Decomposing industry leverage: The special cases of real estate investment trusts and technology & hardware companies

• Published date: 01 September 2023
• Author: Wolfgang Breuer,Linh D. Nguyen,Bertram I. Steininger
• Date: 01 September 2023
• DOI: http://doi.org/10.1111/jfir.12332

Received: 24 September 2021

|

Accepted: 3 May 2023

DOI: 10.1111/jfir.12332

ORIGINAL ARTICLE

Decomposing industry leverage: The special

cases of real estate investment trusts and

technology & hardware companies

Wolfgang Breuer

1

|Linh D. Nguyen

2

|Bertram I. Steininger

3

1

Department of Finance, RWTH Aachen

University, Aachen, Germany

2

Department of Finance, Banking University

of Ho Chi Minh City, Ho Chi Minh City,

Vietnam

3

Real Estate Economics and Finance, KTH

Royal Institute of Technology, Stockholm,

Sweden

Correspondence

Bertram I. Steininger, Real Estate Economics

and Finance, KTH Royal Institute of

Technology, Teknikringen 10 B, 100 44

Stockholm, Sweden.

Email: bertram.steininger@abe.kth.se

Funding information

Deutscher Akademischer Austauschdienst;

BộGiáo dụcvàĐào tạo

Abstract

Different industries exhibit significantly different leverage;

companies in the real estate investment trust (REIT) and

technology/hardware sectors are extreme examples. In the

United States, the leverage ratio is twice as high for REITs

(50%) as compared to non‐real‐estate firms (around 25%),

and the technology/hardware sector has the lowest ratio

(around 17%). We theoretically and empirically analyze

their differences. By decomposing the difference into three

channels, we find that the industry‐specific channel

explains around 67% for REITs and 68% for technology/

hardware firms; the value‐based channel is mostly respon-

sible for the remaining portion. Taking the nonlinear

influences of extreme values into account, the relevance

of the industry‐specific channel is considerably reduced.

JEL CLASSIFICATION

G21, G32, H25

1|INTRODUCTION

Capital structure theories try to explain which sources managers should use to raise money to finance their projects.

The most prominent theories are the trade‐off theory, pecking order theory, and market timing theory. Each of

these theories aims at determining the most important drivers of corporate capital structure choices. However, it

J Financ Res. 2023;46:791–823. wileyonlinelibrary.com/journal/JFIR

|

791

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and

reproduction in any medium, provided the original work is properly cited.

© 2023 The Authors. Journal of Financial Research published by Wiley Periodicals LLC on behalf of The Southern Finance

Association and the Southwestern Finance Association.

remains a challenge to explain theoretically the huge differences in average leverage ratios across industries (e.g.,

Lemmon et al., 2008; Ross et al., 2016). Excluding the banking and insurance sector, the average industry leverage

ratio is 25% and ranges from 17% for technology hardware and equipment firms (THEs) to 50% for real estate

investment trusts (REITs) in the United States from 1999 to 2015.

Hence, it comes as no surprise that the impact of capital structure determinants has been broadly examined in

the literature. However, to the best of our knowledge, these investigations have mostly been conducted within one

industry or with simple industry dummies across industries. Where single industries haven been examined, such as

insurance firms (see, e.g., Frank & Goyal, 2009; Rajan & Zingales, 1995) or REITs (see, e.g., Harrison et al., 2011;

Morri & Beretta, 2008), studies have focused on the sign and the statistical as well as economic significance of

estimated coefficients but not on reasons for the difference in leverage, which goes beyond such an analysis. In this

article, we dig a bit deeper. We pick the REIT industry as the main object and the THE industry as the secondary

object of our study for the following reasons. First, REITs stand out because they exhibit the highest leverage ratios

outside the financial services sector. Second, because of this feature, the literature and industry have coined the

term “REIT debt puzzle,”which renders such an investigation interesting. Third, by looking at THE industry, which

has particularly low debt levels, we verify our approach with a second sector.

Although we want to contribute methodologically to open up avenues for general industry‐specific capital

structure analyses, we also aim to contribute to the more specific literature on REIT and THE debt levels. Until now,

REITs’capital structure determinants have attracted much more interest by scholars than those of THEs, justifying

our approach of putting REITs into the center of our analysis.

In contrast to the literature, we examine three potential reasons for different leverage ratios by distinguishing

among value‐based (different average values of capital structure determinants), coefficient‐based (different

sensitivities to variations in the size of capital structure determinants), and intercept‐based (residual effect not

explainable as being value or coefficient based) explanations of capital structure differences based on (linear)

regression analyses. Although the coefficient‐based explanation has been examined in prior studies, the further

explicit breakdown of determinants of capital structure differences into value‐based and intercept‐based

explanations is our innovation. The intercept‐based explanation may also be referred to as an industry‐specific

fixed effect, which separates the financing behavior of different industries and which cannot be explained by value‐

related or coefficient‐related approaches. In a general sense, such an industry‐specific fixed effect is known to be

relevant from the general literature on corporate capital structure. For example, Frank and Goyal (2009) document

that the industry‐specific fixed effect is one of the most important factors for explaining leverage.

Our study sample covers 29,391 firm‐year observations for 192 equity REITs and 2815 non‐real‐estate firms

(non‐REs) in the United States from 1999 to 2015. According to our analyses of the three channels (i.e., value based,

coefficient based, and intercept based), we find that 67% of the total leverage difference between REITs and non‐

REs of around 25.5 percentage points (pp) is explained by a REIT‐specific fixed effect; the remaining portion is

mostly explained by the value‐based channel, and the coefficient‐based channel implies lower leverage ratios of

REITs in comparison to non‐REs. The most important drivers on the value‐based side are differences in tangibility

and operating risk, whereas the most relevant countervailing effect on the coefficient‐based side stems from the

impact of firm size. Other important capital structure determinants (i.e., profitability, growth opportunities, and

dividend payments) are of minor relevance for capital structure differences between REITs and non‐REs. This

general decomposition is the first main contribution of our article to the literature on REIT capital structure.

Nevertheless, this is bad news for finance theory, as only about one third of the capital structure difference

between REITs and non‐REs seems to be explained by capital structure theories: The REIT debt problem remains,

now changing from a mean‐value REIT debt puzzle (regarding the high average value of REIT debt levels) to a fixed‐

effect REIT debt puzzle. To address this remaining difference, we investigate in more detail the REIT‐specific fixed

effect and find two remarkable results. First, this effect does not vanish when taking into account potential

differences in debt maturity between REITs and non‐REs. Second, however, the REIT‐specific fixed effect is almost

completely absent when comparing REITs to corresponding non‐REs in a matched sample built based on operating

792

|

JOURNAL OF FINANCIAL RESEARCH

risk and tangibility features. This finding might be of general interest for the development of capital structure

theory, as it hints at the possibility that industry‐specific capital structure effects actually do not exhibit the high

importance that might be deduced from traditional (linear) regression models. As a possible explanation for this

reduced relevance of the REIT‐specific fixed effect, we propagate (even minor) nonlinearities in the function

governing corporate leverage choices (particularly with respect to operating risk). For comparable samples of REITs

and non‐REs, these nonlinearities lose relevance, which suggests that the importance of a REIT‐specific fixed effect

is strongly exaggerated. This finding is the second main contribution of our article regarding REIT debt levels. It is

good news for finance theory, as significantly different magnitudes of tangible assets and stock return risk

characteristics driving value‐based differences in capital structure are indeed able to explain the differences in

leverage between REITs and non‐REs.

As another example, we look at the debt levels of THEs. The value‐based and coefficient‐based explanations

for capital structure differences betweenTHEs and non‐THEs almost cancel each other out, leading to an intercept‐

based difference that is nearly identical to the average total difference in leverage ratios between THEs and non‐

THEs. After applying the same matching strategy as for the REIT sample, the intercept‐based explanation is reduced

by around 72%. We conclude once again that nonlinearities between leverage and its determinants seem to

determine to a large extent capital structure of this industry.

In line with Welch (2007), Fattouh et al. (2008), and Vatavu (2016), we thus suggest that the linear regression

approaches do not sufficiently control for the nonlinear relation between leverage and its determinants, thereby

incorrectly identifying a fixed‐effect REIT debt puzzle, a conclusion that may carry over in an analogous way to

analyses of other industry fixed effects in corporate finance.

2|VALUE‐BASED, COEFFICIENT‐BASED, AND INTERCEPT‐BASED

EXPLANATIONS OF CAPITAL STRUCTURE DIFFERENCES

As pointed out earlier, there are three main reasons why the capital structures of chosen industries (e.g., REITs and

THEs) might differ. First, the main firm characteristics driving corporate capital structure choice could have a

systematically different magnitude; for example, the average firm size may differ. Therefore, even if the explaining

factors as independent variables imply the same functional relation to leverage, we observe a divergent leverage

level for them. We call this the “value‐based”explanation of the difference. For a stylized illustration of such a

relation for industry sector X, see Figure 1a. Second, and independent of the first explanation, there could exist a

systematic difference in the sensitivity of the debt level with respect to a capital structure determinant, although

this influential factor has the same average value for different industries. We call this the “coefficient‐based”

explanation for capital structure differences (see Figure 1b). Third, it is possible that neither the value‐based nor the

coefficient‐based approach is able to explain prevailing capital structure differences because the slopes of the two

regression lines and the average values of the capital structure determinant under consideration are identical.

However, the ordinate intercept of both regression lines varies (see Figure 1c). A nonzero intercept can be

interpreted as an industry‐specific effect because capital structure theories seem to fail to explain this part of the

overall difference.

3|MAIN FACTORS EXPLAINING CAPITAL STRUCTURE

3.1 |All industries

The corporate finance literature has identified many factors that may affect a firm's capital structure. Harris and

Raviv (1991) analyze many other general finance studies and report a consensus that leverage is positively

DECOMPOSING INDUSTRY LEVERAGE

|

793