The impact of a country’s legal REIT structure on performance

(1)

The impact of a country’s legal REIT

structure on performance

Joost Boeles

Student No. 5841151

Amsterdam Business School

Master’s Thesis

July 2014

Master of Business Economics – Double Track

Finance and Real-Estate Finance

(2)

2

Abstract

The objective of this research is to obtain a clear view of the various REIT structures around the world and the impact of their differences on the performance of REITs. Specifically, in this research is examined how the components of such a structure affect REITs performance. Companies from nine countries are observed in a panel framework for the time period 2010-2013. It was found that the structures of REITs around the world look similar but that small, subtle differences exist. A summary of the most important components follows. First of all, management restrictions should be minimized, since internal and external management are both favourable in different circumstances. Real-estate investment restrictions have a positive effect on performance and should be high. A gearing limit has no significant effect, and no limit should be incorporated. Payout restrictions are favourable, and the required payout ratio should be at 90% or higher. Finally, ownership requirements should be minimized, since a high ownership concentration results in better returns.

Keywords: REITs, Public real estate, REIT Structure Data: SNL, Compustat

Master: Double track

Supervisor: Professor M. Petrova

Acknowledgements: I want to thank Professor Petrova for her useful input and overall

(3)

3

1. Introduction

Real estate is a major asset class in which investors can participate in two different ways. They can invest in real estate directly and indirectly. Direct real estate has been shown to provide significant diversification benefits with stocks. However, it also has several disadvantages, such as low liquidity and high capital requirements (Hoesli, 2012). Indirect real estate provides a solution to some of these problems. Investing indirectly is mostly done in Real Estate Investment Trusts, also called REITs, the abbreviation used throughout the rest of this thesis. A REIT is an investment vehicle that invests primarily in real estate and

qualifies for a special tax status, so that there is a single tax levied at the investor level. REITs are publicly traded companies that own and manage investment-grade commercial real estate. They should not be confused with mutual funds, closed-end funds, or partnerships. Advantages of REITS are their liquidity, low capital constraints, and tax benefits (Brounen & de Koning, 2012).

Partly due to these favourable characteristics, there has been a tremendous growth in REIT equity-market capitalization around the world in past decades. A market

capitalization of 38 billion dollars in the United Kingdom (U.K.) and more than 630 billion dollars in the United States (U.S.) halfway through 2012 (SNL financial, 2012) is clear evidence of this asset class’s significant size.

Although some countries introduced REITs a long time ago, the global introduction of REITs happened quickly and only recently. Most European countries, for example,

introduced them only in the 21st_{century. There exist differences between REIT systems}

around the globe. These differences are partly due to the novelty of REITs, and they remain despite the fact that real-estate investment has been globalized. Differences exist in terms of legal systems, fund formats, investment restrictions, and REIT requirements.

Much research has already been conducted on the performance of REITs, and specifically on the characteristics of REITs that impact their performance. However, less has been written on the impact of a country’s legal framework on the performance of REITs in that country.

(5)

5 The aim of this thesis is to define the impact of institutional legal structures in countries around the world on the performance of REITs in those specific countries. Once that has been accomplished, the information obtained from these specific countries will be utilized to produce a general recommended structure for REITs. The research question that will be examined is the following:

‘’How do differences within REIT structure affect performance?”

This research question will be tested empirically in a panel framework of REITs from nine different countries. Applicable variables, such as growth rates, risks, and interest rates will be used to control for macroeconomic country influences.

This study expects to find that legal structure impacts the performance of REITs. Each country-specific legal framework consists of multiple requirements that potentially have an impact on performance. Those different requirements will be tested in a regression on firm- level data with excess return as the dependent variable. In this way, the magnitude and impact of those requirements can be calculated and an optimal legal structure can be proposed. This study is a contribution to the on-going debate on the important underlying variables that define REIT performance, and in particular it is intended to give insights into the role of institutional framework on performance.

In this thesis, Stevenson’s article (2013) is used as the standards by which to compare international REIT structures. In his article, REIT structures are compared in terms of

management, real-estate investment requirements, overseas investment, development, use of leverage, payout requirements, investment scheme, and tax treatment. Ownership requirements, which are not discussed by Stevenson’s article, are also added as a component to REIT structure.

This paper is structured as follows: This chapter was a brief introduction to REITs and an explanation of the content of this research. Chapter Two consists of scientific economic background information, such as the existence of REITs, the advantages of REITs, and a summary of the important firm-specific variables that have an impact on REIT performance. In Chapter Three, the different components of REIT structure are examined, and the

expectations and hypotheses are formulated. In Chapter Four, the research design is described. Chapter Five discusses the results. Finally, Chapter Six draws a conclusion and provides possible avenues for future research.

(6)

6

2. Literature review

REITs are investment vehicles that invest primarily in real estate and qualify for a special tax status, so there is a single tax levied at the investor level. They are publicly traded companies that own and manage investment-grade commercial real estate. REITs should not be

confused with mutual funds, closed-end funds, or partnerships.

REITs were invented in the U.S. when Congress decided that the only way for an individual investor to engage in large-scale commercial real-estate investments was through a pooling system. As a result, Congress designed REITs in 1960 to unite the capital of many investors into a single economic enterprise (Brounen en de Koning, 2012). However, REITS played a minor role in real estate investments for more than thirty years. This was partially due to the fact that REITs need to find third parties to operate and manage properties, as a result of constraints. This was abolished by the Tax Reform Act of 1986. The global REIT market has only started to grow in the last decade. Ten years after the introduction of REITs in the U.S., Australia and the Netherlands followed. In the 1990’s, Belgium and Canada also introduced REITs. In the past decade the majority of developed Western and Asian countries have introduced REITs as well. The U.K. and Germany, however, introduced REITs only within the past eight years.See Appendix 1 for a timeline.

To explain the tremendous growth of REITs over the past decades, important

differences between direct and indirect real estate will be discussed. Investing directly in real estate has many advantages. First of all, the investor controls the policy and strategy.

Furthermore, an investor has more feeling with the market and can potentially exploit his own expertise when he is directly involved. In this way, the possibility of adding value through active management arises. Moreover, real estate offers an inflation hedge, since its leasing contracts include rents that increase with inflation. Finally, direct investment in real estate offers a stable cash return. Downsides of investing directly in real estate include the capital-intensive acquisition of a property. A significant amount of capital is needed, and an investor is constrained for a long period of time. Moreover, transactions costs are high, direct real estate is illiquid, and there is a risk of becoming emotionally attached. To

conclude, local expertise is needed which could be difficult to acquire and can be expensive as well (Pagliari & Monopoli, 2005).

(7)

7 REITs offer a solution to some of the problem posed by direct real-estate investments. They enable small capital investment, no local expertise is required, and they are much more liquid. Also, the problem of emotional attachment becomes irrelevant. When investing directly, a high level of knowledge is required, and therefore only a few investors are able to participate directly in the ownership or financing of commercial real-estate properties. Investing indirectly presents an alternative. The ownership of most REITS can be easily transferred with very low transaction costs, since most REITs are publicly listed.

Disadvantages of REITs include the lack of control over the investment policy, the lack of feeling with the market, and the inability to exploit one’s own knowledge. Moreover, REITs are characterized by a higher volatility level. One final drawback is the fees investors sometimes pay to the managers of a fund.

Moreover, there is debate as to whether REITs are really real estate. Is there a

difference for an investor to invest directly or indirectly in real estate? Pagliari, Schrerer, and Monopoli (REE, 2005) found that the average return for NAREIT (indirect) and NCREIF index (direct) over the same period differed 5%, at 13.5% and 8.4%, respectively. However, after they controlled for methodological differences, the difference in average return between direct and indirect real estate was diminished. They concluded that REITs are real estate after all.

Many studies have shown that returns of REITs are affected by multiple underlying variables. Redman et al. found that gross cash flow, leverage, and a firm’s size plus the regional location of properties are determinants of the performance of portfolios (Manakyan et al., 1995). Furthermore, it was found that REIT returns are better explained by stock-market beta and by Fama-French company-specific variables (such as company size), than by macroeconomic variables (Chen, 1998). Nevertheless another empirical study found that country factors are important determinants of international real-estate security returns (Ling and Naranjo 2002).Some literature also claims that past REIT returns are a significant driver of future REIT returns (Chui, 2003). Moreover, Han found that REIT performance varies over time (Han, 1995). Hamelink and Hoesli (2004) suggested that country, scale, and

(8)

8 Beside firm-specific variables, there are other studies that have recognized and recorded the importance of country-specific factors in explaining excess real-estate returns. Edelstein’s article (2011) proposed that cross-country institutional differences might be a significant descriptive variable for the performance of REITs, as well as real-estate-related capital market behaviour in general. Their findings showed that institutional features do not affect firm valuation across industries uniformly and that the influence of institutional

characteristics is real-estate market region-specific (Edelstein, 2011). Edelstein (2011) used three important country-specific variables that may affect excess return:

 Corporate-governance quality

 Legal-system quality

 Accounting-standards quality

This thesis will expand on Edelstein’s article and take it to a more microeconomic level. How do the different aspects of a country’s legal framework affect performance? How does a country’s REIT structure affect the performance of the companies in that country? According to Stevenson (2013), distinctions in REIT structure could be made on real-estate investment requirements, overseas investments and development, gearing limits, payout distributions, capital structure and the level of tax-transparency. Apart from these variables, ownership requirements are added as a component of REIT structure. REIT structure is further discussed in the next chapter.

(9)

9

3. REIT structure

A first impression when analysing the regulations in different countries is that the broad bundle of regulations is similar, but that there are delicate yet very essential differences among them. Those differences within REIT structure are explained in further detail below. The comparison of international REIT structures in Stevenson’s article is used as a standard. All of these different aspects will be intensively discussed in this chapter.

Management

The corporate structure of a REIT is dependent on the management form. There are two property-management structures, namely, internally or externally advised management structures (Ambrose & Linneman, 2012).

This difference exists, because originally, REITs were a passive investment where advisors were asked to carry out tasks similar to those performed by mutual-fund portfolio managers. However, unlike stocks and bonds, real estate requires active management, so property managers were hired. This resulted in numerous inefficiencies and in conflicts between advisors and REIT shareholders. Self-advised/self-managed REITs arose as a result. (Ambrose & Linneman, 2012).

Ambrose and Linneman (2001) tested the hypothesis of whether internally managed REITs are the better choice compared to externally managed REITs, due the fact that

conflicts of interest between REIT management and shareholders are solved more quickly. Results suggested that internally managed REITs constitute an example for externally

managed REITs and that externally managed firms respond to performance standards set by newer, internally advised REITs. Also, internally advised REITs have a cost advantage, which could contribute to their more dominant role.

(10)

10 Real-estate investment requirements

Per country, REIT structure is governed by guidelines on the required rate of real-estate investments. This rate is a percentage of a company’s investments that should be invested in real estate. An examination of those guidelines shows that the required percentages vary widely, from flexible (0% or more, as in France), to strict (a minimum of 100%, as in the Netherlands). The average and median are around 70%.

Overseas investments and development

The next component in the comparison, used in the article of Stevenson, is overseas investment and development. This part shows whether companies are allowed to invest overseas in real estate and whether they are allowed to develop real estate. Since the differences are minor among the countries, and companies in most countries are allowed to do both, this aspect is not further discussed, nor will it return as a variable in the regression.

Gearing limit

The gearing limit of a REIT is the level of debt in relation to the level of equity capital. This number is expressed as a percentage and as a measure of a company’s leverage. It gives information about the relationship between the amount of funding done with borrowed money and shareholder value. Or, in other words, it gives information about a REIT’s ability to debt-finance investments. For example, a gearing ratio of 90% implies that a company’s debt level is equal to 90% of its equity.

When different companies are observed, it is clear that most of the countries examined in this paper have no requirements on the level of leverage. A gearing limit, or a restriction on leverage, is a disadvantage for investors compared to other real-estate

investors, who, for example, invest directly. In general managers should be able to decide on the optimal financing of their company (Eichholtz, 2013).

(11)

11 Tax: Payout distribution

Across countries the payout system differs widely. The payout, or the dividend paid by the REIT, has different restrictions placed upon it across countries, with a couple of important consequences. First of all, clearest consequence is that it implicitly reduces leverage (Stevenson, 2013). This is because, within a REIT, not only debt repayments but also dividends are exempt from corporation tax, which is the key difference with conventional companies. This tax exemption offers investors relatively high dividend yields, which are comparatively safe, given the characteristics of the underlying real-estate asset and its stable income flow. As a result, the income of REITs is comparable to the coupon payments of bonds (Stevenson, 2013).

According to Eichholtz (2007) a high payout requirement (80–100%) is an effective governmental mechanism that mitigates organisational difficulties. Moreover, it does guarantee a maximum tax payment.

That said, there is the issue of depreciation. To what figure does the minimum dividend payout refer? Is depreciation accounted for, or not? (Stevenson, 2013). This is part of a broader range of accounting regulations. For example, in the U.S., which operates under the ‘’United States Generally Accepted Accounting Principles’’ properties are placed on the balance sheet at depreciated costs, where there is clearly no indication of actual current market value of the underlying portfolio. This is done despite the fact that the use of actual market values provides a more adequate and transparent image of dividends payments.

In the U.S., a REIT is obligated to pay out at least 90% of its income as dividends each year. However, in practice most REITs distribute all income, which includes capital gains. U.S. REITs have deduction advantages for dividends paid. When a U.S. REIT does not meet the requirement of 90%, this may be corrected in the next year by so-called deficiency dividends (Taylor and Vermeulen, 2013).

On the other hand, Dutch REITs are required to pay out all of their profits; otherwise they do not qualify as REITs. Moreover, a Dutch REIT should distribute its profits within eight months of the end of the year, and a deduction is not permitted for dividends paid (Taylor and Vermeulen, 2013).

(12)

12 An increasing number of countries have a payout system, under which the income of the REIT is taxed only once, to the owners of the REIT. This includes most European Union countries, Australia, Brazil, Canada, Japan, Singapore and the U.S. (Taylor and Vermeulen, 2013).

Capital structure

According the report of Eichholtz and Kok (2007) open-ended structures are less favourable and more unstable. As a result, closed-ended funds create more shareholder value. For all countries in our dataset, the rule is that a REIT has to be a closed-end fund. This means that there are only a fixed number of shares that can be traded in the market. Since the rules are similar for all the countries in the dataset, this component is not included in the regression.

REITs can be traded as listed, which means they are publicly traded on stock exchanges, or as unlisted, which means they are privately traded. Listed stocks are much more common, which was easily observable when analysing the number of listed REITs versus unlisted REITs. In most countries, only listed REITs are allowed. However, the U.S. and Australia are examples of countries that allow both options. Since the number of unlisted REITs in the SNL indices was negligible, this aspect is not included in the regression as well.

Tax transparency

Stevenson’s final point in his comparison of international REIT structures was tax transparency. This refers to whether a REIT should be tax-transparent to satisfy it legal requirements. The motivation behind tax transparency is the prevention of double taxation of income. Moreover it does create some room to handle direct real-estate ownership (Eichholtz, 2013). Since all countries have the same requirement that REITs should be tax-transparent this is not further discussed or included in the regression.

(13)

13 Shareholder requirements: Ownership concentration

Besides the components mentioned in Stevenson’s article, there is another aspect of REIT structure that should be discussed. This is shareholder requirements, and in particular requirements on ownership concentration. Shareholder requirements are different for all of the countries in our dataset. They do not differ greatly, but small deviations occur. Of how many owners does the national legal framework require a REIT to consist? And to what extent is a shareholder restricted in terms of his ability to hold a substantial share in a REIT?

Many investigations have been done on the relation between ownership concentration and firm performance. However all of these studies have had conflicting results. For example, Morck et al. (2000) found a positive relationship between ownership concentration and firm performance, while Claessens et al. (2002) found a negative relationship. Other studies found mixed relationship or no significant evidence at all.

The mixed outcomes are debatable, since results on ownership requirements could be interpreted to mean that lower concentrations (in other words, fewer shareholders), could lead to better performance, since there are less parties and decision-making is flexible and efficient. On the other hand, one could argue that block ownership could be a good monitoring device, which would have a positive effect on performance. In general, the absence of ownership requirements is better for the corporate governance of a company (Eihholtz, 2013).

In the U.S., REITs must have at least one-hundred owners. However, in practice, this rule does not hold, since ownership, as a measure of the value of shares and voting power, can be in the hands of only a few shareholders. (Taylor, Vermeulen 2013).

(14)

14

4. Research design

4.1 Dataset

For this empirical study, a database provided by SNL is used which consists, among other things, of performance data and governance data on real-estate worldwide. A panel

framework of 269 REITs will be generated. Data is collected from nine countries worldwide, where all developed continents are represented. In this study, data from companies in the U.S., Canada, the Netherlands, Belgium, France, U.K., Australia, Japan and Singapore is used. Furthermore, yearly data is used, and data is collected for years 2009-2013.

4.2 Explanatory Variables and hypotheses

Management

In the regression, “SelfManaged” will be the first explanatory variable. In the SNLxl index it was clear whether the management of a company was internal or external. Therefore, a firm-specific dummy variable is included, which is “0” if the management is internal and “1” if management is external.

According to regulations, Singapore and Japan are the only two countries where REITs are managed externally. REITs in Canada and the Netherlands are managed internally, while in all other countries in the dataset REITs have the freedom to choose how they manage themselves (Stevenson 2013).

The expectation is that internally managed REITs could have an advantage through focus, specialization, and higher expertise. Important decisions are made within the firm. So, internally managed REITs have a positive relationship with performance, and therefore a positive relationship is expected between internally managed REITs and excess return.

Investment restrictions on real-estate investments

The second independent variable is real estate, “Excess_RE_INV,” which is the number of real-estate investments per company minus the percentage required by the country’s legislation – in other words, excess investment in real estate per company. As an input for the percentage of real-estate investments per company, the number of properties divided by its assets is used. This is the investment in properties minus accumulated

(15)

15 depreciation and can also be explained as the percentage of net real estate divided by the assets of a company.

Gearing limit

The third firm-specific dependent variable in the regression is “Debt_To_Equity,”

since the ratio used for the gearing limit is the total debt of a company divided by its total equity. The impact of legislation on this component will be examined by comparing the results of countries with and without a limit.

Tax: Payout distribution

The fourth explanatory variable included in the regression is “Div_Excess,” which is the excess dividend ratio paid per company. The excess dividend ratio is the dividend ratio paid per company minus the rate required by the country’s legal framework. Overall, higher requirements are expected to result in higher returns.

Shareholder requirements: Ownership concentration

As proxy for ownership three variables are incorporated. First of all, the “total %

owned by top 10 insider/stakeholders.” This is the total number of shares held by ten largest

insider/stakeholders by count of shares as a percentage of total shares outstanding. The second variable is “%institutional ownership,” which is the percentage of the total shares owned by institutions according to 13F fillings. Due to options and OP units, the sum of shares owned by institutions may appear greater than 100%. The last ownership variable is “number of institutional Investors,” which is the total number of institutional investors. All three ownership variables are not historical but are fixed as if the date were June 2014. It is expected that performance will be better with lower ownership concentrations. In other words, a negative relationship between ownership requirements and excess return is expected.

(16)

16

4.3 Control variables and hypotheses

According to empirical research, a REIT’s return is influenced by its size (Manakyan et al, 1995). Furthermore, economic literature provides evidence that larger REITs will come to dominate the market (Ambrose & Lineman, 2001). This could be due to the fact that larger REITs have significant advantages, such as economies of scale in revenues, expenses, and capital. On the other hand, larger REITs are more liquid and have lower risk. Because of the low risk, a negative effect is expected.

To control for the impact of REIT size on performance, the log of the market

capitalization is incorporated in the regression. The market capitalization of a company is the total value of the shares issued and is equal to the total number of shares multiplied by its value. It is therefore a clear indication of public opinion of the worthiness of the company. More importantly, market capitalization is a solid proxy for company size.

To control for macroeconomic country-specific aspects, a few control variables are included. Multiple studies exist on the relevance of macroeconomic variables that have a significant role in determining real-estate returns.

GDP growth

The first control variable is “GDPgrowth,” which is the percentage change in real GDP per year in a country. GDP growth is an important economic indicator and a useful proxy for the demand in a country for commercial real estate. It is expected that a higher level of GDP growth implies more demand for real estate, so a positive relationship is expected.

Spreads

The second control variable is “TEDSpread,” which measures the difference between the three-month LIBOR and the three-month T-bond interest rate. This spread is an indicator of credit risks in each country. This follows from the fact that T-bills are risk-free, while LIBOR is the rate that banks mutually charge each other. An increasing “TEDSpread” implies more credit risk. The change in risk has an effect on estate supply, since the success of real-estate developers depends on the availability of funds. It is expected is that a larger spread will result in lower excess returns for real estate (Edelstein, 2010).

(17)

17

4.4 Excess Return

The excess return per company is formulated on the left-hand side of the econometric equation. The excess return is the total return minus the local risk-free rate. The total return as a percentage is the total return of a security over a time period, in this case a year,

including price appreciation and the reinvestment of dividends. The returns collected were index numbers for the period 2008 to 2013, with end of 2007 as reference point. In SNL data, only monthly returns were available. Using a simple formula, yearly returns were calculated. For the risk-free rates, ten-year T-bonds are used.

4.5 Statistical relationship/econometric model

The regression that will be used is derived from the methodology used in Edelstein (2011). The regression will run on firm-level with some country-level control variables:

𝑅

_𝑖

-𝑅

_𝑓

=

α+ 𝛽

₁

[𝐿𝑜𝑔𝑆𝑖𝑧𝑒

_𝑖

]+𝛽

₂

[𝑀𝑎𝑛𝑎𝑔𝑒𝑚𝑒𝑛𝑡

_𝑖

]+ 𝛽

₃

[𝐸𝑥𝑐𝑒𝑠𝑠_𝑅𝐸_𝐼𝑁𝑉

_𝑖

] +

𝛽

₄

[𝐷𝑒𝑏𝑡_𝑇𝑜_𝐸𝑞𝑢𝑖𝑡𝑦

_𝑖

] + 𝛽

₅

[𝐷𝑖𝑣_𝐸𝑥𝑐𝑒𝑠𝑠

_𝑖

] + 𝛽

₆

[𝑂𝑤𝑛𝑒𝑟𝑠ℎ𝑖𝑝 +

β

₇

[𝑇𝑒𝑑𝑆𝑝𝑟𝑒𝑎𝑑

_𝑐

] + β

₈

[𝑅𝑒𝑎𝑙𝐺𝐷𝑃𝑔𝑟𝑜𝑤𝑡ℎ

_𝑐

] + 𝜀

_𝑖

To set up the model we have 𝑅𝑖-𝑅𝑓 on the left-hand side, where 𝑅𝑖 is the annual return of the

REIT and 𝑅𝑓 is the local risk-free rate. The 𝑖 is the firm, and c stands for the specific country.

To summarize, the independent variables are “SelfManaged,” “Excess_RE_INV,”

“Debt_To_Equity,” “Div_Excess,” “Lsize,” “Ownership,” and the country control variables

“TEDSpread” and “Real GDP growth.” In the independent firm-specific explanatory variables, the regulation rules per country are incorporated. Too clarify, for example, in the U.S., the variable “Div_Excess” is the dividend paid by a company minus the required rate of 90%, while in in Canada only 85% is subtracted.

(18)

18

5. Results

5.1 Descriptive statistics

In Table 1 the descriptive statistics are shown. In Appendix 2 more detailed descriptive statistics are shown. The dependent variable is the excess return, or, return minus the local risk-free rate, with a mean of 20.06. The minimum and maximum values differ greatly, as does the standard deviation. The total number of observations is 1,344. This is the number of observations for 269 companies, from nine countries divided over five years (2009-2013). For variables with fewer observations, some data is missing, since it was not monitored or available.

Table 1 Descriptive statistics

Variable Observations Mean SD Min Max

Excess Return (%) 1344 20,06 74,70 -84,13 2299,55 Lsize 1271 6,73 1,65 -0,49 10,80 SelfManaged 1080 0,37 0,48 0 1 Totalownedtop10 1304 23,03 24,61 0 93,73 Institutions (%) 1329 45,14 31,76 0 119,37 #ofinstitutions 1304 147,24 159,91 1 844 Excess_RE_INV (%) 1272 11,50 30,25 -98,89 97,19 Debt_to_Equity 1256 1,28 2,29 0 65,03 Div_Excess (%) 936 25,86 125,38 -100 795 Real GDP growth 1344 1,34 3,30 -5,25 15,06 Tedspread (%) 1344 0,07 0,94 0,09 4,82

“Lsize” is the log of “marketcap” and has a mean of 6.734. Surprisingly, this is a

positive number, in contradiction to Edelstein’s “Lsize.” “SelfManaged” is a dummy for internal and external management. The mean is 0.37, which means that more than half of the companies are internally managed. Real GDP growth is 1.33% per year on average. This is low, but logical, since the years in question were economically difficult. “TEDSpread” is .700, which is comparable with the spread in Edelstein’s article. In the next table the mean per country and region are shown.

(19)

19

Table 2. Mean per country and region

Country Debt to Dividend Excess Market RealGDP ted Excess Log

Equity Excess RE Invest Cap growth spread return size

Australia 1,048 51,444 25,674 1.579,076 2,478 3,984 18,482 6,475 Belgium 0,894 69,242 -19,203 553,226 0,276 0,636 5,110 5,980 Canada 2,392 15,623 14,230 967,691 1,382 0,954 16,055 5,754 France 1,738 39,099 75,443 2.980,387 0,186 0,636 14,121 6,845 Japan 1,101 3,740 14,478 1.493,349 0,348 0,156 19,366 6,923 Netherlands 1,385 16,650 -25,628 1.438,728 -0,636 0,636 -4,170 6,041 Singapore 0,532 -48,695 -1,602 1.954,111 5,288 0,156 28,649 7,070 UK 1,165 -9,324 -1,991 1.290,699 -0,077 0,696 30,190 5,886 USA 1,378 7,090 11,289 4.786,282 1,244 0,290 19,219 7,722 Asia-Pacific region 0,859 -10,144 10,278 1697,921 2,720 0,931 22,859 6,888 Europe Region 1,282 17,847 11,936 1595,822 -0,010 0,668 19,693 6,140 North America region 1,669 59,382 12,189 3624,102 1,295 0,537 18,056 7,123

From Table 2 it is clear that the biggest companies are active in the U.S. The lowest debt level compared to equity is used in Singapore, which also saw the highest growth in real GDP. The Netherlands is the only country with an average negative mean, which maybe is a result of the poor circumstances in the Dutch national real-estate market.

Table 3 reports the cross-correlation matrix for the regression variables. Apart from the ownership variables, all independent variables are negatively correlated with excess return. Size, on the other hand, has a positive correlation with excess return, which is comparable with Edelstein’s findings. However, GDP growth has a negative correlation with excess return, which differs from what was expected. An obvious correlation is clear

between the number of investors and the log of the market capitalization. Furthermore, there clearly is a high correlation between the institutional variables.

(20)

20

Table 3a. Cross-correlation matrix

Excess Return Lsize SelfManaged Totalownedtop10 Institutions

(%) #ofinstitutions Excess Return (%) 1 Lsize 0,046 1,000 SelfManaged -0,003 -0,168 1,000 Totalownedtop10 0,003 -0,290 -0,021 1,000 Institutions (%) 0,041 0,528 -0,300 -0,686 1,000 #ofinstitutions 0,002 0,729 -0,305 -0,451 0,725 1,000 Excess_RE_INV (%) -0,027 0,096 -0,116 -0,075 0,008 -0,040 Debt_to_Equity -0,018 -0,185 0,025 -0,047 -0,056 -0,029 Div_Excess (%) -0,027 0,008 -0,035 -0,176 0,171 0,085 Real GDP growth -0,158 0,111 0,003 0,098 -0,113 -0,040 Tedspread (%) -0,032 -0,014 -0,103 0,032 -0,139 -0,017

Table 3b. Cross-correlation matrix continued

Excess_RE_INV (%) Debt_to _Equity Div_Excess (%) Real GDP growth Tedspread (%) Excess Return (%) Lsize SelfManaged Totalownedtop10 Institutions (%) #ofinstitutions Excess_RE_INV (%) 1,000 Debt_to_Equity 0,058 1,000 Div_Excess (%) 0,097 0,134 1,000 Real GDP growth -0,049 -0,073 -0,135 1,000 Tedspread (%) 0,135 0,019 0,038 -0,010 1,000

(21)

21

5.2 Main results

The main results are discussed in this chapter. To begin, a primary regression analysis is shown in Table 4. Then, the variables are discussed individually. Results are discussed in terms of their economic significance and are related back to the hypotheses. Finally, the outcome is placed in the context of the more general field of research.

Table 4. Primary regression analysis

Excess Return Coefficient SD Z-value P>|z|

Lsize - 3,88** 1,73 (2.23) 0,026 SelfManaged 5,51 3,38 (1.63) 0,103 Totalownedtop10 0,27*** 0,09 (2.86) 0,004 Institutions (%) 0,12 0,08 (1.43) 0,154 #ofinstitutions 0,03* 0,02 (1.82) 0,068 Excess_RE_INV (%) -0,17*** 0,06 (3.03) 0,002 Debt_to_Equity -0,68 1,44 (0.47) 0,637 Div_Excess (%) -0,01 0,01 (1.06) 0,291 Real GDP growth -2,33*** 0,43 (5.43) 0,000 Tedspread (%) -2,21 1,87 (1.18) 0,237 Constant 35,65*** 12,03 (2.96) 0,003 Observations 761 R-sqaured 7.21

Variables are discussed individually below. First of all, “Lsize,” the log of the “marketcap” is significant at the 5% level and has a negative effect on excess return. This implies that overall, the larger the company, the lower the excess returns. This is in alignment with expectations, since a negative effect can arise due to liquidity and risk effects.

Secondly, the variable “SelfManaged” is discussed. The effect is almost significant and shows that when management is external (the dummy variable is “1”) this has a positive effect on excess return. This not in line with expectations, since it was the hypothesized that internally managed REITs had advantages over externally managed REITs. A possibility is that externally managed firms have higher returns due the fact that higher expertise and specific knowledge is acquired outside the company.

(22)

22 The three ownership variables all have a positive effect on excess return. However, the variable “% owned by top 10 insider/stakeholders” is the mostinteresting to examine, since it is the best proxy of ownership concentration. This variable is significant at the 1% level and has a positive effect on excess return. This is in line with expectations, since a higher

percentage of total % owned by top 10 insider/stakeholders implies a lower concentration and better performance. This outcome, however, should be interpreted with care, since the ownership variables are fixed over time.

The next variable that is discussed is “Excess_RE_INV.”In Table 4 it is shown that the effect is negative and significant at the 1% level. This implies that one should minimize excess investment in real estate, and one could argue that countries’ restrictions should be high to achieve a lower excess return.

To continue, the effect of “Debt_To_Equity,” which is the proxy for the gearing limit, is not significant. Most countries have no requirements on the gearing limit. Perhaps this is because managers should be free to decide on the optimal financing of their company.

The variable “Div_Excess” unfortunately is not significant as well. If this fact were neglected, the negative coefficient could be interpreted as in alignment with expectations, since a larger excess has a negative effect on excess return. Therefore, it is favourable to minimize excess returns, and legal structures should set higher requirements, since higher requirements are expected to result in higher returns.

Finally the country control variables are discussed. “GDPgrowth” is significant at the 1% level and has a negative effect on excess returns. This is not consistent with prior studies. “TEDSpread” is negative, which is in alignment with expectations.

(23)

23

5.3 Additional results

In this section, some additional results are discussed. First of all, the primary results are problematic, since the institutional ownership variables are highly correlated. Secondly, it is also problematic that the institutional variables are not exogenous, since institutional investors are attracted by larger and more profitable companies with low risk and low return volatility. This relationship is also seen in the correlation matrix where “InstitutionalOwnership” and

“numberofinstitutionalinvestors” are highly correlated with “Lsize.”

As a solution, some regressions were run, including only one of the explanatory variables and the control variables. These regressions are shown in Table 5. Column “a” has “Div_Excess” as the only explanatory variable. Columns “b,” “c,” and “d” have, respectively, “Excess_RE_INV,”

“Debt _To_Equity,” and “SelfManaged” as the only explanatory variable.

Table 5. Extra regressions

Dependent ( a ) ( b ) ( c ) ( d ) ( e )

variable Ri-Rf Ri-Rf Ri-Rf Ri-Rf Ri-Rf

Lsize 0.69 3.16** 2.58* 1.41* -0.56 (0,85) (2,42) (1.85) (1,73) (0,55) Div_Excess -0.01 -0.011 (1,42) (1,02) Excess_RE_Inv -0.1 -0.17*** (1,48) (3,05) Debt_to_equity -0.65 -1.81 (0,69) (1,37) SelfManaged -1.59 0.69 (0,61) (0,24) RealGDPgrowth -1.96*** -3.76*** -3.72*** -2.49*** -2.34*** (5,26) (5,87) (5,75) (6,98) (5,67) Tedspread -3.19** 0.2 -0.45 -2.98* -3.73** (2,56) (0,09) (0,20) (1,90) (2,06) Constant 17.56*** 4.51 8.56 14.3** 31.99*** (2,93) (0,48) (0,83) (2,32) (3,82) Observations 931 1266 1250 1022 782 Adjusted R-sqaured 3.71 2.96 2.75 4.99 5.55

* stand for 10% significance ** stand for 5 % significance *** stand for 1 % significance

(24)

24 In column “e,” the institutional variables are not included at all, and a regression is made with all the other dependent variables and the control variables. It is arguable that explanatory

variables should be lagged to measure the impact on performance. Therefore, a regression is done where the explanatory variables are lagged. The outcomes are shown in Table 6. Unfortunately, those outcomes are not significant and give no new insights.

Table 6. Regression with lagged explanatory variables

Dependent variable Ri-Rf

Lsize 0.55 (0,40) Div_Excess (lagged) -0.01 (1,24) Excess_RE_Inv (lagged) 0.053 (1,20) Debt_to_equity (lagged) 1.23 (1,15) SelfManaged 2.19 (0,84) Totalownedtop10 -0.04 (0,62) Institutional % -0.01 (0,17) # of institutions 0.002 (0,17) RealGDPgrowth -0.03 (0,07) Tedspread -3.88*** (2,62) Constant 6.13 (0,63) Observations 587 Adjusted R-sqaured 2.52

Finally, a regression analysis was run per country. The outcomes for countries with fewer companies in the dataset, however, were not useful, due to the lack of observations. Most of the results from other countries were not significant and did not give any new insights. The outcomes are shown in Appendix 3.

(25)

25

6. Conclusion

The objective of this thesis is to get a clear picture of the influence of the different components of REIT structure on performance. A panel framework with almost 200

companies from nine different countries is used, for the period 2009-2013. Data is obtained from the SNL database. REIT structure is separated on the basis of real-estate investment requirements, overseas investments and development, gearing limits, payout distributions, capital structure, tax-transparency and ownership requirements. Since most components are the same across the countries used in the database, only management, real-estate

investment requirements, gearing limits, payout distributions, and ownership requirements are tested.

In terms of management requirements, the results were more in favour of external management. However, Table 5 shows that this variable is not significant. Most companies know best if they should outsource their management or not. Therefore, it is advised that the choice of management should be optional, which is already the case in some countries.

Real-estate investment requirements have a positive effect. Since excess investments have a negative effect on returns, high requirements are advised. Apart from France, all countries already incorporate a minimum of 75% on real-estate investment requirements. In line with this result, these figures should be raised to 80-85%.

Furthermore, gearing limits did not prove significant. Managers should decide on their optimal level of debt. Therefore, an optimal REIT structure should include no requirements whatsoever. This is already the case in most countries.

Although findings on payout requirement are not significant, a high payout requirements is advised. First of all, this is because higher requirements are expected to result in higher returns. Secondly, high payout requirements are an effective governmental mechanism guaranteeing a maximum tax payment. Overall a payout requirement of 90% or more is advised.

Finally, a recommendation regarding ownership requirements is made. The results suggest that ownership concentration has a positive effect on REIT performance. Therefore minimal requirements on ownership are advised.

(26)

26 However the above advice should all be taken with care. This study has been conducted with multiple assumptions. Proxies used for the different variables are sometimes simplified. One could do a more intensive research study on the individual aspects of REIT structures

separately, for example focusing only on the management aspect or real-estate investment restrictions. Such research could serve as a take-off point for more investigations on REIT structures.

(27)

27

7. References

Anderson, R., Clayton, J., Mackinnon, G., & Sharma, R. (2005). REIT returns and pricing: the small cap value stock factor. Journal of Property Research,22(04), 267-286.

Ambrose, B. W., & Linneman, P. (2001). REIT organizational structure and operating characteristics. Journal of Real Estate Research, 21(3), 141-162.

Brounen, D., & de Koning, S. (2012). 50 Years of Real Estate Investment Trusts: An International Examination of the Rise and Performance of REITs. Journal of Real Estate Literature, 20(2), 197-223.

Brounen, D., & Eichholtz, P. M. (2003). Property, common stock, and property shares. The journal of portfolio management, 29(5), 129-137.

Chen, S. J., Hsieh, C., Vines, T. W., & Chiou, S. N. (1998). Macroeconomic variables, firm-specific variables and returns to REITs. Journal of Real Estate Research, 16(3), 269-278. Chui, A. C., Titman, S., & Wei, K. C. (2003). The cross section of expected REIT returns. Real Estate Economics, 31(3), 451-479.

Claessens, S., Djankov, S., Fan, P.H., Lang, L.H.P., 2002. Disentangling the incentive and entrenchment effects of large shareholdings, Journal of Finance, 57, 6, 2741-2771

Clayton, J., & MacKinnon, G. (2001). The time-varying nature of the link between REIT, real estate and financial asset returns. Journal of Real Estate Portfolio Management, 7(1), 43-54. Clayton, J., & MacKinnon, G. (2003). The relative importance of stock, bond and real estate factors in explaining REIT returns. The Journal of Real Estate Finance and Economics, 27(1), 39-60.

Derwall, J., Huij, J., Brounen, D., & Marquering, W. (2009). REIT momentum and the performance of real estate mutual funds. Financial Analysts Journal, 24-34.

Edelstein, R., W. Qian, and D. Tsang (2011): “How Do Institutional Factors affect

International Real Estate Returns?,” Journal of Real Estate Finance and Economics 43, 130-151.

(28)

28

Eichholtz, P., & Kok, N. (2007). The EU Reit and the internal market for real estate. Available at SSRN.

Ernst and Young (2012) Reit report 2012. Source: SNL financial

http://www.ey.com/Publication/vwLUAssets/2012_REIT_report/$FILE/2012REITReport.pdf Han, J., & Liang, Y. (1995). The historical performance of real estate investment

trusts. Journal of Real Estate Research, 10(3), 235-262.

Ling, D. C., & Naranjo, A. (2002). Commercial real estate return performance: a cross-country analysis. The Journal of Real Estate Finance and Economics, 24(1-2), 119-142.

Morck, R., Nakamura, M., Shivdasani, A., 2000. Banks, ownership and firm value in Japan, Journal of Business, 73, 3, 539-567.

Neil Myer, F. C., & Webb, J. R. (1993). Return properties of equity REITs, common stocks, and commercial real estate: a comparison. Journal of Real Estate Research, 8(1), 87-106.

Pagliari, J. L., Scherer, K. A., & Monopoli, R. T. (2005). Public Versus Private Real Estate Equities: A More Refined, Long‐Term Comparison. Real Estate Economics, 33(1), 147-187. Redman, A. L., & Manakyan, H. (1995). A multivariate analysis of REIT performance by financial and real asset portfolio characteristics. The Journal of Real Estate Finance and Economics, 10(2), 169-175.

Serrano, C., & Hoesli, M. (2009). Global securitized real estate benchmarks and performance. Journal of Real Estate Portfolio Management, 15(1), 1-19.

Stevenson, S. (2013). The Global Real Estate Investment Trust Market: Development and Growth. In Real Estate Investment Trusts in Europe (pp. 17-25). Springer Berlin Heidelberg. Worzala, E., and C.F. Sirmans (2003): “Investing in International Real

(29)

29

8. Appendix

(30)

30

Appendix 2. Descriptive statistics per country

Country 1 = Australia, Country 2 = Belgium, Country 3 = Canada, Country 4 = France, Country 5 = Japan, Country 6 = Netherlands, Country 7 = Singapore, Country 8 = UK and Country 9 = USA Lsize 85 6.475293 1.618156 1.71702 8.942723 tedspread 85 3.984 .6396859 3.07 4.82 RealGDPgro~h 85 2.478 .6643862 1.54 3.59 MarketCap 85 1579.076 1957.996 5.567911 7652.006 Div_Excess_ 55 51.44426 156.1337 -75.08067 705.5556 Debt_To_Eq~_ 85 1.047647 1.485379 .03 7.72 Excess_RE_~_ 84 25.67406 25.26088 -49.67 49.25997 NumberofIn~r 85 86.35294 83.31019 8 251 Institutio~p 85 34.88588 14.88233 11 60.54 TotalOwned~r 85 25.12294 24.23458 .24 85.85 Selfmanage~o 40 .25 .438529 0 1 Excessreturn 85 18.48181 44.68913 -84.13238 201.9376 Variable Obs Mean Std. Dev. Min Max -> CountryName = 1

> _Equity_ Div_Excess_ MarketCap RealGDPgrowth tedspread Lsize

> er InstitutionalOwnership NumberofInstitutionalInvestor Excess_RE_Inv_ Debt_To . bysort CountryName: sum Excessreturn SelfmanagedYesNo TotalOwnedbyTop10Insid

Lsize 59 5.980195 .8053165 4.25044 7.645123 tedspread 60 .636 .437788 .11 1.3 RealGDPgro~h 60 .276 1.806594 -2.8 2.32 MarketCap 59 553.2257 512.0307 70.13626 2090.426 Div_Excess_ 43 69.24214 177.7612 -80 795 Debt_To_Eq~_ 58 .8941379 .4434278 .04 1.8 Excess_RE_~_ 59 -19.20252 26.83585 -77.21177 -.5207647 NumberofIn~r 55 34.72727 38.85894 3 123 Institutio~p 60 9.870833 10.68624 0 31.02 TotalOwned~r 60 47.60083 25.6035 .29 93.73 Selfmanage~o 30 .5 .5085476 0 1 Excessreturn 60 5.11007 15.02759 -44.4874 43.31654 Variable Obs Mean Std. Dev. Min Max -> CountryName = 2

(31)

31 Lsize 140 5.754043 1.722755 1.590905 9.185433 tedspread 190 .954 .2791309 .59 1.22 RealGDPgro~h 190 1.382 2.127704 -2.71 3.37 MarketCap 140 967.6907 1640.457 4.908189 9754.007 Div_Excess_ 80 15.62275 142.9547 -85 793.6999 Debt_To_Eq~_ 129 2.39186 5.781925 .17 65.03 Excess_RE_~_ 141 14.22969 5.713947 -15.80074 19.35 NumberofIn~r 175 45.51429 46.86577 1 172 Institutio~p 185 27.29892 20.94554 0 67.35 TotalOwned~r 185 16.64892 20.01613 0 67.59 Selfmanage~o 175 .4 .4913037 0 1 Excessreturn 186 16.05522 34.28831 -76.64764 137.1914 Variable Obs Mean Std. Dev. Min Max -> CountryName = 3 Lsize 96 6.845323 1.756344 1.927295 10.12505 tedspread 100 .636 .4363114 .11 1.3 RealGDPgro~h 100 .186 1.807432 -3.07 2.03 MarketCap 96 2980.387 4770.473 6.870902 24960.42 Div_Excess_ 63 39.09854 161.8771 -85 722.6923 Debt_To_Eq~_ 97 1.737938 2.747517 0 17.38 Excess_RE_~_ 97 75.44276 28.87933 .5437606 97.19352 NumberofIn~r 100 92.65 115.4155 1 428 Institutio~p 100 25.5 18.24731 .04 63.81 TotalOwned~r 95 47.73316 21.29576 2.18 84.33 Selfmanage~o 60 .1666667 .375823 0 1 Excessreturn 100 14.121 38.85544 -77.38582 184.3888 Variable Obs Mean Std. Dev. Min Max -> CountryName = 4 Lsize 156 6.922693 .895934 4.261911 9.078675 tedspread 170 .156 .0806365 .09 .31 RealGDPgro~h 170 .348 3.370012 -5.52 4.68 MarketCap 156 1493.349 1504.89 70.94543 8766.343 Div_Excess_ 117 3.740245 21.3135 -79.43758 58.35501 Debt_To_Eq~_ 156 1.100833 .5422057 .16 3.52 Excess_RE_~_ 156 14.47807 9.315429 -33.44477 22.50671 NumberofIn~r 170 83.79412 52.94118 17 192 Institutio~p 170 40.24647 11.50762 8.82 64.02 TotalOwned~r 155 15.8671 17.46559 .14 67.47 Selfmanage~o 150 1 0 1 1 Excessreturn 167 19.36639 35.72229 -52.18336 169.2816 Variable Obs Mean Std. Dev. Min Max -> CountryName = 5

(32)

32 Lsize 35 6.041114 2.59832 -.4936813 8.676105 tedspread 35 .636 .4404623 .11 1.3 RealGDPgro~h 35 -.636 1.852388 -3.65 1.47 MarketCap 35 1438.728 1531.462 .6103753 5861.176 Div_Excess_ 14 16.64957 87.59416 -100 236.6667 Debt_To_Eq~_ 32 1.385 1.939478 .43 10.75 Excess_RE_~_ 35 -25.628 33.77333 -98.89 -.7635103 NumberofIn~r 30 128.3333 77.14891 13 260 Institutio~p 30 49.15667 18.56795 15.92 71.55 TotalOwned~r 25 20.36 22.87174 3.18 62.5 Selfmanage~o 30 .5 .5085476 0 1 Excessreturn 35 -4.169545 40.57606 -83.05 125.3808 Variable Obs Mean Std. Dev. Min Max -> CountryName = 6 Lsize 164 7.069585 1.011111 3.67873 9.463053 tedspread 165 .156 .0806438 .09 .31 RealGDPgro~h 165 5.288 5.375517 -.6 15.06 MarketCap 164 1954.111 2344.301 39.59609 12875.13 Div_Excess_ 130 -48.69549 40.78992 -100 151.5777 Debt_To_Eq~_ 164 .5321951 .3024096 .04 2.18 Excess_RE_~_ 164 -1.602102 29.08566 -61.62997 29.11012 NumberofIn~r 165 81.69697 71.66125 6 243 Institutio~p 165 19.66121 17.15791 .94 70.83 TotalOwned~r 165 43.08364 24.2482 .74 86.26 Selfmanage~o 150 .4666667 .500559 0 1 Excessreturn 165 28.64876 55.63685 -53.65786 366.0036 Variable Obs Mean Std. Dev. Min Max -> CountryName = 7 Lsize 216 5.885645 1.875579 .9735458 9.19385 tedspread 219 .6960274 .1729035 .46 .92 RealGDPgro~h 219 -.0774886 2.614514 -5.17 1.76 MarketCap 216 1290.699 2024.682 2.647315 9836.441 Div_Excess_ 130 -9.323628 111.5927 -90 576.6667 Debt_To_Eq~_ 215 1.165302 1.497498 0 12.27 Excess_RE_~_ 216 -1.990775 26.20233 -74.64 23.90237 NumberofIn~r 204 137.3971 127.8705 1 473 Institutio~p 214 51.82832 30.22483 0 92.47 TotalOwned~r 214 24.04444 23.95309 .08 85.32 Selfmanage~o 125 .16 .3680813 0 1 Excessreturn 216 30.18965 161.2483 -81.20034 2299.547 Variable Obs Mean Std. Dev. Min Max -> CountryName = 8

(33)

33

Appendix 3. Regressions per country

Country 1 = Australia, Country 2 = Belgium, Country 3 = Canada, Country 4 = France, Country 5 = Japan, Country 6 = Netherlands, Country 7 = Singapore, Country 8 = UK and Country 9 = USA. Belgium and Netherlands are missing due too few observations in the data set. Outcomes are shown numerically.

Lsize 320 7.72178 1.286684 2.93624 10.79943 tedspread 320 .29 .0788634 .19 .41 RealGDPgro~h 320 1.244 2.056882 -2.8 2.78 MarketCap 320 4786.282 6892.884 18.84486 48992.7 Div_Excess_ 304 70.89753 130.3362 -90 754.4444 Debt_To_Eq~_ 320 1.377906 1.195731 0 8.97 Excess_RE_~_ 320 11.28926 13.58207 -49.12484 23.74724 NumberofIn~r 320 330.9844 183.9836 1 844 Institutio~p 320 81.80109 24.57788 .31 119.37 TotalOwned~r 320 6.879063 12.02741 .11 58.83 Selfmanage~o 320 .125 .3312369 0 1 Excessreturn 320 19.21891 37.50228 -50.03805 380.6567 Variable Obs Mean Std. Dev. Min Max -> CountryName = 9

rho 0 (fraction of variance due to u_i) sigma_e 8.9970573 sigma_u 0 _cons 106.5554 62.92776 1.69 0.090 -16.7807 229.8916 Lsize -30.47109 9.303992 -3.28 0.001 -48.70658 -12.2356 tedspread -1.853618 3.054538 -0.61 0.544 -7.840402 4.133167 RealGDPgrowth 20.28006 3.97303 5.10 0.000 12.49306 28.06705 MarketCap .0060796 .0034816 1.75 0.081 -.0007442 .0129035 Div_Excess_ -.0074778 .0113943 -0.66 0.512 -.0298103 .0148547 Debt_To_Equ~_ -13.59648 15.40396 -0.88 0.377 -43.78769 16.59474 Excess_RE_I~_ -.1394278 .1821697 -0.77 0.444 -.4964738 .2176182 NumberofIns~r .3105302 .1118914 2.78 0.006 .0912271 .5298333 Institution~p .2912002 .6412021 0.45 0.650 -.9655327 1.547933 TotalOwnedb~r .9509471 .4800446 1.98 0.048 .0100771 1.891817 Selfmanaged~o 19.20417 7.611804 2.52 0.012 4.285311 34.12303 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(11) = 43.37 overall = 0.7430 max = 4 between = 0.9454 avg = 3.4 R-sq: within = 0.7014 Obs per group: min = 2

(34)

34 _cons 59.78888 36.29003 1.65 0.099 -11.33826 130.916 Lsize -5.29591 6.969967 -0.76 0.447 -18.95679 8.364974 tedspread -30.82513 13.01633 -2.37 0.018 -56.33666 -5.313603 RealGDPgrowth -2.745592 2.19448 -1.25 0.211 -7.046694 1.555511 MarketCap -.0057249 .0034873 -1.64 0.101 -.01256 .0011101 Div_Excess_ -.0780914 .0342184 -2.28 0.022 -.1451582 -.0110246 Debt_To_Equ~_ 1.720382 3.19462 0.54 0.590 -4.540958 7.981723 Excess_RE_I~_ -1.364706 1.156552 -1.18 0.238 -3.631505 .9020941 NumberofIns~r .3191829 .2254338 1.42 0.157 -.1226592 .761025 Institution~p .4823172 .3320326 1.45 0.146 -.1684547 1.133089 TotalOwnedb~r .8655012 .3371717 2.57 0.010 .2046567 1.526346 Selfmanaged~o 25.53039 11.09709 2.30 0.021 3.780486 47.28029 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0001 Wald chi2(11) = 37.80 overall = 0.3740 max = 5 between = 0.5467 avg = 3.0 R-sq: within = 0.3486 Obs per group: min = 1 Group variable: SNLInstitu~y Number of groups = 23 Random-effects GLS regression Number of obs = 70 -> CountryName = 3 _cons -7.456239 60.64376 -0.12 0.902 -126.3158 111.4033 Lsize 7.751378 8.841134 0.88 0.381 -9.576926 25.07968 tedspread -21.49244 8.52156 -2.52 0.012 -38.19439 -4.790494 RealGDPgrowth -4.628986 2.27719 -2.03 0.042 -9.092196 -.1657765 MarketCap .0037905 .0053474 0.71 0.478 -.0066902 .0142713 Div_Excess_ -.001765 .0186762 -0.09 0.925 -.0383697 .0348398 Debt_To_Equ~_ 7.183214 6.363479 1.13 0.259 -5.288975 19.6554 Excess_RE_I~_ -.0180148 .1451211 -0.12 0.901 -.3024469 .2664172 NumberofIns~r -.1878237 .1704461 -1.10 0.270 -.521892 .1462446 Institution~p -.4194911 .6404143 -0.66 0.512 -1.67468 .8356979 TotalOwnedb~r -.2159971 .4687104 -0.46 0.645 -1.134653 .7026584 Selfmanaged~o -1.424448 21.85119 -0.07 0.948 -44.25199 41.40309 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0123 Wald chi2(11) = 24.10 overall = 0.4909 max = 5 between = 0.7151 avg = 3.4 R-sq: within = 0.4469 Obs per group: min = 1 Group variable: SNLInstitu~y Number of groups = 11 Random-effects GLS regression Number of obs = 37

(35)

35 _cons 86.72081 64.79687 1.34 0.181 -40.27872 213.7203 Lsize -17.62554 8.2489 -2.14 0.033 -33.79309 -1.457997 tedspread -73.1803 140.7381 -0.52 0.603 -349.0219 202.6613 RealGDPgrowth 2.561528 3.288099 0.78 0.436 -3.883028 9.006084 MarketCap .0018218 .0047181 0.39 0.699 -.0074255 .0110692 Div_Excess_ -.0501705 .2077294 -0.24 0.809 -.4573127 .3569718 Debt_To_Equ~_ -5.647536 9.955181 -0.57 0.571 -25.15933 13.86426 Excess_RE_I~_ -.1084802 .6677574 -0.16 0.871 -1.417261 1.2003 NumberofIns~r .1608427 .1358718 1.18 0.236 -.1054611 .4271465 Institution~p 1.17087 .5554923 2.11 0.035 .0821251 2.259615 TotalOwnedb~r .6192553 .2743311 2.26 0.024 .0815761 1.156934 Selfmanaged~o 0 (omitted) Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = . Wald chi2(10) = . overall = 0.2624 max = 5 between = 0.7028 avg = 3.9 R-sq: within = 0.2006 Obs per group: min = 1 Group variable: SNLInstitu~y Number of groups = 25 Random-effects GLS regression Number of obs = 98 -> CountryName = 5 _cons -35.8816 75.17785 -0.48 0.633 -183.2275 111.4643 Lsize -.2046461 12.55508 -0.02 0.987 -24.81216 24.40286 tedspread 461.3874 81.46795 5.66 0.000 301.7132 621.0617 RealGDPgrowth -.7694748 1.047132 -0.73 0.462 -2.821816 1.282866 MarketCap .0031874 .0045164 0.71 0.480 -.0056646 .0120393 Div_Excess_ -.0522704 .1267247 -0.41 0.680 -.3006463 .1961054 Debt_To_Equ~_ 1.474532 16.92128 0.09 0.931 -31.69056 34.63962 Excess_RE_I~_ -.2100426 .275787 -0.76 0.446 -.7505753 .33049 NumberofIns~r -.132406 .1251928 -1.06 0.290 -.3777794 .1129674 Institution~p -.1043647 .6070241 -0.17 0.863 -1.29411 1.085381 TotalOwnedb~r -.0506061 .5234768 -0.10 0.923 -1.076602 .9753896 Selfmanaged~o 18.477 19.90617 0.93 0.353 -20.53837 57.49238 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(11) = 62.17 overall = 0.3611 max = 5 between = 0.3048 avg = 4.1 R-sq: within = 0.3753 Obs per group: min = 2 Group variable: SNLInstitu~y Number of groups = 30 Random-effects GLS regression Number of obs = 122 -> CountryName = 7

(36)

36

rho .02235955 (fraction of variance due to u_i)

sigma_e 19.973239 sigma_u 3.020581 _cons 83.18654 29.16996 2.85 0.004 26.01447 140.3586 Lsize -.5827764 3.96269 -0.15 0.883 -8.349507 7.183954 tedspread -111.6417 18.71046 -5.97 0.000 -148.3136 -74.96988 RealGDPgrowth -11.35348 1.993943 -5.69 0.000 -15.26154 -7.445425 MarketCap -.001848 .002929 -0.63 0.528 -.0075887 .0038928 Div_Excess_ .023466 .024899 0.94 0.346 -.0253352 .0722672 Debt_To_Equ~_ 1.010457 5.630085 0.18 0.858 -10.02431 12.04522 Excess_RE_I~_ -.1460318 .1256769 -1.16 0.245 -.392354 .1002904 NumberofIns~r .0248081 .0728278 0.34 0.733 -.1179318 .167548 Institution~p .17264 .2235583 0.77 0.440 -.2655262 .6108062 TotalOwnedb~r .3488755 .2192269 1.59 0.112 -.0808013 .7785523 Selfmanaged~o 1.689512 7.291643 0.23 0.817 -12.60184 15.98087 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(11) = 57.74 overall = 0.4601 max = 5 between = 0.3695 avg = 3.2 R-sq: within = 0.4628 Obs per group: min = 1 Group variable: SNLInstitu~y Number of groups = 25 Random-effects GLS regression Number of obs = 80 -> CountryName = 8 more _cons 63.16157 24.59249 2.57 0.010 14.96117 111.362 Lsize -13.5055 4.488866 -3.01 0.003 -22.30351 -4.70748 tedspread -3.691249 35.31449 -0.10 0.917 -72.90638 65.52388 RealGDPgrowth -6.979771 1.398662 -4.99 0.000 -9.721098 -4.238444 MarketCap -.0004369 .0005063 -0.86 0.388 -.0014293 .0005555 Div_Excess_ -.0159137 .0156021 -1.02 0.308 -.0464932 .0146658 Debt_To_Equ~_ 1.939566 1.64357 1.18 0.238 -1.281771 5.160904 Excess_RE_I~_ -.2706078 .1517878 -1.78 0.075 -.5681065 .0268909 NumberofIns~r .0931336 .0305061 3.05 0.002 .0333428 .1529245 Institution~p .4844608 .1433301 3.38 0.001 .2035389 .7653826 TotalOwnedb~r .3474333 .2504515 1.39 0.165 -.1434426 .8383091 Selfmanaged~o 5.801921 6.236816 0.93 0.352 -6.422014 18.02586 Excessreturn Coef. Std. Err. z P>|z| [95% Conf. Interval] corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000 Wald chi2(11) = 107.63 overall = 0.2693 max = 5 between = 0.2230 avg = 4.8 R-sq: within = 0.2783 Obs per group: min = 3 Group variable: SNLInstitu~y Number of groups = 64 Random-effects GLS regression Number of obs = 304 -> CountryName = 9

The impact of a country’s legal REIT structure on performance