Property Market Liquidity and Secured or Unsecured Debt
David H. Downs; Chen Zheng; Bing Zhu
The Journal of Real Estate Finance and Economics · 2025-01-04
CC BY 4.0 · https://creativecommons.org/licenses/by/4.0/
Source: https://link.springer.com/article/10.1007/s11146-024-10005-4

Abstract

Abstract

We examine whether property market liquidity impacts the choice between secured and unsecured debt. A sample of real estate investment trusts (REITs) allows us to estimate the market liquidity of a REIT’s underlying assets and the debt secured by those assets (or unsecured). Using an instrumental variables approach, we find a positive relationship between a REIT’s property market liquidity and its use of unsecured debt relative to secured debt - when a REIT has greater exposure to more liquid underlying property markets, it is more likely to rely on unsecured debt. We investigate several aspects of this relationship including the debt level, issuances, and property loan-to-value ratio. In each case, we find support for our main result. Likewise, our results are robust to (a) using alternative instruments; (b) controlling for REITs’ unencumbered assets, as well as asset quality and redeployability; (c) controlling for credit market conditions; (d) accounting for real estate market conditions; (e) excluding firms that focus on residential real estate; and (f) adding stock market liquidity. Our study highlights the importance of property market liquidity in the debt structure of REITs.

Introduction

Introduction

Deleveraging has become a general trend in the Real Estate Investment Trust (REIT) sector after the global financial crisis (GFC) to reduce financial risk and recoup financial stability. According to the National Association of REIT’s (NAREIT) T-Tracker data on leverage (see Fig. 1), the average debt-to-book assets ratio, which peaked at 58.3% in 2008, has been declining since 2009 and has mostly stayed below 50% since 2013. This downward trend is further amplified when using the market value of assets. A less visible, but more important, change is the composition of total debt - the choice between secured and unsecured debt. In a market report published on June 28, 2019, NAREIT states that “before the financial crisis, total debt was evenly split between secured and unsecured. In 2019Q1, however, unsecured debt encompassed two-thirds of total debt for REITs.”Footnote 1 This suggests that REITs are shifting their debt choices away from secured debt, such as mortgage financing, toward unsecured debt, such as corporate bond issuance. More importantly, the report points out that “the use of unsecured debt. . increased for all but two of the REIT property sectors” between 2013 and 2019, which indicates a common trend for most sectors. This declining trend of secured debt is not REIT-sector specific, but rather a phenomenon across US corporations over the last century.Footnote 2 Arising from this trend is the underexplored question: What drives a firm’s borrowing choices between secured and unsecured debt? The objective of this study is to shed light on this question for REITs and, potentially, to gain insight into this issue for other firms.

Fig. 1

Full size image

Average leverage of all equity REITs, 2000 to 2020

For fifty years, the classic capital structure literature has treated debt homogeneously.Footnote 3 The heterogeneity of debt financing has long been recognized by theoretical research; however, only in the last decade have empirical studies begun to emerge.Footnote 4 While recent work in the debt literature has endeavored to understand the role of tangible assets in corporate debt financing, it primarily centers around assets’ collateral value and its impact on public versus private debt. Little is understood about the choices between secured and unsecured debt, which are the primary sources of REIT debt financing. Moreover, existing studies of firms, in general, overlook the fixed or immobile nature of real estate assets. The fact that real estate assets are immobile means that they are particularly susceptible to local factors, implying that the prevailing property market conditions may play a role in a REIT’s borrowing choices. We attempt to fill this gap in the literature by focusing on the property market liquidity (PML) associated with a REIT’s underlying assets. Our contention is that property market liquidity, a factor of interest to global investors,Footnote 5 is related to a REIT’s use of secured versus unsecured debt.

There are three reasons why REITs provide a unique laboratory to investigate the impact of property market liquidity on a firm’s choices between secured and unsecured debt. First, REITs are (a) required to distribute a large portion of their earnings (90% of taxable income in the US), (b) restricted from actively selling assets to fund new investments (Riddiough & Steiner, 2020), and (c) tend to hold considerably less cash than other public firms (Hardin et al., 2009), therefore they exogenously rely on external financing and possess much less market-timing power than industrial companies, mitigating the selection bias concern faced by studies on financing choices of non-REITs (Ott et al., 2005). Second, holding at least 75% real assets, REITs heavily rely on unsecured debt at the corporate level and secured debt at the asset level (Giambona et al., 2017), with relevant information reported on a regular basis (Ambrose et al., 2010; Giambona et al., 2017). On the contrary, Campello et al. (2022) show that secured debt accounts for only 7% of a firm’s investments for non-financial corporations with limited data availability. More importantly, unlike plants or machinery, the real estate assets held by REITs have a viable secondary market with heterogeneous liquidity across local markets, making it possible to construct a precise measure of a firm’s exposure to property market liquidity. Finally, to achieve unbiased estimation, it is essential to account for the heterogeneous characteristics of physical assets, firms, and industries, which has proven to be an empirical challenge for industrial firms.Footnote 6 This is not a concern for REITs because their primary purpose is to own, manage, and generate income from real estate assets, resulting in a high level of homogeneity and transparency in both firm operation and underlying assets.

We construct a precise measure of an individual REIT’s aggregate exposure to property market liquidity using a metric for the specific liquidity of local markets and the REIT’s investment allocations across different local markets. We adopt an instrumental variable approach to mitigate potential concerns about endogeneity in the PML measure, which could result from the omitted local drivers behind property market liquidity and REIT investment selection bias. We find that REITs tend to use more unsecured debt when their underlying assets are exposed to higher property market liquidity, whereas secured debt is used when property market liquidity is lower. Our results are robust after taking account of a REIT’s unencumbered assets, asset quality and redeployability, credit market conditions, real estate market conditions, stock market liquidity, as well as residential REITs.

Our study contributes to the literature in several ways. The debt literature emphasizes public versus private debt. Most of the REIT literature focuses on the capital structure or debt capacity of REITs in aggregated terms.Footnote 7 Other studies focus on the cost of debt for REITs. For example, Ibrahim and Falkenbach (2023) find that the cost of public debt is associated with both the asset type and geographic diversification of a REIT’s underlying assets. As is the case for non-REITs, a few empirical studies have also recognized asset tangibility as a key determinant of REIT choices between equity and debt (Giambona et al., 2008, 2017; Liu et al., 2019). Due to a lack of high-quality data on firm-level physical assets and debt instruments, the impact of tangible assets on corporate secured versus unsecured debt has rarely been examined.Footnote 8

Allen and Letdin (2020) show that the debt of REITs primarily comprises mortgages and notes payable. They analyze the interest rates on secured debt (mortgages) and unsecured debt (publicly traded notes) for each REIT and find that mortgage debt is the most expensive source of financing due to its non-recourse nature, which results in a strategic default option premium. Campello et al. (2022), to our knowledge, are the first and only to empirically investigate how corporate real assets affect the choice between secured and unsecured debt. Nonetheless, there is limited research on how secured and unsecured debt might be affected by underlying assets, even though these are the primary sources of debt financing for REITs. We extend this line of studies by demonstrating that property market liquidity plays a crucial role in influencing the choice between secured and unsecured debt.

Following the traditional view of Shleifer and Vishny (1992), whereby asset liquidation value affects debt capacity, the role of asset liquidity in corporate borrowing has been theoretically and empirically examined.Footnote 9 However, these empirical studies either focus on asset-level characteristics such as quality and redeployability or use indirect measures of asset liquidity at an aggregated level, resulting in inconclusive findings. Our study proposes a direct measure of a firm’s exposure to the underlying asset market.

More importantly, these studies have largely disregarded another key feature of tangible assets, particularly real estate, which is immobility, exposing these firms to local market conditions. Recent research on the risks of unsecured and secured debt contracts (Attar et al., 2019; Badoer et al., 2019; Demarzo, 2019) argues that real estate assets are exposed to local factors and could ultimately transmit local shocks to holding firms (Tuzel & Zhang, 2017). Campello et al. (2022) imply that while collateral adds security to debt, the economic risks that pledged assets are exposed to can also affect debt financing. We provide empirical evidence for this narrative: Real assets are exposed to local asset market conditions—specifically, market liquidity in our study—which in turn impacts a firm’s borrowing decisions.

The remainder of the paper is organized as follows. Property Market Liquidity and Secured vs. Unsecured Debt discusses the hypothesis development. Sample and data12 describes the data, sample selection, and variable construction. Empirical Analysis describes the methodologies and empirical analysis. Robustness Tests presents robustness tests, and Conclusion concludes.

Property Market Liquidity and Secured vs. Unsecured Debt

Property Market Liquidity and Secured vs. Unsecured Debt

We present three arguments, based on prior studies, that predict a positive relationship between property market liquidity and the use of unsecured (versus secured) debt. We also address a fourth, alternative argument as to why the relationship between PML and unsecured debt may not necessarily be positive.Footnote 10 Overall, our theoretical framework provides the basis for addressing this relationship as an empirical question.

Financial Flexibility

Physical assets can be collateralized, which reduces the risk of borrowing; consequently, firms with more collateralizable assets are more apt to use bank debt as opposed to public debt (Cvijanović, 2014; Lin, 2016). However, collateralization comes with costs - the direct costs to register the collateral and, more importantly, the indirect costs associated with the loss of financial and operational flexibility. Financial flexibility is a valuable financial management tool for addressing future cash flow, funding, and liquidity shocks; overcollateralizing, when it is not absolutely necessary, exhausts borrowing capacity and leaves little or no financial slack. Therefore, assets should be preserved as unencumbered and only used as collateral in times of hardship (see, e.g., Schwarcz (1997); Acharya et al. (2007); Li et al. (2016); Riddiough and Steiner (2020); and Lian and Ma (2021). In a recent study, Benmelech et al. (2022) find that good-quality firms have moved away from secured debt in normal times and only pledge assets against debt when it is truly needed. Since REITs hold considerably less cash than other sectors, preserving financial flexibility is even more important (Hartzell et al., 2019; Riddiough & Steiner, 2020). Secured debt within the US REIT sector is primarily non-recourse mortgage debt, offering no claim on assets other than the ones pledged as collateral. As there are often no covenants associated with this kind of mortgage debt, it is in theory an “unlimited” source as long as there are additional assets, which can potentially lead to dangerously high leverage.Footnote 11 Unsecured debt used by REITs, on the other hand, often incorporates various covenants that require the total leverage to remain below a certain threshold (usually 60%) and restrict the over-collateralization of real estate assets, which implicitly preserves the financial flexibility. Campello et al. (2022) find that firms raise more unsecured debt with an increase in their real estate asset value.

Operational Flexibility

The loss of operational flexibility of collateralized assets is also a concern. Once collateralized, the borrower often needs the creditor’s permission to make alterations to or dispose of the asset (Mello & Ruckes, 2017). In contrast, unsecured debt pools a company’s real assets and uses them only as implicit collateral. The cross-collateralization nature of unsecured debt can generate not only more debt capacity but also enhance operational flexibility (Giambona et al., 2017; Riddiough & Steiner, 2020). For these reasons, unsecured debt is often considered to be a positive signal of REIT quality by rating agencies. Giambona et al. (2017) find that firms that use more unsecured debt present better quality; and Conklin et al. (2018) show that REITs avoid secured debt as a funding source for investments in their primary markets in order to retain operational flexibility and a good credit profile; Feng and Wu (2023) observe a positive relationship between the extent of ESG disclosure and the unsecured debt to total debt ratio in REITs. Considering the financial and operational flexibility, we expect REITs facing good firm and market conditions, i.e., owning assets in liquid markets in our study, to prioritize unsecured debt.

Liquidation Costs

Shleifer and Vishny (1992) argue that the liquidation value of tangible assets positively impacts a firm’s debt capacity. Liu et al. (2019) test this argument in the REIT context and find that REITs with better asset quality are more likely to issue corporate bonds as opposed to equity. Considering the dispersed ownership of public debt, Bolton and Scharfstein (1996) and Bolton and Freixas (2000) argue that the liquidation cost associated with bond financing is higher during times of financial distress because firms have less ability to negotiate with bondholders than with banks. Consequently, bondholders may force firms to rush into inefficient or sub-optimal asset liquidations, whereas banks are more willing to negotiate and reorganize the debt. Building on this line of research, Cantillo and Wright (2000) propose a liquidation cost channel and argue that firms with more illiquid assets have even less negotiating power with bondholders because it is more difficult and takes longer to dispose of these assets; consequently, these firms suffer from higher potential costs of liquidation. Supporting this argument, Cantillo and Wright (2000); Denis and Mihov (2003); Rauh and Sufi (2010); and Chen et al. (2020) find that firms with lower real asset value or asset redeployability tend to use more bank debt. The liquidation cost argument suggests that firms facing low property market liquidity will prefer secured debt to unsecured debt. Our argument is also consistent with recent findings that secured debt is countercyclical in nature (Benmelech et al., 2022), whereas unsecured debt is procyclical (Luk & Zheng, 2022). Moreover, when market liquidity declines the information environment becomes more opaque making it harder to gauge the value; in that case, firms would benefit from negotiation with banks.

Cost of Debt

An alternative argument is that when property markets become more liquid, which has a positive effect on the liquidation value, the cost of secured debt might be lower due to increasing collateral value. Consequently, one might expect a positive relationship between property market liquidity and secured debt. However, when a REIT’s liquidation value benefits from higher market liquidity, the cost of unsecured debt could also drop because unsecured debt is recourse debt secured by the total assets of the firm. By analyzing the interest rates on secured and unsecured debt for each REIT, Allen and Letdin (2020) find that mortgage debt is, on average, much more expensive than unsecured debt, due to the non-recourse nature of secured debt that results in a strategic default option premium. Campello et al. (2022) also find that firms rely more on unsecured debt when their real estate asset value increases. Therefore, how property market liquidity affects the level of unsecured and secured debt via the cost of debt channel is inconclusive. As such, the PML and debt relationship due to the cost of debt is an open empirical question.

In summary, and weighing the four arguments behind our theoretical framework, we hypothesize a positive (negative) relationship between property market liquidity and unsecured (secured) debt.

Sample and Data 

Sample and Data

12

We construct a composite measure of property market liquidity (PML) for each REIT in our sample based on the REIT’s underlying portfolio of properties held and the location-specific market liquidity of each holding. Our liquidity indices are constructed by van Dijk et al. (2022) using Real Capital Analytics (RCA) Footnote 13 transaction data from 2005 to 2018. Fisher et al. (2003); Fisher et al. (2007) propose that the demand-side reservation price index serves as a “constant-liquidity” value index for the asset market. van Dijk et al. (2022) extend their methodology to a repeat-sales framework that separately tracks changes in reservation prices on the demand (potential buyers) and supply (potential sellers) sides of the asset market. van Dijk et al. (2022) also maintain that sellers’ sentiment could also drive market liquidity. Specifically, liquidity is defined as the difference between buyer and seller reservation price indexes. Analogous to a bid-ask spread, this yields an approximation of the trading costs an investor would pay to execute a trade, in addition to taxes and fees. Unlike the method of Fisher et al. (2003); Fisher et al. (2007), which relies on rich hedonic data on property characteristics, the repeat-sale framework used by van Dijk et al. (2022) enables disentangling the reservation prices of buyers and sellers at city level. Based on the RCA property transaction database, which captures 90% of all commercial transactions in the U.S., van Dijk et al. (2022) estimated property market liquidity across 31 U.S. regions. Footnote 14

Compared with volume- or time-based measures, this transaction-cost-based measure reveals the potential loss in the liquidation value of properties and therefore suits our research question. Therefore, we use the RCA liquidity index constructed by van Dijk et al. (2022) as the base for our sample and focus our debt data on the period between 2005 and 2018. Due to missing values in other explanatory variables, we have 142 REITs in our sample, forming an unbalanced panel dataset with 1,071 firm-year observations.

REIT Debt

We collect annual data on unsecured, secured and total debt of REITs in our sample from S&P Global Market Intelligence, which is a comprehensive database that collects REIT firm- and asset-level information. The unsecured debt in this study is recorded as Senior Unsecured Debt in the database, which is all debt for which the lender does not have recourse against specific collateral owned by the borrower, including both long-term senior notes and senior term loans. Secured debt is recorded as Mortgage Debt and Notes in the database, which is all senior debt secured by real estate assets, excluding any lines of credit that may have real estate backing.Footnote 15

Figure 2 presents the average leverage and unsecured debt ratio (as a share of total debt) against PML based on our sample from 2005 to 2018. We observe a declining trend in total debt, from the highest level of 58% to below 45% over the 10 years since the GFC, as well as a sharp increase in the unsecured-to-total debt ratio of around 20% over our sample period, which conforms to the wider sector trend; moreover, the unsecured debt ratio and PML reveal a positive linear relationship, consistent with our conjectures. In addition to the data on debt levels, we collect data on REIT security issuances and identify unsecured credit, unsecured bond, and secured credit issuances. To examine the level of collateralization, we collect property transactions executed by REITs in our sample and measure the property-level loan-to-value (LTV).

Fig. 2

Full size image

Property Market Liquidity, Leverage, and Unsecured Debt: REIT Sample, 2005 to 2018

REIT Property Market Liquidity - An Instrumental Variables Approach

Given that we can retrieve property-specific information on REIT portfolios - i.e., the size and location of individual properties - we construct a measure of a REIT’s property market liquidity (PML) based on the property portfolio of each REIT and the liquidity of the 31 local property markets covered by the RCA liquidity data, as in Eq. (1):

$$\:{PML}_{i,t}=\sum\:_{m=1}^{M}{w}_{m,i,t}{PML}_{m,t},$$

(1)

where \(\:{PML}_{i,t}\) represents the property market liquidity of REIT i in year t. \(\:{PML}_{m,t}\) is the property market liquidity of region m in year t and \(\:{w}_{m.i,t}\) represents the share of properties of REIT i in region m in year t weighted by the adjusted cost of the building. The adjusted cost is the maximum of (1) the reported book value, (2) the initial cost of the property, or (3) the historic cost of the property (Ling et al., 2021b). For a given REIT i, \(\:{w}_{m.i,t}\) is calculated as the cost of properties located in market m to the cost of all properties owned by REIT i;Footnote 16 location data on properties are extracted from the S&P Global Market Intelligence database.

We recognize that endogeneity issues may arise from three sources. First, it is possible that REIT management might demonstrate a preference for investments in less risky and more liquid real estate markets (Ghent, 2021). Thus, the investment weight\(\:\:\left({w}_{m.i,t}\right)\) may be influenced by local property market liquidity, leading to an overestimation of the relationship between local property market performance and debt choice due to self-selection bias. To achieve identification, we address these concerns by using an instrumental variables approach. Specifically, the home bias theory (Coval & Moskowitz, 2001) implies that the distance between assets and a firm’s headquarters is a key determinant of investment allocation. In the cases of REITs Garmaise and Moskowitz (2004); Ling et al. (2021a) find that market participants tend to choose local investments to mitigate information asymmetry. Following this line of thought, we use the distance between a property and a REIT’s headquarters as an instrumental variable for investment allocations (see Eq. 2). The geographic distance between a particular asset and headquarters is independent of the local property market liquidity. Thus, using geographic distance as the instrument can address the concern that investment weights may be influenced by the local property market liquidity:

$$\:{\widehat{w}}_{m,i,t}=\widehat{p}+\widehat{q}\text{ln}{D}_{m,i,t}+\widehat{\vartheta\:}{X}_{i,t-1}+{\widehat{\phi\:}}_{i}+{\widehat{\omega\:}}_{t}\:$$

(2)

where \(\:{\widehat{w}}_{m,i,t}\) represents the instrumented investment weights of REIT i in region m in year t weighted by the adjusted cost of the building. \(\:{D}_{m,i,t}\) is the average distance of individual properties located in region m from the headquarters of REIT i. \(\:\widehat{p}\) and \(\:\widehat{q}\) are estimated coefficients. \(\:{X}_{i,t-1}\) represents all control variables that are included in the second-stage regression, lagged by one year. \(\:{\widehat{\phi\:}}_{i}\) and \(\:{\widehat{\omega\:}}_{t}\) capture the firm and year fixed effects, respectively.

However, we also acknowledge that geographic distance is not completely independent of debt choice, as investing in markets with a longer distance from the home market can lead to information asymmetry and higher risk in foreign markets. This may, in turn, influence REIT debt choices through the cost of debt and the firm’s financial performance. Indeed, as shown by Demirci et al. (2018); Milcheva et al. (2021); and Ling et al. (2021b), geographic diversification or proximity to the home market influences the cost of debt and REIT financial performance. To address this concern, we add geographic concentration (HHI) and the REIT weighted average cost of debt (Interest Rate) as a control variable in all specifications. We argue that our instrument is exogenous when the effects of home bias on the cost of debt and geographic diversification on firm performance are controlled for. Footnote 17

Second, local property market liquidity (\(\:{PML}_{m,t}\)) could be impacted by latent local factors that also affect a firm’s financing decisions, leading to omitted variable bias (Lin, 2016). Finally, local property market liquidity could also respond to local REIT investment activities, leading to reverse causality (Campello et al., 2022). To address this issue, we also use instrumental variables for local property market liquidity.

Saiz (2010) argues that local real estate markets are exogenously affected by land supply elasticity. Campello et al. (2022) argue that both land supply elasticity and the national mortgage rate are plausible instrumental variables for local market conditions. As a macroeconomic variable, the national mortgage rate is less likely to be driven by individual mortgage rates in local markets; however, it does reflect common trends. Therefore, we argue that the national mortgage rate is an exogenous instrumental variable conditional on controlling for time fixed effects. Since land supply elasticity is time-invariant, we follow Campello et al. (2022) and use a joint function of land supply elasticity and the national mortgage rate (see Eq. 3).Footnote 18

$${\widehat{PML}}_{m,q}=\widehat gElasticity_mMort_q+\widehat{\mu\:}Mort_q+{\widehat{\alpha\:}}_m+{\widehat{\omega\:}}_t\:$$

(3)

where \(\:{\widehat{PML}}_{m,q}\) represents the instrumented property market liquidity of region m in quarter q, \(\:Elasticit{y}_{m}\) stands for the Wharton Residential Land Use Regulation Index (WRLURI 2010) of region m, \(\:Mor{t}_{q}\) stands for the national 30-year mortgage rate, \(\:\widehat{g}\) and \(\:\widehat{\mu\:}\) are estimated coefficients, and \(\:{\widehat{\alpha\:}}_{m}\) and \(\:{\widehat{\omega\:}}_{t}\) capture region and year fixed effects, respectively. We then aggregate the quarterly \({\widehat{PML}}_{m,q}\) to the \({\widehat{PML}}_{m,t}\) in region m in year t.

In estimating Eq. (2), we find a negative and statistically significant coefficient for distance (confidence at the 1% level), confirming that REITs allocate assets close to their headquarters. Equation (3) also yields significant results (both confidence at the 1% level), suggesting that local property market liquidity increases with restricted supply and lower mortgage rates. The F-statistics for the instruments in Eqs. (2) and (3) are 7,107 and 9.5 respectively, both of which are significant at the 1% level, confirming the instruments’ validity.Footnote 19

Based on the instrumented investment weights and property market liquidity, we calculate the instrumented average PML for each REIT (PMLT), as in Eq. (4):

$$\:{PML}_{i,t}^{T}=\sum\:_{m=1}^{M}{\widehat{w}}_{m,i,t}{\widehat{PML}}_{m,t}$$

(4)

where \(\:{PML}_{i,t}^{T}\) represents the instrumented property market liquidity of REIT i in year t. \(\:{\widehat{w}}_{m,i,t}\) and \(\:{\widehat{PML}}_{m,t}\) are as defined by Eqs. (2) and (3), respectively.

Firm- and Asset-Level Control Variables

In addition to PML, it is crucial to account for firm characteristics that could drive debt structure outcomes. We include traditional control variables that affect corporate financing decisions, including total leverage, debt maturity, cost of debt, firm size proxied by market capitalization, market-to-book ratio, earnings volatility, and cash flows proxied by funds from operations (FFO). These traditional control variables are widely used in studies on REIT debt financing (see, e.g., Giambona et al. (2008); Harrison et al. (2011); Giambona et al. (2017); Conklin et al. (2018); Riddiough and Steiner (2020); and Feng and Wu (2023).

We follow Riddiough and Steiner (2020) and include investment growth to control for the external financing shortfall, because investment growth predominantly drives the need for external financing. Following Harrison et al. (2011), we include drawn lines of credit as a control variable for REIT financing characteristics. We also control for the geographic diversification of the underlying assets of REITs by using the Herfindahl index, because previous studies find that the cost of debt might be affected by the asset geographic diversification (Ibrahim & Falkenbach, 2023; Ling et al., 2021b). Finally, we control for investment grade rating and REIT status.Footnote 20 We also control for various asset-level characteristics with a detailed discussion provided later in our robustness tests section.

Table 1 reports the summary statistics. Our sample of REITs presents an average PML of 0.003, ranging from − 0.327 to 0.146, with a standard deviation of 0.14.Footnote 21 The average unsecured and secured debt levels are 62.5% and 46.2%, respectively. Among the 1,983 security issuances by REITs in our sample, 39.7% are unsecured debt issuances, and 8% are secured debt issuances. Our sample includes 3,572 property acquisitions with, on average, 59.4% LTV, $51.4 million value, and 9.5% net operating income (NOI)-to-assets ratio. The balance of the summary statistics is as expected for a REIT sample.Footnote 22

Table 1 Summary statistics

Full size table

Empirical Analysis

Empirical Analysis

Fixed-Effects Estimation

We begin by testing the effect of a REIT’s exposure to local property market liquidity (i.e., PML) on the level of its unsecured and secured debt. The estimation is based on an unbalanced panel regression with fixed effects, as in Eq. (5):

$$\:{y}_{i,t}=\alpha\:{PML}_{i,t-1}^{T}+\beta\:{X}_{i,t-1}+{\phi\:}_{i}+{\delta\:}_{t}+{e}_{i,t}$$

(5)

where \(\:{y}_{i,t}\) stands for unsecured-to-total and secured-to-total debt ratios, respectively. All explanatory variables are as defined in Sample and data12 and lagged by one year. We use bootstrapping to generate the standard errors of \(\:{PML}_{i,t-1}^{T}\), i.e., we bootstrap Eqs. (2) and (3) using the bootstrapped \(\:{\widehat{w}}_{m.i,t-1}\) and \(\:{\widehat{PML}}_{m,t-1}\:\)to generate the standard error of α.Footnote 23\(\:\beta\:\) is a vector of coefficients of the control variables. We include firm, and year fixed effects in all specifications.

Estimated results are reported as Models 1 and 2 in Table 2, Panel A. Model 1 shows a 0.34 coefficient for PMLT, significant at 1%, suggesting a positive relation between property market liquidity and the unsecured-to-total-debt ratio. In contrast, the estimate in Model 2 provides evidence of a negative relationship between PML and the level of secured debt.Footnote 24 This supports our argument that a REIT relies more on unsecured debt when its PML is higher and instead finances with secured debt when the PML drops. Our results are economically significant, whereby a one-standard-deviation increase in the level of PMLT is associated with a 4.7 (4.7) percentage-point increase (decrease) in unsecured (secured) debt as a share of total debt. Given the average of $1,835 million unsecured debt and $1,104 million secured debt of our sample REITs, a one-standard-deviation increase in the level of PML translate into an $86 million increase in unsecured debt and a $52 million decrease in secured debt.

Table 2 Property market liquidity and debt structure

Full size table

Property market liquidity and debt structure (continued)

Panel B: Dependent Variable – Weighted Average PML

Fixed Effects

First Difference

Dependent Variable

Model 1

Unsecured Debt

Model 2

Secured Debt

Model 3

\(\:\varDelta\:\)Unsecured Debt

Model 4

\(\:\varDelta\:\)Secured Debt

PML

0.2700***

−0.2549***

0.1035***

−0.0688**

(0.0506)

(0.0585)

(0.0276)

(0.0314)

Leverage

−0.2471***

0.1382*

−0.1393**

0.1479*

(0.0700)

(0.0740)

(0.0684)

(0.0778)

Log(Maturity)

0.0155

−0.0291

0.0508***

−0.0732***

(0.0251)

(0.0244)

(0.0176)

(0.0200)

Interest Rate

−0.0270***

0.0411***

0.0208***

−0.0082

(0.0082)

(0.0087)

(0.0069)

(0.0078)

Log(Size)

0.0249**

−0.0132

0.0135

−0.0092

(0.0122)

(0.0125)

(0.0141)

(0.0161)

M/B

−0.0167**

0.0084

−0.0033

0.0068

(0.0073)

(0.0071)

(0.0061)

(0.0069)

σ(Earning)

−0.5647

0.7785**

0.1247

−0.4793

(0.3986)

(0.3783)

(0.4037)

(0.4589)

FFO

0.4441

−0.4424

0.1448

−0.0002

(0.3704)

(0.3558)

(0.1845)

(0.2097)

Growth

−0.0265**

0.0174

0.0120

−0.0200*

(0.0112)

(0.0110)

(0.0106)

(0.0120)

HHI

0.1252*

−0.1037

−0.0338

0.0187

(0.0736)

(0.0793)

(0.1430)

(0.1626)

LOC

0.0208

−0.0483**

0.0055

0.0057

(0.0205)

(0.0226)

(0.0132)

(0.0151)

Rated

0.1839***

−0.1975***

0.0072

−0.0116

(0.0182)

(0.0178)

(0.0163)

(0.0185)

REIT

−0.1479***

−0.0255

−0.1000*

0.0623

(0.0496)

(0.0489)

(0.0511)

(0.0581)

Year FE

Yes

Yes

Yes

Yes

Firm FE

Yes

Yes

No

No

No. of obs.

1074

1074

933

933

Adj. R2

0.8604

0.8583

0.2476

0.1842

This table reports the results of the unbalanced panel regression with fixed effects in Models 1 and 2, and the results of first-difference regression in Models 3 and 4. In Panel A, PMLT stands for the instrumented weighted average liquidity of the underlying asset markets for each REIT. In Panel B, PML stands for the weighted average liquidity of the underlying asset markets for each REIT. In Models 1 and 2, the dependent variable is unsecured debt and secured debt as a share of total debt, respectively. Control variables include total leverage (Leverage), debt maturity (Maturity), interest rate (Interest Rate), market value (Size), market-to-book ratio (M/B), earning volatility (σ(Earning)), FFO-to-asset ratio (FFO), real estate investment growth rate (Growth), geographic concentration (HHI), drawn credit line (LOC), REIT credit rating (Rated), and REIT status dummy (REIT) Detailed variable definitions are presented in Table 1. In Models 3 and 4, both the dependent and explanatory variables take the form of first difference. All explanatory variables are lagged by 1 year. Year and firm fixed effects are also included. The standard error for PML is measured using bootstrapping. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively

We also find that larger REITs, investment-grade REITs, and REITs that generate higher cash flows have higher unsecured-to-total debt ratios. These results are consistent with the findings of Giambona et al. (2017), Conklin et al. (2018), and Liu et al. (2019) all of which imply that good-quality firms are more likely to issue unsecured debt.

First-Difference Estimation

Although the fixed effects model controls for the unobserved firm characteristics (\(\:{\phi\:}_{i}\) in Eq. 5), it is possible that time-invariant unobserved firm characteristics could simultaneously affect the observable firm-level determinants of unsecured and secured debt. In other words, \(\:{X}_{i,t-1}\) and \(\:{\phi\:}_{i}\) could be highly correlated. To address this concern, we use the first-difference estimator to purge time-invariant unobserved firm characteristics (\(\:{\phi\:}_{i})\) and estimate how the change in the level of unsecured (secured) debt responds to the change in REIT PMLT (ΔPMLT), as in Eq. (6):

$$\:{\varDelta\:y}_{i,t}=\alpha\:\varDelta\:{PML}_{i,t-1}^{T}+\beta\:{\varDelta\:X}_{i,t-1}+{\delta\:}_{t}+{\varDelta\:e}_{i,t}$$

(6)

where \(\:{\varDelta\:y}_{i,t}\) represents the incremental changes in unsecured-to-total debt and secured-to-total debt ratios, respectively. \(\:\varDelta\:{PML}_{i,t-1}^{T}\) is the difference in a firm’s instrumented PML (PMLT) between year t-1 and t-2. \(\:{\varDelta\:X}_{i,t-1}\) stands for the difference in other explanatory variables between year t-1 and t-2. We also control for year fixed effects, \(\:{\delta\:}_{t}\).

Results are reported as Models 3 and 4 in Table 2, Panel A. We find that a 10% increase in PMLT is associated with a 1.2% increase in the unsecured-to-total debt ratio (significant at the 1% level) and a 0.8% decrease in the secured-to-total debt ratio (significant at the 10% level). The first-difference results are in line with the fixed-effects estimation results in Models 1 and 2. As expected, we find that the change in unsecured or secured debt also responds to changes in the overall financing conditions, such as total leverage, debt maturity, and interest rates. Specifically, our results show that the unsecured (secured) debt level declines (increases) when total leverage increases, consistent with Riddiough and Steiner (2020), which argues that unsecured debt comes with covenants that restrict a REIT’s total leverage. Once the debt capacity is exhausted, it is more difficult to raise unsecured debt due to the increased risk and, alternatively, secured debt becomes preferable. Similarly, as debt maturity lengthens overall credit health improves making it easier to access unsecured debt.

Table 2 begins by presenting the results (i.e., Panel A, all models) that allow for identification, because we believe that controlling for endogeneity is a potential concern. In addition, we estimate the same models using the simple average property market liquidity (PML) weighted by the investment shares of a REIT in the 31 regions instead of the instrumented variable. Results are reported in Table 2, Panel B, and are very similar to those in Panel A. Overall, the results in Table 2 support our hypothesis.

Debt Issuances

Next, we examine the effect of the property market liquidity of REITs on their debt issuances in the capital markets. We start with 1,983 security issuances and then manually identify whether an issuance is of secured or unsecured debt - an issuance is regarded as unsecured or secured only if it is clearly stated so, leaving us with 788 unsecured debt issuances and 158 secured debt issuances. We then create two dichotomous variables for secured debt issuance and unsecured debt issuance. The secured (unsecured) debt issuance dummy variable equals 1 if the issuance is secured (unsecured) debt and 0 for all other issuances, such as equities. To control for the fact that REITs are more likely to raise external capital in a year when they make property investments, we create a categorical variable (Invest) which equals 1, −1, or 0, if a REIT has a positive, negative, or zero net investment (acquisitions minus dispositions) in a given year, respectively. We estimate the likelihood of issuance with a probit model as in Eq. (7):

$$\:{Prob(Issue}_{i,t})=\rho\:{PML}_{i,t}^{T}+\beta\:{Invest}_{i,t}+\gamma\:{PML}_{i,t}^{T}X{Invest}_{i,t}+\varrho\:{X}_{i,\:t-1}+{\phi\:}_{i}+{\delta\:}_{t}+{\epsilon\:}_{i}$$

(7)

where \(\:{Issue}_{i,t}\) represents either an unsecured or secured debt issuance. We also include an interaction term between PML and the investment variable (\(\:{PML}_{i,t}^{T}X{Invest}_{i,t}\)). Control variables are defined as in the previous section, lagged by one year. \(\:{\phi\:}_{i}\) and \(\:{\delta\:}_{t}\) represent firm and year fixed effects, respectively.

The results are reported in Table 3, where the dependent variables are unsecured and secure debt issuances in Panel A and B respectively. Models 1 and 3 are base models that include PMLT and the control variables; Models 2 and 4 include the Invest dummy, its interaction with PMLT, and the controls. Although we lose some observations due to missing data, the results in Models 1 and 2 reveal a positive relationship between a REIT’s PML and the likelihood of its issuing unsecured debt, which provides further support to our main results. This finding is in line with Liu et al. (2019) and Chen et al. (2020) who argue that firms are more likely to issue bonds when their real assets have higher liquidation value. The coefficients of PMLT on secured debt issuances are insignificant, which may not be surprising given that most of the security issuances are unsecured bonds and we only observe 158 secured debt issuances. As expected, when a REIT makes a net property investment (Invest) in a year the likelihood of it raising debt capital increases; meanwhile, no moderation effect is detected on PML.

Table 3 Property market liquidity and security issuance

Full size table

Property Acquisition LTV

Secured debt of REITs is primarily mortgages that are collateralized by the underlying real estate assets. REITs with more secured debt—hence, more mortgage debt—are expected to report a higher loan-to-value (LTV) on their properties. On the other hand, when the property market liquidity for a REIT’s properties is higher, we would expect less secured debt, less mortgage financing and, consequently, lower reported LTV at the property level. To test this, we estimate the following model as in Eq. (8):

$$\:{LTV}_{k,t}=\psi\:{\widehat{PML}}_{m,t}+\gamma\:{X}_{k,t}+{\phi\:}_{i}+MS{A}_{k}+{\delta\:}_{t}+{\zeta\:}_{k}+{\epsilon}_{k,t}$$

(8)

where \(\:{LTV}_{k,t}\) stands for the LTV ratio of property k at the time of acquisition. \(\:{\widehat{PML}}_{m,t}\) stands for instrumented PML in region m where the property is located in the year of the transaction t. \(\:{X}_{k,t}\) are control variables for property characteristics that include net operating income as a share of total assets (NOI) and the building’s value. \(\:{\phi\:}_{i}\), \(\:MS{A}_{k}\), \(\:{\delta\:}_{t}\), and \(\:{\zeta\:}_{k}\) represent firm, MSA, time, and property type dummies, respectively. Results are reported in Table 4. Models 1, 2, and 3 are the base models, and include only \(\:\widehat{PML}\) and control for different dummies; Model 4 also includes property characteristics as well as year, firm, property, and MSA dummies. Consistent with our conjectures as stated above, we find that property transactions in regions with higher PML on average present lower LTVs. Specifically, a practical interpretation of Model 4 is that a one-standard-deviation increase in \(\:\widehat{PML}\) is associated with a 17.6% decrease in the LTV ratio. Given an average of 59.4% LTV with a 16% standard deviation in our sample, this effect is economically significant, reinforcing our argument that REITs are less likely to collateralize their assets to raise secured debt when the underlying property markets reflect high liquidity.

Table 4 Property market liquidity and property acquisition LTVs

Full size table

Robustness Tests

Robustness Tests

Alternative Instruments

To address concerns related to using distance to headquarters as the instrument for investment weights \(\:\left({w}_{m.i,t}\right)\), we apply an approach similar to the Bartik instrument. We use the initial weights at the beginning of the sample period, either in 2005 or the first year for which the firm has property location information, and keep these weights constant over time. We then construct an alternative instrumented property market liquidity (\(\:{PML}_{i,t}^{TB}\)), using the initial property allocation weights and the instrumented property market liquidity as follows:

$$\:{PML}_{i,t}^{TB}=\sum\:_{m=1}^{M}{w}_{m,i,1}{\widehat{PML}}_{m,t}$$

(9)

where \(\:{w}_{m,i,1}\) represents the initial weight of the investment allocation of firm i in market m.

Results based on this Bartik-type instrument are reported in Table 5, Columns 1 and 2. These results, which are based on an alternative instrument, are robust relative to our baseline results reported in Table 2. The coefficient for unsecured debt slightly decreases from 0.3361 in the baseline model to 0.3229 in Table 5 and from − 0.3324 in the baseline model to −0.3251 for secured debt. This suggests that the estimated effect is dominantly driven by changes in the underlying property market liquidity rather than by shifts in portfolio allocations.

Table 5 Robustness test—alternative instruments

Full size table

Also, in our baseline, we use the quarterly mortgage rate while including year fixed effects. We also run a specification without year fixed effects. Results for the first-stage regression of property market liquidity are reported in Appendix A, Table A2, Column 3. Since the coefficient for \(\:Elasticit{y}_{m}Mor{t}_{q}\) in Eq. (3) becomes insignificant, we retain only the mortgage rate as the instrument. Given the statistical significance reported in Table A2, Column 3, the mortgage rate is a valid instrument. We then construct an additional alternative instrument (\(\:{PML}_{i,t}^{TA})\) using the re-estimated property market liquidity (\(\:{\widehat{PM{L}^{A}}}_{m,t})\) as follows:

$$\:{PML}_{i,t}^{TA}=\sum\:_{m=1}^{M}{\widehat{w}}_{m,i,1}{\widehat{PM{L}^{A}}}_{m,t}$$

(10)

Results based on this alternative instrumented property market liquidity are shown in Table 6, Columns 3 and 4. The estimated coefficients are noticeably higher than the baseline results, in which we also use the land supply elasticity as a valid instrument. This indicates that using only national monetary policy as the instrument may lead to overestimation of the impact. The IV-estimated coefficient captures the local average treatment effect. For REITs with assets concentrated in markets that are more sensitive to national monetary policy, the relationship between debt choices and underlying asset market liquidity for these firms may be stronger than those that focus on less responsive markets, given the close link between monetary policy and debt decisions. As a result, instruments beyond national monetary policy that can explain variations in liquidity across markets are necessary to avoid overestimating the treatment effect.

Maintenance of Unencumbered Assets

Unsecured debt is implicitly secured by all unencumbered assets, and contracts often limit the amount of total debt a REIT can use. Based on a sample period from 1992 to 2010, Giambona et al. (2017) find that REITs that issue unsecured debt usually have covenants in contracts that restrict total leverage to between 45% and 65%. Our variable of the previous year’s leverage controls for such a situation. In general, another common way to protect unsecured debtholders is to have negative pledge covenants that limit the issuance of secured debt, so that unsecured debtholders are not disadvantaged by other claimants with higher priority. In the REIT context, this is made possible by the requirement that unencumbered assets must be maintained - i.e., REITs must keep a certain amount of unencumbered assets in place in order to cover the principal of unsecured debt by the end of each quarter. Demirci et al. (2023) find that unencumbered assets play a crucial role in REIT asset sales and cost of borrowing. Giambona et al. (2017) document that unsecured debt to total assets should normally be no less than 60% for REITs, which means that a REIT cannot issue more unsecured debt if there are not enough unencumbered assets in place, regardless of property market liquidity. To consider the influence of unencumbered assets, we add a control variable defined as the previous year’s unsecured debt as a share of total assets to proxy for the level of unencumbered assets. As expected, the results reported in Table 6 reveal a positive (negative) relationship between the level of unencumbered assets and the level of unsecured (secured) debt, while our main results on PML are robust.

Table 6 Robustness test—controlling for unencumbered assets

Full size table

Asset Quality and Redeployability

While property market liquidity impacts asset liquidation values, heterogeneous asset characteristics mean that individual assets could respond to local market liquidity differently. To confirm that the effect of PML is on top of the effect of having better assets, we use three proxies to control for asset quality and redeployability. First, we capture asset quality by using the tenant credit and the fire sale discount. Following Lu-Andrews (2017) we obtain annual S&P credit ratings of tenants, ranging from AAA to C and containing 20 ratings in total in our sample. We then create a categorical variable by assigning the value 0 to C, 1 to CC, 2 to CCC-, etc., up to 20 to AAA. The overall tenant quality of each REIT is weighted by the revenue shares from its tenants.

Second, when asset liquidation is needed for a financially constrained REIT whose underlying market is highly concentrated, its potential buyers might face the same financial distress and result in a rapid sale with a greater liquidation discount. Following Liu et al. (2019), who propose the industry concentration of a REIT’s economic base as a proxy for the fire-sale discount, for a given REIT we construct the industry Herfindahl index for MSA-level employment concentration weighted by the number of properties a REIT holds in each MSA. Finally, following Benmelech et al. (2005) and Giambona et al. (2008), we use the recovery rate index to proxy for asset redeployability, which reflects the flexibility of converting a current asset to suit an alternative owner/user. We define the recovery rate index as the weighted average recovery rate based on an S&P study of CMBS recoveries by property type and the REIT’s property type portfolio; following the S&P report (Standard and Poor’s, 2005), the recovery rate is as 70.2% for residential apartments, 61.1% for offices, 60.5% for industrial properties, 59.8% for retail properties, and 51.5% for hotels and motels. Results for unsecured and secured debt are reported in Panels A and B of Table 7, respectively. We find that the impact of REIT PML on unsecured and secured debt financing outcomes remains strongly significant and consistent with the main results in Table 2, mitigating the concern that our results are purely driven by asset quality and redeployability. This extension of our main results also reinforces the argument that, in addition to asset quality and redeployability, PML is a fundamental determinant of REIT debt structure.

Table 7 Robustness test—controlling for asset quality and redeployability

Full size table

PML and Credit Market Conditions

To address the concern that both the debt financing choice by REITs and PML can be influenced by the monetary policy and credit availability, we control for credit market conditions by subtracting the impact of the interest rate, term spread, and willingness to lend of domestic banks from local property market liquidity, as in Eq. (11):

$$\:{\varDelta\:PML}_{m,t}={a}_{m}+{\sum\:}_{q=1}^{4}{b}_{m,q}\varDelta\:{ir}_{t-q}+{\sum\:}_{q=1}^{4}{c}_{m,q}\varDelta\:{sp}_{t-q}+{\sum\:}_{q=1}^{4}{d}_{m,q}\varDelta\:{lend}_{t-q}+\varDelta\:{e}_{m,t}$$

(11)

where \(\:{ir}_{t}\) stands for the federal base rate in period t, \(\:{sp}_{t}\) is the spread between 10-year government bonds and 3-month government bonds, and \(\:{lend}_{t}\) is the tightening standards for business loans from the Senior Loan Officer Opinion Survey. q is the number of lags, and we consider a maximum of 4 lags. \(\:{a}_{m}\), \(\:{b}_{m,q}{,\:c}_{m,q}\), and \(\:{d}_{m,q}\) are the corresponding parameters for region m. \(\:{e}_{m,t}\) captures the idiosyncratic part of the PML that is not affected by aggregated credit market conditions, and thus the PML that accounts for the aggregated credit market conditions (PMLT, e) can be calculated as in Eq. (12):

$$\:{PML}_{i,t}^{T,\:e}=\sum\:_{m=1}^{M}{\widehat{w}}_{m,i,t}{e}_{m,t}$$

(12)

where \(\:{PML}_{i,t}^{T,\:e}\) stands for the instrumented PML of REIT i in year t. \(\:{\widehat{w}}_{m,i,t}\) are the instrumented investment weights, and \(\:{e}_{m,t}\) is the residual value from Eq. (11).

Results are reported in Table 8. We can see that both the coefficients and significance levels remain largely consistent with the main results in Table 3, while the explanatory power of PMLT improves after controlling for credit market conditions.

Table 8 Robustness test—alternative measures of property market liquidity

Full size table

Real Estate Market Conditions and Location Effects

Liquidity in local property markets might be affected by general business conditions, i.e., higher liquidity when the real estate market is booming (Krainer, 2001). Recent studies in the general finance literature suggest that firms tend to rely more on unsecured debt in good business conditions (Benmelech et al., 2022; Campello et al., 2022). To show that the effect of PML is on top of the real estate market cycles, we control for the prevailing real estate market conditions with three variables including the average return (Return), systematic risk (Beta)Footnote 25 and the geographically weighted risk-adjusted return (Alpha).Footnote 26 In particular, for each REIT, we calculate the average return, systematic risk, and the average excess return of all local markets where the REIT’s properties are located weighted by its investment allocations in these markets. Footnote 27 Models 1 to 3 in Table 9 present results controlling for real estate market Return, Beta, and Alpha, respectively. We find a negative (positive) relationship between unsecured (secured) debt and real estate market beta, supporting the argument by Benmelech et al. (2022) that unsecured debt is preferred in good business conditions.

Table 9 Robustness test—controlling for real estate market conditions and MSA location effects

Full size table

In Models 4 and 5, we control for additional location effects. In Model 4, we include a variable 25Getaway to control for the share of properties of a firm located in the 25 getaway markets; in Model 5, we include a variable PML_Head to control for the PML of a firm’s headquarters region. We find that both the signs and sizes of the coefficients of PMLT in Table 9 remain like those in Table 3, reinforcing our findings that the effect of PML remains salient on top of the real estate market conditions.

Other Robustness Tests

According to Allen and Letdin (2020), who find that unsecured debt is on average less expensive than mortgage debt for REITs, we expect a negative relation between the property market liquidity of a REIT and its cost of debt based on our main finding that higher property market liquidity leads to a higher level of unsecured debt. To test this, we regress the weighted average interest rate of REITs on the property market liquidity and include all control variables used in the baseline models. Results are reported in Appendix A, Table A4. We indeed find a negative relationship between PML and the average interest rate, which is significant at the 1% level. Our results are consistent and support those of Allen and Letdin (2020).

Finally, we run four other robustness tests to further strengthen our results. First, to address the concern that local property market liquidity is primarily based on commercial rather than residential real estate transactions, which can distort the PML measure for REITs that mainly invest in residential assets, we exclude residential REITs from our sample. Second, when the stock market is more liquid, there is generally more capital available and higher funding liquidity.Footnote 28 To address the concern that the use of debt might be influenced by the stock market liquidity of REITs, we include REIT stock market liquidity as an extra control variable. Third, we further divide unsecured debt into public debt represented by long-term notes and private debt represented by senior term loan. Finally, given that the PML we use in our regressions are estimated values based on two instruments, we run specifications (1) without instrumenting and (2) with instrumenting only one of two components at a time. Our results remain strongly robust after all these tests.Footnote 29

Conclusion

Conclusion

What determines the financing choice between secured and unsecured debt? This study proposes property market liquidity as an important factor. We construct a precise measure of a firm’s exposure to its underlying asset market liquidity by using REITs, while alleviating potential endogeneity in the measure by using an instrumental variable approach. Specifically, we use private real estate market liquidity at the MSA level weighted by the share of investments a REIT allocates in each MSA as a proxy for its exposure to property market liquidity. The distance from a REIT’s headquarters is used as an instrument for local investment allocations, and the national mortgage rate as well as its interaction with land supply are used as instruments for local property market liquidity.

We find that property market liquidity significantly impacts a REIT’s secured and unsecured debt: Firms with greater balance sheet exposure to liquid underlying property markets rely more on unsecured debt. This result is consistent with our argument that unsecured debt signals firm quality as well as enhancing financial and operational flexibilities. However, unsecured debt is associated with potential liquidation inefficiencies due to the lack of negotiation options with dispersed debtholders. Consequently, when the property markets become less liquid, REITs gravitate toward secured debt in order to capitalize on debt-reorganization capabilities in a time of hardship and elevated information asymmetry. We also find that (1) an increase in a REIT’s property market liquidity leads to an increase (decrease) in the level of unsecured (secured) debt; (2) the likelihood of a REIT issuing unsecured debt is higher when its underlying property market is more liquid; and (3) REITs are less likely to finance property acquisitions via mortgages when the associated property market liquidity is high. Finally, we find that our empirical models perform well when accounting for the critical role that unencumbered assets play in the REIT capital stack. Our results are robust to several potential issues, adjusting PML by alternative instruments; controlling credit market conditions; controlling for stock market liquidity; and excluding residential REITs. Perhaps most interesting is the fact that controlling for asset quality and property redeployability does not subsume the statistical or economic relation between PML and debt choice in our sample.

Much of the existing evidence on the debt structure of firms, in general, does not apply to REITs. This is the case as this evidence is often based on public versus private debt while REITs primarily rely on secured and unsecured debt as long-term debt financing. In addition, current REIT studies often focus on how asset quality affects capital structure outcomes, overlooking the potential impact of the underlying property markets. Our study contributes to the REIT literature by providing evidence that the underlying market liquidity of real estate assets plays a crucial role in the REIT debt structure. Our findings also shed light on possible explanations for the post-GFC uptrend of unsecured debt in the REIT sector.

Our work also adds to recent developments in the general finance literature that seek to understand the determinants of a firm’s borrowing decisions. Although tangible assets clearly play a crucial role in corporate borrowing, the relationship between tangible assets and the adoption of various debt instruments is both theoretically and empirically ambiguous due to the great level of heterogeneity in both debt and firm characteristics, in addition to the complex collateral ownership. Our analysis breaks new ground by examining the use of secured and unsecured debt in the context of a clean empirical setting.

Appendix A – Table A1: Variable Definitions

Appendix A – Table A1: Variable Definitions

Dependent Variables

Unsecured Debt

The ratio of unsecured debt to total debt; unsecured debt is all debt for which the lender does not have recourse against specific collateral owned by the borrower.

Secured Debt

The ratio of mortgage debt and notes to total debt; mortgage debt and notes are senior debt secured by real estate, excluding any lines of credit that may have real estate backing.

Unsecured Debt Issuance

A dummy that equals 1 if the security issuance is unsecured debt issuance and 0 otherwise.

Secured Debt Issuance

A dummy that equals 1 if the security issuance is secured debt issuance and 0 otherwise.

Asset LTV

The loan-to-value ratio at the property asset level.

Explanatory Variables

PMLT

Firm-level property market liquidity is calculated as the average property market liquidity weighted by the share of properties in each region. Property market liquidity is calculated based on the demand and supply reservation prices for the 31 US regions, according to van Dijk and Francke (2021). Regional property market liquidity and investment allocations are instrumented.

Leverage

The ratio of total debt to total assets is based on market value.

Maturity

Debt maturity, which is the weighted average term to maturity on all debt, in the number of months, log transformed.

Interest Rate

The weighted average interest rate on all debt.

Size

Market capitalization (million USD) of the equity, log transformed.

M/B

Market-to-book ratio, which is the price as a percentage of book value per share. Book value is calculated using financial period end common equity and common shares’ outstanding values.

σ (Earnings)

Earnings volatility, which is the standard deviation of EBITA to average total assets over the past 3 years.

FFO

The ratio of FFO to total assets, in percentage.

Growth

The growth rate of real estate investments.

HHI

The geographic concentration of a REIT’s real estate assets.

LOC

A firm’s drawn credit line as a share of the total credit line available.

Rated

A dummy variable that equals 1 if the REIT has an investment grade rating by S&P and 0 otherwise.

REIT

REIT status dummy that equals 1 if the firm is a REIT, and zero otherwise.

Tenant Credit

Average tenant’s long-term S&P rating weighted by the tenant’s revenue for each REIT, log transformed.

Industry HHI

Industry concentration, which is calculated as the Herfindahl index for MSA-level employment concentration weighted by the number of properties a given REIT holds in each MSA.

Recovery Rate

We follow Giambona et al. (2008) and define the recovery rate index as the weighted average recovery rate based on S&P’s study of CMBS recoveries by property type and the REIT’s property type portfolio.

Return

The average return of all local markets where a REIT’s properties are located.

Beta

The average systematic risk of all local markets where a REIT’s properties are located.

Alpha

The average excess return of all local markets where a REIT’s properties are located.

NOI

Net operating income to the property value.

Appendix A – Table A2: First-stage Regressions

Appendix A – Table A2: First-stage Regressions

Dependent Variable

Model 1

Asset Allocation Weights

Model 2

Region Liquidity

Model 3

Region Liquidity

Distance

−0.0133***

(0.0031)

Leverage

−0.0395***

(0.0093)

Maturity

−0.0058*

(0.0031)

Interest Rate

−0.0035***

(0.0013)

Size

−0.0100***

(0.0019)

M/B

0.0001

(0.0007)

σ(Earning)

−0.0666

(0.0695)

FFO

−0.0016

(0.0431)

Growth

−0.0003

(0.0006)

LOC

0.1558***

(0.0534)

Rated

−0.0032*

(0.0017)

REIT

0.0030

(0.0020)

Elasticity X

2.5621***

Mortgage

(0.7263)

Mortgage

−4.7986***

−13.9762***

(1.5423)

(0.8690)

Year FE

Yes

Yes

-

Firm FE

Yes

-

-

MSA FE

-

Yes

Yes

No. of obs.

626,160

1,736

1,736

Adj. R2

0.3702

0.8841

0.1173

F statistic

7082***

9.475***

258.00***

This table reports the results of the first-stage regressions. Model 1 is based on Eq. 2 using distance to headquarters as the instrument. The dependent variable in Model 1 is cost-weighted shares in certain RCA regions. Control variables include total leverage (Leverage), debt maturity (Maturity), interest rate (Interest Rate), market value (Size), market-to-book ratio (M/B), earning volatility (σ(Earning)), FFO-to-asset ratio (FFO), real estate investment growth rate (Growth), drawn credit line (LOC), REIT credit rating (Rated), and REIT status dummy (REIT). Detailed variable definitions are presented in Table 1. The standard errors in Model 1 are clustered at the firm level. Model 2 is based on Eq. 3. The national mortgage rate and the interaction of the national mortgage rate and land elasticity are the instruments. The dependent variable in Model 2 is RCA regional liquidity. Year, firm, and MSA fixed effects are also included. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively

Appendix A – Table A3: Correlation Matrix

Appendix A – Table A3: Correlation Matrix

PMLT

Leverage

Maturity

Interest Rate

Size

M/B

σ

FFO

Growth

HHI

LOC

Rated

REIT

PMLT

1

Leverage

−0.200

1

Maturity

0.120

−0.115

1

Interest Rate

−0.283

0.168

0.176

1

Size

0.160

−0.437

0.175

−0.118

1

M/B

0.184

−0.282

0.204

0.083

0.360

1

σ

0.019

0.069

−0.056

0.009

−0.133

−0.114

1

FFO

0.193

−0.414

0.087

−0.050

0.238

0.305

−0.143

1

Growth

0.049

−0.068

0.031

−0.066

−0.120

−0.082

0.008

−0.227

1

HHI

−0.011

0.015

−0.125

−0.057

−0.020

−0.021

−0.010

−0.021

−0.047

1

LOC

−0.102

0.308

−0.187

−0.060

−0.302

−0.044

0.050

−0.117

0.137

0.003

1

Rated

0.037

−0.253

0.164

−0.015

0.548

0.138

−0.163

0.142

−0.107

0.015

−0.211

1

REIT

0.006

0.034

0.025

−0.092

−0.019

0.008

−0.207

−0.109

0.040

−0.017

−0.016

0.155

1

PMLT

PML

PMLTB

PMLTA

PMLT

1

PML

0.991

1

PMLTB

0.999

0.990

1

PMLTA

0.389

0.391

0.387

1

Appendix A – Table A4: Property Market Liquidity and Interest Rate

Appendix A – Table A4: Property Market Liquidity and Interest Rate

Dependent Variable

Interest Rate

PMLT

−5.0720***

(0.2070)

Leverage

1.2994***

(0.4446)

Log(Maturity)

−0.1539

(0.1072)

Log(Size)

−0.0851

(0.0645)

M/B

0.0378

(0.0350)

σ(Earning)

4.0345***

(1.3131)

FFO

−0.3278

(1.3321)

Growth

−0.0868*

(0.0503)

HHI

0.6441*

(0.3839)

LOC

−0.5238***

(0.0960)

Rated

−0.0054

(0.0633)

REIT

−0.3199**

(0.1568)

Year FE

Yes

Firm FE

Yes

No. of obs.

1090

Adj. R2

0.7437

This table reports the results of the regression on REITs weighted average interest rates. PMLT stands for the instrumented weighted average liquidity of the underlying asset markets each REIT is exposed to. Control variables include total leverage (Leverage), debt maturity (Maturity), market value (Size), market-to-book ratio (M/B), earning volatility (σ(Earning)), FFO-to-asset ratio (FFO), real estate investment growth rate (Growth), geographic concentration (HHI), drawn credit line (LOC), REIT credit rating (Rated), and REIT status dummy (REIT) Detailed variable definitions are presented in Appendix A, Table A1. All explanatory variables are lagged by 1 year. Year and firm fixed effects are also included. The standard error for PMLT is measured using bootstrapping. ***, **, and * denote significance at the 1%, 5%, and 10% level, respectively

Appendix B - Bootstrapping Steps for the Standard Error

Appendix B - Bootstrapping Steps for the Standard Error

Data Availability

Data Availability

Publicly available data (i.e., not vendor data) that supports the findings of this study are available on request from the corresponding author.

Notes

Notes

https://www.reit.com/news/blog/market-commentary/reits-are-an-investment-grade-sector.

For a review of the declining secured debt issuance by US corporations over the twentieth century, please refer to Benmelech et al. (2022).

See, for example, Kraus and Litzenberger (1973), Myers (1977), Myers (1984), and Baker and Wurgler (2002).

Early theoretical studies include, for example, Diamond (1991); Diamond (1993); Park (2000); and DeMarzo and Fishman (2007), Bolton and Scharfstein (1996), and Bolton and Freixas (2000). For empirical studies, see, e.g., Rauh and Sufi (2010), Colla et al. (2013), Lou and Otto (2020), and Chen et al. (2021).

See, for example, Leahy (2023) MSCI report - Loss aversion and property-market liquidity, available at https://www.msci.com/www/blog-posts/loss-aversion-and-property/04076684543#f1-ref.

See, for example, Cvijanović (2014), Lin (2016), Kim and Kung (2017), Chen et al. (2020), and Campello et al. (2022).

See, for example, Feng et al. (2007), Hardin and Wu (2010), Harrison et al. (2011), and Riddiough and Steiner (2020).

Few prior studies on secured versus unsecured debt focus on either the legal perspective of claims priority (e.g., Bebchuk and Fried (1996) and Cohen (2018) or the cyclical nature of the debt market (e.g., Benmelech et al. (2022) and Luk and Zheng (2022).

See, for example, Alderson and Betker (1995), Morellec (2001), Sibilkov (2009), Campello and Giambona (2013); Gavazza (2010), Kim and Kung (2017), and Chen et al. (2020).

We gratefully acknowledge an anonymous reviewer for suggesting further consideration of alternative explanations for the PML and debt relationship for REITs.

Riddiough and Steiner (2020) conducted a case analysis on General Growth Properties (GGP) REIT. They show that by heavily relying on such secured debt, GGP exhausted its debt capacity and found it impossible to access external funds during the financial crisis.

Real Capital Analytics (RCA) is a global real estate data and analytics provider. MSCI acquired RCA in 2021.

The 31 regions are Atlanta, Austin, Baltimore, Boston, Charlotte, Chicago, Dallas, Washington DC, Denver, Detroit, Other Mid-West, Other North East, Other South West, Other West, Other Mid-Atlantic, Other South East, Houston, Los Angeles, Las Vegas, Miami, Minneapolis, New York City, Orlando, Philadelphia, Phoenix, Portland, Sacramento, San Diego, Seattle, San Francisco, and Tampa. van Dijk et al. (2022) liquidity measure is tested in van Dijk and Francke (2021).

We use these variables as primary data to calculate secured and unsecured debt ratios because the definitions explicitly specify whether the debt is secured by particular real estate assets or not. As a robustness test, we also use the SNL-reported secured debt to total debt ratio and calculate unsecured debt as (1- secured debt to total debt) for the unsecured debt ratio. The results are robust, and these tabulated results are available upon request.

Alternatively, we use property size, instead of costs, as weights to measure the liquidity of underlying property markets at the REIT level, the results remain similar.

In tabulated and unreported results, we also include the weighted average distance of underlying assets to the REIT’s headquarters as a control variable. The results are robust, and the coefficient for geographic distance on debt choice is statistically insignificant.

This instrumental approach is similar to that of Mian and Sufi (2011); Chaney et al. (2012); Lin (2016); and Campello et al. (2022).

Full results for first-stage regressions based on Eqs. (2) and (3) are reported in Appendix A, Table A2.

A small number of the firms in our sample were listed real estate investment companies in early years and gained legal REIT status later. We create a REIT dummy to capture this, but it does not affect our results.

The property market liquidity index of van Dijk and Francke (2021) ranges from − 100 to 100; for estimation purposes, we scaled it by 100, ranging from − 1 to 1.

We report a correlation matrix among all the variables in Appendix A, Table A3. Our measure of property market liquidity does not strongly correlate with any control variables. Leverage correlates with FFO and size and size correlates with rating. Therefore, we also estimate our results with leverage and rating removed from the control variables. The results are robust. Also, we observe a high correlation coefficient between the various instrumented PMLs (except for PMLTA) and uninstrumented PML. PMLTA shows less correlation with the others because it relies solely on national macroeconomic variables as instruments for local market liquidity, without accounting for local market supply constraints. As a result, the instrumented PML is less correlated with the actual PML.

Detailed steps for generating the standard error of α are provided in Appendix B.

Although long-term notes and senior term loans are not always unsecured, they represent a large portion of overall unsecured debt. We also estimate the effect of property market liquidity on two key debt instruments: senior term loans and long-term notes. In our data, senior term loans represent the principal amount due to institutional lenders for senior term loans not secured by real estate. The results are robust.

Zhu and Lizieri (2022) show that the REIT return risk can be influenced by the systematic risk of local real estate markets.

Ling et al. (2022) use the geographically weighted risk-adjusted return (alpha) as a proxy for the excess return of REIT property portfolio (αPPR) and find that αPPR predicts the cross-section of returns in public REITs.

We collect the property total return index in 144 core-based statistical areas (CBSAs) and MSAs since 1978 from the NCREIF database. In the NCREIF data, property markets are divided into Metropolitan Divisions (MD). For instance, for the NCREIF the Detroit-Warren-Dearborn, MI MSA is divided into two MDs: Detroit-Dearborn-Livonia, and Warren-Troy-Farmington Hills. However, S&P Global Market Intelligence uses an MSA code for property location that is only at the level of the MSA (Detroit-Warren-Dearborn). Therefore, we convert MD property market to MSA turnover. The MSA turnover is calculated as the average MD turnover weighted by the number of properties in each MD where the MSA belongs to.

Downs and Zhu (2022) find that property market liquidity is positively related to REIT stock liquidity.

For parsimonious reasons, the results of extra robustness tests are not reported but available upon request.

References

References

Acharya, V. V., Almeida, H., & Campello, M. (2007). Is cash negative debt? A hedging perspective on corporate financial policies. Journal of Financial Intermediation, 16(4), 515-554.

Article

Google Scholar

Alderson, M. J., & Betker, B. L. (1995). Liquidation costs and capital structure. Journal of Financial Economics, 39(1), 45-69.

Article

Google Scholar

Allen, L., & Letdin, M. (2020). The cost of debt for REITs: The mortgage puzzle. Journal of Real Estate Research, 42(2), 239-260.

Article

Google Scholar

Ambrose, B., Bond, S., & Ooi, J. T. L. (2010). Secured debt and corporate performance: evidence from REITs.

Attar, A., Casamatta, C., Chassagnon, A., & Décamps, J. P. (2019). Multiple lenders, strategic default, and covenants. American Economic Journal: Microeconomics, 11(2), 98–130.

Google Scholar

Badoer, D. C., Dudley, E., & James, C. M. (2019). Priority spreading of corporate debt. Review of Financial Studies, 33(1), 261-308.

Article

Google Scholar

Baker, M., & Wurgler, J. (2002). Market timing and capital structure. The Journal of Finance, 57(1), 1-32.

Article

Google Scholar

Bebchuk, L. A., & Fried, J. M. (1996). The uneasy case for the priority of secured claims in bankruptcy. The Yale Law Journal, 105(4), 857-934.

Article

Google Scholar

Benmelech, E., Garmaise, M. J., & Moskowitz, T. J. (2005). Do liquidation values affect financial contracts? Evidence from commercial loan contracts and zoning eegulation. The Quarterly Journal of Economics, 120(3), 1121-1154.

Google Scholar

Benmelech, E., Kumar, N., & Rajan, R. G. (2022). The decline of secured debt. The Journal of Finance, Forthcoming.

Bolton, P., & Freixas, X. (2000). Equity, bonds, and bank debt: Capital structure and financial market equilibrium under asymmetric information. Journal of Political Economy, 108(2), 324-351.

Article

Google Scholar

Bolton, P., & Scharfstein, D. S. (1996). Optimal debt structure and the number of creditors. Journal of Political Economy, 104(1), 1-25.

Article

Google Scholar

Campello, M., & Giambona, E. (2013). Real assets and capital structure. Journal of Financial and Quantitative Analysis, 48(5), 1333-1370.

Article

Google Scholar

Campello, M., Connolly, R. A., Kankanhalli, G., & Steiner, E. (2022). Do real estate values boost corporate borrowing? Evidence from contract-level data. Journal of Financial Economics, 144(2), 611-644.

Article

Google Scholar

Cantillo, M., & Wright, J. (2000). How do firms choose their lenders? An empirical investigation. Review of Financial Studies, 13(1), 155-189 .

Article

Google Scholar

Chaney, T., Sraer, D., & Thesmar, D. (2012). The collateral channel: How real estate shocks affect corporate investment. American Economic Review, 102(6), 2381–2409.

Article

Google Scholar

Chen, H., Maslar, D. A., & Serfling, M. (2020). Asset redeployability and the choice between bank debt and public debt. Journal of Corporate Finance, 64.

Chen, Y., Hasan, I., Saffar, W., & Zolotoy, L. (2021). Executive equity risk-taking incentives and firms’ choice of debt structure. Journal of Banking & Finance, 133.

Cohen, N. B. (2018). The private international law of secured transactions: Rules in search of harmonization. Law and Contemporary Problems, 81,203-225.

Google Scholar

Colla, P., Ippolito, F., & Li, K. (2013). Debt specialization. The Journal of Finance, 68(5), 2117-2141.

Article

Google Scholar

Conklin, J., Diop, M., & Qiu, M. (2018). How do firms finance nonprimary market investments? Evidence from REITs. Real Estate Economics, 46(1), 120-159.

Article

Google Scholar

Coval, J. D., & Moskowitz, T. J. (2001). The geography of investment: Informed trading and asset prices. Journal of Political Economy, 109(4), 811-841.

Article

Google Scholar

Cvijanović, D. (2014). Real estate prices and firm capital structure. Review of Financial Studies, 27(9), 2690–2735.

Article

Google Scholar

Demarzo, P. M. (2019). Presidential address: Collateral and commitment. The Journal of Finance, 74(4), 1587-1619.

Article

Google Scholar

DeMarzo, P. M., & Fishman, M. J. (2007). Optimal long-term financial contracting. Review of Financial Studies, 20(6), 2079-2128.

Article

Google Scholar

Demirci, I., Eichholtz, P., & Yönder, E. (2018). Corporate diversification and the cost of debt: Evidence from REIT bank loans and mortgages. Journal of Real Estate Finance and Economics, 61.

Demirci, I., Rasteh, M., & Yönder, E. (2023). Unpledged collateral, REIT liquidity constraints, and asset sales. Journal of Real Estate Finance and Economics.

Denis, D. J., & Mihov, V. T. (2003). The choice among bank debt, non-bank private debt, and public debt: Evidence from new corporate borrowings. Journal of Financial Economics, 70(1), 3-28.

Article

Google Scholar

Diamond, D. W. (1991). Monitoring and reputation: The choice between bank loans and directly placed debt. Journal of Political Economy, 99(4), 689-721.

Article

Google Scholar

Diamond, D. W. (1993). Seniority and maturity of debt contracts. Journal of Financial Economics, 33(3), 341-368.

Article

Google Scholar

Downs, D. H., & Zhu, B. (2022). Property market liquidity and REIT liquidity. Real Estate Economics, 50(6), 1462-1491.

Article

Google Scholar

Feng, Z., & Wu, Z. (2023). ESG disclosure, REIT debt financing and firm value. Journal of Real Estate Finance and Economics, 67(3), 388–422.

Article

Google Scholar

Feng, Z., Ghosh, C., & Sirmans, C. F. (2007). On the capital structure of Real Estate Investment Trusts (REITs). Journal of Real Estate Finance and Economics, 34(1), 81-105.

Article

Google Scholar

Fisher, J., Gatzlaff, D., Geltner, D., & Haurin, D. (2003). Controlling for the impact of variable liquidity in commercial real estate price indices. Real Estate Economics, 31(2), 269-303.

Article

Google Scholar

Fisher, J., Geltner, D., & Pollakowski, H. (2007). A quarterly transactions-based index of institutional real estate investment performance and movements in supply and demand. Journal of Real Estate Finance and Economics, 34(1), 5-33.

Article

Google Scholar

Garmaise, M. J., & Moskowitz, T. J. (2004). Confronting information asymmetries: Evidence from real estate markets. Review of Financial Studies, 17(2), 405-437 .

Article

Google Scholar

Gavazza, A. (2010). Asset liquidity and financial contracts: Evidence from aircraft leases. Journal of Financial Economics, 95(1), 62-84.

Article

Google Scholar

Ghent, A. C. (2021). What’s wrong with Pittsburgh? Delegated investors and liquidity concentration. Journal of Financial Economics, 139(2), 337–358.

Article

Google Scholar

Giambona, E., Harding, J. P., & Sirmans, C. F. (2008). Explaining the variation in REIT capital structure: The role of asset liquidation value. Real Estate Economics, 36(1), 111-137.

Article

Google Scholar

Giambona, E., Mello, A. R., & Riddiough, T. J. (2017). Real assets, collateral and the limits of debt capacity. Real Estate Economics, 46(4), 836-886.

Article

Google Scholar

Hardin, W. G., & Wu, Z. (2010). Banking relationships and REIT capital structure. Real Estate Economics, 38(2), 257-284.

Article

Google Scholar

Hardin, W. G., Highfield, M. J., Hill, M. D., & Kelly, G. W. (2009). The determinants of REIT cash holdings. Journal of Real Estate Finance and Economics, 39(1), 39-57.

Article

Google Scholar

Harrison, D. M., Panasian, C. A., & Seiler, M. J. (2011). Further evidence on the capital structure of REITs. Real Estate Economics, 39(1), 133-166.

Article

Google Scholar

Hartzell, D., Howton, S. D., Howton, S., & Scheick, B. (2019). Financial flexibility and at-the-market (ATM) equity offerings: Evidence from real estate investment trusts. Real Estate Economics, 47(2), 595-636.

Article

Google Scholar

Ibrahim, I., & Falkenbach, H. (2023). Diversification and cost of public debt for REITs: Evidence from the US. Journal of Real Estate Portfolio Management.30(3), 1-18

Google Scholar

Kim, H., & Kung, H. (2017). The Asset redeployability channel: How uncertainty affects corporate investment. Review of Financial Studies, 30(1), 245–280.

Article

Google Scholar

Krainer, J. (2001). A theory of liquidity in residential real estate markets. Journal of Urban Economics, 49(1), 32-53.

Article

Google Scholar

Kraus, A., & Litzenberger, R. H. (1973). A state-preference model of optimal financial leverage. The Journal of Finance, 28(4), 911-922.

Google Scholar

Leahy, T. (2023). Loss aversion and property-market liquidity. MSCI. https://www.msci.com/www/blog-posts/loss-aversion-and-property/04076684543#f1-ref

Li, S., Whited, T. M., & Wu, Y. (2016). Collateral, taxes, and leverage. Review of Financial Studies, 29(6), 1453-1500.

Article

Google Scholar

Lian, C., & Ma, Y. (2021). Anatomy of corporate borrowing constraints. The Quarterly Journal of Economics, 136(1), 229-291.

Article

Google Scholar

Lin, L. (2016). Collateral and the choice between bank debt and public debt. Management Science, 62(1), 111-127.

Article

Google Scholar

Ling, D. C., Naranjo, A., & Scheick, B. (2021a). There is no place like home: Information asymmetries, local asset concentration, and portfolio returns. Real Estate Economics, 49(1).

Ling, D. C., Wang, C., & Zhou, T. (2021b). The geography of real property information and investment: Firm location, asset location and institutional Ownership. Real Estate Economics, 49(1), 287-331.

Article

Google Scholar

Ling, D. C., Wang, C., & Zhou, T. (2022). Asset productivity, local information diffusion, and commercial real estate returns. Real Estate Economics, 50(1), 89-121.

Article

Google Scholar

Liu, C. H., Liu, P., & Zhang, Z. (2019). Real Assets, Liquidation Value and Choice of Financing. Real Estate Economics, 47(2), 478-508.

Article

Google Scholar

Lou, Y., & Otto, C. A. (2020). Debt heterogeneity and covenants. Management Science, 66(1), 70-92.

Article

Google Scholar

Lu-Andrews, R. (2017). Tenant quality and REIT liquidity management. Journal of Real Estate Finance and Economics, 2017(3).

Luk, P., & Zheng, T. (2022). Dynamics of secured and unsecured debt over the business cycle. Review of Economic Dynamics, 44, 284-314.

Article

Google Scholar

Mello, A. S., & Ruckes, M. (2017). Collateral in corporate financing.

Mian, A., & Sufi, A. (2011). House prices, home equity-based borrowing, and the US household leverage crisis. American Economic Review, 101(5), 2132-2156.

Article

Google Scholar

Milcheva, S., Yildirim, Y., & Zhu, B. (2021). Distance to headquarter and real estate equity performance. Journal of Real Estate Finance and Economics, 63(3), 327-353.

Article

Google Scholar

Morellec, E. (2001). Asset liquidity, sapital sructure, and secured debt. Journal of Financial Economics, 61(2), 173-206.

Article

Google Scholar

Myers, S. C. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5(1), 147-175.

Article

Google Scholar

Myers, S. C. (1984). The capital structure puzzle. The Journal of Finance, 39(3), 574-592.

Article

Google Scholar

Ott, S. H., Riddiough, T. J., & Yi, H. C. (2005). Finance, investment and investment performance: Evidence from the REIT sector. Real Estate Economics, 33(1), 203-235.

Article

Google Scholar

Park, C. (2000). Monitoring and structure of debt contracts. The Journal of Finance, 55(5), 2157-2195.

Article

Google Scholar

Rauh, J. D., & Sufi, A. (2010). Capital structure and debt structure. Review of Financial Studies, 23(12), 4242-4280.

Article

Google Scholar

Riddiough, T., & Steiner, E. (2020). Financial flexibility and manager–shareholder conflict: Evidence from REITs. Real Estate Economics, 48(1), 200-239.

Article

Google Scholar

Saiz, A. (2010). The geographic determinants of housing supply. The Quarterly Journal of Economics, 125(3), 1253–1296.

Article

Google Scholar

Schwarcz, S. L. (1997). The easy case for the priority of secured claims in bankruptcy. Duke Law Journal, 47(3), 425-489.

Article

Google Scholar

Shleifer, A., & Vishny, R. W. (1992). Liquidation values and debt capacity: A market equilibrium approach. The Journal of Finance, 47(4), 1343-1366.

Article

Google Scholar

Sibilkov (2009). Asset liquidity and capital structure. Journal of Financial and Quantitative Analysis, 44(5), 1173-1196.

Article

Google Scholar

Standard and Poor’s (2005). Defaults and losses of Standard & Poor’s rated U.S. commercial mortgage loans: year-end 2004.

Tuzel, S., & Zhang, M. B. (2017). Local risk, local factors, and asset prices. The Journal of Finance, 72(1), 325-370.

Article

Google Scholar

van Dijk, D. W., & Francke, M. K. (2021). Commonalities in private commercial real estate market liquidity and price index returns. Journal of Real Estate Finance and Economics.

van Dijk, D. W., Geltner, D. M., & van de Minne, A. M. (2022). The dynamics of liquidity in commercial property markets: Revisiting supply and demand indexes in real estate. Journal of Real Estate Finance and Economics, 64(3), 327-360.

Article

Google Scholar

Zhu, B., & Lizieri, C. (2022). Local beta: has local real estate market risk been priced in REIT returns? Journal of Real Estate Finance and Economics.

Download references

Acknowledgements

Acknowledgements

Our work has benefited from discussions with many of our colleagues as well as conference discussants and participants. We thank Hany Ahmed, Brent Ambrose, Anil Kumar, Chongyu Wang, and René-Ojas Woltering, participants at the 2022 AREUEA International Conference, 2021 ERES Conference, 2021 Hospitality Finance and Economics Conference at École hôtelière de Lausanne, and the 2023 FMA European Conference for helpful comments and suggestions. We are especially grateful to our anonymous referees and Jim Conklin, the editor. Downs acknowledges The Kornblau Institute at Virginia Commonwealth University for support. The authors alone are responsible for any errors.

Funding

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Author information

Author information

Authors and Affiliations

The Kornblau Institute and FIRE Department, School of Business, Virginia Commonwealth University, Richmond, VA, USA

David H Downs

School of Management, University of Bath, Convocation Ave, Claverton Down, Bath, BA2 7AZ, United Kingdom

Chen Zheng

School of Engineering and Design, Technical University of Munich, Munich, Germany

Bing Zhu

Authors

David H Downs

View author publications

Search author on:PubMed Google Scholar

Chen Zheng

View author publications

Search author on:PubMed Google Scholar

Bing Zhu

View author publications

Search author on:PubMed Google Scholar

Contributions

All authors contributed equally to the study conception, design, data collection, analysis, and writing of the first draft. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to

Chen Zheng.

Ethics declarations

Ethics declarations

Ethical Approval

The manuscript is not submitted to any other journal. The submitted work is original and is not published elsewhere in any form or language (partially or in full).

Consent to Participate

Not Applicable.

Consent to Publish

Not Applicable.

Competing Interests

The authors have no financial or non-financial interests to disclose.

Additional information

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

Rights: © The Author(s) 2025
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License
Edition changes: Licensed narrative text edition. Original headings, paragraphs and references are preserved with normalized whitespace and layout. Figures, tables, display equations and separately licensed third-party material are not reproduced; use the linked original full text for those elements and definitive pagination. This is an attributed republication, not newly conducted research.