Get documents about "An Examination of Volatility Spillovers in REIT Returns
Document Sample
An Examination of Volatility Spillovers in
REIT Returns
Executive Summary. This study examines whether vol- by Simon Stevenson*
atility in a variety of equity and fixed income sectors
based in the United States influence the monthly vola-
tility of real estate investment trusts (REITs). The anal- Introduction
ysis is based on two alternative GARCH and EGARCH
specifications and reveals a number of issues in relation The relationship between indirect real estate se-
to volatility spillovers. As with existing evidence with re- curities, such as real estate investment trusts
gard to returns, a causal relationship is revealed from (REITs), and both the general capital markets and
Equity REITs to the other REIT sectors. In addition, the the direct real estate market have been subject to
main influencing asset classes with regard to REITs are a large degree of research in recent years. The vast
small cap stocks and value stocks, which given the char- majority of these studies have examined the rela-
acteristics of REITs, is in line with expectations. Finally,
tionship in the first moment of the return series,
Mortgage REITs are not generally influenced by volatil-
as well as the issue of integration and segmenta-
ity in the fixed income sector.
tion. Liu, Hartzell, Greig and Grissom (1990) find
evidence that the direct real estate market is seg-
mented in terms of the capital markets, while they
cannot reject the hypothesis that REITs are inte-
grated with common equities. Evidence with re-
gard to the integration of REITs and common
stocks has also been found in studies such as Mei
and Lee (1994) and Li and Wang (1995). Ling and
Naranjo (1999) use multi-factor asset pricing tech-
niques to examine whether there is any evidence
of integration between direct real estate, REITs
and common stocks. As with previous studies,
REITs are found to be integrated with non-real es-
tate equities, however, no such evidence is found
in relation to the direct market, even when this
data is adjusted for smoothing. Wilson, Okunev
and Ta (1996) examine the Australian, American
and British indirect real estate and equity mar-
kets, finding in all three markets an absence of any
cointegrating relationships, while He (1998) finds
evidence to support the notion that a causal rela-
tionship exists from Equity REITs to Mortgage
REITs in the United States, with further evidence
finding that the two sectors are cointegrated. The
study also finds evidence of instantaneous linear
* University College Dublin, Dublin, Blackrock, County Dublin, feedback between equity and mortgage REITs
Ireland or simon.stevenson@ucd.ie. through the use of Geweke causality tests.1
Journal of Real Estate Portfolio Management 229
Simon Stevenson
There is also a large literature that has examined using both a Generalized ARCH (GARCH) and Ex-
the relationship between the direct and indirect ponential GARCH specification, and then based on
real estate markets. Studies such as Myer and the initial modeling examines the issue of volatil-
Webb (1994), Barkham and Geltner (1995) and ity spillovers.
Seiler, Webb and Myer (1999) all find evidence of
the indirect market leading the direct, indicating The remainder of the article is laid out as follows.
that real estate securities incorporate information The following section details the data require-
into prices quicker than appraisal-based values.2 ments. This is followed by a discussion of the em-
pirical findings and concluding comments.
The majority of studies that analyze REITs, and
their relationship with equities, have concentrated
on linkages in their returns (e.g., Wang, Erickson Data Requirements
and Chan, 1995), however, no study has as yet ex- Monthly data is utilized over the period from Jan-
amined the volatility linkages between the two uary 1975 to April 2001. The NAREIT indices are
markets. This study aims to examine the linkages used for the REIT sectors, allowing an analysis of
between REITs and other capital market asset not only the overall market, but also the Equity,
classes in the context of the second moment of the Mortgage and Hybrid sub-markets. Those markets
return distribution, thereby examining the effect examined in relation to REITs include a variety of
referred to as volatility spillovers. Volatility spill- equity sectors, U.S. Government Bonds and U.S.
overs have been extensively examined in the gen- Treasury Bills. The analysis of the fixed income
eral financial economics literature, often in the sector is specifically related to the examination of
analysis of international markets. Previous studies Mortgage REITs and the possible linkages be-
of volatility spillovers include Bae and Karolyi tween the two sectors.
(1994), Bekeart and Harvey (1997), De Santis and
Wooldridge (1997), Hamao, Masulis and Ng (1990), The equity series’ analyzed are the S&P 500, the
Karolyi (1995), Koutmos and Booth (1995) and Ng, NASDAQ Composite, the S&P 600 Small Cap In-
Chang and Chou (1991). In most cases there is sig- dex, the MSCI Value Stock Index and the MSCI
nificant evidence of volatility spillover effects being Growth Stock Index. The rationale behind the use
present in the series’ analyzed. In addition, recent of alternative equity indices is that it allows an
studies, such as Booth, Martikainen and Tse examination of whether the relationship differs be-
(1997) have found evidence of asymmetric trans- tween REITs and different sectors of the broader
missions in volatility transmissions.3 capital markets. Studies such as Wang, Erickson
and Chan (1995) have found evidence to support
The analysis of volatility and volatility spillovers the hypothesis that REITs behave more in line
are analyzed through the utilization of Autore- with small stocks than large firms. In addition,
gressive Conditional Heteroscedasticity (ARCH)- Stevenson (2001) finds, based on Style Analysis of
based modes, thereby allowing the simultaneous the property companies in the United Kingdom,
modeling of both the first and second moments of that they derive more of their investment attri-
the return series and providing a more efficient butes from value stocks than growth stocks. The
means of modeling time series. The use of ARCH- use of more detailed equity indices in this study
based models allows us to overcome the common should allow an examination of whether such first
problem that conventional econometric time-series moment derived results also apply with regard to
models that often assume that the variance of the the second moment of the return distribution.
error term is constant. However, this assumption
of homoscedasticity is often problematic in the
analysis of financial time series, with the cluster-
Empirical Analysis
ing of volatility being a prime example of a situa- The principles of Autoregressive Conditional Het-
tion where this assumption may be violated. This eroscedasticity (ARCH) were developed by Engle
study examines the volatility of the REIT market, (1982). The model can be used to generate a series
230 Vol. 8, No. 3, 2002
Volatility Spillover in REIT Returns
Exhibit 1
GARCH (1,1) Model
1 2 1 2
Equity REITs 1.206** 2.528* 0.069** 0.735***
Mortgage REITs 0.680* 6.830*** 0.200*** 0.512***
Hybrid REITs 0.934* 2.055** 0.086*** 0.823***
U.S. Bonds 0.065 0.071** 0.198*** 0.797***
S&P 500 1.383*** 0.400** 0.076** 0.906***
NASDAQ 0.983* 1.239*** 0.100*** 0.880***
U.S. T-Bills 0.006 0.000*** 0.534*** 0.448***
S&P 600 Small Cap 1.144** 6.760*** 0.000 0.752
MSCI Value Index 1.119** 18.794*** 0.109 0.000
MSCI Growth Index 0.570 4.839*** 0.051** 0.843***
Notes: Exhibit 1 reports the primary coefficients for the base GARCH (1,1) model. 1 is the constant in the mean equation, 2 is the constant
in the variance equation, 1 is the coefficient for the lagged squared residuals and 2 is the conditional variance coefficient.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
of changing volatility, essentially suggesting that Nelson (1991) adapted the GARCH form to incor-
large and small forecast errors have a tendency to porate volatility clustering and the leverage effect
occur in clusters. A simple ARCH process can be that exists in the majority of financial data. The
formulated as follows: specification proposed, known as Exponential
GARCH (EGARCH), allows for an asymmetric re-
Rt b0 t , (1) sponse to positive and negative changes in returns.
The motivation for such a variation comes in part
where: from empirical evidence that would suggest that
negative price changes tend to increase volatility
2
t t 1 N(0, ht ) more than positive price changes. Another key ad-
vantage of the EGARCH specification is that there
h2
t a0 a1 2
t 1
is no restriction placed on the parameters of the
conditional variance. Engle and Ng (1993) use
The parameter b0 is a constant term, which is typ-
daily data to examine the asymmetric effect, by
ically estimated to be close to, or equal to, zero and
comparing different conditional variance specifi-
t 1 is the information set available at time t 1.
2 cations, finding that the EGARCH model behaves
It is assumed that the volatility (ht ) is a determin-
remarkably well.4
istic function of past returns and is therefore con-
ditional on 2 1. The Generalized (GARCH) form,
t
Exhibits 1 and 2 detail the coefficients for the
proposed by Bollerslev (1986), allows for lagged
GARCH (1,1) and EGARCH specifications for each
variances and the further lagging of the error
individual asset class. In most cases the coeffi-
term. The primary GARCH form that is examined
cients reported are in line with expectations. The
is a GARCH (1,1,) and can be displayed as:
main exception is with regard to the variables for
which the constant in the mean equation is statis-
Rt b0 t (2) tically significant. With the exception of the small
2 2 2
h t a0 a 1 t 1 ah
2 t 1 cap and value stocks, all of the coefficients with
regard to the conditional variance and the lagged
Where the parameter indicates the significance of squared residuals are significant at conventional
the lagged variance term in the generalized model. levels. Diagnostic results for the two models are
Journal of Real Estate Portfolio Management 231
Simon Stevenson
Exhibit 2
EGARCH Model
1 2 1 2
Equity REITs 1.180** 0.354* 0.178** 0.863***
Mortgage REITs 0.361* 0.457*** 0.213*** 0.855***
Hybrid REITs 1.007** 0.252** 0.136** 0.918***
U.S. Bonds 0.020 0.028* 0.237*** 0.977***
S&P 500 0.014 0.054* 0.079* 0.982***
NASDAQ 1.244** 0.045* 0.141*** 0.991***
U.S. T-Bills 0.018 0.564*** 0.415*** 0.916***
S&P 600 Small Cap 0.986** 3.044*** 0.000 0.061
MSCI Value Index 0.937** 2.928*** 0.138*** 0.011
MSCI Growth Index 0.946** 0.116** 0.074** 0.972***
Notes: Exhibit 2 reports the primary coefficients for the base EGARCH model. 1 is the constant in the mean equation, 2 is the constant in
the variance equation, 1 is the coefficient for the lagged squared residuals and 2 is the conditional variance coefficient.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
displayed in Exhibits 3–5. The results show little of Equity REITs leading the mortgage sector in
evidence of diagnostic problems. The primary issue terms of return movements. The author argues
being the Ljung-Box Q tests with regard to Treas- that this may be due to the equity sector reacting
ury bills. quicker in incorporating real estate fundamentals
into prices. A further factor may relate to the rel-
In order to examine the effect of volatility spill- ative size of the two sectors and the liquidity in
overs, the conditional variance from the indepen- the two. As the equity sector is larger and more
dent variable is added to the equation for each de- liquid, this may also lead to it reacting quicker to
pendent series in turn. The results reported in Ex- changes in either returns or volatility.
hibit 6 detail the coefficient specifically relating to
the exogenous variable.5 The tests were run with One of the primary reasons behind the examina-
the conditional variance included—both contem- tion of a number of different capital market asset
poraneously and lagged one period. The results classes was to examine whether different assets af-
show that there exists some degree of spillover ef- fect REITs differently and whether different asset
fects both between the various REIT sectors and classes affect different REIT sectors. In particular,
from other asset classes to REITs. Under each sce- the inclusion of bonds and Treasury bills was spe-
nario, whether the lag is included or not, and with cifically aimed at examining their potential influ-
either the GARCH or EGARCH specification, Eq- ence on Mortgage REITs, while the inclusion of
uity REITs significantly impact the volatility of small cap, value and growth stocks was aimed at
Mortgage REITs and with one exception, Hybrid seeing whether these sectors have varying degrees
REITs. In contrast, little reverse effects are notice- of influence on REITs in terms of volatility. Due to
able, with significant coefficients only being ob- the nature of the REIT sector and the underlying
served in one case for hybrid REITs and in no case assets of the firms it is not surprising that previous
for the Mortgage sector. These findings illustrate studies, such as Wang, Erickson and Chan (1995),
the dominance of the Equity sector in terms of both have found evidence that REITs behave more in
size and trading volume. In addition, they also line with small and value stocks rather than
support the causality results reported in He (1998) growth stocks. It is of interest therefore that the
with regard to the first moment of the return dis- S&P 500 has a greater influence on volatility in
tribution. In that study the author found evidence each REIT sector than any other equity sector and
232 Vol. 8, No. 3, 2002
Volatility Spillover in REIT Returns
Exhibit 3
Diagnostics for GARCH (1,1) Model
4 8 12 16 20 24
Panel A: Ljung-Box Q Test in Residuals
Equity REITs 1.398 7.292 12.822 14.156 23.534 24.635
Mortgage REITs 1.973 5.456 9.378 12.067 127.305 20.085
Hybrid REITs 0.892 1.827 4.476 5.096 7.124 8.885
U.S. Bonds 1.778 4.174 11.803 20.466 21.840 22.854
S&P 500 0.993 11.396 14.254 16.959 21.515 23.414
NASDAQ 3.288 3.520 4.542 8.015 12.325 13.501
US T-Bills 5.384 13.289* 16.391 21.254 24.723 28.782
S&P 600 Small Cap 5.215 6.019 11.867 14.722 18.600 21.555
MSCI Value Index 5.568 8.709 11.620 12.988 17.208 20.841
MSCI Growth Index 2.582 3.767 8.224 10.879 15.404 17.712
Panel B: F-test of No ARCH vs. ARCH in Residuals
Equity REITs 0.228 0.545 0.433 0.377 0.334 0.338
Mortgage REITs 0.078 0.313 0.942 0.753 0.886 0.808
Hybrid REITs 0.344 0.252 0.244 0.269 0.287 0.307
U.S. Bonds 0.309 0.618 0.715 0.736 0.831 0.826
S&P 500 0.085 0.285 0.421 0.425 0.402 0.340
NASDAQ 0.148 0.186 0.252 0.358 0.297 0.268
U.S. T-Bills 0.694 0.537 1.114 0.829 1.296 1.205
S&P 600 Small Cap 0.028 0.106 0.175 0.202 0.181 0.162
MSCI Value Index 0.017 0.018 0.110 0.114 0.108 0.116
MSCI Growth Index 0.004 0.044 0.114 0.148 0.124 0.112
Panel C: Ljung-Box Q Test in Squared Residuals
Equity REITs 0.850 4.863 5.828 6.963 8.255 10.190
Mortgage REITs 0.345 2.599 12.436 13.942 17.675 19.498
Hybrid REITs 1.230 1.741 3.035 4.079 5.703 8.199
U.S. Bonds 1.319 5.026 9.077 11.493 15.512 20.090
S&P 500 0.472 2.321 5.194 7.085 8.571 9.008
NASDAQ 0.643 1.439 3.145 6.471 6.758 7.346
U.S. T-Bills 2.587 4.318 12.524 13.054 23.425 25.611
S&P 600 Small Cap 0.079 0.871 2.138 3.581 4.149 4.561
MSCI Value Index 0.082 0.159 1.521 2.148 2.564 3.406
MSCI Growth Index 0.020 0.378 1.481 2.658 2.800 3.076
Notes: Exhibit 3 reports the primary diagnostic tests for the GARCH (1,1) model.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
in particular in relation to the Mortgage and Hy- for both Equity and Hybrid REITs are with regard
brid REIT sub-sectors. Significant coefficients are to positive coefficients, the Mortgage sector has a
reported in each case for these two sectors. It is consistently negative and statistically significant
also noticeable that while the significant findings relationship with the S&P.
Journal of Real Estate Portfolio Management 233
Simon Stevenson
Exhibit 4
Diagnostics for EGARCH Model
4 8 12 16 20 24
Panel A: Ljung-Box Q Test in Residuals
Equity REITs 1.634 8.138 13.457 14.766 24.303 25.456
Mortgage REITs 2.328 5.572 9.070 12.833 18.677 22.013
Hybrid REITs 1.114 2.374 4.511 4.824 7.897 9.330
U.S. Bonds 3.271 5.315 12.869 21.078 23.039 24.103
S&P 500 1.782 9.795 13.042 16.544 20.726 21.982
NASDAQ 3.627 3.870 4.822 7.969 12.225 13.583
U.S. T-Bills 4.688 14.129** 21.785*** 26.461** 30.342** 31.114
S&P 600 Small Cap 4.955 6.164 11.185 12.941 18.043 20.090
MSCI Value Index 5.277 9.086 10.830 12.623 19.910 23.242
MSCI Growth Index 3.158 4.585 8.085 10.580 14.842 17.270
Panel B: F-test of No ARCH vs. ARCH in Residuals
Equity REITs 0.297 0.690 0.544 0.468 0.414 0.399
Mortgage REITs 0.148 0.256 0.862 0.702 1.027 0.908
Hybrid REITs 0.307 0.328 0.522 0.509 0.569 0.514
U.S. Bonds 0.301 0.405 0.559 0.612 0.703 0.781
S&P 500 0.072 0.397 0.871 0.712 0.691 0.587
NASDAQ 0.088 0.139 0.254 0.318 0.284 0.272
U.S. T-Bills 0.751 0.906 0.818 0.813 0.908 0.916
S&P 600 Small Cap 0.118 0.165 0.252 0.335 0.337 0.296
MSCI Value Index 0.330 0.195 0.255 0.268 0.250 0.241
MSCI Growth Index 0.053 0.071 0.143 0.164 0.140 0.130
Panel C: Ljung-Box Q Test in Squared Residuals
Equity REITs 1.119 6.071 7.132 8.459 10.176 12.049
Mortgage REITs 0.696 2.266 10.873 12.156 18.655 20.602
Hybrid REITs 1.134 2.228 6.275 7.727 11.714 13.644
U.S. Bonds 1.303 3.451 7.552 10.153 13.222 18.953
S&P 500 0.204 3.294 9.841 10.771 13.818 14.380
NASDAQ 0.391 1.157 3.216 5.849 6.545 7.459
U.S. T-Bills 3.003 6.974 9.231 13.590 17.869 23.596
S&P 600 Small Cap 0.527 1.343 3.063 5.854 7.246 7.556
MSCI Value Index 1.380 1.549 3.296 4.905 5.651 6.847
MSCI Growth Index 0.220 0.628 1.810 2.917 3.180 3.593
Notes: Exhibit 4 reports the primary diagnostic tests for the EGARCH model.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
In terms of the impact from other sectors, there is observed. This is particularly noticeable with re-
a noticeable difference between the findings when gard to the small cap and value stock sectors. In
the GARCH (1,1) model is used in comparison to the original specification, no significant findings
the EGARCH specification. When the original are reported with respect to small cap stocks gen-
model is tested very few significant coefficients are erally and only two are found in relation to value
234 Vol. 8, No. 3, 2002
Volatility Spillover in REIT Returns
Exhibit 5
Sign and Size Bias Tests
Negative Size Positive Size Joint Test for
Sign Bias Test Bias Test Bias Test Three Effects
Panel A: GARCH (1,1) Model
Equity REITs 0.792 0.034 0.343 0.777
Mortgage REITs 0.542 0.536 0.028 0.526
Hybrid REITs 0.121 0.901 0.581 0.852
U.S. Bonds 1.774* 0.924 0.552 1.073
S&P 500 1.218 1.348 1.011 1.835
NASDAQ 2.214** 0.623 0.887 1.855
U.S. T-Bills 2.245** 1.027 1.402 1.775
S&P 600 Small Cap 0.730 0.408 0.740 1.681
MSCI Value Index 0.851 0.221 0.648 1.457
MSCI Growth Index 1.793* 0.722 0.452 1.330
Panel B: EGARCH Model
Equity REITs 0.563 0.190 0.344 0.374
Mortgage REITs 0.696 0.390 0.609 0.379
Hybrid REITs 0.002 0.217 0.357 0.114
U.S. Bonds 1.543 1.031 0.887 0.855
S&P 500 0.370 0.417 1.830 1.704
NASDAQ 2.133** 0.438 0.588 2.032
U.S. T-Bills 2.073 0.343 1.319 1.546
S&P 600 Small Cap 0.364 0.605 0.309 0.166
MSCI Value Index 0.447 0.482 0.302 0.110
MSCI Growth Index 1.680* 0.409 0.188 1.524
Notes: Exhibit 5 reports the sign and sign bias tests for both the GARCH and EGARCH models.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
stocks. These are for Mortgage REITs when no lag idea that REITs are more like value stocks and
is included and Equity REITs when the conditional that they behave in line with small stocks. In ad-
variance is included at one lag. However, with the dition, Equity REITs are not significantly influ-
EGARCH specification, significant findings are re- enced by the broader equity market, as proxied by
ported for each sector in all cases. Equity and Hy- the S&P 500, when the EGARCH specification is
brid REITs see their volatility significantly posi- used.
tively affected by volatility in both small and value
stocks in both EGARCH models. With Mortgage The results with regard to Mortgage REITs are in-
REITs, the only difference is that the coefficient is teresting in a number of respects. The first issue
negative when the variance is included at a lag. It concerns the lack of significant findings in relation
is also of interest that while a number of signifi- to volatility spillovers that originate in the fixed
cant coefficients are found in relation to growth income market. The bond and Treasury bill sectors
stocks, all significant findings are for negatively were included in the analysis to examine whether
signed coefficients. The only positive coefficient re- interest rate linked volatility had an impact on the
ported for growth stocks is for the lagged Equity sector. No significant findings are reported for
REIT coefficient and this is not significant at con- Treasury bills, and while significant coefficients
ventional levels. These findings would confirm the are found with the GARCH specification, both are
Journal of Real Estate Portfolio Management 235
236
Simon Stevenson
Vol. 8, No. 3, 2002
Exhibit 6
Volatility Spillover Results
Equity REITs Mortgage REITs Hybrid REITs Bonds S&P 500 NASDAQ Treasury Bills Small Cap Value Stocks Growth Stocks
Panel A: GARCH Model No Lag
Equity REITs — 0.009 0.016 0.062 0.079* 0.001 249.333 7.794 0.242 0.022
Mortgage REITs 3.299*** — 0.944*** 0.268* 0.167*** 0.009 99.397 4.116 1.183*** 0.090*
Hybrid REITs 0.375*** 0.038 — 0.018 0.276** 0.027 391.730 5.046 0.206 0.013
Panel B: GARCH Model with Lag
Equity REITs — 0.011 0.012 0.023 0.072 0.000 93.628 5.259 0.272** 0.029
Mortgage REITs 0.588*** — 0.183** 0.267* 0.166*** 0.001 75.921 13.860 0.057 0.119***
Hybrid REITs 0.429*** 0.002 — 0.052 0.295** 0.030 91.541 1.563 0.213 0.006
Panel C: EGARCH Model No Lag
Equity REITs — 0.050 0.197* 0.002 0.061 0.003 722.904* 0.350*** 0.337*** 0.024**
Mortgage REITs 1.361*** — 0.801*** 0.134 0.097*** 0.009 19.028 0.372*** 0.248*** 0.037***
Hybrid REITs 0.124*** 0.912*** — 0.089 0.213* 0.008 522.641* 0.061** 0.059** 0.022**
Panel D: EGARCH Model with Lag
Equity REITs — 0.046 0.023 0.018 0.056 0.003 89.194 0.091*** 0.027*** 0.164
Mortgage REITs 1.321*** — 0.708*** 0.154 0.113*** 0.008 28.217 0.175** 0.016* 0.035***
Hybrid REITs 0.530 0.027 — 0.055 1.099*** 0.009 221.548 0.068*** 0.045*** 0.021**
Notes: Exhibit 6 reports coefficient for the exogenous conditional variance variable.
* Significant at the 10% level.
** Significant at the 5% level.
*** Significant at the 1% level.
Volatility Spillover in REIT Returns
negative. The second issue relates to the strong fixed income sector and Mortgage REITs. This has
similarities between the Mortgage results and a number of implications for fund managers who
those reported for the Equity and Hybrid sectors, do differentiate between the different REIT sec-
particularly in relation to the impact of small and tors. It would appear that at least in terms of vol-
value stocks. On a number of occasions, using both atility the markets do not differentiate between
the GARCH and EGARCH specifications, signifi- the REIT sub-sectors and view the entire sector as
cant spillovers are reported from these sectors into a single market. This is further supported by the
Mortgage REITs. The combined findings from the within REIT sector results, with evidence of Eq-
equity and bond markets would indicate that cer- uity REITs heavily influencing the Mortgage and
tainly in terms of volatility, Mortgage REITs gen- Hybrid sectors.
erally behave in line with the rest of the REIT
sector, with no discernable pattern that would in-
dicate differing reactions to volatility in the mar- Endnotes
kets. This could be due to two factors. First, that
1. See also studies such as Gordon and Canter (1999) and
due to the characteristics of the broader mortgage Eichholtz (1996), which have examined the relative stability
market and the importance of factors such as pre- of the correlation coefficients between real estate securities
payments and the growth in the non-pass through and common stocks and the stability of the covariances be-
tween international property stock markets.
market, Mortgage REITs may not behave directly
2. Further studies to have examined various aspects of this
in line with conventional fixed income securities. issue include Gyourko and Keim (1992), Liu and Mei (1992),
The second factor may relate to the fact that the Pagliari and Webb (1995), Quan and Titman (1999) and Wil-
wider capital markets perhaps do not sufficiently son, Okunev and Webb (1998).
differentiate between the different REIT sectors 3. See also studies such as Theodossiou and Lee (1993) and
Kanas (1998).
in terms of the key differences in their asset base
4. For additional information on the methodological issues that
and the impact of wider price and volatility
arise with ARCH-based modeling see studies such as Bol-
movements. lerslev (1990), Bollerslev, Chou and Kroner (1992), Engle
(1995), Engle and Bollerslev (1986), Engle, Ito and Lin
(1990, 1994) and Nelson (1990).
5. The detailed full results for these GARCH models together
Conclusion with the accompanying diagnostic tests are available from
the author on request.
This study has examined influences on volatility in
REITs. The results reveal that volatility in Equity
REITs has a significant influence on the other sub-
sectors of the market and that a number of pat- References
terns emerge with regard to the influence of other Bae, K. and G. A. Karolyi, Good News, Bad News and Inter-
asset classes. The primary results indicate that the national Spillovers of Stock Return Volatility Between Japan
and the U.S., Pacific-Basin Finance Journal, 1994, 2, 405–38.
REIT sector is generally influenced more strongly
Barkham, R. and D. Geltner, Price Discovery in American and
by volatility in small cap stocks and in firms clas- British Property Markets, Real Estate Economics, 1995, 23,
sified as value stocks. These findings are not sur- 21–44.
prising given the average size of REITs and the Bekeart, G. and C. Harvey, Emerging Equity Market Volatility,
nature of them and their asset base. The overall Journal of Financial and Quantitative Analysis, 1997, 25, 203–
15.
S&P 500 has a mixed and inconsistent relationship
Bollerslev, T., Generalized Autoregressive Conditional Heter-
with REITs. A comparison between the results ob- oscedasticity, Journal of Econometrics, 1986, 31, 307–28.
tained through the GARCH and EGARCH specifi- Bollerslev, T., Modeling the Coherence in Short-Run Nominal
cations reveal a number of significant differences, Exchange Rates: A Multivariate Generalized ARCH Model, Re-
with the EGARCH findings generally more justi- view of Economics and Statistics, 1990, 72, 498–505.
fiable in terms of intuition. The results illustrate Bollerslev, T., R. Y. Chou and K. F. Kroner, ARCH Modeling in
that Mortgage REITs are far more influenced in Finance: A Review of the Theory and Empirical Evidence,
Journal of Econometrics, 1992, 52, 5–59.
terms of volatility by the stock market rather than
Booth, G. G., T. Martikainen and Y. Tse, Price and Volatility
the fixed income sector. No evidence is reported of Spillovers in Scandinavian Stock Markets, Journal of Banking
a positive relationship in volatility between the & Finance, 1997, 21, 811–23.
Journal of Real Estate Portfolio Management 237
Simon Stevenson
De Santis, G. and J. M. Wooldridge, Stock Returns and Vola- Liu, C. H., D. J. Hartzell, W. Greig and T. Grissom, The Inte-
tility in Emerging Financial Markets, Journal of International gration of the Real Estate Market and the Stock Market: Some
Money & Finance, 1997, 16, 561–79. Preliminary Evidence, Journal of Real Estate Finance & Eco-
Eichholtz, P. M. A., The Stability of the Covariances of Inter- nomics, 1990, 3, 261–82.
national Property Share Returns, Journal of Real Estate Re- Liu, C. H. and J. Mei, The Predictability of Returns on Equity
search, 1996, 11, 149–58. REITs and their Co-Movement with Other Assets, Journal of
Engle, R. F., Autoregressive Conditional Heteroscedasticity Real Estate Finance and Economics, 1992, 5, 401–18.
with Estimates of the Variance of U.K. Inflation, Econometrica, Mei, J. and A. Lee, Is there a Real Estate Risk Premium?,
1982, 50, 987–1008. Journal of Real Estate Finance and Economics, 1994, 9, 113–
——., (Ed.), ARCH: Selected Readings, Oxford: Oxford Univer- 26.
sity Press, 1995.
Meyer, F. C. N. and J. R. Webb, Statistical Properties of Re-
Engle, R .F. and T. Bollerslev, Modeling the Persistence of Con- turns: Financial Assets versus Commercial Real Estate, Jour-
ditional Variances, Econometric Review, 1986, 5, 1–50. nal of Real Estate Finance and Economics, 1994, 8, 267–82.
Engle, R. F., T. Ito and W. L. Lin, Meteor Showers or Heat Nelson, D. B., Stationarity and Persistence in the GARCH (1,1)
Waves? Heteroscedastic Intra-Day Volatility in the Foreign Model, Econometric Theory, 1990, 6, 318–34.
Exchange Market, Econometrica, 1990, 58, 525–42.
——., Conditional Heteroscedasticity in Asset Returns: A New
——., Do Bulls and Bears Move Across Borders? International
Approach, Econometrics, 1991, 2, 347–70.
Transmissions of Stock Price Volatility, Review of Financial
Studies, 1994, 7, 507–38. Ng., V. K., P. R. Chang and R. Y. Chou, An Examination of the
Engle, R. F. and V. K. Ng, Measuring and Testing the Impact Behavior of Pacific-Basin Stock Market Volatility, In R. Ghon
of News on Volatility, Journal of Finance, 1993, 48, 1749–78. and R. P. Chang, (Eds.), Pacific-Basin Capital Markets Re-
search, Elsevier, 1991.
Gordon, J. N. and T. A. Canter, International Real Estate Se-
curities: A Test of Capital Markets Integration, Journal of Real Pagliari, J. L. and J. R. Webb, A Fundamental Examination of
Estate Portfolio Management, 1999, 5, 161–70. Securitized and Unsecuritized Real Estate, Journal of Real Es-
Gyourko, J. and D. Keim, What Does the Stock Market Tell us tate Research, 1995, 10, 381–426.
About Real Estate Returns?, Journal of the American Real Es- Quan, D. C. and S. Titman, Do Real Estate Prices and Stock
tate and Urban Economics Association, 1992, 20, 457–85. Prices Move Together? An International Analysis, Real Estate
Hamao, Y., R. W. Masulis V. Ng, Correlations in Price Changes Economics, 1999, 27, 183–208.
and Volatility Across International Stock Markets, Review of Seiler, M. J., J. R. Webb and F. C. N. Myer, Are EREITs Real
Financial Studies, 1990, 3, 281–307. Estate?, Journal of Real Estate Portfolio Management, 1999, 5,
He, L. T., Cointegration and Price Discovery Between Equity 171–82.
and Mortgage REITs, Journal of Real Estate Research, 1998, Stevenson, S., Evaluating the Investment Attributes & Per-
16, 327–38. formance of Property Companies, Journal of Property Invest-
Kanas, A., Volatility Spillovers Across Equity Markets: Euro- ment & Finance, 2001, 19, 251–66.
pean Evidence, Applied Financial Economics, 1998, 8, 245–56.
Theodossiou, P. and U. Lee, Mean and Volatility Spillovers
Karolyi, G. A., A Multivariate GARCH Model of International Across Major National Stock Markets: Further Empirical Ev-
Transmissions of Stock Returns and Volatility: The case of the idence, Journal of Financial Research, 1993, 16, 337–50.
United States and Canada, Journal of Business and Economic
Statistics, 1995, 13, 11–25 Wang, K., J. Erickson and S. H. Chan, Does the REIT Stock
Market Resemble the General Stock Market?, Journal of Real
Koutmos, G. and G. G. Booth, Asymmetric Volatility Trans-
Estate Research, 1995, 10, 445–60.
mission in International Stock Markets, Journal of Interna-
tional Money & Finance, 1995, 14, 747–62. Wilson, P., J. Okunev and G. Ta, Are Real Estate & Security
Li, Y. and K. Wang, The Predictability of REIT Returns and Markets Integrated? Some Australian Evidence, Journal of
Market Segmentation, Journal of Real Estate Research, 1995, Property Valuation & Investment, 1996, 14, 7–24.
10, 471–82. Wilson, P., J. Okunev and J. R. Webb, Step Interventions and
Ling, D. and A. Naranjo, The Integration of Commercial Real Market Integration: Tests in the U.S., U.K. and Australian
Estate Markets and Stock Markets, Real Estate Economics, Property Markets, Journal of Real Estate Finance and Eco-
1999, 27, 483–515. nomics, 1998, 16, 91–123.
238 Vol. 8, No. 3, 2002
Related docs
