Investigating the relationship between ESG and REITs

(1)

Investigating the relationship between ESG and REITs

MSc Finance – Real Estate Finance

Abstract

This paper shows that ESG is priced for REITs. Constructing a portfolio by going long in high-rated ESG firms and short in low-rated was unable to outperform the market between 2003 and 2016. ESG had an adverse effect on the return predominantly. However, the results suggest that this relationship is changing towards a positive link. Including ESG as a factor in the multifactor model increases the return of listed real estate between 2014-2016. In addition, it appears that there is a heterogeneous effect of ESG comparing listed real estate to the market. The results suggest that the price of ESG differs between REITs and the market. Investigating the link of ESG and return for listed real estate has received little attention. This thesis illustrates that it is essential to look at the individual effect and that prior studies regarding the whole market cannot just be generalised to REITs.

Robin de Rooy

10667571

Supervised by: Gianluca Marcato

July 2018

(2)

Statement of originality

This document is written by Robin de Rooy who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1. Introduction

In recent decades Socially Responsible Investing (SRI) has been increasingly growing in relevance (Newell, 2009). The share of investments made in firms that take environmental, social and corporate governance (ESG) issues into account is continuously growing. Bansal & Roth (2000) identify four drivers of corporate social responsibility (CSR): regulation, pressure from stakeholders, moral beliefs, and increased financial performance. Climate change and rising attention to social and ethical issues accelerate the transition towards sustainable investments. In addition, it is becoming easier to obtain information regarding a company’s ESG performance. For this reason, investors more often include corporate social performance (CSP) criteria into their decision-making process. The US and European Sustainable Investment Forum (USSIF) believes that investors such as insurance companies and pension funds will use SRI predominantly by 2020 due to their long-term horizon (Geiger et al., 2014). Sustainable indices such as Asset4good, Dow Jones Sustainability Index and the MSCI ESG Indexes illustrate the growing demand for SRI benchmarks. Increased attention and awareness regarding SRI is also noticeable within the real estate sector. Buildings account for approximately 32 percent of the total energy use and 33 percent of the CO2 emission globally (Geiger et al., 2014). It is, therefore, crucial to reduce the emission within the real estate sector in order to prevent temperatures from rising more than two degrees. Initiatives such as LEED, Energy Star and Green Star were established to reduce the emission within the sector (Fuerst, 2015). In addition, new regulation force a large number of properties to reduce their energy consumption. Properties have to become more efficient in using natural resources (Hebb et al., 2010). Within the real estate sector, this is essential to improve the overall CSP. This makes it challenging to compare the social performance from this sector against the market since a single target largely determines the performance. The unique characteristics make it essential to evaluate the sector-specific effects. For this reason, the Global Real Estate Sustainability Benchmark (GRESB) was launched in 2009 to guide real estate investment firms to improve their CSP without damaging stakeholder value. Since its establishment, the number of firms that agreed to have their business analysed by the benchmark has proliferated. However, according to Devine et al. (2017), many real estate managers worry about how sustainable improvements relate to the corporate financial performance (CFP).

Investigating the relation between CSP and CFP is often discussed within the academic literature. Nonetheless, the debate over how investing in ESG factors affects the CFP is still going on (Nagy et al., 2016). More than 2000 papers examine this link, but they repeatedly present contradictive results. For this reason, Friede et al. (2015) combined all the prior research regarding the subject in an attempt

(5)

2

to reach a consensus. The authors found empirical evidence that supports ESG investing. By combining the results of more than 2000 individual studies, they found that approximately 90 percent showed a non-negative relationship in ESG investing. Moreover, most results revealed a positive link between CSP and CFP. Nonetheless, this does not necessarily imply that high ESG scores generate a superior return for investors. A strong ESG commitment is assumed to decrease the systematic and idiosyncratic risk of the investment (Giese et al., 2017). For this reason, investors may demand a lower risk premium from SRI which leads to a lower return.

Hoepner et al. (2010) claim that the link between ESG ratings and CFP is heterogeneous across industries. Most prior literature assumes a similar relationship for the whole market. The authors investigated how CSP relates to the financial performance across industries and found significant differences between them. It suggests that previous results have to be interpreted with caution and that future research has to include industry-specific effects. This outcome may explain the mixed results presented in the academic literature. Another potential source for the misalignment is that endogeneity possibly biased prior results (Brooks & Oikonomou, 2017). Strong financial performance enables firms to improve their CSP. Wang et al. (2016) show inconsistent result from studies that investigated the endogenous relationship. However, it appears that it contributes to the ongoing debate of how CSP is linked to the overall financial performance.

Within the real estate sector, research investigating the link between CSP and CFP received little attention (e.g., Cajias et al., 2014; Fuerst, 2015; Geiger et al., 2014; Newell, 2009). A significant share of prior studies only describes the growing importance of SRI for real estate investors. Others investigated the link between ESG performance and firm value, and only a few on how it associates to the return. By using the MSCI ESG database, Cajias et al. (2014) found that ESG scores are related to lower returns in the period 2003-2010. Another study looks at how ESG associates with the return between 2011 and 2014 (Fuerst, 2015). In this period REITs with high GRESB scores outperform their cohort. However, it is important to keep in mind that both papers use different measures to display the ESG performance for REITs. There is still much unknown of how ESG links to the return of listed real estate. Filling this gap is essential for at least two reasons. First, with the growing interest in SRI, investors would probably appreciate advice from academia regarding the industry-specific effect. Investors can use this information to give more attention to SRI without diminishing their return. Second, there is no research available describing the link in most recent years. REITs are increasingly committing themselves to CSR, and this could affect the pricing of ESG. By looking at how the prices developed over the years, and particularly in the last period, it is possible to predict future trends.

(6)

3

This thesis aims to fill at least some of the existing gap in the literature. It will explore how ESG ratings associate to the return of REITs throughout the years. The ESG performance measure is provided by MSCI ESG KLD STATS, which is commonly used in related literature (e.g., Fatemi et al., 2017; Gillan et al., 2010; Cajias et al., 2014; Hoepner et al., 2010). They offer binary indicators for various criteria which determine a firm's sustainable performance. A weighted index combines these outcomes to illustrate a firms overall ESG rating. All firms are sorted into portfolios, based on their ESG performance, which are redistributed annually. The portfolios will be used to answer three questions. First, it will examine if sorting REITs on their ESG performance enables hedging possibilities. A strategy by going long in the high-ranked portfolio and short in the low-ranked will answer this. Second, is adding a factor for the ESG performance into the model. This method will show if and how ESG is priced for REITs throughout the years. Third, is to examine if ESG is priced equally for REITs as for the market. If this relationship appears to be heterogeneous, then prior work on the subject should not just be generalised to REITs. The multifactor model, with the inclusion of a liquidity factor, is used to estimate the pricing of ESG. With these results, it is possible to answer the following: “How is ESG priced for REITs compared to the market?”

The outcomes show that ESG is priced for REITs. In most years, high ESG ratings are associated with a lower return. Applying the long-short strategy generated negative excess returns in most periods, but this seems to be declining. It appears that the relation of ESG is heterogeneous between the market and REITs since most results show a different price throughout the years. The factor regression illustrates that the pricing of ESG has changed over the years. ESG was predominantly negatively related to the return in most years, but this shifted to a positive effect between 2014 and 2016. This paper is the first to illustrate a heterogeneous effect of ESG when comparing REITs to the market. It reveals the necessity to increase the attention it received so far from academia. Also, it is, as far as known, the only study that examines how the price has developed throughout the years. Investors can use this paper to adopt SRI to increase their profit.

The remainder of this thesis is organized as follows. The next section discusses the most relevant literature regarding the subject. The third section presents the hypotheses following with the methodology. A description of the data is in the fifth section. The sixth section mentions the results, robustness checks and their implication. Finally, is the conclusion.

(7)

4

2. Literature review

Despite the growing importance of SRI there is little known of how it relates to listed real estate. Sustainable investments have grown substantially over the last two decades, but the debate of how it links to the financial performance of a firm is still going on (Friede et al., 2015). While some investors remain cautious about the economic advantages of SRI one might argue that it is a firm’s CSR to include sustainability in their business model. However, not everybody agrees with this statement. According to Friedman (2007), the only social responsibility of a business is to increase its profits. Many investors still consider social concerns as a side issue and maximising the return as their prime objective (Halbritter & Dorfleitner, 2015). Nonetheless, the amount of SRI has grown significantly over the last decades which demonstrates the growing importance of sustainability in the investment decision making. According to USSIF (2016), sustainable investments, in the United States, grew with 33 percent between 2014 and 2016 to a total of 8.72 trillion dollars. The majority of firms, at least, considers to include sustainability in their business model. Actual implementation depends on several things since each company has its motives in this decision-making process.

According to Bansal & Roth (2000), it is essential to understand the motives for social responsiveness for at least two reasons. First, it gives insight into why companies adopt CSR, and this makes it is possible to predict the effectiveness of new initiatives. If, for example, companies would only respond to new legislation it is critical to focus on that. Second, it exposes the mechanisms that encourage the CSR of firms. Policymakers can use this to determine which measures are most useful to involve companies. In their study, they found three primary motivations for companies to improve their CSP: competitiveness, legitimation, and ecological responsibility. Competitiveness means the potential to increase the long-term profitability with sustainable investments. Some examples that may increase the competitiveness are green marketing, energy and waste management, and developing ‘eco products'. Firms motivated by legitimation link their ecological responsiveness to the current regulation, norms, beliefs or values. They react mostly to external factors to avoid fines or unfavourable publicity instead of acting by internal motivation. Motivation driven by ecological responsibility are companies that acknowledge their social obligation. The decisions of such firms are not solely driven by maximising their financial performance. They are characterised by initiatives such as developing less lucrative green products, donating to charity and recycling. Understanding these motives helps to commit a different kind of firms in improving their social footprint. However, in reality, motivations to improve ESG ratings are mixed. A combination of these three motivations probably drives each company.

(8)

5

A report from the Deutsche Bank (2017) describes two fundamental changes that explain the rise of ESG investing. First, investors want an active approach in managing their capital. They want to use their expertise to invest their recourses in an objective-driven way. Second, investors recognise that the profits of sustainable investments are increasing. However, these are not the sole motivations for a company to improve their social footprint. Concerns about global warming fuels the interest of ESG issues for both the public opinion and government. Reducing the CO2 production is critical to prevent temperatures from rising too far. Increased regulation and social pressure help to speed up the transition towards more sustainable investments. Nonetheless, profits appear to remain the primary concern for most investors.

Throughout the years' many researchers investigated how ESG criteria are related to a firm's CFP. Many studies present different outcomes which fuelled the disagreement among investors about the financial benefits of including ESG in the business model. For this reason, Friede et al. (2015) combined the findings of more than 2000 empirical studies discussing the subject. With this, they hoped to provide an unambiguously result about the relation of ESG criteria on the financial performance. The results leave little room for discussion since roughly 90 percent of the studies find a nonnegative relation. Moreover, the majority of prior research found that improving the ESG criteria has a positive effect on the CFP. The positive results are observable across different regions, asset classes and methods. The only exception are studies that include SRI in a mixed portfolio. Overlapping effects of risk, costs for portfolio implementation and construction constraints are presumed to bias the real ESG performance. The authors suspect that portfolio studies are the primary source of misalignment regarding the subject. Another noteworthy finding is that the effects of ESG investing is unusually high in North America compared to other regions. Results are of particular importance in this paper since it combines the outcomes of such a large number of studies. It clearly shows that investing in ESG criteria has a positive effect on the overall financial performance of a firm.

Nonetheless, this does not necessarily imply that high ESG scores improve the excess return. Prior research showed that SRI might have a negative influence on the return (e.g., Lesser et al., 2016; Nofsinger & Varma, 2014). In their studies, they show that, in normal market conditions, SRI funds generate negative abnormal returns, while they perform better during a market crisis compared to conventional funds. They argue that sustainable funds are weakly, negatively, related to systematic risk and are therefore less vulnerable to poor economic times. Cajias et al. (2014) found the same negative relationship of ESG ratings on the excess return of REITs. According to them, the explanation for this is something else. A high ESG score often is associated with reputational benefits. It helps a

(9)

6

firm to improve their relationship with customers, stakeholders, suppliers, employees and the community. This creates more stability, and it reduces the risk perception and thus the cost of capital. Investors might be prepared to accept a lower return from companies with a high ESG score. Research of Geiger et al. (2014) measured the return from sustainable real estate between 2003 and 2010. The market for sustainable real estate exhibited low returns, especially during the latest years of their study. It appealed mostly risk-averse investors who accepted a lower return for the relatively low risk of the investment. These findings are in line with the assumptions from Cajias et al. (2014) who tie SRI to lower risk. There are also papers which claim that ESG scores improve the excess return of a firm. Polbennikov et al. (2016) found that sustainable investments are not negatively related to the excess return of corporate bond portfolios. Moreover, a portfolio consisting of high-rated ESG stocks outperformed the low-rated portfolio over the period 2007 to 2015. Nagy et al. (2016) found similar results for the stock market. They looked if ESG can add alpha by applying the tilt and momentum strategy. The tilt strategy uses stocks with higher current ESG ratings, and the momentum strategy uses stocks that improved their rating over the last 12 months. These strategies attract a different type of investor. High ESG scores reduce the systematic and idiosyncratic risk, which is especially appealing for long-term investors. Looking at the momentum is more for the short term. Improved ratings show that a firm is better suited to avoid potential risk. It reduces possible forthcoming obligations which enhances the value of a firm. The market rapidly adjusts the share prices to its actual value, but it is possible to take advantage of it. Both approaches were able to generate an alpha that outperformed the MSCI World Index benchmark between 2008 and 2015. Fuerst (2015) investigated the link between sustainability and the stock performance of REITs, by using the total score from the GRESB. Results show a positive link between high GRESB scores and excess return of REITs for the period 2011-2014. Altogether, it appears that there is still no clear consensus about how ESG relates to the excess return.

Reversed causality between a firms financial and social performance might explain the mixed results presented in literature (Brooks & Oikonomou, 2017). Outcomes from studies that did not account for endogeneity could well be biased. According to Ng & Rezaee (2015), the CSP could, at least, partly be determined by the financial performance. A strong financial position could, for instance, support more social investments. However, studies that examined the endogenous relationship present conflicting results (Wang et al., 2016). Al-Tuwaijri et al. (2004) investigate the interrelation between environmental performance and economic performance. The results in their paper suggest that high profitability is associated with good environmental performance, supporting the claim for endogeneity issues. Nevertheless, there are also papers that claim that their results do not suffer from

(10)

7

reversed causality problems (e.g., Attig et al., 2013; Ng & Rezaee, 2015; Fatemi et al., 2018). Researchers fail to reach a consensus if it is essential to consider endogeneity while investigating the link between CSP and the financial performance. Looking for the potential bias in real estate studies, Eichholtz et al. (2012) and Deng & Wu (2014) suggest that it tends to overestimate the effect of social investments on the financial performance. However, Cajias et al. (2014) do not detect any endogeneity issues in their study. According to Eichholtz et al. (2012), the market may already reflect the CSP of REITs in the stock price. Using the stock performance would, therefore, reduce the potential threat of endogeneity issues. In addition, most of the prior work regarding the subject does not account for reversed causality. Although endogeneity issues may be present, prior studies show that it is not a vital determinant of the results.

Most studies assume a homogenous effect of CSP on CFP across industries. However, some studies show, at least on some level, diversity of this link between sectors, but they have been mostly ignored (e.g., Barnett, 2007; Godfrey and Hatch, 2007). Hoepner et al. (2010) were the first to investigate if there is a heterogeneous effect of CSP on CFP across industries. In their paper, they state that it is meaningful to understand the causality of this relationship for two reasons. First, if heterogeneity exists between industries, it would influence how to interpret current and future research regarding the subject. Outcomes from studies that assumed a homogenous effect should be evaluated with more caution. Second, with the growing interest in CSR, it has become more important to provide guidance in how firms should implement this in their business model without hampering the CFP. If there is a mixed effect, then stakeholders should make the proper adjustments according to their industry characteristics. By testing for the effect of CSP on CFP across industries, the authors find that the relationship is indeed heterogeneous. According to them, the main reasons causing the variation are the industries relation with their end consumer, the potential damage for environmental and social factors, the reliance on specific stakeholders, and which services or products they offer. Altogether, these results show that most prior research regarding the subject cannot be generalised to particular industries. The research field describing how CSP relates to CFP is still rather young in real estate studies (Friede et al., 2015). Therefore, it is essential to investigate the industry-specific effects of ESG criteria on the financial performance of REITs.

Sustainability is becoming increasingly important within the real estate sector (Fuerst, 2015). Buildings account for approximately a third of the global energy use and CO2 emission. The new regulation is set to reduce the emission, which fuels the demand for sustainable properties. Moreover, stakeholders such as tenants, owners and investors growingly expect more transparency to ESG issues

(11)

8

from the industry. Improving the sustainability in the real sector goes hand in hand with reducing the emission of properties. This sector-specific effect makes it essential to investigate how CSP relates to the financial performance within the industry (Geiger et al., 2014). For this reason, several pension funds launched the GRESB in 2009 to provide more understanding of how environmental, social and governance risks relates to listed real estate. The benchmark globally assesses ESG criteria of real estate assets (GRESB Report, 2015). GRESB members can use this data to include sustainability as an investment criterion. Since 2010, the number of REITs who agreed to have their firm inspected by the GRESB has increased from 198 to 759 (Devine & Yönder, 2017). Participants of the benchmark have become more advanced on ESG issues over the years. It is more common to reduce the consumption of energy, waste, and water than it was only a few years ago (GRESB Report, 2015). The GRESB provides a knowledge centre from the sustainable solutions of their participants. This awareness increases the development of sustainable investments within the sector. Besides, it learns REITs how they can simultaneously improve their ESG rating and financial performance. Some prior work already showed a positive relationship. Cajias et al. (2014) found that high ESG scores have a positive effect on a company’s market value. In addition, Devine & Yönder (2017) present similar results and also claim that sustainable REITs have less systematic risk and higher premiums to NAV. Innovations made it easier to include ESG criteria in the business model without harming the shareholder value. Stakeholders increasingly include sustainability as a criteria while evaluating properties (GRESB Report, 2015). Energy efficient properties positively affect the return on investment (Fuerst, 2015). This is caused, among other things, by rental premiums, higher occupancy rates and a lower cost of capital. Also, the increased attention and regulation regarding the property market influences the risk of the investment (Geiger et al., 2014). Possible future fines for high usage buildings are of importance in the investment decision-making process. Sustainability is, due to new technologies, regulation and increased stakeholder attention, rapidly changing the real estate sector. It is, therefore, important to continuously monitor how this affects the CFP of REITs. There is evidence that the link between sustainability and financial performance is becoming increasingly positive.

However, the research field is still relatively green for the real estate sector. It is still unclear if ESG is priced for REITs, if the pricing has changed over the years, and if the industry is comparable to the market. This thesis aims to answer these questions. With the growing importance of SRI, it is now more important than ever to guide investors in how they should assess a company’s CSP. As far as known, it is the first paper that constructs portfolios based on the ESG performance of listed real estate. Applying the tilt and momentum strategy will show both the short- as long-term possibilities. In addition, the ESG factor shows the price of ESG throughout the years. The relevance of SRI is

(12)

9

increasing, so how can one assume that the price remains equal? Moreover, comparing listed real estate against the market is essential to evaluate the current state of literature regarding the subject. If there appears to be a heterogeneous effect than there remain only a few papers that describe the relationship between ESG and REITs. This outcome would make it essential to fill this gap as soon as possible. Overall, the results will reveal new perspectives on a subject that has received too little attention from academia.

3. Hypotheses

This thesis investigates the relationship of ESG ratings on the return for listed real estate companies. The total ESG rating is constructed by making an index based on the outcomes of the binary indicators. Prior research (e.g., Fuerst et al., 2015; Geiger et al., 2014; Cajias et al., 2014) has primarily focussed on how ESG affects the overall CFP of a firm. Even though proving this relationship by itself provides useful insights, it gives little information about how investors can use this knowledge to outperform the market. This thesis is the first work to implement a long-short strategy to investigate possible arbitrage opportunities. Multiple portfolios are constructed based on the ESG performance of REITs. By going long in the portfolio with the highest ESG score and short in the one with the lowest rating it is possible to detect if such a strategy can outperform the market. Prior research proved that SRI improves the total CFP for a firm (Friede et al., 2015). However, there is still no clear consensus about the sole effect of the ratings on the excess return. There are, for instance, studies that showed that SRI funds underperformed conventional funds during non-crisis periods (Lesser et al., 2016; Nofsinger & Varma, 2014). Looking at the alpha in the long-short portfolio will show if it is possible to use ESG ratings to outperform the market. This leads to the first hypothesis (H1).

H0: The long-short strategy, based on REITs individual ESG performance, does not generate a significant alpha

Most of the prior research regarding ESG scores assumed a homogeneous effect across industries. However, there is reason to assume that this is not always true. Some studies show that the excess return for REITs behaves differently than it does for the market. Anzinger et al. (2017) found, for instance, that a quality factor measuring the profitability, growth, safety and payout is more persistent for real estate companies than it is for other equities. Also, research from Hoepner et al. (2010) showed that the link of CSP and CFP is heterogeneous across industries. The property market is accountable for around a third of the global CO2 emission and energy consumption (Fuerst, 2015).

(13)

10

New regulation and technologies aiming to reduce the pollution are sector-specific. For this reason, it could well be that the effect of ESG differs between REITs and the market. There is still little work available describing the relationship of ESG ratings and the excess return for REITs. If the long-short portfolio follows another trend for REITs as it does for the market, then studies that assumed a homogeneous effect have to be evaluated with care. This information is of great importance for investors that include REITs in their portfolio. This leads to the second hypothesis (H2).

H0: The effect of ESG is the same for REITs as for the market

Interest in ESG criteria has grown substantially over the past years (Nagy et al., 2016). More stakeholders are focussing on ‘going green’, and the environmental regulation has increased. There are more international agreements, such as the Paris agreement, that oblige firms to invest in sustainability. The public opinion is also increasingly growing towards sustainability. More individuals expect that companies meet up to their environmental responsibilities. But not only the environmental awareness has grown over the years. Board diversity, business ethics, employee relations and human rights are only some examples of topics that have become more relevant over the last couple of years (Deutsche Bank, 2017). Moreover, new technologies make it easier to build properties that are more sustainable without damaging shareholder value. Therefore, expectations are that the importance of ESG investing has grown in most recent years for listed real estate. The increased relevance concerning the topic is expected to improve the return for REITs that perform well on ESG criteria. This leads to the third hypothesis (H3).

H0: The relationship between ESG and REITs has not changed over time

4. Methodology

4.1. ESG index

The main goal of this thesis is to investigate the effect of ESG ratings on a firm's excess return and in particular for REITs. It is therefore essential to conduct an index capable of estimating a firms ESG performance, based on the binary indicators provided by Morgan Stanley Capital International (MSCI) ESG Research. By combining the outcomes of these indicators, it is possible to measure the annual ESG performance of each firm. There are two possible ways to construct such an index. The first approach is to create an unweighted index where each firm gets a rating between 0 and 100 for ESG issues. A score of 50 suggests a neutral score, above 50 implies more strengths than concerns and vice

(14)

11

versa for a score below. This formula allows to compare the overall score, based on the strengths and weaknesses of each firm, using a continues scale. However, many researchers criticised using this approach to measure a firms ESG performance (e.g., Griffin and Mahon, 1997; Simpson and Kohers, 2002). Prior studies found that the relevance of each criterion differs across industries. These assumptions are in line with the empirical results that Cajias et al. (2014) present in their study. They used a weighted and un-weighted index to investigate the effect of ESG ratings on the return of REITs. In all regressions, both indexes followed the same pattern, but the un-weighted index presented weaker results than the weighted. For this reason, this thesis will only use a weighted index to measure the ESG performance of each firm. Using a current weighted index is often used in related literature (e.g., Bourassa et al., 2006; Cajias et al., 2014; Geiger et al., 2014). In constructing the weights, this work closely follows their work, but with some adjustments. This study includes more industries than just REITs. It is therefore important to calculate the weight of each indicator by industry:

Wjkt=

∑𝑛_𝑘=1∑𝑛_𝑗=1𝑆it + ∑𝑛_𝑘=1∑𝑛_𝑗=1𝐶it

∑𝑛_𝑘=1∑𝑛_𝑗=1∑𝑛_𝑖=1𝑆it + ∑𝑛_𝑘=1∑𝑛_𝑗=1∑𝑛_𝑖=1𝐶it

This method assigns an industry weight to each ESG indicator. Where Sit and Cit are the individual

binary indicators for firmi at time t representing strengths and concerns respectively, and Wjkt is the

weight for indicator j in industry k at time t. It represents the industry sum of all the individual binary

counts for each indicator divided by the total sum of criteria and i firms within an industry. With this

it is possible to estimate the total ESG score for each company using the following formula:

ESGit= (∑ Sit n j=1 ∙W jkt − ∑ 𝐶it 𝑛 𝑗=1 ∙ 𝑊 jkt ) + 1

Where ESGit indicates the total ESG score for firm i at time t. It is the sum of all the individual binary

strengths and concerns times their industry criterion weight. In this formula, a score of 1 illustrates a neutral position while an outcome above 1 shows more strengths than concerns and vice versa for a score below. It is a current annual industry weighted index to estimate a firms ESG performance.

4.2. Portfolios

Constructing portfolios help to determine if ESG is related to the return. Companies are annually sorted by their ESG score, relative to their industry, and assigned to a portfolio. It is essential to consider the relative industry performance in constructing the portfolios. There are significant

(15)

12

differences between the ESG performance of industries based on regulation or an increased relevance of some indicators (e.g., ESG Ratings Methodology Executive Summary, 2017; Hoepner et al., 2010). Firms cannot reasonably be compared across industries. For this reason, all the firms get sorted into five portfolios based on their relative ESG performance. This paper uses the calendar-time portfolio return, similar to the approach of Anzinger et al. (2017) and Asness et al. (2017), to investigate the pricing of ESG. For each month, the mean excess return of all the portfolios is calculated. To examine if the ESG score relates to the excess return a long-short strategy will be implemented by going long in the portfolio with the highest ESG ratings while going short on the lowest. The spread of REITs portfolio return (RP) between the two portfolios is calculated using the following method:

RP6t = RP5t – RP1t

Where RP6 is constructed by subtracting the mean excess return from the portfolio of REITs with the lowest ESG score (RP1) from the highest portfolio (RP5), for each month (t). A significant alpha, using

RP6 as a dependent variable in the multifactor model, would illustrate that ESG is priced for REITs. Moreover, it would demonstrate if this strategy enables arbitrage opportunities. A positive alpha indicates that the long-short portfolio outperforms the market and vice versa for a negative sign. As stated before, high ESG scores are presumed to reduce the risk and are therefore often used by investors with a long-term horizon. Applying the long-short strategy on the current ESG score is equivalent to the tilt strategy used by Nagy et al. (2016) & Geiger et al. (2017). Following their work, by implementing the momentum strategy, it is possible to examine the short-term potential of ESG investing. Therefore, all firms are also sorted into portfolios based on their annual change in ESG performance. The construction of portfolios and their average excess return is done in the same way as explained above. The following method is used to investigate the price of annual changes in ESG performance for REITs:

RP16t = RP15t – RP11t

Where RP16 is constructed by subtracting the average excess return of REITs in the portfolio with the lowest annual change in ESG rating (RP11) from the highest portfolio (RP15), for each month (t). The

variable of interest is, just as above, the alpha. This method shows if the annual change in REITs ESG performance is applicable for short-term investments.

(16)

13

Another goal of this thesis is to investigate if ESG is priced similar for the market as it is for REITs. If there are differences than this has implications in how to interpret existing work regarding the subject. For this reason, this paper constructs the same long-short portfolios for all the US firms, excluding REITs, for which ESG data is available. For convenience sake, this sample is referred to as the market for the remainder of this thesis. The difference in the market portfolio return (MP) is constructed using the following methods:

MP6t = MP5t – MP1t MP16t = MP15t – MP11t

Where MP(1)6 is constructed by subtracting the average market excess return of the portfolio with the lowest (annual change in) ESG rating (MP(1)1) from the highest portfolio (MP(1)5), for each month (t). Constructing an equivalent portfolio for the market allows investigating if the pricing of ESG follows

the same trend over the years in both samples.

It is the first time that the long-short portfolios are applied to listed real estate. Using this approach will show if it is possible to use ESG criteria to increase the excess return of REITs. Moreover, comparing the results against the market is also unique. This method will, therefore, provide new insights into how investors should assess REITs ESG performance.

4.3. Multifactor model

Fama and French introduced a three-factor model to capture patterns in returns. These factors are the market return, size, and value for a company. Carhart introduced an additional variable to capture the momentum of returns (Fama and French, 2012). Pástor & Stambaugh (2003) showed that liquidity is an important factor for stock prices. Stock returns are related to innovations in aggregate liquidity. Combining these factors will generate the following formula:

Rkt = α + β1(Rmkt)t + β2(SMB)t + β3(HML)t + β4(MOM)t + β5(LIQ)t

Where:

Rkt = the monthly t mean excess return for portfolio k

α = the alpha

(Rmkt)t = the monthly t excess return of the market

(17)

14 (HML)t = the monthly t high-minus-low factor

(MOM)t = the monthly t momentum factor

(LIQ)t = the monthly t liquidity factor

This model is wildly used by academics to explain excess return. In this model, the beta of each variable reveals how the portfolio is related to that factor. The coefficient for the excess return of the market measures the systematic risk of the investment compared to the market. A positive small-minus-big factor implies that the portfolio tilts towards small-cap firms. High-minus-low is the ratio of book value to market value for a portfolio and momentum measures the one-year momentum of the returns (Fama and French, 2012). The four-factor model is wildly used, but it does not always include the liquidity factor. However, a lack of liquidity characterises the real estate sector (Ametefe et al., 2015). For that reason, including this factor could improve the overall explanatory power of the model. The factor measures the portfolio's sensitivity of aggregated liquidity to the return.

4.4. ESG factor

A style factor for ESG is included in the multifactor model to illustrate the pricing over time. Instead of using the calendar long-short portfolio return as a dependent variable, the spread is now included in the multifactor model. This will create the following formula:

Rit = α + β1(Rmkt)t + β2(SMB)t + β3(HML)t + β4(MOM)t + β5(LIQ)t + β6(ESGm)t

Where Rit is the excess return of firm i in month t. The new coefficient ESGm is the average monthly

long-short strategy excess return of portfolio m which is equal to either RP6, RP16, MP6 or MP16. The

inclusion of ESG as a factor in the model is similar to the method used by Anzinger et al. (2017) and Asness et al. (2017) who created a factor to capture how quality relates to return. It is, however, the first time that ESG is included into the model as a factor for REITs. All the other coefficients are similar to the model presented above. This method includes all firms instead of only the monthly average as before. Outcomes of this regression show how the relation of return and ESG developed throughout the years. Furthermore, comparing the results from REITs with the market helps to investigate if the price follows the same trend. As a final point, this second approach using a market risk factor model also offer the opportunity to partially address the endogeneity issue because it measures the exposure to a market wide risk factor as opposed to establishing a link between performance and ESG score at the individual company level.

(18)

15

5. Data

5.1. ESG data

The primary goal of this thesis is to illustrate how ESG is related to the excess return of REITs. It is essential to measure the ESG score for each company, to analyse if there is a significant reaction to the stock return. MSCI ESG KLD STATS offers an annual data set consisting of binary indicators which show how a firm performed on environmental, social and governance issues. It is one of the most extended consecutive data sets available as it was founded in 1991 (MSCI ESG Research Datasets Methodology, 2015). A research team, consisting of 185 employees worldwide, assesses how a firm handles its social responsibility. It is one of the longest existing and extensive sources for ESG data and widely used by academia (e.g., Fatemi et al., 2017; Gillan et al., 2010; Cajias et al., 2014; Hoepner et al., 2010).

Seven pillars determine the overall ESG performance. These are environment, community, diversity, employee relations, human rights, product and corporate governance (MSCI ESG Research Datasets Methodology, 2015). Each of these seven pillars has various criteria that show how well a company performed on it. An indicator can either have a positive or negative effect on the ESG rating. Each indicator represents the binary score for either a strength or a concern. If a firm fits the criteria, it receives a score of one and zero otherwise. The score of one has a positive effect on the ESG score if it represents a strength and vice versa for a concern. Indicators have varied over the years as well in numbers as content. A detailed description regarding all the indicators is included in the appendix.

There are multiple methods to create an index capable of assessing a firms ESG performance. For example, KLD (Kinder, Lydenberg, Domini) created an index based on an industries Key Issue Hierarchy consisting of 37 variables (Melas et al., 2018). The total ESG score, ranging from 0-10, is the sum of adding and subtracting all the strengths and concerns respectively. However, many of these key issues were introduced after the start of this study. This study, therefore, includes all the available indicators of the dataset. The industry relevance of each criterion is determined by assigning weights. It allows to include all the available ESG data, and still consider the industry importance of each criterion (Hoepner et al., 2010). This thesis uses the Global Industry Classification Standard (GICS) to identify industries.

(19)

16

5.2. Data preparation

This thesis uses ESG data from the MSCI ESG KLD STATS database provided by WHARTON. All the available data between 2000 and 2016 is initially included in the sample. There is no data available yet for 2017. This study focusses on REITs located in the United States thus firms from abroad had to be excluded. After eliminating these firms, about 39,000 observations were remaining. Table 1 shows the summary statistics from the sample for each sector, based on the classification of GICS. Illustrating, by year, the number of firms in each sector of which ESG data is available. There is little information available for the years 2000, 2001 and 2002. Especially for the real estate sector since there is only data available for 56 firms during those years. Therefore, this study excludes these years and focusses on how ESG relates to return between 2003 and 2016. Multiple sub-periods are included to investigate the developments throughout the years. These periods are 2003-2006, 2007-2010, 2011-2013, and 2014-2016.

For REITs, the total sum of binary indicators, over each period, and their total ESG score is illustrated in Table 2. Throughout the years, the mean of total strengths is primarily rising for issues regarding environment and social while it decreases for corporate governance. Between 2003 and 2013 the average sum of total concerns increases for social and corporate governance aspects while both decline in the latest period. The number of concerns related to environment decreases throughout the period. After a decline in the ESG score of REITs between 2003 and 2013, it started to increase in the last period. For the years 2014-2016, the weighted index exceeded the value of one, and this

(20)

17

illustrates more strengths than concerns. Investigating the summary statistics give more intuition about the ESG performance of REITs. It shows that the total ESG score changed over the years. Moreover, the composition of the strengths and concerns for the three aspects has also evolved. Table 2, therefore, strengthens the expectation that the link between ESG and the excess return of REITs has changed over time.

WRDS provides monthly data for the stock return, risk-free rate, market proxy, and other factors. Stock return is obtained via CRSP and the others from Fama French & Liquidity factors. The Nareit index used to check for robustness, is collected from Datastream. After subtracting the risk-free rate from the holding period return, the excess return is winsorized at the 0.5% and 99.5% percentiles, similar for trimming the total ESG score. Hereafter, all the variables are combined into one file to prepare the data. First, the annual change in ESG performance is calculated for each firm. The natural logarithm is used to achieve more significance and numerical stability. Second, is the composition of annually rebalanced portfolios using a firms ESG performance, compared to their industry. This method is used

(21)

18

for as well the current ESG score as for the annual change in ratings. For both approaches, companies are sorted into five portfolios from the worst to the best performers respectively. Firms are excluded from the portfolios if the sum of all their strengths and concerns for that year is equal to zero. A score of zero implies that there is no data available to measure their ESG performance thus including them would bias the results. Third, is calculating the average monthly excess return of each portfolio, and the spread of subtracting the return from the portfolio with the lowest ESG performance from the highest. Following the work of Asness et al. (2017), results are collapsed to obtain the calendar-time portfolio return. This file allows performing the portfolio regressions while using the original file for the factor regressions. Standard errors, in all regressions, are adjusted for heteroscedasticity and autocorrelation (Peterson, 2009).

(22)

19

6. Results

This section presents the results of investigating how ESG ratings relate to the excess return by using different methods. The first part discusses the results of the calendar-time series portfolio regression based on the total current ESG score. In the second part, the long-short strategy from the annual change in ESG rating is discussed. Illustrations of the ESG factor for a firms current score and annual change are in the third part. It ends with a discussion about the implications of the results by linking them to the proposed hypotheses for this thesis.

6.1. Results multifactor model portfolio regression using current ESG score

6.1.1. REITs

To investigate the pricing of ESG, and what the effects are, all the REITs are sorted into five portfolios based on their current total ESG score. Table 3 illustrates the regression coefficients for each portfolio over the period 2003 to 2016. In this table, the alpha is the primary source of information to examine the effect of ESG ratings on the excess return. The alphas are significantly positive for RP1, RP2 and RP4 and insignificantly positive for RP3 and RP5. The difference portfolio, i.e. RP6, yields a monthly alpha of -0.491% which is significant at the 1% level. This outcome means that applying the long-short strategy on REITs current ESG score would, on average, underperform the market by 5.74% a year. Furthermore, the results in column 6 of Table 3 show positive coefficients for momentum and liquidity

(23)

20

at the 1% and 10% level respectively. Implying that REITs who perform well on ESG criteria are characterised by more momentum and liquidity than their counterparts.

Multiple sub regressions are included to gain more insights into the relationship of ESG on the excess return of REITs throughout the years. Table 4 present the outcomes of the sub-periods. Looking at the observation period of 2003 till 2006 in column 1, the alpha for the long-short portfolio generates an, at the 5% level significant, alpha of -0.784% a month. Similar to the full period results are the signs and significance of the factors for momentum and liquidity. Also, the small-minus-big coefficient is negatively significant at the 1% level during this time frame. Indicating that, at that time, relatively more REITs with a larger market capitalisation acquired a higher ESG score. Column 2 illustrates the sub regression between 2007 and 2010. Also in those years, is the alpha significantly negative at the 5% level. The momentum factor is, again, positive and significant at the 1% level just as the beta for the excess return on the market. The high-minus-low coefficient is negatively significant at the 5% level which suggests that growth companies had higher ESG scores than value stocks in those years. Looking at the observations of 2011-2013 in column 3, the alpha is, again, negative and significant at the 5% level. Applying this strategy underperformed the market, on average, by -0.669% a month during this period. None of the independent variables differs significantly over this time frame. Column 4 presents the outcomes of the regression between 2014 and 2016. For the first time, there is no significant alpha for the long-short portfolio. The factors for momentum and liquidity are positive at the 5 and 10 percent level.

(24)

21

6.1.2. Market

All the US firms of which ESG data was available, excluding REITs, are included in this model. By using the same methodology, it is possible to compare REITs against other sectors. Table 5 illustrates the results over different periods. Again, the variable of interest is the alpha in each column. Looking at the first column, which represents the full period, the sign of the alpha is negative and significant at the 10 percent level. It shows that the long-short strategy on average underperformed the market by -0.109% per month or -1.3% annually. Despite having a similar sign to the results for REITs over the entire period, both the size and significance of the estimate are smaller. The momentum coefficient is also significantly positive at the 1% level but, likewise, smaller in size than it is for REITs. Unlike as for REITs, is the negative small-minus-big factor which is also significant at the 1% level. Moreover, all sub-periods illustrate the same, negative, sign and significance for the small-minus-big coefficient. This suggests that for the market, firms with a bigger capitalisation score high on ESG issues. Column 2 and 3, illustrating sub-period 2003-2006 and 2007-2010, show a positive but insignificant alpha. This outcome is contradictive to the findings for REITs at the same time span. The results in column 4, covering the period 2011 till 2013, show a negative alpha that is significant at the 1% level. It resembles the outcome for REITs in the same period. Column 5 illustrates some different results. During the years 2014 till 2016, there was no significant alpha for REITs. However, there is a significant, negative, alpha in the market at that moment in time.

(25)

22 Table 6: Main results long-short portfolio regression by sector

This table shows the monthly calendar-time portfolio regressions. All the firms are annually redistributed into five portfolios based on their current ESG score. The dependent variable are all the sectors long-short portfolio return (SP6) in each period. It is the spread of subtracting excess return of the low-rated (SP1) from the high-rated (SP5). Each column shows a different sector. This model includes five factors as independent variables: Excess Return on the Market (Rmkt), Small-Minus-Big Return (SMB), High-Minus-Low Return (HML), Momentum factor (MOM), and Innovations in aggregated liquidity (LIQ). Alpha (α) is the intercept in this model. The sample period runs from 2003 to 2016. Standard errors are corrected for heteroscedasticity and autocorrelation using a lag length of five years. Standard errors are in the parentheses and *_,**_{, and}***_are indicators of statistical significance at the 10%, 5%, and 1% levels, respectively.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

Energy Materials Industrials Consumer Discretionary Consumer Staples Health Care Financials Information Technology Telecommunication Services Utilities REITs Excess Return on the Market -0.0894 -0.0665* _0.0106 _0.0147 _-0.0814 _-0.088*** _0.0541 _0.0519 _-0.239** _-0.0447 _0.0624 (0.0651) (0.0392) (0.0250) (0.0411) (0.0619) (0.0285) (0.0572) (0.0432) (0.108) (0.0600) (0.0451) Small-Minus-Big Return 0.0615 -0.0867 -0.0500 -0.124*** _-0.430*** _-0.282*** _-0.122* _-0.413*** _-0.498** _-0.138 _-0.0624 (0.0743) (0.0728) (0.103) (0.0355) (0.154) (0.0805) (0.0692) (0.0453) (0.199) (0.0925) (0.0440) High-Minus-Low Return -0.0556 0.0296 0.0274 0.0643 -0.0215 -0.121** _0.0893*** _0.0300 _-0.164** _-0.0566 _-0.0126 (0.0635) (0.0572) (0.0540) (0.0646) (0.118) (0.0566) (0.0317) (0.0489) (0.0763) (0.0590) (0.0620) Momentum Factor -0.123** _0.00916 _0.0627 _0.0380 _0.0642 _-0.0270 _0.00640 _0.0992*** _0.0278 _0.0558 _0.123*** (0.0538) (0.0515) (0.0444) (0.0236) (0.0482) (0.0533) (0.0203) (0.0149) (0.0614) (0.0605) (0.0235) Innovations in aggregate liquidity 0.0457 -0.115*** _-0.0285* _-0.0237 _-0.00635 _0.00716 _0.00520 _0.0168 _-0.140** _-0.00874 _0.0536* (0.0299) (0.0402) (0.0155) (0.0252) (0.0390) (0.0326) (0.0263) (0.0235) (0.0699) (0.0188) (0.0306) Alpha -0.005*** _-0.00211 _-0.0034*** _-0.00193** _0.00605** _0.00113 _-0.000142 _-0.00129 _0.00617 _0.00326* _-0.005*** (0.00107) (0.00195) (0.00119) (0.000853) (0.00275) (0.00159) (0.00216) (0.000835) (0.00540) (0.00167) (0.00110) Observations 168 168 168 168 168 168 168 168 168 168 168 R2 _0.033 _0.062 _0.032 _0.033 _0.130 _0.117 _0.045 _0.211 _0.081 _0.086 _0.064

(26)

23

6.1.3. Sector analysis

A sectoral analysis is included to gain more insight in specific effects of ESG ratings on the excess return within sectors. For every sector the long-short portfolio is included which covers 2003 till 2016. Table 6 presents the results, where the alpha is the variable of interest, for each sector. Results show that ESG ratings did not significantly influence the alpha for the sectors Material, Health Care, Financials, Information Technology, and Telecommunication Services. The long-short strategy generated a significant, negative, alpha for, besides REITs, the Energy, Industrial and Consumer Discretionary sectors. However, applying this strategy also had a positive effect in some sectors. Column 5 and 10 show significant positive alphas at the 5 and 10 percent level respectively. The results show that the effect of ESG differs between sectors.

6.2. Results multifactor model portfolio regression using ESG growth

Another approach is not to look at the total current ESG score but instead focus on the yearly change in the ESG score. This approach will give more insights into the short-term potential of ESG investing. In this section, the five portfolios are constructed based on the yearly change in ESG performance.

6.2.1. REITs

Table 7 shows the results from the long-short portfolios for REITs, based on the annual ESG growth, over multiple periods in time. Looking at column 1, the alpha is negative and significant at the 1% level. Between 2003 and 2016 this strategy would, on average, underperform the market by -0.399% a month or -4.68% annually. These results are comparable to the outcomes of the regression for the total ESG score. Likewise similar is the momentum factor which is also positively significant at the 1% level. It also reports a significant positive coefficient for the excess return on the market. This suggests that REITs with growing ESG scores have more co-movement with the market than their counterparts. Furthermore, looking at the sub regressions, only column 3 and 5 show a significant alpha, both having a negative sign and they are significant at the 1 and 5 percent level respectively. The main difference between the outcomes from the tilt and momentum strategy is the alpha coefficient between 2014 and 2016. Where the portfolio constructed on the current ESG score show a positive, insignificant alpha in those years, it is negatively significant in the momentum portfolio.

(27)

(28)

25

6.2.2. Market

The composed long-short portfolios for the market, based on the annual ESG growth, are presented in Table 8. Looking at results of the alpha in all columns, resemble the outcomes in the regressions on the portfolio construction of the markets total ESG score. Despite some variation in the size and significance, the results show the same trend in the alpha movement for both approaches. Column 1 shows that the factor for momentum is positive and significant at the 5% level in the total period. Similar to the tilt strategy, the results appear to differ between REITs and the market in most periods. For both samples, the alpha was insignificant in 2003-2006 but with an opposite sign. The results differ between 2007 and 2013 and are relatively similar in the latest period. Also, the size and significance of the alpha are larger for REITs than it is for the market looking at the full period.

6.3. ESG market factor regression

In this section, a factor for ESG is added to the model. By extending the multifactor model with the independent ESG variable, it is possible to investigate the development of the price for ESG throughout the years. The tables in this section can be found in the appendix.

6.3.1. REITs

The output for the regression, illustrating a factor for the current ESG score of REITs, is in Table A1 where the variable of interest is RP6. Looking at column 1, the ESG scores had an adverse effect on the excess return which is significant at the 1% level. Column 2 and 3 show similar results. For the periods 2003-2006 and 2007-2010, ESG scores had an adverse effect on the excess return. However, this changes in the following periods. The coefficient for ESG is positive but insignificant between the years 2011 till 2013. Moreover, column 5 shows that it becomes significantly positive, at the 1% level, in most recent years. An ESG factor based on the momentum strategy, coefficient RP16, for REITs is presented in Table A2. The results are somewhat contradictive to the current ESG score, except in the latest period. Looking at the full-time frame in column 1, the ESG growth factor was positive but insignificant. It was positively significant in 2003 to 2006 at the 1% level which is the opposite of the results for the current ESG factor at that time. It suggests that, considering the relatively large sign for the coefficient, that improving the ESG score was beneficial during that period. Between the years 2011 and 2013, the ESG momentum factor had a significant adverse effect on the excess return. However, this changed in the following years. Just as with the current ESG factor, the coefficient for the growing ESG factor is positive and significant at the 1% level. Overall, the results suggest that the price of ESG is changing for REITs.

(29)

26

6.3.2. Market

The results for the market, are presented in Table A3 and Table A4. The ESG current score factor, MP6, and the ESG growth factor, MP16, show the pricing of ESG within the market. Neither the current score or the annual change in ESG is priced over the full period as is shown in column 1 of both tables. Looking at Table A3, the coefficient MP6 is significantly negative in 2003 till 2006 and positive between 2011 and 2013. It appears that the price for ESG is different for the market than it is for REITs. Only the outcomes during the sub-period of 2003 till 2006 are comparable. Furthermore, the results in column 5 of Table A3 illustrate that ESG is not priced in the latest period, where it is for REITs. Looking at Table A4, the factor for the annual change in ESG is significantly positive in column 3 while it is negative in column 4. Only between 2011 and 2013 are the results of the sub regression comparable to REITs growth factor. The results for both approaches show that, unlike for REITs, ESG is not priced for the market in the last sub regression.

6.4. Discussion

In this part, the results presented above and their implications are discussed. One of the goals was to inspect potential arbitrage opportunities based on the ESG performance of REITs. The results from the long-short strategy, for both the current ESG rating and the annual change, illustrate that such possibilities exist. However, the outcomes might be counter-intuitive at first. Both approaches show that over the full period the long-short strategy significantly decreased the excess return of REITs. Using this strategy, on REITs current ESG rating, yielded significant negative alphas in all the sub regressions, except in the latest period from 2014 till 2016. The results are comparable to the outcome in the paper of Cajias et al. (2014), who are the only others that investigated the link in those years. Outcomes from the annual change in ESG performance during the sub regression show different results. It generated a significant negative alpha for the periods 2007 till 2010, and between 2014 and 2016. Something else that is noteworthy is the momentum factor. It yields a significant positive coefficient, for both methods, in almost every regression. This outcome suggests that momentum is at least on some level connected to high ESG ratings. Altogether the results show that applying the long-short strategy on the ESG rating for REITs yields a significant alpha. Most studies investigating the link find a significant relationship between CSP and the financial performance (e.g., Lesser et al., 2016; Nofsinger & Varma, 2014; Geiger et al., 2014; Polbennikov et al., 2016). In this study, as well for the entire period as during the sub regressions, at least one of the methods generates a significant alpha. For this reason, the null hypothesis for (H1), the long-short strategy does not generate a significant alpha, can be rejected.

(30)

27

The results also suggest that ESG ratings have a different effect on REITs than for the market. Comparing the current ESG scores for the two samples supports this statement. Despite that, both yield a significant negative alpha in the total time span, the size and significance is smaller for the market than it is for REITs. Also, during the sub regressions, the outcomes are only comparable between 2011 and 2013. Where the sign of the alpha coefficient for the market is positive but insignificant between 2003 and 2010, it is significantly negative for REITs. Looking at the regression of 2014 till 2016 also reveals different effects. While the alpha is insignificant for REITs, it is significantly negative for the market. This trend is also present when performing the long-short strategy on the annual change in ESG scores. Similar to the total ESG score are the differences over the full period. Both the market as REITs yield a significant negative alpha, but both the size and significance are smaller for the market. Between 2003 and 2006 the alpha for both samples was insignificant. Looking at the latest subsample from 2014 to 2016 does show comparable outcomes. Over these years the long-short strategy, for both the market and REITs, yielded a significant negative alpha. The results support prior work claiming that the effect of ESG on the financial performance differs between industries (e.g., Barnett, 2007; Godfrey and Hatch, 2007; Hoepner et al., 2010). Including ESG as a factor only strengthens the assumption of a different effect between the two samples. Comparing both the ESG factor as the ESG growth factor between the market and REITs further illustrate the dissimilarity. Almost every regression presents contradictive results between the market and REITs when compared over the same time frame. Only between 2003 and 2006 are there similarities of the effect of the current ESG factor on the return between the samples. Also, the sign and significance of the factor for the annual change in ESG performance are only comparable in the period 2011 till 2013. Looking at the outcomes from the individual sector regressions gives more intuition about this. It shows that, similar to the results of Hoepner et al. (2010), the effect of ESG ratings differs between sectors. The alpha for the long-short portfolio is for some sectors positive, for others negative or even insignificant. Overall the results imply that the existing literature describing the effects of ESG on the market cannot simply be generalised to listed real estate. By using multiple approaches, there is enough evidence to reject the null hypothesis of (H2), that ESG has the same effect for REITs as for the market.

Including ESG as a factor in the multifactor model provides more insight into how it is priced over the years. The results in Table A1, with a factor for the current ESG score of REITs, illustrate that the effect is changing. Looking at the sub regressions, between 2003 and 2010 the ESG score had an adverse effect on the excess return. This effect became insignificant in the period 2011 till 2013, and positive during the last sub regression of 2014-2016. The results show an upward trend from being significantly

(31)

28

negative in the first seven years to positive at most recent years. Results have also varied when the ESG growth factor is included in Table A2. During the first sub regression in 2003 till 2006 it had a positive effect on the excess return. This effect became insignificant between 2007 and 2010 and significantly negative for the years 2011 till 2013. Looking at the last sub-period, covering 2014 till 2016, illustrate that, similar to the current ESG score factor, the factor has a positive effect on the excess return. Both approaches show that the relationship does change during the investigated period. Geiger et al. (2014) also claim that the importance of sustainable real estate is evolving over the last decade. Increased transparency and more commitment to sustainable investments are only some examples that initiated this shift. Except for the period between 2003 and 2006 for the ESG growth factor, the effect changes from beginning predominantly insignificant or negative to positive. Therefore, rejecting the null hypothesis for (H3), that the relationship between ESG and REITs has not changed over time.

The negative relationship between high ESG scores and excess return might be somewhat counter-intuitive at first. However, it is not the first research that shows that high ESG ratings can decrease the excess return of REITs. Cajias et al. (2014) included the total ESG score, instead of using a factor for ESG, in their model and showed that it had a negative impact on the return for REITs. Data in their study was limited from 2003 to 2010, and the results in this thesis show a similar relationship in that period. Additional studies revealed that socially responsible firms underperformed conventional firms in non-crisis periods (Lesser et al., 2016; Nofsinger & Varma, 2014). Most prior works suggest that investors most likely demand less excess return from firms that have high ESG ratings. The perceived risk for the investment is assumed to be lower for firms that score well on ESG criteria. However, the results from the long-short strategy find no evidence that the systematic risk is lower. The lower systematic risk is expected to result in a negative coefficient for the excess return on the market in the long-short portfolios. Neither the outcomes of the market as REITs find support for this. Halbritter & Dorfleitner (2015) present contradictive results in their paper as firms with a high ESG score have a lower market beta. The authors use the same dataset, but they use a different method to conduct the total ESG score. This may explain the different results. Eichholtz et al. (2012) also found that portfolio greenness is negatively related to the market beta of REITs. In their study, portfolio greenness is measured with certifications that reveal how energy efficient a property is. Using this approach instead of the total ESG performance could be the reason for the different outcome. Despite the extensive existing literature regarding the subject, it is still important to look at the sector-specific effects. Most prior work generalises the effect of ESG on the excess return over the whole market (Hoepner et al., 2010). This thesis shows that it is essential to look at the effects for each sector. It is

Investigating the relationship between ESG and REITs