Swedish National Pension System ... 11
3.1.1 AP Funds 1-4 ... 11
3.1.2 AP7 SÅFA ... 12
3.1.3 The Ethical Council... 13
3.2 Sample Investment Funds ... 13
3.3 Hypotheses ... 14
4
Method ... 15
4.1 Methodology ... 15
4.2 Tests ... 16
4.2.1 Least Squares Linear Regression Analysis ... 16
4.2.2 Pearson Correlation Coefficient ... 16
4.3 Calculations and Data Selection Reasoning ... 17
4.4 Morningstar Sustainability Rating ... 18
4.5 Critical Assessment ... 21
5
Results ... 22
5.1 The Correlation of Ethics and Performance ... 22
5.2 Comparing Fund Performance ... 25
6
Analysis ... 29
6.1 National Pension Funds vs Private Investment Funds ... 29
6.2 Sustainability’s Impact on Performance ... 30
6.3 Screening Practises’ Impact on Performance ... 30
6.4 The Ethical Considerations of the National Pension Funds ... 31
6.5 Limitations of the study ... 31
7
Conclusions ... 33
7.1 Contributions to the Literature ... 33
References ... 35
Appendix ... 47
1
Interview ... 47
1.1 Intervju med Etikrådets Generalsekreterare ... 47
1.2 Interview with the General Secretary of the Ethical Council (translated). ... 48
2
Data ... 51
2.1 Ranked Correlation Statistics ... 51
2.2 Least Squares Linear Regression Statistics ... 52
2.3 Performance Tables ... 58
Glossary
Buffer fund – A fund to which money has been put aside in order to help support
short-term deficits in another part of the system. The AP-funds 1-4 are buffer funds and when the there is an excess in the inflow of payments to the Swedish pension system the buffer funds grow, when there is a deficit they decline.
Negative Screening – A practise in which an investment fund excludes, screens out,
firms in a certain industry/industries, mainly for ethical reasons.
Positive Screening – A practise in which an investment fund focuses on adding firms to
1 Introduction
Theory assumes that investors are only interested in two factors before making an invest-ment, expected risk and expected return (Markowitz 1952). This statement is based on an assumption of rationality and utility maximization. Based on theory it is therefore reason-able to assume people to be more interested in money making than world improvement. However, in today’s business climate corporate social and environmental responsibility, (CSR), and socially responsible investing, (SRI), are more attractive than ever before. This increased interest and awareness in social and environmental issues can be seen in the Paris climate meetings in December 2015, which led to unprecedented strict regula-tions concerning climate change (Harvey, 2015). Furthermore, as of March 2016 because of this increasing interest, the independent firm Morningstar Inc., in a collaboration with the environmental, social, governance (ESG) research and analysis company Sustainalyt-ics, is providing its customers with sustainability ratings on more than 20 000 funds (Morningstar, 2016). The amount of financial institutions signing the Principles for Re-sponsible investment Initiative, which is backed by the United Nations, increased by 29 percent in 2015 (Morningstar, 2016). This shows additional proof of ethical investing being a very timely topic.
Previous research has shown mixed results regarding the correlation of financial perfor-mance and level of CSR of a company; (Hull & Rothenberg, 2008, McWilliams & Siegel, 2000, Humphrey & Lee, 2011, Ferhatovic & Gherab, 2014). The majority of studies how-ever, imply that ethical considerations affect performance negatively (Renneboog, Ter Horst & Zhang, 2007).
What really is in the centre of this study however are the Swedish national pension funds. 18.5 percent of Swedish workers’ income is forcibly invested into pension funds. 16 per-centile units of these are automatically put in AP-funds 1-4 and 6. The other 2.5 percent (so called ‘premium pension’) the worker can choose where to invest him or herself, alt-hough will automatically end up in Sjunde AP-fonden, (AP7), if no action is taken from the investor’s side (Apfonderna, 2013).
This thesis aims to provide a brief background of Swedish national pension funds, how they have performed historically and how they invest with consideration of CSR. The
investment funds and explore the possibility of these (potential) differences in perfor-mance being related to the difference in consideration of CSR. This is important to study because of the fact that workers in Sweden have no say in where 16 percent of their earned money ends up being invested. The people financing the Swedish pension system should be provided an unbiased analysis of the national pension funds.
The AP-funds’ mission, given by the parliament, includes considering ethics and the en-vironment, although without ignoring the larger goal of high returns, the main mission is “to manage the funds to the greatest utility for the income pension system” (p 9 Skr. 2014/15:130). As previous research shows that the AP-funds perform below the market (Keskin & Uludag, 2010) it is interesting to explore whether or not the underlying reason for this is the funds’ ethical considerations. The mission statement after all clearly states that the work with CSR must not work in conflict of the goal of high returns. The problem then should be whether or not SRI is causing the AP-funds to perform worse than they could. Hence, the research questions for this thesis are:
How do Swedish national pension funds invest with regards to CSR?
How do Swedish national pension funds perform in comparison to Swedish pri-vate investment funds?
If there are differences in performance, can these differences be explained by SRI? If differences in performance due to SRI can be observed, is this correlation
neg-ative or positive?
1.1 Purpose
The purpose of this study is to analyze and compare, Swedish national pension funds’, and the Swedish private investment funds’, ethical considerations and financial perfor-mances, measured as risk adjusted returns. Moreover, the study will determine whether these differences in performance, if there are any, are correlated with the funds’ ethical considerations. The level of ethical consideration is in this study determined by the sus-tainability score provided by Morningstar and Sustainalytics, as well as whether the funds conduct negative and/or positive screening. What problem this thesis aims at answering is; do the Swedish national pension funds’ ethical considerations cause them to perform
1.2 Perspective
This study is performed from the investor perspective, that is the tax payers who in this study will be called ‘investors’. The people whose pensions are in the care of the AP-funds should be properly informed of how their money is being invested. These people who uphold this current system should have an interest in how the AP-funds have inter-preted the government given mission, “…to maximize utility for the income pension sys-tem.” (p 9 Skr. 2014/15:130). What the greatest utility is, according to the author, is after all up to the people affected by the system.
1.3 Delimitations
This thesis will include parts on five Swedish national pension funds as well as 34 Swe-dish private investment funds. Only SweSwe-dish funds are included in the study because only a Swedish fund would be likely to receive responsibility for the investments of the current pension fund system (2 kap. 2 § andra stycket lagen om allmänna pensionsfonder, LAP 2000:192). The study will provide a brief background on the concepts of corporate social and environmental responsibility, and socially responsible investing, though not go so deep as to cover the philosophical views on ethics and morality in history and/or today. Data will be gathered and analyzed to sufficiently answer the research questions. The author has chosen to exclude AP 6 from the study because it is the smallest national pen-sion fund and because it does not share ethical council with AP 1-4, besides it only invests in unlisted companies, making AP 6 excessive (Apfond6, 2016). AP7 is included in the study, however only AP7s equity fund is used to calculate its performance as the interest fund is of no significant interest for this study.
The chosen measures for assessing the performance of the investment funds are; the Sharpe ratio, the Treynor ratio, expected return or ‘mean annual returns’, and volatility or ‘standard deviation’. There are other measures possible for use, but as this study only aims at comparing the performance of investment funds, and see if there is a relation between these and the funds’ consideration of ethics, said ratios should suffice as a meas-urement for risk adjusted returns (Scholz & Wilkens, 2005). To measure ethical consid-erations, the author intends to analyze the funds’ Morningstar sustainability scores, and
differences in other specific screening practises (Reneboog et al., 2007), this study will not replicate those tests but instead discover if certain inclusion (positive screening) and exclusion (negative screening) play a significant role for Swedish funds’ performance. As the Swedish pension system was reformed in 2001 (Settergren, 2001) data will not be collected from before January 2001 and not later than December 2015. This is because the AP funds 1-4 did not exist by then, hence the assets were allocated by other fund managers and therefore irrelevant to this study. Modern portfolio theory is described to provide the reader with the basic rationale of investors but as this study will not construct its own portfolio the model is not described in detail.
2 Theoretical Framework and Previous Empirical
Stud-ies
This chapter presents some of the existing theories on portfolio management to provide the reader with an understanding of the basic rationale of an investor. It also includes previous studies on the AP-funds and on socially responsible investing in relation to per-formance, to provide the reader with a sufficient background of what has been studied before and what those studies found.
2.1 Classifying Performance Measurement
Evaluating performance is not quite as simple as one might believe. What is meant by performance can vary, hence there are a number of different measures for calculating performance. This study utilizes two parameter performance measures. According to Jobson and Korkie (1981) there are three general classes within the two parameter per-formance measures branch. The classes are divided by their definition and usage of risk. The first class is based on overall risk (standard deviation). The second class uses sys-tematic risk, which is the beta or covariance. The third class does not need a measurement for risk.
2.1.1 Investor Specific Performance Measurement
By calculating returns with regards to risk we get what is called, risk adjusted returns. There are many different models and ratios that can be used for calculating risk adjusted
returns (Hübner, 2005). Scholz and Wilkens (2005) discuss the Sharpe ratio and the Trey-nor ratio, these are so called investor specific performance measurements. The Sharpe and Treynor measures are easily used as they do not use a benchmark as reference, mak-ing it possible to compare portfolios with different risk levels (Le Sourd, 2007). Further-more they have been extensively tested and found sufficiently reliable, making them op-timal for non-professional investors and therefore used in this study. A key difference between the two is that the Sharpe ratio is based on total risk whilst the Treynor ratio is based on systematic risk, or as mentioned in Section 2.1, the Sharpe ratio belongs to class 1 and the Treynor ratio to class 2. When using these two performance measures one should always pick the fund that is superior according to one of these measures as long as it is not inferior to the alternative according to the other measure (Scholz & Wilkens, 2005). Basing investment decisions on more than one type of risk therefore increases the probability of making a good investment decision.
2.1.2 Sharpe Ratio
The Sharpe ratio which was originally called the ‘reward to variability ratio’ has since then become named after its originator William F. Sharpe and is maybe the most com-monly used risk adjusted measurement for returns. The ratio shows how large of a return the investor has performed in relation to total risk, risk being measured as the standard deviation of the stock. The rational investor wants to maximize the Sharpe ratio, which is achieved through constructing a portfolio with the highest possible ratio between ex-pected return and risk. (Sharpe, 1994)
The main weakness of the Sharpe ratio is that it only assumes volatility as the measure for risk and that volatility is only considered as being bad, even though higher volatility also can mean higher returns. Another problem with the Sharpe ratio is that it fails unless there is a normal distribution, making it unsuitable as a risk measure when there is skew-ness or kurtosis in the distribution of returns. This is often the case with hedge funds as these can have modest levels of volatility for long periods of time which then suddenly increase substantially. (Auer & Schuhmacher, 2013)
The Sharpe ratio is calculated by subtracting the risk free rate of return from the expected return of the portfolio and then dividing that by the standard deviation of the same
port-Formula 11
𝑆𝜌 = 𝑟𝜌 − 𝑟𝑓 𝜎𝜌
2.1.3 Treynor Ratio
The Treynor ratio is similar to the Sharpe ratio in that it measures excess returns (returns above the risk free rate) but differs in how it considers risk. The Treynor ratio divides the excess returns by the beta of the portfolio (see Formula 2 below), in other words it adjusts returns with the systematic risk compared to the Sharpe ratio which considers overall risk.
(Le Sourd, 2007)
According to Scholz and Wilkens (2005) the Treynor ratio is the most appropriate meas-ure of a portfolio that has been well diversified. The reason for this being that it eliminates the systematic risk which still exists after the diversification.
Formula 22
𝑇𝜌 =𝑟𝜌− 𝑟𝑓 𝛽𝜌
2.2 Corporate Ethics
There are many different terms used in studies on the ethical aspects of investments and corporate governance. Most common are CSR, SRI, and ESG. ESG is short for Economic, Social, and governance (corporate) and it basically means the same thing as CSR (Gillan, et al., 2010). As there does not exist one entity which is the absolute authority on corpo-rate ethics in the world, although there are several institutions for this such as the U.N and the OECD, there are also different names for these ethical practices. However, to avoid confusion this study will use the term CSR when discussing companies’ ethical practises, except for when referring to another author’s work who specifically used the term ESG in his/her research, and SRI when discussing investment funds’ ethical prac-tises.
1 Sharpe, 1994
2.2.1 Corporate Social and Environmental Responsibility
There are many definitions of CSR which causes it to be a slightly confusing concept to grasp (Van Marrewijk, 2003), this thesis uses the definition as stated by the European Commission (EC). The European Commission defined CSR in 2011 as “…the responsi-bility of enterprises for their impacts on society.” (p 6. European Commission Brussels, 14.7.2015). That ESG is explained in almost the exact same words by the Swedish gov-ernment in Ds 2015:34 further adds to the eligibility of using said definition for this study on solely Swedish investment funds, and that either of the two expressions CSR and ESG could probably be used without confusion.
Examples of acting responsibly are; promoting equality in the workplace, efforts of re-ducing any type of dangerous emissions, and following the law. The Commission states that CSR is important for the European economy because it upholds sustainability, inno-vation, and competitiveness of European companies. CSR is based on several interna-tional guidelines and principles such as the United Nations Global Compact, OECD Guidelines for Multinational Enterprises, and many more. (European Commission, 2016) It was in 1994 that one of the founders of the company SustainAbility, John Elington, coined the term ‘triple bottom line’ also knows as the three Ps. Triple bottom line refers to people, planet, and profit which are also the basis of CSR. ‘People’ means respecting labor and the community and region where the company conducts its business, ‘Planet’ of course referring to the environment, ‘Profit’ may seem to be the most obvious as it is the core of any business, but in the case of triple bottom line it refers to what is the most economically beneficial for the society as a whole (Forbes, 2012).
2.2.2 Socially Responsible Investing
Socially responsible investing has been defined differently many times since its origin in the 1940s. It seems that most definitions agree on that SRI means that the investor pro-motes ideas or concepts that hold values aligning with the investor’s own (Gillan, Hart-zell, Koch & Starks, 2010). This study will commit to the definition provided by The European Sustainable Investment Forum (Eurosif), which states “…a generic term cov-ering any type of investment process that combines investors’ financial objectives with their concerns about Environmental, Social and Governance (ESG) issues.” (p 8.
Euro-and environmental responsibility, such as environment, labor conditions Euro-and similar, but SRI also includes religious beliefs which is an important difference (Hill et al., 2006). In a more practical setting, SRI is what we today call it when investors refrain from in-vesting in firms associated with misconduct or involvement in the production of certain goods and services, such as tobacco, gambling, weapons, alcohol, and/or pornography. The reasons for SRI are purely ethical and philosophical, as it is legal to invest in the above listed goods and services in most countries. Most people today hold views in the normative morality sense, knowing what is good and bad, although still sometimes acting against their own sense of morality. But what is moral is not universal and different cul-tures share different values, this means that SRI cannot be universal either and hence always deserves to be questioned. (Ransome & Sampford, 2013)
2.3 Modern Portfolio Theory
In 1952 Harry Markowitz came up with a model, the Harry Markowitz model, today more known as the mean variance model in modern portfolio theory (Elton & Gruber, 1997). This theory started an era of diversification. Harry Markowitz realized the insufficiency of calculating the expected return and risk of only one stock and that it would be more efficient to calculate what we now call ‘an efficient frontier’. The idea is based on ex-pected returns, the mean returns over a period of time, and the standard deviation, how much the returns deviate from the mean, of a portfolio. The efficient frontier is the line showing the lowest possible risk, using the covariance of the stocks’ standard deviations as the measure of risk, achievable at the highest possible expected returns for different combinations of asset allocation (Markowitz, 1952). In other words one must consider the co-movements of several securities and through this construct a portfolio which min-imizes the risk at the highest achievable expected returns, something which is not achiev-able by simply analysing single stocks against other stocks (Elton & Gruber, 1997). The model makes use of a number of assumptions:
1) The portfolio risk is based on the portfolio’s variability of returns. 2) Investors are risk averse.
4) The utility function of an investor is concave and increases, this because of the inves-tor’s consumption preference and risk aversion.
5) Analysis of the portfolio returns and volatility is based on a single period model. 6) An investor has two choices, either he maximizes expected return at a given risk level or he maximizes his return for the minimum risk level.
7) The investor is rational and needs to make two decisions when choosing portfolio. Creating a set of efficient portfolios, followed by choosing a portfolio of the set to invest in. (Markowitz, 1952)
2.4 Previous Empirical Studies
According to Reneboog et al. (2007) studies have hinted but not proven that SRI invest-ment funds perform worse than conventional investinvest-ment funds. The study conducted by Reneboog et al. (2007) was based on numerous earlier empirical studies within the areas of SRI, CSR, and different screening practices. Reneboog et al. (2007) conducted a fur-ther study in which they found that SRI-funds’ risk adjusted returns underperform their domestic markets in European and Asia-pacific countries by roughly 5 percent per an-num, measured as four-factor- and conditional four-factor-adjusted returns. In the UK and US they underperformed as well but not significantly. The most interesting finding was that SRI-funds’ performance is highly influenced by what sort of screening they practise. Corporate governance and social screens had a positive effect on risk adjusted returns whilst environmental screens had a strong negative impact.
Keskin and Uludag (2010) studied the performance of the AP-funds 1-4 and 7 during the time period 2003-2009 and found that AP 1-4 performed similarly in both expected return and risk. AP7 had the highest annual returns but all five of them performed worse than the average returns at the Stockholm Stock Exchange (SIXRX), which yielded twice the returns during the same time frame. They also compared the funds to SIXRX based on Treynor ratio, where the differences were not as large as in the average return comparison but SIXRX still performed considerably better. SIXRX was the only index used to calcu-late all beta values.
Humphrey and Lee (2011) studied the effects of negative and positive screening on funds risk and performance, in Australia. Their research found no significant effect on
perfor-mance, measured as one- and four factor RAP as well as total return model. Claesson and
Slettvold (2013) looked into the issues of using positive or negative screening as a process for investing or not investing in a company. They found that positive screening produces too subjective judgements from the fund managers, and that positive screening can cause the fund manager to overlook the fact that a company is not ethically responsible by being too focused on the company’s strengths. Negative screening on the other hand they found to be contradictory as it allows for certain extents of the returns to come from the same companies that have been blacklisted. Negative screening also completely disregards all companies in the gaming, pornographic, and alcohol industry out of moral reasons, re-gardless of how these companies conduct their business otherwise. Claesson and Slettvold conducted their study in a qualitative manner, in this case using only interviews as basis for all their conclusions.
Nygren (2013) looked into Sjunde AP-fonden which has been labelled environmental-ethical. His research, which was conducted by looking into AP7s responsibilities and guidelines, as well as its holdings, discovered that there were several companies in AP7s portfolio who do not qualify as ethics friendly and concluded that these were still invested in to achieve profit maximization. Nygren’s findings also led him to consider whether
actions to achieve change within unethical companies can have any positive effect
what-soever. In a comparison between 25 ethical and 25 traditional funds’ performance be-tween 2009-2013 Ferhatovic and Gherab (2014) found no significant difference in risk adjusted returns between ethical and traditional funds, although the traditional funds did perform slightly better. Their research used Sharpe and M2 as performance measures, however only compared M2 averages between the two groups in a t-test for significance.
3 Empirical Framework
This chapter first provides the reader with necessary background knowledge on the Swe-dish pension system and the national pension funds. The chapter ends with a part on the author’s hypotheses regarding the thesis’ results.
3.1 The Swedish National Pension System
The modern Swedish pension system was introduced in 2001 and its national retirement pension consists of three parts, an income-based pension, a premium pension and, for some, a guarantee pension (Apfonderna, 2016). The guarantee pension part exists for the purpose of giving all retirees a sufficiently large pension to live on it (Pensionsmyn-digheten, 2016). 18.5 percent of tax payers’ income is invested in this pension system. 16 percent is allocated in the income based pension which is managed by AP-funds 1-4, and 6. The other 2.5 percent goes in the premium pension, managed by AP7. It is only the premium pension part that the tax payers themselves can choose to invest elsewhere (Ap-fonderna, 2016).
The four largest buffer funds (AP1-4) were allocated SEK 134 billion each when the new system was introduced. This buffer capital is meant to even out deficits between money coming in and money going out that are caused by conjuncture and demographic changes within the Swedish state. These four buffer funds comprise roughly 10 percent of the assets in the pension system, and make up the part that is exposed to financial market risk. The other 90 percent mirrors the value of future pension payments (inflow), and is not affected by the changes on the financial market. Instead this part is affected by salaries, employment rate, and retirement age. (AP2, 2016)
3.1.1 AP Funds 1-4
As previously stated (see Section 1) The AP-funds’ mission includes to consider ethics and environment, although without ignoring the larger goal of high returns (P 9 Skr. 2014/15:130). The ethical part in this mission has been interpreted by AP funds 1-4 as, if it does not violate international conventions we can invest in it (J. Howchin, personal communication, March 9, 2016).
Each of the AP-funds 1-4 must delegate 10 percent of their capital to external investment managers (4 kap. 18 § LAP). The reasoning behind this can be found in prop. 1999/2000:46 p. 111 f. which says that in markets where Swedish fund managers have lacking expertise it is necessary to hire external expertise, in order to create a cost effec-tive fund managing.However, the funds may only delegate this responsibility of
discre-tionary management to an asset manager who in their home country stand under “...ade-quate supervision by an authority or any other competent bodies…” (p 126 Ds 2015:34). Furthermore a minimum of 30 percent of the market value of the assets which each of the AP funds 1-4 hold must be invested in low risk entities, 4 kap. 4 § LAP. The AP-funds are also limited in the amount of shares they are allowed to own in a single firm. AP1-4 are not allowed to have holdings in Swedish companies larger than 2 percent of that com-pany’s total market value (4 kap. 6 § LAP).
3.1.2 AP7 SÅFA
AP7 SÅFA (Statens årskullsförvaltningsalternativ) is comprised of two parts, the AP7 equity fund and interest fund. As you get closer to your pension a larger part of your capital is moved from the equity fund and put in the interest fund (see Figure 1). This is done in order to maintain high returns for as long as possible while still lowering the risk for the investor, so that his/her pension is as large as possible when it is needed. The previously mentioned government given mission only applies to AP1-4 but 5 kap. 1 § LAP clearly states that the management of funds must be exclusively in the interest of the pension savers. AP7’s mission is simply to manage the premium pension part in the pre-selected state alternative AP7 SÅFA, but its goal is to provide its investors with a pension that is at least as good as if they had chosen another alternative. (AP7, 2016)
Figure 13 The gradual change in the AP7-investor’s pension fund as he/she gets older.
AP7 describes itself as an ethically and environmental responsible investor and has a M/E label which translates to taking environmental and ethical considerations, however AP7 does not fit in the standard definition of an ethical or environmental fund often used in
the private market (AP7, 2016). An M/E label simply states that a fund takes environ-mental and/or ethical considerations, there are however no strict guidelines for these con-siderations (Pensionsmyndigheten, 2016). Like AP1-4, AP7 also does not exclude busi-nesses unless it is bound by law or international conventions to do so. Funds that do fit in under the standard definition of an ethical or environmental friendly usually exclude firms of questionable ethics. Examples of companies of questionable ethics includes but is not limited to such businesses that are associated with alcohol, tobacco, oil, pornography, gambling, and weapons. (AP7, 2016)
3.1.3 The Ethical Council
The ethical council is comprised of a secretary general, working only for the council, as well as members from each of the four AP-funds. The council provides AP 1-4 with eth-ical guidelines for investing, and engages companies who misconduct in order to change them for the better (Etikrådet, 2016). The ethical council focuses its resources where they are believed to maximize utility. There is no specific process for determining this, but it is based on the council’s previous experiences and beliefs. The AP-funds are not obliged to follow the recommendations laid out by the council but so far they have done so in 100 percent of the cases. The council follows the international conventions signed by the Swe-dish state, which means that no industries are recommended for exclusion unless they violate the conventions. Out of the limited amount of cases where exclusions have been recommended, all funds have ridden themselves of their possession in these companies within months. The AP-funds own considerable amounts of shares in the tobacco busi-ness. These possessions are continuously checked for possible violations of the WHO convention of tobacco control. As the ethical council is not actually part of the AP-funds, it does not necessarily consider returns when making recommendations. The council will simply recommend to exclude companies if they violate signed conventions regardless of whether or not these are assumed to be profitable investments. (J. Howchin, personal communication, March 9, 2016)
3.2 Sample Investment Funds
The full list of investment funds analyzed in this study including the 34 private investment funds can be found in Section 5.1, Table 3. The table displays the funds’; sustainability
ratings and scores, annual returns, volatility, Sharpe and Treynor ratios, and if they con-duct positive and/or negative screening. No in-depth analysis of these funds have been conducted other than the amount necessary to produce results regarding these just men-tioned variables.
3.3 Hypotheses
Based on the previous empirical studies and the author’s personal beliefs, the hypotheses for this paper’s results are as follows:
1) SRI lowers performance.
In the case of AP 1-4 there are no sustainability scores available instead the author will analyze these funds’ ethical considerations based on a mix of qualitative and quantitative data. The other investment funds do have sustainability scores and so they will be tested against performance measures in a Pearson correlation test and by conducting least squares linear regression the author will see if the correlation levels are of significance. The first hypothesis is that the higher the fund’s sus-tainability score the lower the fund’s performance will be, in other words that sustainability and performance are negatively correlated. As the sustainability score does not consider negative and/or positive screening I will add a separate variable for this for the correlation test. Fund performance is measured as Sharpe ratio and Treynor ratio, but volatility and annual returns are also tested.
2) A majority of the national pension funds perform below the market.
This will be tested by measuring the AP-funds performances in the same way as explained in the first hypothesis, however if the performances ranks below the mean then a significance test will conducted to measure if the AP-funds perform significantly below the market.
3) The fund that performs the best is the least socially responsible.
Being the least considerable towards social and environmental factors should log-ically result in being increasingly focused on performing well financially. After all, unethical practises have been conducted in all times in all kinds of fields all in
order to get ahead. If an ethical company seems like a good investment oppor-tunity this can still be invested in without automatically creating a positive screen-ing practise. A similar argument works for not investscreen-ing in a particular company. To test this hypothesis the author only has to rank the funds based on their sus-tainability scores and performance, then look at the top performing fund and com-pare that rank to its sustainability rank.
4 Method
The methods chapter contains the methodology used for collecting all the research data. It also contains the calculations used to obtain all the necessary results for answering the hypotheses and why they were calculated as they were.
4.1 Methodology
This study follows a mainly positivistic research paradigm as quantitative data often is less likely of producing biased results, and because the main results of this study are most efficiently found using a quantitative approach. According to Collier and Mahoney (1996), a study using mainly qualitative data is more likely to exaggerate the causal ef-fects of two variables. The author of this study strongly believes that a researcher has a responsibility to produce unbiased, non-exaggerated results as doing otherwise could be called a ‘research sin’. It is in the author’s beliefs that the use of quantitative data in order to achieve results is best suited for a study which aims to test the correlation between different variables. If the correlation is found too low to be significant but still hinting that the correlation may be causal, because of the use of quantitative data, then that creates an opportunity in the analysis for the author to discuss that possibility, however also de-prives him of the possibility to draw a definite conclusion on the matter.
Qualitative primary data will be gathered through an interview with the general secretary of AP-funds 1-4 shared ethical council. Interviewing the general secretary of the ethical council should provide a better understanding of the reasoning behind the AP-funds’ eth-ical considerations behind their investment decisions, as well as increased clarity regard-ing their interpretation of their government given mission.
Quantitative secondary data will be gathered from the AP-funds’ websites, Pensionsmyn-digheten, Morningstar, MSCI, Tradingeconomics, Hållbarhetsprofilen, and from the pri-vate investment funds’ websites included in the study. The quantitative data sources were chosen based on relevance and accessibility.
The quantitative data gathered is in the form of; historical prices, so called net asset val-ues, used for calculating the performance of the sample, sustainability ratings and scores, that will be tested against the funds’ performances to study a possible correlation between the variables, as well as annual index returns to calculate beta values for the funds, and risk free rates in Sweden during the periods of 2001 to 2015 and 2010 to 2015.
4.2 Tests
Sampled data has been analyzed using two tools; Least squares linear regression analysis and Pearson correlation. These can easily be used in MS Excel and are efficient for com-paring different variables and finding relations between these.
4.2.1 Least Squares Linear Regression Analysis
The least squares linear regression method is used to show how a dependent variable is affected by one or more independent variables. In the case of this study sustainability score becomes the independent variable and the author tests how it is affects the factors volatility, annual returns, Sharpe ratio, and Treynor ratio. Positive and negative screening are also tested as independent variables and using the Sharpe and Treynor ratios as de-pendent variables. The regression results, show not only how strongly the dede-pendent var-iable is affected by the independent varvar-iables but also provides different measures for whether or not this relation is significant. This study utilizes the р-value to interpret the significance level, and assumes very high significance if that value is lower than an α = 0.01, a definite significance if p < α = 0.05 and a possible significance if p < α = 0.10.
(Anderson, Sweeney, Williams, Camm, & Cochran, 2013)
4.2.2 Pearson Correlation Coefficient
The Pearson correlation coefficient, (PCC), measures to which extent two variables co-vary. It is different from covariance however, as the PCC is scaled such that the value resulting from the test is independent of the units in which the two measurement variables
are expressed. Hence, the PCC will still provide the same value even if one variable is changed from imperial to the metric system. The PCC is always between -1 and +1, where -1 means a perfect negative correlation and +1 means a perfect positive correlation. Val-ues close to 0 would indicate no correlation, that the two variables are unrelated. A posi-tive correlation means that large values of one variable is associated with high values of another variable. A negative correlation means instead that high values of one variable is associated with low variables of the other. (Anderson, et al., 2013)
According to Fenton and Neil (2012) the strength of a correlation can be defined as low or weak if it in absolute values is lower than 0,4, medium if it ranges between 0,4 and 0,6 and strong if the coefficient is higher than 0,6. The author also believes it is necessary to mention the issue of drawing conclusions based on statistics. In a study like this, where ethics are tested alongside performance, a nearly perfect correlation between the two is not assumed. Rather, it would be more believable if the correlation is small. However, smaller amounts of correlation is statistically insignificant unless the sample size is very large, (Fenton & Neil, 2012). For a study with such limited time where gathering high amounts of data is very time consuming, a sample size of such magnitude that significance is reached even in low correlations, is difficult to achieve.
4.3 Calculations and Data Selection Reasoning
The historical data gathered from the 39 funds include daily net asset values, (NAV), from January 1st 2001 to December 31st 2015. 32 funds have NAV values from that date but the rest are only used in the later sample period 2010 to 2015 as it is in 2010 that the fund with the least amount of data starts. Both data periods are calculated in an identical manner except for the amount of observations and of course, time range of the sample. The sample is assumed to be normally distributed as it exceeds 30 observations, which according to the Central Limit Theorem is sufficiently large (Anderson, et al., 2013). By using MS Excel to calculate the logarithmic returns for the daily NAV values and finding the arithmetic average of these, mean daily returns are calculated. The standard deviation is based on the logarithmic daily returns. The average arithmetic returns and the standard deviation of the sample are then both multiplied by the average number of trad-ing days in a year (252) to receive mean annual returns and annual standard deviation.
These annual returns are in turn used to calculate the covariance or ‘the beta value’ of each fund by using the ‘slope function’ on the annual returns against the returns of each fund’s corresponding index’s returns. The Sharpe ratio is calculated as described in Sec-tion 2.1.2, Formula 1, and the Treynor ratio is calculated as described in SecSec-tion 2.1.3, Formula 2.
Sustainability rating, sustainability score, and negative and positive screening require no processing as their values are gathered directly from an external source. Index values could have been used in raw form as well, however the author combined two different indices for a limited number of funds in order to as precisely as possible calculate the most exact systematic risk. Depending on the fund’s holdings the beta was based on dif-ferent percentages of two indices in those cases. The author decided which indices that were gathered for the calculation of betas based on the recommendation of Morningstar and in cases where Morningstar had no recommendation, on the author’s analysis of that fund’s holdings.
If a fund has a negative and/or positive screening practise, this is treated as a dummy variable and is written as a ‘1’ in Excel if the fund does and a ‘0’ if it does not.
As the sustainability ratings cannot be transferrable between categories the sustainability score is used to test for correlation between sustainability and performance in the regres-sion tests. Sustainability rating would have been used within ‘single category’ regresregres-sion tests, if the funds were divided by which market they invested in. Regressions are run with only single X-variables to avoid multicollinearity from reducing the reliability of the results.
As correlations may be small and end up being regarded insignificant despite the possi-bility of being directly causative the funds are also ranked based on their sustainapossi-bility score and Sharpe-, and Treynor ratios. By ranking the funds it creates an opportunity to run a regression analysis based on ranks where small differences are amplified and large differences are normalized.
4.4 Morningstar Sustainability Rating
The sustainability rating measures a fund’s management of ESG issues in relation to other funds in the same category or market (Sweden, global, small companies etc.). Depending
on how well the fund does in its category it will receive a rating between 1-5 earth globes. The rating is based on the fund’s sustainability score, which in turn is calculated by the portfolio ESG score deducted by portfolio controversy deduction. The ESG score is pro-vided by Sustainalytics and is also based on industry peer groups. The controversy de-duction means that a portfolio receives a lower sustainability score if companies it has invested in are subject of ESG-controversies or ‘scandals’. (Morningstar, 2016)
“Portfolio Sustainability Score = Portfolio ESG Score – Portfolio Controversy Deduc-tion”(p 1 Morningstar Sustainability Rating, 16 Feb 2016)
The ESG score has been given based on the variables, preparedness, disclosure, and
per-formance measuring how well a company manages ESG issues. The received scores range
from 0-100 and are calculated by normalizing the score in each category. 50 is the peer group average and then each additional 10 points above or below are equal to a standard deviation away from the average. A portfolio then receives an ESG score based on the asset weighted average of the companies’ scores in the portfolio. A minimum of 50 per-cent of the portfolio’s assets must be ESG rated to receive a portfolio ESG score. This is also rescaled to 100 percent before the calculation of the portfolio ESG score. (Morn-ingstar, 2016)
Formula 34
𝑍𝑝𝑒𝑒𝑟 = 𝐸𝑆𝐺𝑥 − 𝜇 𝑝𝑒𝑒𝑟 𝜎 𝑝𝑒𝑒𝑟 Where
:
ESGx = Sustainalytics company ESG score
µ Peer = Peer-group mean ESG score
σ Peer = Peer-group standard deviation of ESG score
Formula 45
𝑃𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜𝐸𝑆𝐺 = ∑𝐸𝑆𝐺𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑𝑥𝑊𝑒𝑖𝑔ℎ𝑡𝑒𝑑𝑎𝑑𝑗
𝑛
𝑥=1
The overall controversy deduction score is calculated by rescaling the weighted average, although this score is based on the gravity controversies committed by a firm. There are six categories of controversy impact on environment and/or society, 0 means no evidence of controversy and provides the company a score of 100, and 5 is the most severe and provides a score of 0. After rescaling the highest scoring portfolio receives a score of 0 hence, deduces 0 from its ESG score. Accordingly, the lowest scoring portfolio receives the highest score and are deducted 20 points from its ESG score. A minimum of 10 rated funds must exist in a category for a fund to receive its final Morningstar sustainability rating. (Morningstar, 2016)
Table 16 Sustainalytics Controversy Score: The worse impact a company has on the environment or society the
higher controversy score it receives and will be deducted points from its ESG score.
Formula 57
𝑃𝑜𝑟𝑡𝐶𝑜𝑛𝑡𝑟𝑜𝑣𝑒𝑟𝑠𝑦= 100 − ∑ 𝐶𝑜𝑛𝑡𝑟𝑜𝑣𝑒𝑟𝑠𝑦𝑥𝑊𝑒𝑖𝑔ℎ𝑡𝑒𝑑𝑎𝑑𝑗
𝑛
𝑥=1
The funds receive their final category rank based on how high their portfolio sustainabil-ity score is in relation to its peers. The top 10 percent receive five globes, the next 22.5 percent 4 globes, the next 35 percent 3 globes, the next 22.5 percent 2 globes, and the bottom 10 percent receive 1 globe. (Morningstar, 2016)
5 Used with the written consent of Morningstar, Inc.
6 Used with the written consent of Morningstar, Inc.
Table 28 Morningstar Sustainability Rating: The table shows that the rating a portfolio receives is dependent on its portfolio’s sustainability score in relation to its peers.
4.5 Critical Assessment
The arithmetic average return is somewhat faulty because in reality if a fund has a nega-tive logarithmic return then it would need a larger posinega-tive logarithmic return to compen-sate for that downfall, but in the case of average arithmetic returns negative returns are slightly underrated. However I still believe this to be a good measurement as fund shares are bought and sold basically every day and not on an annual basis. The differences from day to day are also very small which almost eliminates this discrepancy.
Necessary to mention is that funds that only screen out inhumane weapons are still as-signed a 0 dummy variable as all funds in the sample screen out these weapons. The data received on AP1-4 was lumped together which is why these funds’ are treated as a single fund in the results section.
A final lesser issue is that two of the seven indices contain historic returns from 2002 first and not 2001 like the other five do. The reason for this is lack of availability of this data. The indices are gathered from two different sources and one of the two only contained data from 2002 onwards. However, fortunately the indices with data from 2002 are only used to calculate the beta value of one fund each. The author also tried calculating another fund’s beta value from the year 2002 instead of 2001, and the difference in beta was so small it is barely worth mentioning.
The author believes the sources used in the thesis to contain reliable data and information to make fair conclusions about the results. The only data with a moderate possibility of bias would be from the interview conducted, but as the author suspected this possibility the questions asked were very straight forward with little room for biased answers or interpretations.
5 Results
This part features and explains the results created through the data analysis in MS Excel, which has been done in order to test this thesis’ research questions and hypotheses.
5.1 The Correlation of Ethics and Performance
Starting with the results on the correlation of ethics and performance for the years 2001-2015, Table 3 below shows a medium strength positive correlation between sustainability score and performance. Least squares regression shows that this correlation is significant on the 1 percent level for both the Sharpe and Treynor ratios as seen in Table 4 below. Positive screening shows an insignificant correlation to performance on all levels. Nega-tive screening however showed a posiNega-tive correlation, which was significant on the 10 percent level for the Sharpe ratio and on the 5 percent level for the Treynor ratio. These results are based on a slightly smaller sample, 28 observations instead of the full sample of 36 (AP1-4 count as one observation) observations. The reason for this is that not all funds have data available from 2001 and onwards and so only the ones that do are in-cluded. AP1-4 is not included in either time period’s correlation tests as it lacks data on sustainability score. Looking at the ranked correlations between sustainability score and performance (see Table 12 in Appendix 2.1), the results are nearly identical. Both Sharpe and Treynor ratios are significantly affected by sustainability score on the 1 percent level according to the regression performed on the ranked list (see Table 16 in Appendix 2.2).
Table 3 Pearson Correlation 2001-2015
Sustainability scoreMean Annual ReturnsVolatility Sharpe Treynor Positive Negative
Sustainability score 1,0000
Mean Annual Returns 0,5956 1,0000
Volatility 0,3428 0,0306 1,0000
Sharpe 0,5639 0,9855 0,0428 1,0000
Treynor 0,5601 0,9898 0,0072 0,9924 1,0000
Positive screening -0,2855 -0,0954 -0,0900 -0,1282 -0,1085 1,0000
Table 49 Least Squares linear regression estimates 2001-2015:
Running the same tests on the full sample for the period 2010-2015, Table 5 below shows a low to medium negative correlation between sustainability score and performance. The correlation is significant on the 5 percent level for the Treynor ratio and the 10 percent level for the Sharpe ratio (see Table 7). The ranked correlations are actually stronger in this period. Both Sharpe and Treynor ratios are found significant on the 5 percent level in the regression run on the ranked fund list (see Table 19 in Appendix 2.2). Negative
screen-Regression statistics 2001-2015
Sustainability score as independent (X) variable
Y-Variable Coefficient P-Value Multiple R Intercept Observations
Sharpe 0,0227 ***0,002 (P) 0,5638 -1,2879 28
Treynor 0,0046 ***0,002 (P) 0,5601 -0,2609 28
Mean Annnual Returns 0,0045 ***0,001 (P) 0,5955 -0,2214 28
Volatility 0,0031 *0,0742 (P) 0,3428 0,0073 28
Positive screening as independent (X) variable
Y-Variable Coefficient P-Value Multiple R Intercept Observations
Sharpe -0,0552 0,5669 (N) 0,113011961891910,0152 28
Treynor -0,0093 0,6387 (N) 0,09275730371102740,0039 28
Negative screening as independent (X) variable
Y-Variable Coefficient P-Value Multiple R Intercept Observations
Sharpe 0,1234 *0,0714 (P) 0,3459 -0,0720 28
ing has a medium strength negative correlation to both Sharpe and Treynor which is sig-nificant on the 1 percent level as can be seen in Tables 6 and 7. Positive screening shows an insignificant correlation to either performance measures.
Table 5 Pearson Correlation 2010-2015 top table is full sample (35 observations) bottom table is smaller due to lack of data (28 observations).
Table 6 Pearson Correlation 2010-2015 all variables on the smaller sample (28 observations)
These results neither prove nor disprove the first hypothesis (see Section 3.3) of sustain-ability being negatively correlated with performance, as it was positively correlated in the longer time period but negatively correlated in the shorter period. Volatility however is found to be positively correlated with sustainability score during both periods, within 10
Sustainability score Mean Annual Returns Volatility Sharpe Treynor
Sustainability score 1,0000
Mean Annual Returns -0,0513 1,0000
Volatility 0,4994 0,3911 1,0000
Sharpe -0,3017 0,8774 -0,0762 1,0000
Treynor -0,3464 0,8424 -0,0051 0,9316 1,0000
Treynor Sharpe Negative scr Positive scr
Treynor 1,0000
Sharpe 0,9352 1,0000
Negative screening -0,5426 -0,5414 1,0000
Positive screening -0,2580 -0,1666 0,3043 1,0000
Sustainability score Mean Annual Returns Volatility Sharpe Treynor Positive Negative
Sustainability score 1,0000
Mean Annual Returns -0,0107 1,0000
Volatility 0,4830 0,4395 1,0000
Sharpe -0,2668 0,7410 -0,2149 1,0000
Treynor -0,2868 0,7847 -0,0909 0,9352 1,0000
Positive screening -0,0943 -0,1329 -0,0143 -0,1666 -0,2580 1,0000
percent significance 2001-2015 (see Table 5) and 1 percent significance 2010-2015 (see Table 7).
Table 710 Least Squares linear regression estimates 2010-2015:
5.2 Comparing Fund Performance
For the second hypothesis, comparing the performance of the sampled funds whose his-torical data could be analyzed (29 observations) from 2001-2015 in Table 8, one can ob-serve that 4 perform better than the average fund on all performance variables. AP1-4 have a higher mean annual return (0.062 vs 0.035), lower volatility (0.120 vs 0.182), higher Sharpe ratio (0.250 vs 0.016) and higher Treynor ratio (0.051 vs 0.004) than the sample mean.
Regression statistics 2010-2015
Sustainability score as X-variable
Y-Variable Coefficient P-Value Multiple R Intercept Observations Sharpe -0,0143 *0,0782 (N) 0,3017 1,3645 35 Treynor -0,0028 **0,0415 (N) 0,3464 0,2572 35 Mean Annnual Returns -0,0004 0,7697 (N) 0,0513 0,1270 35 Volatility 0,0035 ***0,0023 (P) 0,4994 -0,0440 35 Positive screening as independent (X) variable
Y-Variable Coefficient P-Value Multiple R Intercept Observations Sharpe -0,1159 0,3969 (N) 0,1666 0,6303 28 Treynor -0,0292 0,1850 (N) 0,2580 0,1064 28 Negative screening as independent (X) variable
Y-Variable Coefficient P-Value Multiple R Intercept
Sharpe -0,2751 ***0,0029 (N) 0,5414 0,7906 28 Treynor -0,0449 ***0,0029 (N) 0,5426 0,1311 28
Table 8 Performance data 2001-2015: The table features the mean values for AP-funds 1-4, as well as the sample average in the same period. Full data can be found in Table 20 in Appendix 2.3.
In the time period 2010-2015 (see Table 9) AP1-4 also perform better than the mean regarding volatility, Sharpe, and Treynor, however with a slightly lower annual return, roughly 9 percent vs the sample mean of almost 10.4 percent. AP7 performs better in terms of annual returns (0.133) and Sharpe (0.615 vs 0.574), but has a higher volatility (0.187 vs 0.153) and slightly lower Treynor ratio (0.091 vs 0.099) than the mean. These findings disprove the second hypothesis, that the majority of the national pension funds perform below the market (see Section 3.3).
Table 9 Performance data 2010-2015: The table features the mean values for AP-funds 1-4 and 7, as well as the sample average in the same period. Full data can be found in Table 21 in Appendix 2.3.
Answering the third hypothesis, in the period 2001-2015, Lannebo Småbolag is the fund performing best both in Sharpe- and Treynor ratio (see Table 10 below). Regarding sus-tainability it is ranked 14th which is just about in the middle. Didner & Gerge Aktiefond is ranked highest in sustainability score and is ranked 5th in Treynor ratio and 4th in Sharpe ratio.
Table 11 below shows that funds ranking well regarding sustainability do not rank well in the performance variables during 2010-2015. SEB Läkemedelsfond ranks highest in overall performance and has the lowest sustainability score (as hypothesized), Lannebo
Vision ranks the second highest in overall performance rank and 27th in sustainability.
AP1-4 is a close third regarding performance and should in any case not be ranked highly regarding sustainability based on the author’s judgement.
These results both prove and disprove the third hypothesis, (see Section 3.3), it holds for the period 2010-2015 but not for 2001-2015.
2001-2015 Annual Returns Volatility Sharpe Treynor
AP 1-4 0,0620 0,1202 0,2500 0,0513
Sample average 0,0352 0,1824 0,0157 0,0043
2010-2015 Annual Returns Volatility Sharpe Treynor
AP7 Aktiefond 0,1328 0,1867 0,6149 0,0912
AP 1-4 0,0894 0,0609 1,1713 0,1567
Table 1011 2001-2015 Funds ranked on Sustainability Score, Treynor ratio and Sharpe Ratio.
11 Funds with the same sustainability scores receive the same rank. AP1-4 lacks some data regarding
sustaina-bility, and has therefore been ranked last in this. All funds with data 2001-2015
Ranked list lowest total points first Sustainability RankTreynor Sharpe Total points = sum of ranks
Didner & Gerge Aktiefond 1 5 4 10
Handelsbanken Svenska Småbolag 10 2 2 14
AMF Aktiefond Sverige 3 6 6 15
Lannebo Småbolag 14 1 1 16 Lannebo Sverige 2 7 7 16 Carnegie Sverige 10 4 5 19 Skandia Sverige 3 8 8 19 Sverigefond MEGA 5 9 9 23 SEB Sverigefond 5 12 12 29 AMF Balansfond 10 10 10 30
Handelsbanken Sverigefond Index 5 13 13 31
AMF Aktiefond Världen 10 11 11 32
AP 1-4 29 3 3 35
Inst Aktiefonden Sverige 5 17 17 39
Nordea Futura 15 15 15 45
Aktiefond Pension 15 16 16 47
Nordea ALFA 9 20 19 48
Swedbank Robur Allemansfond Komplett 15 18 18 51
SEB Läkemedelsfond 28 14 14 56
IP Aktiefond 15 22 20 57
AMF Aktiefond Global 22 19 21 62
Kapitalinvest 15 25 23 63
SEB Aktiesparfond 22 21 22 65
Skandia SMART Balanserad 20 24 24 68
SEB Världenfond 21 23 25 69
Skandia Världen 24 27 26 77
Handelsbanken Global Tema (A1 EUR) 26 26 28 80
Lannebo Vision 24 28 29 81
Table 1112 2010-2015 Funds ranked on Sustainability Score, Treynor ratio and Sharpe Ratio.
12 Funds with the same sustainability scores receive the same rank. AP1-4 lacks some data regarding
sustaina-bility, and has therefore been ranked last in this.
All funds 2010-2015
Ranked list lowest total points first Sustainability scoreTreynor Sharpe Total points = sum of ranks Handelsbanken Svenska Småbolag 12 3 5 20
Didner & Gerge Småbolag 16 4 3 23
Didner & Gerge Aktiefond 1 11 12 24
Lannebo SmÃ¥bolag 16 7 6 29 Lannebo Vision 27 1 4 32 Nordea Futura 17 10 7 34 Lannebo Sverige 2 20 15 37 SEB Läkemedelsfond 35 2 2 39 Carnegie Sverige 12 13 14 39 AP 1-4 36 5 1 42 SEB Aktiesparfond 25 9 8 42 SEB Världenfond 24 12 10 46 Technology 32 6 9 47 SEB Sverigefond 6 19 22 47
SEB Sverige Indexfond 6 18 24 48
Nordea Indexfond Sverige 3 21 27 51
Handelsbanken Sverigefond Index 6 23 23 52
Handelsbanken Global Tema (A1 EUR) 32 8 16 56
AMF Balansfond 12 25 19 56
AP7 Aktiefond 32 14 11 57
Inst Aktiefonden Sverige 6 26 25 57
AMF Aktiefond Sverige 3 27 28 58
Skandia Sverige 3 31 26 60
Aktiefond Pension 17 15 29 61
Nordea Global 29 16 17 62
Handelsbanken Global Index Criteria 29 22 13 64
AMF Aktiefond Världen 12 32 21 65
Nordea Global Value Fund 29 17 20 66
Sverigefond MEGA 6 34 31 71
AMF Aktiefond Global 25 30 18 73
IP Aktiefond 19 24 30 73
Kapitalinvest 19 28 32 79
Nordea ALFA 11 33 35 79
Swedbank Robur Allemansfond Komplett 19 29 33 81
Skandia SMART Balanserad 23 36 36 95
6 Analysis
The analysis section covers all the author’s thoughts and reflections concerning the study’s findings. The author analyzes the significant results and what they indicate, as well as the insignificant results and if these could possibly still be meaningful.
6.1 National Pension Funds vs Private Investment Funds
To begin with, one of the most interesting findings of this study would be that the pension funds performed better than the market. This was much unexpected because of the study that Keskin and Uludag (2010) had conducted, which also used the Treynor ratio for cal-culating performance, but found that the AP-funds performed below the market. The rea-son this study’s results and Keskin and Uludag’s results do not match could be that they used a bad reference index. SIXRX is not suitable as a reference, because the Swedish market is not a good representation of the world market and therefore causes the Treynor ratio to be miscalculated. The Swedish market also performed a lot better than the world market during the years of their research (see Table 22 in Appendix 2.4), which creates a large bias when comparing against funds that invest closer to the world market index, MSCI World.
When comparing the performance of private investment funds and the AP-funds, one must also not forget that the AP-funds are heavily regulated and are thus required to invest a certain part of its assets abroad, as well as not being allowed to have too large holdings in the Swedish market. The AP-funds 1-4 must also invest a minimum of 30 percent into low risk entities which should limit their possibility of high returns. Having such regula-tions is only natural as people’s pensions depend on the funds’ stability. But because of these limitations the AP-funds’ historical performance becomes all the more impressive. Based only on their financial performance one should be able to say that they accomplish their mission rather well. Regarding the ethical part it is harder to tell. Out of the variables that this study analyzes the AP-funds are all among the least ethical. However, one must not forget that they try to affect companies to change for the better and that the impact of this could possibly be stronger than other ethical considerations.
6.2 Sustainability’s Impact on Performance
Starting by discussing the correlation estimates, one can see that the two sample periods show almost opposite findings. In the longer time period ethical considerations seem to have a positive effect on performance, while in the shorter period it instead negatively affects performance. During the shorter period, all funds performed considerably better. This is true for the indices as well (see Table 22 in Appendix 2.4), meaning that these years were obviously better overall for the market as a whole. A possible explanation of these findings would be that firms that are more ethical and sustainable are sturdier during times of financial distress. By that logic a less ethical firm may instead perform increas-ingly well during good years but also correspondincreas-ingly worse than others during years of financial distress. Another possibility would be that it is not just those firms but their whole industry that is affected as Howchin (2016) said that certain sectors are more prob-lematic than other (see Appendix 1.2).
Volatility was found positively correlated with sustainability during both periods. Instinc-tively one would think that this correlation should be negative, given that volatility is most commonly associated with risk and that the firms with a higher sustainability score performed better in the period when the whole market was more volatile. However, as mentioned in Section 2.1.2 volatility is not necessarily a bad thing.
6.3 Screening Practises’ Impact on Performance
Regarding negative and positive screening, negative screening has had a larger impact on performance during both periods. This seems reasonable as excluding a whole industry should affect performance more than just when possible adding a firm to the portfolio based on certain characteristics. Only the negative screening is found significant in either of the periods and just like in the case of sustainability it was positively correlated with performance during the longer period, and negatively correlated during the shorter period. This may be an indication that it is indeed certain industries that suffer the most during times financial distress. This finding shows that it was relevant to study the possibility of negative and positive screening having an impact on Swedish funds’ performance, even though Humphrey and Lee (2011) found no significance for this on Australian funds. The
whether screening affects performance positively or negatively could possibly be rein-forced by these results, but one would need to do a thorough analysis of each funds’ screenings.
6.4 The Ethical Considerations of the National Pension Funds
As Morningstar had no sustainability score available for the AP funds 1-4, the author can only base his analysis of these on the data gathered from the funds’ websites and the interview conducted with John Howchin. AP 1-4 conduct neither negative nor positive screening. As this thesis does not study the effects of engagement that factor cannot be considered either. The ethical council does however continuously check their holdings for possible breaches of conventions and could as such, if they are successful in this, possibly have few points deducted in the Morningstar controversy score, should they be included.
AP7 ranked 34th regarding the Morningstar sustainability score, although conducts both negative and positive screening. As Nygren (2013) found though, AP7 had holdings in unethical corporations and this study supports his findings. As Tables 3 and 5 shows (see Section 5.1), sustainability score and screening practises are not necessarily positively correlated. This could be an indication of poorly conducted screening or, more likely, that one needs to analyze the amount of screening before testing the correlation between these variables.
6.5 Limitations of the study
A first limitation would be how negative and positive screening are treated in the study. Because there is no template for measuring the amount of screening a fund conducts, this study cannot differ between large and small amounts of screening. Had the level of screening been obvious in the data analysis it is possible the results would have come out stronger.
The Morningstar sustainability score does not consider or reward negative screening. It also does not show whether or not the fund managers have had a positive effect on firms’ governance. This is why the author decided to gather data on screening from Hållbar-hetsprofilen. Morningstar had also mainly provided sustainability scores on large stock
This also caused the sample to be smaller because of lack of availability. The author would have preferred a greater sample size in order to see if the smaller correlations held significance.
Finally, the Morningstar sustainability score is completely new and because of that does not perfectly mirror the reality of the sustainability of the funds over the entire period. However, the rating takes into consideration historical controversies and should therefore definitely, to some extent, reflect the funds ethical practises historically as well. Unfortu-nately Morningstar could not provide a sustainability score for the income based pension funds (AP 1-4), and this makes it impossible for this study to draw a fair conclusion re-garding whether AP-funds 1-4 are ethical or not.
