InterestRate Sensitivity of Real Estate Investment Trusts
Post on: 24 Июль, 2015 No Comment
Page 1
THE JOURNAL OF REAL ESTATE RESEARCH
Interest-Rate Sensitivity K c. Chen‘
of Real Estate ‘’“‘e’’’- “8
Investment Trusts
Abstract. This paper addresses the issue of whether RElTs are sensitive to changes
in short—term and long-term interest rates. REITS were found to be sensitive to
Introduction
RElTs are pooled real estate funds that provide individual investors, as well as institutions,
the opportunity to invest in income—producing real estate properties, mortgages, joint ventures,
and other hybrid structures. In general, REITS take two forms; equity and mortgage REITS.
The equity REITS specialize in the ownership of income—producing real estate properties,
whereas the mortgage REITs primarily hold long-terrn as well as short—term construction
loans, and mortgages on commercial properties.
The REIT industry grew slowly until the late 1960s. By the mid—1970s, the economic slowdown
and the level of overbuilding sent the real estate industry into a severe recession. Many
mortgage REITS went into bankruptcy under the burden of non—eaming real estate assets.
However, many of the equity—oriented trusts survived the problems of the mid—1970s. This
result has been attributed to their adherence to property ownership, rather than mortgage-
lending activities, and because they were reluctant to leverage with short—term, floating-rate
debt.
Historically, dividend yield has always been a significant component of the return from a
REIT share investment. By complying with IRS regulations, REITs are exempt from corporate
income taxes if they distribute 95% of net income in the form of dividends to shareholders.
By virtue of this pass—through feature, REITs generally have relatively high dividend yields,
with 8.6% and 12.1% in 1985 for equity and mortgage REITS, respectively.‘ Consequently,
REITS, like other high—dividend—yield stocks such as utilities, may possess a high degree of
sensitivity to interest—rate fluctuations.
The objective of this paper is twofold. First, based on Mertons (1973) intertemporal capital
‘Theodore F. Bnx Professor of Finance, Department of Finance, California State University-Fresno, Fresno,
California 93740.
”Associate Professor of Finance, University of Pacific, Stockton, California 95204.
Date Revised——August 1988; Accepted—September 1988
asset pricing model (ICAPM), we attempt to investigate empirically whether both equity and
mortgage RElTs are sensitive to changes in interest rates. In the context of the ICAPM, the
The rest of the paper is organized as follows. The second section reviews the relevant
There has been an abundant number of studies on RElTs in recent years. Most of them
have focused on evaluating the performance of RElTs. Smith and Shulman [1976] used a
CAPM framework in comparing the performance of RElTs and closed—end investment funds.
They found that, in general, RElTs did not offer significantly better retums than closed—end
funds over the period 1963
1976 when compared with the S&P 500 index. Kuhle and Walther
[1986] and Kuhle, Walther and Wurtzebach [1986], however, provided evidence that REIT
stocks had been a significantly better investment asset than common stocks over the time
period 1977 to 1985. Burns and Epley [1982] found that the mixed-asset efficient frontier
containing both RElTs and stock returns was superior to both of the single—asset efficient
frontiers at every risk and return level during the 1970-1979 period. Miles and Estey [1982]
and Miles and McCue [1982] examined the portfolios of the RElTs and compared them to
those of the commingled real estate funds [CREFS] managed by insurance companies and
bank trust departments. Kuhle [1987] showed that equity REIT stocks provided better benefits
in portfolio risk reduction than common stocks and mortgage REIT stock over the period
1980-1985. However, the overall performance of mixed portfolios of common stock and RElTs
was not significantly different from that of portfolios of only common stock.
Recently, Titman and Warga [1986] analyzed the retums of RElTs and examined their risk-
adjusted performance using_ both single index (i.e. CAPM) and multiple index (i.e. the
arbitrage pricing theory (AP1)) models. Their findings indicated that the performance rankings
of RElTs were not sensitive to the risk—adjustment model, and neither the CAPM nor APT-
based techniques were powerful enough to provide reliable evaluations of real estate portfolio
managers.
Explicit Model and Testable Hypotheses
The following analysis is based on Mertons [1973] intertemporal capital asset pricing model:
E(R.) — or = B. lE(R. ) — a] + I32lE(R;..)
01], (1)
where
E(R,) expected retum on an asset at time t;
(0.62) (7.44) (— 0.35)
1YTB 0.0032 1 .29 — 0.20 0.46
(0.53) (7.71 ) (— 0.09)
20YTB 0.0075 1 .16 — 88.77 0.49
(1.23) (7.35) “ (— 2.09)‘
3MTB 0.0083 0.60 — 15.23 0.61
(2.24) (7.85) ‘ (— 4.08)“
6MTB 0.0086 0.58 -18.19 0.62
(2.31)- (7.52)- (—4.39)
1980-1985
1Y1B 0.0088 0.57 — 21.53 0.62
(2.36)* (7.34) (— 4.42)
20YlB 0.0087 0.63 — 19.06 0.56
(2.36) (7.39) (— 2.43)‘
t—values are in parentheses.
Significant at 99% level
‘Significant at 95% level
E(R. ) = expected retum on a hedge portfolio constructed to have a
covariance with each assets return that is identical to the
covariance between the changes in the state variable of interest
and the assets retum;
a = the risk-free rate of interest; and
B. [3, = multiple regression coefficients.
Equation (1) states that, in equilibrium, investors receive higher expected return for bearing
market (systematic) risk and for bearing the risk of unfavorable shifts in the investment
opportunity set. The ICAPM is a more generalized model because it reduces to the traditional
Page 4
16 THE JOURNAL OF REAL ESTATE RESEARCH
capital theory, and to practitioners. It is observable, satisfies the condition of being
stochastic over time, and while it is surely not the sole determinant of yields on
I, = E17, + Er. (4)
Taking the first difference in variables of (4) and substituting into (3) yields:
R: = 30 + BIRMI + B2AE7: + B3AE7: + 51- (5)
The variable, AEr. deserves special attention here, because the empirical evidence of Fama
[1976] and Fama and Gibbons [1982; 1984] suggests that the expected real rate is a random
walk. In addition B1 and B, can help detect whether interest—rate sensitivity is attributable
to changes in expected inflation, changes in the expected real rate, or both.
Why are RElTs likely to be sensitive to changes in expected inflation and changes in the
expected real rate, respectively? We provide the following explanations and hope that they
can further enhance our understanding of the effect of these variables on REITs performance.
First, higher—than—expected inflation generally results in a lower real present value of existing
mortgage loans and would thus reduce the mortgage REITs stock price. Likewise, higher
inflation will also increase the equity RElTs expenses. If regulators do not allow a complete
pass—through of expenses, the earnings of those RElTs will be reduced.
Secondly, RElTs offer higher dividend yields than the average corporation because they
must pay out at least 95% of net earnings to shareholders to avoid payment of corporate
income taxes. The evidence from previous studies, that REITs have performed well over past
years, suggests that RElTs’ shareholders pay a premium for high dividends. If this premium
is based on the present values of dividends, a rise in the real interest rate will reduce the
present value of RElTs dividends more than other low—dividend—paying stocks.
Period Rate 8. B, B2 Fl’
3MTB — 0.0017 1.40 — 23.78 0.41
(— 0.23) (6.91) (— 0.94)
6MTB — 0.0021 1.40 — 19.56 0.41
(— 0.28) (6.77) (-0.76)
1YTB — 0.0027 1.43 -14.13 0.41
(-0.36) (7.00) (-0.53)
20YlB 0.0075 1.33 — 105.93 0.43
(0.18) (6.88)“ (— 2.04)’
3MTB 0.0073 0.73 — 20.71 0.61
(1.85) (7.77)“ (— 4.76)
6MTB 0.0076 0.70 — 24.41 0.62
(1.95) (7.40)“ (—5.06)*
1980-1985
1YTB 0.0081 0.66 — 31.30 0.64
(2.08) (7.15) (- 5.62)
20YTB 0.0094 0.65 -44.03 0.62
(2.29)‘ (6.70)“ (- 4.98)
t-values are in parentheses.
Significant at 99% level
‘Significant at 95% level
changed its operating procedures in October 1979, shifting from interest—rate targeting toward
money—supply targeting.
To be included in the sample, RElTs must satisfy the following requirements: (1) They can
be classified as either equity or mortgage REITs.‘ Company annual reports, Moodys Bank
and Finance Manual and Value Line Investment Survey, are used to aid classifications; and (2)
they must have a sufficiently large number of monthly retums during the same period. The
sample consists of thrity-two equity REITS and twenty-two mortgage REITS. Monthly returns
are obtained from the CRSP Monthly Return File. We then construct an equally weighted
portfolio of equity and mortgage REITS, respectively.
For the state variable, four measures of the interest rate are used. The first measure is an
index of yields on twenty-year U.S. government bonds (ZOYTB), because Merton [1973]
suggested the use of a long—term interest rate for the state variable. In addition, we also use
three short-term interest rates: Indexes of yields on three—month (SMTB), six—month (6MTB),