A cross-country REIT stock performance comparison : the influence of different REIT legislation regimes

(1)

A Cross-Country REIT Stock Performance Comparison

The influence of different REIT legislation regimes

UNIVERSITY OF AMSTERDAM

MSc Finance

Specialisation Corporate Finance & Real Estate Finance

Master Thesis

Author:

S. Calis

Student number:

10351272

Thesis supervisor: Dr. J.J.G. Lemmen

Finish date:

July-2017

(2)

This document is written by Sven Calis who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is 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)

ACKNOWLEDGEMENTS

I would like to special thank my thesis advisor Dr. J.J.G. Lemmen of the Economics and Business Faculty at the University of Amsterdam. He fully supported me when writing this thesis and has steered me in the right direction throughout the whole process.

(4)

ABSTRACT

REITs operate within different regimes around the globe. However, very little empirical research has addressed whether the existence of the different REIT regimes might affect the REIT stock performance. Also, the impact of the Global Financial Crisis on REIT stock performance has not been subjected to research yet. Using a sample of 217 publicly-listed equity REITs divided over four countries, the existence of differences in REIT performance is tested using standard Jensen’s Alpha, Treynor Index, Sharpe Ratio and Treynor-Mazuy Measure. The Fama-Macbeth model is conducted to see whether a relationship exists between the REIT regimes and the difference in performance. While results show that significant differences exist when various sectors and countries are compared, the differences are not consistent over time. While the difference in excess returns changed by the Global Financial Crisis, the results present no evidence for a relationship between REIT regimes and obtained excess returns.

Keywords: REITs, performance analysis, policy, international financial markets, Fama-Macbeth JEL Classification: R28; K25; F18

(5)

LIST OF TABLES

Table 1: Summary of findings on REIT performance 8

Table 2: Overview MSCI definitions of REIT market GICS 14

Table 3: Overview number of REITs per country full period sample 17

Table 4: Description of independent variables 18

Table 5: Overview regulations and adjustments related to the U.S. REIT market 21 Table 6: Overview regulations and adjustments related to the Australia REIT market 23 Table 7: Overview regulations and adjustments related to the Japanese REIT 25 Table 8: Overview regulations and adjustments related to the French REIT market 27 Table 9: Overview United States obtained performance measurement statistics 39 Table 10: Overview Australia obtained performance measurement statistics 39 Table 11: Overview France obtained performance measurement statistics 40 Table 12: Overview comparison Japan obtained performance measurement statistics 40 Table 13: Summary of Treynor-Mazuy Measure estimates per period and benchmark 41 Table 14: Cross-sectional regression of before-GFC excess returns using

country-related benchmark 45

Table 15: Cross-sectional regression of before-GFC excess returns using

Kenneth benchmark 45

Table 16: Cross-sectional regression of during-GFC excess returns using

Table 17: Cross-sectional regression of during-GFC excess returns using

Table 18: Cross-sectional regression of after-GFC excess return using

Table 19: Cross-sectional regression of after-GFC excess returns using

Table 20: Robustness regression on term structure before the GFC using

Table 21: Robustness regression on term structure during the GFC using

(8)

LIST OF FIGURES

Figure 1: REIT market capitalization by market, 2010 vs. 2016 (US$b) 4 Figure 2: Mean excess return of France REITs over time 19

Figure 3: Mean excess return of France REITs over time 19

Figure 4: Mean excess return of Australian REITs over time 19

(9)

CHAPTER 1 Introduction

Over the last few decades, real estate investment firms have become increasingly important for investors in financial markets. Starting with a signature of President Eisenhower on the REIT act title contained in the Cigar Excise Tax Extension of 1960, REITs have obtained global market capitalization of approximately $1.7 trillion (EY, 2016). There are three major assets classes in which investors could participate. As result of this market capitalization, REITs have become, accompanied by other investment funds, the third generation of these largest assets class available to investors behind stocks and bonds. Stocks and bonds are the first generation of asset classes whereas the second generation includes indexes, futures, options and swaps (Imperiale, 2002). The exponential growth in market capitalization can be attributed to the new investment opportunities that came with the introduction of REITs. Before the existence of REITs, investing directly in commercial real estate was only possible for a few high-capitalized investors which also had significant knowledge of the real estate market.

REITs trade as stocks, therefore, it vanishes the capital threshold of direct property investment by allowing investors to invest in commercial real estate by purchasing REIT ownership. Another positive occurrence that comes with this sort of trading is that the purchased ownership of publicly traded REITs can easily be transferred at a low cost. This results in an opportunity for small-scale investors to invest in direct commercial property without having the large and long-term financial obligation aligned with direct commercial real estate investment (Han and Liang, 1995). In essence, REITs are pools of properties and/or mortgages that trade on stock exchanges (Chan, 2003). For investors, this can be seen as a form of securitized real estate with claims on mortgage payments and/or real property (Chan, 2003).

Not all real estate companies can obtain a ‘REIT’ status, which allow them to be subjected to tax-exemption and other financial benefits. Companies who want to obtain this status have to fulfill specific requirements. These requirements are (mostly) based to the following criteria related to direct property ownership: the income generation and distribution, asset ownership, management structure and corporate ownership (Chan, 2003). These requirements differ widely per country (EPRA, 2016).

Real Estate Investment Trusts (REIT) are an investment vehicle created for common investors which enables them to invest in income-producing real estate. Direct investments in real estate are costly and illiquid while REITs are structured as mutual funds stocks and, therefore, more easily accessible. The REIT industry has grown at a high pace since their introduction. It grew from 34 REITs with a market capitalization of roughly 1.5 billion dollars in 1971 to 216 REITs with a market capitalization of 907,5 billion in 2014 in the U.S. only. The main reason for investors to incorporate REITs in their portfolios is to diversify their risk (Fidelity, 2003).

Diversification aims to maximize the returns by investing in different asset classes that react differently to certain events. Investing in REITs can provide diversification benefits to a portfolio.

(10)

However, many investors have remained underexposed to this asset class despite its low correlation and commendable track record of performance relative to other assets (Fidelity, 2003).

REITs behave differently around the globe. These significant differences are related to structure, strategy and operations of REITs. In some countries, REITs can be privately held (Private REITs), such as in the U.S., where in other countries REITs have to be listed on a stock exchange (Public REITs), as in France and Australia. In Japan, REITs are even required to be listed on a domestic stock exchange such as the Tokyo Stock Exchange (EPRA, 2016). Publicly traded REITs can be diversified further by the REITs choice of investment, resulting in the existence of equity REITs, mortgage REITs and hybrid REITs. Equity REITs invest in real property only, whereas mortgage REITs acquire mortgage obligations. Therefore, those REITs become the creditor with mortgage liens given to priority equity holders (Jackson, 2009). These mortgage obligations are securitized by the underlying value of the real property. Hybrid REITs invest in both real property and mortgage obligations which allow them to obtain the advantages and disadvantages of both sectors.

This study evaluates the difference in REIT stock performance between countries, and, therefore, uses publicly listed-equity REITs. The reasons for this is twofold. First, equity REITs have the largest market capitalization share of the global REIT market and, as second, publicly listed-equity REITs are the most appropriate to compare their returns with the returns of common stock because they are influenced by macroeconomic circumstances in the same way as other listed companies (Buttimer, Chen and Chiang, 2012). Listed-equity REITs can be diversified even further into specific subsectors. This study makes a difference between eight divisions, namely: Residential REITs, Office REITs, Industrial REITs, Lodging REITs, Retail REITs, Healthcare REITs, Specialized REITs and Diversified REITs. The REIT stock performance of these sectors are compared between four countries. The countries chosen for this study are the United States, Australia, Japan and France.

According to Yunus (2015), the U.S. Housing Market Crisis accumulated into the Global Financial Crisis (GFC). This has resulted in a change of investors and governments perspective towards the real estate market in general (Eichholtz, 2011; Ratcliff, 2013). REITs are securitized by the underlying real estate value. So, when the perspective towards real estate changes due to the GFC a couple of question arise relating to the REIT stock market. How does this change in perspective due to

the GFC affect REIT stock performance? If there are any significant effects, are those positive or negative? Do these effects differ between different REIT sectors of various countries? Which legislation framework has the best ingredients for the REIT to operate?

As pointed out by Eichholtz (2011), the GFC has changed the perspective towards the real estate market in general. Therefore, a valid hypothesis for the first question is that the GFC has affected the REIT stock performance.

To address a hypothesis to the second question, the finding of a study by Heany and Srianathakumar (2012) is used. According to Heany and Srianathakumar (2012), a period of financial distress results in a higher correlation between various sectors of the REIT market compared to the

(11)

pre-crisis period. Therefore, this study expects that the GFC will diminish the differences in stock performance if these are observed before the GFC.

According to Brounen et al. (2010), the individual REIT returns is intricately linked to the performance of the underlying asset in which the REIT has invested in. This implies that the sector in which REITs belong, affect the REITs return performances. Therefore, the hypothesis related to the third question will be that the effects of the GFC indeed differs between REIT sectors of various countries.

When a low distribution obligation is set, REITs can invest more of their income into new investments and, in this way, they can strengthen their future stream of dividends resulting in an increase in value (Gordon, 1959). On the other hand, the greater the uncertainty of the relation an additional investment to the safety of the dividend would overshadow this effect and thus will result in a decrease in value of the REIT (Gordon, 1959). Another regulation difference to consider when formulating a hypothesis of the fourth question would be the restriction on the use of leverage. The use of leverage is influencing the performance of the REIT. If a REIT regime provides a ceiling on the use of leverage, a reduction in systematic risk is expected because they are related to each other (Brounen, Mahieu and Op ‘t Veld, 2013). In this way, one would expect that leverage-constrained publicly-listed REITs are less risky than their stock counterparts. However, a restriction on the use of leverage could imply that REITs that are financially constraint up until their own resources. This restriction could result in missing a possible profitable investment opportunity. For these reasons, a general hypothesis for the fourth question difficult to make.

In most studies, a comparison has been made on the performance of different REIT sectors within a country. Also, the change in U.S. REIT performance by introducing a new regulation, the return differences between acquiring and target firms in the event of a M&A or the differences in performance before and after obtaining a REIT status are common subjects of research. Since 2010, the U.S. REIT market has grown by almost 150 percent to a market capitalization of around 1.07 billion U.S. dollars. Therefore, this market has become increasingly interesting to focus a study on. Simultaneously, the non-U.S. REIT market has more than doubled in terms of non-U.S. dollars to a market capitalization of around $600 billion as presented in Figure 1. France and the U.K. have taken the spot from Australia and Japan to become the owner of the second- and third-largest REIT markets (EY, 2016). U.K. REITs were created in 2007 and cannot be used to evaluate the differences before, during and after the GFC (EPRA, 2016). Therefore, these REITs are not used during this study.

(12)

Figure 1: REIT market capitalization by market, 2010 vs. 2016 (US$b)

Source: Ernst & Young (2016)

Most studies about REIT performance are related to the U.S.. To date, there has not been an empirical study made to compare various global REIT regimes. This study fills this gap by investigating which country possesses the best ingredients for REITs subjected to different investment sectors. Therefore, the REIT legislation framework of the countries studied should be evaluated before conducting various performance analyses. Due to these differences, it is plausible that the impact of the GFC on REIT performance varies per country.

As previously explained, this research focuses on the publicly listed-equity REIT market. Schwert (1981) argues that tests on regulations with stock price data are more powerful then tests with accounting data. Stock price data has an advantage over accounting data because it is more accurate and generally provide a great number of observations. Additionally, well-specified models of expected return can be used to isolate company-specific effects from marketwise shocks (Schwert, 1981).

This study analyzes the changes made in the REIT regulation and the changing global perspective of the investors and governments towards the REIT market. These changes affect important determinants of REIT performance. By analyzing the changes and performance per different legislation framework, this study intends to examine which framework has the best ingredients for the specific investment vehicle based on how different characteristics influence the excess returns of the different REIT sectors. By doing this, knowledge about the effects of future changes within the regulatory framework or economic circumstances on the REIT performance is offered to investors who allocate their recourses across the global REIT market.

This study uses monthly performance to test whether significant differences exist between excess returns of different countries by comparing various sectors of the REIT market. When conducting the performance analysis, two different benchmarks are used. First, a country-specific benchmark is used where the one-year Treasury yield related to the country studied is used as the risk-free proxy. For

0 200 400 600 800 1000 1200 2010 without

U.S. 2016 without U.S. 2010 US 2016 U.S.

(13)

This study cannot use the one-month Treasury yield because this rate is for two of the four countries included in the sample not available. Therefore, to make a valid comparison, the one-year Treasury yield for each country is used.

For the second benchmark, this study uses the data on market returns obtained by Kenneth R. French which is available on his website (French, 2016). These databases are commonly used in previous studies on REIT performance (Eichholtz, Kok and Yonder, 2010; Buttimer, Cheng and Chiang, 2012; Striewe, Rottke and Zietz, 2013; Abugri and Dutta, 2014). From here, the data for each continent where the country studied is located is subtracted to compare the performance of different REIT markets. Existing literature lacks a multiple cross-country performance comparison for the REIT market. Especially, how the performance of REITs is affected by the GFC within a specify country. This study intends to fill this gap by examining the excess returns with different performance measurements to determine whether there is a significant difference in excess stock returns between countries when different sectors are evaluated. The Sharpe Ratio, Treynor Index, Jensen’s Alpha and Treynor-Mazuy Measure are the performance measures chosen. The obtained values are compared to determine if a difference exists between countries using the Welch’s t-test. These performance measurements are conducted on three different time periods, which are pre-GFC period, during-GFC period and after-GFC period. This study uses the GFC period from the result of the Chow-test conducted by Eichholtz, Kok and Yonder (2010). The Chow-test is used to see if a structural break exists within a period can be assumed to know a priori (Eichholtz, Kok and Yonder, 2010). Conducting the Chow-test on the NAREIT equity index, they found that the GFC-period starts at January 2007, and ended at the end of May 2009. Therefore, the period from January 2003 to January 2007 will be referred during this study as the before-GFC period, January 2007 to May 2009 as the during before-GFC-period and May 2009 to December 2015 as the after-GFC period.

This study intends to relate regulation to performance and, therefore, different variables are tested on the excess returns using the Fama-Macbeth Two-Stage regression model. This model first regresses the market returns and two other factors, accounting for size and market exposure, on the difference between the return of the REIT and the risk-free proxy. The constant term, representing the abnormal return, is then used as the dependent variable in the second step where various REIT related variables are regressed on to see which variables affect these abnormal returns.

For analyzing the impact of different regulations and changes within a legislation framework, the Welch’s t-test is being used to test if a regulation adjustment has any significant impact on the excess returns of REITs. Normally, event-studies are conducted to test whether a significant difference occurs when a policy change is implemented (Binder, 1985). The main focus of this study is to relate various REIT regimes to differences in REIT stock performance. Therefore, the less time-consuming Welch’s t-test is used to see if a significant difference arises from a regulation adjustment.

This study proceeds as follows. Chapter 2 will discuss the relevant literature for studying REIT return performances. Chapter 3 describes the data used whereas Chapter 4 evaluates the legislation

(14)

frameworks per country. Chapter 5 explains how the models are implemented. Chapter 6 will give an interpretation of the results followed by the conclusion in Chapter 7. This study ends with an examination of the limitations of this study and suggestions for further research.

(15)

CHAPTER 2 Literature review

2.1 REIT Evolution

The evolution and history of the REIT market can be described by one word: explosive. In the relatively short lifetime of the real estate investment vehicle, the global REIT industry has been subjected to multiple failures and enjoyed various successes.

The last two decades, governments all over the globe implemented changes in tax laws and made regulatory adjustments. These adjustments have significantly affected the growth and development of the global REIT industry which has resulted in the REIT market becoming more attractive, viable and efficient (Jackson, 2009). Due to the significant impact of government regulations there can be concluded that a REIT is an investment vehicle which is directly related to government legislation. Consequently, the changes in government legislation have contributed to the rapid expansion of the REIT market (Jackson, 2009).

Today, as result of the popularity and efficiency of the REIT market, this market has attracted the interest of large institutional investors. REITs were initially developed to allow low-resource investors to invest in the real estate market. Due to the revolutionary shift in interest, participation of large institutional investors caused the REIT market to gain their important share which resulted in REITs becoming one of the three most important asset classes for today’s global investment.

2.2 REIT Performance

Starting late 1970s, many studies were related to the REIT market and these studies resulted in a large contribution to the understanding of this important investment vehicle (Han and Liang, 1995). The increasingly large capitalization of the market, combined with the occurrence of various different dramatic events associated with the industry, caused an increase in interest of many researchers towards the performance of REIT stocks (Han and Liang, 1995). However, the results of these studies have not showed consistent findings on the performance of un-securitized real estate (Wendt and Wong, 1965; Fieldman, 1970; Miles and McCue, 1984; Brueggeman, Chen and Thibodau, 1984; Lusht, 1988). This inconsistency arises by using differences in approaches to test the REIT performances, the use of different benchmarks and the use of different risk-free rates as proxies for determining out- and underperformance. The results of various studies comparing equity REITs performance relative to the stock market portfolio, have been mixed (Han and Liang, 1995). Furthermore, the period, and the length of it, which is chosen to be the period of interest differs widely and, therefore, it has a large influence on the test results (Buttimer, Chen and Chiang, 2012). Table 1 summarizes ten different studies relating to REIT performance and shows the variety in the data and period chosen, performance measurement(s) used and their main findings. These articles are discussed during this chapter.

(16)

Table 1: Summary of findings on REIT performance

Author(s) Data Performance

Measure Findings

Liu, Hartzell and Grissom (1990)

Quarterly returns on 18 survivor U.S. equity REITs, 1978-1986

Modified Jensen’s alpha

Performance are sensitive to the choice of six value-weighted indices

Chan, Hershott, Sanders (1990)

Monthly returns on 30 U.S. equity REITs, 1973-1987

CAPM and APM With CAPM, evidence of excess returns in

80s. With APM, this evidence disappears Martin and Cook

(1991) 27 U.S. equity REITs, 7 U.S. finite-life REITs,

1980-1990

Stochastic dominance In both post-tax reform sub periods, stock

portfolios dominated traditional equity REITs

Redman and Manakyan (1995)

48 U.S. REITs daily returns, 1986-1990

Sharpe Ratio Real property characteristics of REIT

portfolios influence returns Han and Liang

(1995) Different REIT sample per period, trading on NASDAQ, AMEX or

NYSE, 1970-1993

Two-factor performance benchmark model

REIT performance was unstable over time

Jackson (2009) Monthly returns on 297 U.S. equity

REITs, 1993-2005

Jensen’s alpha Different REIT performance results per

benchmark used Zhou and

Ziobrowski (2009)

Monthly returns on different CRSP/Ziman equity REIT sample

Modified Carhart 4-factor model

No evidence for performance persistence Huston and Elliot

(2011)

5 A-REITs with residential exposure

Treynor Index No correlation between risk management

score and performance Wu, Huang, Chiu

(2011) Daily returns on 100 U.S. REITs, 1993-2001 ARJI model Tax reform increased REIT value and thereby returns.

Buttimer, Chen, Chiang (2012)

NAREIT listed equity REITs monthly return, 1987-2009 Sharpe’s index Fama-French 3 and 5 factor model Carhart 4-factor Treynor measure

Various equity REIT subcategories perform differently. No evidence for equity REIT to outperform market

2.2.1 Aim of REIT performance

The aims of past research relating to REIT performance can be divided into three categories. The first category aims to identify the general characteristics that influence risk-adjusted returns. The second category uses applications of the Asset Pricing Theory (APT) model relative to Capital Asset Pricing Model (CAPM) to determine the economic characteristics affecting performance whereas the third category compares REIT performance to that of common stock and other real estate investments.

The study of Liu, Hartzell, Grissom and Wylie (1990) investigates whether ranking investment performance of equity REITs is meaningless given that the omission of assets in a market proxy leads to an inaccurate measurement of positive excess returns. Using a modified Jensen’s index, they find that even when the investment performance is inaccurately measured, the ranking of investment performance is useful. Another finding suggests that the composition of the market proxy does not necessarily lead to different inferences on real estate investment.

Redman and Manakyan (1995) examine the risk-adjusted performance of REITs in relation to financial and property characteristics of their portfolio. Using the Sharpe Ratio, they show that various

(17)

financial ratios, location of properties and types of real estate investments are significantly influencing the risk-adjusted performance of REITs. A second finding refers to the testing of individual financial variables where the results show that these variables are not significantly influencing REIT performance.

Buttimer, Chen and Chiang (2012) study the performance of equity REITs with the use of different models. The results of these classic regression-based frameworks, multi-index and multifactor extensions are validated by a non-parametric test. They find that various equity REIT subcategories perform differently and that the pure index model may not be the appropriate way to capture REIT return dynamics. Other findings of this study show that size, value term and default premiums have a significant influence on REIT returns whereas momentum and liquidity factors do not have significant explanatory impact.

The study of Chan, Hendershott and Sanders (1990) is related to the second category. They analyze monthly returns of equity REITs on an equally-weighted index using a multifactor Arbitrage Pricing model which they extended with pre-specified macroeconomic factors. They use appraisal-based real estate return data and when transaction-based equity REIT returns are utilized in a CAPM model with a single factor loading, excess returns still seem to exist. However, when a five-factor model is used, the evidence of excess returns vanishes. They also divided equity REITs into two groups related to their use of leverage. Using the five-factor model, they found that REITs with a higher leverage ratio seem to have a stronger and more consistent relationship with the macroeconomic factors used.

Wu, Huang and Chiu (2011) use daily data from 1989-2008 to examine the risk characteristics of REIT returns and how a tax reform (1993 TAR) influences these characteristics. Their evidence suggests that the tax reform of 1993 attracted institutional investors towards the REIT market. Institutional investors’ participation resulted in increasing monitoring of REIT value, decreasing risk exposure in the bond and stock markets and slowing the response of REIT returns to new market information. This implies that allowing institutional investors to obtain large ownership percentages of a REIT will result in reducing market efficiency and, thereby, having a negative impact on the relationship between REITs market and other asset classes. As explanation of this phenomenon, is the loose policy surrounding the 5/50 rule which implies that market makers may have strong influence on the prices. Therefore, it may reduce the market efficiency resulting in a negative impact on the relationships between REIT markets and other asset classes (Wu, Huan and Chiu, 2011).

A study of Han and Liang (1995) is related to the third category. They investigate the long-term performance of REITs, the stability of the performance and the sensitivity of the REIT performance using the Jensen’s alpha. They find that that the use of the S&P 500 as a benchmark leads to results that overstate the REIT performance relative to the stock market. Furthermore, their results show that short-term (six-year) variations in REIT performance were substantial and, in some circumstances, statistically significant. This implies that different outcomes of various studies are highly attributable to the chosen period of interest when it comes to short-term REIT performance studies.

(18)

Jackson (2009) examines the performance of a specific REIT division, namely lodging REITs. The performance of these REITs was examined relative to standard industry market portfolios and other equity REITs. Jackson found that the performance of lodging REITs was inferior to the other REIT sectors as well as the market portfolios used during his research. According to Jackson, there are several explanations for his findings. However, the foremost reason is the finding out of an article of Brounen

et al. (2010). According to Brounen et al. (2010), the individual REIT returns is intricately linked to the

performance of the underlying asset in which the REIT has invested in. This implies that the sector in which REITs belong, affect the REITs return performances.

To conclude, researchers have presented divergent views and outcomes when it comes to the evaluation of REIT performance. This is mainly due to the differences in periods, models, benchmarks and samples used conducting the performance analysis. The differences in results arising from the use of different periods can be seen when a comparison is made between studies. (Kuhle, 1987; Sagalyn, 1990; Titman and Warga, 1986; Goebel and Kim, 1989). According to Kuhle (1987) and Sagalyn (1990), most notable equity REITs were outperforming the stock market portfolio while the results of the studies of Titman & Warga (1986) and Goebel & Kim (1989) showed that these equity REITs have performed worse compared to the stock market portfolio. An explanation for this might be that most of these studies conduct a short-term study. According to Han and Liang (1995), the use of a short-term study results in inconclusive findings due to the booms and busts the REIT market was, and is, subjected to.

2.2.2 Testing of REIT performance

Different results also arise when different approaches are used. Most studies are focusing, either directly or indirectly, on the application of CAPM or APT to evaluate relative performance of REIT shares (Redman and Manakyan, 1995). Modifications of these classical unconditional one-factor CAPM models (Sharpe, 1964; Lintner, 1965) are also used to determine REIT returns. The market-timing ability modifications, such as Treynor-Mazuy Measure, are widely used when these returns are modeled with a time-varying portfolio (Treynor and Mazuy, 1966; Buttimer, Cheng and Chiang, 2012). These models use Jensen’s alpha to measure abnormal returns of the REITs (Jensen, 1968). One-factor CAPM models and the time-varying modifications are frequently used as a baseline model which are, occasionally, extended to multifactor settings. This approach is used because single-factor market models may not be appropriate to explain REIT market performance (Liang and Webb, 1995). In a study of Titman and Warga (1986), single-factor models as well as multifactor models are used to analyze REIT performance. They show that the multifactor model produces lower performance estimates that the CAPM-based counterpart.

When it comes to comparing different model outcomes, a recent study of Buttimer, Chen and Chiang (2012) uses six different models on all sort of REITs available in the REIT market, namely the

(19)

Treynor Index; Sharpe-Lintner; Fama-French three factor model (1993); Carhart four-factor model (1997); Fama-French five-factor model (1993) and the Pástor-Stambaugh model (2003) to analyze REIT performance. They argue that including other risk factors increases explanatory power of the models substantially. Their results show that the Treynor-Mazuy Measure and the Fama-French three-factor models are all significant at a 1% significance level and the adjusted-R2’s range from 0.40 to 0.71. These adjusted-R22_{statistic show the explanatory power of a model which is relatively high in the study of}

Buttimer, Chen and Chiang (2012). Therefore, the Fama-Macbeth model used explains the variation of the observations relatively well. Buttimer, Chen and Chiang (2012) show that momentum factors only matter for lodging and retail REITs and the liquidity factor of Pástor and Stambaugh (2003) also affects REIT returns in a negative way. They claim that, overall speaking, stock market factors and bond market factors, such as the term-structure of interest rates and credit rating are important to include in evaluation models to capture systematic risk (Buttimer, Chen and Chiang, 2012).

2.2.3 Benchmarking of REIT performance

While many REIT performance evaluations use a broad stock market index such as the S&P 500 or a national stock market index as a market proxy, this is only reasonable when the performance of an equity fund is assessed (Buttimer, Chen and Chiang, 2012). Multiple studies show that the use of a benchmark will make this benchmark another risk factor to take into consideration because different benchmarks present different results when performance is evaluated (Becker et al., 1999; Stutzer, 2003). During the period 1977-1987, two different benchmarks were used to study the performance of REITs. Sagalyn (1990) uses the S&P 500 index and shows that equity REITs outperformed this index, whereas Titman and Warga (1986) used the equally- and value-weighted CRSP indices. They did not find any significant difference when comparing returns.

These results are in line with the findings of Anderson, Clayton, Mackinnon and Sharma (2006) who use a variance decomposition approach to explore the investment characteristics of equity REITs. This is done with to use of a multifactor model relating REIT returns to the returns of small-capitalized value stocks, small cap growth stocks, large cap growth stock, bonds and private real estate. The outcome of this study shows that REITs have a significant small capital value component and the correlation between large capital stock indices and REITs have decreased over time (Anderson, Clayton, Mackinnon and Sharma, 2006). This implies that stock index returns containing constituents with a very large capitalization, are not appropriate anymore to use as a benchmark when conducting REIT performance evaluations.

According to Buttimer, Chen and Chiang (2012), the issue on which benchmark to use when analyzing REIT performance can be addressed in two ways. One way is to use the relevant benchmark, such as the market index related to the subcategory used for comparison. Within their own study, they use a specific hotel real estate index when they evaluate lodging REIT performance. Another possibility

(20)

is to make use of multiple index models. Block and French (2002) argue that such two-index models clearly identify those who out- and underperform the market and, therefore, have a higher explanatory factor compared to single-index models when REIT performance is evaluated. This is demonstrated in an article of Elton et al. (1993) where they include several relevant benchmarks simultaneously when conducting the performance analysis and present the same results.

2.2.4 REIT performance sample selection

Sample selection is another major factor explaining the difference in the obtained results. The choice of sample can be divided into three categories, namely survivors-only REITs vs all REITs, mortgage REITs vs. equity REITs vs hybrid REITs and the subcategories of the equity REIT market which are diversified REITs, industrial REITs, lodging REITs, office REITs, health care REITs, residential REITs; retail REITs and specialized REITs.

A survivors-only sample is a sample of REITs where REITs that are short-lived are being excluded from the sample. Using a survivors-only sample may result in an overestimation of REIT performance (Han and Liang, 1995). Especially for the REIT market, which is characterized by a high exit and entry rate, the sample used to fulfill the continuity requirement may result in a biased representation of the overall REIT industry performance. This survivors-only sample was frequently seen by more dated studies (Liu, Hartzell, Grissom and Wylie, 1990). More recent studies use a mild form of data continuity requirement which implies that a REIT should exist for 12 months before added to the sample. This sort of data continuation solves the counter part of survivorship bias, namely negative selection bias. Negative selection bias has a negative effect on the true outcome of REIT performance. Most equity REITs invest in hard real estate assets. Equity REITs therefore generate income from rent on properties as well as by obtaining undervalued properties and selling them for a profit or, depending on countries, develop properties to earn some profit. Equity REITs are obligated to distribute a significant percentage of the generated profits to their investors as dividends. The exact percentage which REITs are obligated to distribute differs per country as a result of the regulatory framework in which they operate. When interest rates are low and/or property prices are rising, equity REITs tend to perform better compared to other types of REITs (Tsai and Chiang, 2013).

Mortgage REITs invest in mortgages only. They capitalize less than ten percent of the total REIT market (EY, 2016). Mortgage REITs obtain their returns by charging interest on money lent to borrowers to finance property purchases. They also trade and invest in mortgage-backed securities. As with equity REITs, mortgage REITs can be divided into subcategories. There are commercial mortgage REITs, residential mortgage REITs and REITs that combine mortgages for both property types. As with equity REITs, the majority of mortgage REIT profits are paid to investors as dividends. Mortgage REITs obtain money by interests on loans, which are positively related with the interest rate, instead of rents

(21)

produced by the underlying property. Thus, mortgage REITs tend to perform better in times of rising interest rates which differs from equity REITs (Tsai and Chiang, 2013).

Hybrid REITs invest in both properties and mortgages. There are only a few REITs that actually engage in both types of business activity. Not specializing on the choice of investing results in a more balanced approach. As a result, hybrid REITs may be able to profit in both rising and falling interest rate environments where traditional equity only or mortgage only REITs can face difficulties. However, profits resulting from this hybrid-construction are, in general, relatively moderated (Tsai and Chiang, 2013).

The above three sorts of REITs present the main differences per sector. This study uses equity REITs only to make a global comparison of the REIT performances over a thirteen-year time period. The reason behind the choice of equity REITs is twofold. First, equity REITs dominate the REIT industry thereby contributing to most of the growth and capitalization of the REIT market (Najand et

al., 2006). The second reason is related to liquidity differences. Mortgage REITs invest in mortgages

and hold them to maturity without actively timing the market (Buttimer, Chen and Chiang, 2012). Therefore, equity REITs are more actively managed. As a result, equity REITs have smaller bid-ask spreads than mortgage REITs (Bhasin et al., 1997). This implies that equity REITs are therefore more appropriate to compare with other common stocks than mortgage and hybrid REITs.

During this section, an article of Jackson (2009) has been discussed. Jackson is not the only researcher that focused his study on a particular segment of the REIT market. REITs have become more specialized. As a result, many investors have chosen to focus on particular segments of the REIT industry (Chan et al., 2003). According to Chan et al. (2003), this specialization has enabled REITs to become more efficient and thereby making it more attractive for investors and interesting for researchers to focus studies on. This study uses the definitions of the MSCI Global Industry Classifisation Standard (GICS) to relate equity REITs to their appropriate sub-industry. On the following page, the GICS definitions for the REIT sector used by the MSCI are presented in Table 2 (MSCI, 2016).

To summarize, research findings on REIT performance vary in large proportion, especially when it comes to equity REITs. These inconclusive and contradicted findings exist for various reasons. No study has been conducted on the issue whether REITs perform differently due to their regulatory framework where REITs have to operate in. With the use performance measures, one-factor models and multiple factor models, the hypotheses of an existing difference between the performance of subcategories of the equity REITs will be tested, and, if there are significant differences, how these differences can be related to the regulation in specific.

(22)

Table 2: Overview MSCI definitions of REIT market Global Industry Classifisation Standard

Sub-Industry MSCI Definition of GICS

Office Companies or trusts engaged in the acquisition, development, ownership, leasing, management and operation

of office properties.

Industrial Companies or trusts engaged in the acquisition, development, ownership, leasing, management and operation

of industrial and distribution properties.

Lodging Companies or trusts engaged in the acquisition, development, ownership, leasing, management and operation

of hotel and resort properties.

Heath Care Companies or Trusts engaged in the acquisition, development, ownership, leasing, management and operation

of properties serving the health care industry, including hospitals, nursing homes, and assisted living properties.

Retail Companies or trusts engaged in the acquisition, development, ownership, leasing, management and operation

of shopping malls, outlet malls, neighborhood and community shopping centers.

Specialized Companies or trusts engaged in the acquisition, development, ownership, leasing, management and operation

of properties not classified elsewhere. Includes trusts that operate and invest in storage properties. It also includes REITs that do not generate most of their revenues and income from real estate rental and leasing operations.

Diversified Companies or trusts with significantly diversified operations across two or more property types.

Source: S&P Capital IQ MSCI GICS Mapbook Version 2, 2015

2.3 Financial regulation on REIT performance

Studies on the impact on REIT performance due to changes in financial regulations is scarce. This is puzzling because there is no doubt that differences in regulatory regimes has a large impact on the performance of certain REITs (La Porta, Lopex-de-Silanes, Shleifer and Vishny, 2001). A possible explanation for this is that there is a difficulty involved in testing whether certain performance effects can be related to a regulatory regime in specific (Brounen, Mahieu and Op ‘t Veld, 2013). To illustrate, if an adjustment in regulation has been done and there has a reduction in return volatility observed simultaneously, how can someone ascertain which one has occurred because of the other happening? Avgouleas and Degiannakis (2008) insert control variables to counter this problem. They investigated the effects of Financial Instruments Markets Directive (FIMD) on liquidity and found that measures of liquidity generally increase over time. So, inserting control variables for these measurements has a de-trending effect on these series. By inserting different control variables, a particular effect of a regulatory change can be isolated (Brounen, Mahieu and Op ‘t Veld, 2013).

As previously stated, studies about effects of a regulatory adjustment on REIT performance is limited. There are some studies about these changes, but they are mainly focussed on tax reforms in the U.S., related to the Initial Public Offerings of the REITs and performed on the change in performance arising when obtaining a REIT status (Brounen, Mahieu and Op ‘t Veld, 2013). The latter is on the change in status but names a couple of important determinants. Those determinants are the distribution obligations, requirement to be exchange listed, permitted activities, leverage limitations and the ownership structure of the REIT. This study uses exchange-listed REITs only, therefore, the study does not investigate the effect of listing requirements before obtaining a REIT status. The other determinants are used to compare their effects of on the performance of the REITs. Because these determinants differ per country, and some of them also differ per year within a country, the timing of these adjustments will

(23)

be incorporated in the study for making a valid comparison between REIT regimes. The different legislation frameworks will be evaluated in Chapter 3.

2.4 Dividend pay-out policy and leverage restrictions

Dividend pay-out policy has been at the heart of the finance literature for a long time (Brounen, Mahieu and Op ‘t Veld, 2013). In all countries, REITs have the obligation to distribute a large share of their income to their shareholders. These percentages to which extent REITs are obligated to distribute their income vary widely. This determinant can have a major influence on the REIT performance because a sudden shift in dividend pay-out policy might affect the REIT stock performance directly. When a low distribution obligation is set, REITs can invest more of their income into new investments and, in this way, they can strengthen their future stream of dividends resulting in an increase in value (Gordon, 1959). On the other hand, the greater the uncertainty related to the additional investment to the safety of the dividend would overshadow this effect and thus will result in a decrease in value of the REIT (Gordon, 1959).

This study concentrates on the differences between regulation per country in which the REIT has to operate. Therefore, the impact of the differences in the dividend pay-out distribution obligation is important to evaluate. According to the studies of Pettit (1972), Charest (1978) and Michaely, Thaler and Womack (1995), an increase in the obligation of the share which has to be distributed will result in a positive stock return whereas a decrease in the obligation will cause a negative market reaction. These studies also show that the magnitude of the change in dividend is directly related to the price reaction. The direct reaction is caused by the fact that dividends can be used as an ex-ante signal for the future cash-flows, assuming the management of the firm can properly assess corporate earnings for following years (Brounen, Mahieu and Op ‘t Veld, 2013). The announcements of the dividends policies of the REITs are subjected to research already (Wang et al., 1993; Hardin and Hill, 2008). Still, the effects of various legal REIT requirements have not been subjected to much of research but, in theory, these requirements seem to have a large influence on the returns of REITs (Brounen, Mahieu and Op ‘t Veld, 2013).

The differences in regulatory acceptance for the use of leverage are also influencing the performance of the REIT. If a REIT regime provides a ceiling on the use of leverage, a reduction in systematic risk is expected because they are related to each other (Brounen, Mahieu and Op ‘t Veld, 2013). In this way, one would expect that leverage-constrained publicly-listed REITs are less risky than their stock counterparts.

This study does not investigate these dividends announcements or changes in the limitations on the use of leverage in specific but focusses on the differences of various REIT regime key attributes between countries in general. Therefore, various empirical tests will be conducted to acquire knowledge about how these differences affect the REIT performance in specific.

(24)

CHAPTER 3 Data description

This section describes the construction of the sample and presents the choice of the main independent variables.

3.1 Sample Construction

The full sample contains only publicly listed-equity REITs which are available in the CRSP Daily Security and Compustat-IQ databases. From here, the Standard Industrial Classification code (SIC) and the Global Industry Classification Standards (GICS) were used to obtain only listed-equity REITs, respectively the SIC for Real Estate Investment Trusts (6798) and the GICS for equity-REITs. The GICS codes used are between 600000 and 700000.

The annual financial data from Compustat-IQ had to be available for the REIT to be included in the sample. Additionally, this study requires that REITs are incorporated in one of the four countries studied and are listed on one of the four specific stock exchanges. These four countries are the United States of America, France, Australia and Japan. These countries are chosen based on the number of available listed-equity REITs, moment of implementation of the real estate vehicle and their REIT market status which is either developed or matured (PWC, 2013). The NASDAQ, Euronext Paris (ENP), Australian stock exchange (ASX) and the Tokyo stock exchange (TSX) are the market benchmarks for the U.S., France, Australia and Japan, respectively. The data for the monthly performance of these stock exchanges are obtained from Yahoo Finance and Investing.com. Data for the one-year Treasury yields are obtained from the governmental websites or, if they were not available for the time period studied, from Yahoo Finance and Investing.com. The data obtained from the governmental websites, Yahoo Finance or Investing.com databases is referred to as the “Country-specific benchmark” during this study. For the second benchmark, referred to as the “Kenneth benchmark” during this study, the data available on the Kenneth R. French website is obtained for the different continentals and related to the appropriate countries. Therefore, this study uses the data samples for North-America, Europe, Japan and Asia-Pacific for the U.S., France, Japan and Australia, respectively.

The final REIT sample contains various listed-equity REIT for which there is at least one year accounting data available and had twelve monthly stock price observations to prevent bad-sample bias (Zhou and Ziobrowski, 2009). Including REITs that did not survive the whole periods studied prevents this study from containing survivorship-bias (Zhou and Ziobrowski, 2009)

Before conducting the Fama-Macbeth Two-Stage regression model, the Small-Minus-Big (SMB) and High-Minus-Low (HML) factors had to be determined. These factors seem to explain a large part of the variation between excess return observation by accounting for size and market exposure (Fama and French, 1993). Therefore, this study follows the method of Eichholtz, Kok and Yonder (2010), Buttimer, Cheng and Chiang (2012), Striewe, Rottke and Zietz, (2013) and Abugri and Dutta, (2014) and uses the data on these factors which are also subtracted from the Kenneth R. French website.

(25)

From there, the Fama-French factors for North-America, Europe, Japan and Asia-Pacific are obtained and used in the performance analysis which are in line with the market and risk-free rate proxies obtained from the same website as described previously.

To conclude, the sample includes 287 listed-equity REITs divided over four countries, eight sectors and are listed on one of the four broad stock exchanges. Table 3 shows the distribution of the REITs per country included in the sample.

Table 3: Overview number of REITs per country full period sample

United States France Australia Japan

All 153 30 62 42

Diversified REITs 20 14 13 12

Industrial REITs 9 0 3 4

Lodging REITs 14 1 1 3

Office REITs 21 7 9 13

Health Care REITs 13 0 2 0

Residential REITs 19 1 2 6

Retail REITs 29 5 7 4

Specialized REITs 28 2 7 0

3.2 REIT characteristics

This analysis examines the differences between legislation frameworks and how they influence stock performances of listed-equity REITs around the world. Consequently, this study uses yearly excess returns as dependent variable in a panel regression. In the following, the independent variables will be described.

First, following the approach of Eichholtz, Kok and Yonder (2010), this study includes the leverage ratio into the regression equation, defined as debt divided by assets, to control for the use of debt and to see how the use of debt affects the excess return of REIT stocks. By interpreting the sign of the estimates regression coefficient related to the leverage ratio, a conclusion can be made on the effect of the use of debt within the REIT market. This is important due to the difference whether the use of debt is allowed. As second, the return on assets ratio, defined as net income over assets, will be included as a risk control variable (Striewe, Rottke and Zietz, 2013). The third control variable included in the regression will be the dividend yield which is defined as dividend distributed in year t divided by the price of the average REITs stock within year t. In this way, the effect of a higher dividend yield will can be evaluated. This variable is important to make an appropriate conclusion regarding the differences in the obligated dividend distribution percentage of the REITs income stream.

Next to the ratios, dummy variables are included to control for which country the REIT is incorporated. The regression analysis will be conducted on each REIT market sector described in Chapter 1. The division of these sectors are based on the Global Industry Classification Standards as stated previously (Benefield, Anderson and Leonard, 2009). The natural logarithm of a REITs total assets at the end of the fiscal year will be used as a control for differences in size (Redman and Manakyan, 1995).

(26)

Another control for size which is implemented into the regression analysis will be the book-to-market ratio whereas for the Fama-French three-factor regression analysis the ‘SMB’ and ‘HML’ will be the controlling factors for explaining excess returns. Peterson and Hiesh (1997) find that these risk proxies are frequently used to explain REIT stock return but they criticize the effectiveness of it (Eichholtz, Kok and Yonder, 2010). However, this study follows the procedure of Eichholtz, Kok and Yonder (2010) and therefore also avoids the discussion of this topic.

To make an appropriate comparison, all variables which in displayed in currencies, are transformed into U.S. dollars using the historical exchange rates provided by the governmental sites used or, when those sites did not provide the exchange rates needed, the historical exchange rate database of OFX group is used for converting the various currencies into U.S. dollars (OFX, 2017).

Table 4 will give an overview of all independent variables used in the regression analyses. A summary of the financial data of REITs used during this study related to the independent variables is given in Appendix A. Appendix B provides a list of the REITs used during this study sorted on country and market sector

Table 4: Description of independent variables

Name Description

Leverage Ratio Total liabilities divided by total assets

Dividend Yield Dividend per share distributed divided by average price of share

ROA Net income divided by total assets

Size The natural logarithm of the total assets

BTM Book-to-market ratio

Country Dummy indicating in which country a REIT is incorporated

HML High-Minus-Low factor obtained from Kenneth. R. French website

SMB Small-Minus-Big factor obtained from Kenneth. R. French website

3.3 Regulatory and macroeconomic environment

The focus of this thesis lies on making an analysis of the influence of different legislation frameworks on various REIT sector stock performances and intends to examine which country has the best ingredients for a specific REIT sector, respectively. In particular, this study investigates whether difference between REIT stock performance can be explained by difference in the country-specific regulatory environment which the REIT market sector is subjected to and how the GFC has affected these differences.

Using the database from the EPRA, this study obtains the main determinants of the REIT legislation over four different countries during the time period of 2003-2015. By comparing the differences between the main legislation determinants related to the REIT sector with the results of the regression analysis, a statement can be made which country has the best ingredients for a specific REIT sector based on stock performance.

(27)

Adjustments of the regulatory environment are not uncommon, but are a result from a relatively slow political process (Bart, Caprio and Levine 2004). To see whether a specific regulation has any significant impact, the Welch’s t-test is computed on the mean excess returns. When comparing different legislation frameworks, one could argue that a stricter environment will cause a more stable return of the REITs. However, more powerful supervision could also limit REITs in their range of investment opportunities (Pelster et al., 2009). Therefore, an expectation about the signs of the coefficients resulting from regression analyses cannot be made with appropriate economic reasoning.

Before evaluating the differences between countries REIT market regulations, Figures 2 to 4 presented below show the average excess returns per country and benchmark used during this study.

Figure 2: Mean excess return of France REITs over time Figure 3: Mean excess return of Japanese REITs over time

Figure 4: Mean excess return of Australian REITs over time Figure 5: Mean excess return of U.S. REITs over time

Interpreting Figures 2 to 4, there can be said that the excess returns differ for the two benchmarks proxies. Australia shows the largest deviation between both benchmarks, while Japan has only minimum difference between both market and risk-free proxies. For all countries considered, both benchmarks show the same negative excess returns during the GFC.

In 2005, the Japanese REIT market reaches an all-time high when the country-benchmark is employed. Also, there should be noted that for France and the United States, the peak in excess returns is reached directly after the GFC.

From the Figures, one could conclude that the Japanese REIT sector, on average, performs best whereas the Australia REIT market shows inferior performance, even when both benchmarks proxies are evaluated. -, 0 2 -, 0 1 0 ,01 ,02 Exce ss re tu rn 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Returns in logarithm are used to determine the excess return

Excess return of France

Mean Country-benchmark Mean French-benchmark

Year -, 0 5 0 ,05 ,1 ,15 Exce ss re tu rn 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Excess return of Japan

Year -, 1 -, 0 5 0 ,05 Exce ss re tu rn 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Excess return of Australia

Year -, 0 6 -, 0 4 -, 0 2 0 ,02 ,04 Exce ss re tu rn 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Excess return of United States

(28)

CHAPTER 4 REIT Legislation Framework

This section will give an overview on the legislation frameworks per country. Per country studied, a brief history of the REIT sector will be given followed by examination of the regulations related to the main determinants of REIT stock performance. When changes in the regulations related to the studied variables occur, these will be examined and a hypothesis will be added about the impact of this change on the REITs performance. This study assumes that the Efficient Market Hypothesis holds in general, implicating that the price of a REIT stock reflects all information which is publicly available at that point in time (Fama and French, 1970). All information of this section is obtained from the EPRA Global REIT surveys 2003-2015, otherwise, it will be indicated. The following four countries are used in this study: United States of America; Australia; France and Japan. At the end of each section, a quick overview of the regulations and their adjustments per country is given. The determinant(s) affected by policy adjustments are indicated by the bold numbers.

4.1 United States

The United States were the first to introduce the REIT (U.S.-REIT) as early as 1960. Starting with a signature of President Eisenhower on the REIT act title contained in the Cigar Excise Tax Extension of 1960, this vehicle created an opportunity for the average investor to invest in large-scale commercial properties just as if it were any other kind of equity investment. The U.S. REIT regime, which is governed by several tax and regulatory laws, has been modified several times since its creation. The most recent change has been implemented by the PATH Act, signed into law on 18 December 2015. The REIT sector of the U.S. is the largest REIT market in the world and accounts from almost 1 trillion U.S. dollars.

Minimum share capital & shareholder requirements

The U.S. REIT regime does not require any specific initial minimum share capital. However, there is a minimum of the number of shareholders required. The minimum number of shareholders to benefit from the mutual ownership is set at 100 (Internal Revenue Code 856). Also, there is a restriction made to some of the investors because five or fewer individuals may have no more than 50% of the REIT shares held at the end of the taxable year. There are no explicit regulations related to foreign investors.

To trade on the NASDAQ, REITs must have a minimum of 1,250,000 publicly-traded shares upon listing which are trading at a regular bid price of 4 U.S. dollars at time of listing. This implies that the minimum share capital has to be 5 million U.S. dollars (NASDAQ, 2106). The minimum number of shareholders is equal to the number set in the REIT framework, namely 100 (NASDAQ, 2016).

(29)

Table 5: Overview regulations and adjustments related to the U.S. REIT market

Determinant Variable 01/2003 07/2008

Minimum share capital Minimum share capital ($ mln) 5 5

Shareholder requirements Minimum number of shareholders 100 100

Restriction of institutional investors (%) 50 50

Restriction on foreign shareholders (%) 0 0

Asset level Restriction on ownership other company (%) 10 10

Maximum non-REIT activity (%) 5 5

Income test real estate activity (%) 75 75

Maximum ownership TRS (%) 20 25

Leverage Leverage limitation (yes = 1; no =0) 0 0

Leverage limitation (%) 0 0

Development Allowed (yes = 1; no =0) 1 1

Maximum development (%) 25 25

Foreign investment Allowed (yes=1; no=0) 1 1

Financial lease Financial leases (% or allowed=1) 1 1

Profit distribution obligation Income (%) 90 90

Capital gains (%) 0 0

Source: EPRA Global Survey’s 2003-2015

Asset level, leverage & development

According to U.S. REIT legislation, a REIT cannot own more than 10% of the value of a single entity (Internal Revenue Code 856). Nor can a REIT own stock in a corporation if the value of the entity exceeds more than 5% of a REITs’ assets. Since January 2001, the REIT modernization act (RMA) opened the possibility for REITs to create Taxable REIT subsidiaries (TRS). These TRS can provide services to tenants for which a REIT is permitted. A REIT can own up to 100% of the ownership of the TRS but its value cannot compromise more than 20% of the REITs’ assets. This ceiling has been lifted the major housing bill in July 2008, increasing the size of the ceiling from 20% to 25%. Recently, this ceiling has been downgraded towards its former size. This regulatory adjustment was only introduced recently by the PATH act and is effective from January 2018. This study makes a performance analysis on the period between 2003-2015 whereas the announcement of the downgrade was during 2016. Therefore, this policy adjustment has no effect on the results of this study. However, the impact of the policy change may be interesting for further research.

If a REIT is allowed to own more of a TRS, more services for which a REIT is permitted can be supplied to the tenants thereby increasing the possibility to create an income stream. However, this income stream is subjected to corporate income tax and, thus, maybe generate a lower income after tax resulting in a lower profit.

A U.S. REIT is allowed to operate, own, manage and develop for its own portfolio. Only, at least 75% of the gross income must be derived from real estate property rental when it comes to equity REITs. Therefore, this study assumes a ceiling of REITs own development of 25% due to this gross income

A cross-country REIT stock performance comparison : the influence of different REIT legislation regimes