2003

V31 3: pp. 313-346

REAL ESTATE ECONOMICS

Financing Choice and Liability Structure of Real Estate Investment Trusts

David T. Brown* and Timothy J. Riddiough" We conduct an analysis of public financial offerings of equity Real Estate Investment Trusts (REITs), with a focus on liability structure effects and whether or not firms target longer-run debt ratios. Our major findings are that (1) proceeds from equity offers are more likely to fund investment, whereas public debt offer proceeds are typically used to reconfigure the liability structure of the firm; (2) public debt issuers are often capital constrained and target total leverage ratios to retain an investment grade credit rating; and (3) the preoffer liability structure affects the issuance choice decision, in that firms with higher preoffer levels of secured (unsecured) debt tend to issue equity (public debt). Other notable findings are that the market for public REIT debt is integrated with the broader debt markets and that higher credit quality firms issue longer-maturing bonds. Capital structure of the firm is one of the most researched topics in all of economics. Yet little work has been done on Real Estate Investment Trusts (REITs) or commercial real estate operating companies.' The previous literature has generally defined capital structure in terms of total debt and equity. A contribution of this paper is that we conduct a detailed examination of the REIT's liability structure, by emphasizing the various types of financial claims that exist and that are issued by firms, rather than focusing exclusively on total debt and equity that exist at a particular point in time. For example, we distinguish between secured debt, public unsecured debt and 'University of Florida, Wardngton College of Business, Gainesville, FL 32611-7168 or david.brown@cba.ufl.edu. ''University of Wisconsin-Madison, Graduate School of Business, Madison, WI53706 or tdddiough@bus.wisc.edu. ' A sample of the relevant research includes Howe and Shilling (1988), who focus on the tax-exempt status of REITs to argue that firms should use little or no debt in the capital structure; Gau and Wang (1990), who examine a sample of income property transactions in Vancouver and find that transaction-based debt levels are sensitive to nondebt tax shields, costs of financial distress, the absolute cost of debt finance and borrower capital constraints; Bradley, Capozza and Seguin (1998), who find interaction between dividend payouts and debt levels; and Capozza and Seguin (1999), who find that the cost of both debt and equity are higher for more diversified (less focused) REITs.

314

Brown and Riddiough

common stock, all of which differ in their seniority as a claim on tirm assets. These finer distinctions provide insight into complex liability structures and the trade offs involved with REIT tinancial management. We are also interested in whether firms actively target debt ratios in their search for an optimal capital structure. Debt ratio targeting, once an accepted part of the corporate finance canon, has lately become a controversial topic. Standard trade-off theories in corporate finance suggest that optimal capital structures exist, and therefore that debt ratio targeting is a productive activity for financial managers.^ In contrast, recent literature suggests that debt ratio targeting is nonexistent or at best a low-priority activity. For example, ShyamSunder and Myers (1999) test the trade-off theory against the pecking-order theory of Myers and Majluf (1984), which suggests that "changes in debt ratios are driven by the need for external funds, not by any attempt to reach an optimal capital structure" (Shyam-Sunder and Myers, p. 221). Shyam-Sunder and Myers present evidence favoring the pecking-order theory over the trade-off theory. Baker and Wurgler (2002) argue that capital structure is the cumulative outcome of previous attempts to opportunistically time tinancial markets, and therefore that debt ratio targeting is of little importance to financial managers. An extreme case is Welch (2002), who presents evidence that observed capital structures cannot be explained by any of the existing theories of value maximization. To address questions of liability structure and debt ratio targeting, this paper provides a systematic empirical analysis of financial policy of equity Real Estate Investment Trusts. We focus on public debt and equity issuances by REITs and relate these issuances to their pre- and postoffer liability structures.-' We structure our empirical analysis around public debt and equity issuances. We do this for three reasons. First, public offerings are major financing events. They occur infrequently and are large in the sense that offering proceeds typically exceed $50 million. Because of their size, public offerings provide important information regarding the firm's postoffer liability structure. Public offerings also depend importantly on the preoffer liability structure of the issuing firm. Bank debt and mortgages are typically issued at much higher frequencies and

^ The simplest form of the trade-off theory is that tax benefits of increased debt are traded off against the direct and indirect costs offinancialdistress. See Brealey and Myers (2000, Chapter 18) for additional background. Also see Shyam-Sunder and Myers (1999) for a cogent review of the relevant literature. ^ By public debt and equity issuances, we mean financial offerings registered with the SEC for broad distribution in the market. Public debt claims are credit-rated unsecured bonds that can be traded with relative ease in a secondary market. Public equity clahns are common stock that is listed on a major exchange. Public debt is generally junior to secured debt in terms of claim priority, but it is senior to common stock.

Financing Choice and Liability Structure of REITs

315

in much smaller increments, and therefore they are less significant "events". Second, certain claims such as preferred stock or privately negotiated equity offerings, which are significant transactions when they occur, are much less important than public debt and equity offers as sources of finance for REITs. And third, data availability is simply better with public as compared to private offerings. Mortgage debt is difficult to identify accurately, and data definition problems plague a systematic analysis of bank debt. In contrast, public debt and equity offering information are standardized and readily accessible. As an initial step in our empirical analysis, a database of public debt offerings made by equity REITs from late 1993 through early 1998 is constructed. A preliminary analysis indicates that the average public debt credit rating is BBB by Standard and Poors, the average maturity of the debt issued is 10.5 years, the average issue size is $133 million and the average offer spread over similar maturity Treasury bonds is 124 basis points. There is evidence that public REIT debt prices and corporate bond prices are cointegrated, and that the credit quality of the issuer and debt maturity are positively related. This latter finding is consistent with predictions of Diamond (1991), who argues that lower credit quality borrowers are often forced to rely on short-term debt. The preliminary evidence indicates that public bond issuers target longer-run debt ratios that result in an investment-grade credit rating. This evidence comes from three sources. First, we document a positive relation between the bond's offer spread and debt maturity, which is consistent with predictions of the target debt ratio model of CoUin-Dufrense and Goldstein (2001). Second, we find that the relation between the bond's offer spread and issuer credit quality is nonlineara change in classification from an investment grade to a noninvestment grade credit rating results in almost a 1 % jump in bond yield. It is therefore extremely expensive for REITs to issue junk bonds. Third, we find that REITs that issue public debt have credit ratings that cluster just above the minimum investment-grade credit rating. To more fully analyze REIT liability structure and debt ratio targeting, we augment the public bond database to include equity offers. The stated use of offer proceeds is considered first. Public debt offer proceeds are most often used to buy back bank and other senior secured debt, rather than fund new investment. Equity offer proceeds are also used to buy back senior debt, but are more often used to fund investment. When a REIT issues either public debt or equity to fund investment, average issue size exceeds that of other use-ofproceeds categories. This suggests that it is easier to access the capital markets when investment opportunities exist than when proceeds are used to reconfigure the firm's liability structure.

316

Brown and Riddiough

Total leverage ratios tend to rise for firms that issue debt, whereas they tend to decline for firms that issue equity. Postoffering secured debt levels as a percentage of total assets are found to decline for both public debt and equity issuers. Analysis shows that, in the year of security issuance, the total leverage ratio of public debt issuers regresses toward that which produces a minimum investment-grade credit rating. This finding complements earlier evidence, suggesting that public debt issuers target a longer-run debt ratio and that targeting is driven by a desire to maintain a minimum investment-grade credit rating. There is less compelling evidence that equity issuers target debt ratios, at least based on data in the year of security issuance. To further examine the determinants of financing choice, we estimate a probit model of the likelihood that a firm issues debt, as opposed to equity, conditional on the firm raising capital from the public markets. The most important finding relates to liability structure at the time of the security offering. Firms with higher total debt levels are less likely to issue debt. However, the composition of debt in the liability structure is more predictive than the total level of debt. Firms with a higher preoffer ratio of senior secured debt are more likely to issue common stock, while firms with a higher preoffer ratio of junior unsecured debt are more likely to issue public debt. REITs with large amounts of secured debt are therefore unable to, or choose not to, use public debt financing. Furthermore, REITs that have never issued public debt are unlikely to issue public debt in the near future. The evidence therefore suggests a bifurcation in REIT liability structures: Some REITs primarily use secured (bank and mortgage) debt and equity to finance themselves, whereas other REITs choose a more complex capital structure with various types of secured and unsecured claims. The main body of the paper is organized as follows. The next section provides a detailed analysis of the public debt issue characteristics, including an examination of factors that influence offer spreads and debt maturity. A comprehensive analysis of debt-equity issuance choice is then provided. Stated use of proceeds, changes in financial leverage and the determinants of issuance choice are examined and discussed. Concluding remarks can be found in a final section. Appendices provide a significant amount of supporting detail onfinancingtransactions. Preliminaries: Public Debt Issue Characteristics Data and Summary Statisties Relative to public equity and even private mortgage debt, little is know about the issue characteristics of public REIT debt. To learn more about issue

Financing Choice and Liability Structure of REiTs

317

characteristics, in this section we examine a sample of 174 fixed-rate public debt issuances by non-health care equity REITs occurring between September 1993 and March 1998. Offerings are identified from Morgan Stanley Dean Witter's Equity REIT Monthly Statistical Review. The data set includes all public debt offerings made during the sample period, with the exception of six fioating rate offers (which lack offer spread information). The maturity, credit rating and offer spread over similar-maturity Treasury securities are obtained for each issue as well as information about whether the debt is callable or putable. Total proceeds from these offerings equal $16.1 billion. In a number of instances, a firm issued more than one bond on the same date. Multiple-issue bonds typically varied by maturity date and offer spread, but are otherwise identical. To distinguish between individual bond offerings and multiple bond offerings made by a particular firm at a particular time, we denote the firm-level issuance as an offering event. There were a total of 120 offering events resulting in 174 individual public bond issuances. Of these 120 offering events, 38 were multiple-debt issuances, implying that 2.4 bonds were issued on average in a multiple-issue offering. In the case of offering events with multiple bond issuances, the maturity and offer spread is determined by weighting maturities and offer spreads of the individual bonds by their face values.'' Summary statistics for the sample of 120 public debt offering events are shown in Table 1. The majority of issuances are straight fixed-rate coupon bonds. Nine of the bonds include a provision that allows the issuer to call the bonds. These are make-whole call provisions that allow redemption at a price equal to the present value of the remaining coupons, discounted at a yield of between 15 and 25 basis points over the prevailing Treasury yield. Such provisions limit the gains of bondholders if they attempt to hold out when the issuer tries to buy back its bonds in the open market (see Mann and Powers (2001) for a discussion of make-whole call provisions). Six of the bonds in the sample are "mandatory par put remarketed securities," or MOPPERS. These bonds are both callable and putable at par on a particular remarket date prior to legal maturity. From an investor's and issuer's perspective, the maturity of MOPPERS is the remarket date (which we use as the maturity date for these bonds).

'' A review of financial covenants reveals little variation across investment grade issues, so we do not to include these data. Nearly all offerings stipulate four financial ratio tests that must be met: (1) the ratio of total debt to undepreciated total assets does not exceed 60%, (2) secured debt does not exceed 40% of undepreciated total assets, (3) the interest coverage ratio must exceed 1.5 and (4) total unencumbered undepreciated assets must exceed 150% of unsecured debt outstanding.

318

Brown and Riddiough

Table 1 Summary statistics for the sample of public debt offers. Size (Millions) Mean Median Maximum 4th quintile 1st quintile Minimum Std. Dev. $133 $110 $1,500 $150 $75 $10 $142 Maturity (Years) 10.6 10.0 30.0 10.2 7.0 4.3 5.7 Offer Spread (Basis Points) 124 116 415 145 93 40 51 Rating" (Moody's) 10.1 10.0 17.0 II.O 9.0 5.0 1.3 Rating"" (S&P) 10.1 10.0 17.0 11.0 9.0 6.0 1.3

" Maximum, average and minimum expressed as numerical values. Moody's ratings correspond to 5 = Aa3, 10 = Baa2, 17 = B2. '' Maximum, average and minimum expressed as numerical values. S&P ratings correspond to 6 = A4-, 10 = BBB, 17 = B - . The table provides summary statistics for a sample of 120 public debt offering events by non-health care REITs from September 1993 through March 1998. Thirty-eight offering events had multiple bonds with different maturities and offer spreads. SIZE of offering is total amount of debt issued, MATURITY is weighted-average maturity of the bonds and OFFER SPREAD is weighted-average offer yield minus yield on comparable maturity Treasuries. Moody's and S&P bond ratings are numerical values between 1 (Aaa-l-/AAA-f) and 23 (D/D) assigned to each offering. Appendix A provides a conversion from alphabetic to numeric ratings.

The average issue size is $133 million. Issues range in size from a $10 million offering by Associated Estates Realty in November 1995 to a $ 1.5 billion offering by Equity Office Property in February 1998. The average maturity is 10.6 years, where maturities vary between 4.3 and 30 years. Seventy-eight percent of the bond offering events have maturities between 5 and 10 years. The average offer spread is 124 basis points, with a range of 40 to 415 basis points. Bond ratings are assigned numerical codes from 1 (Aaa+ rated by Moody's and AAA-I- rated by S&P) to 23 (D rated by Moody's and S&P). The average bond rating is 10.1 by Moody's (Baa2 = 10) and S&P (BBB = 10). The highest-rated issuer in the sample is Summit Properties, rated Aa3 by Moody's and A-f- by S&P. The lowest rated issuer in the sample is Saul Centers with a rating of B2 by Moody's and B - by S&P. See Appendix A for detail on the assignment of numeric values to bond ratings. Analysis ofOjfer Spreads We will now examine the effects of maturity and credit market conditions on the offer spread of newly issued public REIT debt. Longstaff and Schwartz (1995) and Leland and Toft (1996) examine the debt maturity offer spread

Financing Choice and Liability Structure of REITs 319

relation by modeling stochastic asset values as the cause of bond default. In their models default occurs when asset value falls below a fixed threshold value, which predicts that credit spreads will increase with maturity for higher credit quality firms and decline with maturity for lower credit quality firms. Credit spreads are predicted to decline with maturity for lower-rated bonds because the credit quality of the representative issuer is expected to increase significantly over time when debt is fixed in the capital structure. In contrast, Collin-Dufrense and Goldstein (2001) incorporate the assumption that the default boundary is not fixed, in the sense that firms choose a target leverage ratio. They assume that when asset values rise, firms tend to raise their debt levels (or when asset values fall, firms reduce their debt levels) so as to adjust toward a longer-run target leverage ratio. Such a model generates credit spreads that unambiguously increase with maturity for lower-rated firms, as firms raise their debt levels in response to an increase in value as they move toward their target. Given that the directional relation between the offer spread and maturity depends on the market's assessment of future capital structure policy, it is interesting to examine the relafion in the case of public debt issued by REITs.^ If credit market participants assess that REITs issue debt when they are aggressively leveraged, and if they anticipate that REIT balance sheets will strengthen in the future, then credit spreads are predicted to decline with maturity. If, however, credit market participants assess that REITs issue public debt at their longterm target leverage ratios, then credit spreads are predicted to increase with maturity. The second issue we examine is whether general market conditions affect REIT borrowing costs. To structure the analysis, we recognize that there are three components to unsecured debt credit spreads: (1) the default risk premium, which compensates for expected losses in default; (2) the market risk premium, which compensates for the undiversifiable risk of uncertain defaults; and (3) the liquidity risk premium, which compensates for the fact that corporate bonds are less liquid than Treasury bonds.* If rating agencies are quick to incorporate information into their ratings of new issues, one would expect a positive relation between market credit spread

' Previous empirical examinations of the relation between credit spread and maturity do not focus on specific industries (see He, Hu and Lang 2001 and Helwege and Turner 1999). ^ See Brown (2001) for a detailed discussion and empirical analysis of the three components of corporate bond spreads.

320

Brown and Riddiough

and the offer spread on market and liquidity risk portance of time-varying Fama (1986), Keim and Brown (2001).

same-rated new issues only if changing prices of cause credit spread changes.^ Evidence on the immarket and liquidity risk premiums is provided by Stambaugh (1986), Huang and Huang (2000) and

We examine these issues by regressing the REIT bond offer spread against the maturity of the issue, market credit spreads and control variables. We control for the assessed credit quality of the public debt by including Moody's rating of the issue. RATING is a numerical value attached to each credit rating, where a higher value corresponds to lower credit quality issues (see Appendix A). An ordinal credit rating measure implicitly assumes a linear relation between the credit ratings and expected default losses. Moody's historical default rate and default recovery analysis supports this assumption for investment-grade bonds. Expected default losses are dramatically higher for below-investmentgrade issues, however. We capture the nonlinear relation between credit rating and expected default loss by including the indicator variable, JUNK, which takes on a value of one for all new issues rated below BBB. We also control for offering scale effects, where SIZE is the value of the issue in millions of dollars. To the extent that larger issues tend to be more liquid, and liquidity is valued, a negative relation between the offer spread and SIZE is expected. The estimated model is reported in Table 2.^ The coefficients on the RATING and JUNK variables provide evidence of a piecewise-linear relation between credit rating and offer spread. Estimates indicate that a one-notch decline in credit rating results in a 20 basis point increase in the offer yield when the bond

' When the time series variation of market spreads is caused by changing default loss expectations, there will be a relation between market credit spread and offer spread on same-rated new issues only if the rating agencies do not incorporate new information about market conditions into ratings of new issues. Although rating agencies have been criticized for lags in rerating previously issued bonds, they are relatively quick to react to information at the time of new debt security issuances, * The model is estimated using all 174 separately issued bonds. Using the full sample of bonds introduces the potential that the observations are not independent, since several of the observations come from the same issuer on the same date. We address this concern in two ways. First, we examine the model residuals from the observations that come from the same issuer, and we find that they are not correlated. Second, we estimate the models using 120 issuance event observations. Similar results are obtained using the full sample and the smaller sample of 120 issue events.

Financing Choice and Liability Structure of REITs 321

Table 2 Determinants of the offer spread. Variable INTERCEPT RATING JUNK CREDIT SPREAD SIZE MATURITY N R2 Coefficient -290.49" (25.59) 19.94" (1.79) 97.15" (11.32) 1 90" (0.14) -0.096" (0.034) 2.06" (0.40) 174 0.722

The table provides coefficient estimates for the determinants of the offer spread for a sample of 174 fixed-rate public debt issues. The dependent variable, OFFER SPREAD, is the offer yield minus the yield on a comparable-maturity Treasury security, in basis points. RATING is a numerical value attached to the Moody's bond ratings between I (AAA-f) and 23 (D; see Appendix A); JUNK is a dummy variable that takes a value of 1 if the bond is rated below investment grade and 0 otherwise; CREDIT SPREAD is the difference between the yield on Baa rated corporate bonds with maturity between 7 and 10 years and the yield on comparable Treasury bonds at the time of bond issuance, in basis points; SIZE of the issue is the total amount of debt issued on the offer date, in millions of dollars and MATURITY is the maturity of the issued bonds, in years. Standard errors are reported in parentheses. " Indicates the estimated coefficient is significantly different from zero at the 1% level.

remains in either the investment or non-investment-grade category range. When credit quality changes from investment grade to non-investment grade, however, estimates indicate a jump in the offer yield of almost 1%. This is a substantial increase in the cost of capital, which may explain why investment-grade credit ratings are coveted by publicly traded firms. We find a statistically significant positive relation between market credit spread and the offer spread, and a statistically significant negative relation between the size of the issuance and the offer spread. Presuming that the credit rating is an accurate proxy for default risk and that size proxies for liquidity of the bond, the credit spread coefficient of 1.90 indicates that the public REIT debt market is integrated with the broader corporate bond market and that it is highly sensitive to changes in market risk. Exactly why public REIT debt is so sensitive to changes in credit spreads is not obvious, however. The coefficient on the MATURITY variable is positive and statistically significant. The estimate indicates that extending the maturity of public REIT debt by one year increases the offer spread by two basis points. Since we have controlled for credit quahty through the bond rating, our interpretation of this result is that it is consistent with the target debt ratio approach of Collin-Dufrense and

322

Brown and Riddiough

Goldstein (2001), which predicts an unambiguously positive relation between bond maturity and offer spread.' Debt Maturity Structure To examine whether issuer credit quality and market conditions are related to bond maturity, we specify bond maturity as a function of the REIT's credit rating and financial market conditions. The results are reported in Equation (1):' MATURITY = 26.6 - 0.05 * CREDIT SPREAD - 1 , 1 6 * RATING (5.07)** (0.029) N = 174, /?2 = .26 (0.35)** (1)

We find a statistically insignificant relation between maturity and the credit spread, but a statistically significant negative relation between maturity and the firm's credit rating. The estimated coefficient is very close to one improving the credit rating one notch results in one year longer debt maturity. To more clearly highlight the credit rating-bond maturity relation. Table 3 provides a breakdown of issue maturity by bond rating category. Other than the one B rated bond, there is a monotonic relation between credit rating and debt maturity. Below-investment-grade (BB and BB-I-) borrowers issue bonds with an average maturity of less than 7 years. The average maturity of bonds increases to 9.1 years for B B B - rated borrowers and then to 12.7 years for

' Unobserved variation in firm credit quality could have contaminated the offer spreadmaturity relation (e.g., higher quality firms in the same rating category might issue shorter-term debt at a lower offer yield). To examine this, we followed the approach of Helwege and Turner (1999) to exploit the fact that numerous firms in our sample issue bonds of more than one maturity at a given date. Using the existing database of 174 fixed-rate coupon bond issuances, we created a subsample of 54 intra-issue offer spread-maturity combinations. We found that the offer spread on the longer term bond is larger than the offer spread on the shorter term bond for each of the 54 observations. We also estimated a model of the incremental offer spread as a function of the bond rating and the increment in debt maturity. The estimates suggest that a one-year increment in debt maturity increases the offer spread by 2.41 basis points, which is similar to the results presented in Table 2, There is also a statistically significant positive relation between the incremental offer spread and the issuer's bond rating, implying that the term structure of credit spreads is more steeply sloped for lower credit-rated issuers. The finding that credit spreads increase more quickly with maturity for lower-credit-rated issuers is more consistent with credit spread models that incorporate target debt ratios than models that assume a fixed debt level, '" We also ran the regression with A = 120 offer events, where maturity was a weighted ^ average in the case of multiple bond offerings. The results are nearly identical.

Financing Choice and Liability Structure of REITs 323

Table 3 Bond maturity by issue credit rating. Bond Rating BAverage maturity Number of issues 10,0 1 BB 6,7 4 BB-f 6,8 6 BBB9,1 45 BBB 12,3 35 BBB-I12,7 20 A 14,1 9

The table categorizes debt maturity by Standard & Poor's bond rating category. N = 120 bond offer events. Maturity is measured in years.

BBB-I- rated borrowers. The average maturity of the bonds issued by A rated borrowers is 14.1 years." Our finding of a monotonic relation for firms in the low to moderate credit quality range is consistent with predictions of Diamond (1991), who argues that lower credit quality firms borrow using short-term debt and in the extreme may be rationed out of the public capital markets. Note that firms cluster just above the investment-grade rating cutoff of BBB. Sixty-seven percent of the firms in our sample are rated either BBB or BBB, and 83.3% of thefirmsfall into the B B B - to BBB+ rating range. This clustering suggests that REITs that issue public debt may intentionally target total debt levels to obtain credit ratings just above the investment-grade cutoff point. The coefficient on the JUNK dummy variable in the offer spread regression reported in Table 2 provides additional credence to investment-grade debt targeting, as results indicate that a below-investment-grade rating causes almost a 1% increase in the cost of debt capital for REITs, Target Debt Ratios and the Choice Between Public Debt and Equity Issuance This section builds on prior results along two dimensions, Eirst, we extend previous analysis on debt ratio targeting by examining the use of issuance proceeds and how the liability structure of the firm changes during the year in which an offering occurs. Second, we attempt to develop a more detailed understanding of the liability structure of REITs. We focus specifically on alternative types of debt and equity claims that REITs use to finance themselves, as well as whether financial claims are publicly issued or privately placed.

" Also see Guedes and Opier (1996) and Stohs and Mauer (1996) for evidence on the relation between credit quality and debt maturity for industrial firms.

324

Brown and Riddiough

Our modus operandi will be to conduct a side-by-side analysis of public debt and equity offers. Use of offer proceeds, the preoffer liability structure and changes in the seniority and total amount of liabilities in the year of security issuance are compared and contrasted. A direct comparison of debt and equity offers highlights the causes and consequences of issuance choice, which in turn provides insight into debt ratio targeting and liability structure effects. The first step in our analysis is to augment the data to include equity offerings. The sample for analysis includes public debt and equity offers occurring between September 1993 and December 1997. We restrict the analysis to firms issuing only debt or equity in the same year, but not both. The sample includes only offers for which year-end prior and year-end after balance sheet data are available from SNL Financial's SNL REIT Database. Ending the sample period in 1997 reduces the number of public debt offering events from 120 to 95, and missing SNL data further reduces the sample to 73. Two additional observations are eliminated because of suspect financial data. A final sample size of 40 public debt offer events is obtained when we eliminate 31 bond offers made in the same year as an equity offer. A similar screening process results in a final sample of 140 equity offers. Thus, our final sample consists of 40 public debt and 140 common stock offerings occurring between September 1993 and year-end 1997. Use of Offer Proceeds and Changes in Eiability Structure We now consider the announced use of public debt and equity offer proceeds. Classification of offer proceeds is based on news stories obtained from the Dow Jones News Retrieval Service. We describe the use of offer proceeds for both public debt and common stock offers as falling into one of four categories: (1) repay debt, (2) repay debt and fund investment, (3) fund investment and (4) no stated use. Table 4 displays the categorized results, and also reports average offer size. In a large number of debt and equity offers, the financing proceeds are used to repay debt. When the type of debt repurchased is specified, it is typically bank debt. This is consistent with the idea that REITs largely fund investment with bank lines of credit and other sources of private debt. When these lines of credit are exhausted, REITs access the public capital markets and use the issue proceeds to pay down credit lines in order to prepare for the next round of financings. The detailed financial information on selected issuing firms in Appendix B strongly supports this point. Borrowings against lines of credit are often very large prior to issuance, while similariy large amounts are repaid once the issuance occurs.

Financing Choice and Liability Structure of REITs

325

Table 4 Stated use of proceeds as a result of public debt and common stock issuances. Public Debt Stated Use of Proceeds Repay debt Repay debt and fund investment Fund investment No stated use Total N 17 16 1 6 40 Av. Size $123.9 $125.6 $200.0 $156.3 $133.3 Common Stock N 25 48 15 52 140 Av. Size $80.8 $76.6 $140.9 $91.6 $89.1

The table classifies the stated use of proceeds from public debt and common stock issuances to either: (1) repay debt, (2) repay debt and fund investment, (3) fund investment or (4) no stated use of proceeds. The table also displays the average size of the issuance for each category in millions of dollars. Issuance data include 40 public debt offers and 140 common stock offers by non-health care equity REITs between September 1993 and December 1997. Stated use of proceeds data are obtained from the Dow Jones News Retrieval Service.

For public debt offers, a clear use of proceeds is determined in 34 of the 40 cases. In 17 of 34 offers (50%) for which there was a clear use of proceeds, the stated use was to retire debt. In 16 of 34 offers (47%), the stated use of proceeds was to repay debt and fund investment in new property or improvements to existing property. In only 1 of 34 cases (3%) was the offer proceeds used only to fund investment. Given that public debt offer issue proceeds are largely used to retire existing debt, as opposed to funding new investment, these offers are unlikely to have a great effect on total financial leverage. At the same time, public debt offers will typically change the character of the firm's liability structure by extending the average maturity of debt outstanding and/or reducing the amount of secured debt outstanding. Case study evidence presented in Appendix B supports these conclusions. Proceeds from public debt offerings are used to pay down secured debt in all five of the cases analyzed. Mortgage loans are often repaid to result in a net reduction of mortgage debt outstanding. Most of these repaid mortgage loans have maturities of 5 years or less, whereas newly issued unsecured debt has maturities of 5 years or more (10 years on average). Consequently, the average maturity of debt outstanding lengthens as a result of the public debt issuance. As previously noted, offer proceeds are also often used to pay down bank lines of credit, which are typically one-year renewable facilities. However, bank lines are often either reconfigured or redrawn by year-end so that the net effect varies across firms.

326

Brown and Riddiough

For example, TriNet Corporate Realty paid down both mortgage debt and bank lines with its unsecured debt issuance, but then it ended the year (1996) with more bank line borrowings than repayments. For the sample of common stock offers, the use of proceeds is clearly described in 88 of the 140 offers. In 25 of the 88 offers (28%) for which there is a clear use of proceeds, the stated use was to repay debt. In 48 of 88 offers (55%), the issue proceeds were used to repay debt and fund investment. Finally, in 15 of 88 offers (17%), the proceeds were used to fund investment. Although proceeds are often used to repay debt, common stock offers are far more likely to fund new investment. Using equity to fund new investment, especially when accompanied with the use of bank lines and mortgage debt, as opposed to paying down debt, suggests that total leverage levels do not decline drastically as the result of an issuance. Reference to detailed case study evidence in Appendix B again supports these conclusions. Significant increases in asset book values imply that all of the five firms analyzed made acquisitions in the year of common stock issuance. Most also issued mortgage debt to partially fund their acquisitions, implying that total leverage ratios did not drop significantly in the year of equity issuance. Indeed, only Arden Realty experienced a significant decline in its total leverage ratio. Finally, observe in Table 4 that security issue size is considerably larger when proceeds are used to fund investment. This suggests that it is easier to access the public capital markets when there are investment opportunities than when proceeds are used to reconfigure the firm's liability structure. To further examine the effects of public debt and equity issuance on REITs' liability structure, we calculate total debt and senior secured debt as a percentage of the book value of assets. Summary statistics for these ratios are reported in Table 5. Total leverage ratios at the year-end prior to the offer reveal that common stock issuers are more leveraged than public debt issuers, but the difference is statistically insignificant. However, public debt issuers have significantly less senior secured debt on their balance sheets prior to issuance. Thus, while there is an insignificant difference in total leverage, common stock issuers rely much more on secured debt financing in the form of bank loans and mortgages. Another way to characterize this difference is to observe that the mean preoffer ratio of junior debt to total assets is 6.9% for equity issuers versus 16.5% for public debt issuers.

Financing Choice and Liability Structure of REiTs 327

Table 5 Changes in liability structure during the year of public debt and equity issues. Year-end Prior to Offer Panel A: Total Leverage Ratio Public debt issuers Mean Median Standard deviation Common stock issuers Mean Median Standard deviation Panel B: Secured Debt Ratio Public debt issuers Mean Median Standard deviation Common stock issuers Mean Median Standard deviation 23.61% 18.40% 19.79% 34.85%** 36.19% 20.17% 15.40% 12.72% 12.67% 30.15%** 30.16% 18.10% -8.21% -4.70% 11.42% -4.70% -3.72% 11.06% 40.12% 43.37% 15.37% 41.79% 43.05% 17.73% 45.46% 44.80% 10.07% 39.09%' 39.43% 15.40% 5.34% 4.56% 4.15% -2.69%** -2.66% 10.37% Year-end of Offer Year Change During the Year

The table reports summary statistics for variables that describe the liability structure of public debt and common stock issuers. The sample includes 40 firms that issued public debt but did not issue common stock and 140 firms that issued common stock but did not issue public debt in the same year. The total leverage ratio is total debt divided by the book value of assets, and the secured debt ratio is secured debt divided by the book value of assets. These ratios are presented for the year-end prior to the year of the offer and the year-end of the year of the offer. The mean, median and standard deviation are reported for each variable. We compare total leverage and secured debt ratio means of public debt and common stock issuers at year-end prior to the offer, year-end of the offer year and the change during the year in order to determine whether differences are statistically significant. * Denotes mean difference between public debt and equity issuers significant at the 5% level, and ** denotes mean difference significant at the 1% level.

Interestingly, in a separate analysis we tind that 85 of the 140 common stock issuers have no unsecured debt on their balance sheets at the start of the year that they issue common stock. These relations suggest that causality between the REIT's liability structure and issuance choice is complex: The current liability structure is the cumulative outcome of previous security issuance, which in turn influences the current security issuance policy.

328

Brown and Riddiough

The finding that firms with large amounts of senior secured debt are less likely to issue unsecured debt is consistent with predictions of Bulow and Shoven (1978) and Stulz and Johnson (1985). They emphasize that conflicts and coalitions among existing claimholders affect incentives to issue subsequent claims on the firm's assets.'^ Credit rating agency analysis of REIT debt reflects the same idea. According to Moody's Investors Service, public credit ratings are affected by both the borrower's total financial leverage and the extent of secured debt financing, in that secured debt financing reduces the amount of "unencumbered assets" and "cash flows from unencumbered assets" available to unsecured creditors (see Moody's Investors Services 2002). Rating agency concern over secured debt is highlighted in the Moody's report by a clear negative relation between the ratio of secured debt to total debt (a measure of the composition of the borrowings) and the issuer's public credit rating,'^ The year-end balance sheets of public debt and equity issuers are consistent with our analysis of the use of proceeds. The leverage increase for public debt issuers is relatively modest: The mean leverage ratio increases from 40.1 % to 45.5% while the median leverage ratio increases from 43.3% to 44.8%. The major balance sheet impact of public debt issuance is to reduce the reliance on secured debt financing, where the mean ratio of secured debt to total assets falls from 23.6% to 15.4%. By year-end of the offer year, equity issuers have significantly lower leverage ratios than public debt issuers, although the decline in average total leverage is relatively modest, going from 41.8% to 39.1%. This is coupled with a decline in secured debt levels of more than four percentage points on average.

'^ To see this, consider two REITs with identical financial leverage and identical assets, one whose debt is entirely secured and the other whose debt is entirely unsecured. The relative cost of issuing equity across these two firms depends on the extent to which the equity coinsures the outstanding debt. Equity financing is relatively attractive for the firm with secured debt because the coinsurance of the debt as a result of the equity issue is less than for the firm with unsecured debt. In comparison, unsecured debt is relatively more costly to issue for the firm with entirely secured debt. When there are three sets of claimholders with differing priority and maturity structures (equity, long-term unsecured debt, short-term secured debt), agency problems grow considerably, and it is usually the case that unsecured debtholders fare most poorly in the case of financial distress. These problems will be anticipated, and will result in a higher cost of capital for the firm that issues unsecured debt with all secured debt in its preoffer liability structure. '* For example, the discussion provided in the 1 OK report for Oasis Residential for 1997 (the year the firm issued public debt) points out that after the proceeds from the issuance of unsecured debt were used to repay secured debt balances, the company was able to generate 64% of its EBITDA from unencumbered properties (see Appendix B),

Financing Choice and Liability Structure of REiTs

329

What accounts for the increase in junior unsecured debt in the year of an equity issuance? Further analysis reveals that firms that issue equity during our sample period sometimes convert secured bank lines to unsecured lines. This outcome is detailed in Appendix B, where we note three cases of equity-issuing firms that convert bank line priority during the offering year. There are also unreported cases in which preferred stock or privately placed unsecured debt is issued during the year of a common stock issuance. Additional Evidence of Debt Ratio Targeting We have presented evidence that public debt issuers target a longer-run total leverage ratio and that an investment grade credit rating has significant value. Also recall that most REITs that issue public debt are at or just above the minimum investment-grade credit rating. These findings suggest that public debt issuers are debt capital constrained and that they pay close attention to managing their debt levels. To further pursue this issue, in Figures 1 and 2 we plot total leverage ratios at year-beginning versus year-end for public debt and equity issues. The vertical and horizontal lines bracket the 35 to 55% total leverage ratio range, which, according to Moody's Investors Service (1998), corresponds to the range in which most investment-grade rated REITs exist. Notice that public debt issuers that begin the year within the target range typically remain in that range (see Figure 1). Only 3 REITs leave the target leverage range out of 20 that begin the year in that range, and all 3 end the year with leverage ratios above 55%. Ten of 15 issuers that begin the year with total leverage ratios of less than 35% end the year with leverage ratios in the target range. Only 5 REITs issue public debt with year-beginning leverage ratios in excess of 55%, and 2 of those firms are able to finish the year in the target range. The presence of only 5 high-leverage issuers in the data implies that REITs generally cannot or choose not to issue public debt when leverage ratios are particularly high. When the year-end total leverage ratio of public debt issuers is regressed against the year-beginning ratio (shown by the heavy dashed line), we observe significant reversion toward a total debt ratio range of 35 to 55%. This line shows that issuers with year-beginning leverage levels of between 21 and 58% are predicted to end the year within the target range of 35 to 55%. These results suggest that aggressively leveraged public debt issuers are constrained in their ability to undertake leverage-increasing activities. Borrowing right up to the investment-grade credit rating leverage limit would appear not

330 Brown and Riddiough

Figure 1 Total leverage ratios at year-beginning and year-end: public debt issuances. The figure plots the total leverage ratio at the beginning of the year versus the total leverage ratio at the end of the year for firms that issue public debt in a given year. Vertical and horizontal dotted lines are drawn at the 35 and 55% leverage ratio points. REITs within the 35-55% leverage ratio range are often rated BBB-)- (Baal), BBB (Baa2) or BBB- (Baa3) by S&P (Moody's). The 45 line is drawn to help distinguish firms that increase or decrease their leverage ratios in the issuance year. The heavy dashed line is fitted by regressing the year-end leverage ratio against the year-beginning leverage ratio.

0.6-

O (P

0.4UJ
^^ o -'

'O

QO 0

0.2-

9-'

no0.0

0.2

0.4 BEGIN

0.6

0.8

only to constrain actions in the year of the financing, but may also limit a firm's ability to address unanticipated adverse changes in market conditions going forward. These facts also explain why public debt offer proceeds are often used to repay debt: Many REITs have little room to take on additional leverage without significantly increasing their cost of debt capital. Relations are less clear-cut in the case of equity issuers (see Figure 2). There are 44 REITs below, 66 within and 30 above the target range at year-beginning. Seventeen firms that begin the year within the target range are outside the range at year-end, and 15 of those firms end the year with total leverage ratios below 35%. A total of 23 firms that begin the year outside the target range end the year within the range. Unlike public debt issuers, a significant proportion of REITs that issue equity are highly leveraged (30 of 140), and they remain so

Financing Choice and Liability Structure of REiTs

331

Figure 2 Total leverage ratios at year-beginning and year-end: equity issuances. The figure plots the total leverage ratio at the beginning of the year versus the total leverage ratio at the end of the year for firms that issue equity in a given year. Vertical and horizontal dotted lines are drawn at the 35 and 55% leverage ratio points, REITs within the 35-55% leverage ratio range are often rated BBB-I- (Baal), BBB (Baa2) or B B B (Baa3) by S&P (Moody's), The 45 line is drawn to help distinguish firms that increase or decrease their leverage ratios in the issuance year. The heavy dashed line is fitted by regressing the year-end leverage ratio against the year-beginning leverage ratio.

0,80,6Q LU

o //5'i
0,40,2000,0
y^'''oo Oo

0,2

0,4

0,6

0,8

1,0

BEGIN

at year-end (16 of 30 retain leverage ratios above the 55% level). The dashed regression line in Figure 2 indicates that reversion to the target leverage ratio range is significantly weaker than in the case of public debt issuers. These findings indicate that equity issuers are less debt constrained than public debt issuers, and that debt ratio targeting is less prominent. Previous analysis indicates that equity issuers have better investment opportunities than public debt issuers, implying less focus on liability management and more focus on growth. Only a handful of equity issuers (10 of 140) had publicly registered debt at the time of the offering, implying that most equity issuers did not have a credit rating to consider. All in all, the data suggest that many equity issuers did not anticipate issuing public debt in the near future and therefore were less concerned than public

332

Brown and Riddiough

debt issuers about total debt levels and how their liability structures would be perceived by outsiders such as rating agencies. Probit Analysis of Financing Choice Financial statements at the time of debt and equity offers indicate that public debt issuers have slightly less total debt and significantly less senior secured debt as a proportion of total assets (see Table 5). This suggests that a firm's preoffer liability structure may affect issuance choice. The firm's total debt capacity (as distinct from its current total debt level) may also affect issuance choice. An example demonstrates why. Suppose that the typical apartment REIT that issues debt has a leverage ratio of 47%, and the typical hotel REIT that issues equity has a leverage ratio of 45%. If property type is ignored, comparing these two observations would suggest that leverage has a minor impact on the external financing decision. In fact, it may very well be the case that apartment REITs, which had relatively stable cash flows over the sample period, have greater debt capacity than hotel REITs. Hence, the apartment REIT may be well below its total debt capacity at the time of issuance, while the hotel REIT may be at its debt capacity. We estimate a probit model of the decision to issue public unsecured debt (1) versus common stock (0), conditional on issuance of one type or the other (exclusively) in a given year. We are particularly interested in how the firm's liability structure (secured and total debt ratios) relates to the issuance choice decision. Other right-hand-side variables proxy for debt capacity, and they include firm size, asset holdings by property type and annual revenue as a percentage of total assets.''* Coefficient estimates for two probit model specifications are provided in Table 6. In the first model the SECURED DEBT variable is excluded in order to better illustrate its effect on the issuance choice decision. In the first model, there is a statistically significant relation between issuance choice and the preoffer amount of total financial leverage, where higher total leverage reduces the likelihood of public debt issuance. This result is consistent with thefindingsshowing that REITs with high leverage levels cannot or choose not to issue public debt (see Figures 1 and 2).

'' The model is estimated with year dummies. Use of proceeds indicator variables are also included in the model specification (see Table 4 for a description of the use of proceeds variables), but estimated variable coefficients are not reported.

Financing Choice and Liability Structure of REiTs 333

Table 6 Probit model of the choice to issue public debt versus equity. Variable INTERCEPT TOTAL DEBT SECURED DEBT TOTAL ASSETS INDUSTRIAL MULTIFAMILY OFFICE RETAIL SELF STORAGE TOTAL REVENUE N McFaddenR" Coefficient -2.68"* (0.94) -3.34" (1.46) 7.32** (3.28) 0.30 (1.14) 1.15* (0.67) -0.56 (0.92) 0.66 (0.70) -0.27 (1.03)) 7.50* (4.U) 180 0.215 Coefficient -3.81"* (1.12) 1.22 (2.11) -6.03*** (1.68) 8.44** (3.54) 0.31 (1.28) 1.38* (0.72) -0.93 (1.23) 0.71 (0.73) -0.27 (1.09) 10.15" (4.58) 180 0.300

The table presents the estimates of a probit model of the likelihood that a firm issues public unsecured debt (1) versus equity (0), conditional on issuance of one type or the other (exclusively) in a given year. The sample includes 40 public debt offers and 140 equity offers. The explanatory variables are TOTAL DEBT, as a percentage of total assets; SECURED DEBT, as a percentage of total assets; TOTAL ASSETS, book value in millions of dollars; dummies for the issuer's property type as designated by SNL, which includes INDUSTRIAL, MULTIFAMILY, OFFICE, RETAIL and SELF STORAGE {HOTEL is the omitted category) and TOTAL REVENUE, as a percent of total assets. Year dummies are included in the specification, but coefficient estimates are not reported. Indicators for reported use of proceeds are also included in the specification (see Table 4), but coefficient estimates are not reported. *** indicates a coefficient significantly different from zero at the 1 % level, ** indicates a coefficient significantly different from zero at the 5% level and * indicates a coefficient significantly different from zero at the 10% level. Standard errors are shown in parentheses.

The probability of a public debt offer is positively related to the size of the firm as measured by total book assets. We offer two explanations for this finding. First, the size of a public debt issue influences the liquidity of the issue and the offer spread. Smaller potential issuers may be unable to borrow the minimum amount ($50 to $100 million) required for the debt to be liquid in the secondary market. Second, the assets of larger REITs are typically more diversified than those of smaller REITs and hence support increased debt capacity.

334

Brown and Riddiough

The coefficient on the multifamily dummy variable is positive and significant at the 10% level. In comparison to other property types, multifamily property exhibited relative stability over the sample period, in the sense that it was less susceptible to oversupply problems and experienced less asset price volatility. Stable cash flows and asset values are attractive to outside debt providers, thus increasing debt capacity and reducing public debt costs. There is a positive relation between the likelihood that the firm issues debt and total revenue as a percentage of total assets. Higher total revenues make it easier to meet interest coverage tests and perhaps indicate greater management efficiency. The second model includes SECURED DEBT in the specification. There is a negative and statistically significant relation between the likelihood of a public debt issue and the preoffer secured debt level. Total leverage is insignificant in this specification. Thus, REITs with high secured debt levels are unlikely to access the public debt market, but total debt per se does not seem to be a barrier to entry. This suggests that the composition of liabilities, and not simply total leverage, is a crucial factor in the issuance choice decision. Another way to interpret the secured debt-total leverage result is to recognize that there is a statistically significant positive relation between the percentage of unsecured debt in a REIT's capital structure and the likelihood of a public debt issuance. Once a REIT issues unsecured debt, and public debt is part of the capital structure, the REIT has a tendency to conduct follow-on public debt offerings. Other variable coefficients are more or less unaffected by the inclusion of secured debt in the second model specification. If anything, statistical significance improves with the inclusion of secured debt. In summary, the probit estimates suggest that a firm's asset, liability and income structure is related to the REIT's decision whether to issue public debt or equity. Directional relations are generally as expected. Our most important result is that the issuance decision is related more to the seniority of debt in a firm's capital structure than the overall level of debt.

Summary of Findings and Directions for Future Research We have undertaken a systematic empirical analysis of REIT financing choice and liability structure. We find that REITs that issue equity are more likely to use offer proceeds to fund investment than REITs that issue public debt.

Financing Choice and Liability Structure of REITs 335

Issuance choice is further found to depend critically on secured debt levels, where firms with higher preoffer levels of secured debt tend to issue equity as opposed to public debt. We also provide evidence that public debt issuers target a longer-run leverage ratio in order to preserve a minimum investment-grade credit rating. Further work in this area may prove fruitful. One important finding is that the type of debt is as important as the amount of debt in a REIT's capital structure. Further refinements that identify differences between and among secured and nonsecured bank lines, mortgage debt, preferred stock, convertible bonds and privately negotiated debt and equity would be useful. A more detailed analysis of debt maturity, perhaps in conjunction with the firm's leasing structure, would also be interesting. A closer study of the dynamics of REIT capital structure and financial policy, especially in the context of industry growth and consolidation, is warranted as well. Finally, in our view, the big questions in REIT corporate finance are why debt is used at all to finance the ownership of real estate assets and which types of debt are most advantageous. REITs are not equity capital constrained in the same way that private real estate investors are. Moreover, there is no obvious tax advantage to debt since REITs do not pay corporate taxes, and there are certainly deadweight costs of financial distress associated with the use of debt. Thus the standard trade-off theory implies that REITs should employ little or no debt to finance themselves. Further, pecking order and free cash-fiow rationales for debt are muted by REIT dividend payout requirements. It is true that bank debt provides near-term fiexibility to fund operations and investment when it is difficult to retain earnings. Beyond fiexibility, it may be that the use of certain types of debt provide monitoring benefits that are useful to outside shareholders. Pinning down how much debt to use, which types of debt and equity are optimal and how the monitoring channel works are important challenges to future research.
We gratefully acknowledge research assistance provided by Matthew Hintze and the comments of Dennis Capozza, David Ling and three anonymous referees.

References
Baker, M.P. and J. Wurgler. 2002. Market Timing and Capital Structure. Journal of Finance 57: 1-32. Bradley, M., D.R. Capozza and P.J. Seguin. 1998. Dividend Policy and Cash-Flow Uncertainty. Real Estate Economics 26: 555-580. Brealey, R.A. and S.C. Myers. 2000. Principles of Corporate Finance. 6th Edition. Irwin McGraw-Hill: New York.

336

Brown and Riddiough

Brown, D.T. 2001. Maturity, Credit Quality and the Term Structure of Credit Spreads. Joumal of Fixed Income 11: 9-27. Buiow, J.L and J.B. Shoven. 1978. The Bankruptcy Decision. Bell Journal of Economics 48: 437^56. Capozza, D.R. and P.J. Seguin. 1999. Focus, Transparency and Value: The REIT Evidence. Real Estate Economics 27: 587-619. Collin-Dufrense, P. and R. Goldstein. 2001. Do Credit Spreads Reflect Stationary Leverage Ratios? Journal of Finance 56: 1929-1957. Diamond, D. 1991. Debt Maturity Structure and Liquidity Risk. Quarterly Journal of Economics 106: 709-737. Fama, E. 1986. Term Premiums and Default Premiums in Money Markets. Journal of Financial Economics 17: 175-196. Gau, G.W. and K. Wang. 1990. Capital Structure Decisions in Real Estate Investment. AREUEA Journal 18: 501-521. Guedes, J. and T. Opler. 1996. The Determinants of the Maturity of Corporate Debt Issues. Journal of Finance 51:1809-1834. He, J., W. Hu and L.H.P. Lang. 2001. Credit Spread Curves and Credit Ratings. Unpublished manuscript. Chinese University of Hong Kong. Helwege, J. and CM. Turner. 1999. The Slope of the Credit Curve for Speculative-Grade Issuers. Journal of Finance 54: 1869-1884. Howe, J.S. and J.D. Shilling. 1988. Capital Structure Theory and REIT Security Offerings. Journal of Finance 43: 983-993. Huang, J. and M. Huang. 2000. How Much of the Corporate-Treasury Yield Spread Is Due to Credit Risk? Unpublished manuscript. Pennsylvania State University. Keim, D. and R. Stambaugh. 1986. Predicting Returns in the Stock and Bond Markets. Journal of Financial Economics 17: 357-390. Leland, H. and K. Toft. 1996. Corporate Debt Value, Bond Covenants, and Optimal Capital Structure. Joumal of Finance 51: 987-1019. Longstaff, F.A. and E.S. Schwartz. 1995. A Simple Approach to Valuing Risky Fixed and Floating Rate Debt. Journal of Finance 50: 789-819. Mann, S. and E. Powers. 2001. Indexing a Bond's Call Price: An Analysis of MakeWhole Call Provisions. Unpublished manuscript. University of South Carolina. Moody's Investors Services. 1998. Credit Evaluation of Real Estate Investment Trusts (REITs) and Real Estate Operating Companies (REOCs). September. New York. . 2002. Moody's Views on Secured Versus Unsecured Debt in REIT's Capital Structures. April. New York. Myers, S. and N. Majluf. 1984. Corporate Financing and Investment Decisions When Firms Have Information that Investors Do Not Have. Journal of Financial Economics 13: 187-221. Shyam-Sunder, L. and S.C. Myers. 1999. Testing Static Tradeoff Against Pecking Order Models of Capital Structure. Journal of Financial Economics 51: 219244. Stohs, M. and D. Mauer. 1996. The Determinants of Corporate Debt Maturity Structure. Joumal of Business 69: 279-312. Stulz, R.M. and H. Johnson. 1985. An Analysis of Secured Debt. Journal of Financial Economics 14:501-521. Welch, 1.2002. Columbus' Egg: The Real Determinant of Capital Structure. Unpublished manuscript. Yale School of Management.

Financing Choice and Liability Structure of REITs 337

Appendix A Table A1 lists numerical conversions for various credit quality measures which are an ordinal ranking of the credit ratings assigned to issuers by Moody's and Standard and Poors (S&P), Table Al Numerical conversions.
Numerical Conversion 1 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20
21 22 23

Moody's Aaa+ Aaa Aal Aa2 Aa3 Al A2 A3 Baal Baa2 Baa3 Bal Ba2 Ba3 Bl B2 B3 Caal Caa2 Caa3 Ca C D

S&P AAA+ AAA AA+ AA AAA+ A ABBB+ BBB BBBBB+ BB BBB+ B BCCC+ CCC CCCCC C D

Appendix B This appendix reports information on the capital structure, flow of funds and a chronology of major financing transactions during the year of the public debt or equity offer. These items, obtained from company filings, are reported for five public debt issuers and five common stock issuers. The presentation of the flow of funds from financing does not include all items reported in the company filings. Only significant financing transactions are reported. Quantities reported in the tables are in millions of dollars. Oasis Residential: Issued $150 million of public debt in November 1996,

338

Brown and Riddiough

Capital Structure

12/95 Assets (millions) Debt (millions) Debt/assets $623.18 $250.82 40.25%

12/96 $745.42 $394.27 52.

At the end of 1995, Oasis had $189.79 tnillion of total debt in the form of mortgage loans. The remainder of the debt was borrowings against a line of credit. By the end of 1996, the mortgage debt had dropped to $135.06 million. Flow of Funds from Financing

1995 Debt issuance Debt repayments Common stock issuance $232.23 ($193.50) $99.20

1996 $329.28 ($185.84)

Financing Chronology 8/96 Increased funding capacity on unsecured credit facility from $150 million to $200 million. 11/96 Completed the issuance of $150 million in unsecured debt. Used approximately $54 million of the proceeds to pay off mortgage debt. This enabled the company to generate 64% of its EBITDA from unencumbered properties during the fourth quarter of 1996. Tri-Net Corporate Realty: Issued $150 million in public debt during May 1996. Capital Structure

12/95 Assets (millions) Debt (millions) Debt/assets $559.73 $270.39 48.31%

12/96 $708.24 $342.19 48.32%

Financing Choice and Liability Structure of REITs

339

In the year prior to issuing unsecured public debt, Tri-Net reduced its reliance on secured debt financing. Flow of Funds from Financing
1995 Borrowings under acquisition facility Payment of acquisition facility Mortgage note borrowings Mortgage note repayment Common stock issuance Unsecured debt issuance Preferred stock issuance $156.90 ($79.80) $114.91 ($151.58) $120.40 1996 $288.51 ($263.31) ($112.74) $149.69 $78.77

Financing Chronology 5/96 Issued $150 million in 5- and 12-year unsecured notes. Used the proceeds to repay borrowings under the acquisition facility. 6/96 Issued $47.7 million in preferred stock and used the proceeds to repay borrowers under the acquisition facility. 6/96 Borrowed $35 million under acquisition facility, partly to repay mortgage loans. 8/96 Issued $31.1 million in preferred stock and used the proceeds to repay borrowers under the acquisition facility. Simon DeBartolo Group: Issued $ 150 million of public debt in July 1997 and $150 million of unsecured debt in October 1997. Capital Structure
12/96 Assets (millions) Debt (millions) Debt/assets $5,895.9 $3,682.0 62.45% 12/97 $7,662.7 $5,078.0 66.27%

Much of the Simon DeBartolo asset growth in 1997 was the result of the acquisition of Retail Property Trust and Shopping Center Properties. Simon Property

340

Brown and Riddiough

assumed $474 million in unsecured debt in the Retail Property Trust acquisition. During 1997, the firm negotiated a $1.25 billion line of credit and it ended 1997 with almost $ 1 billion in borrowings against the line. At the same time, a preexisting line of credit was retired, which partially accounts for the large inand outflows of funds during the year. At the end of 1996, 82% of Simon Property Group's $3.7 billion in debt was secured. By the end of 1997, only 50% of its debt was secured. Borrowings increased from 1996 to 1997 by over $1.3 billioti while secured borrowings actually fell by $230 million. Flow of Funds from Financing

1996 Debt issuances Debt repayments 1,293.6 (1,267.9)

1997 2,976.2 (2,030.8)

Financing Chronology 5/97 Issued $100 million of privately placed notes and used the proceeds to pay down borrowings under the credit facility. 7/97 Issued $150 million in preferred stock and used the proceeds to pay down borrowings under the credit facility. 7/97 Issued $150 million of unsecured debt. The use of proceeds was not disclosed. 10/97 Issued $150 million of unsecured debt and used $115 million of the proceeds to retire mortgages on three malls. The remainder of the proceeds was used to pay down borrowings under the credit facility. Post Properties: Issued $123 million of public debt in October 1996. Capital Structure
12/95 Assets (millions) Debt (millions) Debt/assets $812.98 $349.72 43.01% 12/96 $958.67 $434.32 45.30%

Financing Choice and Liability Structure of REITs

341

While the firm's total indebtedness increased during 1996, secured debt fell from $189 million at the end of 1995 to $170 million at the end of the 1996, Most of the secured debt is tax-exempt notes secured by real property. Flow of Funds from Financing
1995 Borrowings under credit line Credit line repayments Unsecured debt issuance Preferred stock issuance $362.20 ($374.52) $105.28 1996 $236.83 ($277,23) $123,44 $48.61

Financing Chronology 10/96 Issued $49 million worth of preferred stock and used the proceeds to buy operating partnership units. 10/96 Issued $123 million in unsecured debt and used the proceeds to repay borrowings under the revolving line of credit. Camden Property Trust: Issued $ 170 million in public debt in February 1996. Capital Structure
12/95 Assets (millions) Debt (millions) Debt/assets $570.80
$235,45 41,25%

12/96 $590,16 $244.18 41,38%

While the firm's total indebtedness and leverage ratios were largely unchanged, its amount of secured debt fell from $112.67 million at the end of 1995 to $58.38 million at the end of 1996. Flow of Funds from Financing
1996 Net borrowings from line of credit Proceeds from debt issuance $50,78 $39,86 1997 ($110.78) $170.00

342

Brown and Riddiough

Financing Chronology 2/96 Issued $170 million of unsecured debt and used the proceeds to repay borrowings under a line of credit and to retire remaining secured construction loans. Arden Realty: Issued $377 million in common stock in October 1996. Capital Structure

12/95 Assets (millions) Debt (millions) Debt/assets $340.12 $173.61 51.04%

12/96 $551.26 $174.16 31.59%

While Arden's total borrowings were largely unchanged from the end of 1995 to the end of 1996, there was a marked reduction in its secured borrowings. Concurrently, the firm's unsecured line of credit went from nearly nothing to $51 million. Flow of Funds from Financing

1995 Mortgage loan proceeds Mortgage loan repayment Borrowings under credit line Proceeds from equity issuance

1996

$120.48 ($5.05)

$104.00 ($368.47) $51.00 $377.31

Financing Chronology 10/96 Issued $377 million in common stock and used the proceeds to repay mortgage debt financing. Beacon Properties: Issued a total of $755 million in common stock in three separate offerings during 1996. The first two offerings were used to fund acquisitions. The third offering was used to repay debt.

Financing Choice and Liabiiity Structure of REITs 343

Capital Structure
12/95 Assets (millions) Debt (millions) Debt/assets $534.80 $239.01 44.69% 12/96 $1,779.01 $676.71 38.04%

The growth in debt on Beacon's balance sheet during 1996 primarily came from increased secured debt financing. Specifically, Beacon went from $70.53 million in mortgage debt at the end of 1995 to $452 million in secured debt at the end of 1996. Flow of Funds from Financing
1995 Proceeds from equity issue Credit facility borrowing Repayment of credit facility Mortgage note borrowing Repayment of mortgage notes $160.03 $124.70 ($124.50) $608.56 ($281.8) 1996 $754.78 $460.00 ($445.50) ($20.40)

Financing Chronology 1/96 Converted $55 million credit facility into permanent mortgage loan. 2/96 Converted $60 million credit facility into permanent mortgage loan. 3/96 Issued $173.8 million in common stock and used the proceeds to acquire units in an operating partnership. 3/96 Closed a $218 million mortgage loan and used the proceeds to repay borrowings under the acquisition loan facility. 8/96 Issued $139.4 million in common stock and used the proceeds to purchase a portfolio of properties from New York Life. 11/96 Issued $436.7 million in common stock and used the proceeds to finance acquisitions and generate cash. 12/96 Closed a $15 million mortgage loan and used the proceeds to pay off loans under the revolving line of credit.

344

Brown and Riddiough

General Growth Properties: Issued $166 million in common stock in October 1997. Capital Structure

12/97 Assets (millions) Debt (millions) Debt/assets $1,757.32 $1,169.88 66.57% $2,098.01 $1,276.71 60.85%

Nearly all of General Growth Properties' debt obligations are mortgages secured by specific properties. Flow of Funds from Financing

1996 Proceeds from equity issues Proceeds from mortgage loans Repayment of mortgage loans

1997

$705.8 ($668.0)

$165.8 $832.5 ($802.9)

Financing Chronology 1/97 Gompleted the sale of GenterMall for $130 million and used the proceeds to repay $12.6 million in mortgage debt and borrowings under a nonrecourse borrowing facility. 9/97 Negotiated a $200 unsecured credit facility and borrowed $88 million against the facility during 1997. 10/97 Issued $165.8 million in common stock, used primarily to reduce borrowings under its development loan and unsecured debt facilities. 11 /97 Closed on a $560 million private placement of a loan secured by the firm's interests in 13 regional malls. Storage Trust Properties: Issued $84 million in common stock in June 1996.

Financing Choice and Liabiiity Structure of REiTs

345

Capital Structure
12/95 12/96

Assets (millions) Debt (millions) Debt/assets

$190.02 $39.28 20.67%

$304.11 $63.25 20.80%

Debt at the end of 1995 was nearly all borrowings under a $50 million secured bank line of credit. At the end of 1996. the debt consisted of borrowing under a $75 million unsecured Hne of credit. Flow of Funds from Financing 1995 Borrowings under credit line Repayment of line of credit Equity issuance proceeds $96.87 ($70.82) $57.50 1996 $113.73 ($86.16) $83.83

Financing Chronology 2/96 Received a two-year $75 million unsecured line of credit. The facility replaced a $50 million secured line of credit. 6/96 Issued $83 million in common stock and used the proceeds to fund acquisitions. Summit Properties: The firm issued $97 million in common stock in August 1996. Capital Structure 12/95 Assets (millions) Debt (millions) Debt/assets $533.25 $297.01 55.70% 12/96 $634.91 $309.93 48.81%

The firm's borrowings at the end of 1995 are entirely secured debt (mortgages and borrowings under lines of credit). By the end of 1996. almost half the borrowings are unsecured.

346

Brown and Riddiough

Flow of Funds from Financing

1995 Debt proceeds Repayment of line of credit Equity issuance $97.07 ($ 101.65) $65.94

1996 $89.90 ($90.70) $96.60

Financing Chronology 7/96 Privately placed $31 million in unsecured notes. 8/96 Issued $97.6 million in common stock. Some $97.07 million of the proceeds from the debt private placement and equity issuance were used to pay off secured borrowings under a line of credit and secured development loans. 11/96 Replaced a $50 million secured line of credit with a $150 million unsecured line of credit.