Drivers of U.S.

REIT Returns and its Volatility:

A Bayesian Model Averaging Approach
Tjeerd M. Boonman∗, Vivek Sah†, Tom G. Geurts‡§
February 6, 2025

Abstract
Obtaining insight in the determinants of investment returns and their volatility has
a long tradition in asset pricing modeling. For Real Estate Investment Trusts (RE-
ITs) the results have been mixed, with no clear consensus emerging. We apply the
Bayesian Model Averaging method to let the data select the most robust indicators.
Using a comprehensive sample of U.S. office, industrial, and healthcare REITs, we
identify the most robust drivers of their excess returns and volatility for the period
2001-2024. We find that returns are mostly driven by domestic financial markets
indicators, while the volatility of the returns are driven mainly by REIT-specific in-
dicators. By analyzing three subperiods, we find that the Global Financial Crisis
(GFC) had a disruptive impact on both the returns and volatilities, as the number of
robust variables in the post-GFC period is roughly half of the pre-GFC period, sug-
gesting a shakeout due to the economic shock, leading to more consolidation within
the sector. In the COVID period, the drivers show resilience as these resemble more
to the pre-GFC period, suggesting that the GFC has had a more disruptive impact
than the COVID crisis.

Keywords: REITs, Bayesian Model Averaging, Excess returns, Volatility of returns,

Global Financial Crisis, COVID
JEL-code: G01, F31, E47, C11, C23.

∗
Leon Hess Business School, Monmouth University, West Long Branch, NJ, [email protected].
†
Daniels College of Business, University of Denver, Denver, CO, [email protected]
‡
Freeman College of Management, Bucknell University, Lewisburg, PA, [email protected]
§
Technical University of Berlin, Germany

1
1 Introduction

Real estate is an important part of the economy and is considered an attractive investment
alternative for diversification purposes within a portfolio containing other investments.
Disadvantages of direct real estate investments are its low liquidity and transparency, as
well as the fact that one real estate investment requires a large capital outlay, which
hampers creating an optimal portfolio. This is partly offset by Real Estate Investment
Trusts (REITs), that offer liquid shares in an investment vehicle that invests in a portfolio
of real estate assets1 . REITs are a hybrid form of investments, with characteristics from
both shares and bonds, and have the risk-return profile of small-cap stocks (Ross and
Zisler, 1991). Given its public character, information about market prices and indicators,
availability of financial statements, and more importantly, information transparency on its
invested real estate assets, makes it feasible to use data-driven estimation techniques.
REITs are predominantly owned by institutions (Chung et al. 2012), and providing
more insights on the drivers of their returns can lead to a better understanding of the
mechanisms, and useful to predict returns, as well as better asset allocation for such big
investors. For the purpose of this study, returns are calculated as the excess returns over
the T-bill rate, as is typical in these type of studies. Furthermore, it is also important
to understand the volatility of REIT returns, which has received less attention in the
literature and indeed no clear consensus has emerged on the determinants for REIT returns
and volatility of these returns. Typically, in most previous research a pre-selection of
indicators is based on meta studies, stylized facts, or theoretical models. In our study, we
let the data lead and decide the primary drivers. We use the Bayesian Model Averaging
1
REIT definition: In order to qualify as a REIT, a company must comply with certain provisions within
the U.S. Internal Revenue Code: A REIT must be an entity that is taxable as a corporation; be managed
by a board of directors or trustees; have shares that are fully transferable; have no more than 50% of its
shares held by five or fewer individuals during the last half of the taxable year; invest at least 75% of its
total assets in qualifying real estate assets; derive at least 75% of its gross income from real estate related
services; have no more than 25% of its assets consist of stock in taxable REIT subsidiaries; pay annually
at least 90% of its taxable income in the form of shareholder dividends (taken from Ghysels et al., 2013).

1
(BMA) method, since it avoids the all-or-nothing mentality that is associated with classical
hypothesis testing, in which a model is either accepted or rejected wholesale. Additionally,
BMA is robust for adding or eliminating observations, and the method is relatively robust
to model misspecification (Hinne et al., 2020). BMA is found to better estimate the true
volatility series than does the traditional maximum likelihood approach (Ward, 2008).
BMA outperforms other methods2 in simulations and out-of-sample tests (Porwal and
Raftery, 2022).
Identifying the drivers of REIT returns is part of the question on return predictability.
For practitioners, predictable asset returns can affect the asset allocation and hedging
decisions of investors and the timing of security issuance by firms. For academics, the
existence of predictable returns has implications for market efficiency. Although traditional
formulation of the efficient market hypothesis precludes predictable asset prices, most
financial economists agree that this view of market efficiency is overly restrictive. Advances
in asset pricing theory and computing power have more recently provided an abundance of
empirical evidence that stock prices are predictable, to some extent, using publicly available
information. Two explanations are offered for this return predictability. On the one hand,
predictability results from business cycle movements and changes in investors’ perceptions
of risk that are reflected in time-varying risk premiums. On the other hand, predictability
reflects an inefficient market populated with overreacting and irrational investors, that is,
a market full of “animal spirits,” as characterized by Kaul (1996).
There is mixed evidence of drivers of REIT returns and volatility in the empirical lit-
erature. In general the REIT literature has followed the non-REIT asset pricing literature
to similarly motivate the variables. Asset returns are modeled in CAPM, adjusted CAPM
(including the Fama French three factor model), and the Arbitrage Pricing Theory (APT).
The APT, created by Ross (1976), is built on the premise that asset returns can be deter-
2
BMA outperforms penalized likelihood methods including LASSO, smoothly clipped absolute devia-
tion, minimax concave penalty, and elastic net methods.

2
mined by macroeconomic and firm-specific factors. Since the seminal paper of Ross (1976),
numerous studies have been conducted to determine the performance of the share price
using firm fundamentals, macroeconomic information and market conditions (Arora et al.,
2019).
Chui et al. (2003) find that REIT returns in the pre-1990 period are impacted by
momentum from past returns, size, turnover, and analyst coverage, while after 1990, mo-
mentum is the dominant predictor of REIT returns. Hung and Glascock (2010) examine
REIT returns over the 1993–2009 period and find that momentum, book-to-market ratio,
institutional ownership, and illiquidity are highly correlated with REIT returns. Goebel et
al. (2013) explores the drivers of REIT returns in the characteristic framework developed
by Daniel and Titman (1997, 1998), while accounting for the well-established momentum
effect. They find size and analyst coverage are not correlated with REIT returns (Goebel
et al., 2013). Bond and Xue (2017) find that two fundamental factors enhance predictive
power in explaining REIT returns, namely investment and profitability.
There is another set of studies that focus on macroeconomic factors as drivers for REIT
returns. Naranjo and Ling (1997) find that growth rate in real per capita consumption, the
real T-bill rate, the term structure of interest rates, and unexpected inflation are fundamen-
tal drivers of direct non-securitized real estate returns. Payne (2003) uses the generalized
impulse response analysis to identify the response of their excess returns to unexpected
changes in the broader stock market, real output growth, inflation, term structure of inter-
est rates, default risk, and the federal funds rate. For equity REITs, they find that excess
returns respond positively to an unexpected shock to the broad stock market index, while
the initial impact of an unexpected shock to the term structure is negative. Li and Lei
(2011) find that REIT total returns are positively related to the next quarter GDP growth
rate, and negatively related to changes in term spread and T-bill rates. In a more recent
study by Akinsomi et al. (2016), the authors use several models, including the Bayesian

3
Model Averaging (BMA), to forecast returns of the U.S. NAREIT index. They find that
good REIT return predictors vary over time and forecast horizons. The three most domi-
nant predictors they find are economy-wide indicators, monetary policy instruments, and
market sentiment related.
In contrast to the literature that has examined the return behavior of REITs, few have
examined volatility of these returns. Cotter and Stevenson (2007) use a GARCH model
and daily data from 1992 to 2005, to model the volatility of the returns on SNL U.S.
REIT Indices. They find that the volatility is influenced by other U.S. equity series such
as large caps and value stock. Fei et al. (2010) use a multivariate asymmetric dynamic
diagonal conditional correlation GARCH model, and find that the time-varying correlations
and volatility of REITs, stocks and direct real estate can be explained by macroeconomic
variables, such as term and credit spreads, inflation, and the unemployment rate. Huerta
et al. (2016) also employ GARCH models to examine the impact of the 2007–2009 liquidity
crisis and investor sentiment on REIT returns and volatility. They find that the liquidity
crisis led to higher volatility, while positive (negative) changes in investor sentiment will
affect REIT returns’ volatility negatively (positively). Delisle et al. (2013) examine the
pricing of volatility risk in the cross-section of daily REIT returns from 1996 to 2010, for
all equity REITs that are publicly traded on the three major exchanges (NYSE, AMEX,
and NASDAQ). They find that REIT volatility is not sensitive to the aggregate systematic
volatility (measured by VIX), but has a significant and negative firm-specific (idiosyncratic)
volatility, as measured by the Fama-French three factor model. They also consider returns
skewness, firm size, book-to-market, momentum, institutional ownership, and liquidity.
This study combines a large number of domestic and REIT specific factors to deter-
mine the most robust drivers of REIT excess returns, as well as its volatility. Using a
sample of total 63 office, industrial, and healthcare REITs over a time period from 2001
to 2024, we find strong dominance of domestic financial variables, namely volatility on

4
stock index, monetary policy uncertainty, stock market returns, and the performance of
income producing real estate over REIT-specific indicators as drivers of REIT excess re-
turns. Representing the overall industry, only the construction variable is robust. Unlike
previous studies, we do not find evidence that domestic macroeconomic variables play a
decisive role in returns. This can be attributed to the use of exhaustive list of other pre-
dictors, which override the power of domestic factors. Our sub-sample analysis of small
and large REITs, shows similar results but more robust drivers for the larger REITs. In
addition, we find that returns of large REITs are driven by stock turnover, volatility of
stock price and analyst rating suggesting more integration for these entities with the stock
market than for smaller REITs. For our sub-time period analysis, the results show that the
number of robust variables drops from a high of 12 pre-GFC to just 7 post-GFC suggesting
a shakeout of the economic shock leading to more consolidation within the sector, as well as
more fundamental driven investing by REITs. Key robust variables funds from operations
(FFO), size, and stock turnover become significant drivers post-GFC, while none of the
construction-sector variables are significant. For our analysis of the COVID period, we
find that the number of significant drivers increases to 10, and returns are mainly driven
by domestic variables.
Our results on the determinants of the volatility of REIT returns reveal that REIT-
specific variables dominate. These are stock turnover, size, short-term debt to total debt,
expected analyst rating and operating margin. For the sub-sample of small and large RE-
ITs, the result is asymmetrical. While both are influenced by REIT specific factors, for
small REITs, the financial ratios namely leverage and profitability matter more. When
we look at the sample broken down by various time-periods, we find that the three peri-
ods analyzed in our study have distinctly different drivers. Pre-GFC volatility is driven by
REIT-specific factors based on financial statement ratios, financial markets data, and loca-
tion dummies while the number of total drivers of volatility are reduced significantly from

5
11 to 4, when analyzing the post-GFC time period. Lastly, during the COVID time-frame,
volatility is driven by REIT-specific and domestic indicators in the same proportion.
We contribute in four ways to the existing literature. First, we provide more insight in
the scant literature on identifying determinants of the volatility of REIT returns. The ex-
isting studies typically relate the volatility on REIT index returns to other asset categories,
such as small cap stocks, value stocks and real estate stocks. The dynamics are usually
captured in time series models, in particular GARCH models, using daily or monthly fre-
quency returns of REIT indices. We base the volatility on the returns of a large number of
individual REITs, and we use a large number of potential underlying variables to explain
the volatility. In this respect, our work is closely related to Delisle et al. (2013). However,
we use quarterly data so that we can include a battery of underlying domestic and REIT-
specific data that is only available at a lower (quarterly) frequency. We also use a different
methodology which enables us to include a large number of potential determinants.
Our second contribution is that we apply the BMA approach to a large sample of
individual REITs, as opposed to Akinsomi et al. (2016), who use the BMA approach to
the REIT index. REITs are very heterogeneous entities with focused property types, and
operate on high concentrated geographical niche areas. What drives their returns could
potentially vary depending upon the property type and other idiosyncratic factors. As an
example, while most REITs underperformed in the aftermath of the Great Financial Crisis
of 2008-2009, apartment REITs experienced one of the best time periods with the lowest
vacancy rates in their history of operations. To have a clearer interpretation of the results,
we focus on three distinct property types, namely office, industrial, and healthcare REITs
and do not include retail and residential REITs. Ex-ante, we expect the two categories
(office, industrial, and healthcare versus retail and residential) to have different drivers.
The latter category is directly related to individual/retail consumption as compared to
office, industrial, and healthcare, which are consumed/tenanted by corporate entities and

6
are therefore more institutional. Consequently, these are more likely affected by multiple
determinants, which are not known, and therefore require the analysis performed in this
paper. This helps in covering the sector types as listed by the North American Industry
Classification System (NAICS), which is the standard used by Federal statistical agencies in
classifying business establishments for the purpose of collecting, analyzing, and publishing
statistical data related to the U.S. business economy.
Our third contribution lies in the fact that we do not only include a larger number of
individual REITs, but we also employ a larger number of potential determinants. Using
the BMA approach, we are able to add a much larger set of determinants to test in our
model. We base the pre-selection of indicators on the surveys of Ghysels et al. (2013)
and Letdin et al. (2019), as well as Ling et al. (2000), Liu et al. (2012), Akinsomi et al.
(2016), Giacomini, et al. (2017), Alcock and Steiner (2018) and Arora et al. (2019). Our
variables are divided into two broad categories, one REIT-specific, and the other domestic
macroeconomic and financial factors. In total we have 34 variables to test, making it a
very comprehensive set of factors. By contrast, Akinsomi et al. (2016) use just 13 in the
their study and therefore this paper studies 21 additional variables.
Finally, the dimension of the time frame of the data, which encompasses two boom to
bust economic cycles and two major disruptive events (GFC and COVID) also contributes
to the literature. The first boom-bust cycle runs from 2001 up to the Global Financial Crisis
of 2007-2009, and the second cycle from the depth of the ensuing recession to the end of the
largest economic expansion in U.S. history, just ending at the beginning of the pandemic in
2020. This also provides the opportunity to see whether the drivers of REIT returns and
volatility have changed over time especially especially after the economic shock of 2008. In
addition, our study is the first one to analyze how post-COVID the determinants of REIT
returns and volatility have changed. Other studies looked at the impact of COVID on
returns during the pandemic, for example Akinsomi (2021), however, with the passing of

7
more time there are enough data to do this additional analysis. This post-COVID analysis
is a crucial time for our sample as it is unprecedented and a black swan event in our financial
history and the findings that we present for that particular time frame contributes to the
literature. Indeed, the COVID time-period forced certain structural changes in the real
estate space markets, thereby changing the narrative on the property types that we study.
Analyzing the impact of these changes on drivers of REIT returns provides a comparison
with the time-period preceding this mammoth event in our financial markets.
The paper is organized as follows. In Section 2, we discuss the methodology and model
specification, followed by a discussion on the data and variables in Section 3. In Section 4,
we discuss the results followed by conclusion in Section 5.

2 Methodology

Following Goebel et al. (2013), we use the Daniel and Titman (1997, 1998) framework
for explaining cross-sectional REIT return patterns. In this framework, one focuses on
the underlying characteristics of the stocks (e.g., low book-to-market versus high book-to-
market ratios), rather than the return covariance with a risk factor, such as common in
for example Fama and French (1993) factor models. One of the criticisms is that these
factor models are empirically motivated and have no theoretical linkage to the underlying
fundamental drivers of returns.
Our model specification is described in Section 2.2, and distinguishes among REIT-
specific indicators, and domestic macroeconomic and financial indicators. To identify the
most robust determinants of REIT returns, we use the Bayesian Model Averaging (BMA),
which is described in Section 2.3.

8
2.1 Dependent variables: Excess returns and Volatility

For the quarterly excess return we use the arithmetic return that consist of capital gains
and cash dividend yield, and subtract the 3 month T-bill rate.

Pi,t + Di,t − Pi,t−1 T Bi,t

Ri,t = − , (1)
Pi,t−1 4

with Pi,t the price of the i-th REIT at the end of period t, Di,t the cash dividend from the
i-th REIT during period t, T Bi,t is annualized 3 month T-bill rate.
For the volatility of the returns, we use a GARCH(1,1) model, in order to account for
heteroskedasticity and autocorrelation that is present in returns of time series of financial
assets. Using quarterly data implies a low number of observations, which does not fit well
with the GARCH model. For this reason, we use weekly returns.

ht = ω + αe2t−1 + βht−1 , (2)

where et is the difference between the observed variable, yt , and its conditional expected
mean. α > 0, β > 0, ω > 0, α + β < 1, and

ω
, (3)
1−α−β

is the unconditional variance of yt .

In the literature α is often interpreted as the short run persistence (value positive and
close to zero; lower α means more stable time series, and higher α means more jumpy or
nervous market) and β denotes the long run persistence (β positive and close to 1; when
α is higher, then β is lower).
We employ a GARCH(1,1) model to construct the volatility measure, based on weekly
returns over the entire time horizon. Using weekly observations guarantees a large enough

9
number of observations to use the GARCH model. Next, we calculate the average weekly
volatility in the quarter, and convert this into a quarterly one by multiplying the average
by the squared root of 13.

2.2 Model specification

Ghyssels et al. (2013) mention that the extensive predictability literature in finance and
real estate considers variations of the following linear predictive regression:

Rt+1 = α + βXt + t+1 , (4)

where Rt+1 is a return (or price change) and Xt is a vector of variables, observable at time t.
However, linear models are deceptively simple. Statistical complications arise because the
predictor Xt is often persistent and its innovations are correlated with t+1 , which induces
bias in the estimation of β. However, at a lower frequency (quarterly), this persistence and
correlation with error terms is less an issue than at the daily or weekly frequency.
In their survey, Ghyssels et al. (2013) distinguish between the use of serial dependence,
valuation ratios, and economic ratios to predict returns in real estate. Regarding serial
dependence, we should test for serial correlation in returns and weak-form market effi-
ciency. Several studies in the real estate literature find that returns exhibit positive serial
correlation, including Gau (1984), Guntermann and Smith (1987), Case and Shiller (1989),
Gyourko and Voith (1992), Hill et al. (1997, 1999), Gu (2002), and Schindler (2013). How-
ever, the evidence on whether this predictability can be exploited for financial gains (i.e. to
predict returns) is less clear. The empirical literature has used AR(1), ARIMA, GARCH,
and EGARCH models.
The second group of indicators, valuation ratios (dividend-price, book-to-market, earnings-
price, etc.) have consistently found to be reliable predictors of equity returns (see Rapach

10
and Zhou, 2013). Analogous ratios have been used in the real estate literature, some of
which are the rent-price ratio (Hamilton and Schwab, 1985; Geltner and Mei, 1995; Camp-
bell et al., 2009; Himmelberg et al., 2005; Gallin, 2008; Plazzi et al., 2010), the loan-to-value
ratio (Lamont and Stein, 1999), and the price-to-income ratio (Malpezzi, 1999).
The third group consists of economic variables. Ghyssels et al. (2013) mention that
there is considerable evidence that economic variables are associated with future appreci-
ations in property values, as shown by Rosen (1984), Case and Shiller (1990), Abraham
and Hendershott (1996), MacKinnon and Zaman (2009), and Plazzi et al. (2010). The
empirical framework in most of these papers is the predictive regression model (4) with
conditioning information Xt that includes demographic variables (population growth, per-
centage of population within a certain age), income and employment variables, construction
costs, housing starts, tax rates, zoning restrictions, and other regulatory variables. The
selection of the conditioning information is dictated by the data, the level of aggregation,
and the methodology.
To identify the determinants or drivers of the REIT returns and the volatility of these
returns, we use the Bayesian Model Averaging method, which is described next.

2.3 Bayesian Model Averaging

Bayesian Model Averaging (BMA) is an empirical tool to deal with model uncertainty in
various fields of applied science. Typically, BMA focuses on which regressors to include
in the analysis, by determining the robustness of the explanatory variables. By averaging
across a large set of models one can determine the variables that are relevant. Each model,
thus each possible combination of explanatory variables, is regressed on the dependent
variable. Each model receives a weight and the final estimates are constructed as a weighted
average of the parameter estimates from each of the models. BMA includes all of the

11
variables within the analysis, but shrinks the impact of certain variables towards zero
through the model weights.
Using the normal linear regression model, as shown in equation (4), model uncertainty
comes from the selection of q regressors (q = K + M ). There are a total of 2q candidate
models to be estimated, indexed by Mj for j = 1, . . . , 2q , which all seek to explain y.
The posterior for the parameters β j calculated using Mj is written as:

j f (y|β j , Mj )g(β j |Mj )

g(β |y, Mj ) = (5)
f (y|Mj )

Given a prior model probability P(Mj ), we can calculate the posterior model probability
using Bayes Rule as:
f (y|Mj )P (Mj )
P (Mj |y) = (6)
f (y)

Following Leamer (1985), we can consider β a function of β j for each j = 1, ... ,

2q and then calculate the posterior density of the parameters for all the models under
consideration by the law of total probability:

2 q
X
g(β|y) = P (Mj |y)g(β|y, Mj ) (7)
j=1

Therefore, the full posterior distribution of β is a weighted average of its posterior

distributions under each of the models, where the weights are given by P (Mj |y).
The expected coefficient for β, the posterior mean, is calculated as:

2 q
X
E(β|y) = P (Mj |y)E(β|y, Mj ) (8)
j=1

12
 and the posterior variance as:

2q 2q
X X
V (β|y) = P (Mj |y)V (β|y, Mj ) + P (Mj |y)[E(β|y, Mj ) − E(β|y)]2 (9)
j=1 j=1

Bayesian model averaging requires prior distributions for the unknown parameters un-
der the various models M j for j = 1, ... , 2q . We follow the standard choice in the BMA
literature, by putting a conditionally normal prior on coefficients βj . We use a conditional
prior for the j-th model’s parameters (β j |σ 2 ) with zero mean and the variance proposed
by Zellner (1986), a prior covariance given by g(Xj0 Xj )−1 . This prior covariance is pro-
portional to the posterior covariance arising from the sample (Xj0 Xj )−1 with the scalar g
determining how much importance is attributed to the prior beliefs of the researcher. Many
options for choosing g have been proposed in the literature. We employ the widely used
Unit Information Prior, proposed by Kass and Wasserman (1995), which corresponds to
taking g = N , and leading to Bayes factors that behave like the BIC (Moral-Benito, 2013).
Additionally, we have to make assumptions about the model space, that is, which type of
models are a priori more likely. Following the standard in the literature, the assumption
of uniform model priors, so that, ex ante, each model is presumed equally likely (Eicher et
al., 2012).
To determine whether a regressor is robust, we calculate the Posterior Inclusion Proba-
bility (PIP), which is the sum of the posterior model probabilities for all models including
that variable. In other words, the PIP is calculated as the proportion of models that con-
tain the corresponding explanatory variable. The PIP is a useful diagnostic for deciding
whether an individual explanatory variable has an important role (Desbordes et al., 2018).
We use the rule of thumb proposed by Kass and Raftery (1995), and described by Moral-
Benito and Roehn (2016): The evidence of a regressor having an effect is weak, positive,

13
strong, or decisive if the posterior inclusion probabilities lie between 50–75%, 75–95%,
95–99% or are greater than 99%, respectively.
Since enumerating all potential variable combinations becomes infeasible quickly for a
large number of covariates (Amini and Parmeter (2011) consider large as more than 14),
we use the Markov Chain Monte Carlo (MCMC) algorithm. All of the computations are
performed using the R package BMS from Amini and Parmeter (2011).

3 Data

We focus on U.S. office, industrial, and healthcare REITs from 2001Q3 to 2024Q1, with
quarterly observations. The REITs in our sample are listed in Appendix B. We avoid
survivorship bias by including also REITs that cease to exist, either because they were
discontinued, went bankrupt, or merged. We exclude REITs with a lifetime of 2 years or
less, which leaves us 63 REITs, of which 24 are listed during the entire sample period, 23
started on a later date, and are listed on 2024Q1, and 16 REITs are no longer listed in
2024Q1.

3.1 Returns and volatility of the returns

Table 1 compares the excess returns on the selected REITs with the excess returns on
two REIT indices and the S&P500 stock market index for the period 2001Q3-20240Q1.
Returns on the selected REITs are on average positive, and are characterized by a relative
large standard deviation. Relative to the stock market index, the returns of U.S. REITs
are more dispersed.

(INSERT TABLE 1 HERE)

14
3.2 Potential determinants of the returns

For the independent indicators, or the determinants of the returns, we created a broad
sample. Based on the literature and stylized facts we pre-selected 18 REIT-specific in-
dicators and 19 domestic macroeconomic and financial factors. Details on definition and
source are presented in Appendix A.

REIT-specific indicators

We distinguish accounting ratios, that are based on financial statements, and market ratios,
that are based on information from capital markets. None of the indicators are highly
correlated. Data availability also plays a role in the definitive selection. All variables have
been inspected for non-stationarity and where necessary, transformed into stationary series.
All indicators enter with a one quarter lag, following Chen et al. (1998) and Giacomini et
al. (2017), for two reasons. First, to exclude possible endogeneity of the ratios based on
capital markets and the return on REITs. Second, to acknowledge the reporting lag, as
information on the quarter is not available until well after the quarter has finished.

• The debt-to-equity ratio represents the financial leverage, and is often included as
determinant since REITs rely heavily on leverage to finance commercial properties.
Modigliani and Miller’s (1958) second proposition posits that increases in financial
leverage directly increase the riskiness of the cash flows to equity holders and thus
raise the required rate of return on equity. However, empirical research is mixed. Gi-
acomini et al. (2017) find that highly levered REITs have both lower average returns
and higher return variances. Arora et al. (2019) also find a negative relationship.

15
 According to Ooi and Liow (2004), unadjusted returns tend to be positively related
to debt-to-equity ratio, while the evidence for risk-adjusted returns is mixed. The
expected sign for volatility in REIT returns is positive.

• Growth of the assets. Bond and Xue (2017) use the annual growth rate in total
investments in non-cash assets, and find that REITs with the higher investment
rates underperform relative to REITs with the lower investment rates. The expected
sign for returns is negative.

• Return on Assets (ROA) and Return on Equity (ROE) as proxies for profitability.
An increase in the profitability is likely to result in positive share returns. Bond and
Xue (2017) find that more profitable REITs (higher ROA and ROE) outperform less
profitable REITs, hence we expect a positive sign for returns.

• Operating margin, as an alternative proxy for profitability. A higher profitability is

expected to have a positive impact on the return.

• Cash Holdings to Total Assets. Higher cash holdings have a positive effect on the
value of the REIT, as financial flexibility increases with more liquidity, which is valued
by investors (Hill et al., 2012).

• Short-term debt to Total debt is a measure for liquidity. Short-term debt has a
lower cost than long-term debt, but is riskier: when interest rates increase, rolling
over becomes much more expensive. Adams et al. (2015) argues that companies
that have the majority of their debt denominated as short-term debt are generally
in greater need of cash and hence more sensitive to losses and market distress. We
therefore expect a negative sign for the returns.

16
 • Asset turnover, defined as the revenues to assets ratio. A higher turnover implies
a more efficient use of the assets, that is expected to have a positive effect on the
return.

• Capex to total assets. REITs that invest more, are expected to have a higher return.

• Funds from Operations (FFO). Graham and Knight (2000) established that this
variable provides the best information with respect to earnings with an expected
positive sign. See also the discussion in Chacon et al. (2021).

• Free Cash Flow to Enterprise Value. A higher ratio implies more financial flexibility,
which is associated with higher returns.

• Property type dummies. The REIT returns and their volatilities depend partially on
the property types, which became very clear during the COVID and post-COVID era,
when offices were used less, while interest for healthcare increased. We distinguish
three non-residential types (number of REITs in parenthesis): office (28), healthcare
(17), and industrial (18).

• Location dummies. Location of the real estate objects plays an important role for
the returns of REITs. We use headquarter location as a proxy for geographic con-
centration, according to the classification used by REITs when filing with the SEC.
Accordingly, we group the REITs into one of four regions (with the number of RE-
ITs in parenthesis): East (28), West (15), South (13), and Midwest (7). Ling et
al. (2022) measure the level of diversification of a REIT portfolio by calculating the
weighted geographic distance of each property to the location of the headquarter of
the REIT. They find that REITs that are concentrated in their headquarters states,
need to underprice more in order to attract a larger pool of investors.

REIT-specific indicators that are related to the market are presented next.

17
 • Size, in terms of market capitalization. Since the seminal Fama and French three
factor model, size is considered a substantial contributor to returns. Ooi and Liow
(2004) confirm that small real estate stocks outperform large real estate stocks in
terms of returns, and we expect that the same holds for REITs, so the expected sign
is negative.

• Price to Book value. From an asset pricing perspective, in particular the three factor
model of Fama and French, the premium attached to value stocks indicates that
investors require higher returns from stocks with low Price to Book values (Ooi and
Liow, 2004). Therefore a negative sign is expected.

• Stock turnover. This ratio is a proxy for stock liquidity. Yung and Nafar (2017) use
stock turnover or the trading volume as proxy for stock’s visibility.

• Volatility stock price (30 days standard deviation of the stock returns). A higher
volatility in the returns would be accompanied by a higher return when markets are
efficient. Akinsomi et al. (2016) includes a 12 months moving standard deviation.

• Expected Analyst ratings. A higher rating implies that the analyst recommendation
are more positive, which would lead to a higher return.

All indicators are retrieved from Bloomberg. Table 2 presents descriptive statistics of
these independent indicators.

(INSERT TABLE 2 HERE)

18
Domestic indicators

To examine the time-varying returns of U.S. REITs, both macroeconomic and financial
factors are incorporated in the regression model. We split these indicators in three sub-
categories, distinguishing (i) macro-economic, (ii) financial market indicators, and (iii)
construction and real estate sector.
We include the following (seven) macro-economic indicators.

• Real GDP growth is informative about general economic conditions and their impact
on real estate activity. It is expected to have a positive influence on REIT returns
(Arora et al., 2019; Liu et al., 2012).

• Inflation and inflation expectations. According to Arora et al. (2019) and others,
inflation has a negative impact on returns, because revenues are not adjusted imme-
diately, and higher inflation is likely to cause an increase in nominal interest rates,
which affects the financing, particularly the high leveraged REITs. On the other
hand, Liu et al. (2012) and others note that real estate provides a hedge against in-
flation, and thus inflation is expected to be positively correlated with REIT returns.

• Unemployment. Demand for real estate is likely to be lower with higher unemploy-
ment levels (Liu et al, 2012).

• Wage growth. The operation of REITs does not require a high proportion of labor,
so wages only marginally affect the profits of the REIT. Wage increases when the
economy is booming would coincide with high REIT returns.

• Consumer Sentiment. Difference of the sentiment compared to the trend (measured

as the moving average of past 12 months of consumer sentiment). An increase in
consumer sentiment would be associated with favorable business conditions and thus
a positive sign is to be expected.

19
 • The percentage change in the monetary base. Money supply M2 is the stock of money
available to be circulated in the economy. We expect a positive relationship between
money supply (M2) and REIT returns (Ling et al., 2000).

Financial variables are forward looking, respond quicker to news and are available at higher
frequencies.

• Real interest rate, calculated as the 10 year treasury bond yield minus consumer
price inflation, both annualized. A higher real interest rate leads to a higher cost of
borrowing, and makes investments in real estate less attractive compared to govern-
ment bonds. Therefore, we expect a downward effect on the REIT returns (Liu et
al., 2012; Ling et al., 2000).

• Time spread, measured as the spread between the 10 year T-bond yield and the 3
months T-bill rate (Payne, 2003; Ling et al., 2000). The time spread affects real
estate supply since profitability of developers and real estate companies depends
on the cost and availability of borrowed funds. We expect the larger the spread,
the larger the supply in the real estate markets, and the lower the returns for real
estate firms (Edelstein et al., 2011). Since most REITs have a capital structure with
significant long-term debt, we expect that increases in the yield curve will result in
higher borrowing costs and have a negative impact on REIT returns (Arora et al.,
2019; Liu et al., 2012).

• Stock market returns have an ambiguous effect on real estate markets. On one hand,
stock market returns reflect perceptions of the health of the economy, which affects
the demand for housing and construction. Equities represent a major component of
household wealth, which determines household attitudes towards future consumption.
On the other hand, as a main competing asset class for fund managers and investors,

20
 equity returns will impact on relative demand for real estate investments (Liu et al.,
2012). We use the quarterly return on the S&P 500 stock index.

• VIX, the implied volatility on stock index, is widely used as a measure for risk
aversion in the market. In episodes of turmoil, risk aversion increases and investors
flee from risky assets to safe havens. When REITs are considered a safe, low risk
investment, then the relation between VIX and REIT returns would be positive,
otherwise negative.

• U.S. Monetary Policy Uncertainty (MPU). This measure has a high value when
uncertainty about monetary policy is high. Monetary policy is important for REITs,
because of its sensitivity for interest rates and lending.

• The NCREIF Property Index (NPI) measures the performance of income producing
real estate, and is the primary index used by institutional investors in the United
States to analyze the performance of commercial real estate and use as a benchmark
for actively managed real estate portfolios. We expect a positive relationship with
the individual REIT returns.

The third and last subcategory contains (six) indicators linked to the construction and
real estate sector. The percentage change in housing construction starts would positively
impact REIT returns (Ling et al., 2000). Secondly, changes in construction spending (total,
office, commercial, health care, education) can have an impact, namely an increase points
towards more attractive returns. On the other hand it could also lead to lower profit
margins as the supply of real estate increases. Since the four segments, office, commercial,
health care, and education, are to a certain point substitutions, the increase in construction
spending in one segment, say health care, can lead to lower investments in another segment,
say offices. In this respect construction spending can reveal dynamics among the segments.

21
 Domestic and sector variables are taken from Federal Reserve Bank of St. Louis, EPU,
Bloomberg, and NCREIF (see Appendix A for complete definitions, sources and transfor-
mations). Table 3 presents descriptive statistics of these independent indicators.

(INSERT TABLE 3 HERE)

4 Empirical results

4.1 Determinants of REIT returns

The robust indicators according to the BMA approach are shown in Table 4 for all 63
REITs during the full time horizon, 2001Q3 to 2024Q1. The complete results are shown
in Appendix C, Table C1.

(INSERT TABLE 4 HERE)

The results show that the traditional REIT-specific indicators, namely size, funds from
operations, liquidity as measured by stock turnover, and expected analyst rating matter.
These results, which have the expected signs, are in line with previous research. However,
what stands out from this analysis is the strong showing of the domestic financial variables
as drivers of REIT excess returns. Our results imply that REIT investors are concerned
about financial variables such as volatility, uncertainty about the U.S. Monetary Policy,
and their performance vis-à-vis market indices. We observe that the signs are as expected
for most variables. The negative sign of VIX implies that investors consider REITs a

22
risky asset, and collectively part of the overall stock market. The positive sign of the
S&P stock index confirms that real estate and equity should be considered a similar asset
class for fund managers and investors, see Liu et al. (2012) for a discussion. Finally,
the positive coefficient on the time spread variable can indicate that REITs are able to
lock in a low interest rate before the yield curve gets steeper. The steeper yield curve
indicates a booming economy with rising rents and lower vacancy rates, leading to increased
profitability for the REITs. Conversely, the opposite is true during recessions. Only one
construction-sector variable, namely construction in the total commercial sector, is robust.
The negative sign indicates that more construction results in lower returns, as more supply
enters the market. Of the domestic macroeconomic variables only inflation is significant
with the expected negative sign. It is hypothesized that although the other domestic
macroeconomic variables most likely affect the direct real estate in the portfolio, they have
no impact at the REIT investor level. Lastly, it should be noted that none of the location
or property type dummies are significant.
We compare the performance of the set of robust drivers with the traditional four factor
model, in Appendix C, Table C7. The model based on the robust drivers outperforms the
factor model.

Next, we separate our sample to analyze whether the drivers are different for large vs.
small REITs. We use the market capitalization as our criteria for size. We split up the
REITs in two parts, with H1 the 50% largest REITs and H2 the 50% smallest REITs.
The halves are adjusted every 3 years, on March 31. So, March 31 2002 for the period
2002-2004, March 31 2005 for the period 2005-2008, and so on. Table 5 shows only the
robust determinants; the complete results can be found in Appendix C, Table C2.

23
 (INSERT TABLE 5 HERE)

The robust drivers for large REITs outnumber the robust drivers for small REITs. In
terms of the REIT specific indicators, one result that stands out is the robustness of stock
turnover, volatility of the stock price, and analyst rating for the largest REITs as compared
to the smallest REITs. This suggests more integration of the large REITs with the stock
market than the smaller REITs. It would also suggest more institutional holdings and more
market awareness among those institutions for such REITs as compared to smaller REITs.
Also the location matters for large REITs, as the MidWest dummy demonstrates. Indeed,
investors in small REITs focus only on the size, where the smaller the REIT the higher
the return, and a limited number of financial variables, but no individual characteristics.
This is in line with previous research using APT and other factor models.
Turning to the domestic drivers, we see that large REITs are influenced by two macroe-
conomic trends: real GDP growth and consumer sentiment. The positive sign on the latter
is expected and the negative sign on the former indicates that REITs are counter-cyclical.
This is an important finding, because when looking at the total sample, see Table 4, we
did not find the two variables to be significant. The domestic financial drivers once again
dominate, albeit more so for large REITs than for small REITs.
A significant and fundamental difference between large and small REITs that is reflected
and supported by the results is the real estate strategy. Large REITs tend to have the high-
est quality of real estate assets (Grade A properties only, highest quality of sustainability
standards and practices incorporated) primarily focusing on global, reputed and only high-
quality credit rated tenants (Fortune 500) that are impacted by a larger swathe of macro
variables. Besides that, their portfolio is largely concentrated in central business districts
(or emerging CBDs) of mostly gateway cities (https://www.cbre.com/insights/reports/global-

24
gateway-cities). Small REITs on the other hand, tend to be regionally focused, mostly own
grade B and C buildings in peripheral locations (non-CBD), target all types of tenants and
are in tier II and tier III cities (and lower).

One of the largest economic shocks to the capital markets that brought a structural change
to it, was the Great Financial Crisis (GFC) in 2008-2009. A second major disruptive event
was the COVID pandemic in 2020. By analyzing a large time-period, we can shine light
on possible differences in the dynamics of REITs in these different periods. We therefore
compare the drivers of REIT returns in the pre-GFC period (2001Q1-2007Q2), the post-
GFC period (2009Q3-2019Q4), and the COVID period (2020Q1-2024Q1). Table 6 shows
the results for the robust determinants; the complete results can be found in Appendix C,
Table C3.

(INSERT TABLE 6 HERE)

The results show that the GFC had a substantial impact on the determinants that drive
REIT returns. First, the number of robust variables drops from a high of twelve pre-GFC
to just seven post-GFC. A possible explanation is that the shakeout of the economic shock
led to more consolidation within the sector. Second, in the pre-GFC period the domestic
variables dominate with 11 out of 12 indicators, while in the post-GFC period REIT-
specific determinants (funds from operations, size, and stock turnover) matter more, and
none of the construction-sector variables is robust. The REIT sector was one of the worst
hit by the GFC, which led to structural changes that were implemented by them. In the
aftermath of the GFC, the REITs focused shifted towards quality real estate assets (and
maintaining those assets to preserve value), location driven investment and following real
estate market fundamentals of supply and demand which essentially translates to a “real

25
estate specific” strategy than driven by just other factors such as lending environment,
herding and more speculative driven investments . It also led to consolidation in the in-
dustry with firms who had followed the former strategy were bankrupt and absorbed by
private equity or other REITs (Equity Office and General Growth Properties are two such
examples). This is supported by our results (Table 6), where the number of factors has
been reduced to just 7. Reiterating, the focus has shifted to three REIT specific factors
from just one pre-GFC, while macro factors have reduced to just two from four in the
pre-GFC time frame.
In the COVID period, a return to some of the pre-GFC variables can be observed.
The number of significant drivers increases to ten, while REIT returns are once again
driven mainly by domestic variables, including total construction spending. Also, there
is a remarkable overlap of five specific variables between pre-GFC and COVID, namely
the volatility of the stock price, the percentage change in the monetary base, the time
spread, the performance of the NCREIF index, and total construction spending. This
implies that the investment strategy seems to return to the pre-GFC period strategy. The
pre-GFC period was similar to the COVID period as in both periods excess risk-taking
led to speculative bubbles in the stock markets and residential real estate, driven by low
interest rates and easy money (prior to the GFC from the financial sector, and in the
COVID period from fiscal stimulus and reduced consumer spending) (Agaton Lombera
et al., 2024). Laborda and Olmo (2021) measure volatility spillovers between sectors of
economic activity using network connectivity measures. Their study shows that the sectoral
volatility connectedness is a relevant measure during periods of high uncertainty. Leading
up to the Global Financial Crisis, their index starts to increase and reaches a peak of 78.55%
on June 15, 2009. It reaches similar high levels during the COVID pandemic. Finally, we
can conclude that in the COVID period, industrial REITs have been outperforming office
and healthcare REITs.

26
4.2 Determinants of the volatility of REIT returns

In this section we present the results from the analysis of the volatility of the returns. The
most robust drivers are shown in Table 7, and the complete results in Appendix C, Table
C4.

(INSERT TABLE 7 HERE)

While REIT returns are driven predominantly by domestic conditions in the domes-
tic macroeconomic and financial conditions, the volatility of these returns is dominated
by REIT-specific indicators. Stock turnover, size and expected analyst rating are three
REIT-specific indicators that are robust drivers for both returns (see Table 4) and their
volatilities (see Table 7). All three are derived from financial market data. Larger REITs
have lower returns, but also lower volatility. Shares with higher turnover in the stock mar-
kets have higher returns, but also higher volatility. These results align with the expected
pattern of higher (lower) returns and higher (lower) risk, where risk is proxied by volatility.
A better analyst rating leads to higher returns and lower volatility, which underscores the
relevance of these ratings. Also the other two robust REIT-specific drivers have the ex-
pected sign. REITs with a relatively higher amount of short-term debt are more volatile,
as rolling over debt increases the interest rate risk. A higher operating margin leads to a
reduction in volatility, which may be explained by the increased resilience to face adverse
situations for REITs that have a higher operating profit margin.

Repeating similar analysis from the previous section, we explore possible size effects on the
volatility. Table 8 shows the robust drivers only; for the complete results see Appendix
C, Table C5. The result that stands out is the asymmetrical impact between large and

27
small REITs. For small REITs the financial ratios, in particular leverage and profitabil-
ity, matter more for the volatility on returns. In general, the volatilities on these smaller
REITs are driven more by REIT-specific conditions and features. Only three out of nine
robust determinants for small REITs are also robust for large REITs, namely M2 growth,
NCREIF returns, and analyst rating. This confirms the asymmetry between large and
small REITs for the volatility. The second finding from Table 8 is that the volatility is
mainly driven by REIT-specific indicators. We can now draw the same conclusion as we
did for the whole sample size, namely that the REIT-specific indicators matter more for the
volatility of the returns, while the actual returns are driven more by domestic indicators,
regardless of the size of the REIT. This confirms the robustness of our findings.
.
(INSERT TABLE 8 HERE)

Next we compare the drivers for return volatility pre-GFC, post-GFC, and COVID. The
results are shown in Table 9, with the complete results in Appendix C, Table C6. We find
that the dynamics in each period are different, and comparable to the dynamics of the
returns. The number of robust drivers for the pre-GFC period is the largest, similar to
the results found for returns (see Table 6). REIT-specific drivers, based on financial state-
ment ratios, financial markets data and location and property type dummies outnumber
the domestic indicators. In the post-GFC period the number of significant drivers drops
from 11 to 4, although the REIT-specific indicators continue to dominate. In the COVID
period the number of robust drivers rebounces. In this period, the volatility is driven by
REIT-specific and domestic indicators in the same proportion. M2 growth is the only
robust driver of both return and volatility in this period, which reflects the influence of the

28
strong monetary expansion and fiscal stimulus in 2020-2022.

(INSERT TABLE 9 HERE)

5 Conclusion

REITs are important derivative products that allow investors access to the benefits of in-
vesting in real estate assets without significant capital contribution and operational exper-
tise. In the last decade, REITs have become increasingly relevant to institutional portfolio
due to their performance and risk measures (Pagan, 2023). This makes understanding
drivers of REIT returns even more important as their asset allocation in a mixed asset
portfolio has been increasing since 2000, with the current allocation close to 10.2% (Pen-
sion Real Estate Association, 2023). The topic has long been researched in the real estate
literature but has shown mixed results with no clear consensus emerging from previous
studies. While some studies in the literature have utilized an index to measure the REIT
returns, others have used all the publicly listed REITs. Since most REITs tend to specialize
in a property type, identifying the factors driving their returns tend to be scattered. Our
study focuses on specific property types REITs, to mitigate this concern and have a clearer
interpretation of the results. Using a sample of REITs specializing on three property types,
namely office, industrial, and healthcare REITs, we are also able to cover all sector types
in the economy as listed by the North American Industry Classification System. Over an
expansive time horizon from 2001 to 2024, encompassing at least two boom to bust eco-
nomic cycles and the turmoil of the COVID era, this study uses Bayesian Model Averaging
(BMA), a data-driven process, to identify robust determinants of REIT returns and their

29
volatility. Our exhaustive list of predictors consists of 18 REIT-specific indicators and 19
domestic macroeconomic and financial factors. For our overall sample and over the entire
time horizon, we find strong dominance of domestic financial variables over REIT-specific
indicators as drivers of REIT excess returns. Unlike previous studies, we don’t find sig-
nificant results for the domestic macroeconomic variables. This is possibly due to the use
of more exhaustive set of variables in our study which dominate others used in previous
studies (2021). Our sub-sample analysis of small and large REITs, shows similar results
but more robust drivers for the larger REITs. In addition, we find that returns of large
REITs are driven by stock turnover, volatility of stock price, and analyst rating suggesting
more integration for these entities with the stock market than for smaller REITs. For our
sub-time period analysis, the results show that the number of robust variables drops from
a high of 12 pre-GFC to just 7 post-GFC suggesting a shakeout of the economic shock
leading to more consolidation within the sector, as well as more fundamental driven invest-
ing by REITs. Key robust variables funds from operations (FFO), size, and stock turnover
become significant drivers post-GFC, while none of the construction-sector variables are
significant. For our analysis of the COVID period, we find that the number of significant
drivers increases to 10, and returns are mainly driven by domestic variables. A key con-
tribution of this study is identifying the drivers of REIT return volatility. We find those
are dominated by REIT specific variables namely stock turnover, size, short-term debt to
total debt, expected analyst rating, and operating margin. For the sub-sample of small
and large REITs, the result is asymmetrical. While both are influenced by REIT spe-
cific factors, for small REITs, the financial ratios namely leverage and profitability matter
more. When we look at the sample broken down by various time-periods, we find that the
three periods analyzed in our study have distinctly different drivers. Pre-GFC volatility
is driven by REIT-specific factors based on financial statement ratios, financial markets
data, and location dummies while the number of total drivers of volatility are reduced sig-

30
nificantly from 11 to 4, when analyzing the post-GFC time period. Analyzing the COVID
time-frame, we find that volatility is driven by REIT-specific and domestic indicators in
the same proportion. This drop in number of robust variables mimics the drop that was
observed for returns.
The results from our paper provide further insights into the relatively black-box of
determinants of REIT returns and particularly in their volatility. Returns and volatility
follow very different patterns. While factors that drive REIT returns have been identi-
fied with more consensus around most factors in literature, volatility on the other hand
is influenced by different factors, as in several finance theories and models. For REITs
specifically, those have been mired by confounding factors given the dual nature of REITs,
with real estate assets in Main Street while being traded in Wall Street. We expect that
this study can further contribute to the identification of the drivers of volatility and open
up more research on this topic. Overall, our study will give institutional investors more
information, and help them build robust portfolios and better asset allocation strategies.

Acknowledgements

We thank Swati Pattel and Jeremy van Denack for excellent research assistance, LHBS
Business Council and the McMullen Family for support, and attendants of NBEA 2022,
ARES 2022 and research seminars at Monmouth University and Bucknell University for
their valuable feedback. We particularly want to thank the two anonymous reviewers for
their valuable suggestions, which have greatly improved the paper.

31
References

Abraham, J. M. and P.H. Hendershott (1996), “Bubbles in metropolitan housing markets”,

Journal of Housing Research, 191–207.

Adams, Z., R. Füss, and F. Schindler (2015), “The sources of risk spillovers among US
REITs: financial characteristics and regional proximity”, Real Estate Economics, 43(1),
67–100.

Agatón Lombera, D. I., D. A. Cardoso López, J. A López Cabrera, and J. A. Nuñez Mora
(2024), “Market reactions to U.S. financial indices: A comparison of the GFC versus
the COVID-19 pandemic crisis”, Economies, 12(7).

Akinsomi, O. (2021), “How resilient are REITs to a pandemic? The COVID-19 effect”,
Journal of Property Investment and Finance, 39, 19–24.

Akinsomi, O., G. C. Aye, V. Babalos, F. Economou, and R. Gupta (2016), “Real estate
returns predictability revisited: novel evidence from the US REITs market”, Empirical
Economics, 51, 1165–1190.

Alcock, J. and E. Steiner (2018), “Fundamental drivers of dependence in REIT returns”,

J Real Estate Finan Econ, 57, 4–42.

Amini, S. M. and C. F. Parmeter (2011), “Bayesian Model Averaging in R”, Journal of

Economic and Social Measurement, 36, 253–287.

Arora, P., R. Killins, and P. Gangineni (2019), “REIT-specific and macroeconomic deter-
minants of REIT returns: Evidence from Singapore”, Accounting and Finance Research,
8(3), 27–42.

Bond, S. and C. Xue (2017), “The cross section of expected real estate returns: Insights
from investment-based asset pricing”, J Real Estate Finan Econ, 54, 403–428.

32
Campbell, S. D., M. A. Davis, J. Gallin, and R. F. Martin (2009), “What moves housing
markets: A variance decomposition of the rent–price ratio”, Journal of Urban Eco-
nomics, 66(2), 90–102.

Carhart, M. M. (1997), “On persistence in mutual fund performance”, The Journal of

Finance, 52(1), 57–82.

Case, K. E. and R. J. Shiller (1989), “The efficiency of the market for single-family homes”,
The American Economic Review, 125–137.

Case, K. E. and R. J. Shiller (1990), “Forecasting prices and excess returns in the housing
market”, Real Estate Economics, 18(3), 253–273.

Chacon, R. G., D. W. French, and K. Pukthuanthong (2021), “The information content

of NAV estimates”, The Journal of Real Estate Finance and Economics, 63, 598–629.

Chen, S.-J., C. Hsieh, T. Vines, and S.-N. Chiou (1998), “Macroeconomic variables, firm-
specific variables and returns to REITs”, Journal of Real Estate Research, 16(3), 269–
278.

Chui, A. C. W., S. Titman, and K. C. J. Wei (2003), “The cross section of expected REIT
returns”, Real Estate Economics, 31, 451–479.

Chung, R., S. Fung, and S-Y. K. Hung (2012), “Institutional investors and firm efficiency
of real estate investment trusts”, The Journal of Real Estate Finance and Economics,
45, 171–211.

Cotter, J. and S. Stevenson (2007), “Uncovering volatility dynamics in daily reit returns”,
Journal of Real Estate Portfolio Management, 13(2), 119–128.

Daniel, K. and S. Titman (1997), “Evidence on the characteristics of cross-sectional vari-

ation in stock returns”, Journal of Finance, 46, 1739–1764.

33
Daniel, K. and S. Titman (1998), “Characteristics or covariances?”, Journal of Portfolio
Management, 24, 24–33.

DeLisle, R. J., S. P. Price, and C. F. Sirmans (2013), “Pricing of volatility risk in REITs”,
Journal of Real Estate Research, 35(2), 223–248.

Desbordes, R., G. Koop, and V. Vicard (2018), “One size does not fit all ... panel data:
Bayesian Model Averaging and data poolability”, Economic Modelling, 75, 364–376.

Edelstein, R., W. Qian, and D. Tsang (2011), “How do institutional factors affect inter-
national real estate returns?”, The Journal of Real Estate Finance and Economics, 43,
130–151.

Eicher, T. S., L. Helfman, and A. Lenkoski (2012), “Robust FDI determinants: Bayesian
Model Averaging in the presence of selection bias”, Journal of Macroeconomics, 34,
637–651.

Fama, E. F. and K. R. French (1993), “Common risk factors in the returns on stocks and
bonds”, Journal of Financial Economics, 33, 23–49.

Fei, P., L. Ding, and Y. Deng (2010), “Correlation and volatility dynamics in REIT re-
turns: performance and portfolio considerations”, The Journal of Portfolio Manage-
ment, 36(2), 113–125.

Gallin, J. (2008), “The long-run relationship between house prices and rents”, Real Estate
Economics, 36(4), 635–658.

Gau, G. W. (1984), “Weak form tests of the efficiency of real estate investment markets”,
Financial Review, 19(4), 301–320.

Geltner, D. and J. Mei (1995), “The present value model with time-varying discount rates:
Implications for commercial property valuation and investment decisions”, The Journal
of Real Estate Finance and Economics, 11, 119–135.

34
Ghysels, E., A. Plazzi, R. Valkanov, and W. Torous (2013), “Chapter 9: Forecasting
real estate prices”, in F. Fukuyama and N. Birdsall, editors, Handbook of Economic
Forecasting, Volume 2A, Elsevier B.V.

Giacomini, E., D. C. Ling, and A. Naranjo (2017), “REIT leverage and return performance:
Keep your eye on the target”, Real Estate Economics, 45, 930–978.

Goebel, P. R., D. M. Harrison, J. M. Mercer, and R. J. Whitby (2013), “REIT momentum

and characteristic-related REIT returns”, J Real Estate Finan Econ, 47, 564–581.

Graham, C. and J. Knight (2000), “Cash flows vs. earnings in the valuation of equity
reits”, Journal of Real Estate Portfolio Management, 6(1), 17–25.

Gu, A. (2002), “The predictability of house prices”, Journal of Real Estate Research, 24(3),
213–234.

Guntermann, K. L. and R. L. Smith (1987), “Efficiency of the market for residential real
estate”, Land Economics, 63(1), 34–45.

Gyourko, J. and R. Voith (1992), “Local market and national components in house price
appreciation”, Journal of Urban Economics, 32(1), 52–69.

Hamilton, B. W. and R. M. Schwab (1985), “Expected appreciation in urban housing

markets”, Journal of Urban Economics, 18(1), 103–118.

Hill, M. D., G. W. Kelly, and W. G. Hardin (2012), “Market value of reit liquidity”, The
Journal of Real Estate Finance and Economics, 45, 383–401.

Hill, R. C., J. R. Knight, and C. F. Sirmans (1997), “Estimating capital asset price in-
dexes”, Review of Economics and Statistics, 79(2), 226–233.

Hill, R. C., C. F. Sirmans, and J. R. Knight (1999), “A random walk down main street?”,
Regional Science and Urban Economics, 29(1), 89–103.

35
Himmelberg, C., C. Mayer, and T. Sinai (2005), “Assessing high house prices: Bubbles,
fundamentals and misperceptions”, Journal of Economic Perspectives, 19(4), 67–92.

Hinne, M., Q. F. Gronau, D. van den Bergh, and E-J. Wagenmakers (2020), “A conceptual
introduction to Bayesian Model Averaging”, Advances in Methods and Practices in
Psychological Science, 3(2), 200–215.

Huerta, D., P.V. Egly, and D. Escobari (2016), “The liquidity crisis, investor sentiment,
and REIT returns and volatility”, Journal of Real Estate Portfolio Management, 22(1),
47–62.

Hung, S. and J. Glascock (2010), “Volatilities and momentum returns in Real Estate
Investment Trusts”, Journal of Real Estate Finance and Economics, 41(2), 126–149.

Kass, R. and A. Raftery (1995), “Bayes factors”, Journal of the American Statistical As-
sociation, 90, 773–795.

Kass, R. and L. Wasserman (1995), “A reference Bayesian test for nested hypothesis with
large samples”, Journal of the American Statistical Association, 90, 928–934.

Kaul, G. (1996), “Predictable components in stock returns”, Handbook of Statistics, 14,

269–296.

Laborda, R. and J. Olmo (2021), “Volatility spillover between economic sectors in financial
crisis prediction: Evidence spanning the Great Financial Crisis and Covid-19 pandemic”,
Research in International Business and Finance, 57(C), 1–13.

Lamont, O. and J. C. Stein (1999), “Leverage and house-price dynamics in US cities”, The
Rand Journal of Economics, 30(3), 498–514.

Leamer, E. A. (1985), “Sensitivity analyses would help”, American Economic Review,

75(3), 308–313.

36
Letdin, M., C. S. Sirmans, G. S. Sirmans, and E. N. Zietz (2019), “Explaining REIT
returns”, Journal of Real Estate Literature, 27(1), 1–25.

Li, J. and L. Lei (2011), “Determinants and information of reit pricing”, Applied Economics
Letters, 18(15), 1501–1505.

Ling, D. C., G. Marcato, and C. Zheng (2022), “Does asset location and concentration
explain reit ipo valuation?”, Real Estate Economics, 50(3), 672–706.

Ling, D. C., A. Naranjo, and M. D. Ryngaert (2000), “The predictability of equity REIT
returns: Time variation and economic significance”, Journal of Real Estate Finance and
Economics, 20(2), 117–136.

Liu, J., G. Loudon, and G. Milunovich (2012), “Linkages between international REITs:
The role of economic factors”, Journal of Property Investment & Finance, 30(5), 473–
492.

MacKinnon, G. H. and A. Al Zaman (2009), “Real estate for the long term: the effect of
return predictability on long-horizon allocations”, Real Estate Economics, 37(1), 117–
153.

Malpezzi, S. (1999), “A simple error correction model of house prices”, Journal of housing
economics, 8(1), 27–62.

Modigliani, F. and M. H. Miller (1958), “The cost of capital, corporation finance and the
theory of investment”, The American Economic Review, 48(3), 261–297.

Moral-Benito, E. (2013), “Model Averaging in economics: An overview”, Journal of Eco-

nomic Surveys, 29(1), 46–75.

Moral-Benito, E. and O. Roehn (2016), “The impact of financial regulation on current

account balances”, European Economic Review, 81, 148–166.

37
Naranjo, A. and D. C. Ling (1997), “Economic risk factors and commercial real estate
returns”, The Journal of Real Estate Finance and Economics, 14, 283–307.

Ooi, J. T. L. and K-H. Liow (2004), “Risk-adjusted performance of real estate stocks:
Evidence from developing markets”, The Journal of Real Estate Research, 26(4), 371–
396.

Pagan, J. (January 11, 2023), “The role of real estate investments in a portfolio”, Morning
Star, https://www.morningstar.com/funds/role–real–estate–investments–portfolio.

Payne, J. E. (2003), “Shocks to macroeconomic state variables and the risk premium of
reits”, Applied Economics Letters, 10(11), 671–677.

Plazzi, A., W. Torous, and R. Valkanov (2010), “Expected returns and expected growth
in rents of commercial real estate”, The Review of Financial Studies, 23(9), 3469–3519.

Porwal, A. and A.E. Raftery (2022), “Comparing methods for statistical inference with
model uncertainty”, PNAS, 119(16), 1–8.

PREA (2023), “Consensus Forecast Survey Q3 2023”, Survey Report, https://www.prea.org/

research/industry-surveys/, Pension Real Estate Association, Hartford, CT.

Rapach, D. and G. Zhou (2013), “Forecasting stock returns”, in Handbook of economic

forecasting, volume 2, 328–383.

Rosen, K. T. (1984), “Toward a model of the office building sector”, Real Estate Economics,
12(3), 261–269.

Ross, S. A. (1976), “The arbitrage theory of capital asset pricing”, Journal of of Economic
Theory, 13, 341–360.

Ross, S. A. and R. C. Zisler (1991), “Risk and return in real estate”, The Journal of Real
Estate Finance and Economics, 4, 175–190.

38
Schindler, F. (2013), “Predictability and persistence of the price movements of the S&P/Case-
Shiller house price indices”, The Journal of Real Estate Finance and Economics, 46,
44–90.

Ward, C. (2008), “Bayesian REIT volatility estimation and institutional portfolio alloca-
tion”, Journal of Real Estate Portfolio Management, 14(4), 425–442.

Yung, K. and N. Nafar (2017), “Investor attention and the expected returns of REITs”,
International Review of Economics & Finance, 48, 423–439.

Zellner, A. (1986), “On assessing prior distributions and Bayesian regression analysis with
g-prior distributions”, in P.K. Goel and A. Zellner, editors, Bayesian Inference and De-
cision Techniques: Essays in Honor of Bruno De Finetti (Studies in Bayesian Econo-
metrics Statistics), volume 6, Elsevier Science Ltd, North Holland, 233–243.

39
Appendix

A: Explanatory variables: definitions and sources

40
Symbol Definition Transformation
Source

1. REIT-specific indicators
Debt-to- (Short and Long Term Debt / Sharehold- None Bloomberg
Equity ratio ers’ Equity) * 100
ST debt to Short term debt / (Short and Long Term Ratio Bloomberg
Total debt Debt) * 100
Asset growth Change in Total assets value Quarterly Bloomberg
change
Cash to Total (Cash and Cash Equivalents / Total As- Ratio Bloomberg
Assets sets) * 100
Asset Amount of revenues generated per dollar None Bloomberg
Turnover of assets. Calculation: TTM3 Net Sales
/ ((Total Assets Current Period + Total
Assets Prior Year Period) / 2)
Capex to To- (-Capital Expenditures / Total Assets) * Ratio Bloomberg
tal Assets 100
Funds from Net Income + Depreciation + Amortiza- None Bloomberg
Operations tion + Losses on Sales of Assets - Gains
on Sales of Assets - Interest Income
FCF to (TTM Free Cash Flow + TTM Oper- None Bloomberg
Enterprise ating Lease Liability Payment + FCF
Value Annualization Transition Adjustment)/
Current Enterprise Value) * 100
ROA TTM EBITDA / Average Assets * 100 None Bloomberg
ROE TTM Net Income / Average Total Com- None Bloomberg
mon Equity * 100
Operating EBIT / Total Revenue * 100 None Bloomberg
margin
Size Total current market capitalization ln(market Bloomberg
cap, in
billions)
Price to Book Ratio of the (last) stock price to the book None Bloomberg
ratio value per share
Stock Ratio of total amount traded in the se- Ratio trad- Bloomberg
turnover curity, as percentage of total market cap ing volume
to current
shares out-
standing

41
Volatility Annualized standard deviation of the rel- None Bloomberg
stock return ative price change for the 30 most recent
(30d) trading days closing price, expressed as
a percentage.
Expected an- Stock analysts’ ratings on a scale of 1 None Bloomberg
alyst rating to 5, with 5 being optimistic and 1 pes-
simistic on expected stock performance
Property Property type in which REIT invests in Binary dum- Bloomberg
type dum- predominantly, with Office REITs the mies for
mies base healthcare
and indus-
trial REITs
Location Dummies for location of the headquar- Binary dum- Bloomberg
dummies ters of the REIT, with East the base case mies for
West, South
and MidWest

2. Domestic indicators
Real GDP U.S. Gross Domestic Product by Expen- None FRBSL
growth diture in Constant Prices: growth rate,
quarterly, seasonally adjusted
Unemployment U.S. unemployment Rate, Percent, Last month FRBSL
Monthly, Seasonally Adjusted of the quarter
Consumer Consumer Sentiment, Index Sentiment FRBSL
sentiment 1966:Q1=100, Monthly, Not Seasonally minus TTM
Adjusted (University of Michigan) average
sentiment
Inflation (Q- Consumer Price Index for All Urban Last month FRBSL
on-Q) Consumers: All Items in U.S. City Av- of the quar-
erage, Index 1982-1984=100, Monthly, ter; Percent-
Seasonally Adjusted age change
in CPI-U
(quarterly)
Expected In- University of Michigan: U.S. inflation ex- None FRBSL
flation pectation, Percent, Monthly, Not Sea-
sonally Adjusted

42
M2 growth U.S. Real M2 Money Stock, Billions Last month FRBSL
of 1982-84 Dollars, Monthly, Seasonally of the quar-
Adjusted ter; Per-
centage
change in M2
(quarterly)
Wage growth Employment Cost Index: Wages and Percentage FRBSL
Salaries: Private Industry Workers, In- change in
dex Dec 2005=100, Quarterly, Season- wages (quar-
ally Adjusted terly)
Real interest 10 year yield minus consumer price in- Last day of FRBSL
rate (10y) flation, with 10-Year Treasury Constant the quarter
Maturity Rate, Percent, Daily, Not Sea- interest rate,
sonally Adjusted adjusted
for infla-
tion (both
annualized)
Time spread 10-Year Treasury Constant Maturity Mi- Last day of FRBSL
nus 3-Month Treasury Constant Matu- the quarter
rity, Percent, Daily, Not Seasonally Ad-
justed
VIX CBOE Volatility Index (VIX), Index, Last day of FRBSL
Daily, Not Seasonally Adjusted the quarter
MPU Monetary Policy Uncertainty (MPU): Last month EPU
Proportion of articles that meet Econ- of the quarter
omy, Politics, Uncertainty and Monetary
Policy criteria across newspapers.
SP500 stock S&P 500 stock market index, Daily, Not Percentage Bloomberg
return Seasonally Adjusted change in
index (quar-
terly)
NCREIF re- Total return on office real estate invest- None NCREIF
turn ments: sum of income return (net oper-
ating income / equity market value) and
capital return (increase in value of the
property market value / equity market
value) NCREIF Property Index (NPI)
∆ House Housing Starts: Total: New Privately Change FRBSL
starts Owned Housing Units Started, Thou- (quarterly)
sands of Units, Monthly, Seasonally Ad-
justed Annual Rate

43
∆ Construc- Total Construction Spending, Millions of Percentage FRBSL
tion total Dollars, Monthly, Seasonally Adjusted change
Annual Rate (quarterly)
∆ Construc- Total Construction Spending: Commer- Percentage FRBSL
tion commerc cial, Millions of Dollars, Monthly, Sea- change
sonally Adjusted Annual Rate (quarterly)
∆ Construc- Total Construction Spending: Health Percentage FRBSL
tion health Care, Millions of Dollars, Monthly, Sea- change
sonally Adjusted Annual Rate (quarterly)
∆ Construc- Total Construction Spending: Educa- Percentage FRBSL
tion educat tional, Millions of Dollars, Monthly, Sea- change
sonally Adjusted Annual Rate (quarterly)
∆ Construc- Total Construction Spending: Office, Percentage FRBSL
tion office Millions of Dollars, Monthly, Seasonally change
Adjusted Annual Rate (quarterly)

44
B: REITs included in the sample

45
Office REITs
# Ticker Name
1 FSP FRANKLIN STREET PROPERTIES C
2 1993649D BIOMED REALTY TRUST INC
3 DEI DOUGLAS EMMETT INC
4 OPI OFFICE PROPERTIES INCOME TRU
5 PDM PIEDMONT OFFICE REALTY TRU-A
6 HPP HUDSON PACIFIC PROPERTIES IN
7 EQC EQUITY COMMONWEALTH
8 VNO VORNADO REALTY TRUST
9 PCFO PACIFIC OFFICE PROPERTIES TR
10 BDN BRANDYWINE REALTY TRUST
11 CUZ COUSINS PROPERTIES INC
12 1448021D PARKWAY PROPERTIES INC
13 CDP COPT DEFENSE PROPERTIES
14 CMCT CREATIVE MEDIA & COMMUNITY T
15 HIW HIGHWOODS PROPERTIES INC
16 KRC KILROY REALTY CORP
17 ARE ALEXANDRIA REAL ESTATE EQUIT
18 BXP BOSTON PROPERTIES INC
19 SLG SL GREEN REALTY CORP
20 9876555D GRAMERCY PROPERTY TRUST INC
21 DEA EASTERLY GOVERNMENT PROPERTI
22 PGRE PARAMOUNT GROUP INC
23 CXP COLUMBIA PROPERTY TRUST INC
24 SIR SELECT INCOME REIT
25 JBGS JBG SMITH PROPERTIES
26 TIER TIER REIT INC
27 NYRT NEW YORK REIT LIQUIDATING LL
28 PSTL POSTAL REALTY TRUST INC- A

Healthcare REITs
29 SBRA SABRA HEALTH CARE REIT INC
30 MPW MEDICAL PROPERTIES TRUST INC
31 HR HEALTHCARE REALTY TRUST INC
32 2392595D PHYSICIANS REALTY TRUST
33 DOC HEALTHPEAK PROPERTIES INC
34 NHI NATL HEALTH INVESTORS INC
35 UHT UNIVERSAL HEALTH RLTY INCOME
36 VTR VENTAS INC
37 WELL WELLTOWER INC
38 LTC LTC PROPERTIES INC
39 OHI OMEGA HEALTHCARE INVESTORS
40 9990721D HEALTHCARE REALTY TRUST INC
41 DHC DIVERSIFIED HEALTHCARE TRUST
42 CTRE CARETRUST REIT INC
43 SNR NEW SENIOR INVESTMENT GROUP
44 CHCT COMMUNITY HEALTHCARE TRUST I
45 GMRE GLOBAL MEDICAL REIT INC
46
Industrial REITs
# Ticker Name
46 1817382D DCT INDUSTRIAL TRUST INC
47 TRNO TERRENO REALTY CORP
48 STAG STAG INDUSTRIAL INC
49 REXR REXFORD INDUSTRIAL REALTY IN
50 DRE DUKE REALTY CORP
51 EGP EASTGROUP PROPERTIES INC
52 2314374D MONMOUTH REAL ESTATE INV COR
53 LXP LXP INDUSTRIAL TRUST
54 FR FIRST INDUSTRIAL REALTY TR
55 PLD PROLOGIS INC
56 COLD AMERICOLD REALTY TRUST INC
57 GPT GRAMERCY PROPERTY TRUST
58 IIPR INNOVATIVE INDUSTRIAL PROPER
59 PLYM PLYMOUTH INDUSTRIAL REIT INC
60 ILPT INDUSTRIAL LOGISTICS PROPERT
61 INDT INDUS REALTY TRUST INC
62 QTS QTS REALTY TRUST INC-CL A
63 CONE CYRUSONE HOLDCO LLC

47
C: Determinants of REIT excess returns and volatili-

ties, 2001Q3-2024Q1

(INSERT TABLE C1 HERE)

(INSERT TABLE C2 HERE)

(INSERT TABLE C3 HERE)

(INSERT TABLE C4 HERE)

(INSERT TABLE C5 HERE)

(INSERT TABLE C6 HERE)

(INSERT TABLE C7 HERE)

48
Table 1: Excess returns on selected U.S. REITs, REIT indices, and S&P 500 index, 2001Q3-
2024Q1: Stylized facts
Our selection NCREIF Equity REIT S&P500
63 REITs Index Index
Mean 2.85 1.87 2.74 1.97
Median 3.02 2.22 2.68 3.15
Standard Deviation 18.87 2.53 10.75 8.31
Kurtosis 224.72 4.29 3.38 0.52
Skewness 7.97 -1.74 -0.90 -0.86
Range 644.32 14.54 72.08 38.80
Minimum -77.65 -8.40 -38.80 -22.56
Maximum 566.67 6.14 33.28 16.24
Count 3682 91 91 91
Notes: BSG (Boonman, Sah and Geurts) selection refers to the 63 office,
industrial, and healthcare REITs of our sample.
Data sources: NCREIF, NAREIT, and Bloomberg, respectively.

Table 2: Descriptive statistics for 63 U.S. office, industrial, and healthcare REITs, 2001Q3-
2024Q1: REIT-specific indicators

mean median standard 10% 90%

deviation percentile percentile
Debt-to-Equity 130.657 94.984 280.061 44.637 159.665
ST debt / Total debt 55.917 65.698 34.623 0.000 95.149
Asset growth 3.282 0.697 18.835 -2.563 8.857
Cash / Total Assets 2.819 0.950 7.880 0.188 5.861
Asset Turnover 0.126 0.122 0.042 0.088 0.162
Capex / Total Assets -2.803 -1.375 4.283 -6.565 -0.154
FCF / Enterprise Value -4.053 -1.959 18.303 -15.029 3.494
Funds from Operations 70.344 40.665 135.907 4.832 161.827
ROA 2.850 2.451 3.927 -0.548 6.348
ROE 4.124 5.033 18.059 -2.349 13.539
Operating margin 29.092 27.933 24.396 6.705 61.373
Size 7.588 7.663 1.389 5.932 9.306
P-B ratio 1.993 1.710 2.086 0.890 2.989
Stock turnover 46.923 40.643 32.611 18.399 77.978
Volatility Stock Price 34.655 22.572 63.978 14.142 50.283
Expected Analyst Rating 3.649 3.667 0.677 2.857 4.538
Notes: 3682 observations; ROA only used for volatility;
ROE and Volatility 30 Day stock price only used for Returns.

49
Table 3: Descriptive statistics for domestic macroeconomic, domestic financial, and
construction-sector indicators, 2001Q3-2024Q1
mean median std dev 10% perc 90% perc
Real GDP growth 0.52 0.60 1.32 -0.32 1.19
Unemployment 5.81 5.40 1.90 3.62 9.08
Consumer sentiment -0.21 0.90 6.67 -10.20 6.83
Inflation (q-on-q) 0.62 0.61 0.76 -0.11 1.17
Expected inflation 3.09 3.00 0.74 2.32 4.30
M2 growth 0.93 0.87 1.98 -0.87 2.12
Wage growth 0.70 0.67 0.28 0.36 1.09
Real interest rate (10y) 1.00 0.95 0.76 0.00 2.06
Time spread 1.57 1.73 1.29 -0.24 3.24
VIX 19.86 17.29 8.65 12.05 31.75
MPU 96.39 79.93 66.15 33.22 169.26
SP500 return 1.97 3.15 8.31 -11.82 11.12
NCREIF return 1.87 2.22 2.53 -1.72 4.50
∆ House starts -7.48 19.00 301.57 -391.40 341.20
∆ Construction total 1.03 1.47 2.55 -2.64 3.72
∆ Construction commerc 0.87 1.89 4.85 -5.98 5.84
∆ Construction educat 0.68 0.74 3.03 -3.14 4.46
∆ Construction health 1.12 1.31 2.55 -2.04 4.17
∆ Construction office 0.95 1.39 4.58 -4.58 7.01
Number of observation: 91 quarters
Data from Federal Reserve Bank of St. Louis, EPU (MPU),
Bloomberg (S&P500), and NCREIF (NCREIF returns).

50
 Table 4: REIT excess returns, 2001Q3-2024Q1: robust indicators

Variable PIP P.Mean P.Std.Dev CP.Sign

Size 1.000 -1.975 0.291 0.000
Time spread 1.000 1.690 0.289 1.000
VIX 1.000 -0.377 0.058 0.000
MPU 1.000 0.037 0.006 1.000
SP500 return 1.000 0.830 0.049 1.000
NCREIF return 1.000 0.850 0.187 1.000
∆ Construction commerc 0.999 -0.400 0.082 0.000
Stock turnover 0.998 0.046 0.011 1.000
Funds from Operations 0.960 0.009 0.003 1.000
Expected Analyst Rating 0.769 1.062 0.691 1.000
Inflation (q-on-q) 0.503 -0.678 0.755 0.000
Notes:
PIP: Post-Inclusion Probability; P.Mean: Posterior Mean; P.SD: Posterior
Standard Deviation; Cond.Sign: proportion of positive signs in regression;
specifications.
Only robust variables are shown: an effect is weak, positive, strong, or decisive
if the posterior inclusion probabilities lie between 50–75%, 75–95%, 95–99% or
are greater than 99%, respectively. The variables are ranked for descending PIP.
Complete results shown in Appendix C, Table C1 .

51
Table 5: Robust determinants of REIT excess returns, 2001Q3-2024Q1: PIP and sign for
largest vs. smallest REITs

H1 (largest REITs) H2 (smallest REITs)

Variable PIP sign PIP sign
PIP Sign PIP Sign
Funds from Operations 0.952 + 0.144 +
Size 0.993 - 1.000 -
Stock turnover 1.000 + 0.110 +
Volatility Stock Price 0.695 - 0.030 +
Expected Analyst Rating 0.987 + 0.117 +
Dummy MidWest 0.643 + 0.036 +
Real GDP growth 0.956 - 0.038 -
Consumer sentiment 0.972 + 0.025 -
Time spread 1.000 + 0.872 +
VIX 1.000 - 0.590 -
MPU 1.000 + 0.586 +
SP500 return 1.000 + 1.000 +
NCREIF return 0.995 + 0.272 +
∆ Construction commerc 1.000 - 0.088 -
PIP: Post-Inclusion Probability; sign is based on the posterior mean.
All robust variables are shown; for completeness we also show the PIP
and sign for the other subcategory.
An effect is weak, positive, strong, or decisive if the posterior inclusion
probabilities lie between 50–75%, 75–95%, 95–99% or are greater than
99%, respectively.
Variables that have a positive, strong, or decisive effect (PIP greater
than 0.75 are dark-grey shaded, variables with a weak effect (PIP
between 0.50 and 0.75) are light-grey shaded.

52
Table 6: Robust determinants of REIT excess returns, pre-GFC (2001Q3-2007Q2), post-
GFC (2009Q3-2024Q1) and COVID (2020Q1-2024Q1): PIP and sign

pre-GFC post-GFC COVID

(2001Q3-2007Q2) (2009Q3-2019Q4) (2020Q1-2024Q1)
Variable PIP sign PIP sign PIP sign
Funds from Operations 0.049 + 0.782 + 0.033 +
Size 0.064 - 0.955 - 0.115 -
Stock turnover 0.045 - 0.993 + 0.557 +
Volatility Stock Price 1.000 + 0.217 + 0.783 +
Dummy industrial 0.057 - 0.071 + 0.998 +
Real GDP growth 0.818 + 0.785 - 0.269 -
Unemployment 0.249 - 0.059 - 0.823 -
Consumer sentiment 0.992 + 0.404 - 0.293 +
Inflation (q-on-q) 0.730 + 1.000 - 0.140 -
Expected inflation 0.092 - 0.078 + 0.516 +
M2 growth 0.988 + 0.060 - 0.577 +
Real interest rate (10y) 0.973 - 0.089 - 0.245 -
Time spread 0.900 - 0.070 - 0.644 +
VIX 1.000 - 0.061 - 0.434 +
MPU 0.450 + 0.647 + 0.444 -
SP500 return 0.141 + 1.000 + 0.999 +
NCREIF return 1.000 - 0.128 - 0.950 -
∆ Construction total 0.609 + 0.276 - 0.996 +
∆ Construction commerc 1.000 - 0.043 -
∆ Construction office 0.988 - 0.040 -
PIP: Post-Inclusion Probability; sign is based on the posterior mean.
All robust variables are shown, i.e. variables with a PIP greater than 0.50;
for completeness we also show the PIP and sign for the other subcategories.
Variables that have a positive, strong, or decisive effect (PIP greater than 0.75)
are dark-grey shaded, variables with a weak effect (PIP between 0.50 and 0.75)
are light-grey shaded. Due to the low number of observations in the COVID
period, we excluded the four subcategories for construction.

53
Table 7: Determinants of REIT excess return volatility, 2001Q3-2024Q1: robust indicators

Variable PIP P.Mean P.Std.Dev CP.Sign

Size 1.000 -10.328 2.027 0.000
Stock turnover 1.000 0.864 0.072 1.000
Expected Analyst Rating 1.000 -18.501 3.077 0.000
M2 growth 1.000 11.520 1.199 1.000
NCREIF return 1.000 -14.768 0.993 0.000
Consumer sentiment 1.000 -1.845 0.367 0.000
Operating margin 0.954 -0.298 0.105 0.000
ST debt / Total debt 0.901 0.196 0.087 1.000
Notes:
PIP: Post-Inclusion Probability; P.Mean: Posterior Mean; P.SD: Posterior
Standard Deviation; Cond.Sign: proportion of positive signs in regression;
specifications.
Only robust variables are shown: an effect is weak, positive, strong, or
decisive if the posterior inclusion probabilities lie between 50–75%, 75–95%,
95–99% or are greater than 99%, respectively. The variables are ranked for
descending PIP.

Table 8: Robust determinants of REIT excess return volatility, 2001Q3-2024Q1: PIP and
sign for largest vs. smallest REITs

H1 (largest REITs) H2 (smallest REITs)

Variable PIP sign PIP sign
Debt-to-Equity 0.057 + 0.552 -
ST debt / Total debt 0.035 + 0.978 +
Operating margin 0.134 - 0.552 -
Size 0.489 - 0.988 -
Stock turnover 1.000 + 0.375 +
Expected Analyst Rating 0.965 - 1.000 -
Dummy industrial 0.987 + 0.028 +
Dummy MidWest 0.038 - 0.726 +
Dummy West 0.774 + 0.049 -
Consumer sentiment 1.000 - 0.046 -
M2 growth 1.000 + 0.998 +
Wage growth 0.058 + 0.697 +
NCREIF return 1.000 - 1.000 -
Notes: see notes in Table 7

54
Table 9: Robust determinants of REIT excess return volatility, pre-GFC (2001Q3-2007Q2),
post-GFC (2009Q3-2024Q1) and COVID (2020Q1-2024Q1): PIP and sign

pre-GFC post-GFC COVID

(2001Q3-2007Q2) (2009Q3-2019Q4) (2020Q1-2024Q1)
Variable PIP sign PIP sign PIP sign
ST debt / Total debt 0.253 - 0.112 + 0.751 +
Asset Turnover 0.808 + 0.028 + 0.214 +
Funds from Operations 0.754 + 0.024 - 0.046 +
Operating margin 0.998 - 0.132 - 0.073 -
Size 0.970 - 0.999 - 0.220 -
Stock turnover 0.160 + 1.000 + 0.849 +
Expected Analyst Rating 1.000 - 0.139 - 0.996 -
Dummy healthcare 1.000 + 0.734 - 0.114 +
Dummy West 0.852 + 0.360 + 0.115 -
Dummy South 0.542 - 0.068 + 0.092 -
Real GDP growth 0.661 - 0.026 + 0.112 -
Consumer sentiment 0.049 - 0.041 - 0.818 -
Expected inflation 0.068 + 0.914 - 0.055 -
M2 growth 0.369 + 0.036 + 0.904 +
Real interest rate (10y) 0.142 + 0.022 + 0.766 +
Time spread 0.533 + 0.024 + 0.111 -
∆ Construction health 0.532 + 0.048 - -
Notes: see notes in Table 7

55
 Table C1: Determinants of REIT excess returns, 2001Q3-2024Q1

Variable PIP P.Mean P.Std.Dev CP.Sign

Debt-to-Equity 0.027 0.000 0.000 1.000
ST debt / Total debt 0.067 0.001 0.004 1.000
Asset growth 0.018 0.000 0.002 0.000
Cash / Total Assets 0.016 0.000 0.005 0.390
Asset Turnover 0.084 1.015 3.894 1.000
Capex / Total Assets 0.027 0.001 0.014 0.980
FCF / Enterprise Value 0.017 0.000 0.003 1.000
Funds from Operations 0.960 0.009 0.003 1.000
ROA 0.027 0.002 0.016 0.990
Operating margin 0.146 0.004 0.010 1.000
Size 1.000 -1.975 0.291 0.000
P-B ratio 0.034 -0.005 0.036 0.000
Stock turnover 0.998 0.046 0.011 1.000
Volatility Stock Price 0.078 0.001 0.003 1.000
Expected Analyst Rating 0.769 1.062 0.691 1.000
Dummy healthcare 0.162 0.254 0.648 1.000
Dummy industrial 0.187 0.325 0.762 1.000
Dummy MidWest 0.177 0.332 0.798 1.000
Dummy West 0.044 0.043 0.252 1.000
Dummy South 0.036 -0.028 0.198 0.000
Real GDP growth 0.213 -0.115 0.246 0.000
Unemployment 0.070 -0.031 0.132 0.000
Consumer sentiment 0.205 0.028 0.062 1.000
Inflation (q-on-q) 0.503 -0.678 0.755 0.000
Expected inflation 0.039 0.016 0.156 0.768
M2 growth 0.053 0.015 0.083 0.945
Wage growth 0.071 0.164 0.712 0.999
Real interest rate (10y) 0.026 -0.011 0.104 0.000
Time spread 1.000 1.690 0.289 1.000
VIX 1.000 -0.377 0.058 0.000
MPU 1.000 0.037 0.006 1.000
SP500 return 1.000 0.830 0.049 1.000
NCREIF return 1.000 0.850 0.187 1.000
∆ House starts 0.028 0.000 0.000 0.250
∆ Construction total 0.031 -0.005 0.038 0.002
∆ Construction commerc 0.999 -0.400 0.082 0.000
∆ Construction educat 0.020 0.002 0.022 1.000
∆ Construction health 0.016 -0.002 0.021 0.000
∆ Construction office 0.109 -0.018 0.059 0.000
Notes:
PIP: Post-Inclusion Probability; P.Mean: Posterior Mean; P.SD: Posterior
56
Standard Deviation; Cond.Sign: proportion of positive signs in regression
specifications.
Variables that have a positive, strong, or decisive effect (PIP greater
than 0.75) are dark grey shaded, variables with a weak effect (PIP
between 0.50 and 0.75) are light grey shaded.
Table C2: Determinants of REIT excess returns, 2001Q3-2024Q1: Posterior Inclusion
Probability (PIP), with sample split in 50% largest REITs (measured in market cap) and
50% smallest REITs
Largest 50% Smallest 50%
Variable PIP sign PIP sign
Debt-to-Equity 0.042 + 0.030 -
ST debt / Total debt 0.037 + 0.047 +
Asset growth 0.041 - 0.023 +
Cash / Total Assets 0.031 - 0.021 +
Asset Turnover 0.048 + 0.042 +
Capex / Total Assets 0.043 + 0.026 -
FCF / Enterprise Value 0.304 + 0.021 -
Funds from Operations 0.952 + 0.144 +
ROA 0.217 + 0.026 -
Operating margin 0.327 + 0.034 +
Size 0.993 - 1.000 -
P-B ratio 0.032 + 0.067 -
Stock turnover 1.000 + 0.110 +
Volatility Stock Price 0.695 - 0.030 +
Expected Analyst Rating 0.987 + 0.117 +
Dummy healthcare 0.033 + 0.188 +
Dummy industrial 0.043 + 0.255 +
Dummy MidWest 0.643 + 0.036 +
Dummy West 0.028 + 0.094 +
Dummy South 0.028 - 0.029 -
Real GDP growth 0.956 - 0.038 -
Unemployment 0.034 - 0.110 -
Consumer sentiment 0.972 + 0.025 -
Inflation (q-on-q) 0.045 - 0.037 -
Expected inflation 0.093 + 0.018 -
M2 growth 0.062 + 0.029 +
Wage growth 0.026 + 0.038 +
Real interest rate (10y) 0.027 - 0.043 -
Time spread 1.000 + 0.872 +
VIX 1.000 - 0.590 -
MPU 1.000 + 0.586 +
SP500 return 1.000 + 1.000 +
NCREIF return 0.995 + 0.272 +
∆ House starts 0.031 - 0.025 -
∆ Construction total 0.027 - 0.049 -
∆ Construction commerc 1.000 - 0.088 -
∆ Construction educat 0.020 + 0.028 -
∆ Construction health 0.047 - 0.038 -
∆ Construction office 57
0.029 - 0.192 -
Notes: See notes in Table C1.
Table C3: A3: Determinants of REIT excess returns, pre-GFC (2001Q3-2007Q2), GFC
(2007Q3-2009Q2), post-GFC (2009Q3-2019Q4) and COVID (2020Q1-2024Q1): PIP and
sign

pre-GFC post-GFC COVID

(2001Q3-2007Q2) (2009Q3-2019Q4) (2020Q1-2024Q1)
Variable PIP sign PIP sign PIP sign
Debt-to-Equity 0.085 + 0.026 + 0.042 +
ST debt / Total debt 0.059 - 0.030 + 0.047 +
Asset growth 0.038 - 0.030 + 0.039 -
Cash / Total Assets 0.038 + 0.020 - 0.039 +
Asset Turnover 0.135 + 0.021 - 0.044 -
Capex / Total Assets 0.036 + 0.017 + 0.032 -
FCF / Enterprise Value 0.060 + 0.049 + 0.040 -
Funds from Operations 0.049 + 0.782 + 0.033 +
ROA 0.084 - 0.015 + 0.033 +
Operating margin 0.048 - 0.019 + 0.043 +
Size 0.064 - 0.955 - 0.115 -
P-B ratio 0.056 + 0.020 - 0.075 -
Stock turnover 0.045 - 0.993 + 0.557 +
Volatility Stock Price 1.000 + 0.217 + 0.783 +
Expected Analyst Rating 0.067 + 0.142 + 0.063 +
Dummy healthcare 0.047 + 0.026 + 0.304 +
Dummy industrial 0.057 - 0.071 + 0.998 +
Dummy MidWest 0.058 - 0.040 + 0.102 +
Dummy West 0.048 + 0.064 + 0.035 +
Dummy South 0.129 - 0.023 + 0.053 -
Real GDP growth 0.818 + 0.785 - 0.269 -
Unemployment 0.249 - 0.059 - 0.823 -
Consumer sentiment 0.992 + 0.404 - 0.293 +
Inflation (q-on-q) 0.730 + 1.000 - 0.140 -
Expected inflation 0.092 - 0.078 + 0.516 +
M2 growth 0.988 + 0.060 - 0.577 +
Wage growth 0.188 + 0.129 + 0.091 -
Real interest rate (10y) 0.973 - 0.089 - 0.245 -
Time spread 0.900 - 0.070 - 0.644 +
VIX 1.000 - 0.061 - 0.434 +
MPU 0.450 + 0.647 + 0.444 -
SP500 return 0.141 + 1.000 + 0.999 +
NCREIF return 1.000 - 0.128 - 0.950 -
∆ House starts 0.097 + 0.406 - 0.257 -
∆ Construction total 0.609 + 0.276 - 0.996 +
∆ Construction commerc 1.000 - 0.043 -
∆ Construction educat 0.140 - 0.031 +
58
∆ Construction health 0.091 - 0.044 -
∆ Construction office 0.988 - 0.040 -
Notes: See notes in Table C1.
Table C4: Determinants of Volatility of REIT excess returns, 2001Q3-2024Q1

Variable PIP P.Mean P.Std.Dev CP.Sign

Debt-to-Equity 0.021 0.000 0.001 0.119
ST debt / Total debt 0.901 0.196 0.087 1.000
Asset growth 0.015 0.000 0.014 0.000
Cash / Total Assets 0.154 -0.089 0.232 0.000
Asset Turnover 0.118 11.325 35.072 1.000
Capex / Total Assets 0.021 0.004 0.074 1.000
FCF / Enterprise Value 0.023 -0.002 0.022 0.000
Funds from Operations 0.290 0.013 0.022 1.000
ROE 0.031 -0.004 0.035 0.000
Operating margin 0.954 -0.298 0.105 0.000
Size 1.000 -10.328 2.027 0.000
P-B ratio 0.018 0.003 0.131 0.934
Stock turnover 1.000 0.864 0.072 1.000
Expected Analyst Rating 1.000 -18.501 3.077 0.000
Dummy healthcare 0.020 -0.045 0.768 0.130
Dummy industrial 0.084 0.741 2.796 1.000
Dummy MidWest 0.025 0.130 1.279 1.000
Dummy West 0.024 0.138 1.159 1.000
Dummy South 0.024 -0.126 1.107 0.000
Real GDP growth 0.023 -0.031 0.368 0.095
Unemployment 0.037 -0.070 0.461 0.026
Consumer sentiment 1.000 -1.845 0.367 0.000
Inflation (q-on-q) 0.015 -0.023 0.531 0.134
Expected inflation 0.052 0.302 1.541 1.000
M2 growth 1.000 11.520 1.199 1.000
Wage growth 0.346 7.650 11.619 1.000
Real interest rate (10y) 0.026 0.074 0.710 0.961
Time spread 0.017 -0.014 0.283 0.246
VIX 0.011 0.002 0.039 0.936
MPU 0.118 -0.009 0.029 0.000
SP500 return 0.018 -0.004 0.045 0.002
NCREIF return 1.000 -14.768 0.993 0.000
∆ House starts 0.096 -0.002 0.006 0.000
∆ Construction total 0.113 0.251 0.823 1.000
∆ Construction commerc 0.018 0.002 0.081 0.638
∆ Construction educat 0.047 0.048 0.260 1.000
∆ Construction health 0.045 0.056 0.324 1.000
∆ Construction office 0.018 0.003 0.076 0.786
See Notes in Table C1

59
Table C5: Determinants of REIT excess return volatility, 2001Q3-2024Q1: Posterior In-
clusion Probability (PIP), large vs. small REITs

Largest 50% Smallest 50%

Variable PIP sign PIP sign
Debt-to-Equity 0.057 + 0.552 -
ST debt / Total debt 0.035 + 0.978 +
Asset growth 0.029 + 0.021 -
Cash / Total Assets 0.210 - 0.036 -
Asset Turnover 0.025 + 0.079 +
Capex / Total Assets 0.031 + 0.052 -
FCF / Enterprise Value 0.023 + 0.276 -
Funds from Operations 0.065 + 0.036 +
ROE 0.033 + 0.214 -
Operating margin 0.134 - 0.552 -
Size 0.489 - 0.988 -
P-B ratio 0.039 + 0.094 -
Stock turnover 1.000 + 0.375 +
Expected Analyst Rating 0.965 - 1.000 -
Dummy healthcare 0.075 - 0.060 +
Dummy industrial 0.987 + 0.028 +
Dummy MidWest 0.038 - 0.726 +
Dummy West 0.774 + 0.049 -
Dummy South 0.020 - 0.044 -
Real GDP growth 0.024 - 0.029 -
Unemployment 0.223 - 0.055 -
Consumer sentiment 1.000 - 0.046 -
Inflation (q-on-q) 0.016 - 0.029 +
Expected inflation 0.028 + 0.114 +
M2 growth 1.000 + 0.998 +
Wage growth 0.058 + 0.697 +
Real interest rate (10y) 0.040 + 0.026 -
Time spread 0.043 + 0.150 -
VIX 0.026 + 0.037 +
MPU 0.499 - 0.030 -
SP500 return 0.034 - 0.040 -
NCREIF return 1.000 - 1.000 -
∆ House starts 0.043 - 0.029 -
∆ Construction total 0.061 + 0.047 +
∆ Construction commerc 0.018 + 0.028 -
∆ Construction educat 0.026 + 0.225 +
∆ Construction health 0.159 + 0.028 +
∆ Construction office 0.021 + 0.027 +
Notes: See Table C1.
60
Table C6: Determinants of REIT excess return volatility, pre-GFC (2001Q3-2007Q2), post-
GFC (2009Q3-2024Q1) and COVID (2020Q1-2024Q1): PIP and sign

pre-GFC post-GFC COVID

(2001Q3-2007Q2) (2009Q3-2019Q4) (2020Q1-2024Q1)
Variable PIP sign PIP sign PIP sign
Debt-to-Equity 0.054 + 0.037 - 0.047 +
ST debt / Total debt 0.253 - 0.112 + 0.751 +
Asset growth 0.029 + 0.028 + 0.034 -
Cash / Total Assets 0.042 - 0.216 - 0.185 -
Asset Turnover 0.808 + 0.028 + 0.214 +
Capex / Total Assets 0.046 + 0.031 - 0.104 +
FCF / Enterprise Value 0.054 + 0.030 - 0.058 -
Funds from Operations 0.754 + 0.024 - 0.046 +
ROE 0.099 - 0.066 - 0.068 -
Operating margin 0.998 - 0.132 - 0.073 -
Size 0.970 - 0.999 - 0.220 -
P-B ratio 0.128 + 0.246 - 0.043 +
Stock turnover 0.160 + 1.000 + 0.849 +
Expected Analyst Rating 1.000 - 0.139 - 0.996 -
Dummy healthcare 1.000 + 0.734 - 0.114 +
Dummy industrial 0.499 - 0.029 + 0.036 -
Dummy MidWest 0.204 - 0.090 + 0.031 -
Dummy West 0.852 + 0.360 + 0.115 -
Dummy South 0.542 - 0.068 + 0.092 -
Real GDP growth 0.661 - 0.026 + 0.112 -
Unemployment 0.103 + 0.038 - 0.126 +
Consumer sentiment 0.049 - 0.041 - 0.818 -
Inflation (q-on-q) 0.082 + 0.053 - 0.051 -
Expected inflation 0.068 + 0.914 - 0.055 -
M2 growth 0.369 + 0.036 + 0.904 +
Wage growth 0.057 - 0.031 + 0.245 -
Real interest rate (10y) 0.142 + 0.022 + 0.766 +
Time spread 0.533 + 0.024 + 0.111 -
VIX 0.178 + 0.483 + 0.064 -
MPU 0.051 + 0.042 + 0.070 +
SP500 return 0.058 - 0.034 + 0.052 +
NCREIF return 0.374 - 0.160 - 0.171 -
∆ House starts 0.066 - 0.027 + 0.114 -
∆ Construction total 0.074 - 0.025 + 0.032 +
∆ Construction commerc 0.344 + 0.019 -
∆ Construction educat 0.138 + 0.051 +
∆ Construction health 0.532 + 0.048 -
∆ Construction office 0.056 + 0.093 +
Notes: See notes in Table C1. 61
Table C7: Comparison of excess returns according to the BMA approach vis-à-vis the
traditional Four Factor Model (FFM), for 63 U.S. office, industrial, and healthcare REITs,
2001Q3-2024Q1

BMA FFM
Variable estimate std.error estimate std.error
Size -4.4604 0.4153 ***
Time spread 0.8745 0.2739 ***
VIX -0.3576 0.0540 ***
MPU 0.0316 0.0060 ***
S&P 500 0.8367 0.0440 ***
NCREIF 0.7861 0.1696 ***
∆ Construction commerc -0.3209 0.0781 ***
Stock turnover 0.0349 0.0117 ***
Funds from Operations 0.0102 0.0027 ***
Expected Analyst Rating 0.9228 0.5311 *
Inflation (q-on-q) -0.9255 0.4286 **
Market return 2.3563 0.1211 ***
SMB 0.5581 0.2286 **
HML 1.2219 0.1295 ***
Momentum -0.0656 0.1230
R squared 0.2210 0.1828
Adjusted R squared 0.2055 0.1681
F-statistic 93.0762 202.18
Degrees of Freedom (11, 3609) (4, 3616)
p-value 0.0000 0.0000
Firm fixed effects Yes Yes

Carhart (1997) Four Factor Model, with factors: (i) Market risk factor, or market
excess return (S&P500 return minus risk free rate); (ii) SMB (Small Minus Big:
the average return on small stock portfolios minus the average return on big
stock portfolios; (iii) HML (High Minus Low: the average return on value stock
portfolios minus the average return on growth stock portfolios); and (iv) Momentum
(difference in returns between stocks that have performed well in the past and those
that have performed poorly). The signs are as expected, except for the momentum,
which is not significant.
Data retrieved from the webpage of Kenneth French, link:
https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/index.html

62