AN ANALYSIS OF THE FINANCING DECISIONS OF REITS: FROM A CAPITAL MARKET PERSPECTIVE Joseph T.L OOI # Department of Real Estate National University of Singapore 4 Architecture Drive, Singapore 117 566. Tel: (65) 6516 3564 Fax: (65) 6774 8684 E‐mail: rstooitl@nus.edu.sg Seow‐Eng, ONG Department of Real Estate National University of Singapore 4 Architecture Drive, Singapore 117 566. Tel: (65) 6516 3552 Fax: (65) 6774 8684 E‐mail: seong@nus.edu.sg Lin, LI GIC Real Estate International China Co., Ltd #805 Azia Center, 1233 Lujiazui Ring Road Shanghai, 200120, P.R. China Tel: (86) 21 6165 1978 Fax: (86) 21 6165 1900 Email: lukeli@gic.com.sg December 22, 2007 # corresponding author – can be contacted via e‐mail at rstooitl@nus.edu.sg AN ANALYSIS OF THE FINANCING DECISIONS OF REITS: A CAPITAL MARKET PERSPECTIVE Abstract This paper examines the relationship between capital market conditions and the financing activities of REITs, which include both their capital raising and capital reduction events. Tracking the financing activities of the REIT sector between 1986 and 2003, the study shows that the market‐timing story describes REITs financing activities much better than the traditional theories of capital structure. The evidence shows that REITs exhibit strong market‐timing behavior, in terms of when and what type of capital to issue or reduce, to take advantage of variations in their relative costs in the capital market. Specifically, REITs time their equity offerings to coincide with periods of high stock valuation. Debt securities, on the other hand, are preferred when the long‐term rate is low and the credit spread is narrow. REITs appear to issue both debt and equity securities when investors are more risk‐averse. We also find that highly geared REITs are more likely to engage in net debt reduction activities, suggesting that they have an optimal target debt level. Hence, whilst the findings favor the market‐timing theory, they do not imply that the traditional theories of capital structure are irrelevant. 1
1. Introduction The traditional way to examine the capital structure decisions of firms is to carry out cross sectional studies to identify the key determinants of leverage. However, the main limitation of such cross‐sectional studies is that a firm’s deliberate financing activities are not separated from any passive change in the leverage, which could arise due to the accumulation of earnings or losses, and adjustment to its property holdings due to value appreciation or diminution. Welch (2004) observe that prior stock returns are the main determinants of leverage and that firms do not actively offset the effects of stock returns on their capital structure. Strebulaev (2007), similarly, show that in a dynamic setting with frictions, the leverage of most firms is likely to deviate from the optimal leverage most of the time because firms adjust their leverage by issuing or retiring securities infrequently. Consequently, even if firms follow a certain model of financing, any static cross‐sectional model cannot explain differences between firms since their actual and optimal leverage may differ. Thus, such balance‐sheet approach may not be suited to test theories that rely on state variables that fluctuate over time (see Guedes and Opler, 1996). An alternative and a more direct approach would be to examine the marginal financing decisions of firms. Discrete choice models, which have been employed to predict the debt‐equity choices of firms (Marsh, 1982; Jung, Kim and Stulz, 1996), eliminate the problem associated with passive changes in leverage associated with cross‐sectional regressions. However, these models have their own limitations too. In particular, the debt‐
equity choices are conditional on the firms having decided to seek funding in the capital market. In practice, firms usually consider simultaneously when and which type of security to issue. Huang and Ritter (2004), therefore, contend that studies on financing decisions should consider the joint determinants of when and which security to issue. For the joint decision, they employed a multinomial logistic (MNL) model. Furthermore, the focus on 2
debt‐equity choice does not embrace all the refinancing options available to the firms. Firstly, firms may raise both debt and equity capital within the same period, but they are omitted from most studies on the grounds that their inclusion may lead to confounding results. Dual issues, in reality, are important financing events that can induce substantial changes to the firm’s capital structure (Hovakimiam, Hovakimian and Tehranian, 2004). Secondly, private debt, which is a significant source of funding for many firms, is often disregarded in prior studies. Given that private debt, in the form of bank loans, account for one‐third to half of total capital issued by Real Estate Investment Trusts (REITs) each year (Wu and Riddiough, 2005), it is important for both private and public sources of funds to be considered together when examining the marginal financing decisions of REITs. Thirdly, the focus on security issuance only covers half of the story since capital reduction activities such as share repurchases and debt retirement also affect the capital structure of a firm. Leary and Roberts (2005) argue that although firms may appear inactive most of the times, they may still be buying back securities in clusters. The aim of our paper is to provide a more holistic examination of the financing points of a firm. Unlike prior studies, we do not confine our study solely to the debt‐equity choices. Financing events which involved funding from private sources as well as dual issues are incorporated as possible financing options of the firms. We also examine their decisions to reduce capital through debt reduction and share repurchases. Following Huang and Ritter (2004), the simultaneity of these decisions is incorporated within the MNL framework with the base financing option being status quo. In other words, the marginal financing decisions of the firm are examined simultaneously within a framework which recognizes that due to adjustment costs, firms do not engage in financial activities all the time. This is the main contribution of this paper. This study also contributes to the real estate literature by analyzing the financing decisions of REITs from a capital market perspective. The theories of capital structure views 3
a firm’s financing decisions as either a trade‐off between the benefits and costs of using debt, or as a pecking‐order where internal source of funding is preferred and should the firm need to use external funding, debt capital will be preferred over equity capital. Whilst these traditional theories assume an efficient capital market, a new stream of research has examined the implications of an imperfect capital market on managers’ behavior. Departing from the conventional theory, recent studies have examined how capital market conditions dictate the firms financing decisions. The market timing theory of capital structure, in particular, prescribes that firms attempt to minimize their cost of capital by timing their financing activities to take advantage of any temporary mispricing in their securities. This paper examines the relationship between capital market conditions and the financing activities of REITs, which are typically excluded in studies on capital structure of firms in general. Since they rely primarily on external sources of funding for their capital investment and asset acquisitions, REITs provide unique opportunities to examine how corporate financing activities and the capital markets are related. Moreover, REITs are usually highly leveraged and consequently, interest charges generally constitute the single largest expense item of most REITs.1 Thus, REIT managers have more incentives to monitor the debt and equity capital market closely because any savings on their cost of capital would have a greater impact on their corporate performance.2 REITs also provide a useful setting to test capital structure theories because the key drivers behind the pecking order and trade‐
off theories of capital structure are not that relevant in the case of REITs. First, REITs do not pay corporate tax; consequently, the tax benefit associated with using debt, which is central 1
Our data shows that interest charges account for between 30% and 70% of the total expenses incurred by a REIT. 2
In contrast to event studies which seek to measure the effects of financing decisions on the capital market (as reflected by the stock price reaction), the capital market perspective involves a reverse approach. Instead, we investigate how conditions in the capital market, such as equity market valuations and returns or debt market yields and spread, influence the financing decisions of REITs through the timing and choice of capital. 4
to the trade‐off models, is not applicable in the REIT context. This leads to the fundamental question as to why debt is used at all to finance REIT investment. Second, equity REITs buy and hold properties, which are tangible and not firm‐specific assets (Gentry and Mayer, 2002). Consequently, they are less exposed to bankruptcy and agency costs as compared to firms in other industries. Third, the transparent nature of REITs’ operations implies lower information asymmetry between the insiders and outsiders. REITs also operate under more stringent corporate governance and reporting rules. Fourth, due to the high distribution requirement3, REIT managers have less discretion to engage in managerial opportunism or over‐investment activities. Even successful REITs have to raise capital externally and are, hence, subjected to frequent monitoring and disciplining in the capital market (Ghosh, Nag and Sirmans, 1997). Given these conditions, we posit that a capital structure theory based on market timing behavior will provide a better explanation on the financial decisions of REITs. Analyzing the financing activities of the REIT sector at the aggregate level, we observe a close correlation between the financing decisions and the conditions in the debt and equity markets. At the firm level, the MNL estimation provide further empirical evidence that REITs time and choose their financing activities based on the time‐varying cost of debt and equity capital. We find that equity offerings of REITs are more likely to be issued during a hot market when stocks are high valued. Debt securities, on the hand, tend to be issued during periods of low long‐term interest and narrow credit spread. Besides the debt‐equity choice, the market‐timing behavior is also prevalent in the dual issuance and capital reduction activities of REITs. This paper is organized as follows: Section II reviews previous studies on the financing decisions of REITs. Section III analyzes the financial patterns of REITs over the study 3
The Internal Revenue Code requires that REIT pays dividend of at least 90 per cent of their taxable income. Prior to 2000, the distribution requirement was higher at 95 per cent. 5
period. Section IV and Section V presents the estimation model and describes the data. Section VI discusses the estimation results, whilst Section VII concludes. 2. Literature Review Following the seminal contribution of Modigliani and Miller (1958), development of the capital structure theory has generally followed two tracks. Firms are viewed to make their financing decisions either by balancing the cost and benefits of debt, or by following a pecking order with internal financing preferred over external financing, and if they need to raise external capital, debt will be preferred over equity (Myers and Majluf, 1984). Despite the volume of research that has been carried out, there is still disagreement as to which of the two theories better explain the capital structure decisions of firms. In a departure from the traditional theories of capital structure, Baker and Wurgler (2002) propose that the firm’s observed capital structure is nothing more than a cumulative outcome of its past attempts to time the equity market.4 Baker, Greenwood and Wurgler (2003) subsequently broadened the market‐timing theory to cover debt issues. They defined market timing as “raising finance in whatever form is currently available at the lowest‐risk adjusted cost”. Ritter (2002), similarly, shows that firms follow different pecking‐orders in different windows of opportunity in the capital market. In his model, equity can move temporarily to the top of the pecking‐order if the market overpriced the shares to the extent that equity capital 4
The notion that capital structure decisions are related to the equity market is not new though. Early studies which observed that firms tend to issue equity following stock price increase include Masulis and Korwar (1986) and Asquith and Mullins (1986). Motivated to minimize their cost of capital, rational managers would take advantage of investor exuberance by issuing more shares when the firm’s stock price is high. Conversely, when the stock is underpriced, firms would buyback their shares (Stein, 1996). 6
becomes truly cheap. Alternatively, if debt is really cheap in certain period, debt issues can also move temporarily to the top of the pecking‐order. A number of recent studies have provided empirical support for the market‐timing theory. Examining corporate financing activities of industrial firms between 1928 and 1997, Baker and Wurgler (2000) observe that the proportion of new equity issues is higher when the stock market is more highly valued. They also present evidence that managers time the equity issues successfully to coincide with “hot” phases of the market before the stock valuation returns to a more realistic level. In an extensive survey involving 392 CFOs, Graham and Harvey (2001) reported that many of the CFOs time their financing decisions to take advantage of temporary mispricing in the capital market. More than two thirds of the respondents said that they would issue equity when their share price has risen. Similarly, they would time their debt issuance to coincide with periods when the interest rates are low. Huang and Ritter (2004) also document that market‐timing behavior, based on variations in the relative cost of equity, provided a more satisfactory explanation for the observed variations in the financing patterns of industrial firms as compared to the static trade‐off theory or the pecking‐order theory. In the real estate literature, binary discrete choice models have been employed to study the debt‐equity choices of property companies (Ooi, 2000) and REITs (Brown and Riddiough, 2003). Analyzing the public security offerings of equity‐REITs between 1993 and 1998, Brown and Riddiough observe that REITs with higher pre‐offer levels of secured debt tend to issue equity, while those with higher pre‐offer levels of unsecured debt tend to issue public debt. Whilst Brown and Riddiough find that equity offerings are more likely to be used for investment and debt offerings are normally used for adjusting capital structure, they did not explicitly test for any capital structure theory. Gentry and Mayer (2002) find that REITs appear to finance marginal projects with a mix of debt and equity that is similar to their average debt‐equity mix. They interpret the result to be consistent with REITs having a 7
target debt ratio. On the other hand, Feng, Ghosh and Sirmans (2007) find that REITs with high market‐to‐book ratios tend to have high leverage ratios. They also observe that historical market‐to‐book has a long‐term persistent impact on current leverage ratio, which they interpret to be supportive of the pecking order theory. Two recent studies (Li et al., 2007; Boudry, Kallberg and Liu, 2007) have examined the ability of the market‐timing theory of capital structure to explain the issuance decisions of REITs. Consistent with the results in the mainstream finance literature, both studies recorded strong evidence supporting the market‐timing theory. Li et al. (2007) also find some support for the trade‐off and pecking order theories and conclude that there is no single theory that fully explains the capital structure of REITs. In contrast to the two studies, which focused primarily on the debt‐equity choices of REITs, the current study provides a more holistic analysis of the financing decisions of REITs which include not only debt and equity issues, but also dual issues as well as capital reduction activities thorough debt reduction and share repurchases. To our knowledge, our study is the first to cover a wider range of marginal financing options available to REITs. The MNL framework, adopted in this study, recognizes that REITs may not undertake financing activities all the time and that managers may tilt their marginal source of capital decisions towards a higher leverage by issuing more debt or repurchasing shares. Similarly, REITs wishing to reduce their leverage can choose different paths by either issuing new equity or retire debt. Our primary focus is to examine how the capital market conditions affect these financing decisions at the margin. 3 Analysis of the Financing Patterns of REITs Figure 1 presents the initial public offerings (IPO), seasoned equity offerings (SEO) and debt offerings of the sector over the study period. Between 1988 and 1991, the aggregate amount of capital issued by REITs was below $ 5 billion a year. The capital raising activities of 8
the sector started to climb progressively from 1992, reaching a peak in 1997 with 463 issues raising $ 45.4 billion. Both the number of offerings and the funding amount, however, declined over the next few years with the IPO activity disappearing completely from the REIT market in 2000 and 2001. From 2001 onwards, the capital raising activity of the sector started to increase again. Despite the lack of tax incentives to employ debt, many equity REITs are traditionally highly geared. Feng, Ghosh and Sirmans (2007), for example, note that REITs have average debt ratios of over 65% ten years after their IPO. Figure 1, interestingly show that from early to mid 1990s (with the exception of 1992) debt actually played a secondary role to equity capital. In particular, SEOs constituted around 70% of the total external fund raised by REITs between 1991 and 1997.5 From 1998 onwards, REITs turned increasingly to the debt capital market for their funding requirements. Since 1999, debt financing has outpaced SEO as the major form of external finance for REITs. As a result, the sector experienced a sharp rise in its aggregate debt level. A pertinent question to ask is how do the sector’s observed patterns of financing fit with the existing capital structure theories? Clearly, REIT’s preference for equity capital between 1991 and 1996 violates the standard pecking order of financing. Ghosh, Nag and Sirmans (1997, 1999) attribute REIT’s preference for equity capital to firstly, the lack of tax incentive to use debt and secondly, the problem of adverse information commonly associated with equity offering is mitigated by the fact that REITs have little retained earnings. Whilst this may be generally true, it still does not explain the popularity of debt capital in 1992 as well as in the latter periods. Was 1991‐1996 a special era in terms for REITs’ equity financing? We believe the change in preference over the sources of external 5
Ghosh, Nag and Sirmans (1997) also observe that REITs issued equity three times more frequently than debt and raised almost twice as much through equity than debt between 1991 and 1996. 9
capital through time exhibited by the REIT sector is consistent with a market‐timing behavior. The windows‐of‐opportunity theory proposed by Ritter (2002) hypothesizes that firm follows different pecking‐orders in different windows‐of‐opportunities in the capital market. For instance, equity can move temporarily to the top of the pecking‐order if the market overpriced shares to the extent that equity capital becomes truly cheap. Alternatively, if debt is really cheap in certain period, debt issues can also move temporarily to the top of the pecking‐order. Figure 2 charts the relative cost of equity and debt capital over the study period. Consistent with the market‐timing theory of capital structure, we observe that REITs are more likely to issue equity following a stock price run‐up. Note that during 1991‐1996 when equity offerings were in favor, the price‐earning multiples of the REIT sector were at a record high. Even within this window period, a dramatic fall in the price‐earning ratio can be seen in 1992. Reflecting the escalating cost of equity capital, Figure 1 shows a parallel slow down in the REIT SEO segment in the same year. Similarly, the second chart in Figure 2 shows that cost of borrowings has been declining in the later period, marking a return to favor of debt capital by REITs. To provide a clearer picture, the two charts in Figure 3 superimpose the capital market conditions against the fund raising activities of REITs. Consistent with the market‐
timing theory, REITs financing activities have a strong correlation with the equity and debt capital market conditions. It is interesting to observe that whilst REIT stock returns and general stock returns have moved in tandem with each other in the earlier years, they have started to diverge from each other in the mid 1990s. The divergence allows us to examine the correlation between the level of REIT equity issues and the broader stock market condition. The pattern exhibited in Figure 3 appears to indicate that the aggregate dollar amount of equity offerings by REITs tend to follow more closely to the general stock market performance than to the REIT sector performance. For instance, the equity offering surge 10
observed between 1995 and 1998 corresponded with above 20% annual return in S&P 500, whilst a sharp drop in the REIT sector’s performance in 1998 did not slow the pace of REITs equity offerings. On the other hand, the sharp decline in equity issuance activities during 2000‐2002 coincided with below ‐10% returns in S&P 500, whereas the NAREIT equity‐REITs Index registered healthy returns during the same period. The bottom chart in Figure 3 shows that REIT’s debt offering also increased in tandem with equity issues during 1992 to 1998, though it played a secondary role compared to equity issues. However, since 1999, REITs have turned increasingly to the debt capital market to meet their capital requirements. As a result, debt financing outpaced equity as the major form of external finance for the period of 1999‐2002. This pattern can be attributed to a combination of push and pull factors. On the one hand, the general weakness in the equity market, caused by the burst of the internet bubble, greatly reduces the potential demand for new equity offerings. This occurred at the same time as the REIT industry was experiencing an increase in the number of mergers, acquisitions and joint ventures, which accelerated the industry’s need for additional capital. On the other hand, the historically low interest rates provide ample liquidity in the debt market.6 Consistent with the market‐timing theory, the sector’s increased reliance on debt capital from 1997 onwards also corresponded with a low interest rate regime. Figure 4 tracks the aggregate equity buyback activities as well as the flow of debt funding into the REIT sector over the study period. Using cash flow statement data from the COMPUSTAT database, we identified a total of 101 such cases during the period 1986‐2002. A commonly expressed view in the market suggests that the sharp price decline of REITs stock between 1999 and 2002 may be the primary reason for the increased activities in 6
Rapid development of the commercial‐mortgage‐backed‐securities (CMBS) market and corporate bond market also contribute to this shift in the financing pattern. 11
share buybacks. The top chart in Figure 4 shows a raising trend in the number of share buybacks by REITs since 1999, coinciding with the sharp decline in the stock market. The bottom chart in Figure 4 shows that the net flow‐of‐fund from bank financing to the REIT sector has been consistently negative from 1999 onwards.7 In contrast, more debt capital is channeled into REITs through the public debt market. The drift from private to public sources of debt capital can be attributed to firstly, historically low interest rates which prompt REITs to lock‐in the favorable financing cost by substituting short‐term bank loans with public debts which are usually of longer maturity. Secondly, innovations in the CMBS market provides more options for REITs as well as reduces the yield spread over benchmark Government bonds. The financing patterns are consistent with Wu and Riddiough’s (2005) observation that REIT engage in “bridging finance” strategy where bank debt is used as a temporary financing vehicle by REITs and retired subsequently with proceeds from long‐
term unsecured debt. In summary, the capital market conditions appear to have a strong relationship with the aggregate time‐varying financing activities exhibited by the REIT sector. In the following sections, we will examine the relationship in more detailed by controlling for firm level attributes in the MNL model. 4. The Multinomial Logit Model In modeling the financing activities of REITs, we extend the feasible set to six financing options, namely no material change, equity issuances, equity repurchases, net debt issuance, net debt reduction, and dual offerings where the REIT issued both debt and equity 7
Although NAREIT and COMPUSTAT do not provide data on bank commitment of REITs, we employ the flow‐of‐fund data published by Federal Reserve to provide an overview of the aggregate private bank versus public debt fund flow into the REIT sector. 12
securities in the same period. The MNL model is applied to estimate simultaneously the probability of the occurrence and the choices between different forms of financing activities.8 Essentially, the MNL model is represented as follows: Let y denotes a random variable taking the values {0, 1, 2, … , J} where J = 5 in our case, x denotes a set of conditioning variables, and (xi , yi) is a random drawn from the population. Our primary interest is to explain how ceteris paribus changes in the elements of x affect the response probabilities, P ( y = j|x), j = 0, 1, 2, 3, 4, 5. In other words, we are interested to know how x changes the probability that REITn chooses option J. Since the probabilities must sum to unity, P ( y = 0|x ) is determined once we know the probabilities for j = 1, 2, 3, 4, 5. Let x be a 1 x K vector, the response probabilities of the MNL model 5
⎡
⎤
P ( y = j | x ) = exp ( xβ j ) / ⎢1 + ∑ exp ( xβ h )⎥ , j = 1, 2, 3, 4, 5 ⎣ h =1
⎦
where β j is K x 1, j = 1, …, 5. The MNL model expresses the probability that a specific alternative is chosen is the exponent of the utility of the chosen alternative divided by the exponent of the sum of all alternatives. Because the response probabilities are bounded by zero and one and must sum to unity, 5
⎡
⎤
P ( y = 0 | x ) = 1 / ⎢1 + ∑ exp ( xβ h )⎥ ⎣ h =1
⎦
The dependent variable in the model is the mutually exclusive financial activities; namely, equity issuance/repurchase, net debt increase/reductionand dual offering, which are coded with an unordered integer ranging from 1 to 5. In particular, the dependent variable equals one if the REIT issues external equity, two if it repurchases shares, three if it issues net debt, four if it retires net debt and five if it engages in both debt and equity 8
Terza (1985), Guedes and Opler (1996) and Huang and Ritter (2004) has used the MNL model to examine assignment of bond ratings, debt maturity and financing choices of corporations. 13
offerings in the same quarter. REITs that did not experienced any material changes in their financing structure during the review period are taken as baseline scenario (0). 4.1 Capital Market Variables. Having identified the feasible financing choice subsets for the individual REITs, we proceed to specify the explanatory variables that may influence the decision process. Our explanatory variables essentially comprise factors reflecting the relative cost of equity or debt capitals for the individual REITs as well as market‐wide costs. The market‐timing behavior postulates that firms would take advantage of the relative costs of debt and equity capital when making their financing decisions. The first set of independent variables is included to detect if REITs exhibit any equity market timing behavior. The variables include both individual stock price valuation as well as market‐wide sentiments associated with equity capital. The firm‐specific variables are market‐to‐book ratio, price‐earning ratio and price appreciation of the individual REIT stock with higher stock prices generally translating to lower cost of funds.9 The preceding twelve months return of the S&P 500 Index and the National Association of Real Estate Investmetn Trust (NAREIT) Index for equity REITs are adopted to reflect the general sentiment in the stock market and the REIT sector, respectively. The returns associated with Fama‐French size and growth factors are also incorporated to control for changes in investors’ preference for different types of equity, namely small versus large cap stocks and value versus growth stocks. 9
Another common metric used for valuing REIT stocks is dividend yield, which is measured as the annualized dividend rate divided by the REIT’s closing stock price. Due to its high correlation with price‐earnings ratio, this variable was not included in our regression model. Nevertheless, we ran a separate MNL regression by substituting the price‐earning ratios of the REIT with its dividend yield, and found the results to be robust. 14
Market timing managers would be expected to take advantage of any apparently low interest by issuing more debt. Thus, the second set of independent variables represents the relative costs of debt capital, which include debt capital market related variables (namely, the long‐term government bond yield, term‐spread of interest rate, real risk‐free rate and credit‐spread of corporate bond yield) and the credit rating of the individual REIT. Nevertheless, there is a major difference between debt market timing and equity market timing. Whilst equity market timing is typically connected to inside information, debt market timing can only be driven by publicly available information because firms do not have inside information about future interest rates (Baker, Greenwood and Wurgler, 2003). The two set of independent variables are summarized in Table 1 together with their definitions, whilst Table 2 shows the correlation matrix of the capital market variables. Overall, the correlations are reasonably low. In particular, general stock market return is only moderately correlated (0.243) with REIT sector returns. In addition, correlations among the four debt capital market variables are also moderate, indicating that the four variables specified capture different aspects of risk factors in the debt capital market. 4.3
Firm Characteristics We also incorporate three firm‐specific attributes that have been identified in previous studies as significant determinants of capital structure decisions. They are the size, profitability, and leverage of the individual REITs. Even though all the REITs may face the same external capital market conditions, these intrinsic factors could also influence the firm’s financing decisions. For instance, a highly leveraged REIT with poor corporate performance may find it costly to issue new debt securities even in a favorable debt capital market condition. Similarly, larger firms may have higher bargaining power over investors and possess certain advantages in timing their issuance. They also tend to have less volatile 15
cash flows and are therefore less likely to become financially distressed (Rajan and Zingales, 1995). 5. Data and Sample Our study covers 17 years starting from 1986, which coincides with the Tax Reform Act introduction that has fundamentally changed the REIT landscape.10 We restrict the analysis to equity REITs, which primarily own and hold investment properties. Names of the individual REITs are first identified and verified against the NAREIT database. Financial information on the individual REITs is then extracted from their balance‐sheet, income‐
expense and cash flow statements in Standard and Poor’s COMPUSTAT database. Information on equity market performance is obtained from the Center for Research in Security Prices of University of Chicago (CRSP) database, whilst data on the debt capital market is extracted from the Federal Reserve database. The final sample covers 144 equity REITs with a total market capitalization of $205 billion. It accounts for nearly 94% of the total capitalization of the REIT sector. Whilst most studies are constrained to annual frequency due to data limitation, we were able to examine the market timing behavior of REITs on a quarterly interval which is advantageous given that capital market conditions can fluctuate quickly and that firms often engaged in more than one financing activity within a particular calendar year (Leary and Roberts, 2005). This allows us to capture the dynamics in the capital market and offer more insights into the firms financing activities. 10
Prior to the legislation, REITs functioned primarily as passive income‐producing asset owner and operator under a third‐party management structure. However, after 1986, REITs were allowed to self‐manage. Since then, equity REITs have increasingly become like real estate operating corporations that engage in a wide range of real estate activities, including leasing, development of real property and tenant services. 16
Through the cash flow statements of the individual REITs, a detailed picture of their financing activities is compiled by identifying the exact amount of financing issued and retired in every quarter. Although the statements of cash flow do not pick up any non‐cash transactions, such as exchange offers, this limitation does not detract us from focusing on the normal financing activities of REITs, instead of exceptional equity or debt issuance events related to major mergers or corporate restructuring. Based on the information extracted from their cash flow statements, the financing activities of the individual REITs are then classified into the following categories: (1) equity issuance, (2) equity repurchase, (3) net debt increase, and (4) net debt reduction activities. In addition, we categorize instances when REITs engage in both debt and equity offerings as dual issuances (5). The following filters are applied to include only material financing activities in the sample: The sum involved must, firstly, be larger than US$ 1 million. For equity issue or repurchases, the amount must also constitute more than 1% and 5% of the REIT’s total asset and equity capitalization, respectively. For debt activities, the amount must constitute at least 2% of the REIT’s total assets.11 In other words, a debt issue or reduction is defined as having occurred in a given quarter if the net change in debt, normalized by the book value of assets at the end of the previous period, is greater than 2%. When accounting debt financing activities, we reflect the “net” amount because REITs, like any other corporations, do engage in debt refinancing activities which may not change the capital structure of the companies.12 11
Hovakimian, Opler and Titman (2001), Baker and Wurgler (2002), Frank and Goyal (2003), Huang and Ritter (2004) and Leary and Roberts (2005) adopted similar classifications and filtering criterion. Our filtering criteria for debt issues are less restrictive than equity issues because they are reported as “net” amount as compared to “gross” amount in the case equity issues. 12
For illustration, the cash flow statement of Equity Office Properties Trust recorded that it drew $5.169 billion from its lines of credit, whilst paying back $ 5.987 billion in 2000. In our classification, the financing activities would be categorized as a net reduction in debt amounting to $ 818 million. Note that the distinction between ‘net” and “gross” is less critical for equity financing because seasoned equity offerings and stock repurchases are often made separately. The filtering process also sieves out the small increases in share outstanding due to compensation (such as exercise of options by employees). 17
We also classify proceeds from preferred stocks under debt to be consistent with Baker and Wurgler (2002) and Fama and French (2002). Firms that do not clear the filtering process will be categorized as REITs that did not engaged in any material financing activities in the particular period (0). Following the filtering exercise, we ended up with a sample of 3,230 financing events, comprising 767 equity issuances, 101 equity repurchases, 1,570 net debt issuance, 622 net debt reduction events and 170 incidents where a REIT issues debt and equity securities in the same quarter. The data shows that capital reduction activities constitute a significant proportion of the financing activities of REITs. 6. Estimation Results A likelihood ratio test is first conducted to examine whether each of the independent variables are significantly associated with the dependent variables. The results, which are presented in Table 3, show that all the variables are statistically significant with the exception of NAREIT price index. This is consistent with our earlier observation that aggregate price movement in the REIT sector does not appear to dictate the financing decisions of REITs as much as the aggregate price movement in the broader stock market. The MNL model estimation results are presented in Table 4. Between 17.0% and 19.1% of the log likelihood is explained by the explanatory variables in the MNL model. The reported coefficient estimates compare the likelihood of issuing equity (1), repurchasing equity (2), increasing net debt (3), reducing net debt (4) and issuing both debt and equity capital (5) relative to the likelihood of not executing any material financial activities in that quarter. For example, a significantly positive coefficient estimate in the equity issuance equation would indicate that high values of the variable increase the probability of an equity issue vis‐à‐vis zero financing activities in the same quarter. The sign of the coefficients for 18
the constant show that, all else being equal, net debt increase and dual issuance are the most common fund raising routes for REITs. 6.1
Capital Raising Activities In this section, we focus our discussion on the estimation results for specifications 1, 3, and 5 in Table 4. They reflect the likelihood of an equity issue, a debt issue, and a dual issue, respectively vis‐à‐vis a non‐financing event. Overall, both the equity issuance and debt issuance decisions of REIT exhibit strong market‐timing behavior. Specifically, REITs are more likely to raise funds through equity offerings or debt issues or both, in a bullish stock market and in a low interest rate environment. The negative coefficients observed for the Fama‐
French size factor and the credit spread of corporate bonds suggest that REITs have lower propensity to engage in fund raising activities when the market is more risk averse.13 The occurrence of a dual issuance is also strongly related to the marketing timing variables. REITs are more likely to engage in both debt and equity offerings when the economy is expanding and when their individual stock prices have appreciated considerably in the preceding twelve months. The incident of dual issuance also rises when there is a high appetite for small cap and growth stocks and in a low interest rate environment. Specification 3 shows that REITs are likely to add more debt in their capital structure when the long‐term interest rate and the credit spread of corporate bonds are low. Interestingly, we find that their debt issuance decisions are also sensitive to the equity capital market conditions. Specifically, the propensity for REITs to issue debt increases with a price run‐up experienced by the individual stock and the overall stock market. This suggests 13
Note that a large Fama‐French size factor reflects a market sentiment that requires a bigger return premium for small cap stocks, whilst a high credit spread indicates the market’s preference for lower risk bonds. 19
that REITs do attempt to offset the effects of stock returns on their capital structure. Both the Fama‐French size and growth factors also have significant inverse effect on their propensity to take on more debt. Likewise, the debt capital market conditions also have significant influence on the equity issuance decisions of REITs in Specification 1. Their propensity to issue equity capital increases with current as well as future movements in interest rates. It is not surprising to find that long‐term interest rate has a significant influence on when REITs seek funding from the capital market. Interest rates do not only affect the costs of debt capital but also the valuation of equity shares indirectly through the required rate of return demanded by investors. Furthermore, a stock market run‐up could also reflect more growth opportunities for the firm, and hence, more capital requirements. Overall, these results indicate that REITs opportunistically switch between debt and equity offerings according to their relative costs over time. At the firm level, the price‐earning multiple and individual stock price performance have positive effect on a REIT’s decision to issue equity capital. REITs, therefore, take into consideration sentiments in their own stock as well as that in the broader stock market when making their equity market‐timing decisions. Interestingly, credit rating has a positive relation with the propensity to issue equity capital. This observation coupled with the insignificant coefficient observed for net debt increase indicate that REITs take advantage of better credit rating by issuing new equity rather than increasing their net debt. Note that this does not necessarily mean that an improvement in the credit rating does not preclude the firm from refinancing their existing debt commitments.14 14
No material debt financing activities would have been recorded if the proceeds of the new debt issues, no matter how large, is contra off by an equally huge debt retirement in the same period. In contrast, the net debt of a REIT would decline if the amount of retired debt is more than the amount of new debt issued. 20
6.2
Capital Reduction Activities We do not know what drives REITs decision to reduce debt. Similarly, it is not clear why REITs should engage in stock repurchase? The distribution requirement virtually eliminate the tax advantages of repurchasing shares (instead of paying dividends) to distribute cash. REITs also have less flexibility to substitute repurchases for dividends. Brau and Holmes (2006) observe that REIT managers initiate repurchases when they perceive that the stocks are undervalued. Similarly, Ghosh et al (2007) find that REIT uses repurchase announcements to signal undervaluation.15 Both set of results are not inconsistent with REITs exhibiting market‐timing behavior in their stock repurchases. Specification 2 in Table 4 shows that REITs are more likely to engage in equity repurchase after a prolonged poor performance in the general stock market. Furthermore, REITs are more inclined to repurchase their shares when the market premium for value stocks is low. However, judging by the insignificant coefficients for individual stock valuation metrics, it appears that the repurchase decision of REITs is triggered by a general decline in the market‐wide sentiment rather than the outlook of individual stocks. Furthermore, on the basis that credit default risk is higher during economic recessions, the estimation results indicate that REITs are more likely to repurchase share when the economy is in a downturn. Like Ghosh et al, we also observe a significant positive coefficient on firm size. This is a puzzling result since large firms tend to be widely covered by analyst and hence, suffer less from information asymmetry. Consequently, their need to repurchase shares to signal 15
Meyer and Gentry (2003) and Ghosh et al (2007) are the only two studies we know that has examined the determinants of REITs stock repurchase. Meyer and Gentry find that the ratio of share price to net asset value (NAV) strongly predicts whether managers issue or repurchase shares. They find that managers rarely repurchase shares when price to NAV exceeds unity. Using data of repurchase from 1997 to 1999, Ghosh et al estimate a logit model of the decision to repurchase. Their main focus was on the role of executive stock options on REIT repurchase. They did not include any capital market variables in their model and given the narrow study period (1997‐1999). Both studies did not examine the role of capital market conditions directly in the REITs decision to repurchase. 21
private information should be lower than smaller‐sized firms. Ghosh et al attribute this to the fact that REITs tend to be smaller than industrial firms and small REITs might not have the requisite market depth to support an open market repurchase program.16 We also find a significant positive relationship between firm profitability and the propensity of REITs to engage in stock repurchases. This finding is consistent with the free cash flow argument despite that REITs have to distribute at least 90% of their net earnings. However, consistent with Ghosh et al., we did not find a significant relationship between firm leverage and the likelihood of a repurchase program. Turning to debt reduction activities, we find a significant negative coefficient for term structure of interest rate. This finding indicates that REITs are more likely to retire some of their existing debts in expectation of future rising interest rates. None of the broad equity capital market conditions are significant in explaining net debt reduction decisions of REITs. Nevertheless, firm leverage appears to have a role in REITs net debt reduction decisions. Specifically, highly geared REITs are more probable to engage in net debt reduction activities, suggesting that they may be adjusting to target debt level. 6.3 The Role of Firm Characteristics The literature commonly prescribes that large firms are less susceptible to bankruptcy because they tend to be more diversified than smaller companies (Warner, 1977; Ang and McConnell, 1982). Hence, the trade‐off model of capital structure prescribes that large firms should employ more debt in their capital structure. On the basis that large firms are more transparent and subjected to less information asymmetries, the pecking order theory 16
Dittmar (2000) also found a positive relationship between firm size and repurchase activity and interpret the results as indication that large firms are better able to use share repurchase to take advantage of misevaluations. 22
prescribes that they are less adverse to equity issues. The MNL regression shows a mixed result. We observe that firm size has a negative but insignificant relationship with the firm’s propensity to issue equity. Furthermore, REITs that engaged in stock repurchase activities are, however, on average larger that REITs in the other financial categories. The net effect of a share repurchase, all else being equal, would be an increase in the firm’s leverage. However, we also find that REITs which engaged in net debt increases and dual issues tend to be significantly smaller. This result is consistent with Titman and Wessels (1988) who posit that small firms rely less on equity issues because they face a higher per unit cost. Marsh (1982) also argues that small companies, due to their limited access to the equity capital market, tend to rely heavily on bank loans for their funding requirements. The effect of corporate profitability on the debt‐equity choice has often been used to test which of the alternative theories best describe the financing decisions of firms. A positive relationship would be supportive of the trade‐off model, whilst a negative relationship would be consistent with the pecking order theory (Shyam‐Sunder and Myers, 1999).17 The MNL estimation results show that profitable REITs have lower propensity to engage in fund raising activities, which is consistent with the pecking order theory that profitable firms avoid costly external capital. Corporate profitability, however, have opposite on equity repurchases and net debt reduction activities of REITs. The combined results suggest that highly profitable REITs not only engage in fewer fund raising activities but they are also more likely to engage in capital reduction activities, and in particular, prefer equity repurchase over debt retirement. This has the effect of increasing the leverage ratio of the firm, which is consistent with the trade‐off model of capital structure. 17
Many studies find that higher profitability leads to lower leverage, inconsistent with the trade‐off prediction that more profitable firms should borrow more to reduce tax liabilities (see Huang and Ritter, 2004; p. 6). Note that the market timing theory does not make any predictions regarding the effects of corporate profitability. 23
Whilst leverage does not play any significant role in the fund raising and share repurchases decisions of REITs, it has a strong positive effect on their net debt reduction activities. In particular, we find that highly leveraged REITs are more likely to reduce their debt commitments, rather than issuing new shares. The results show that deviations from a target debt ratio might be a more important factor in debt reduction decisions than in security issuance decisions due to lower adjustment costs.18 7. Conclusion This paper provides empirical evidence on how equity and debt capital market conditions influence the market‐timing and financing activities of REITs. On the basis that the key drivers in the traditional trade‐off and pecking‐order theories of capital structure are mostly circumvented in the case of REIT, we posit that the market‐timing behavior offers a better explanation to the observed financing patterns of REITs over time. Since REITs have little financial slack and depend heavily on external sources of capital, REIT managers have to monitor the debt and equity capital markets closely to take advantage of any temporary windows of mispricing, either at the market‐wide level or at the firm‐level. The results of our analysis, both at the aggregate industry‐level as well as at the firm‐level, support the market‐timing story. We find evidence that REITs exhibit the market‐
timing behavior with respect to their decisions on when to enter the capital market and what type of capital they prefer, as well as their capital reduction decisions. Whilst REITs opportunistically time their financing activities based on capital market conditions, the empirical results also appear to suggest that REITs do have a target capital structure in mind. On the whole, REITs are also more active in the debt market than in the equity market. This 18
Hovakiam, Opler and Titman (2001) find that, all else being equal, the cost of adjustment is generally more expensive for security issues. 24
indicates that are more likely to adjust any imbalance in their capital structure through the debt market, whilst their activities in the equity market are confined to periods when the cost of equity is low. On the success of market timing practices, Baker and Wurgler (2000) observe that firms were able to time their equity issues to coincide with “hot” phases of the market before the stock valuation returns to a more realistic level. They also argue that the firm’s observed capital structure reflects the cumulative outcome of its past attempts to time the equity market. With regards to the efforts to time the debt market, Baker, Greenwood and Wurgler (2003) conclude that it is hard to prove that their efforts actual reduce the overall cost of capital. There is still some disagreement in the general literature on the success and persistence of the effect of market‐timing on capital structure. Future research can therefore explore two follow‐up questions on market‐timing, namely how successful are REIT market‐timing initiatives, and their cumulative effect on the capital structure of the individual REITs over time. We address these two questions in a separate paper. 25
References Ang, J.J. and J. McConnell, 1982, "The Administrative Costs of Corporate Bankruptcy: A Note", Journal of Finance 37, 219‐26. Asquith, P. and D.W. Mullins Jr, 1986, “Equity Issues and Offering Dilution”, Journal of Financial Economics 15, 61‐89. Baker Malcolm, Greenwood Robin, and Wurgler Jeffrey, 2003, “The Maturity of Debt Issues and Predictable Variation in Bond Returns”, Journal of Financial Economics, 70, 261‐291. Baker Malcolm, and Wurgler Jeffrey, 2000, “The Equity Share in New Issues and Aggregate Stock Returns”, Journal of Finance, 55, 2219‐2257. Baker Malcolm and Wurgler Jeffrey, 2002, “Market Timing and Capital Structure”, Journal of Finance, 57, 1‐32. Brau, James C and Andrews Holmes, 2006, “Why Do REITs Repurchase Stocks? Extricating the Effect of Managerial Signaling in Open Market Share Repurchase Announcements”, Journal of Real Estate Research 28(1), 1‐23. Brown, David T., and Riddiough, Timonthy J., 2003, “Financing Choice and Liability Structure of Real Estate Investment Trusts”, Real Estate Economics, 31:3, 313‐346. Boudry, W.I., J.G. Kallberg and C.H. Liu, 2007, “An Analysis of REIT Security Issuance Decisions”, Working Paper. Butler, A.W., G.Grullon and J.P. Weston, 2006, “Can Managers Successfully Time the Maturit of their Debt Issues?” Journal of Finance 56(4), 1731‐1758. Dittmar, A.K. “Stock Repurchase Wave”, Journal of Business 73(3), 331‐355. Dittmar, A. and A. Thakor, 2007, “Why Do Firms Issue Equity?”, Journal of Finance 57(1), 1‐54. Fama F., Eugene and French, R., Kenneth, 1993, “Common Risk Factors in the Returns on Stocks and Bonds”, Journal of Financial Economics, 33, 3‐56. Fama F., Eugene, and French, R., Kenneth, 2002, “Testing Trade‐Off and Pecking Order Predictions about Dividends and Debt”, The Review of Financial Studies, 15, 1‐33. Feng, Zhilan, Chinmoy Ghosh and C.F. Sirmans, 2007, “On the Capital Structure of Real Estate Investment Trusts”, Journal of Real Estate Finance and Economics 34(1), 81‐105. Fischer, E. O., R. Heinkel and J. Zechner, 1989, “Dynamic Capital Structure Choice: Theory and Tests”, Journal of Finance, 44, 19‐40. Frank Murray Z., and Goyal Vidhan K., 2003, “Testing the Pecking Order Theory of Capital Structure”, Journal of Financial Economics, 67, 217‐248. Gentry, William M. and Christopher J. Mayer, 2002, “What Can We Learn about Investment & Capital Structure with Better Measure of q?” Working paper, University of Columbia. Ghosh, Chinmoy, Raja Nag, and C. F. Sirmans, 1997, “Financing Choice by Equity REITs in the 1990s”, Real Estate Finance, 14:3, 41‐50. 26
Ghosh, Chinmoy, Raja Nag, and C. F. Sirmans, 1999, “An Analysis of Seasoned Equity Offerings by Equity REITs (1991‐1995)”, Journal of Real Estate Finance and Economics, 19:3, 175‐192. Ghosh, C., J.Harding, O. Sezer and C.F. Sirmans, 2007, “The Role of Executive Stock Options in REIT Repurchases”, Journal of Real Estate Research (forthcoming). Graham, John R. and Campbell R. Harvey, 2001, “The Theory and Practice of Corporate Finance: Evidence from the Field”, Journal of Financial Economics, 60, 187‐243. Guedes, Jose, and Tim Opler, 1996. “The Determinants of the Maturity of Corporate Debt Issues”, Journal of Finance 51(5), 1809‐1833. Hovakimian, A., T. Opler, and S. Titman, 2001, “The Debt‐Equity Choice”, Journal of Financial and Quantitative Analysis, 36:1, 1‐24. Hovakimian Armen., Hovakimian Gayane., and Tehranian Hassan, 2004, “Determinants of Target Capital Structure: The Case of Dual Debt and Equity Issuers”, Journal of Financial Economics, 71, 517‐
540. Howe, John, and Shilling, James., 1988, “Capital Structure Theory and REIT Security Offering”, Journal of Finance, 43, 983‐993. Huang Rongbing and Ritter, Jay R., 2004, “Testing the Windows‐of‐Opportunity Theory of Capital Structure”, Working Paper, University of Florida. Jaffe, Jeffrey, 1991, “Taxes and the Capital Structure of Partnerships, REITs, and Related Entities”, Journal of Finance, 46, 401‐407. Jung, Kooyul, Yong‐Cheol Kim and Rene M. Stulz, 1996, “Timing, Investment Opportunities, Managerial Discretion, and the Securities Issue Decision”, Journal of Financial Economics, 42, 159‐185. Leary, Mark and Roberts, Michael R., 2005 “Do Firms Rebalance their Capital Structure?” Journal of Finance 60(6), 2575‐2619. Li, Ying, M.F. Ibrahim, S.E. Ong and J.T.L. Ooi, 2006 “Market Timing and REIT Capital Structure Changes”, Working Paper, National University of Singapore. Maris, Brian and Elayan Fayez, 1990, “Capital Structure and the Cost of Capital for Untaxed Firms: The Case of REITs”, AREUEA Journal, 18:1, 22‐29. Marsh, Paul, 1982, “The Choice Between Equity and Debt: An Empirical Study”, Journal of Finance, 37, 121‐144. Masulis, Ronald and A. Korwar, 1986, “Seasoned Equity Offerings: An Empirical Investigation”, Journal of Financial Economics 15, 91‐118. Modigliani, Franco and Merton, Miller, 1958, “The Cost of Capital, Corporation Finance, and the Theory of Investment”, American Economic Review, 48, 261‐297. Myers, Stewart C., 1977, “The Determinants of Corporate Borrowing”, Journal of Financial Econnomics 5, 147‐175. Myers, Stewart C., 1984, “The Capital Structure Puzzle”, Journal of Finance, 39, 575‐592. Myers, Steward C. and Nicholas S. Majluf, 1984, “Corporate Financing and Investment Decisions when Firms Have Information that Investors Do Not Have”, Journal of Financial Economics 13, 187‐221. 27
Ooi, J.T.L., 2000, “Managerial Opportunism and the Capital Structure Decisions of Property Companies”, Journal of Property Investment & Finance 18(3), 316‐331. Rajan, R.G. and L. Zingales, 1995, “What do We Know about Capital Structure? Some Evidence from International Data”, Journal of Finance 50, 1421‐1460. Ritter, Jay R., 2002, “The Windows of Opportunity Theory of Capital Structure”, Working Paper, University of Florida. Shyam‐Sunder, Lakshmi, and Myers, Stewart, C., 1999, “Testing Static Tradeoff against Pecking Order Models of Capital Structure”, Journal of Financial Economics, 51, 219‐244. Stein, Jeremy C., 1996, “Rational Capital Budgeting in an Irrational Market”, Journal of Business, 69, 429‐455. Strebulaev, I.A., 2007, “Do Test of Capital Structure Theory Mean What They Say?”, Journal of Finance 57(4), 1747‐1787. Titman, S. and R. Wessels, 1988, “The Determinants of Capital Structure” Journal of Finance 43, 1‐19. Warner, J., 1977, "Bankruptcy Costs: Some Evidence", Journal of Finance 32, 337‐47. Wu, Zhonghua and Timothy Riddiough, 2005, “Bank Line of Credit, REIT Investment and Banking Relationship”, Working Paper, University of Wisconsin‐Madison. 28
Figure 1: Fund raising exercises by REITs The activities include initial public offerings, secondary equity offerings and public debt issues. Debt securities include both secured and unsecured debt offerings but not private sources of debt and bank loans. Decomposation of Historical U.S REITs Financing
30,000
25,000
20,000
15,000
10,000
Source: NAREITs
IPO
SEO
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
0
1988
US$ Million
5,000
Public Debt
Percentage of US REITs Financing
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
Source: NAREIT
Percentage of IPO
Percentage of SEO
Percentage of Debt
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
0%
29
Figure 2: Relative cost of equity capital and debt capital Real short‐term rate is proxied by the difference between 3‐month T‐bill rate and the inflation during the corresponding period, nominal short‐term rate is the 3‐month T‐bill rate, while long‐
term rate is proxied by 10‐year government bond yield. U.S. REIT and Broader Equity Mkt P/E Ratio
40
35
30
25
20
15
DS U.S. REITs P/E Ratio
Q1 2004
Q3 2003
Q1 2003
Q3 2002
Q1 2002
Q3 2001
Q1 2001
Q3 2000
Q1 2000
Q3 1999
Q1 1999
Q3 1998
Q1 1998
Q3 1997
Q1 1997
Q3 1996
Q1 1996
Q3 1995
Q1 1995
Q3 1994
Q1 1994
Q3 1993
Q1 1993
Q3 1992
Q1 1992
Q3 1991
Q1 1991
Q3 1990
Q1 1990
Q3 1989
Q1 1989
Q3 1988
5
Q1 1988
10
DS U.S. Mkt P/E Ratio
U.S. Interest Rate 1986Q1--2003Q2
12.0
10.0
8.0
4.0
2003Q2
2002Q3
2001Q4
2001Q1
2000Q2
1999Q3
1998Q4
1998Q1
1997Q2
1996Q3
1995Q4
1995Q1
1994Q2
1993Q3
1992Q4
1992Q1
1991Q2
1990Q3
1989Q4
1989Q1
1988Q2
-2.0
1987Q3
0.0
1986Q4
2.0
1986Q1
Percent
6.0
-4.0
-6.0
-8.0
Real Short Term Rate
Nominal Short Term Rate
Long Term Rate
Source: DataStream & Federal Reserve Database 30
Figure 3: Capital market conditions and fund raising activities by REITs The activities include initial public offerings, secondary equity offerings and public debt issues. Debt securities include both secured and unsecured debt offerings but not private sources of debt and bank loans. Equity Offering and Stock Return
50.0
30,000
40.0
25,000
30.0
20.0
10.0
15,000
(%)
Million US$
20,000
0.0
10,000
-10.0
5,000
Source: NAREITs & EIU
IPO
SEO
S&P 500 Return
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
0
-20.0
-30.0
NAREITs E-REITs Index
US REITs Debt Issuance and Bond Yield
16,000
10.00
14,000
9.00
8.00
12,000
6.00
8,000
5.00
6,000
4.00
(%)
Million US$
7.00
10,000
3.00
4,000
2.00
Source: NAREITs & EIU
Unsecured Debt
Secured Debt
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
0.00
1990
0
1989
1.00
1988
2,000
US 10Y Bond Yield
31
Figure 4: Capital Reduction Activities of REITs The right chart tracks the aggregate flow of fund into the REIT sector through private bank loans vis‐à‐vis public debt issues between 1986 and 2003. The left chart shows the share buyback activities of the sector. U.S. REITs Equity Repurchase (1986-2002)
2,500
25
2,000
20
1,500
15
1,000
10
5
fR
0
0
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Amount of REITs Equity Repurchase
No. of REITs Equity Repurchase
REITs Industry Flow of Fund (1986-2003)
25,000
20,000
Millions USD
15,000
10,000
5,000
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
-5,000
1986
0
-10,000
Source: Federal Reserve Flow of Funds
Corporate Bond
Bank Loan
Sources: COMPUSTAT and author compilation. 32
N
Millions USD
500
Table 1. Explanatory variables in the multinomial logistic regression model Explanatory Variable Definition 1) Equity Market Timing Market‐to‐Book Ratio Closing share price divided by book value per share Price‐Earning Ratio Closing share price divided by the 12‐month moving‐average of annualized earning per share Price performance Price appreciation of the REIT stock in the preceding 4 quarters. Price Return of S&P 500 Index E‐REITs Price Return Price appreciation of the common stock market in the preceding 4 quarters Price appreciation of the equity REITs sector in the preceding 4 quarters Fama‐French Size Factor Difference between small and big size equity‐portfolio return Fama‐French Growth Factor Difference between high and low book‐to‐market equity‐portfolio return 2) Debt Market Timing 10‐Year Gov. Bond Yield 10‐Year Government bond yield from Federal Reserve Data Base Real Short Term Interest Rate 3‐month T‐bill rate less inflation rate over the corresponding quarter Term Spread of Gov. Bond Yield Difference between yields of 10‐Year and 1‐Year Gov. bond Credit Spread of Bond Yield Difference between yields of Aaa‐ and Baa‐rated corporate bonds. S&P Long‐Term Debt Rating Binary variable equals 1 if the REIT’s long‐term debt rating by S&P falls within the investment grade, 0 if it falls within non‐investment grade or if it is not rated. 3) Firm‐Specific Variables Firm Profitability Net‐income scaled by total‐asset of a REIT at the end of each quarter Firm Size The natural logarithm of a REIT’s total‐asset at the beginning of the quarter. Total debt divided by total assets at the beginning of each quarter. Leverage Ratio 33
Table 2.Correlation Matrix of Capital Market Variables (1986Q1—2003Q2) SP_R4Q is the S&P500 index price return of previous 4 quarters; NAREIT_R4Q is the NAREIT e‐REITs index price return of previous 4 quarters; FF_SMB is the return for Fama‐French SMB(small‐minus‐big) factor, which is the difference between small and big size equity portfolio return; FF_HML is the return for Fama‐French HML(high‐minus‐low) factor, which is the difference between high and low boon‐to‐market equity portfolio return; GB_10Y is the 10‐year Government bond yield; GB_TS is the term spread of interest rate proxied by the difference between the yields of 10‐
year and 1‐year Government bond yield; REAL_GB_3M is the real short‐term interest rate, proxied by the difference between 3‐month Treasury bill rate and the inflation rate of corresponding quarter; CBS is the credit spread of corporate bond yield, proxied by the difference between the yields of high‐quality (Aaa rated) and high‐yield (Baa rated) U.S. corporate bond. All data are of quarterly frequency from 1986Q1 to 2003Q2. SP_R4Q NAREIT_R4Q FF_SMB FF_HML GB_10Y GB_TS REAL_GB_3M NAREIT_R4Q FF_SMB FF_HML GB_10Y GB_TS REAL_GB_3M CBS 0.243 1.000 ‐0.110 0.118 1.000 ‐0.248 0.364 ‐0.211 1.000 0.244 ‐0.118 ‐0.112 ‐0.071 1.000 ‐0.400 0.210 0.295 0.077 ‐0.163 1.000 0.399 0.037 ‐0.216 ‐0.113 0.269 ‐0.482 1.000 ‐0.302 ‐0.171 0.053 ‐0.047 0.180 0.281 0.038 34
Table 3. Likelihood Ratio Tests of the Multinomial Logistic Model The likelihood ratio test examines whether the independent variables specified in the MNL model is significantly related to the dependent variable. The Chi‐Square statistics is the difference in log‐likelihoods between the general MNL model and a reduced model which exclude the particular factor from the final MNL model. The null hypothesis is that all parameters of that effect are 0. Variables M/B Ratio P/E Ratio Firm Price Return Previous 4Q S&P 500 Return Previous 4Q NAREIT Return Previous 4Q Return for Fama‐French Size Factor Return for Fama‐French Growth Factor 10‐Year Gov. Bond Yield Term Spread of Interest Rate Real Short Term Interest Rate Credit Spread of Corp. Bond Yield Long Term Debt Rating Firm Profitability Firm Size Firm Leverage Chi‐Square Significance 22.345 26.824 37.944 74.543 4.813 40.122 19.048 34.788 30.713 9.634 31.317 14.633 128.213 28.273 38.422 0.000 0.000 0.000 0.000 0.439 0.000 0.002 0.000 0.000 0.086 0.000 0.012 0.000 0.000 0.000 35
Table 4. Estimation Results of Multinomial Logistic Regression This table presents the results of multinomial‐logistic‐regression modeling the probability of the occurrence of a certain type of financing activity in a given quarter. The probability of such financing activity taking place is linked to two groups of explanatory variables reflecting debt and equity capital market conditions, as well as one group of firm‐characteristic controlling variables. Firm‐quarter observations during which no financing activities are observed are taken as baseline scenario. Equity Issuance [Dependent Variable=1] Constant M/B Ratio P/E Ratio Firm Price Return Previous 4Q S&P 500 Return Previous 4Q NAREIT Return Previous 4Q Return for Fama‐French Size Factor Return for Fama‐French Growth Factor 10‐Year Gov. Bond Yield Term Spread of Interest Rate Real Short Term Interest Rate Credit Spread of Corp. Bond Yield Long Term Debt Rating Firm Profitability Firm Size Firm Leverage ‐1.928 0.053 0.013 1.300 3.272 0.151 ‐4.864 1.175 ‐0.160 0.473 0.111 ‐0.007 0.600 ‐7.299 ‐0.039 ‐0.265 ** *** *** *** *** ** *** *** * *** *** Equity Repurchase [Dependent Variable=2] ‐10.259 0.081 ‐0.006 0.928 ‐3.142 1.515 ‐1.060 ‐6.823 0.091 ‐0.893 ‐0.077 0.027 0.224 2.985 0.550 0.393 *** ** ** *** *** * *** Net Debt Increase [Dependent Variable=3] *** 1.258 *** 0.094 0.000 0.370 1.114 ‐0.087 ‐3.632 ‐1.969 ‐0.116 ‐0.015 0.008 ‐0.008 0.064 ‐5.923 ‐0.129 0.099 * *** *** *** *** *** *** *** Net Debt Reduction [Dependent Variable=4] ‐2.821 ‐0.130 ‐0.031 0.946 0.074 ‐0.915 0.164 1.267 0.072 0.168 0.026 ‐0.010 0.021 ‐2.293 0.033 2.394 *** ** *** *** * *** * *** Dual Issuance [Dependent Variable=5] 4.185 0.011 0.004 1.226 5.068 1.358 ‐8.364 ‐3.832 ‐0.604 0.250 0.057 ‐0.018 0.325 ‐23.352 ‐0.493 0.748 Pseudo R‐Square Cox and Snell Nagelkerke 0.170 0.191 *, **, *** denote significance at 10%, 5% and 1% respectively Model Fitting Information McFadden Chi‐Square 0.084 860.476*** Total Observations 4610 *** *** *** *** * *** *** *** ***