1 Introduction

The capital structure of real estate investment trusts (REITs) is different from that of industrial firms because of their unique regulatory environment. Generally, REITs have higher leverage ratios and are forced to access external financing more frequently since REITs pay out most of their earnings as cash dividends. Similar to industrial firms, the capital sources for REITs include equity, public debt, syndicated loans, and mortgages, where over the past two decades the dominant financing source has been syndicated loans. Table 1 shows that syndicated loans for equity REITs increase from $85 million (8.94% of the capital offering) in 1989, to $44.29 billion (45.00% of the capital offering) in 2015. Figure 1 shows this trend by amount and percent of shares from 1989 to 2015. After 1994, we observe that REITs tend to be financed through syndicated loans rather than through public debts or seasoned equity offerings (SEOs). The amount of syndicated loans reaches a peak of $58.70 billion in 2013 (50.45% of the capital offering). There are only four years after 1993 (years 1996, 1997, 2009, and 2010) where syndicated loans do not dominate the financing sources in terms of percentage of shares. Despite this increased reliance on syndicated loan financing, there are very few studies on REIT syndicated loans.

Table 1 Total amount of capital raised from different sources by equity REITs
Full size table
Fig. 1
figure 1

Total Amount of Capital Raised from Different Sources by Equity REITs. This figure represents the total amount of capital raised from different sources by equity REITs. Private Debt is syndicated loans borrowed from banks, with data obtained from the DealScan database. SEO includes both common stock and preferred stock issuances, with data obtained from the NAREIT capital offering database. Public Debt includes long-term notes and mortgage-based securities issued in the public market, with data obtained from the NAREIT capital offering database. The sample period is from 1987 to 2015, while NAREIT’s capital offering data starts from 1989

Full size image

A syndicated loan is credit granted by a group of banks to a borrower. In general, a syndicated loan arises when a borrower requires a loan that is so large that a single lender is not able to grant it due to the risk issue or credit line limits. A group of lenders, including at least one lead bank, is in such cases syndicated to offer the loan to the borrower with the same loan conditions. Syndicated loans are quite popular in the European and U.S. markets, and have evolved since the 1990s to become one of the main sources of funding for corporate borrowers.

Previous literature on REIT financing sources largely focuses on public debt and equity offerings (Brown and Riddiough 2003), while few studies are on bank debt. Among the bank debt studies, most focus on REITs’ lines of credit. For example, Hardin III et al. (2009) find that there is a negative relationship between cash holdings and lines of credit. Hardin III and Hill (2011) suggest that lines of credit increase REITs’ external liquidity, allowing them to lower their cash holdings to mitigate the agency problem. Hardin III and Wu (2010) show that bank debt is an important financing source for REITs. They find that REITs with banking relationships tend to have lower leverage, less secured debts, and access the public debt market more frequently. Moreover, their empirical results explain why REIT financing has shifted from traditional mortgages to bank debts. While these studies examine how banking relationships affect REIT syndicated loan facility conditions and capital structures, we still do not know how banking relationships affect syndicated loan pricing.

Previous literature suggests that relationships between borrowers and lenders result in favorable loan conditions and help reduce information asymmetry between the loan parties as well as improve a borrower’s corporate governance. For example, Bharath et al. (2011) find that borrowing companies that have relationships with their banks have favorable syndicated loan conditions compared to those without banking relationships. They find that banking relationships lower loan spreads by 10–17 basis points, which is valuable for borrowing firms that have low transparency. Alexandre et al. (2014) find that firms with banking relationship before 2008 have lower loan spreads and longer maturity during the financial crisis. Dass and Massa (2011) find that stronger relationships between borrowing companies and banks improve the corporate governance of the borrowing firms. Yildirim (2020) finds that relationship loans reduce the default risks and improve the efficiency of borrowing companies, especially for inefficient and less creditworthy ones. However, the reliance on syndicated loans as a financing source may cause a firm to suffer larger valuation losses and a higher decline in both capital expenditures and profitability during a financial crisis. This is because banks are often required to reduce their lending amounts and ask for higher loan interest rates during such periods (Chava and Purnanandam 2011). From previous literature, whether REITs with banking relationships distinguish themselves from those without banking relationships in terms of borrowing costs during financial crisis periods remains unclear.

This study focuses on REIT syndicated loan facilities and investigates whether REITs with banking relationships have favorable loan conditions compared to those without banking relationships. We also investigate whether the strength of the relationship affects loan conditions. In addition, the subprime crisis that resulted from the bursting of the housing bubble in the U.S. caused significant losses in the REIT industry. This economic shock should affect REIT capital costs. We therefore further investigate how banking relationships affect REIT financing costs before, during, and after the subprime crisis given that past studies show that bank lending activities decreased during the crisis.Footnote 1 Lastly, how the offer of a new syndicated loan affects the following cost of capital is also examined.

This research contributes to the literature on how banking relationships impact the spreads of REIT syndicated loans, as banks are an important financing source for growing REITs. We further investigate the effect of financial crises on borrowing costs, which has not been examined before. The results provide insight for REITs to understand the lending behavior of banks and the importance of banking relationship management. Finally, the linkage between a new syndicated loan and the following public financing cost has not previously been investigated. The evidence provides further insight on the cost effect of capital structure management for REITs.

Using REIT syndicated loans from 1987 through 2015, we find that REITs with banking relationships benefit from significantly lower spreads, longer maturities, larger loan amounts, and less collateral requirements. In addition, there are more banks participate in the loans and lead banks retain smaller shares of the loans. We also find that the financial crisis causes the spread to go up significantly but only in the short run. After the financial crisis, the spread level declines but does not go back to the pre-crisis level. This evidence indicates that banks become more conservative after the crisis. More importantly, banking relationships reduce the spread in every period, including that of the financial crisis. Last, we find that the effect of banking relationships on loan spreads is greater for term loans than for credit lines. Furthermore, a new syndicated loan results in lower bond spread and underpricing for REITs with banking relationships.

The literature background is provided in Sect. 2, data collection and methodology are described in Sect. 3, and empirical results are presented in Sect. 4. Finally, we conclude the paper in Sect. 5.

2 Literature

Previous studies on syndicated loans show that greater information asymmetry and moral hazard result in less favorable loan conditions (see Diamond 1984). Sufi (2007) investigates how information asymmetry between lenders and borrowers influences syndicated loan structures and what lenders become syndicate members. He finds that when information asymmetry is severe, lead lenders retain larger loan shares. He further finds that participant lenders that are closer to the borrower, both geographically and in terms of previous lending relationships, are more likely to become syndicate members. These participant lenders are invited to mitigate information asymmetry. Focarellia et al. (2008) find that the announcement of a syndicated loan facility has a positive effect on the borrowers’ stock price and that this effect is an increasing function of the share of the loan retained by the arranger. Lead banks tend to increase their share of the loan held to mitigate the agency problem when there is greater information asymmetry between the borrowing company and the lenders.

Bharath et al. (2007) indicate that a borrower with a relationship to a lender has a higher probability of obtaining a loan from the same lender in the future. Bharath et al. (2011) find that borrowing companies with a banking relationship have favorable syndicated loan conditions compared to those without a banking relationship. These conditions include lower spreads, larger loan amounts, and less collateral requirements. This banking relationship is especially valuable when borrowers face a higher degree of information asymmetry and moral hazard among syndicated lenders. Their results also suggest that borrowing companies obtain favorable loan conditions even when they have multiple external financing sources. Zhang et al. (2022) also find that lending relationships facilitates the pricing of syndicated loans in terms of fewer adjustment frequency and shorter syndication time period. Gustafso et al. (2021) directly measure the monitoring of lead lenders in the syndicated loan market and find that more monitoring leads to lower loan spreads and shorter maturity. Chava and Purnanandam (2011) find that borrowing companies who mainly rely on bank loans suffer larger valuation losses during financial crises and experience a higher decline in both capital expenditures and profitability. Their empirical results further show that financial crises result in a lower quantity of lending and higher loan interest rates in the post-crisis period. However, Schwert (2018) documents different financing behaviors between bank-dependent firms and firms that can access the public debt market. They show that the matching of banks and firms is due to the informational frictions and borrowers’ access to outside funding rather than the risk management policy of banks.

Past literature on lines of credit finds a negative relationship between lines of credit and firms’ cash holdings. Since REITs have limited cash holdings due to the payout requirement, this finding implies that REITs may tend to increase their lines of credit. Whether this inference applies to REITs is not clear; only a few studies examine lines of credit or syndicated loans within the context of REITs. An example is Hardin III and Wu (2010), who investigate the impact of banking relationships on REIT capital structures. Their sample contains syndicated loans of equity REITs from 1992 through 2003. They collect data on 1,061 bank lines of credit and revolvers, 303 term loans, and 70 other loans and find that REITs with banking relationships have lower leverage and less secured debts. Their results further show how REIT financing has moved from traditional mortgages to bank debts; their focus is on how banking relationships affect REITs’ capital structures and their access to the public capital market. However, how banking relationships affect REITs’ syndicated loan conditions and whether the impact changes after the subprime crisis remains unknown. The other study on lines of credit is Ooi et al. (2012). They indicate that credit lines protect REITs from firm-level credit quality deterioration and that REITs are more likely to draw down on their credit lines in tight credit markets. They also find that the REIT sector relies more heavily on bank lines of credit as compared to firms operating in other sectors. Their observations evidence the importance of bank debt for REITs, though how the cost of financing is related to banking relationships is still unclear.

This study examines the syndicated loan pricing criteria, and the effect of the subprime crisis on REITs’ cost of capital. Dass and Massa (2011) and Bharath et al. (2011) show that relationship loans are better monitored, lead to better corporate governance, have a lower degree of information asymmetry, have lower costs of borrowing, and have better loan conditions. Therefore, we expect that REITs with banking relationships will pay lower costs compared to REITs without banking relationships. In addition, we expect that strong banking relationships should protect REITs from paying extremely high spreads during the financial crisis period. Finally, we expect to observe lower loan interest rates and underpricing after REITs are granted syndicated loans.

3 Data and methodology

To determine the REIT sample, we first collect syndicated loan facility data from the Loan Pricing Corporation’s (LPC) DealScan database over the period of 1987 through 2015. The DealScan database contains information about the loans (mainly syndicated loans) of large global borrowing companies. The information includes loan conditions (e.g., spread, maturity, loan amount, and collateral requirements), loan structure (e.g., information on lead banks and participant banks and their shares of the loans), and information about the borrowing and lending companies (e.g., name, industry, and SIC code). There are 316 REITs, 2,125 loan packages, and 4,152 lead bank facility-level observations obtained from DealScan database from 1987 to 2015. Second, financial information is retrieved from Compustat. We use DealScan-Compustat Linking data provided by Chava and Roberts (2008)Footnote 2 to merge the dataset between DealScan and Compustat. The CRSP Ziman database is used to identify equity REITs (SIC code 6798) as our sample. We exclude observations when data on all-in-drawn spreads, loan amounts, or maturities are not available, since they are important regression variables; REITs with a negative book value of total assets are also excluded. Finally, we winsorize the top and the bottom 1% of the observations according to Spread, Leverage, and M/B ratio. The final sample contains 238 REITs (around 75% of the total data), 1,587 loan packages (around 75% of the total data), and 3,173 lead bank facility-level observations.

Daily stock prices and market values are taken from CRSP. Other available information about SEOs is obtained from SDC, including issue date, offer price, principle amount, and underwriters. Our final sample is matched with offering information obtained from SDC and CRSP through both CUSIP numbers and issuer names. The data on corporate bond offerings, including issue date, coupon rate, maturity, currency, amount issued, and bond price at issue date, is collected from the Bloomberg Terminal. The loan type include U.S. domestic and domestic medium-term notes.

We adopt lead bank facility-level observations related to the REIT syndicated loans since lead banks are more powerful and have larger effects on loan conditions than other participant banks. Relationship loans are expected to have favorable syndicated loan conditions compared to non-relationship loans (Bharath et al. 2011). To examine the impact of prior lending relationships on REIT syndicated loan conditions, we group REIT syndicated loans into those with banking relationships and those without banking relationships according to the following three proxies: REL(Dummy), REL(Number), and REL(Amount). Following Hardin III and Wu (2010) and Bharath et al. (2011), REL(Dummy) receives a value of one if the REIT borrowed from the same lead bank in the preceding 5 years, and zero otherwise. REL(Number) and REL(Amount) measure banking relationship strength, following both Dahiya et al. (2003) and Bharath et al. (2011). The equations are shown as follows:

$$REL\left( {Number} \right)_{{m,i}} = \frac{{{\text{Number}}\;{\text{of}}\;{\text{loans}}\;{\text{by}}\;{\text{bank}}\;m\;{\text{to}}\;{\text{borrower}}\;i\,{\text{in}}\;{\text{the}}\;{\text{preceding}}\;5\;{\text{years}}}}{{{\text{Total}}\;{\text{number}}\;{\text{of}}\;{\text{loans}}\;{\text{by}}\;{\text{borrower}}\;i\;{\text{in}}\;{\text{the}}\;{\text{preceding}}\;5\;{\text{years}}}}$$
(1)
$$REL(Amount)_{{m,i}} = \frac{{\$ {\text{amount}}\;{\text{of}}\;{\text{loans}}\;{\text{by}}\;{\text{bank}}\;m\;{\text{to}}\;{\text{borrower}}\;i\;{\text{in}}\;{\text{the}}\;{\text{preceding}}\;5\;{\text{years}}}}{{{\text{Total}}\;\$ \;{\text{amount}}\;{\text{of}}\;{\text{loans}}\;{\text{by}}\;{\text{borrower}}\;i\;{\text{in}}\;{\text{the}}\;{\text{preceding}}\;5\;{\text{years}}}}$$
(2)

where REL(Number) measures the number of loans lead bank m has lent to borrower i in the preceding 5 years relative to borrower i’s total number of loans, and REL(Amount) measures the total size of the loan(s) lead bank m lent to borrower i in the preceding 5 years relative to borrower i’s total loan amounts. A higher REL(Number) and REL(Amount) means a stronger banking relationship.Footnote 3

We apply Bharath et al.’s (2011) syndicated loan model to examine the impact of banking relationships on REIT syndicated loan conditions. We also investigate how spread changes before, during, and after the subprime crisis that caused both REITs and banks to suffer large losses. The equation is as follows.

$${Spread}_{i,t}=\alpha +{\beta }_{1}{REL}_{i,t}+{\beta }_{2}{Pre-Crisis}_{i,t}+{\beta }_{3}Post-{Crisis}_{i,t}+{\beta }_{4}{LoanControl}_{i,t}+{\beta }_{5}{X}_{i,t-1}+{\beta }_{6}{LoanPurpose}_{i,t}+\varepsilon$$
(3)

where Spread denotes the all-in-spread drawn in basis points, and is calculated as the difference between the prime rate or London Inter-bank Offered Rate (LIBOR) and the facility’s all-in-drawn loan rate plus annual fees; REL denotes the banking relationship, which is either REL(Dummy), REL(Number), or REL(Amount), as defined above, or REL(Number)2 and REL(Amount)2, which are the squares of REL(Number) and REL(Amount), respectively. Pre-crisis and Post-crisis are dummy variables. When both variables equal zero, the loan year is during the subprime crisis period, i.e., from year 2008 through 2009. When Pre-crisis equals one and Post-crisis equals zero, the loan year is before the crisis, i.e., from year 1987 through 2007. When Pre-crisis equals zero and Post-crisis equals one, the loan year is after the crisis, i.e., from year 2010 through 2015. The first coefficient, β1, measures whether a banking relationship helps reduce the borrowing costs; the second coefficient, β2, examines whether the spread before the crisis is lower than the spread during the crisis; the third coefficient, β3, tests whether the spread after the crisis is lower than the spread during the crisis. All three variables are expected to be negative for the following reasons. First, since REITs significantly rely on private loans and are usually highly leveraged, keeping a good banking relationship should lead to reduced REIT financing costs. Second, in the panic of 2008, new lending fell (see Ivashina and Scharfstein 2010) and the economy became more uncertain. We expect that banks thus become more conservative and ask for higher spreads during the crisis. Last, after the crisis, the spread should bounce back, though the level may not be the same as before. LoanControl denotes a set of control variables relevant for syndicated loans as follows: ln(Loan amount) is calculated as the natural log of the total deal amount of a facility in millions of dollars, ln(Maturity) is calculated as the natural log of the number of months to maturity from a facility’s start, # of lead banks measures the number of lead lenders for a facility, and lead bank shares measures the percentage of loan amount held by lead banks. X refers to a set of REIT specific control variables as follows: Investment grade equals one if the S&P long term issuer credit rating is investment grade, and zero otherwise, FFO/Assets is the ratio of funds from operations to book value of total assets, STD(FFO) is the standard deviation of funds from operations over the preceding 5 years, M/B ratio is the ratio of market value of total assets to book value of total assets, Leverage measures market leverage calculated as total debts over the sum of total debt and the market value of equity, and Size measures the natural log of the book value of total assets.Footnote 4LoanPurpose denotes loan purpose fixed effects.

We also investigate how spread changes during the subprime crisis when REITs have banking relationships, since both REITs and banks suffer large losses. The equation is as follows.

$${Spread}_{i,t}=\alpha +{\beta }_{1}{REL}_{i,t}+{\beta }_{2}{Crisis}_{i,t}+{\beta }_{3}{Crisis}_{i,t}\times {REL}_{i,t}+{\beta }_{4}{LoanControl}_{i,t}+{\beta }_{5}{X}_{i,t-1}+{\beta }_{6}{LoanPurpose}_{i,t}+\varepsilon$$
(4)

where Crisis is a dummy variable that takes the value of one if the loan year is during the financial crisis period, i.e., year 2008 to 2009, and zero otherwise. Crisis × REL is the interaction term for REITs with banking relationships and loan years during the crisis. β1 and β1+β3 measure whether a banking relationship helps reduce the borrowing costs in the non-crisis and crisis periods, respectively; β2+β3 and β2 examine whether the spread during the crisis is higher than the spread during the non-crisis for REITs with and without banking relationships, respectively; β3 examines whether banking relationships protect REITs from incurring increasing costs and having lower spreads during the crisis. Thus, β1 and β3 are expected to be negative, while β2, and is expected to be positive.

4 Empirical results

4.1 Capital sources of equity REITs

Table 1 and Fig. 1 describe the amount and the percentage of different types of capital sources issued by equity REITs from 1989 through 2015. The capital sources include public debts, syndicated loans, and SEOs. Public debts are long-term notes and mortgage-backed securities issued in the public market, syndicated loans are private debts from banks, and SEOs include common and preferred equity. Both data on public debts and SEOs are collected from the NAREIT Capital Offering database, while data on private debts are obtained from the DealScan database.

We observe that REITs increase their reliance on syndicated loans after 1994. In Table 1, SEOs and public debts make up more than 90% of the capital sources of REITs from 1989 to 1993, except for in 1990; SEOs are the major resource during these years. After this period, we observe a large change in financing for modern REITs. REITs shift to use a larger portion of syndicated bank loans (private debts) for their capital needs. This trend becomes especially prominent from 1998 onward. We further observe that there are large decreases in public debts and syndicated loans for REITs in 2008, where both banks and REITs suffer large valuation losses. After the crisis, REITs change their capital sourcing policies, and rely more on public equity. This is largely because crisis-affected banks tend to charge higher loan interest rates and decrease their lending quantities during post-crisis periods (Chava and Purnanandam 2011). However, syndicated loans dominate the financing sources again after 2011. We observe a pro-cyclicality in syndicated lending. The amount of syndicated lending decreases as credit tightens during the economic downturn of 2001, and the financial crisis of 2008–2009. In addition, the percentages of funding from SEOs and syndicated loans are negatively correlated.

Figure 2 shows the change in banking relationships and in the strength of the banking relationships for each year. We observe an increasing trend that indicates banking relationships become stronger for REL(Dummy) and REL(Number) but not for REL(Amount). This indicates that REITs might tend to keep their relationships with the same banks through borrowing frequency but that they might not want to increase their loan amounts from the banks to avoid the liquidity risk of banks’ insufficient funds. Figure 3 shows the difference in firm characteristics between REITs with and without banking relationship for each year. It indicates that the loan size and maturity increase over time and that for most of the years the loan spread is lower for REITs with banking relationships.Footnote 5 The statistical comparison of the loan characteristics is provided in the following 4.3 section. Figure 4 shows that syndicated loans are the main funding source relative to corporate bonds for REITs with and without banking relationships. Figure 5 reports the average amount of capital, which include REIT corporate bond offerings and SEOs, in the years after a new syndicated loan is borrowed. We observe that banking relationships help REITs access the capital market and obtain more corporate bonds and equity compared to REITs without banking relationships.

Fig. 2
figure 2

Banking Relationship Variables of Equity REITs from 1987 to 2015. This figure represents the changes in the average banking relationship variable of equity REITs from 1987 to 2015. For each new syndicated loan, we compute three banking relationship variables: REL(Dummy), REL(Number), and REL(Amount). REL(Dummy) is a dummy variable that receives the value of one if a REIT borrows from the same lead bank in the preceding 5 years, and zero otherwise. REL(Number) is calculated as the number of loans provided by the same lead bank(s) to borrower i in the preceding 5 years divided by borrower i’s total number of loans in the preceding 5 years. REL(Amount) is calculated as the total size of the loans provided by the same lead bank(s) to borrower i in the preceding 5 years divided by the total size of borrower i’s loans in the preceding 5 years. Syndicated loan data are collected from the DealScan database. The final sample contains 238 REITs, 1,587 loan packages, and 3,173 lead bank facility-level observations. The figure is shown as lead bank-facility level observations

Full size image
Fig. 3
figure 3

Average Size, Spread, and Maturity of REIT Syndicated Loans. This figure represents the average size, spread, and maturity of syndicated loans borrowed by equity REITs with or without banking relationships. Syndicated loan data are collected from the DealScan database. The final sample contains 238 REITs, 1,587 loan packages, and 3,173 lead bank facility-level observations. The figure is shown as lead bank-facility level observations. REITs with (without) banking relationships are the REITs that (do not) borrow from the same lead bank in the preceding 5 years

Full size image
Fig. 4
figure 4

Average Amount of Capital Raised from Different Sources by REITs. This figure reports the average amount of capital REITs raise from different sources. The sample period is between 1988 and 2015. REITs with (without) banking relationships are the REITs that (do not) borrow from the same lead bank in the preceding 5 years. The information on syndicated loans is from the DealScan database. The information on corporate bond offerings is from the Bloomberg Terminal. The information on SEOs is from the SDC Global New Issues database and the CRSP Ziman Real Estate Database

Full size image
Fig. 5
figure 5

Average Amount of Capital Raised by REITs in Each Corporate Bond Offering and Each Seasoned Equity Offering in the Years after a New Syndicated Loan Is Borrowed. This figure reports the average amount of capital raised by REITs in each corporate bond offering and each seasoned equity offering (SEO) in the years after a new syndicated loan is borrowed. Year 0 is the year when a new syndicated loan is borrowed, and thus Year 1 is one year after the bank loan. REITs with (without) banking relationships are the REITs that (do not) borrow from the same lead bank in the preceding 5 years. The information on syndicated loans is from the DealScan database. The information on corporate bond offerings is from the Bloomberg Terminal. The information on SEOs is from the SDC Global New Issues database and the CRSP Ziman Real Estate Database. The sample period is between 1988 and 2015

Full size image

4.2 Summary statistics

Table 2 provides loan type distributions for the equity REIT syndicated loans over the period 1987 through 2015. The whole syndicated loan market reaches around $44.29 billion in 2015. Of the syndicated loans, credit lines are the major lending type, approaching 54.19% in year 2015, followed by term loans and others.

Table 2 Sample distribution of REIT syndicated loans by year
Full size table

Table 3 shows the summary statistics. Panel A shows the banking relationships of the sample REITs. The mean of REL(Dummy) is 0.504, which implies that 50.40% of the lead bank facility-level observations are relationship loans. Moreover, the means of REL(Number) and REL(Amount) are 0.284 and 0.138, respectively, implying that REITs borrow on average 28.40% of their loans and 13.80% of their loan amounts from the same lender in the 5 years prior to the observation loan. In untabulated results, we find that REITs are more likely to borrow from the same lenders compared to industrial firms.Footnote 6

Table 3 Summary statistics
Full size table

Panel B shows syndicated loan conditions, loan types, loan purposes, and financial covenant types. Loan conditions include price terms, such as spread, and non-price terms. Non-price term conditions include loan amounts, number of lenders, number of lead lenders, lead bank shares, and financial covenants. The mean (median) of the spreads of REIT syndicated loans is 166.840 (150.000) basis points or 1.67% (1.50%). Mean loan amounts and maturities are $412.73 million and 42.64 months, respectively. The lenders require collateral in 36.50% of the loans. The average loan has 10.106 lenders and 2.316 lead lenders. Furthermore, lead banks, on average, retain 43.99% of the loan amount. We also observe that 58.80% of the REITs provide covenants. As for loan types, 60.60% of the loans are revolver and line of credit loans, while 31.30% are term loans. In terms of loan purpose, the most common reasons are corporate (54.30%), working capital (18.60%), and debt repayment (11.00%).

Panel C shows the summary statistics of specific control variables; these variables might be expected to affect the loan pricing. We observe that 43.40% of the REITs are not rated and that 37.60% of the REITs are classified as investment grade by the S&P long term issuer credit rating. The mean of FFO/MVE and FFO/Assets are 8.90% and 5.60%, respectively. Finally, the mean leverage ratio is 42.30%.

4.3 REIT banking relationships and loan conditions

We examine loan condition differences between REITs with and without banking relationships in this section. REL(Dummy) indicates the existence of a banking relationship while the banking relationship strength is measured by REL(Number) and REL(Amount).

Table 4 presents the results of the difference in means tests for syndicated loan conditions based on REIT banking relationships. We show that REITs with banking relationships have significantly lower spreads, larger loan amounts, longer loan maturities, less collateral requirements, a larger number of lenders and lead banks, and lower shares held by lead banks compared to REITs without banking relationships, at the 1% significance level. REITs with banking relationships have favorable loan conditions, where the cost of loans are lower by 21.51 basis points, the loan sizes are larger by $105.70 million, and they are less likely to be required to provide collateral for their loans. In terms of the loan structure, relationship loans tend to have a greater number of lenders and the lead lenders tend to retain a 5.65% lower share of the loans compared to non-relationship loans.

Table 4 Loan characteristic comparison based on banking relationship
Full size table

4.4 Effects of REIT banking relationships on loan spreads

Table 5 presents the results from the regression of loan spreads on banking relationships. We show the results for the whole sample as well as separately for the credit line and term loan subsamples, respectively. There are 1,914 credit line loan observations and 990 term loan observations. The coefficients on banking relationship, proxied by REL(Dummy), REL(Number), and REL(Amount), are all negative for the whole sample as well as for each of the two subsamples. These results support our hypothesis that banking relationships benefit REITs in reducing the borrowing costs for all types of syndicated loans. For the whole sample, on average, REITs with banking relationships pay 13.53 basis points less than REITs without banking relationships. For credit lines and term loans, this number is 11.13 basis points and 15.41 basis points, respectively.

Table 5 Effects of REIT banking relationships on loan spreads in different periods across different loan types
Full size table

As we expect, the Pre-crisis coefficients are negative for the whole sample as well as each of the two subsamples. In addition, the Post-crisis coefficients are negative and significant, though relatively smaller, for the whole sample and the credit line subsample. For example, for the whole sample, the coefficient for the pre-crisis period is -103.1, i.e., the pre-crisis period basis point spread is lower compared to the spread during the crisis period, while the coefficient for the post-crisis period is -52.72. This evidence is consistent with the univariate test resultsFootnote 7 and shows that lenders become more careful after an economic shock, though relationship loans still prove to be less expensive.

In addition, for the whole sample, we observe that REITs pay significantly lower spreads when they have investment credit ratings, have higher FFO relative to assets, and have a larger firm size. REITs with higher FFO volatility and higher leverage ratios are riskier and bear higher costs. These findings indicate that banks appreciate cash-rich REITs as well as larger REITs, since large REITs with high cash flows can defend themselves from crises and have the capacity to pay off the debt. On the other hand, REITs with higher leverage ratios are not able to obtain lower borrowing costs since they may run out of financial slack.

For the credit line subsample, the results are similar, though larger loan amounts, longer maturities, and higher M/B ratios further help to reduce the spread, while the percentage of FFO to assets does not. This evidence shows that REITs with growth opportunities are expected to have the potential to generate profits and pay back the credit line borrowed in the future. For the term loan subsample, the results are similar to what we observe for the whole sample, with the exception that loan amounts are positively associated with spread while FFO volatility does not affect the spread.

Table 6 tests whether banking relationships during the crisis period further help to reduce loan costs. We find that the results are similar to those in Table 5. During the non-crisis period, REITs with banking relationships pay 8.996 basis points less than REITs without banking relationships. Then, during the crisis period, the spreads for all REITs (i.e., REITs with and without banking relationships) increase by 95.92 basis points, before taking banking relationships into consideration. Banking relationships are found to help REITs reduce this increase in borrowing cost by 36.56 basis points. Thus, taken together, for REITs with banking relationships, the borrowing cost increases by only 59.36 (95.92 minus 36.56) basis points during the crisis period compared.

Table 6 Regression of loan spread on REIT banking relationship
Full size table

4.5 REIT banking relationships and cost of public capital

In addition to the cost of syndicated loans, we further examine whether banking relationships improve the cost of public capital in terms of bond issuances and equity offerings. Datta et al. (1999) suggest that banking relationships significantly reduce the initial public straight bond offering yield spreads by about 68 basis points for industrial firms. In addition, Schenone (2004) show that firms with a pre-IPO banking relationship with a prospective underwriter receive lower IPO underpricing. For whether a new syndicated loan affect the difference in cost of public capital between REITs with and without banking relationships deserve further investigation. REITs with banking relationships signal better firm quality and transparency to the market and are expected to bear lower funding costs.

Table 7 presents the bond issuance spread in the years following a new syndicated loan. The results in Panel A show that, compared to REITs without banking relationships, REITs with banking relationships pay 38.5 basis points lower in bond spreads in the year when a new syndicated loan is granted. Panel B indicates that the same results persist throughout all the examined periods, though the difference is extremely large during the crisis period (293.20 basis points lower for REITs with banking relationships) and narrows after the crisis (13.70 basis points lower).

Table 7 Bond issuance spread in the years following a new syndicated loan
Full size table

Table 8 shows the results by regressing bond spread on the bond control variables as well as variables similar to those tested in Table 6 and show similar findings. During the non-crisis period, banking relationships help reduce the borrowing cost of public debt by around 33.99 basis points. The crisis period increases the borrowing cost for all REITs by 328.0 basis points, before banking relationships are taken into consideration. Banking relationships reduce this increase significantly, by 271.9 basis points, leaving an increase of 56.1 basis points for REITs with banking relationships during the crisis period. Overall, the evidence proves that banking relationships are very helpful in reducing bondholders’ required rate of return, especially during crisis periods. In addition, bond spreads decrease by about 90 basis points for investment grade rating bonds.

Table 8 Regression of bond issuance spread on REIT banking relationships
Full size table

We further examine the underpricing of SEOs after a REIT borrows a new syndicated loan. REITs with banking relationships are expected to be more transparent and be able to mitigate underpricing caused by asymmetric information. Panel A in Table 9 indicates that the underpricing for REITs without banking relationships is significantly larger than REITs with banking relationships in the year a new syndicated loan is offered as well as in the following year. The difference exists for the whole sample period as well as for just during the crisis, as shown in Panel B. We further test the regression of underpricing on banking relationships in Table 10. The evidence shows that banking relationships help enhance pricing accuracy and lower the underpricing of equity offerings during the crisis period. However, significant spread differences between REITs with and without banking relationships are not observed either before or after the crisis. During the crisis period, the degree of underpricing for REITs with banking relationships is significantly lower (13.2%) compared to REITs without banking relationships. Overall, the evidence shows that banking relationships help reduce equity underpricing. In addition, REITs with higher stock volatility are riskier and are offered at a lower price and end up with a greater underpricing.

Table 9 SEO underpricing in the years following a new syndicated loan
Full size table
Table 10 Regression of SEO underpricing on REIT banking relationships
Full size table

5 Conclusions

The evolution of REIT capital structures is an interesting issue, given access to capital markets is a critical decision for a REIT to grow. REITs are forced to access external capital markets more frequently than industrial firms since they pay out most of their earnings as dividends. Among the various capital sources, bank debt is a major channel and is increasingly favored by REITs, where syndicated loans are the most popular type applied. The ratio of syndicated loans to overall capital sources increases from only 8.94% ($85 million) in year 1989 to 45.00% ($44 billion) in year 2015. Previous literature on REIT financing sources largely focuses on public debt and equity offerings. There are very few studies on bank debt or syndicated loans, and most of them focus on REITs’ lines of credit. We investigate REITs’ syndicated loan costs and how REIT banking relationships affect the cost of capital before, during, and after the 2008–2009 financial crisis. In addition, we examine whether a new syndicated loan affects the cost of the following bond issuance and equity offering for REITs with banking relationships.

Our results show that REITs derive significant benefits from banking relationships. REITs with banking relationships have significantly lower spreads during the sample period as a whole and for all types of syndicated loans. This lower spread is also found for the periods before, during, and after the subprime crisis, individually. The results show that REITs with banking relationships obtain larger loan amounts and longer loan terms and are less required to offer collateral. More lenders and more lead banks are willing to participate in relationship loans and lead banks retain smaller shares.

After considering loan contract control variables and REIT specific control variables, for the whole sample, banking relationships result in a lowering of syndicated loan spreads by 13.53 basis points; the spread difference increases to 23.53 basis points and 23.78 basis points when the two banking relationship strength proxies, REL(Number) and REL(Amount), are applied. We also find that during the subprime crisis, the spread is increased by 103.1 basis points or 103.8 basis points depending on which banking relationship proxy is applied. Post crisis, the spread difference narrows, decreasing by only 52.72 basis points when REL(Dummy) is used or around 54 basis points when the two banking relationship strength proxies are applied.

Our evidence further shows that during the non-crisis period, REITs with banking relationships pay significantly lower spreads (8.996 basis points) to the lending banks. We find that the financial crisis increases the borrowing cost for REITs with banking relationships by 59.36 basis points, while it increases by 95.92 basis points for REITs without banking relationships. In summary, banking relationships offer significant benefits in lowering bank borrowing costs, especially during the subprime crisis.

In addition to the cost of syndicated loans, our results also provide evidence of reduced bond spreads and SEO underpricing after the granting of a new syndicated loan. During the non-crisis periods, banking relationships help reduce the borrowing cost of public debt by around 34.00 basis points. For REITs with banking relationships, the crisis increases the bond spread by only 56.10 basis points, whereas the increase in bond spread due to the crisis is 328 basis points for REITs without banking relationships. Overall, the evidence shows that banking relationships are very helpful in reducing bondholders’ required rate of return, especially during the crisis period. The evidence also shows that during the crisis period, the degree of SEO underpricing for REITs with prior banking relationships is significantly lowered (13.20%) compared to REITs without banking relationships.

Although our sample period is only up to 2015, the findings should still be relevant up to the current market. Referring to the most updated study by Dahiya et al. (2022), they find the banking relationship proxies by REL(Dummy), REL(Number), and REL(Amount) for 5,811 companies up to the year 2019 are 0.562, 0.437, and 0.421, respectively, which are higher than what we estimate for industrial firms, 0.391, 0.246, and 0.157, up to the year 2015. In addition, the overall syndicated loan size has doubled from $1.3 trillion in 2015 to $2.4 trillion in 2019. The REIT industry is very stable in terms of company numbers. There are 182 equity REITs in 2015 while the number is 179 in 2019, though the market capitalization increases from $886.5 billion to $1.246 trillion. This evidence indicates that syndicated loans and banking relationships continue to grow and become more and more important.

Overall, our empirical results support our hypothesis that banking relationships benefit REITs as they experience lower borrowing costs, and that this benefit exists for credit lines as well as for term loans. Relationship loans are always less expensive, even when an economic shock occurs, though banks adjust the spread upward during the crisis period. Finally, banking relationships also lead to lower bond yields for REITs and improve equity pricing accuracy after syndicated loans are obtained.