The capital structure determinants of REITs around the world

(1)

Amsterdam Business School

The Capital Structure Determinants of REITs

around the World

Supervisor:

Dr. Milena Petrova

Name:

Yunus Yasin Dogan

(2)

1. Introduction

When investing into a new project, managers usually need do decide how to finance the project. That is, managers are confronted with the problem whether to raise funds thorough debt, equity, or other types of securities. The relative proportions of this debt, equity, and other types of securities constitute the capital structure of a corporation. There are various factors, like business risk, a company’s tax exposure, and market conditions etc., which influence the constitution of the capital structure. In general, managers try to choose the capital structure, which maximizes the total value of the securities outstanding.

There is a lot of research about financing decisions of corporations. However, most of the existing literature does exclude Real Estate Investment Trusts (REITs). Yet, due to the increasing popularity of REITs around the world, more research on the financing decisions of REITs is needed.

The trade-off, pecking order and market timing theories try to explain how corporations choose among different financing options when raising external capital.

The trade-off theory posits that there is an optimal amount of debt financing, which is determined by weighing the benefits and cost of debt. The most obvious benefit of debt is that it allows exploiting the interest tax shield. However at the other hand, debt increases the probability and, thus, the cost of financial distress. Thus, according to the trade-off theory, a firm should increase leverage only until the point where the difference between the benefits and cost of debt financing is largest. The pecking order theory, however, claims that asymmetric information is the most important determinant of financing decisions. Thus, informational asymmetries increase the cost of outside financing, especially the cost of equity issuance. Hence, a firm should first use its retained earning, then debt financing, and, as a last resort, equity financing, when the need for funds arises. Lastly, the market timing theory just claims that the financing decisions depend on the market conditions, at a given point in time. One of the main differences of REITs and non-regulated firms is that REITs are exempt from paying corporate taxes. Thus, there does not exist an interest tax shield, which REITs can exploit. In that regard the trade-off theory would predict that REITs should have a very low debt ratio. However, due to the favorable tax treatment, most of the REITs have to distribute large amounts of their taxable earnings, which make REITs more dependent on external financing. Thus, the pecking order theory would predict that REITs should have a high debt ratio, especially, if informational asymmetries are high. In line with the pecking order theory, historical evidence shows that REITs usually issue high amounts of debt.

(4)

Riddiough (2003) find that REITs that issue equity are highly leveraged, despite the fact that there are no obvious tax advantages. They demonstrate that US-based REITs issue equity only as a last resort. Thus, their findings suggest that REITs largely fund investments with bank credits or other sources of public debt. However, Morri and Cristanziani (2009), on the other hand show, using panel data on European REIT and non-REIT firms, that non-REIT firms are significantly more leveraged than REITs. This shows that the tax-exemption of REITs is an important factor in determining the capital structure of European REITs.

To the point, as the results indicate, the regulator environment, that is, mainly the tax-exempt status and the dividend payout requirement, determine, to a large extent, the constitution of a REITs capital structure. Differences in the legal requirements may thus lead, holding everything else constant, to different leverage ratios of REITs across countries.

Besides the unique regulations, there are also various other factors, which may influence the financing decisions of REITs. First of all, the traditional firm-specific factors, like the size or profitability of a firm, are one of the main determinants of the capital structure choice. Thus, for instance, Morri and Beretta (2008) demonstrate that most of the capital structure determinants, like firm size and profitability, have a statistically significant effect on European REITs. Harrison, Panasian and Seiler (2011) also show in their research, that the capital structure of US based REITs is driven by many of the same traditional factors, which influence non-REIT corporate borrowing. In addition to firm-specific factors, various country-specific factors may also have an influence on financing decisions. Demirgüc-Kunt and Maksimovic (1996; 1999) and Booth et al. (2001) illustrate that the institutional and economic environment of a country is an important factor, which may influence the financing decisions of a corporation. The country-specific factors may be especially important in developing countries. Additionally, the institutional and economic environment may also have an impact on the influence of firm-specific factors, which may even cause that some firm-specific factors have influences in opposite directions, in different countries. The findings of De Jong, Kabir and Nguyen (2008), who use date on 42 countries from 1997-2001 to analyze the importance of firm- and country-specific determinants in the leverage choice of non-REIT firms, suggest that the firm-specific determinants of capital structure differ across countries and, in addition, that there is a direct, as well as an indirect impact of the country-specific factors on the leverage choice of firms.

Needless to say, like the country-specific factors, the legal requirements of REITs, which often differ across countries, may also have an impact on the influence of the firm-specific variables.

(5)

Using firm-year observations for the years 2002-2013 of 313 REITs, located in 12 different countries, I will first investigate whether the differences in the legal restrictions across countries have an impact on the leverage ratio of REITs. Here we distinguish among three types of countries. First, countries, which have to payout most of their income and are also required not to exceed a maximum leverage ratio. In the second group we include countries, which also have to payout most of their operating income, but, which can employ unlimited amounts of debt. And lastly, the country, which has no dividend payout requirement, while it still retains its tax-exempt status. The results indicate that there is statistically significant evidence that REITs in countries with high dividend payout requirements but no restrictions concerning their debt financing have the highest leverage ratios. Furthermore, the findings illustrate that the country with no payout requirement has the lowest leverage ratio among the three categories. First of all, the results illustrate that the differences in the legal requirements, indeed, have an impact on the leverage ratio. However, the former finding also implies that REITs prefer to issue debt rather than equity, when the need for external financing arises. The latter result, in addition, also indicates that retained earnings are preferred to external sources. However, the latter result cannot be concluded confidently, since there is only one country, namely, Turkey in this category. Thus, the low leverage ratio may be also a result of the institutional and economic environment of Turkey.

Subsequently, we investigate the influence of the firm-specific variables on the leverage ratio for each country individually. The findings illustrate that the magnitudes and also signs of various firm-specific factors vary across countries and that, thus, no single model for all countries can be used to analyze the influence of the firm-specific variables on the leverage ratio. This would wrongly force the coefficients of the firm-specific factors to be equal across countries. We also discuss the firm-specific coefficients, in the lights of the differences in the legal restrictions across countries. The results are, however, not very indicative of whether the differences in the legal requirements influence the firm-specific variables.

As country-specific variables we used stock and bond market development and GDP growth. The results suggest that country specific variable do not have a major role in determining the capital structure choice of REITs, in the majority of the countries.

Holding everything constant, the specialization of a REIT in a specific property type may also influence the constitution of the capital structure. In line with this argument, Brown and Riddiough (2003), for instance, argue that debt capacity may vary across

(6)

property types, as more stable cash flows will support higher debt levels. Hence, we also investigate whether the specialization of REITs in a specific property type influences the decision of how much debt to employ in the capital structure. The findings indicate that, indeed, the different property types have an impact on the leverage ratio of REITs.

First and foremost, the contribution of this paper is that it analyses some countries, which have, to my knowledge, not been investigated individually before. Second, it illustrates that the differences in the REIT regimes, across the countries, influence, holding all else constant, the capital structure choice of REITs. Third, it indicates that the coefficients of firm-specific variables vary across the countries and that, thus, no single model can be used to describe REIT’s financing decisions. Last but not least, the large sample gives the opportunity to discuss the coefficients of firm-specific factors, in the lights of the different legal requirements across countries.

The remainder of the paper is organized as follows. In section 2, we will present the institutional background of REITs. In the subsequent section, we will discuss the theory of capital structure and capital structure determinants, and the existing evidence about the theory and determinants. In the end of section 3, we will derive the hypotheses. In Section 4, we will present the data and methodology of the empirical analysis. Subsequently, in section 5, we will discuss the empirical results. Section 6 concludes.

(7)

2. REITs: Institutional Background and History

In 1960 a Congress in the US introduced the REIT investment vehicle, through a legislation called the Real Estate Investment Trust Act. A Real Estate Investment Trust (REIT) is a company with the purpose of owning and operating income producing real estate or real estate-related assets. The main features that distinguish REITs from other real estate operating firms is that REITs are exempt from corporate income taxation but, in turn, have to distribute a specified percentage of its taxable income through dividends. For instance in the US, a REIT has to distribute at least 90% of its taxable income. REITs are an ideal way to invest and to achieve the whole benefits of commercial real estate. It provides all investors, that is, also investors for whom direct investment into a real property is beyond their means, a liquid way of investing into diversified portfolios of investment properties. REITs may invest in different kind of commercial property, that is, they may, e.g., invest in residential property, office buildings, shopping malls, hotels etc. While there are REITs, which own and operate several kinds of investment properties, many also specialize only in one type. Furthermore, it is distinguished between three types of REITs: Equity REITs, mortgage REITs, and hybrid REITs. Equity REITs are the most popular type of REITs. The strategy of these REITs are to own and operate income producing real estate. On the other hand, the purpose of mortgage REITs is to provide debt capital to housing and commercial real estate either directly through investing into mortgages or other types of real estate loans, or indirectly through investing into mortgage backed securities. Hybrid REITs are corporations that combine the investment strategies of both equity REITs and mortgage REITs.

Until the 1990s the role of REITs in the real estate sector was very limited. However starting in 1992 equity REITs began to experience an explosive growth in their market capitalization. This was also the beginning of the modern REIT era.

Since until the 1990s REITs played a limited role in real estate investment, even in the US, many countries started to introduce the REIT investment vehicle few years after the beginning of the modern REIT era. Before the 2000s only very few governments legislated the REIT system in their countries. Still in many countries the REIT system does not exist, but its popularity continues to grow.

In the following table we present the main characteristics of the REIT regimes across the countries in the sample. As you can see the legal requirements for REITs may differ across countries. While the minimum dividend payout requirement is 100% for some countries, the requirement is 80% or even only 20% in other countries. For instance, the dividend payout

(8)

were exempt from the minimum dividend pay out requirement until 2008 and are now only required to payout 20% of income, while they could and still can benefit from the corporate income tax exemption.

Table 1

REIT regimes of the countries in the sample

Australia Belgium Canada France Hong Kong Japan

Enacted year 1970 1995 1994 2003 2003 2000

Minimum

share capital No min. requirement €1.25m €15m No min. requirement JPY 100m

Mandatory listing on stock exchange Optional requirement to list on Australian stock exchange Belgian stock

exchange Canadian Stock Exchange On a French or foreign stock exchange before the first day of application of the tax regime Hong Kong

Stock Exchange Optional to list on Japanese Stock Exchange Leverage Restriction 75% of the adjusted Australian asset base Max. 65% of the assets at the time when the loan agreement is concluded

No

restrictions No restrictions Max. 45% of total gross asset value Unlimited. But borrowings must be from qualified financial institutions Min. Dividend payout requirement No requirement. However, undistributed income is taxed. Often 100% is paid out. At least 80%

of net profit No requirement. But 100% of income in order to avoid tax liability 85% of the tax-exempt profit 90% of annual audited net income 90% of accounting income Distribution on capital gain on disposed investments No requirement. However, undistributed gains are taxed.

Tax free if reinvested within 4 years 100% of any capital gains in order to avoid tax liability. 50% of gains arising from the sale of RE or shares in qualifying subsidiaries or shares in RE partnerships No

requirement At least 90% of accounting income

Income tax Trustee not taxable provided that unit holders are entitled to income of the trust at year-end. The qualifying real property income is exempt from taxation Taxable income that is not paid or payable to unit-holders is subject to tax Eligible activities are exempt Income derived from RE outside Hong Kong exempt from tax Subject to tax (42%) but distributions are deductible Capital gains tax

Net capital gain included in taxable income of trust. 50% capital gains tax may be available to individuals and trusts The real property gains are exempt from taxation Must include 50% of any capital gain for purposes of computing taxable income Exempt if capital gains from disposal of qualifying assets Capital gains from RE outside Hong Kong exempt from tax Treated same as ordinary income Source: KPMG; PWC; CMS

(9)

Table 1 (continued)

REIT regimes of the countries in the sample

Netherlands Singapore South Africa Turkey USA UK

Enacted year 1970 2002 2013 1985 1960 2007 Minimum share capital NV €45000 BV €18000 US$20m minimum assets size TRY 30m (For year 2013) £50000 Mandatory listing on stock exchange Optional Singapore stock exchange On an organized stock market in in Istanbul

None Must be listed on a recognized stock exchange Leverage Restriction 60% of fiscal book value of real property and 20% of fiscal book value of all other investments 35% of deposited property Total debt to total asset ratio of 60% Short term credits limited to 3 times the net asset values

No restriction Tax charge on REIT if interest coverage is less than 1.25 Min. Dividend payout requirement 100% of taxable

profit 90% of income from properties in Singapore. No requirement otherwise. 75% of earnings before tax. Was 100% prior to 2013 Prior 2008, no requirement. After 2008, 20% of income 90% of taxable income 90% of the profits of the exempt business Distribution on capital gain on disposed investments

Capital gains are allocated to a tax free reserve and do not form part of the taxable profit

No

requirement No requirement Will be regarded within the distributable profit. No requirement but if not distributed corporate tax is due No requirement to distribute exempt gains.

Income tax Taxed at 0% Exempt if taxable income from properties located in Singapore is distributed within the year

Exempt from corporate income tax Exempt from corporate income tax Deduction is allowed for dividends paid to shareholders. Corporate level tax applies on any taxable income that is not distributed Profit from tax-exempt property business not subject to tax Capital gains tax

Capital gains can be allocated to a tax free capital gains reserve

No capital

gains tax No capital gains tax No capital gains tax Follow the same system as ordinary income Gains from property of tax-exempt business is exempt Source: KPMG; PWC; CMS

Furthermore, REITs in some countries are additionally restricted by a maximum debt-to-asset ratio, while other countries can issue unlimited amounts of debt. Thus, for instance, REITs in Singapore are not allowed to borrow more than 35% of the deposited property value, whereas REITs in France have no restriction concerning leverage. Lastly, while in some countries REITs are completely exempt for income and capital gains tax, in other countries, for instance, only eligible activities or assets are exempt.

In order not to confuse the reader, a note on South Africa is necessary. The internationally recognized REIT system was introduced in South Africa in the year 2013.

(10)

which operated in a similar manner as the internationally known REIT. These were called Property Unit Trust (PUT) and Property Loan Stock (PLS). Both PUTs and PLSs were exempt from corporate income taxation and were required to distribute 100% of their taxable earnings. There were, however, some differences in other legal requirements between the two property companies, which we will, though, not discuss here. Due to some inconsistencies between the two systems, the foreign investors were hesitant to invest in these property vehicles. Consequently, PUTs and PLSs were replaced by the widely known and accepted REIT system.

(11)

3. Theory and Empirical Evidence

3.1. Capital Structure Theories and Findings

There are various theories and many empirical researches, which try to explain the capital structure choice of firms. As we already said, a rational firm should constitute the capital structure in a way, which benefits the shareholders the most. In a very influential paper, the Nobel Prize winning authors Modigliani and Miller (1958) proposed that in a perfect capital market the capital structure does not have any impact on the market value of a firm. So their proposition implies that in an environment where taxes, bankruptcy costs, and asymmetric information does not exist, and where the market is efficient, it is irrelevant for the market value of a firm whether debt, equity or other types of securities are used to raise funds. Where there is, actually, nothing like a perfect capital market in the real world, the proposition of Modigliani and Miller (1958) is still a very important attempt to explain the constitution of capital structure, since the proposition implies that any effect of financing decisions on the market value of a firm is due to capital market imperfections, and thus that the true role of a firm’s financial policy is to optimize its capital structure with regard to these market imperfections.

The trade-off-, pecking order-, and market timing theories further explore capital structure decisions. The trade-off theory postulates that there is an optimal level of debt, which is determined by weighing the benefits of debt against the potential costs associated with leverage. On the one hand debt creates value through the interest tax shield, which is the gain to investors due to the tax deductibility of interest payments. On the other hand, however, debt also increases the financial distress cost of a firm. Thus, a greater amount of liabilities or a higher volatility of a firm’s cash flows and asset values increases the cost of financial distress, since the probability of financial distress increases as the liabilities or the volatility of a firm’s cash flows or asset values increase. With regard to real estate this would imply that firm’s, which are specialized in real estate classes with a higher volatility of cash flows and asset values (like industrial- or office real estate), the financial distress costs should be higher. This is one of the hypotheses we will test later on. The magnitude of the financial distress cost, when there is financial distress, may depend on many factors. For instance, cost incurred during the bankruptcy process or cost due to the loss of, e.g., customers, suppliers and employees largely determine the magnitude of the financial distress cost. Besides that, firms with higher proportions of tangible assets have lower

(12)

Thus, real estate companies should have a relatively low level of financial distress cost since most of their value is derived from tangible assets, i.e., the properties they own.

Agency cost, which is the cost that arises due to the conflict of bondholders and shareholders, may also discourage the use of leverage. Especially when a firm faces financial distress, managers may take actions, which benefit shareholders but harm creditors, and they may also decide not to finance a positive NPV project because the creditor may be the one who benefits from this action. There may be, however, also agency benefits of using leverage. That is, the manager of a firm, e.g., may have greater incentives to reduce wasteful investments when debt is used and, hence, the cost of financial distress is faced.

According to the trade-off theory, a firm should increase debt until the point where the difference between the benefits of using leverage and the costs of using leverage is highest, so that the value of the firm is maximized.

The pecking order model, developed by Myers and Majluf (1984), argues that there exist asymmetric information between the managers of a firm and the investors. The model claims that managers, who have private information, will issue equity only when they know the firm is overvalued. However, rational investors, who know that managers have private information, would discount the price of the existing and the new equity issues, to prevent adverse selection. On the other hand, managers who are aware of the price discount may decide not to issue equity but use their retained earnings, at the first place, when there is need to raise funds. Debt will be issued if there are not enough retained earnings and equity is only issued as a last resort.

The market timing theory posits that the market condition is the most important determinant of the capital structure of a firm. According this theory, firms generally do not care whether to issue debt, equity or other types of securities but try to use the most profitable (or appropriate) source of funding, given the market conditions. Thus, when the market is performing badly, managers avoid issuing equity, whereas at more favorable times equity issues will be large.

There is a wide range of empirical research, which explores the predictions of the capital structure theories, using real world data. Fama and French (2002) studied both the trade-off- and pecking order theory. Their findings indicate that the leverage ratio of firms is mean reverting. Thus, their results suggest that firms have leverage targets and that the leverage tend to move toward this target level. However, the results suggest that the mean reversion is 7-17% per year, which they find suspiciously low. Furthermore, Fama and French (2002) illustrate that there is no positive correlation between profits and debt of a company.

(13)

This finding suggests that firms do not exhaustively exploit the tax advantage of debt. In contrast to Fama and French (2002), the results of Welch (2002) imply that firms do not return to a previously chosen debt ratio. Furthermore, he does also not find any statistical evidence to support the predictions of the pecking order- and market timing theory. Welch (2002) proposes that capital structure is primarily determined by external stock market influences. Shyam-Sunder and Myers (1998) found that the pecking order model performs much better than the trade-off theory, when these two models are tested jointly. Thus, their results indicate that firms follow a pecking order of financing choices rather than determining their capital structure based on an optimal debt ratio. Ozkan (2001), who studies the determinants of capital structure and the adjustment of debt to a long-run target based on a panel data set for 390 UK companies, show that UK companies, indeed, have long term target leverage ratios and that they adjust to their target ratio relatively fast. Hovakimian et al. (2001) found that in the short-run pecking order considerations determine the capital structure of firms, however in the long run their results indicate that firms tend to move towards a target debt ratio, which is consistent with the trade-off theory. Frank and Goyal (2003) test the pecking order theory of corporate leverage based on publicly traded American firms for the period 1971 to 1998. Their findings imply, unlike the pecking order theory, that debt financing does not dominate equity financing. They found evidence that net equity issues closely follow the financing deficit, while net debt does not do so. However, Frank and Goyal (2003), found statistical evidence of a pecking order for firms in the 1970s. Thus, their results suggest that pecking order considerations declined over the years. As possible reasons Frank and Goyal (2003) propose that more small firms were traded publicly during the later years and because equity became more important during time even for the larger firms. Baker and Wurgler (2002) test the relation between market timing behavior and the capital structure choice of firms. According to the market timing theory one would expect that firms raise funds through equity issues when their valuation is high and repurchase equity when the valuation is low. Confirming this behavior, Baker and Wurgler (2002) found that the low leveraged firms tend to be those that raised funds when their market valuation, measured by the market to book ratio, were high, and that the high leveraged firms tend to be those that raised funds when market valuations were low. Alti (2006) investigates how persistence the impact of market timing is on the capital structure choice of firms. By identifying market timers as firms that go public in hot issue markets, Alti (2006) illustrates that hot-market IPO firms issue significantly more equity than cold-market firms, and, thus also have lower leverage ratios. However, their results indicate that

(14)

hot-market firms increase their leverage ratios immediately after going public. Hence, her findings suggest that market timing is a significant determinant of the capital structure choice of a firm in the short-run, but in the long run it has only a limited effect on financing activity.

3.2. Capital Structure and the REIT Regulatory Environment

In the beginning of this section we illustrated the REIT systems across the countries. As already said, there are several distinguishing features of Real Estate Investment Trusts (REITs) from other real estate companies, or, generally, from other corporations. The most notable difference is that REITs are exempt from corporate income tax. Due to the favorable tax treatment, REITs, however, are required to distribute large amounts of their taxable income to investors as dividends. In the majority of the countries where the REIT system is established the dividend payout requirement is at least 90% or more (e.g. Singapore-, US-, UK-based REITs 90%, Dutch REITs 100%). As an entity exempt from taxation, REITs cannot benefit from the tax deductibility of interest payments when using debt. Since there is no interest tax shield, the trade-Off Theory would predict that REITs should have no or very low levels of debt in their capital structure. However, empirical evidence shows that REITs use significant amounts of debt. So, the following question arises: Why is it that REITs still hold large amounts of debt, despite the fact that there are no obvious tax benefits? Pecking order considerations provide a rational explanation for this behavior. As already said, due to asymmetric information, shareholders tend to discount the value of the new equity issues. As Han (2006) argues, it is especially problematic to value REITs, due to the difficulties of appraising real estate properties. Consequently, informational asymmetries are likely to be more severe in the REIT sector, or in the real estate market in general. Thus, managers of REITs may prefer to issue debt rather than equity. While the pecking order explanation for capital structure decisions of REITs is intuitively appealing, the relevance for REIT capital structure is still questionable. The pecking order theory requires that firms have access to the full range of funding options, i.e, retained earnings, debt, and equity. Since many REITs must distribute most of their taxable income to shareholders, they are less able than other firms to retain earnings for investment purposes. This implies that REITs have to seek financing in the capital market more frequently than other firms when the need for liquidity arises. Confirming this, Brown and Riddiough (2003) and Ott, Riddiough, and Yi (2005) show that REITs mostly use debt or equity to finance new investment projects. This fact makes

(15)

the application of the pecking order theory, to describe financing decisions of REITs, problematic.

REIT capital structure may be also determined by market timing behavior. Thus, REITs may issue equity only during times when the market conditions are favorable, and use debt financing when the market is performing badly. Confirming this, the findings of Boudry, Kallberg, and Lui (2009) suggest that REIT capital structure is largely determined by market timing behavior. Feng, Ghosh, and Sirmans (2007) show, however, that REITs, which have a historically high market-to-book ratio, also tend to have a high leverage ratio. Thus, their findings suggest that REITs with high growth opportunities and high market valuation mainly use debt to finance a real estate investment. This is not consistent with the predictions of the market timing theory, as well as the trade-off theory.

The capital structure choice may also vary within the REIT sector, i.e., when we control for the specialization of a REIT in a specific property type. As already mentioned, the trade-off theory would predict that REITs specialized in industrial- or office real estate would use less debt than REITs specialized in residential property, because industrial- and office real estate have more volatile cash flows and asset values than residential properties. Confirming this notion, Harrison, Panasian and Seiler (2011) show that US REITs concentrating their investments in Regional Malls or Manufactured Homes use relatively high amounts of leverage, whereas firms specializing in Self-Storage properties exhibit low leverage ratios. Giambona, Harding and Sirmans (2007) test in their paper the Shleifer-Vishny hypothesis, which states that the liquidity of assets influences both firm leverage and the choice of debt maturity, using a sample of REITs specialized in distinct property types. The findings suggest that REITs, which specialize in the most liquid property type, use more leverage and have longer maturities.

3.3. Capital Structure Determinants

The capital structure of a firm is influenced by many factors. Firm specific characteristics are the most important determinants of financing decisions, however, country specific factors may also have a huge impact on the capital structure choice. In this section we will point out the most important determinants of the capital structure of a firm and review the existing empirical evidence about these determinants.

(16)

3.3.1. Firm-Specific Determinants

In our analysis we will use five firm-specific determinants, namely, asset tangibility, profitability, growth opportunities, firm size, and interest coverage. These will be discussed in turn.

Asset Tangibility

Tangible assets provide collateral for the lender and mitigate the agency cost, which result from the conflict of the bondholder and shareholder. Furthermore, tangible assets can be liquidated relatively easier than intangible assets, which, thus, should also mitigate the financial distress cost of a firm. Hence, an increased use of tangible assets is associated with greater debt capacity for the firm (see Myers 1977;1984; Shyam-Sunders and Myers, 1999; Baker and Wurgler, 2002; Feng et al. 2007 etc.).

Profitability

The trade-off theory predicts that profitability will have a positive impact on leverage. First of all, higher profitability increases the value of the interest tax shield. Thus, firms, which experience an increase in profits, will increase the leverage ratio to fully exploit the interest tax deductibility. Second, more profits will decrease the probability of financial distress and, hence, financial distress cost. A decrease in financial distress cost will increase the optimal leverage ratio. The pecking order theory, however, predicts that an increase in profitability should decrease the leverage ratio. This is because more profitable firms can retain more earnings for new investment projects. Hence, firms will seek less financing from capital markets and, consequently, the leverage ratio should decrease. Consistent with the pecking order theory, existing evidence about non-REIT and REIT firms suggests that there is a negative relationship between profitability and the use of debt (see for example Titman and Wessels, 1988; Rajan and Zingales, 1995; Fama and French, 2002; Morri and Cristanziani, 2009; Harrison, Panasian, and Seiler, 2011).

Growth Opportunities

The trade-off theory predicts that the relationship between growth opportunities and leverage is negative. As we already said, agency cost may cause a debt overhang problem. There is a debt overhang problem when a firm does not decide to undertake a positive NPV project because it will only benefit the debt holder. This agency cost of debt is higher for

(17)

firms with high future growth opportunities. Thus, high growth firms would want to avoid using debt ex-ante, and hence their leverage ratio should be lower. The pecking order theory, however, predicts a positive relationship because future growth opportunities usually require large capital investments, which may exceed the retained earnings. Since equity is used only as a last resort, high-growth firms will issue debt.

Several papers find that an increase in the market to book ratio (i.e. growth opportunities) decrease the leverage ratio (Myers, 1977; Baker and Wurgler, 2002; Jong, Kabir, and Nguyen, 2008), thus, confirming the tradeoff theory. Using a sample of US REITs Harrison, Panasian, and Seiler (2011) also show that higher growth opportunities are associated with lower leverage ratios. However, Feng et al. (2007) and Morri and Beretta (2008) show that US-based REITs with more growth opportunities have higher leverage ratios. Morri and Cristanziani (2009) illustrate, using a sample of European REITs that growth opportunities tend to follow the predictions of the trade-off theory, though the effect is not statistically significant.

Firm Size

There are various possible explanations for why firm size may influence capital structure. One of the possible explanations is that an increase in the firm size would mitigate bankruptcy cost (“too big to fail”), because it is more diversified, has more stable future cash flows etc. Thus, based on the trade-off model, one would predict that debt capacity increase, as a firm gets larger. On the other hand, informational asymmetries are also expected to decrease with the size of a firm, because it is easier to acquire information about larger firms. Consequently, managers of larger firms have greater incentives to issue equity. Hence, the pecking order theory predicts that the leverage ratio should decrease as firm size increases.

The financial literature documents a positive relationship between firm size and leverage (see Titman and Wessels 1988, Rajan and Zingales, 1995; Fama and French, 2002), thus, supporting the trade-off theory. In line with these findings, Maris and Elayan (1990), which examine the factors affecting the capital structure and the cost of capital for REITs, show that leverage is positively related to firm size of both equity and mortgage REITs.

(18)

Interest Coverage

A high interest coverage ratio means that a firm can more easily cover the interests on the debts outstanding. The trade-off theory holds that firms with high coverage ratios have relatively low bankruptcy costs. Therefore, interest coverage should have a positive impact on leverage. Rovolis and Feidakis (2013) find a significant positive relationship of interest coverage and leverage in most of the econometric models they utilize. Harrison et al. (2011) use the lagged coverage ratio in their regressions and find an insignificant negative relationship between (lagged) interest coverage and market leverage.

3.3.2. Country Specific Determinants

We will use three country specific determinants in the regressions, namely, stock market development, bond market development, and GDP growth. The last variable, i.e., GDP growth, will be included only as a control variable and thus the impact will not be discussed. Stock Market Development

A less developed equity market will constrain the financing choices of firms. Thus, firms in countries where the stock market is less developed will have a suboptimal high debt to equity ratio (Demirgüc-Kunt and Maksimovic, 1996). Accordingly, it is argued that if the stock market starts to develop, the debt to equity ratio will necessarily decrease. Furthermore, Demirgüc-Kunt and Maksimovic (1996) argues that the development of the stock market has also an indirect effect on leverage. They assume that the development of the stock market makes it less costly for investors and financial intermediaries to monitor firms, which makes both external equity and debt less risky.

The results of Demirgüc-Kunt and Maksimovic (1996) show that an improvement in the functioning of a developing stock market increases the leverage ratio of firms. However, in countries where the stock market is already developed, a further development would lead to a substitution of equity for debt financing, thus, decreasing the leverage ratio a company. Lastly, they show that the leverage of large firms increase as the developing stock market further improves but that small firms do not seem to be affected by this development.

Bond Market Development

Rajan and Zingales (1995) argue in their paper, studying the determinants of capital structure in the G-7 countries that institutional factors, like development of bond market,

(19)

tax code, etc. may have an impact on the financing decisions of firms. A development in the bond market is expected to have a positive impact on the leverage ratio of firms. This is because in more developed bond markets, firms have more choice for borrowing and are willing to take in more debt (De Jong et al. 2008). Consistent with this notion, De Jong et al. (2008) find a positive and statistically significant relationship between bond market development and the leverage ratio of firms. On the other hand, the findings of Fan, Titman and Twite (2012) illustrate that firms in countries with larger government bond markets have lower debt ratios and shorter maturity debt. They, thus, argue that government bonds tend to crowd out long-term corporate debt.

3.4. Hypotheses

We are going to test the following hypotheses:

H1: REITs located in countries where a regulation concerning the leverage ratio exist, in addition to the requirement of distributing high amounts of their operating income, have lower leverage ratios than REITs located in countries where no leverage restriction exist. H2: REITs in countries with no legal requirement or less strict legal requirement, with regard to the distribution of operating income, have the lowest leverage ratios.

If hypotheses 1 and 2 can be confirmed then we can also say that REITs would prefer internal financing to external financing; and, when external financing is necessary, REITs would prefer debt issuance to equity issuance.

Holding everything else constant, the high dividend payout requirement and, additionally, the leverage restrictions may influence the importance of the profitability variable. Thus, given the different legal requirement concerning the distribution of profits and the maximum amount of leverage to employ, we may observe very different effects of profitability on the leverage ratio across the countries in the sample. We expect in countries where there is a very high dividend payout requirement that profitability is associated with an increase in leverage. This is because firms with very high payout requirements, usually, have not enough retained earnings to finance future investments, so that REITs are dependent on external financing. A high profitability then signals the creditors that the REIT is financially stable, thus, making it easier and less costly for firms to employ debt. We

(20)

should also not here that the dividend payout requirement of REITs refers to the net income, as computed under the generally accepted accounting principles (GAAP). According to these principles depreciation is also considered as an expense. But since depreciation is not a cash outflow, REITs usually have more cash at their disposal, as the accounting income would suggest, and thus more funds can be plowed back for new investments. Though we think that in most cases the retained earnings may still not be enough, it still could have an impact. Thus, we may observe a positive coefficient only in countries with the highest payout requirement. Hence:

H3: Holding everything else constant, we expect in countries where REITs payout 100% of their taxable income that an increase in profitability will increase debt financing.

Furthermore, we think that in countries where the leverage ratio is restricted, REITs are more dependent on internal financing or equity issuance. Thus:

H4: In all other countries (than in H6) profitability has a negative impact on leverage. In countries with leverage restrictions, however, profitability is likely to be more important. H5: There is a negative relationship between firm size and leverage in countries with high dividend payout requirements. In countries with leverage restrictions, however, the coefficient of firm size is likely to be greater (in absolute values).

We define no hypotheses for the other firm-specific variables. However, we will discuss their impact in the lights of the existing theories and empirical evidences.

Economic intuition and results of empirical studies imply a mixed impact with regard to the above-discussed country-specific variables. However, we propose the following hypotheses about stock and bond market development for REITs in every country.

H6: Stock market development has a negative impact on leverage H7: Bond market development has a positive impact on leverage

We propose these hypotheses because the first thoughts of people about the impact of the stock and bond market development would be that the former has a negative and that the latter has a positive impact on the leverage ratio.

(21)

Lastly, I specify the following hypothesis to test the impact of the different property types on leverage:

H8: Property types with more stable cash flows (i.e. lower overall risk) have higher leverage ratios.

This hypothesis follows from Brown and Riddiough (2003) and, basically, from the trade-off theory, which state, as already mentioned, that debt capacity vary across property types because property types with more stable cash flows will support a higher level of debt.

(22)

4. Data and Methodology

4.1. Data

The data necessary to conduct the analysis will be collected from the SNL database and the World Development Indicators and Financial Structure Database of the World Bank. The SNL database contains the information needed to construct the firm-specific variables, whereas the data from the World Bank databases is necessary to construct the country-specific determinants. In the analysis I will use data from REITs based in 12 different countries, namely, Australia, Belgium, Canada, France, Hong Kong, Japan, The Netherlands, Singapore, South Africa, Turkey, USA, and UK. The sample will cover the period 2002-2013.

In order to define the firm-specific variables mentioned in the previous section, the following financial information was collected: Total assets, total liabilities, market capitalization, EBITDA, interest expense, net real estate investments, net property plant and equipment, tangible assets, current assets. Firm-year observations were dropped whenever there were missing data on total assets or market capitalization. We further restrict the sample to firms, which have data for at least three years over the study period. The final sample consists of 313 firms and a total of 2971 firm-year observations. The following table shows how much firms and firm-year observations are available per country.

Table 3

Firms and observations per country

Country Firms Observations

Australia 21 192 Belgium 10 106 Canada 24 232 France 16 149 Hong Kong 7 47 Japan 30 230 Netherlands 5 60 Singapore 23 173 South Africa 9 80 Turkey 6 43 USA 138 1418 UK 24 241 Total 313 2971

The (market) leverage ratio is defined as following: total book debt over the sum of book debt and market capitalization, whereas book debt is defined as total assets minus book equity. We will use this definition of leverage, since Welch (2011) argues that the

(23)

debt-to-asset ratio is a flawed measure of leverage. Asset tangibility is defined as the ratio of net real estate investments over total assets. Whenever firm-year observations are missing for net real estate investments, it is replaced by net property, plant and equipment. Any remaining missing observation is then replaced by fixed assets, which is defined as tangible assets minus current assets.

The ratio of EBITDA over total assets measures the ‘profitability’ of a firm. Growth opportunities is defined as the market to book ratio, which the sum of book debt and market capitalization over total assets. Firm size is defined as the natural logarithm of total assets, and lastly, interest coverage is measured by the ratio of EBITDA over interest expense of debt. The country-specific variables, which will be used in the regressions, are defined as follows: stock market development is defined as the ratio of stock market capitalization to GDP, whereas bond market development is defined as the ratio of private and public bond market capitalization to GDP.

The following table shows the mean values of the firm specific variables for each country in the sample.

Table 4

Descriptive statistics of the firm-specific variables

Country LEV TANG PROFIT GROWTH SIZE COVERAGE

Australia 0.401 0.907 0.047 0.908 14.831 5.73 Belgium 0.411 0.943 0.054 0.938 13.355 26.229 Canada 0.555 0.932 0.074 1.216 13.928 2.837 France 0.550 0.884 0.060 0.933 14.788 26.394 Hong Kong 0.394 0.958 0.094 0.716 16.777 26.313 Japan 0.468 0.914 0.044 0.969 12.212 11.113 Netherlands 0.454 0.959 0.058 0.891 14.679 3.875 Singapore 0.356 0.933 0.074 0.892 14.474 12.391 South Africa 0.298 0.935 0.133 1.064 16.075 8.104 Turkey 0.152 0.634 0.069 0.681 13.286 10.618 USA 0.440 0.881 0.077 1.243 14.284 5.008 UK All Countries 0.451 0.441 0.921 0.898 0.041 0.069 0.877 1.093 13.709 14.169 2.836 7.546

Closely followed by France, Canada is the country with the highest leverage ratio. Their average market leverage ratio is 55% and 55.5%, respectively. The country with the lowest leverage ratio is Turkey, which amounts to 15.2%. This is result is not very surprising, since REITs in Turkey have no minimum dividend payout requirement, which implies that they can retain their earnings for investment purposes. The average leverage ratio of REITs in South

(24)

Africa is 29.8%, which is the second lowest leverage ratio in the sample. However, REITs in South Africa were, unlike Turkish REITs, required to distribute 100% (prior to 2013) of their distributable earnings. Considering the fact that Turkey and South Africa are emerging markets, this observation implies, holding everything else constant, that REITs in emerging markets have lower leverage ratios. However, there could be, of course, also various other reasons why Turkish and South African REITs have a lower leverage ratios than the REITs based in other countries. Whereas the firm-specific and country specific factors are likely to be the most important reasons for the difference in the leverage ratio, it could be also caused by, e.g., cultural attitudes or other differences in the REIT regimes. As mentioned before, many countries are required not to exceed a specified loan-to-value ratio. Thus it is not surprising that we observe a low leverage ratio for, e.g., Singapore. For almost all countries in the sample, tangibility is close or above 90%, which is a very expected result for REITs. However, the tangibility of Turkish REITs is surprisingly low, i.e., 63.4%. Profitability is lowest for REITs in the countries Australia, Japan, and UK. The highest (average) profitability is observed in South Africa, which could be due to the overall higher risk, and thus, higher capitalization rates in this country. Growth opportunity values (or market-to-book ratios) above one are observed in the countries Canada, South Africa, and USA. The country for which growth opportunity is lowest is Turkey (0.68). The interest coverage ratio varies a lot across the countries. The highest values of the interest coverage ratio is observed in Belgium, France, and Hong Kong, with values of 26.2, 26.4, and 26.3, respectively.

The following table shows the descriptive statistics of the two variables, market leverage- and interest coverage ratio, for the different property types.

Table 5

Descriptive Statistics of leverage and interest coverage

Leverage Interest Coverage

Property Type Obs. Mean Std. Min Max Mean Std. Min Max Diversified 854 0.439 0.211 0 0.991 10.703 120.56 -2423 1570.5 Health Care 169 0.351 0.188 0 0.947 9.831 36.31 -117 376.8 Hotel 195 0.517 0.213 0 0.966 3.447 9.064 -72.2 73.68 Industry 182 0.445 0.175 0 0.952 6.52 15.57 -10.2 184.12 Manufa. Home 36 0.520 0.142 0.256 0.826 2.59 0.978 1.08 5.275 Multi-Family 251 0.525 0.160 0.016 0.994 -4.787 83.244 -1122 36.36 Office 463 0.459 0.164 0 0.998 5.36 6.66 -19.8 51.87 Other Retail 172 0.389 0.200 0 0.889 13.19 50.55 -23.8 369.8 Regional Mall 127 0.491 0.183 0.158 0.986 6.988 11.36 -9.44 79.03 Self-Storage 63 0.354 0.198 0.017 0.792 34.58 121.36 -25.5 791.14 Shopping Centre 323 0.429 0.164 0 0.973 5.743 17.73 -10.3 259.43 Specialty 136 0.297 0.159 0 0.901 9.655 46.27 -115 423.61

(25)

As you can see the average leverage ratio is lowest for the property type specialty (29.7%), followed by health care (35.1%) and self-storage properties (35.4%). This observation was expected since these properties are the most risky asset types, in terms of cash flow and asset value volatility, and hence should have lower leverage ratios. The property types with the highest leverage ratios are multi-family and manufactured homes, with ratios of 52.5% and 52%, respectively. This is also not surprising since these property types have relatively low risk.

4.2. Methodology

We will first analyze whether the legal restrictions are a critical factor in determining the leverage ratio of REITs. In that regard, we will pool all the available data and use dummy variables to control for different legal restrictions. Thus, we will first estimate the following model:

𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖= 𝛽𝛽0+ 𝛽𝛽1𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽2𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽3𝐺𝐺𝑃𝑃𝑃𝑃𝐺𝐺𝑇𝑇ℎ 𝑂𝑂𝑂𝑂𝑂𝑂𝑃𝑃𝑃𝑃𝑇𝑇𝑂𝑂𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑂𝑂𝑂𝑂𝑖𝑖 + 𝛽𝛽4 𝐹𝐹𝑇𝑇𝑃𝑃𝐹𝐹 𝑆𝑆𝑇𝑇𝑆𝑆𝑂𝑂𝑖𝑖+ 𝛽𝛽5𝐼𝐼𝑇𝑇𝑇𝑇𝑂𝑂𝑃𝑃𝑂𝑂𝑂𝑂𝑇𝑇 𝐶𝐶𝑃𝑃𝐶𝐶𝑂𝑂𝑃𝑃𝑇𝑇𝑇𝑇𝑂𝑂𝑖𝑖

+ � 𝐿𝐿𝑂𝑂𝑇𝑇𝑇𝑇𝑇𝑇 𝑅𝑅𝑂𝑂𝑂𝑂𝑇𝑇𝑃𝑃𝑇𝑇𝑅𝑅𝑇𝑇𝑇𝑇𝑃𝑃𝑇𝑇 𝐷𝐷𝑂𝑂𝐹𝐹𝐹𝐹𝑇𝑇𝑂𝑂𝑂𝑂 + 𝑂𝑂𝑖𝑖

where i denotes the individual REIT. The dependent variable is market leverage, which is defined as the ratio of total book debt to the sum of book debt and the market value of equity. The last term, ui, is the residual. There will be three dummy variables, which will

represent the following categories: One dummy variable will represent all countries with high dividend payout requirements and leverage restrictions. One dummy variable will be created for countries with high leverage restrictions but no maximum leverage requirement. And lastly, one categorical variable will be created for the countries with no or very low requirement concerning income distribution. The emphasis in this analysis will be only on the coefficients of the dummy variables.

Next, in order to analyze the impact of the firm-specific variables on the capital structure choice of REITs, I will estimate the following firm-level ordinary-least squares regression for each of the 12 countries in the sample:

(26)

where i denotes the individual REIT and j denotes the country. The dependent variable is leverage, which is defined as the ratio of total book debt to the sum of book debt and the market value of equity. The last term, ui, is the residual. This is basically the same model as

above, except that there are no dummy variables. Following the existing literature, we will analyze the contemporaneous relationship between market leverage and the firm-specific variables (see for example De Jong et al., 2008; Morri and Cristanziani, 2009; Harrison et al., 2011; Rovolis and Feidakis, 2013).

Subsequently, following de Jong et al. (2008), we will investigate whether the firm-specific coefficients are equal across the countries. Thus, using an unrestricted model, where the coefficients are allowed to vary across countries, and five restricted models, where in each model one firm-specific variable is restricted to be equal across countries (which will be determined by the results of the regression using all countries), we will test whether the firm-specific coefficients have the same values for all countries in the data set. More specifically, the following hypotheses will be tested

𝐻𝐻0: 𝛽𝛽𝑘𝑘,𝑖𝑖= 𝛽𝛽𝑘𝑘 𝑃𝑃𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇 𝑗𝑗 = 1, 2, 3, … ,12

𝐻𝐻1: 𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑇𝑇𝑂𝑂 𝑃𝑃𝑃𝑃 𝐹𝐹𝑃𝑃𝑃𝑃𝑂𝑂 𝑗𝑗 𝑇𝑇ℎ𝑂𝑂 𝑃𝑃𝑂𝑂𝑂𝑂𝑇𝑇𝑇𝑇𝑅𝑅𝑇𝑇𝑇𝑇𝑃𝑃𝑇𝑇 𝑂𝑂𝑇𝑇𝑢𝑢𝑂𝑂𝑃𝑃 𝐻𝐻0 𝑢𝑢𝑃𝑃𝑂𝑂𝑂𝑂 𝑇𝑇𝑃𝑃𝑇𝑇 ℎ𝑃𝑃𝑇𝑇𝑢𝑢

where k denotes the coefficient of a firm specific variable and j denotes the country, whereas 𝛽𝛽𝑘𝑘 is the estimated coefficient when all the countries are used.

Furthermore, using a single restricted model, i.e. where all firm-specific variables are restricted to be equal across countries, I will test whether all firm-specific coefficients have the same value for the countries in the sample. According to Jong et al. (2008), the latter test is more important since it illustrates whether a single model for firms in all countries can be used. The following joint hypothesis will be tested

𝐻𝐻0: 𝛽𝛽𝑘𝑘,𝑖𝑖= 𝛽𝛽𝑘𝑘 𝑃𝑃𝑃𝑃𝑃𝑃 𝑇𝑇𝑇𝑇𝑇𝑇 𝑗𝑗 = 1, 2, 3 … ,12 𝑇𝑇𝑇𝑇𝑢𝑢 𝑇𝑇𝑇𝑇𝑇𝑇 𝑘𝑘 = 1, 2, 3, 4, 5 𝐻𝐻1: 𝑃𝑃𝑃𝑃𝑃𝑃 𝑃𝑃𝑇𝑇𝑂𝑂 𝑃𝑃𝑃𝑃 𝐹𝐹𝑃𝑃𝑃𝑃𝑂𝑂 𝑗𝑗, 𝑃𝑃𝑇𝑇𝑂𝑂 𝑃𝑃𝑃𝑃 𝐹𝐹𝑃𝑃𝑃𝑃𝑂𝑂 𝑃𝑃𝑃𝑃 𝑇𝑇ℎ𝑂𝑂 𝑃𝑃𝑂𝑂𝑂𝑂𝑇𝑇𝑃𝑃𝑇𝑇𝑅𝑅𝑇𝑇𝑇𝑇𝑃𝑃𝑇𝑇𝑂𝑂 𝑂𝑂𝑇𝑇𝑢𝑢𝑂𝑂𝑃𝑃 𝐻𝐻0 𝑢𝑢𝑃𝑃𝑂𝑂𝑂𝑂 𝑇𝑇𝑃𝑃𝑇𝑇 ℎ𝑃𝑃𝑇𝑇𝑢𝑢 The statistical significance whether the coefficients are equal or not across the countries will be determined using the joint test of significance of regression coefficients, that is, the F-test. The F-test is defined as follows,

(27)

𝑃𝑃 = (𝑅𝑅𝑆𝑆𝑆𝑆_{𝑅𝑅𝑆𝑆𝑆𝑆}𝑅𝑅− 𝑅𝑅𝑆𝑆𝑆𝑆𝑈𝑈𝑅𝑅)/𝑞𝑞 𝑈𝑈𝑅𝑅/(𝑁𝑁 − 𝐾𝐾)

where RSSR and RSSUR are the residual sum of squares for the restricted model and for the

unrestricted model, respectively, q is the number of restrictions, N is the number of observations, and K is the number of regressors remaining in the restricted model. The RSSR

is found by adding all of the sum-of-squared residuals in each restricted model for a country together. Similarly, the RSSUR that will be used in the F-test is the total of the sum of squared

residuals of each unrestricted model of the countries.

In a subsequent regression, I will include the various country-specific variables into the model, in order to investigate the impact of the institutional and economic environment on the capital structure choice of REITs.

𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖 = 𝛽𝛽0𝑖𝑖+ 𝛽𝛽1𝑖𝑖𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽2𝑖𝑖𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽3𝑖𝑖𝐺𝐺𝑃𝑃𝑃𝑃𝐺𝐺𝑇𝑇ℎ 𝑂𝑂𝑂𝑂𝑂𝑂𝑃𝑃𝑃𝑃𝑇𝑇𝑂𝑂𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑂𝑂𝑂𝑂𝑖𝑖 + 𝛽𝛽4𝑖𝑖 𝐹𝐹𝑇𝑇𝑃𝑃𝐹𝐹 𝑆𝑆𝑇𝑇𝑆𝑆𝑂𝑂𝑖𝑖+ 𝛽𝛽5𝑖𝑖𝐼𝐼𝑇𝑇𝑇𝑇𝑂𝑂𝑃𝑃𝑂𝑂𝑂𝑂𝑇𝑇 𝐶𝐶𝑃𝑃𝐶𝐶𝑂𝑂𝑃𝑃𝑇𝑇𝑇𝑇𝑂𝑂𝑖𝑖+ 𝛽𝛽6𝑖𝑖 𝐺𝐺𝐷𝐷𝑃𝑃 𝐺𝐺𝑃𝑃𝑃𝑃𝐺𝐺𝑇𝑇ℎ𝑖𝑖

+ 𝛽𝛽7𝑖𝑖 𝑆𝑆𝑇𝑇𝑃𝑃𝑅𝑅𝑘𝑘 𝐹𝐹𝑇𝑇𝑃𝑃𝑂𝑂𝑘𝑘𝑇𝑇 𝐷𝐷𝑂𝑂𝐶𝐶𝑂𝑂𝑇𝑇𝑃𝑃𝑂𝑂𝐹𝐹𝑂𝑂𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽8𝑖𝑖 𝐵𝐵𝑃𝑃𝑇𝑇𝑢𝑢 𝑀𝑀𝑇𝑇𝑃𝑃𝑘𝑘𝑂𝑂𝑇𝑇 𝐷𝐷𝑂𝑂𝐶𝐶𝑂𝑂𝑇𝑇𝑃𝑃𝑂𝑂𝑂𝑂𝐹𝐹𝑂𝑂𝑇𝑇𝑇𝑇𝑖𝑖 + 𝑂𝑂𝑖𝑖 As above, this analysis will be conducted for each country individually.

Finally, in order to see how the different property types influence the market leverage ratio of REITs, I will pool all the available data and use dummy variables to control for the property types, illustrated in the ‘Data’ section.

𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽2𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖+ 𝛽𝛽3𝐺𝐺𝑃𝑃𝑃𝑃𝐺𝐺𝑇𝑇ℎ 𝑂𝑂𝑂𝑂𝑂𝑂𝑃𝑃𝑃𝑃𝑇𝑇𝑂𝑂𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑂𝑂𝑂𝑂𝑖𝑖

+ 𝛽𝛽4 𝐹𝐹𝑇𝑇𝑃𝑃𝐹𝐹 𝑆𝑆𝑇𝑇𝑆𝑆𝑂𝑂𝑖𝑖+ 𝛽𝛽5𝐼𝐼𝑇𝑇𝑇𝑇𝑂𝑂𝑃𝑃𝑂𝑂𝑂𝑂𝑇𝑇 𝐶𝐶𝑃𝑃𝐶𝐶𝑂𝑂𝑃𝑃𝑇𝑇𝑇𝑇𝑂𝑂𝑖𝑖+ � 𝑃𝑃𝑃𝑃𝑃𝑃𝑂𝑂𝑂𝑂𝑃𝑃𝑇𝑇𝑇𝑇 𝑇𝑇𝑇𝑇𝑂𝑂𝑂𝑂 𝑢𝑢𝑂𝑂𝐹𝐹𝐹𝐹𝑇𝑇𝑂𝑂𝑂𝑂 + 𝑂𝑂𝑖𝑖

(28)

5. Empirical Results

5.1. Legal Restrictions and Firm-Specific Variables

We will start the analysis with a regression where we pool all of the countries in the sample, in order to analyze how the different dividend payout and leverage regulations influence the leverage ratio. Since the emphasis in this analysis is not on the firm-specific factors, we can neglect for now the coefficients of the firm-specific variables and force them to be equal across countries, though, they may be very different. To analyze how the restrictions influence the market leverage we categorize the countries into three groups. One group will consist of the countries, which do not have a minimum dividend payout requirement. In our sample this is, only, Turkey. The other group is constituted of the countries, which have a minimum dividend payout requirement of 85% or more, but which are allowed to issue as much debt as they want. This group is formed of the countries: Canada, France, Japan, US, and UK.1_{The last group consists of Australia, Belgium, Hong Kong, The Netherlands,}

Singapore, and South Africa.2_{This is the group of countries, which are restricted by a}

minimum dividend payout requirement, as well as a maximum debt-to-asset ratio. Below you see the results of the regressions. A coefficient is considered to be statistically significant when the p-value is equal or lower than 10%. Robust standard errors were used in all of the regressions.

The two models are, actually, the same, with the only exception that the omitted category in the first regression is the countries with no dividend payout requirement and in the second regression the countries, which are restricted by a dividend payout requirement but have no leverage restriction. Due to the reason that we just use a different combination of the categorical variables, the firm-specific coefficients are the same in both regressions. As the first model illustrates, the group which is restricted by a dividend payout requirement but which has no leverage restriction, has a significantly higher leverage ratio than the group with no payout requirement. The estimate is also statistically different from zero at the 1% level. The other group, i.e., the countries, which have a payout requirement and a leverage restriction, are also predicted to have slightly higher leverage ratio, which is statistically significant at the 5% level. Furthermore, the second regression illustrates that a leverage

1_{Though there is a restriction in UK, it is not as strict as in the other countries. Thus, we put UK in this}

group.

2_{There is no payout requirement in Australia and Canada. But in both countries undistributed dividends} are taxed. So, usually 100% is paid out.

(29)

restriction is a significant determinant of the market leverage ratio. Thus, the group, which has to payout dividends and is also restricted by a maximum amount of debt financing, has a lower market leverage ratio than the countries, which can freely determine their debt strategy. Hence, the results confirm hypotheses 1 and 2, and, thus, show that the legal requirements influence the capital structure strategy of REITs. The results, consequently, also imply that REITs prefer debt issuance to equity issuance since REITs in the countries with no leverage restriction have the highest leverage ratio. Furthermore, the results also imply that internal financing is preferred to external financing since the country with no payout requirement has the lowest leverage ratio. However, since we have only one country in this category, which is Turkey, we cannot confidently say that the low leverage ratio is due to the no payout requirement, because, especially in the case of Turkey, many other factors may have caused such a low leverage ratio. Though, we do not think that this will cause such a huge difference, and still conclude that internal financing is preferred over external financing.

Table 6

The impact of legal requirements on the leverage ratio

Variable Model 1 Model 2

Intercept 0.3642*** (0.0605) 0.6273*** (0.0448) Tangibility 0.1971*** (0.0343) 0.1971*** (0.0343) Profitability -0.3798*** (0.0428) -0.3798*** (0.0428) Growth Opportunities -0.2502*** (0.0101) -0.2502*** (0.0101) Firm Size -0.0017 (0.0021) -0.0017 (0.0021) Interest Coverage -0.00016 (0.0002) -0.00016 (0.0002) Payout req. and lev. restr. 0.1187**

(0.0145) -0.1443*** (0.0078) Payout req. but no lev. restr. 0.2630***

(0.0137) ___

No payout requirement ___ -0.2630***

(0.0137)

R-square 0.3404 0.3404

Standard errors are given in parentheses under the coefficients. Individual coefficients are statistically significant *10% level, **5% level and ***1% level.

Finally we should note that the legal requirements could also, in addition to many other factors, determine the magnitude and direction of the impact of the firm-specific

(30)

variables. In that regard, a comparison of countries, with differences in their legal requirements, would be interesting.

In the following analysis we will breakdown the data into the various countries. Table 7 shows the impact of the firm-specific variables on the leverage ratio of REITs based in the various countries. The last row shows the results of the regression when data from all countries are used.

Table 7

The impact of firm-specific variables across countries

Country Intercept TANG PROFIT GROWTH SIZE COVERAGE R2

Australia 1.387*** (0.196) -0.200* (0.120) -0.330* (0.193) -0.39*** (0.072) -0.03*** (0.009) -0.0016** (0.0007) 0.385 Belgium 0.821*** (0.303) -0.571* (0.289) -1.3*** (0.483) -0.200 (0.149) 0.03*** (0.009) -0.001*** (0.0002) 0.573 Canada 0.934*** (0.117) 0.35*** (0.107) 1.35*** (0.409) -0.37*** (0.034) -0.02*** (0.0063) -0.0496*** (0.0112) 0.603 France 1.664*** (0.197) -0.26** (0.114) -0.417 (0.262) -0.55*** (0.063) -0.024** (0.011) -0.0004*** (0.0001) 0.681 Hong Kong 0.299 (0.538) 1.058** (0.502) -0.167 (0.129) -0.205** (0.0932) -0.04*** (0.0121) -0.0024*** (0.0003) 0.756 Japan 1.386*** (0.207) -0.239 (0.189) 0.391 (0.584) -0.43*** (0.0282) -0.022* (0.0116) -0.0033*** (0.0012) 0.679 Netherlands 1.552*** (0.358) -0.142 (0.191) 0.929 (0.747) -0.39*** (0.121) -0.039** (0.017) -0.0227* (0.0118) 0.470 Singapore 0.467*** (0.147) 0.033 (0.039) -0.200 (0.197) -0.31*** (0.045) 0.012 (0.008) -0.0025*** (0.0004) 0.378 South Africa -0.0235 (0.345) -0.204 (0.199) 0.222 (0.229) 0.267*** (0.072) 0.0154 (0.0193) -0.006 (0.001) 0.559 Turkey -1.65*** (0.305) 0.48*** (0.117) 0.744 (0.508) -0.30*** (0.077) 0.127*** (0.021) -0.0002*** (0.00004) 0.805 USA 0.619*** (0.064) 0.24*** (0.05) -0.4*** (0.049) -0.24*** (0.013) -0.0038 (0.0034) 0.00003 (0.00023) 0.339 UK 1.078*** (0.155) 0.21** (0.102) -0.3*** (0.102) -0.66*** (0.063) -0.014** (0.007) -0.0017 (0.0012) 0.424 All countries 0.656*** (0.047) 0.19*** (0.037) -0.4*** (0.046) -0.19*** (0.009) -0.01*** (0.0023) -0.0002 (0.00021) 0.240

Standard errors are given in parentheses under the coefficients. Individual coefficients are statistically significant *10% level, **5% level and ***1% level.

As you can see, for eight out of 12 countries the coefficient on tangibility is statistically significant. As discussed before, the effect of tangibility on leverage is expected to be positive. Confirming this, the coefficient estimates of tangibility for REITs based in Canada, Hong Kong, Singapore, Turkey, USA and UK is positive. For REITs based in Singapore, this effect is, however, not significantly different from zero. In the remaining six countries the

The capital structure determinants of REITs around the world

Amsterdam Business School