Market timing and REIT capital structure changes

4.2 Test the Hypotheses of the Information-based Market Timing…35 4.3 Market Efficiency and Market Timing in REIT Equity Offerings…46 4.4 Market Timing in REIT Debt Issuances………51 5 Se

MARKET TIMING AND REIT CAPITAL STRUCTURE CHANGES

LI YING

NATIONAL UNIVERSITY OF

SINGAPORE

2006

MARKET TIMING AND REIT CAPITAL STRUCTURE CHANGES

DEPARTMENT OF REAL ESTATE

NATIONAL UNIVERSITY OF SINGAPORE

2006

Acknowledgement

First, I would like to say thanks to my advisor, Associate Professor Ong Seow Eng, because this study is carried out under his careful supervision Professor Ong has provided me continuous support during my master studies He is always there to help me with my study, and has given me a lot of insightful suggestions, in both asking the research question and answering the question This study could not be finished without Professor Ong’s help Besides, his rigorous attitude toward academic research sets a good example for me and many other people to follow

Special thanks should be given to another advisor of mine, Dr Muhammad Faishal Bin Ibrahim He is both a good teacher and a good friend He has confidence in me even when I doubt myself

I also want to say thanks to my family and friends, who always give me courage and support in both my life and my study

Table of Contents

Summary 1

List of Tables 3

List of Figures 4

1 I n t r o d u c t i o n 5

1.1 Capital Structure Theories………5

1.2 About Real Estate Investment Trust……….7

1.3 Research Objective………9

1.4 Organization of the Study………10

2 Literature Review 11

2.1 Trade-off Theory and Pecking Order Theory………11

2.2 Market Timing Theory………13

3 Research Methods and Data Sources 23

3.1 VAR Model………23

3.2 Probit Model………25

3.3 Data Sources………26

4 Market Timing of REITs 29

4.2 Test the Hypotheses of the Information-based Market Timing…35 4.3 Market Efficiency and Market Timing in REIT Equity Offerings…46

4.4 Market Timing in REIT Debt Issuances………51

5 Security Choice of REITs Offerings 56

6 Conclusion 65

Bibliography 70

Appendices 76

Summary

This paper seeks to determine if REIT equity and debt offerings depend on capital market conditions Specifically, we utilize a vector autoregressive (VAR) model to analyze the relationship between capital issues of REITs and relevant capital market conditions to see whether market timing behavior is supported In addition, we seek to understand the fundamental mechanism behind the market timing behavior where we test two different hypotheses of market timing- the information-based market timing and market inefficiency-based market timing Last but not least, we examine the security choice of REITs to test between different capital structure theories- trade-off theory, pecking-order theory and market timing theory

The empirical results from 1980 through 2004 support market timing in REIT equity and debt issues In addition, support is also found for the information-based market timing theory for equity offerings and backward looking market timing in debt issuances Our results have two implications First, although REITs are relatively more transparent than industrial firms, REITs managers still could time the equity market Asymmetric information and adverse selection still exist for REITs, but with less obvious timing patterns Second, the absence of long-run return anomaly after SEOs of REITs means that the market efficiency is not violated This is different from studies about general stocks, in which the market inefficiency-based market timing is preferred compared to the information-based market timing With regard to the debt

issuances, two hypotheses are tested- backward-looking and forward-looking, and backward-looking debt market timing is supported

In addition, we find that, although market conditions are important in the security choice of REITs, they are not the only factors that determine the capital structure changes of REITs Firm characteristics, such as size, leverage, debt capacity, etc, all have significant effects on the financing activities of REITs The trade-off theory and the pecking-order theory, also find partial support from our analysis We conclude that there is no single theory that fully explains the capital structure of REITs

List of Tables

Table 1: Equity and Debt Offerings by Year……….……… 27

Table 2: Statistic Summary……… …………33

Table 3: Correlation Matrix………33

Table 4: VAR Result………34

Table 5: Event Day Return on Announcement of Equity Offerings………36

Table 6: Event Day Return on Announcement of Debt Issues………… 37

Table 7: Distribution of Time Lags… ………44

Table 8: Event Day Return Regression………45

Table 9: Long-run Performance Calculation………49

Table 10: Effect of Interest Rate and Interest Rank on REIT Debt Offering… 54

Table 11: Interest Rates Subsequent to Debt Issues………55

Table 12: Probit Regression……… 60

Table 13: Summary of the Predictions of Different Capital Structure Theories… 64

Table 14: Logit Regression ……….………76

Table 15: Equity Market Capitalization Outstanding of REITs………77

List of Figures

Figure 1: Cumulative Normal Distribution ……….……….………25

Figure 2: Distribution of Time Lags……… 40

Figure 3: Distribution of Time Lags in Different Decades………… ………41

Figure 4: NAREIT Equity REITs Price Index……….…… 78

Figure 5: S & P 500 Price Index……….………79

Figure 6: 10-Year Government Bond Yield……….………79

Chapter 1

Introduction

Recent evidence in capital structure research has supported the market timing explanation for capital structure where capital financing decisions depend on capital

market conditions (Graham and Harvey, 2001; Baker and Wurgler, 2002) Real

Estate Investment Trusts (REITs) are typically excluded in studies for general stocks

as they are special investment vehicles This study links market timing with REITs to understand the capital financing and capital structure changes of REITs An introduction to different capital structure theories, including the market timing theory, and a description of REITs is given in this chapter Then the research objective and

the organization of this study are introduced

1 1 Capital Structure Theories

The development of capital structure theories starts with the capital structure irrelevance theory, which was developed by Modigliani and Miller in 1958 This theory is based on a number of assumptions, which include: (1) capital markets are frictionless, (2) individuals can borrow and lend at the risk free rate, (3) there are no bankruptcy costs, (4) firms only issue risk-free debt and risky equity, (5) all firms are

in the same risk class, (6) there are no personal taxes, (7) all cash flows are perpetuities, (8) information is symmetric, and (9) managers always maximize shareholders’ wealth Based on above assumptions, the MM model concludes that the capital structure of the firm is irrelevant and does not influence the market value

of the firm

However, in the real world, taxes and bankrupt costs exist and firms are leveraged This means that some assumptions of the MM propositions should be relaxed if the real world situations are considered With corrections to the capital structure theory, capital structure is no longer irrelevant to the firm value There is a trade-off between the tax benefits of debt and the bankruptcy risk, and there is an optimal capital structure for firms, although this optimal capital structure may be unobservable This is the trade-off theory Trade-off theory is still a very important capital structure theory in corporate finance

Instead of the trade-off between tax benefits of debt and the bankruptcy risk, the information asymmetry between firm managers and investors is the primary determinant of capital structure changes according to the pecking-order theory Managers are assumed to act in interest of “old” shareholders, and choose not to issue equity when the prospects of the firms are good Investors, who have less information about firms than managers, will consider the financing decisions made

by managers as signals Equity issues made by managers are bad signals that indicate bad news about the firms So managers avoid issuing equity if they have internal capital or they could issue debt The pecking order for managers is first internal capital, then debt, and the last resort is equity

The market timing theory is a dynamic capital structure theory and implies that managers make financing decisions according to the capital market conditions Generally, managers will issue equity when their stocks are overvalued and issue debt when interest rates are low The market timing concept is also supported by the long-run underperformance of stocks after initial public offerings or secondary equity offerings compared to non issuing stocks

The market timing theory is different from traditional capital structure theories not only because it examines the capital structure changes from the capital market perspective, but also it relaxes the assumption of the market efficiency As one important capital structure theory, the market timing concept is being studied by more and more researchers However, most of these studies concentrate on the timing behavior of general stocks REITs are always excluded in general corporate research In this study, we choose REITs as the research focus and examine the market timing and capital structure changes of REITs

1.2 About Real Estate Investment Trust

According to the definition given by National Association of Real Estate Investment Trust (NAREIT), Real Estate Investment Trust, or REIT is a company that owns, and in most cases, operates income-producing real estate REITs are created to give smaller investors the ability to invest in large-scale commercial properties During the past decades, REITs have grown dramatically in size and

importance in both U.S and other countries Among all the REITs in different countries, US REITs have the longest history and are the most developed

US REITs were created in the 1960’s by the Congress and aimed at facilitating the investment in large-scale real estate The market capitalization of REITs has increased from 1,494.3 millions of dollars in 1971 to 307,894.73 millions of dollars

in 2004 (with reference to the Appendix) The total number of REITs also increased from 34 to 193 during the period Besides the rapid growth, REITs also have good investment performance and attracted more and more investors, especially since 1990’s REITs are generally considered as total return investments with high dividends and long-term capital appreciation The attractive risk/reward tradeoff offered by REITs, the low correlation between REITs returns and other asset classes, and the diversification offered by REITs to the investment portfolio, are the reasons

of the popularity of REITs

We examine the market timing of REITs, which are usually not included in the studies for general stocks as they are special investment vehicles However, it is precisely because REITs are special vehicles that make them more interesting test bed for alternative capital structure explanations This is for several reasons First, because REITs face higher dividend payout requirements, they have to look to external funding for capital investment and asset acquisitions and capital structure adjustments (Brown and Riddiough, 2003) Second, the higher tangibility of REITs assets and better predictability of cash flows imply that less asymmetric information

is expected from REITs compared to industrial stocks Given that asymmetric

information is fundamental to capital structure theories, evidence from the REIT industry would provide even stronger support for information-based theories

There are three different types of US REITs: equity REITs, mortgage REITs and hybrid REITs We choose equity REITs because equity REITs own or have an equity interest in rental real estate, and the operation of equity REITs are close to general stocks Mortgage REITs, which make or own loans and other obligations that are secured by real estate collateral, and hybrid REITs, which combine the investment strategies of equity REITs and mortgage REITs, are excluded from this study At the same time, equity REITs take the largest market share of the whole US REITs market (Refer to the Appendix) The REITs mentioned in following parts of the study represent equity REITs

1.3 Research Objective

The focus of this study is on testing the market timing and capital structure changes of US equity REITs The first research question is: do REITs time the market in their capital issues? If the answer is yes, what is the underlying mechanism

of this timing behavior, the information-based market timing or market based market timing? The answer to the above research questions could help us learn more about the performance of REITs around capital issues Investors could choose their investment strategies if they have better knowledge of the performance of REITs, and REITs managers and practitioners could learn from others and know what is happening in the REITs industry For researchers, the findings of this study could help develop the relevant capital structure theory

inefficiency-The second part of the study examines the security choice of REITs against different capital structure theories The key question in this part is: which capital structure theory could be used to explain the security choice of REITs, trade-off theory, pecking-order theory or market timing theory? The answer to this question allows us to understand the influences of market conditions (according to market timing theory), and the asymmetric information between investors and REITs managers (according to the pecking-order theory), and the bankruptcy risk (according to the trade-off theory) on the security choice of REITs At the same time, the security choice of REITs offers a new test ground of different capital structure theories because REITs have little retained earnings to resort to

1.4 Organization of the Study

Chapter 2 reviews the related literature to the topic of this study The research methods and data sources are introduced in Chapter 3 The analysis of market timing

of REITs is carried out in Chapter 4 Whether there is market timing behavior in REITs security offerings, and whether this behavior is supported by asymmetric information or market inefficiency are studied Chapter 5 examines the security choice of REITs and tries to understand which capital structure theory could be used

to explain the security choice of REITs Chapter 6 concludes this study

Chapter 2

Literature Review

2.1 Trade-off theory and pecking order theory

The trade-off theory implies that there should be an optimal debt level, although this optimal debt level is not directly observable, and could vary through time Empirical evidence that supports the trade-off theory and the optimal capital structure comes from Schwartz and Aronson (1967), Bradley, Jarrell and Kim (1984), etc With regard to the security choice that we analyze in the empirical part of this study, the trade-off theory implies that all capital financing decisions are made to adjust the leverage ratio to the optimal leverage ratio

For pecking order theory, firms do not need to keep a target debt level The driving forces of the capital structure changes of firms are the financing needs of investment opportunities The pecking order theory is developed by Myers and Majluf (1984) in which there is no well-defined optimal capital structure When there is a good investment opportunity, managers would avoid using equity to finance the investment, because equity issues would jeopardize the benefits of current

needs, managers would use internal funds first If there are not enough internal funds, they would use debt Equity is the last resort in the financing activity This pecking order that firms follow in choosing financing resources explains why more profitable firms borrow less: because they have more internal funds

Shyam-Sunder and Myers (1999) attempts to differentiate between the pecking order theory and the static trade-off theory Based on the premise that changes in debt ratios result from the need for external funds rather than a desire to maintain an optimal capital structure, Shyam-Sunder and Myers (1999) estimate the need for funds as cash flow deficits They also test the static trade-off theory by regressing the difference in the previous period’s debt from the firm’s target debt level on the amount of debt issued or retired What they find is that the pecking order model has superior explanatory power

Chirinko and Singha (2000) improve on Shyam-Sunder and Myers (1999)’s work by considering equity issues and debt capacity of firms Their result could not reject the pecking order theory, and alternative tests are needed The Shyam-Sunder and Myers model is also examined by Frank and Goyal (2003) with more current data They did not find evidence consistent with the adverse selection and the asymmetric information implications of the pecking order theory With the data of the financing choices of firms that had recently gone public, Helwege and Liang (1996) found no evidence of a clear pecking order for firms to raise external capital

Aside from direct studies of the pecking order theory, the announcement effects

of the financing decisions are also studied to test the pecking order hypothesis Under

the pecking order theory, the announcement of a secondary equity offering (SEO) is perceived as a negative signal by investors, which leads to negative reactions of investors at the announcement of new stock issues As debt is less sensitive to information asymmetry, the announcement of a risk debt issue is somewhat less negative

Harris and Raviv (1991) find evidence consistent with the pecking order theory with the documentation of significant negative announcement effects for SEOs and insignificant effects for debt issues In more recent literature, Jung, Kim and Stulz (1996) find mixed evidence on the pecking order with negative announcement returns for stocks and insignificant returns for debts What they also find is, even with debt capacity, some firms use equity to fund good investment opportunities, which is inconsistent with the pecking order hypothesis D’ Mello and Ferris (2000) find support of the pecking order theory, as more negative announcement period returns are observed for firms with more severe information asymmetry

The market reaction to capital offerings of REITs is studied by Howe and Shilling (1988), in which paper, negative equity-issuance effect is found by REITs Similarly, in Francis, Lys and Vincent (2004), reaction to REITs equity offerings is also found to be negative, although more favorable than reaction to industrial equity offerings The asymmetric information is believed to be an important factor in determining the reaction

2.2 Market timing theory

In the Graham and Harvey (2001) survey involving of CFOs, it is found that more than two thirds of the managers issue equity when it is overvalued and would issue debt when the interest rates are low This empirical evidence offers direct support to the market timing theory Market timing theory implies that firms make capital financing decisions according to the capital market conditions Equity market timing means that equity issues decisions are made according to the stock prices, and debt market timing means that debt issuing decisions are made based on the interest rate levels The optimal leverage ratio does not exist in the market timing theory According to the market timing theory, the security choice of REITs depends on the market conditions, which means that choosing equity or debt is not fixed with some order

Research findings about equity and debt market timing are different Equity market timing gains supports from two different timing theories: information-based market timing and market inefficiency-based market timing Studies of information-based equity timing theory were carried out by Lucas and McDonald (1990), Korajczyk, Lucas and McDonald (1991) and Korajczyk, Lucas and McDonald (1992) These studies focus on the variations of adverse selection costs of equity issues with the existence of asymmetric information Since the equity issues only occur when the stock prices are overvalued, the stock prices would decrease on announcement of the equity issues Outsiders would lower their evaluation of the issuing firm’s quality, and this creates a “lemons market” in new equity issues In Korajczyk, Lucas and McDonald (1991) and Korajczyk, Lucas and McDonald (1992), the asymmetric information is not fixed over time, and firms tend to issue equity when the market is most informed about the quality of the firm, for example, after earnings releases They

find that managers can control the informational disadvantage of the market by choosing the timing of an equity issue, which is called market timing

The market inefficiency-based equity market timing arises from the long-run performance study after equity issues The underperformances after IPO (Ritter, 1991), the long-run underperformances after equity issues (Loughran and Ritter, 1995), etc seem to indicate that market timing is the only explanation for the long-run performance anomaly after equity issues The anomalies observed come from the timing of inefficient capital markets Besides the market timing behavior of managers before equity issues, other corporate decisions, for example spin-offs, debt offerings, and stock splits also exhibit long-run abnormal returns (Desai and Jain (1999), Spiess and Affleck-Graves (1999), and Ikenberry, Rankine and Stice (1996))

At the same time, some doubts are also cast on this version of market inefficiency Other researchers also performed similar tests but they do not find strong evidence against market efficiency (Fama and French (1988) and Kothari and Shanken (1997)) Besides, on-going debate exists to give methodological suggestions

on the long-run anomaly (Fama (1998), Eckbo, Masulis, and Norli (2000), and Brav, Geczy, and Gompers (2000)) The robustness of the long-run returns calculations are questioned by researchers

In Baker and Wurgler (2002), the equity market timing theory gained strong support by finding that market timing has persistent effects on the capital structure, and the capital structure is the cumulative outcome of past attempts to time the equity market The market-to-book ratio is used to measure the market timing opportunities

perceived by managers and managers are believed to time the market in equity issues

In addition, Baker and Wurgler (2002) find that leverage is strongly negatively related

to the market-to-book ratio Since trade-off theory and pecking-order theory could not

be used to explain this long-run relationship, it is believed that market timing is the only plausible explanation However, the underlying reason of this result is not clear

in Baker and Wurgler (2002)

Both two versions of equity market timing could be used to explain this long-run relationship between leverage and market-to-book ratio In the information-based market timing, which is the dynamic version of Myers and Majluf (1984), the market-to-book ratio is inversely related to adverse selection, and temporary fluctuations in the market-to-book ratio measure variations in adverse selection The second equity market timing theory assumes that managers believe that they can time the equity market and issue equity when they perceive that investors overvalue the firm The market-to-book ratio is a proxy of market misevaluations As market-to-book ratio could represent both adverse selection and market misevaluations, Baker and Wurgler (2002) could not tell which market timing theory dominates

With regard to the debt market timing, Guedes and Opler (1996) find that the maturity of issues is negatively related to the term spread Baker, Greenwood and Wurgler (2003) also support the market timing theory by finding that firms tend to issue long-term debt when inflation is low and term spread is narrow On the other hand, firms issue short-term debt when inflation is high and term spread is wide Barry et al (2003) investigate the debt issuing decision of firms in relation to

historical interest rates, and conclude that the amounts and numbers of debt issues are higher when current interest rate is lower compared to historical interest rates

However, most of these studies concentrate on the timing behavior of general stocks REITs are always excluded in general corporate research In this study, we choose REITs as the research focus and examine the market timing and capital structure changes of REITs

REITs are chosen as the focus of study because they are quite different from general stocks in the following aspects First, REITs are highly correlated with real estate properties The heavy dependence of real properties on capital has differentiated REITs from other stocks Second, US REITs are required to pay out at least 90% of the taxable income to shareholders Such a high dividend payout requirement may influence the finance decisions of REITs managers as there would

be less retained earnings for REITs Third, as a kind of indirect investment tool, REITs are required to disclose more information about the firm performance compared to general listed firms The more information is disclosed, the less asymmetric information is expected between REITs managers and investors Under this condition, what kind of influence does the information asymmetry have on the equity issues of REITs? All the above differences between REITs and general stocks make the capital structure changes of REITs more interesting and distinctive

Previous studies on REIT capital structure changes include Brown and Riddiough (2003), who analyze the public financial offerings of equity REITs They find that equity offerings are more likely to be used for investment, while debt

offerings are normally used for adjusting the capital structure A probit model is used

to examine the choice to issue public debt versus equity It is found that a firm’s asset, liability and income structure is related to the REIT’s decision whether to issue public debt or equity When facing the choice between equity and debt, REITs with higher debt levels would choose not to issue public debt The larger the REIT, the more likely it would issue debt This paper examines the capital structure changes of REITs, taking into account different characteristics of REITs, but the market timing theory is not tested Instead, this paper focuses on testing classic capital structure theories

Previous research works about the market timing behavior of REITs include Li and Ooi (2004) They find that US REITs time both the equity and debt market conditions using a sample of equity REITs from 1986 to 2003 The M/B ratio is considered as the proxy of market misevaluations and equity/debt issues/repurchases are analyzed They conclude that REITs time the market in both equity and debt offerings However, neither asymmetric information nor market inefficiency assumptions are tested in their study At the same time, their conclusions are based on separate studies of equity and debt offerings

Feng, Ghosh and Sirmans (2004) use the same research methodology as Baker and Wurgler (2002) to examine the relationship between market-to-book and leverage ratios of REITs They find that REITs with high market-to-book ratios have high leverage ratios and historical market-to-book has long-term persistent impact on current leverage ratio Pecking order theory is believed to be more relevant in explaining REITs capital structure theories Their study is similar to Baker and Wurgler (2002), in which two versions of market timing could not be differentiated

In this paper, both the equity and debt market timing of REITs are tested Then hypotheses from both the information-based and market inefficiency-based equity market timing are tested to see which theory could be used to explain the equity market timing of REITs Last but not least, to understand the capital structure changes

of REITs more clearly, we include three capital structure theories (trade-off theory, pecking order and market timing) in our probit model to test the financing pattern and security choice of REITs Our study is complementary to previous literature in REITs capital structure studies

This study seeks to make four contributions to the extant literature First we test whether market timing exists for REITs by evaluating if REITs capital issues depend

on capital market conditions A vector autoregressive (VAR) model is applied as it allows us to know how capital structure changes Typically, the total leverage ratio is examined to test different capital structure theories Focusing on the leverage ratio could not, however, differentiate between different capital-structure adjustment paths For example, debt issues and equity repurchases have the same effect on the capital structure ratios If we use the total leverage ratio, then it is impossible to know which adjustment path is used by REITs With VAR model, we could overcome this shortcoming and keep track of the changes of both equity and debt Another advantage of the VAR model is the choice of the variables does not to be determined before estimation This is allows more relevant variables to be included in the regression A detailed description of the VAR model is given in Chapter 3

The second contribution is to examine the short-run and long-run performance of REITs after equity issues to test the hypotheses of information-based market timing theory (Korajczyk, Lucas and McDonald, 1991) and market inefficiency-based market timing theory (Loughran and Ritter, 1995) According to market timing under the asymmetric information framework, managers have superior information about the stock prices compared to investors Under such conditions, managers choose to issue equity when the stock prices are overvalued and repurchase them when stock prices are undervalued Thus the positive abnormal returns preceding the equity issues would be used as one method to test the market timing abilities If managers could time the market when they issue the equity, the stock prices should be increasing before the equity issues As the stock prices have reached its peak when new equity is issues, the stock prices would decline at the announcement or issues of the new shares Another hypothesis is that managers would choose to issue equity when the market is best informed of the quality of the firm to reduce the adverse selection problem

The market inefficiency-based market timing theory is supported by the long-run return anomaly after equity issues as documented by Ritter (1991) and Loughran and Ritter (1995) To test whether the market efficiency is challenged by REITs equity issues, the long-run performance after secondary equity issues of REITs is tested Previous research findings about the long-run performance of REITs SEOs include Howton, Howton and Friday (2000), which found long-run underperformance of REITs SEOs when comparing the holding period return of equity issuing REITs with non-equity offering REITs The method applied by Howton, Howton and Friday (2000) is to compares equity issuing REITs with non-equity offering REITs As most

of the REITs (91 out of 123) in the sample have at least one equity offering during the

study period, we will use the Fama and French four factor model to calculate the

long-term returns This approach is preferred over the Horton et al.’s approach in that we

control for relevant risk factors

Our third contribution is to examine the factors influencing the choice of debt offering by REITs In particular, we appeal to the idea that the current level of interest rates in relation to the past history of interest rates matters in debt market timing, and that managers are backward-looking We also test the forward-looking debt market timing hypothesis by examining whether interest rates rise after REITs issue debt

The fourth contribution is to differentiate between competing capital structure theories by examining the security choice of REITs when financing decisions are made Which security type to choose when firms need external capital, equity or debt? Different capital structure theories have different assumptions and explanations

Trade-off theory implies that all capital financing decisions are made to adjust the leverage ratio to the optimal leverage ratio, and the optimal leverage ratio is correlated with the bankruptcy costs The pecking-order theory suggests that should firms need external capital, they would use debt first, as debt suffers less asymmetric information problem The market timing theory focuses on the capital market conditions and implies that firms would choose one type of security when it is relatively “cheap”

As motivated earlier, REITs offer a good test ground of the security choice because REITs have high dividend payout requirements, and limited retained earnings

availability Such restrictions effectively limit the security choice of REITs to issue either equity or debt As such, we differentiate between the different capital structure theories by examining the security choice of REITs between equity and debt This allows us to conduct a “cleaner” test of different capital structure theories

The VAR model is employed to test the relationship between relevant variables

to examine the market timing behavior of REITs in Chapter 4 Now we briefly introduce the models here The autoregressive model is

t p t p t

Eε ε This equation system is a vector autoregression, or VAR Vector means that we are dealing with a vector of variables while autoregressive means the appearance of the lagged value of the dependent variable

Akaike or Schwarz Criterion can be used to decide how many lags to take

One of the virtues of the VAR is that it obviates a decision as to what contemporaneous variables are exogenous; it has only lagged (predetermined) variables on the right-hand side, and all the variables are endogenous (although in some circumstances exogenous variables can also be included) VARs are special, because in simultaneous or structural equation models, the equations in the system should be determined before the estimation of the model This kind of identification is often subjective Using VAR this subjectivity could be avoided

Employing VAR model allows us to test which variables are more relevant in determining the equity issues or stock returns, and we do not need to specify the relationships before the estimation VARs have been used primarily in macroeconomics for forecasting Besides, VARs can also be used for testing Granger Causality and studying the effects of the policy

Although VAR model has many advantages, critics of VAR point out that some problems exist in VAR modeling:

(1) VAR model is a-theoretic because it lacks prior information

(2) Although VAR model is good at forecasting, it is not suitable for policy analysis

(3) Choosing the appropriate lag length is not very easy

(4) Strictly speaking, all the variables should be stationary If not, we have to transform the data to stationary ones

3.2 Probit Model

In Chapter 5, a probit model is used to examine the security choice of REITs

The probit model is used when the dependent variable is a binary variable, and the

probit model could be used to predict the dependent variable on the basis of

independent variables

The probit regression equation is:

)'()

1

Pr(y i = =F x i b (2)

Here b is a parameter to be estimated, and F is the normal cumulative distribution

function of the standard normal distribution, as shown in the following figure

Figure 1 Cumulative Normal Distribution

The probit model can be obtained from a single latent variable model Suppose

that

And suppose that Y is an indicator for whether the latent variable y* is positive,

then it is easy to show that

)'()

1

Pr(y i = =F x i b (4)

The coefficients of the probit model can be estimated using maximum likelihood estimation instead of ordinary least squares Maximum likelihood estimation is a method for estimating the parameters of the population from which sample data is drawn, and the maximum likelihood estimate is the parameter associated with the highest probability of observing the sample statistic

In our test, y=0 when equity issues are made and y=1 when debt issues are made

The coefficient of the model represents the effect of changes of the independent variables (such as market conditions, REITs characteristics, etc) on the probability of debt/equity issues

3.3 Data Sources

All the US equity REITs captured in this study are reported by National Association of Real Estate Investment Trust US REITs are of relatively long history, and this allows us to know the capital structure theories in more developed REITs markets There are 123 equity REITs captured in this study, which comprise a great majority of the equity REITs reported by NAREIT

Table 1: Equity and debt offerings by year

Year Number

of Equity Offerings

Amount

of Equity Offered (mil $)

Number

of Debt Offerings

Amount

of Debt Offered (mil $)

Total REITs captured in the year

data of equity REITs that are captured spread from January 1980 to March 2004, 279 months in total

The sample of REITs SEOs obtained from the SDC database includes equity REITs SEOs between January 1980 and March 2004 The final sample contains 424 SEOs from 91 REITs The final sample of REITs debt offerings includes 326 debt issues from 56 equity REITs from January 1980 to March 20041 (see Table 1 for details)

4.1 The equity and debt market timing of REITs

Equity market timing theory implies that managers choose to issue equity when their stock prices are (or believed to be) overvalued and repurchase the stocks when they are (or believed to be) undervalued Thus equity issues always occur after positive stock returns Debt market timing implies that debt issues are made when the interest rates are low We employ the VAR model here to find whether there is market timing in REITs capital issues For REITs captured in this part, the following hypotheses are tested: will net capital issuing decisions be made according to the capital market conditions?

The VAR test is ran according to regression equation (1) shown below:

( )

t t t

t t

n t n

t n t n

t t

t t

t t

AVMB E

TS D SP C I B NAREIT

Debt Equity

A

NAREIT Debt

Equity A

NAREIT Debt

Equity

ε

µ

++

++

++

1 1 1 1 1 1

1 1

1 1

),

,(

),

,(

,,

AVMB : The average of the M/B ratio of all the equity REITs in quarter t

The detailed explanations of these variables are:

(a) Market value of equity= price closed at the end of the third month of each quarter

* shares outstanding (splits adjusted)

(b) Long-term debt= total long-term debt

(c) Market-to-book ratio= market value of assets/ total assets= (market value of equity + long-term debt+ preferred stock + debt in current liabilities)/total assets

(d) Long-term interest rate= 10 year interest rate per year (%)

2 As long-term debt is believed to be closely correlated with the long-term interest rate in market timing, we use

The statistical summary of some cross sectional variables is listed in the following Table The vast majority of the debt issues by equity REITs are fixed-rate issues The time series description of the S & P500 index return and the 10-year interest rate is plotted in Appendix

The VAR model is estimated with panel data Because of the special characteristics of the panel data, individual heterogeneity should be considered and appropriate econometric techniques should be applied Then the data from different observations could be pooled to estimate the coefficient Holtz-Eakin, Newey and Rosen (1988) give detailed technical explanation about using panel data in VAR model

The VAR model results are listed in Table 43 It is evident from Table 4 column

1 that REIT index returns in the previous quarters have positive effects on equity issues When the stock price of the REIT market is increasing, REIT equity issues are more likely This is consistent with the market timing theory

Another finding is that the equity issues have negative effects on the stock return

in the subsequent quarters, although such effects are not significant This can be observed from the last column in Table 4, where the NAREIT return is the dependent variable This negative effect implies that equity offerings are made when stock prices are peaking These results indicate that equity market prices are important for equity financing decisions of REITs Similar findings were made in Baker and Wurgler (2000)-the equity share in the total equity and debt issues is a good predictor of future

aggregate market returns Higher equity share will lead to lower stock returns in the future Our findings are complementary to Baker and Wurgler (2000) as we find the negative relationship between current equity issues and future market returns is also valid for the REIT sector

Next we examine the effects of the interest rates on the issues of debt We find that increasing interest rate leads to decrease of debt issues (column 2).In other words, REITs also time the debt market This is consistent with one of the conclusions of Li and Ooi (2001) The above tests clearly show that market timing behavior exist in REIT financing activities4 Put differently, capital market conditions are important determinants of the capital issues of REITs

In order to better understand the market timing of REITs, we seek to determine the fundamental mechanism for timing behavior Is this market timing based on the asymmetric information between investors and managers (Korajczyk, Lucas and McDonald, 1991), or that managers capitalize on market inefficiency to time the market (Loughran and Ritter, 1995)? Hypotheses from both would be tested in the following sections

4

Table 2: Statistical Summary

EQUITY t : Net change in equity of all REITs in quarter t; DEBT t : Net change in book value of total long-term debt of all REITs in quarter t; ASSET t: Size of the REIT at the

end of the previous quarter (log value of total assets); AVMB t : Average of the M/B ratio of all the equity REITs in quarter t; NAREIT t: NAREIT equity REIT index return in

quarter t; INT t : Long-term (10 year) interest rate in quarter t; S&P t : S&P 500 composite index return in quarter t; TS t: Term-spread between long-term interest rate and

short-term interest rate (3-month T-bill rate) in quarter t; LEV t : Leverage ratio of the REIT at the end of the previous quarter t;

Table 3-1: Correlation Matrix between Time Series Variables

Table 3-2: Correlation Matrix Between Cross Sectional Variables

Term-spread between long-term interest rate and short-term interest rate (3-month T-bill rate) at end of

previous quarter; DC: Debt capacity dummy variable that shows whether the debt capacity is reached when the new equity/debt issue is made; STK_AP: REIT stock price appreciation in the previous quarter; SVAR:

Variance of the stock returns in the past year

Table 4: VAR Result

Vector Autoregression Estimates

EQUITY t : Net change in equity of all REITs in quarter t; DEBT t : Net change in book value of total

NAREIT t : NAREIT equity REIT index return in quarter t; INT t: Long-term (10 year) interest rate in quarter

t; TS t: Term-spread between long-term interest rate and short-term interest rate (3-month T-bill rate) in

Hypothesis 1: Stock prices fall at the equity issue announcement

Howe and Shilling (1988) examine the stock price reactions to the announcement of new security offerings by REITs from 1970-1985, and find a positive stock price reaction

to debt offerings, and a negative stock price reaction to equity offerings In this paper we use a more recent dataset to reconfirm and update the evidence