The Capital Structure of North American REITs and REOCs A Panel Data Regression

1
The Capital Structure of North American REITs
and REOCsA Panel Data Regression

• Nicolai C. Striewe
• Nico B. Rottke

2
Introduction

• Motivation
• Larger dataset for REITs than those of past
  studies.
• Extension of sample by REOCs
• Including Canadian REITs and REOCs
• More efficient methodology, as data records have
  improved

3
Introduction

• Contribution to literature
• More determinants of leverage included
• Dynamic dimension (panel approach)
• Differentiation between REITs and REOCs
• Detailed robustness tests

4
Introduction

• Overview of results
• Application of pecking order theory for
  REITs/REOCs concerning growth opportunities
• Application of trade-off theory concerning size
  and asset tangibility
• Concerning profitability only REITs follow the
  pecking order relationship
• Compensation style and directors stake in
  company influences the choice of leverage
• REOCs prefer higher leverage as they profit from
  tax-shield (trade -off theory)

5
Data

• Database SNL Financial
• 418 REITs(355) and REOCs (63)
• Over 41 periods quarterly from Q4, 1998 to Q4,
  2008
• Data preparation
• Hybrid REITs and mortgage REITs excluded (19
  units).
• Missing data 148 units (missing data or
  incomplete time series)
• Units for analysis 251 units.

6
Results

• Preferred Model

Sum squared resid 1263975 S.E. of regression 15.33344
No. of observations 5430 Akaike criterion 45113.6
'Within' variance 41.3928 'Between' variance 152.76
coefficient std. error t-ratio p-value
Constant -28.366 4.410 -6.432 0.000
Market to book ratio 0.161 0.022 7.450 0.000
ln(assets) 4.464 0.297 15.050 0.000
Real estate to assets ratio 0.141 0.012 12.090 0.000
REOC dummy 5.984 2.688 2.227 0.026
External mgmt dummy 4.940 2.556 1.932 0.053
Return on assets -0.107 0.013 -8.180 0.000
Insider ownership 0.063 0.012 5.165 0.000
Residential dummy 14.639 2.720 5.382 0.000
Shopping dummy 9.354 2.409 3.882 0.000
Office dummy -0.442 2.693 -0.164 0.870
Hotel dummy -3.227 3.134 -1.030 0.303
Diversified dummy 3.372 3.100 1.088 0.277
Age 0.000 0.000 -1.909 0.056
Canada dummy 8.666 7.579 1.143 0.253
Bonus to total compensation -0.344 0.205 -1.676 0.094

Hausman test 0.689342
Notes The table shows the results of the panel
data regression using random effects with debt to
assets ratio as dependent variable. The
time-series are from 1 to 41 periods long for 251
cross-sectional units. Time dummies are not
displayed. They are jointly significant with a
p-value of 0.0000 according to Wald test.
Breusch-Pagan test identifies the variance of the
unit-specific error to be significantly different
from zero (p0.0000). indicates significance at
the 10 level, indicates significance at the
5 level, indicates significance at the 1
level.
7
Time Trend of Leverage

• Coefficients of time dummies in preferred model

8
Sensitivity Analysis

• Impact of endogeneity
• Random effects model assumes endogeneity of all
  independent variables
• Endogenous variables market to book ratio,
  ln(Assets), real estate to assets and return on
  assets.
• Testing robustness of results by lagging (t-1)
  the endogenous variables.
• Alternative dependent variable
• Welch (2007) The opposite of financial debt is
  not equity
• Alternative proxy for leverage liabilities to
  assets ratio
• Alternative independent variables
• Alternative proxy for growth opportunities
  ln(asset growth)
• Alternative proxy for size ln(rental revenue)
• Alternative proxy for profitability return on
  sales
• Special cases of residential and shopping
  REITs/REOCs
• Check changing influence of determinants for
  residential and shopping REITs/REOCs respectively.

9
Findings

Trade-off theory (TO) Pecking order theory (PO) REITs (previous research) Corporate Gover-nance This Study Comment
Profitability - - -(REIT) (REOC) REITs are tax-exempt follow PO. REOCs benefit from tax-shield (TO).
Growth - /- 1 Information gap is larger for growth REITs/REOCs (PO).
Tangibility - 2 Asset tangibility reduces risk of financial distress (TO)
Operating risk - - Not covered in this study
Size - 2 Size reduces risk of financial distress(TO)
External Management 2 Compensation by assets under management incentivizes to lever
Insider Ownership Directors with stake in company concentrate their ownership and reduce risk of hostile takeover with debt.
REOC REOCs benefit from tax shield.
1 Lower influence for residential and shopping
REITs/REOCs 2 Higher influence for residential
and shopping REITs/REOCs
10
Discussion

• Nicolai C. Striewe
• Real Estate Management Institute
• EUROPEAN BUSINESS SCHOOL
• Söhnleinstraße 8D
• 65201 Wiesbaden, Germany
• E-Mail striewe.ebs_at_rem-institute.org

11
Backup

• Nicolai C. Striewe
• Real Estate Management Institute
• EUROPEAN BUSINESS SCHOOL
• Söhnleinstraße 8D
• 65201 Wiesbaden, Germany
• E-Mail striewe.ebs_at_rem-institute.org

12
Variable definitions and basic descriptives
Measure Description Abbreviation Mean Min Max
Leverage Total debt divided by total assets Debt to assets 50.89 0 96.33
Alternative Liabilities divided by total assets Liabilities to assets 56.923 0 99.76
Growth opportunities Market to book ratio. Market capitalization divided by shareholders equity. Market to book 1.82 0 276.86
Alternative Natural logarithm of asset growth ln(asset growth) 2.207 -4.61 6.73
Size Natural logarithm of total assets ln(assets) 14.04 5.25 17.17
Alternative Natural logarithm of Rental Revenue ln(rental revenue) 10.55 2.20 13.74
Asset Tangibility Property Investment divided by Total Assets Real estate to assets 79.55 0 110.73
REOC Dummy variable for REOC status. (1REOC, 0REIT) REOC dummy 0.13 0 1
Management structure Dummy variable for external management. (1externally managed, 0internally managed) External mgmt dummy 0.15 0 1
Profitability Net income divided by total average assets. Return on assets 3.11 -177.59 177.35
Alternative Return on Sales Return on sales 0.11 -122.50 15.58
Insider ownership Percent ownership in shares of directors. Insider ownerships 15.59 0 100.00
Notes Basic descriptives from 251 observations
of the preferred model. Variance Inflation
Factors (VIF) all below 2.4 for the variables.
13
Dataset
Full Sample Full Sample Reduced Sample (Data Preparation) Reduced Sample (Data Preparation)
Property Focus Number Percentage Number Percentage
Residential 65 16 36 14
Shopping 59 15 43 17
Diversified 55 14 27 11
Hotel 54 14 42 17
Office 53 13 36 14
Storage 26 7 7 3
Industrial 24 6 16 6
Health Care 23 6 15 6
Speciality 22 6 17 7
Retail 17 4 12 5
Not Available 1 0 0 0
Total 399 100 251 100
REITs 337 84 207 82
REOCs 62 16 44 18
14
Sensitivity Analysis Model specification
Preferred Model Preferred Model Preferred Model Model 2 (alt. dep. Var) Model 2 (alt. dep. Var) Model 2 (alt. dep. Var) Model 3 Model 3 Model 3
coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value
Constant -28.366 0.000 -10.662 0.015 -15.977 0.001
Market to book 0.161 0.000 0.181 0.000
Market to book (t-1) 0.142 0.000
ln(Assets) 4.464 0.000 3.838 0.000
ln(Assets) (t-1) 3.625 0.000
Real estate to assets 0.141 0.000 0.085 0.000
Real estate to assets (t-1) 0.148 0.000
REOC dummy 5.984 0.026 11.245 0.000 5.413 0.046
External mgmt dummy 4.940 0.053 0.720 0.782 2.980 0.252
Return on assets -0.107 0.000 -0.135 0.000
Return on assets (t-1) -0.110 0.000
Insider Ownership 0.063 0.000 0.057 0.000 0.047 0.000
Residential dummy 14.639 0.000 11.993 0.000 12.915 0.000
Shopping dummy 9.354 0.000 7.755 0.002 9.239 0.000
Office dummy -0.442 0.870 -1.802 0.512 -0.137 0.960
Hotel dummy -3.227 0.303 0.748 0.815 -2.120 0.502
Diversified dummy 3.372 0.277 1.240 0.695 3.921 0.209
Age 0.000 0.056 0.000 0.756 0.000 0.018
Canada dummy 8.666 0.253 3.239 0.675 9.944 0.192
Bonus to total compensation -0.344 0.094 -0.143 0.479 -0.369 0.080
No. of observations 5430 5430 5211
Cross-sectional units 251 251 248
S.E. of regression 15.333 15.159 15.121
Akaike criterion 45113.60 44989.53 43151.050
'Within' variance 43.393 40.005 40.922
'Between' variance 152.76 159.541 153.927
Wald (joint) time dummies (?2) 139.638 0.000 190.722 0.000 138.122 0.000
Hausman test (?2) 40.798 0.689 137.287 0.000 13.911 1.000
Notes The table shows the results of the panel
data regressions using random effects with debt
to assets ratio as dependent variable. In model 2
the alternative dependent variable liabilities to
assets is used instead. Model 3 tests the impact
of endogeneity by lagging the affected variables
by lagging them (t-1). The time-series are from 1
to 41 periods. Time dummies are not displayed.
They are jointly significant with a p-value of
0.000 according to Wald test. indicates
significance at the 10 level, indicates
significance at the 5 level, indicates
significance at the 1 level.
15
Sensitivity Analysis Proxy robustness
Preferred Model Preferred Model Preferred Model Model 4 Model 4 Model 4 Model 5 Model 5 Model 5 Model 6 Model 6 Model 6 Model 7 Model 7 Model 7
coefficient p-value p-value coefficient p-value p-value Coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value
Constant -28.366 0.000 2.567 0.646 17.573 0.000 -22.716 0.000 -17.802 0.000
Market to book 0.161 0.000 0.129 0.000 0.155 0.000 0.153 0.000
ln(asset growth) 0.191 0.028
ln(assets) 4.464 0.000 2.153 0.000 3.913 0.000 3.618 0.000
ln(rental revenue) 1.889 0.000
Real estate to assets 0.141 0.000 0.151 0.000 0.092 0.000 0.150 0.000 0.142 0.000
REOC dummy 5.984 0.026 5.468 0.063 2.839 0.343 6.775 0.018 6.059 0.031
External mgmt dummy 4.940 0.053 2.106 0.455 1.010 0.703 4.310 0.104 3.897 0.133
Return on assets -0.107 0.000 -0.138 0.000 -0.213 0.000
Return on sales 0.116 0.001 -1.636 0.000
REOC dummy 1.825 0.000
Insider Ownership 0.063 0.000 0.046 0.002 0.083 0.000 0.069 0.000 0.067 0.000
Residential dummy 14.639 0.000 12.819 0.000 12.965 0.000 14.749 0.000 14.755 0.000
Shopping dummy 9.354 0.000 10.045 0.000 7.861 0.001 9.523 0.000 9.451 0.000
Office dummy -0.442 0.870 0.870 0.765 -0.226 0.932 -0.324 0.908 -0.339 0.902
Hotel dummy -3.227 0.303 -1.388 0.684 1.580 0.666 -2.669 0.419 -2.622 0.417
Diversified dummy 3.372 0.277 3.627 0.282 2.903 0.348 3.240 0.316 3.282 0.299
Age 0.000 0.056 0.000 0.033 0.000 0.312 0.000 0.033 0.000 0.029
Canada dummy 8.666 0.253 9.312 0.254 6.958 0.340 8.395 0.283 8.348 0.276
Bonus to total compensation -0.344 0.094 -0.458 0.109 -0.180 0.405 -0.199 0.336 -0.150 0.462
No. of observations 5430 3555 4733 5372 5372
Cross-sectional units 251 248 221 248 248
S.E. of regression 15.333 15.034 14.346 15.326 15.138
Akaike criterion 45113.60 29414.390 38699.870 44627.380 44495.600
'Within' variance 43.393 42.759 37.266 41.753 41.063
'Between' variance 152.76 179.785 140.866 163.261 156.360
Wald (joint) time dummies (?2) 139.638 0.000 145.879 0.000 247.071 0.000 175.046 0.000 178.145 0.000
Hausman test (?2) 40.798 0.689 28.233 0.982 14.778 1.000 145.98 1.000 9.977 1.000
Notes The table shows the results of the panel
data regressions using random effects with debt
to assets ratio as dependent variable. The impact
of an alternative proxy is tested for growth
opportunities in model 4, for size in model 5,
for profitability in model 6 and in model 7 with
an additional interaction variable to
differentiate the impact of the alternative proxy
for profitability on REOCs. The time-series are
from 1 to 41 periods. Time dummies are not
displayed. indicates significance at the 10
level, indicates significance at the 5 level,
indicates significance at the 1 level.
16
Sensitivity Analysis Property type sensitivity
Preferred Model Preferred Model Preferred Model Model 8 Model 8 Model 8 Model 9 Model 9 Model 9 Model 10 Model 10 Model 10 Model 11 Model 11 Model 11
coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value
Constant -28.366 0.000 -26.934 0.000 -22.821 0.000 -19.826 0.000 -23.124 0.000
Market to book 0.161 0.000 0.971 0.000 0.160 0.000 0.159 0.000 0.162 0.000
Residential dummy -0.909 0.000
Shopping dummy -0.693 0.000
ln(assets) 4.464 0.000 4.642 0.000 4.101 0.000 4.092 0.000 4.485 0.000
Residential dummy 1.491 0.000
Shopping dummy 0.378 0.010
Real estate to assets 0.141 0.000 0.148 0.000 0.137 0.000 0.099 0.000 0.140 0.000
Residential dummy 0.251 0.000
Shopping dummy 0.055 0.004
REOC dummy 5.984 0.026 0.331 0.882 4.065 0.077 2.944 0.181 -1.667 0.472
Residential dummy 31.981 0.000
Shopping dummy -2.644 0.840
External mgmt dummy 4.940 0.053 0.876 0.693 3.646 0.106 3.436 0.118 0.055 0.980
Return on assets -0.107 0.000 -0.110 0.000 -0.114 0.000 -0.127 0.000 -0.106 0.000
Insider Ownership 0.063 0.000 0.072 0.000 0.066 0.000 0.070 0.000 0.065 0.000
Age 0.000 0.056 0.000 0.198 0.000 0.178 0.000 0.154 0.000 0.148
Canada dummy 8.666 0.253 18.339 0.017 6.338 0.412 4.604 0.542 7.220 0.364
Bonus to total compensation -0.344 0.094 -0.382 0.061 -0.350 0.087 -0.302 0.138 -0.371 0.072
No. of observations 5430 5430 5430 5430 5430
Cross-sectional units 251 251 251 251 251
S.E. of regression 15.333 16.096 15.934 15.830 15.762
Akaike criterion 45113.60 45637.430 45528.180 45457.000 45409.650
'Within' variance 43.393 40.330 39.177 40.455 41.393
'Between' variance 152.76 161.928 152.839 152.609 160.569
Wald (joint) time dummies (?2) 139.638 0.000 137.918 0.000 136.382 0.000 143.867 0.000 134.338 0.000
Hausman test (?2) 40.798 0.689 38.977 0.820 319.225 0.000 87.530 0,000 42.479 0.621
Notes The table shows the results of the panel
data regressions using random effects with debt
to assets ratio as dependent variable. The
changing impact of growth opportunities for the
special cases of residential and shopping
REITs/REOCs is measured in model 8, of size in
model 9, of asset tangibility in model 10 and of
REOCs in model 11. The time-series are from 1 to
41 periods. Time dummies are not displayed.
indicates significance at the 10 level,
indicates significance at the 5 level,
indicates significance at the 1 level.
17
Sensitivity Analysis Property type sensitivity
(contd)
Preferred Model Preferred Model Preferred Model Model 12 Model 12 Model 12 Model 13 Model 13 Model 13 Model 14 Model 14 Model 14
coefficient p-value p-value Coefficient p-value p-value coefficient p-value p-value coefficient p-value p-value
Constant -28.366 0.000 -22.411 0.000 -9.778 0.024 -22.686 0.000
Market to book 0.161 0.000 0.161 0.000 0.159 0.000 0.166 0.000
ln(assets) 4.464 0.000 4.411 0.000 3.532 0.000 4.400 0.000
Real estate to assets 0.141 0.000 0.142 0.000 0.133 0.000 0.143 0.000
REOC dummy 5.984 0.026 0.510 0.815 0.515 0.815 2.481 0.262
External mgmt dummy 4.940 0.053 -2.800 0.224 -0.066 0.976 1.102 0.614
Residential dummy 39.472 0.000
Shopping dummy 11.470 0.135
Return on assets -0.107 0.000 -0.107 0.000 -0.173 0.000 -0.104 0.000
Residential dummy 0.278 0.000
Shopping dummy -0.127 0.001
Insider Ownership 0.063 0.000 0.065 0.000 0.066 0.000 0.016 0.309
Residential dummy 0.051 0.248
Shopping dummy 0.116 0.000
Age 0.000 0.056 0.000 0.090 0.000 0.152 0.000 0.185
Canada dummy 8.666 0.253 17.568 0.019 16.759 0.027 16.496 0.030
Bonus to total compensation -0.344 0.094 -0.352 0.088 -0.324 0.111 -0.342 0.096
No. of observations 5430 5430 5430 5430
Cross-sectional units 251 251 251 251
S.E. of regression 15.333 15.588 15.558 15.867
Akaike criterion 45113.60 45289.320 45268.780 45481.820
'Within' variance 43.393 41.393 40.600 41.195
'Between' variance 152.76 154.607 158.613 156.534
Wald (joint) time dummies (?2) 139.638 0.000 138.765 0.000 155.335 0.000 135.704 0.000
Hausman test (?2) 40.798 0.689 56.032 0.147 31.459 0.969 103.426 0.000
Notes The table shows the results of the panel
data regressions using random effects with debt
to assets ratio as dependent variable. The
changing impact of externally managed REITs/REOCs
for the special cases of residential and shopping
REITs/REOCs is measured in model 12, of
profitability in model 13 and of insider
ownership in model 14. The time-series are from 1
to 41 periods. Time dummies are not displayed.
indicates significance at the 10 level,
indicates significance at the 5 level,
indicates significance at the 1 level.
18
Variance Inflation Factors
Minimum possible value 1.0 Minimum possible value 1.0
Values gt 10.0 may indicate a collinearity problem Values gt 10.0 may indicate a collinearity problem

Market to book 1.012
ln(assets) 1.383
Real estate to assets 1.44
REOC dummy 2.34
External mgmt dummy 1.657
Return on assets 1.059
Insider ownership 1.319
Residential dummy 1.607
Shopping dummy 1.831
Office dummy 1.602
Hotel dummy 2.398
Diversified dummy 1.499
Age 1.117
Canada dummy 1.044
Bonus to total compensation 1.286

VIF(j) 1/(1 - R(j)2), where R(j) is the multiple correlation coefficient VIF(j) 1/(1 - R(j)2), where R(j) is the multiple correlation coefficient
between variable j and the other independent variables between variable j and the other independent variables

Properties of matrix X'X Properties of matrix X'X

1-norm 7.7927016e011 1-norm 7.7927016e011
Determinant 7.8064223e065 Determinant 7.8064223e065
Reciprocal condition number 2.2254893e-011 Reciprocal condition number 2.2254893e-011