Multi Factor Models I Cap M

This article appeared in a journal published by Elsevier.
The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Journal of Financial Economics 106 (2012) 586613
Contents lists available at SciVerse ScienceDirect
Journal of Financial Economics

journal homepage: www.elsevier.com/locate/jfec
Multifactor models and their consistency with the ICAPM$

Paulo Maio a,n, Pedro Santa-Clara b,c,d
a
Hanken School of Economics, Finland

NOVA School of Business and Economics, Portugal
c
NBER, United States
d
CEPR, United Kingdom
b
a r t i c l e in f o
abstract
Article history:
Received 30 August 2011
Received in revised form
2 December 2011
Accepted 17 January 2012
Available online 22 July 2012
Can any multifactor model be interpreted as a variant of the Intertemporal CAPM

(ICAPM)? The ICAPM places restrictions on time-series and cross-sectional behavior of
state variables and factors. If a state variable forecasts positive (negative) changes in
investment opportunities in time-series regressions, its innovation should earn a
positive (negative) risk price in the cross-sectional test of the respective multifactor
model. Second, the market (covariance) price of risk must be economically plausible as
an estimate of the coefcient of relative risk aversion (RRA). We apply our ICAPM
criteria to eight popular multifactor models and the results show that most models do
not satisfy the ICAPM restrictions. Specically, the hedging risk prices have the wrong
sign and the estimates of RRA are not economically plausible. Overall, the Fama and
French (1993) and Carhart (1997) models perform the best in consistently meeting the
ICAPM restrictions. The remaining models, which represent some of the most relevant
examples presented in the empirical asset pricing literature, can still empirically
explain the size, value, and momentum anomalies, but they are generally inconsistent
with the ICAPM.
& 2012 Elsevier B.V. All rights reserved.
JEL classication:
G11
G12
G14
Keywords:
Asset pricing models
Intertemporal CAPM
Predictability of stock returns
Cross-section of stock returns
Value and momentum
1. Introduction
Explaining the dispersion in average excess returns in
the cross-section of stocks has been one of the most
important topics in the asset pricing literature. The
$
We thank an anonymous referee, Antonio Antunes, Matti Keloharju,
Timo Korkeamaki, Anders Lound,

Peter Nyberg, Bill Schwert (the
editor), and seminar participants at the Helsinki Finance Seminar, Bank
of Portugal, and the 2012 SGF Conference (Zurich) for helpful comments.
We are grateful to Kenneth French, Amit Goyal, Lubos Pastor, and Robert
Shiller for making data available on their webpages. A previous version
of this paper was entitled The time-series and cross-sectional consistency of the ICAPM. Maio acknowledges nancial support from the
Hanken Foundation. Santa-Clara is supported by a grant from the
Fundac- a~ o para a Ciencia e Tecnologia (PTDC/EGE-GES/101414/2008).
All errors are ours.
n
Corresponding author.
E-mail address: paulmaio@hanken. (P. Maio).
0304-405X/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jneco.2012.07.001
inability of the Sharpe (1964)Lintner (1965) CAPM to

price portfolios sorted on size, book-to-market, momentum, and other stock characteristics has led to so-called
size, value, and momentum anomalies (Fama and French,
1992, 1993, 1996, among others). In response, several
multifactor models seeking to explain these various
anomalies have emerged in the literature. Typically, these
models include factors in addition to the market return
whose betas help match the dispersion in excess portfolio
returns observed in the cross-section. Many of these
multifactor models have been justied as empirical applications of the Intertemporal CAPM (ICAPM) (Merton,
1973), leading Fama (1991) to interpret the ICAPM as a
shing license to the extent that authors claim it
provides a theoretical background for relatively ad hoc
risk factors in their models. However, Cochrane (2005,
Chapter 9) notes that although the ICAPM does not
directly identify the state variables underlying the risk

P. Maio, P. Santa-Clara / Journal of Financial Economics 106 (2012) 586613
factors, there are some restrictions that these state variables must satisfy. According to Merton, the state variables relate to changes in the investment opportunity set,
which implies that they should forecast the distribution of
future aggregate stock returns. Moreover, the innovations
in these state variables should be priced factors in the
cross-section.
We examine the restrictions associated with the
ICAPM that prevent it from being a shing license for
any multifactor model that seeks to explain the crosssection of stock returns. We identify three main conditions that a multifactor model must meet to be justiable
by the ICAPM and nd that most multifactor models in
the literature do not satisfy these restrictions.
First, the candidates for ICAPM state variables must
forecast the rst or second moments of aggregate stock
returns. We assess the forecasting power of each variable
by conducting time-series long-horizon regressions.
Second, if a given state variable forecasts positive
expected aggregate returns, its innovation (the risk factor)
should earn a positive risk price in cross-sectional tests,
while state variables that forecast negative expected
aggregate returns should earn a negative risk price. Risk
premiums with opposite signs should accrue to innovations to state variables that forecast market volatility.
Thus, it is not enough that the candidate state variables
forecast future aggregate expected returns or the volatility of returns, the corresponding factors should also be
priced in the cross-section with the correct sign. The
intuition for this result is simple. An asset that covaries
positively with innovations to the state variable also
covaries positively with future expected returns. It does
not provide a hedge for reinvestment risk because it offers
lower returns when aggregate returns are expected to be
lower. Hence, a risk-averse rational investor will require a
positive risk premium to invest in such an asset, implying
a positive price of risk for the factor. A similar argument
applies to assets that covary with innovations to market
volatility.
The third restriction associated with the ICAPM is that
the market (covariance) price of risk estimated from the
cross-sectional tests must be economically plausible as an
estimate of the coefcient of relative risk aversion (RRA)
of the representative investor.
Most of the empirical literature on the ICAPM uses
state variables from the predictability literature (shortterm interest rates, bond yields, and aggregate nancial
ratios) in order to meet the rst ICAPM restriction that the
state variables should forecast expected market returns.
Yet, authors largely neglect the other constraints of the
ICAPM: that the market price of risk corresponds to the
risk aversion of the representative investor and especially
that there must be consistency between the hedging
factor risk prices and the corresponding slopes from the
predictive regressions.
In Campbell (1996), the risk prices associated with the
vector autoregressive (VAR) state variables that forecast
market returns are constrained in the sense that they are
linked with the estimated slopes from the VAR. However,
Campbell only tests a specic parametrization with
Epstein-Zin preferences and a VAR to estimate market
587
discount rate news. This paper extends this work, focusing on whether commonly used empirical factor models
satisfy the consistency between time-series slopes and
cross-sectional risk prices to be justiable as ICAPM
applications. Our work is also related to Lewellen, Nagel,
and Shanken (2010) and Lewellen and Nagel (2006), who
advocate that cross-sectional tests of asset pricing models
in general, and the conditional CAPM in particular, should
impose the models theoretical restrictions on the factor
risk prices.
We apply our ICAPM criteria to eight multifactor
models, tested over 25 portfolios sorted on size and
book-to-market (SBM25) and 25 portfolios sorted on size
and momentum (SM25). We include the market return in
the set of testing assets, which enables us to merge the
cross-sectional literature on the ICAPM with the literature
on the time-series aggregate risk-return trade-off. Hence,
we have a total of 16 empirical tests in the cross-section:
eight models and two sets of portfolios.
Table 1 summarizes the main results regarding the
multifactor models satisfying the ICAPM criteria. When
investment opportunities are driven by changing
expected market returns, our results show that only two
modelsthe Fama and French (1993) three-factor model
tested over SBM25, and the Carhart (1997) model tested
Table 1
Consistency of multifactor models with the ICAPM.
This table reports the consistency of the factor risk prices from
multifactor models with the ICAPM criteria. The criteria are associated
with the magnitude of the market risk price (or risk-aversion coefcient)
(g), and the consistency in sign of the risk prices of the hedging factors
with the corresponding predictive slopes over the excess market return
(gz ,Er) and the market variance (gz , s2 r). The multifactor models are
Hahn and Lee (2006) (HL), Petkova (2006) (P), Campbell and
Vuolteenaho (2004) (CV), Koijen, Lustig, and Van Nieuwerburgh (2010)
(KLVN), Fama and French (1993) (FF3), Carhart (1997) (C), Pastor and
Stambaugh (2003) (PS), and the Fama and French (1993) ve-factor
model (FF5). The testing assets in the cross-sectional tests are the 25
size/book-to-market portfolios (SBM25, Panel A), and 25 size/momentum portfolios (SM25, Panel B). A | indicates that the ICAPM criteria
are satised.
gz , Er
gz , s2 r

|
|

|
|
|
|

|

|
|

|

|

Panel A: SBM25
HL
P
CV
KLVN
FF3
C
PS
FF5
Panel B: SM25
HL
P
CV
KLVN
FF3
C
PS
FF5

588
over SBM25 and SM25meet the ICAPM consistency

criteria.
When we consider changes in the investment opportunity set driven by the second moment of aggregate
returns, only the Fama and French (1993) model satises
the ICAPM criteria when tested with the SBM25
portfolios.
In most other models the hedging risk prices have
the wrong sign and the estimates of RRA are not economically plausible. The Koijen, Lustig, and Van Nieuwerburgh
(2010) and Pastor and Stambaugh (2003) models in tests
with SBM25 meet the sign restriction on the hedging risk
prices but do not produce a reasonable estimate for the
risk-aversion coefcient. The Hahn and Lee (2006) model
in the test with SM25 produces a plausible estimate for
RRA, but fails the sign restriction on the risk prices for the
state variable factors. These rejections show that the
ICAPM is not really a shing license after all.
Our ndings are robust to estimating the multifactor
models with an intercept, estimating each model by
second-stage generalized method of moments (GMM),
using alternative measures of the innovations in the state
variables, using alternative test equity portfolios, adding
bond risk premiums to the menu of test assets, estimating
the models in expected return-beta form, conducting a
bootstrap simulation for the slopes in the predictive
regressions, and using alternative proxies for the expected
market return.
Our paper is related to the growing empirical literature
on the ICAPM. An incomplete list of empirical tests of the
ICAPM over the cross-section of stock returns includes
Shanken (1990), Brennan, Wang, and Xia (2004), Ang,
Hodrick, Xing, and Zhang (2006), Gerard and Wu (2006),
Hahn and Lee (2006), Lo and Wang (2006), Petkova
(2006), Bali (2008), Guo and Savickas (2008), Ozoguz
(2009), and Bali and Engle (2010). In related work,
Campbell (1993) develops a theoretical model based on
a representative agent with Epstein and Zin (1991) preferences that leads to two risk factorsthe excess market
return (as in the standard CAPM) and expectations about
future market returns (discount rate news). Campbell
(1996), Chen (2003), Guo (2006a), Campbell and
Vuolteenaho (2004), Chen and Zhao (2009), and Maio
(unpublished) represent variants or extensions of the
two-factor model developed in Campbell (1993). A related
literature focuses on the time-series aggregate risk-return
trade-off. An incomplete list of recent papers includes
Scruggs (1998), Whitelaw (2000), Brandt and Kang
(2004), Ghysels, SantaClara, and Valkanov (2005), Guo
and Whitelaw (2006), Lundblad (2007), Pastor, Sinha, and
Swaminathan (2008), Bali, Demirtas, and Levy (2009), and
Guo, Savickas, Wang, and Yang (2009).
This paper is organized as follows. In Section 2 we
discuss the theoretical restrictions associated with the
ICAPM. In Section 3 we analyze the forecasting power of
ICAPM state variable candidates with regards to the
expected market return. In Section 4 we analyze whether
the factor risk prices from cross-sectional asset pricing
tests are consistent with the time-series predictability of
market returns. In Section 5 we conduct a sensitivity
analysis. In Section 6 we evaluate the predictive ability of
the ICAPM state variables with regards to the volatility of

market return. In Section 7, we conduct a Monte Carlo
simulation experiment to assess the plausibility of our
results.
2. Time-series and cross-sectional implications of the
ICAPM
A simplied version of the Merton (1973) Intertemporal CAPM (ICAPM) is based on the consumption/portfolio choice problem of a representative investor in
continuous time.1 We discuss the restrictions this model
imposes on multifactor asset pricing models.
There are N risky assets, and asset i has an instantaneous rate of return given by
dSi
mi z,t dt si z,t dxi ,
Si
i 1, . . . ,N,
where Si denotes the price of asset i; dxi is a Wiener

process; and the covariance between two arbitrary risky
assets is equal to sij dt.
In this model, investment opportunities are timevarying since both the mean (mi ) and volatility (si ) of
asset returns are functions of a single state variable, z,
which also evolves as a diffusion process:
dz az,t dt bz,t dz,
where dz denotes another Wiener process, and the covariance with the return on risky asset i is equal to siz dt.2
The N 1th asset is a risk-free asset with instantaneous
rate of return equal to r:
dB
r dt:
B
To simplify the exposition, we assume that the risk-free

rate is constant.
The dynamics of wealth (W) are given by
dW
N
X
oi mi rW dt rWC dt
i1
N
X
oi W si dxi ,
i1
where oi denotes the portfolio weight for asset i, and C

stands for consumption. The investor maximizes lifetime
utility:
Z 1

JW,z,t maxEt
UC,s ds ,
5
C, oi
st
subject to the intertemporal budget constraint (4), where

JW,z,t denotes the value function.
It can be shown that the ICAPM equilibrium relation
between expected return and risk is given by
mi r gsim gz siz ,
where g WJ WW W,z,t=J W W,z,t denotes the parameter of relative risk aversion; sim and siz denote the
covariances between the return on asset i and the market
return and state variable, respectively; and gz denotes the
(covariance) risk price associated with the state variable,
1
For a textbook treatment, see Pennacchi (2008, Chapter 13).

The Merton (1973) ICAPM does not directly identify the state
variables. The main restriction in this simple model is that the state
variables forecast the rst two moments of stock returns.
2

which is given by
gz
J Wz W,z,t
,
JW W,z,t
where J W denotes the marginal value of wealth; JWW

is the growth in the marginal value of wealth; and JWz
represents a second-order cross-derivative relative to
wealth and the state variable.
As in Cochrane (2005, Chapter 9), we can approximate
Eq. (6) in discrete time, leading to the pricing equation:
Et Ri,t 1 Rf ,t 1 g Covt Ri,t 1 ,Rm,t 1
gz Covt Ri,t 1 , Dzt 1 ,
where Ri,t 1 is the return on asset i between t and t 1;

Rf ,t 1 denotes the risk-free rate, which is known at t;
Rm,t 1 is the market return; and Dzt 1 denotes the
innovation or change in the state variable. In Eq. (8), the
novelty relative to the standard CAPM (Sharpe, 1964;
Lintner, 1965) is the second term on the right-hand side,
gz Covt Ri,t 1 , Dzt 1 . This means that if the risk price
associated with the state variable is zero, gz J Wz 0,
we are back to the standard static CAPM. This pricing
equation represents the theory behind many multifactor
models in the empirical asset pricing literature, leading
Fama (1991) to call the ICAPM a shing license. Yet, as
should be clear from the derivation of (8), the non-market
factors in such models should proxy for the innovation in
some state variable, Dzt 1 , so they cannot be just
anything.
By using the law of iterated expectations, we can
obtain the ICAPM in unconditional form:
ERi,t 1 Rf ,t 1 g CovRi,t 1 ,Rm,t 1 gz CovRi,t 1 , Dzt 1 :
9
In Eq. (9) there are two sources of risk that explain
average risk premiums.3 The rst is captured by the
static market risk premium associated with the CAPM,
g CovRi,t 1 ,Rm,t 1 , which postulates that an asset that
covaries positively with the market return earns a positive risk premium over the risk-free rate. The intuition is
that such an asset does not provide a hedge against
changes in current aggregate wealth, as it pays in good
times (periods with high returns on wealth), so a riskaverse investor is willing to hold such an asset only if it
offers a premium over the risk-free rate. The estimate for
the relative risk aversion (RRA) coefcient should be
between one and ten (see Mehra and Prescott, 1985, for
example).
To understand that the second source of risk in Eq. (9)
is captured by the term, gz CovRi,t 1 , Dzt 1 , consider a
state variable that predicts positive future market returns.
If the risk price for intertemporal risk (gz ) is positive, an
3
If the factor risk prices were time-varying (as a function of state
variables), there would be additional (scaled) risk factors in the pricing
equation from the interaction between the original factor and the state
variables (see Jagannathan and Wang, 1996; Lettau and Ludvigson,
2001; Cochrane, 2005, Chapter 8, among others). In our case, and
following most of the empirical literature on the ICAPM, we assume
that the risk prices are constant through time. In the next sections, we
test only the unconditional versions of the empirical multifactor models
that are candidates for ICAPM applications.
589
asset that covaries positively with changes in the state

variable (and is thus positively correlated with the future
market expected return) earns a risk premium. The
intuition is that the asset does not provide a hedge against
future negative shocks in the returns of aggregate wealth
(reinvestment risk), as it offers low returns when future
aggregate returns are also expected to be low. Therefore, a
rational investor is willing to hold such an asset only if it
offers an expected return in excess of the risk-free rate.
This central economic intuition of the ICAPM puts a
constraint on the sign of gz , when the model is forced to
price a set of assets in the cross-section. Specically, if the
state variable is positively correlated with future aggregate returns,
Covt Rm,t 2 ,zt 1 Covt Et 1 Rm,t 2 ,zt 1
Covt Et 1 Rm,t 2 , Dzt 1 4 0,
10
then the intertemporal risk price must be positive.

To see this point, assume without loss of generality
that the return on asset i is positively correlated with the
(innovation in the) state variable:
Covt Ri,t 1 ,zt 1 Covt Ri,t 1 , Dzt 1 4 0:
11
These two conditions imply that the return on asset i is

also positively correlated with the future expected market
return:
Covt Ri,t 1 ,Rm,t 2 Covt Ri,t 1 ,Et 1 Rm,t 2 40:
12
This last condition has an important economic content:

This asset does not provide a hedge for reinvestment risk,
and should earn a higher risk premium than an asset with
CovRi,t 1 ,Rm,t 2 0, that is, gz Covt Ri,t 1 , Dzt 1 4 0.
This in turn implies that gz 40, given the assumption
that Covt Ri,t 1 , Dzt 1 40. If we assume instead that the
state variable forecasts negative expected market returns,
then the intertemporal risk price must be negative, and
the argument is just symmetric.4
The main practical implication of this result is that if a
state variable positively forecasts expected returns, the
assets covariance with its innovation should earn a risk
premium in the cross-section. Thus, it is not enough that
the candidate state variables forecast future aggregate
returns. It must be the case that the (covariance) risk
prices with (the innovations in) those state variables have
the correct sign. Otherwise, the risk-based explanation
4
This positive correlation between the factor risk price and the
forecasting slope is also valid under the (more restrictive) Campbell
(1993) version of the ICAPM, as long as the coefcient of relative risk
aversion is greater than one, g 41. In the (implausible) case of an
investor less risk-averse than the log investor, the risk price of discount
rate news would be negative, that is, an asset that is positively
correlated with good news about future market returns (or alternatively,
an asset that is positively correlated with a state variable that forecasts
positive market returns) would earn a lower risk premium than an asset
that is uncorrelated with future market discount rates. The intuition is
that for an investor with very low risk aversion, the upside effect of a
positive correlation between a given asset and future market returns (in
the sense that it allows the investor to prot from the improvement in
future investment opportunities) outweighs the downside effect (a
reduced ability to hedge changes in future investment opportunities).
However, we follow the extant literature and assume that the representative investor is more risk-averse than the log investor, thus ruling
out this perverse effect.

590
associated with those factors is inconsistent with the

rational explanation underlying the ICAPM.
In the predictability literature, most predictive variables forecast positive expected equity market returns.
This is the case for aggregate nancial ratios (dividend-toprice ratio; earnings-to-price ratio; book-to-market ratio),
or bond yield spreads like the slope of the yield curve or
the default spread.
Next, we consider a state variable that forecasts the
future variance of market return, and reexamine the
corresponding implications for the sign of gz in the
cross-sectional asset pricing tests. The main result in this
case is the following: If the state variable is positively
correlated with the future volatility of aggregate returns,
Covt R2m,t 2 ,zt 1 Covt Et 1 R2m,t 2 ,zt 1
Covt Et 1 R2m,t 2 , Dzt 1 40,
13
then the risk price of intertemporal risk must be negative.5

To see this point, assume again without loss of generality that the return on asset i is positively correlated
with the (innovation in the) state variable, which implies
that the return on asset i is also positively correlated with
the future volatility of market return:
Covt Ri,t 1 ,R2m,t 2 Covt Ri,t 1 ,Et 1 R2m,t 2 4 0:
14
The economic implication of this last condition is that

such an asset provides a hedge for reinvestment risk, as it
pays high returns when future aggregate volatility is also
high. Thus, such an asset should earn a lower risk
premium than an asset with Covt Ri,t 1 ,R2m,t 2 0, that
is, gz Covt Ri,t 1 , Dzt 1 o0, which implies that gz o0,
given the assumption that Covt Ri,t 1 , Dzt 1 40. If we
assume instead that the state variable forecasts negative
market volatility, then the intertemporal risk price must
be positive, and the argument is just symmetric.
Thus, we have an opposite result relative to the case of
expected returns: If a state variable negatively forecasts
the volatility of returns, the assets covariance with its
innovation should earn a risk premium in the crosssection. The intuition is that an asset that covaries
negatively with future aggregate volatility does not provide a hedge for risk-averse investors (who dislike
increased uncertainty in their future wealth), who in turn
are willing to invest in such an asset if it offers a premium.
Because Rf ,t 1 is known at the beginning of the period,
we have Covt Rf ,t 1 ,Rm,t 1 Covt Rf ,t 1 , Dzt 1 0, leading to the pricing equation that we test on our data:
ERi,t 1 Rf ,t 1 g CovRi,t 1 Rf ,t 1 ,Rm,t 1
gz CovRi,t 1 Rf ,t 1 , Dzt 1 :
15
3. Do state variables forecast future investment

opportunities?
In this section, we test the rst restriction associated
with the Intertemporal CAPM (ICAPM), that is, do the
5
This proposition is consistent with the model developed by Chen
(2003).
candidates for state variables forecast future investment

opportunities? Second, and most important, we want to
assess the sign of the correlation between the state
variables and future aggregate returns, for comparison
with the factor risk prices we estimate, which represent
our main criteria to evaluate the ICAPM. Our proxy for the
investment opportunity set is the aggregate equity market, which is captured by the monthly return on the
value-weighted stock market index available from the
Chicago Center for Research in Security Prices (CRSP).
We conduct long-horizon predictive regressions,
which are commonly used in the predictability literature
to assess the forecasting power of the state variables over
future expected market returns (Keim and Stambaugh,
1986; Campbell, 1987; Fama and French, 1988, 1989,
among others):
r t,t q aq bq zt ut,t q ,
16
where r t,t q r t 1 r t q is the continuously compounded return over q periods (from t 1 to t q), and
ut,t q denotes a forecasting error with zero conditional
mean, Et ut,t q 0. It follows that the conditional
expected return at time t is given by Et r t,t q aq bq zt .
The sign of the slope coefcient, bq, indicates whether a
given state variable forecasts positive or negative changes
in future expected aggregate stock returns, and the
associated t-statistic indicates whether this effect is
statistically signicant. We use forecasting horizons of 1,
3, 12, 24, 36, 48, and 60 months ahead.6 The original
sample is 1963:072008:12, which corresponds to the
time span used in most empirical asset pricing studies of
the cross-section. We evaluate the statistical signicance
of the regression coefcients by using both Newey and
West (1987) and Hansen and Hodrick (1980) asymptotic
standard errors with q lags.7
The rst set of state variables we use in our empirical
test are variables from the predictability literature. The
rst two variables are bond yield spreads: The slope of the
Treasury yield curve (TERM) (Campbell, 1987; Fama and
French, 1989) and the corporate bond default spread
(DEF) (Keim and Stambaugh, 1986; Fama and French,
1989). We measure TERM as the yield spread between
the ten-year and the one-year Treasury bonds, while DEF
represents the yield spread between BAA and AAA corporate bonds from Moodys. The yield data are available
from the FRED (St. Louis Fed) database.
We also use the market dividend-to-price ratio (DY)
(Fama and French, 1988, 1989; Campbell and Shiller,
1988a), and the aggregate priceearnings ratio (PE)
(Campbell and Shiller, 1988b; Campbell and Vuolteenaho,
2004), both for the Standard and Poors (S&P) 500 index. DY is
6
There is a recent debate between in-sample versus out-of-sample
predictability of stock market returns (see Campbell and Thompson,
2008; Cochrane, 2008; Goyal and Welch, 2008, among others). In our
case, it makes sense to use in-sample regressions since we are interested
in the long-run predictive power of the state variables, and also to use
the same time-series that is used in the estimation of the factor
covariances (betas) and average excess returns, which are employed in
the cross-sectional tests.
7
We use q lags to correct for the serial correlation in the residuals
caused by the overlapping returns.

computed as the log ratio of the sum of annual dividends to

the level of the S&P 500 index, and PE corresponds to the log
ratio of the price of the S&P 500 index to a ten-year moving
average of earnings. The price, dividend, and earnings data for
the S&P 500 index are available from Robert Shillers
Web site.
The next two state variables are the one-month Treasury bill rate (RF) (Fama and Schwert, 1977; Campbell,
1991; Hodrick, 1992), which is available from Kenneth
Frenchs Web site, and the value spread (VS) (Campbell
and Vuolteenaho, 2004). The value spread is computed as
the difference between the monthly log book-to-market
ratios of small-value and small-growth stocks.8 The last
state variable is the Cochrane and Piazzesi (2005) factor
(CP), which is related to bond risk premia.9
Hahn and Lee (2006) use TERM and DEF in their ICAPM
application. Petkova (2006) uses DY and RF, in addition to
TERM and DEF, to proxy for changes in future investment
opportunities. Campbell and Vuolteenaho (2004) use PE
rather than DY and the value spread in addition to TERM
to forecast aggregate returns in their ICAPM application.
In Koijen, Lustig, and Van Nieuwerburgh (2010), the two
state variables are TERM and CP.10 Therefore, in addition
to single-variable forecasting regressions, we conduct
multiple-variable regressions to assess the joint forecasting power of the state variables in the four ICAPM
applications11:
r t,t q aq bq TERM t cq DEF t ut,t q ,
17
These three factors are obtained from Kenneth Frenchs Web

site. The fourth empirical factor we use is the liquidity factor
from Pastor and Stambaugh (2003) (L), which is obtained
from Lubos Pastors Web site.13
To obtain the associated state variables, we use the
cumulative sums on the factors for UMD and L. For
example, in the case of L, the cumulative sum is obtained
as
t
X
CLt
18
r t,t q aq bq TERM t cq PEt dq VSt ut,t q ,
19
r t,t q aq bq TERM t cq CP t ut,t q :
20
The second set of state variables is based on risk factors

widely used in the empirical asset pricing literature.
Although several of these multifactor models are justied
as applications of the ICAPM, it is not clear whether the
associated state variables do actually forecast future stock
returns. The rst two variables are the size (SMB) and value
(HML) factors used by Fama and French (1993, 1996).12 We
also use the momentum factor (UMD) from Carhart (1997).
Ls :
21
s t59
We use the cumulative sum over the last 60 months since

the total cumulative sum is close to being non-stationary
(autoregressive coefcients around one).
In the case of SMB, the corresponding state variable,
SMBn, is constructed as the difference between the
monthly market-to-book ratios of small and big stocks,
using the six portfolios sorted on both size and book-tomarket (BM), available from Kenneth Frenchs Web site:
SMBn
MBSL MBSM MBSH MBBL MBBM MBBH

,
3
3
22
where MBSL, MBSM, MBSH, MBBL, MBBM, and MBBH denote

the monthly market-to-book ratios of small-growth,
small-middle BM, small-value, big-growth, big-middle
BM, and big-value portfolios, respectively. Similarly, HMLn
corresponds to the difference between the monthly market-to-book ratios of value and growth stocks:
HMLn
r t,t q aq bq TERM t cq DEF t dq DY t eq RF t ut,t q ,
591
MBSH MBBH MBSL MBBL

:
2
2
23
Since SMB C DSMBn and HML C DHMLn , this procedure

enables us to interpret the original factors as innovations
in the state variables, which we use in the cross-sectional
regressions conducted later.14
Thus, we conduct multiple regressions corresponding
to three different multifactor models:
r t,t q aq bq SMBnt cq HMLnt ut,t q ,
24
r t,t q aq bq SMBnt cq HMLnt dq CUMDt ut,t q ,
25
r t,t q aq bq SMBnt cq HMLnt dq CLt ut,t q :
26
Also we estimate a predictive regression associated with the

augmented model estimated in Fama and French (1993):
r t,t q aq bq SMBnt cq HMLnt dq TERM t eq DEF t ut,t q :
The value spread is calculated from six portfolios sorted on both

size and book-to-market, available from Kenneth Frenchs Web site. For
details on the construction of VS, see the Appendix in Campbell and
Vuolteenaho (2004).
9
CP is the tted value from a regression of an average of excess
bond returns on forward rates. For details on the construction of CP, see
Cochrane and Piazzesi (2005). We thank an anonymous referee for
suggesting the inclusion of CP in the empirical analysis conducted in the
paper.
10
The proxy for the level of the yield curve in Koijen, Lustig, and
Van Nieuwerburgh (2010) is constructed differently than TERM, however, both proxies are highly correlated.
11
When there are multiple state variables, we should focus on the
marginal predictive role of each variable for changes in future investment opportunities, conditional on all other variables. This is why we
use multivariate forecasting regressions, instead of univariate regressions for each variable separately.
12
Vassalou (2003) provides evidence that both SMB and HML
convey information about future Gross Domestic Product (GDP) growth,
27
Table 2 presents summary statistics for the state
variables described above. Most state variables are highly
persistent, with autoregressive coefcients above 0.90.
The least persistent variable is CP with an autoregressive
coefcient of 0.85.
(footnote continued)
while Da and Schaumburg (2011) nd that these factors are related to
market volatility.
13
We use the non-traded liquidity factor (Pastor and Stambaugh,
2003, Eq. (8)).
14
The relation is only approximate, since we are ignoring the
dividend component of returns. The construction of SMBn and HMLn is
similar in spirit to the value spread computed by Campbell and
Vuolteenaho (2004).

592
Table 2
Descriptive statistics for state variables.
This table reports descriptive statistics for the state variables used in
the predictive regressions. The forecasting variables are the termstructure spread (TERM); default spread (DEF); market dividend yield
(DY); one-month Treasury bill rate (RF); market priceearnings ratio
(PE); value spread (VS); Cochrane-Piazzesi factor (CP); size premium
(SMBn); value premium (HMLn); momentum premium (CUMD); and
liquidity factor (CL). The sample is 1963:072008:12. f designates the
rst-order autocorrelation coefcient.
Variable
Mean
Stdev.
Min.
Max.
TERM
DEF
DY
RF
PE
VS
CP
SMBn
HMLn
CUMD
CL
0.009
0.010
3.580
0.005
2.871
1.559
0.010
0.122
2.837
0.508
0.033
0.011
0.005
0.413
0.002
0.448
0.159
0.016
0.250
0.981
0.199
0.582
0.031
0.003
4.495
0.000
1.893
1.201
0.052
1.033
6.445
0.132
1.496
0.033
0.034
2.801
0.014
3.789
2.222
0.070
0.780
0.895
1.081
0.904
0.967
0.992
0.996
0.955
0.997
0.938
0.845
0.924
0.966
0.961
0.991
Table 3 presents the results for the single long-horizon

regressions associated with TERM, DEF, DY, RF, PE, VS, and
CP. We can see that TERM, DEF, DY, RF, and CP forecast
positive market returns, although only in the cases of DY,
DEF, RF, and CP do the asymptotic t-stats indicate statistical signicance (and for DEF and RF, only at longer
horizons). On the other hand, PE consistently forecasts
negative market returns at all horizons, and this effect is
statistically signicant at horizons of and beyond three
months. The value spread also forecasts negative market
returns at all horizons, and the slopes are signicant at
the 5% level for horizons between three and 48 months.15
Basically, all the variables forecast positive market returns
with the exception of PE and VS and the forecasting power of
most variables increases with the horizon as indicated by the
approximately monotonic pattern in the R2 estimates.
The results for the multiple long-horizon regressions
(17)(20) are displayed in Table 4. To save space, we report
results only for horizons of 1, 12, and 60 months. The slope
estimates indicate that both TERM and DEF forecast positive
market returns, but only the slope associated with DEF is
statistically signicant at horizons of 12 and 60 months. In
the case of regression (18), at q60, for which there is
greater evidence of predictability as indicated by the adjusted
R2 estimates, TERM, DY, and RF are statistically signicant and
forecast positive market returns but the slope associated with
DEF is negatively estimated, although not signicant. In the
case of regression (19), TERM forecasts positive market
returns, while PE is negatively correlated with future
expected returns at all three horizons. VS forecasts negative
market returns at q 1,12, but the predictive slope is positive
at the 60-month horizon. The three forecasting coefcients
are statistically signicant at q60. Regarding the state
variables in regression (20), CP forecasts positive market
returns, conditional on TERM, and the slopes are signicant
15
The negative slopes associated with VS are in line with the results
obtained in Campbell and Vuolteenaho (2004).
at the 5% or 1% levels for q1 and q60. TERM forecasts

negative market returns for q1 and positive returns thereafter, but none of the slopes is signicant at the 10% level.
The results for the multiple long-horizon regressions
(24)(27) are displayed in Table 5. Both SMBn and HMLn
forecast positive market returns at all horizons; the slope
associated with HMLn is statistically signicant at all horizons, while SMBn is only marginally signicant at q60.
CUMD forecasts positive market returns at horizons of 12 and
60 months, conditional on both SMBn and HMLn, but the
slopes are statistically signicant only at very long horizons
(q 60). The liquidity factor is positively correlated with
future expected returns at all horizons, but the coefcients
are statistically signicant only at q12. In the fourth
multiple regression (27), we can see that the slopes associated with TERM and DEF, conditional on SMBn and HMLn,
are positive, although only TERM is marginally signicant in
predicting the market return for q 60 (based on the NeweyWest standard errors).16
Overall, these results show that most candidates for
ICAPM state variables can forecast market returns,
although the evidence of predictability is much stronger
at longer horizons.
4. Are factor risk prices consistent with the ICAPM?
To assess the ICAPM restrictions regarding factor risk
prices, we test different multifactor models in the crosssection of stock returns. In each model, the rst factor is
the market equity premium, RM, computed as the
monthly return of the value-weighted index in excess of
the one-month Treasury bill rate.
The rst group of multifactor models corresponds to
models explicitly justied as ICAPM applications in which
the risk factors represent innovations to state variables
used in the predictability of returns literature to forecast
aggregate equity returns. The rst model we estimate is
the ICAPM version of Hahn and Lee (2006), in which the
additional risk factors relative to the market factor are the
innovations in TERM and DEF:
ERi,t 1 Rf ,t 1 g CovRi,t 1 Rf ,t 1 ,RM t 1
gTERM CovRi,t 1 Rf ,t 1 , DTERM t 1
gDEF CovRi,t 1 Rf ,t 1 , DDEF t 1 :
28
The second model is the ICAPM of Petkova (2006) that

consists of the innovations to RF and DY, in addition to
TERM and DEF:
gDEF CovRi,t 1 Rf ,t 1 , DDEF t 1
gDY CovRi,t 1 Rf ,t 1 , DDY t 1
gRF CovRi,t 1 Rf ,t 1 , DRF t 1 :
29
The third model is an unrestricted version of the Campbell

and Vuolteenaho (2004) ICAPM, in which investment
16
We also conduct multiple predictive regressions for the equity
premium. In most cases, the signs of the slopes are the same as in the
regressions for the market return. In the few cases in which the sign
ips, the point estimates of the slopes are not statistically signicant.

593
Table 3
Single predictive regressions for ICAPM state variables.
This table reports the results for single long-horizon regressions for the monthly continuously compounded return on the value-weighted market
index, at horizons of 1, 3, 12, 24, 36, 48, and 60 months ahead. The forecasting variables are the current values of the term-structure spread (TERM);
default spread (DEF); market dividend yield (DY); one-month Treasury bill rate (RF); market priceearnings ratio (PE); value spread (VS); and the
Cochrane-Piazzesi factor (CP). The original sample is 1963:072008:12, and q observations are lost in each of the respective q-horizon regressions, for
q 1, 3, 12, 24, 36, 48, 60. For each regression, in line 1 are reported the slope estimates, and in lines 2 and 3 are reported Newey-West (in parentheses)
and Hansen-Hodrick t-ratios (in brackets) computed with q lags. Italic, underlined, and bold t-statistics denote statistical signicance at the 10%, 5%, and
1% levels, respectively. R2 (%) denotes the adjusted coefcient of determination (in %).
Predictor
q 1
q 3
q 12
q 24
q 36
q 48
q 60
TERM
0.16
(0.88)
[0.89]
0.16
0.37
(0.79)
[0.71]
0.26
1.78
(1.07)
[0.98]
1.56
2.28
(1.15)
[1.16]
1.45
3.65
(1.59)
[1.44]
2.71
5.43
(1.44)
[1.33]
4.83
7.20
(1.28)
[1.16]
6.29
0.67
(1.30)
[1.45]
0.43
2.04
(1.41)
[1.25]
1.12
7.30
(1.85)
[1.74]
3.74
7.89
(1.49)
[1.30]
2.49
11.58
(1.50)
[1.27]
4.00
17.96
(1.93)
[1.82]
7.71
27.82
(2.91)
[3.23]
14.23
0.01
(1.87)
[1.88]
0.77
0.03
2:27
2:01
2.23
0.11
2:32
1:96
8.18
0.19
2:23
2:00
13.02
0.25
2:50
[2.75]
17.25
0.33
(3.39)
[5.57]
21.83
0.43
(5.66)
[5.61]
27.55
0.65
(0.68)
[0.72]
0.11
1.88
(0.74)
[0.66]
0.27
6.78
(0.79)
[0.78]
0.89
18.87
2:28
2:27
3.91
25.55
2:10
2:00
5.33
30.54
2:04
[1.92]
5.99
41.95
2:13
2:13
8.16
0.01
( 1.63)
[ 1.63]
0.57
0.02
2:01
[ 1.78]
1.73
0.09
2:21
[ 1.89]
7.12
0.17
2:35
2:09
13.16
0.24
( 2.76)
[ 2.78]
19.49
0.32
( 3.83)
[ 4.75]
25.77
0.41
( 6.25)
[ 9.67]
32.42
0.02
( 1.41)
[ 1.41]
0.37
0.06
2:37
2:12
1.59
0.28
( 2.94)
[ 2.59]
8.09
0.39
2:13
[ 1.94]
8.69
0.51
2:07
2:14
10.80
0.44
( 1.85)
2:12
6.20
0.45
( 1.81)
[ 1.92]
4.98
0.35
(2.63)
[2.62]
1.47
0.84
2:51
2:25
2.61
1.84
2:00
[1.83]
3.24
2.34
(1.82)
[1.71]
2.88
3.59
2:13
2:07
4.98
5.47
(2.58)
2:44
9.29
6.39
2:08
[1.87]
9.78
R2 (%)
DEF
R2 (%)
DY
R2 (%)
RF
R2 (%)
PE
R2 (%)
VS
R2 (%)
CP
R2 (%)
opportunities are described by PE, TERM, and VS17:

gPE CovRi,t 1 Rf ,t 1 , DPEt 1
gVS CovRi,t 1 Rf ,t 1 , DVSt 1 :
30
The fourth model is the three-factor model from Koijen,

Lustig, and Van Nieuwerburgh (2010), in which the state
variables are TERM and CP18:
gCP CovRi,t 1 Rf ,t 1 , DCP t 1 :
31
17
Campbell and Vuolteenaho estimate a model with two factors,
cash ow news (NCF) and discount rate news (NDR). Both NCF and NDR are
linear functions of the innovations in the state variables, so that the two
specications are equivalent. For further details, see Campbell (1996)
and Maio (2012).
18
The Koijen, Lustig, and Van Nieuwerburgh (2010) model is not
motivated by the authors as an ICAPM application. However, since the
respective factors (other than the market factor) are related to state
variables widely used in the predictability literature (TERM and CP), it
makes sense to test whether this model satises the ICAPM restrictions.
Next, we use multifactor models with less theoretical

justication, but that some authors consider as possible
applications of the ICAPM. The fth model is the Fama and
French (1993, 1996) three-factor model (FF3), in which
the factors are SMB and HML in addition to the market risk
premium:
gSMB CovRi,t 1 Rf ,t 1 ,SMBt 1
gHML CovRi,t 1 Rf ,t 1 ,HMLt 1 :
32
SMB is used to explain the size premium, the positive

spread in average returns between small and big stocks.
HML seeks to explain the value premium, that value
stocks (stocks with high book-to-market) have larger
average returns than growth stocks (stocks with low
book-to-market).
The sixth model is the Fama-French three-factor model
augmented by TERM and DEF (FF5):
gHML CovRi,t 1 Rf ,t 1 ,HMLt 1

594
Table 4
Multiple predictive regressions for ICAPM state variables.
This table reports the results for multiple long-horizon regressions for the monthly continuously compounded return on the value-weighted market
index, at horizons of 1, 12, and 60 months ahead. The forecasting variables are the current values of the term-structure spread (TERM); default spread
(DEF); market dividend yield (DY); one-month Treasury bill rate (RF); market priceearnings ratio (PE); value spread (VS); and the Cochrane-Piazzesi
factor (CP). The original sample is 1963:072008:12, and q observations are lost in each of the respective q-horizon regressions, for q 1, 12, 60. For each
regression, in line 1 are reported the slope estimates, and in lines 2 and 3 are reported Newey-West (in parentheses) and Hansen-Hodrick t-ratios (in
brackets) computed with q lags. Italic, underlined, and bold t-statistics denote statistical signicance at the 10%, 5%, and 1% levels, respectively. R2 (%)
denotes the adjusted coefcient of determination (in %).
Row
TERM
DEF
0.12
(0.69)
[0.70]
0.21
(0.81)
[0.86]
0.21
(1.18)
[1.19]
0.10
( 0.48)
[ 0.50]
0.62
(1.23)
[1.35]
0.18
(0.23)
[0.27]
1.48
(0.88)
[0.81]
2.09
(1.13)
[0.98]
2.64
(1.82)
[1.70]
0.61
(0.29)
[0.26]
6.84
(1.76)
[1.70]
1.75
(0.39)
[0.35]
5.86
(1.10)
[0.96]
17.11
(2.75)
[2.76]
8.77
2:49
2:49
3.39
(0.68)
[0.63]
25.81
2:56
[2.78]
10.04
( 0.74)
[ 0.74]
DY
RF
PE
VS
CP
R2 (%)
Panel A: q 1
1
0.34
0.01
(1.43)
[1.43]
0.01
( 0.01)
[ 0.01]
0.55
0.01
( 1.20)
[ 1.22]
0.01
( 0.84)
[ 0.86]
0.56
0.38
2:49
2:54
1.34
Panel B: q 12
1
4.62
0.12
2:01
[1.64]
3.26
( 0.29)
[ 0.25]
10.85
0.06
( 1.37)
[ 1.22]
0.24
2:27
2:19
13.34
1.62
(1.29)
[1.15]
3.19
Panel C: q 60
1
18.15
0.37
(4.97)
[4.97]
62.35
2:06
2:06
45.40
0.48
( 7.74)
[ 7.71]
0.23
2:04
2:05
43.08
5.11
2:45
[3.14]
10.60
Fama and French (1993) use this model to explain both

equity and bond risk premiums.
The seventh model is Carharts (1997) four-factor
model (C):
the momentum anomaly; that is, past short-term winners

tend to have higher average returns than past losers
(Jegadeesh and Titman, 1993).
Finally, we use the four-factor model employed by
Pastor and Stambaugh (2003), which incorporates a
liquidity-related risk factor (L):
ERi,t 1 Rf ,t 1 g CovRi,t 1 Rf ,t 1 ,RMt 1
gDEF CovRi,t 1 Rf ,t 1 , DDEF t 1 :
33

gUMD CovRi,t 1 Rf ,t 1 ,UMDt 1 :
34
Added to the Fama and French (1993) model is the

momentum (UMD) factor. The role of UMD is to explain

gL CovRi,t 1 Rf ,t 1 ,Lt 1 ,
which is denoted as PS.
35

Table 5
Multiple predictive regressions for state variables constructed from
empirical factors.
This table reports the results for multiple long-horizon regressions for
the monthly continuously compounded return on the value-weighted
market index, at horizons of 1, 12, and 60 months ahead. The forecasting
variables are the current values of the term-structure spread (TERM);
default spread (DEF); size premium (SMBn); value premium (HMLn);
momentum factor (CUMD); and liquidity factor (CL). The original sample
is 1963:072008:12, and q observations are lost in each of the respective
q-horizon regressions, for q 1, 12, 60. For each regression, in line 1 are
reported the slope estimates, and in lines 2 and 3 are reported NeweyWest (in parentheses) and Hansen-Hodrick t-ratios (in brackets) computed with q lags. Italic, underlined, and bold t-statistics denote
statistical signicance at the 10%, 5%, and 1% levels, respectively. R2
(%) denotes the adjusted coefcient of determination (in %).
Row
SMBn
HMLn
CUMD
CL
TERM
DEF
R2 (%)
Panel A: q 1
1
0.00
(0.00)
[0.00]
0.00
(0.02)
[0.02]
0.01
( 0.58)
[ 0.61]
0.00
( 0.02)
[ 0.02]
0.00
1:98
[1.92]
0.00
(1.92)
[1.87]
0.01
2:50
2:53
0.00
(1.63)
[1.62]
0.65
0.00
( 0.42)
[ 0.44]
0.50
0.01
(1.52)
[1.58]
0.98
0.16
(0.90)
[0.91]
0.27
(0.47)
[0.53]
0.53
Panel B: q 12
1
0.01
(0.12)
[0.11]
0.01
(0.12)
[0.10]
0.05
( 0.64)
[ 0.56]
0.01
(0.14)
[0.12]
0.05
(3.04)
2:56
0.05
(3.01)
2:56
0.08
(4.02)
[3.70]
0.05
(2.70)
2:28
9.91
0.02
(0.22)
[0.20]
9.81
0.07
2:11
[1.88]
14.64
1.69
(0.41)
[0.40]
0.32
(1.71)
[1.77]
0.27
(1.69)
[1.69]
0.27
(1.85)
[1.85]
0.31
(1.76)
[1.79]
0.15
(3.27)
[3.82]
0.17
(3.67)
[3.67]
0.17
(5.12)
[5.08]
0.14
(4.89)
[6.49]
36.99
0.45
(2.97)
[2.99]
42.49
0.05
(0.90)
[0.90]
37.38
7.99
(1.72)
[1.43]
6.11
(0.60)
[0.57]
36
11.68
Panel C: q 60
1
empirical test, and allows us to combine the literature

on cross-sectional asset pricing with the literature on the
market risk-return trade-off. Thus, the estimates of risk
aversion and the intertemporal risk prices incorporate
information from both the cross-section of equity premiums and the aggregate equity premium.19 Moreover,
adding factors to the menu of testing assets when the
factors are returns themselves represents a more appropriate test of asset pricing models, as suggested by
Lewellen, Nagel, and Shanken (2010).
We estimate each multifactor model in expected
return-covariance form by a one-stage generalized
method of moments (GMM) procedure (Hansen, 1982).
This method uses equally weighted moments, which is
conceptually equivalent to running an ordinary least
squares (OLS) cross-sectional regression of average excess
returns on factor covariances (right-hand side variables).
The advantage of the GMM procedure is that we do not
need to have previous estimates of the individual covariances, as these are implied in the GMM moment conditions. This estimation allows us to assess whether each
model can explain the returns of a set of economically
interesting portfolios (e.g., SBM25).20
The GMM system includes N K 1 moment conditions. The rst N sample moments correspond to the
pricing errors for each of the N testing returns:
8
Ri,t 1 Rf ,t 1 gRi,t 1 Rf ,t 1 RMt 1 mm
>
>
>
>
>
g1 Ri,t 1 Rf ,t 1 f 1,t 1 m1
>
>
>
>
>

g2 Ri,t 1 Rf ,t 1 f 2,t 1 m2
>
>
>
<
T 1 >
X

gK Ri,t 1 Rf ,t 1 f K,t 1 mK
1
0,
g T b
RM
>
T t 0>
t 1 mm
>
>
>
> f 1,t 1 m1
>
>
>
>
>^
>
>
>
:f
K,t 1 mK
i 1, . . . ,N:
1.90
(1.24)
[1.16]
595
K denotes the number of factors in addition to the market

factor (that is, each model has K 1 factors).
In this system, the market return (RM t 1 ) is the rst
factor, with unconditional mean given by mm . The remaining factors are given by f 1,t 1 , . . . ,f K,t 1 , and the respective means are denoted by m1 , . . . , mK . For example, in the
case of the Fama-French model, we have K 2 with
f 1,t 1 SMBt 1 ,f 2,t 1 HMLt 1 . g1 , . . . , gK denote the
(covariance) risk prices associated with the hedging
factors. The last K 1 moment conditions in system (36)
enable us to estimate the factor means.
45.79
We use two sets of portfolio returns for testing assets.

The rst group is the 25 portfolios sorted on size and
book-to-market (SBM25). The second group is 25 portfolios sorted on size and momentum (SM25). The portfolio
return data are obtained from Kenneth Frenchs Web site.
We add the market return to each group of portfolios.
Adding the market return enables a more powerful
19
Note that it is common practice in the literature on the aggregate
risk-return trade-off to estimate only the risk-aversion parameter by
assuming (in opposition with the underlying theory of the ICAPM) that
the risk prices associated with time-varying investment opportunities
are negligible, i.e., that hedging motives are marginal. In our case, the
hedging factor risk prices are the core of the analysis.
20
The procedure is also more convenient than the two-pass timeseries/cross-sectional regressions approach (Cochrane, 2005, Brennan,
Wang, and Xia, 2004), since we want to estimate the model in expected
return-covariance form rather than in expected return-beta form to
obtain the covariance risk prices for each factor, and specically the
market (covariance) risk price, which represents an estimate of the
relative risk aversion.

596
Thus, the estimated covariance risk prices from the

rst N moment conditions do account for the estimation
error in the factor means, as in Cochrane (2005,
Chapter 13) and Yogo (2006). In this system, there are
NK1 overidentifying conditions (N K 1 moments
and 2(K 1) parameters to estimate). The standard errors
for the parameter estimates, and the remaining GMM
formulas are presented in Appendix A.
To assess the robustness of the asymptotic standard
errors, we conduct a bootstrap simulation to produce
empirical p-values for the tests of individual signicance
of the factor risk prices. The bootstrap simulation consists
of 10,000 replications in which the portfolio return data
and the factors are simulated independently, that is, the
data are simulated under the hypothesis that the model is
not true. Full details of this bootstrap simulation are
available in Appendix B.
Although our focus is on the analysis of the factor risk
prices, for completeness we also present some measures
for the overall explanatory power of each model. The idea
is to assess whether each model satises the economic
restrictions associated with the ICAPM, in addition to
explaining the dispersion in equity premiums over the
cross-section.
The rst measure is the mean absolute pricing error:
MAE
N
1X
b 9,
9a
Ni1 i
37
b i ,i 1, . . . ,N represents the rst N moments, i.e.,

where a
the pricing errors associated with the N testing assets.
The second goodness-of-t measure is the cross-sectional OLS coefcient of determination:
VarN a^ i
,
38
VarN R i
P
where R i 1=T T1
t 0 Ri,t 1 Rf ,t 1 denotes the average
excess return for asset i, and VarN stands for the crosssectional variance. R2OLS measures the fraction of the crosssectional variance in average excess returns explained by
the model.21
We start the empirical analysis by estimating twofactor models that include a hedging risk factor in
addition to the market return:
R2OLS 1
ERi,t 1 Rf ,t 1 g CovRi,t 1 Rf ,t 1 ,RMt 1

gz CovRi,t 1 Rf ,t 1 , Dzt 1 ,
39
where z TERM, DEF, DY, RF, PE, VS, and CP. The objective is
to analyze whether some of the most relevant predictors
of market returns proposed in the predictability literature
can be justied in two-factor ICAPM specications.
The factor risk price estimates are displayed in Table 6.
In the tests with SBM25 (Panel A), the point estimates for
the relative risk aversion (RRA) parameter are negative in
most specications; the exceptions are the models with
DDEF, DVS, and DCP. Moreover, the risk price estimates
associated with the intertemporal factor are negative in
21
We do not present the values for the asymptotic w2 test of
overidentifying restrictions, since both the mean absolute error (MAE)
and R2OLS represent more robust measures of the models global t.
the specications with DDEF, DDY, and DRF, which are

inconsistent with the positive forecasting slopes over the
market return associated with the level of these variables,
as shown in the last section. In the model with DPE, the
positive risk price estimate is also inconsistent with the
corresponding negative slope estimated in the single
predictive regression (16). Only in the specications with
DTERM, DVS, and DCP is there consistency in sign between
the factor risk price estimates from the cross-sectional
regressions and the slopes from the forecasting regressions over the market return. Overall, only two two-factor
models (including DVS or DCP) satisfy the restrictions on
the market (covariance) risk price (RRA parameter) and on
the hedging factor risk price.
In the tests with SM25 (Panel B), only in the specications with DRF and DVS is there an implausible negative
estimate for g. The factor risk prices associated with
DTERM, DDEF, DRF, and DCP are negative in all cases,
which is inconsistent with the corresponding positive
slopes estimated in the single predictive regressions.
Moreover, the risk price estimate for DVS is estimated
positively, which is inconsistent with the negative slopes
from the single forecasting regressions.
On the other hand, the signs of the risk price estimates
associated with DDY and DPE are consistent with the
forecasting ability of these variables over market returns,
although in both cases the risk price estimates are largely
insignicant. Furthermore, in both cases there is no
explanatory power over the cross-section of returns, as
illustrated by the negative estimates for the cross-sectional R2, indicating that these two models perform more
poorly than a model that predicts constant expected risk
premiums in the cross-section of equities.
The results in Table 6 show overall that most of the
state variables in the literature are not valid risk factors
under the ICAPM. Specically, only two two-factor models
satisfy the economic restrictions underlying the ICAPM,
jointly with a reasonable explanatory power over the
cross-section of excess stock returns (DVS and DCP in the
test with SBM25, with R2 38% and 42%, respectively).
We next estimate factor models presented in the literature as empirical applications of the ICAPM, which include
combinations of the state variables analyzed above, specically the three-factor Hahn and Lee (2006) model (henceforth, HL); the ve-factor Petkova (2006) model (P); a fourfactor model that corresponds to an unrestricted version of
Campbell and Vuolteenaho (2004) (CV); and the three-factor
model from Koijen, Lustig, and Van Nieuwerburgh (2010)
(KLVN). The results are displayed in Table 7.
In the tests with SBM25 (Panel A), we have the usual
result that the CAPM cannot price the SBM25 portfolios,
as shown by the negative cross-sectional R2 of 42%,
while the four multifactor models show a considerable
explanatory power over the dispersion in risk premiums
within these portfolios, with R2 estimates above 70%. The
point estimates for the RRA parameter, however, are
negative and statistically insignicant in all four factor
models, compared to a positive estimate of 2.79 in the
CAPM, which is statistically signicant.
In the case of HL, the estimates for gTERM and gDEF are
positive and negative, respectively, although only the risk

597
Table 6
Factor risk premiums for ICAPM state variables.
This table reports the estimation of the factor risk premiums from rst-stage GMM with equally weighted errors. The testing assets are the 25 size/
book-to-market portfolios (SBM25, Panel A) and 25 size/momentum portfolios (SM25, Panel B). g represents the risk price for the market factor. gTERM ,
gDEF , gDY , gRF , gPE , gVS , gCP represent the risk prices associated with the term-structure spread, default spread, market dividend yield, one-month Treasury
bill rate, market priceearnings ratio, value spread, and the Cochrane-Piazzesi factor, respectively. The rst line associated with each row presents the
covariance risk price estimates, and the second line reports the asymptotic GMM robust t-statistics (in parentheses). The column MAE % presents the
average absolute pricing error (in %). The column R2OLS denotes the OLS cross-sectional R2. The sample is 1963:072008:12. Italic, underlined, and bold
numbers denote statistical signicance at the 10%, 5%, and 1% levels, respectively.
Row
gTERM
gDEF
gDY
gRF
gPE
gVS
gCP
MAE %
R2OLS
0.11
0.72
0.22
0.38
0.17
0.01
0.12
0.43
0.18
0.04
0.17
0.38
0.14
0.42
0.27
0.43
0.31
0.08
0.34
0.09
0.31
0.05
0.34
0.08
0.32
0.04
0.34
0.09
Panel A: SBM25
1
2
3
4
5
6
7
1.86
( 0.61)
2.28
(1.76)
7.62
2:01
3.71
( 0.96)
7.66
( 1.87)
4.73
(3.96)
4.54
(1.62)
614.12
1:98
272.13
( 0.93)
17.13
( 2.72)
2675.52
( 1.52)
17.00
2:50
6.21
( 3.35)
227.39
2:33
Panel B: SM25
1
2
3
4
5
6
7
6.78
2:52
0.99
(0.43)
4.48
(1.74)
2.29
( 0.99)
4.89
(1.86)
0.16
( 0.12)
2.10
(1.59)
547.92
2:22
763.32
2:32
3.53
(0.79)
1947.25
2:32
price associated with DTERM is statistically signicant. Thus,

the sign of gDEF is inconsistent with evidence above that DEF
(conditional on TERM) forecasts positive market returns.
In the case of P, both gDY and gRF are negative, while
both gTERM and gDEF are positive. All four risk price
estimates are not statistically signicant, however, which
indicates evidence of multicollinearity. Thus, the signs of
gDEF , gDY , and gRF are inconsistent with the evidence from
the multiple regressions (18) at a horizon of 60 months
(for which there is stronger evidence of predictability)
that both DY and RF (conditional on the other variables)
forecast positive market returns, while DEF forecasts
negative market returns (although with no signicance).
In the case of CV, the estimate for gPE is positive, and
thus inconsistent with the corresponding negative slopes
found in the predictive regressions (19), while gVS is
estimated negatively, which also goes against the sign of
the predictive slope at q 60. Both gPE and gVS are not
statistically signicant at the 10% level, suggesting the
presence of multicollinearity.
Regarding the KLVN model, the point estimates for
gTERM and gCP are both positive, and thus consistent with
4.15
( 0.92)
8.14
(2.87)
26.81
( 0.52)
the corresponding slopes in the multiple regression (20)

at q 12,60. However, the point estimate for gCP is largely
insignicant based on both the asymptotic t-stat and
empirical p-value.
When the testing portfolios are SM25 (Panel B), the
explanatory power of the CAPM for the cross-section of
excess returns is also negative (R2 10%), and the corresponding RRA estimate is 2.34, which is signicant at the 5%
level. The four multifactor models are signicantly better
than the CAPM in pricing the SM25 portfolios, with crosssectional R2 estimates varying between 50% (HL) and 67% (P).
Contrary to the tests with SBM25, the point estimates for g
are positive for the HL, P, and KLVN models, but only in the
cases of HL and KLVN are there estimates above one (5.56
and 10.77, respectively), which are signicant (only marginally in the case of HL, based on the asymptotic t-stats). For
risk price estimates associated with the intertemporal factor,
there are inconsistencies with the slopes from the corresponding predictive regressions in all four models.
In the case of HL, the point estimates for both gTERM and
gDEF are negative, which goes against the positive slopes
estimated from the multiple long-horizon regressions. In

598
Table 7
Factor risk premiums for ICAPM specications.
gDEF , gDY , gRF , gPE , gVS , gCP represent the risk prices associated with the term-structure spread, default spread, market dividend yield, one-month Treasury
bill rate, market priceearnings ratio, value spread, and the Cochrane-Piazzesi factor, respectively. The models are the baseline CAPM (CAPM) and the
multifactor models from Hahn and Lee (2006) (HL), Petkova (2006) (P), Campbell and Vuolteenaho (2004) (CV), and Koijen, Lustig, and Van Nieuwerburgh
(2010) (KLVN). The rst line associated with each model presents the covariance risk price estimates, the second line reports the asymptotic GMM robust
t-statistics (in parentheses), and the third line shows empirical p-values from a bootstrap simulation (in brackets). The column MAE % presents the
average absolute pricing error (in %). The column R2OLS denotes the OLS cross-sectional R2. The sample is 1963:072008:12. Italic, underlined, and bold
numbers denote statistical signicance at the 10%, 5%, and 1% levels, respectively.
Model
gTERM
gDEF
gDY
gRF
gPE
gVS
gCP
MAE %
R2OLS
0.23
0.42
0.10
0.74
0.09
0.77
0.08
0.78
0.08
0.77
0.34
0.10
0.25
0.50
0.20
0.67
0.21
0.60
0.21
0.62
Panel A: SBM25
CAPM
HL
CV
KLVN
2.79
2:52
[0.03]
2.15
( 0.85)
[0.58]
4.66
( 0.56)
[0.66]
4.83
( 0.85)
[0.52]
0.28
( 0.08)
[0.96]
608.84
2:45
[0.04]
436.86
(1.52)
[0.20]
380.65
(2.66)
[0.02]
485.40
(1.69)
[0.21]
173.10
( 0.30)
[0.85]
234.60
(0.57)
[0.66]
3.35
( 0.19)
[0.89]
1049.02
( 0.90)
[0.46]
8.88
(0.98)
[0.46]
2.29
( 0.72)
[0.60]
78.68
(0.76)
[0.61]
Panel B: SM25
CAPM
HL
CV
KLVN
2.34
2:16
[0.05]
5.56
(1.72)
[0.17]
0.51
(0.08)
[0.95]
0.90
( 0.09)
[0.94]
10.77
2:13
[0.08]
503.90
( 1.96)
[0.09]
614.81
2:16
[0.04]
882.29
2:12
[0.05]
778.67
2:35
[0.04]
490.50
( 0.98)
[0.46]
56.50
( 0.11)
[0.93]
2.29
( 0.17)
[0.89]
the case of P, the estimates for gTERM , gDY , and gRF are
negative; again, these estimates are inconsistent with the
positive correlation between each of the corresponding
state variables and future market return found in the
multiple predictive regressions. Thus, only for one factor
(DEF) is the sign of the risk price consistent with the sign
of the slope in the forecasting regressions, but the point
estimate is insignicant (t-stat 0.11). In the case of CV,
the risk price associated with DTERM is estimated negatively and is statistically signicant, which goes against
the positive slope from the long-horizon regressions, while
both gPE and gVS have the wrong sign as in the test with
SBM25. Finally, for KLVN, the negative risk price associated with DTERM, which is signicant at the 5% level, is
at odds with the positive slopes (for longer horizons) of
TERM in the multiple predictive regressions.
If we compare the hedging factor risk prices with the
slopes from the single forecasting regressions in Table 3,
2242.96
( 1.61)
[0.15]
19.58
(1.22)
[0.34]
5.37
( 0.77)
[0.53]
241.79
(1.53)
[0.23]
instead of the multiple regressions, then the same qualitative results hold. That is, in each of the four models
tested in both sets of portfolios, there is inconsistency in
sign with the forecasting slopes for at least one state
variable in the model. The only exception is the KLVN
model tested on SBM25, which satises the sign consistency (as in the comparison with the multiple regressions). However, this model generates a negative estimate
for the risk-aversion coefcient as discussed above.
Overall, these results show that these four multifactor
models are inconsistent with the ICAPM in pricing either the
SBM25 or SM25 portfolios, despite signicant explanatory
power over the cross-section. This inconsistency shows up
in both the risk price estimates associated with the hedging
risk factors and the risk-aversion estimates. Hence, these
models can serve as good empirical models that explain the
size, value, and momentum anomalies, but there is no
underlying justication in line with the ICAPM theory.

599
Table 8
Factor risk premiums for empirical risk factors.
gDEF , gSMB , gHML , gUMD , gL represent the risk prices associated with the term-structure spread, default spread, size factor, value factor, momentum factor,
and liquidity factor, respectively. The multifactor models are Fama and French (1993) (FF3), Carhart (1997) (C), Pastor and Stambaugh (2003) (PS), and
the Fama and French (1993) ve-factor model (FF5). The rst line associated with each model presents the covariance risk price estimates, the second
line reports the asymptotic GMM robust t-statistics (in parentheses), and the third line shows empirical p-values from a bootstrap simulation (in
brackets). The column MAE % presents the average absolute pricing error (in %). The column R2OLS denotes the OLS cross-sectional R2. The sample is
1963:072008:12. Italic, underlined, and bold numbers denote statistical signicance at the 10%, 5%, and 1% levels, respectively.
gSMB
gHML
3.31
(2.69)
[0.02]
7.14
(3.38)
[0.00]
0.72
( 0.33)
[0.81]
0.32
( 0.12)
[0.92]
2.84
(1.87)
[0.13]
2.59
(0.94)
[0.47]
1.88
(0.90)
[0.49]
1.57
(0.61)
[0.63]
8.67
(5.49)
[0.00]
15.37
(4.45)
[0.00]
7.56
(3.81)
[0.00]
3.27
(0.91)
[0.47]
0.25
(0.19)
[0.90]
5.00
(3.10)
[0.00]
21.17
2:10
[0.06]
10.58
(1.81)
[0.11]
2.76
(1.78)
[0.15]
2.08
(1.14)
[0.36]
2.63
( 0.49)
[0.71]
6.16
(1.16)
[0.32]
4.62
( 1.58)
[0.22]
10.57
(3.36)
[0.00]
18.04
( 1.24)
[0.32]
16.41
(1.76)
[0.12]
Model
gUMD
gL
gTERM
gDEF
MAE %
R2OLS
0.10
0.69
0.09
0.78
0.10
0.73
0.09
0.76
0.32
0.01
0.13
0.84
0.31
0.19
0.16
0.73
Panel A: SBM25
FF3
PS
FF5
22.64
(3.23)
[0.00]
8.78
(1.86)
[0.10]
407.79
2:54
[0.02]
1.51
( 0.01)
[1.00]
Panel B: SM25
FF3
PS
FF5
7.24
(4.24)
[0.00]
Next, we estimate the four empirical-based multifactor

models. The results are displayed in Table 8, which is
similar to Table 7.
In the tests with SBM25, all four models have a
signicant explanatory power over the cross-section of
equity premiums, with cross-sectional R2 estimates
around 70%. On the other hand, in the tests with SM25,
only C and FF5 have relevant explanatory power over the
cross-section, with R2 estimates above 70%, thus conrming that the FF3 model cannot price the momentum
portfolios (Fama and French, 1996; Cochrane, 2007,
among others). The RRA estimates are positive and statistically signicant in the cases of FF3 and C, but insignicantly negative in the cases of PS and FF5, when the
test portfolios are SBM25. In the tests with SM25, g is
implausibly below one in the FF3 model, and negative in
the liquidity model, while in the cases of C and FF5, the
RRA estimates are positive and statistically signicant (in
the case of FF5, only marginally and based on the
asymptotic t-stat).
For hedging risk factors, in the case of FF3 tested
with SBM25, the risk price estimates associated with
SMB and HML are both positive and statistically signicant
(in the case of SMB, only marginally and based on the
43.01
2:56
[0.01]
865.52
2:17
[0.04]
183.26
(0.25)
[0.83]
Newey-West t-stat), which is consistent with the positive

slopes estimated in the multiple predictive regressions
(24). In the tests with SM25, the point estimate for gHML is
negative, although not statistically signicant. In the case
of the C model, the three factor risk prices are estimated
positively in both tests, with HML and UMD signicantly
priced, while gSMB is not statistically signicant. Thus, all
three risk price estimates are consistent with the corresponding positive slopes from the long-horizon regressions in (25). In the case of the liquidity model, the
estimates associated with gL are positive in both tests,
which are compatible with the positive slopes estimated
in the long-horizon regression (26) at q60. Thus, the
restriction on the hedging risk prices is satised in the
test of this model with SBM25, but the associated RRA
estimate is negative as referred above.
Finally, in the case of FF5, the point estimates for
gTERM and gDEF are positive and negative, respectively, in
the test with SBM25, and these estimates ip signs in
the test with SM25. Only gTERM is statistically signicant,
however. This implies that the signs of gDEF (in the
tests with SBM25) and gTERM (in the tests with SM25)
are incompatible with the positive slopes associated
with TERM and DEF in the predictive regressions (27).

600
Therefore, only two multifactor models admit an ICAPM

interpretationFF3 tested with the SBM25 portfolios,
and C tested with both sets of equity portfolios. When
tested on the SM25 portfolios, the FF3 model is no longer
consistent with the ICAPM, besides having no explanatory power for the cross-section of equity premiums
(R2OLS 1%).
Overall, considering the 16 cross-sectional tests (eight
models times two sets of test assets), only in three cases
can the multifactor model be justied as an empirical
application of the ICAPM. As for pricing both sets of equity
portfolios, only the Carhart (1997) four-factor model is
able to satisfy the ICAPM restrictions. The Fama-French
model, on the other hand, satises the ICAPM criteria only
when pricing the size-value portfolios. Moreover, none of
the four multifactor models a priori more likely to be
consistent with the ICAPM (HL, P, CV, and KLVN), satisfy
the ICAPM restrictions.
5. Additional results
To assess the robustness of the results in Sections 3
and 4, we add an intercept to the estimation of the
multifactor models; estimate the models by second-stage
GMM; add bond risk premiums to the test assets; estimate the models in expected return-beta form; conduct a
bootstrap simulation for the predictive slopes; use different state variables associated with SMB and HML; and
employ a different measure of expected market return.
We also conduct additional robustness checks that are
available in an addendum to this paper: we estimate the
multifactor models by rst orthogonalizing the hedging
factors relative to the market factor; exclude the excess
market return from the menu of test assets; estimate the
models with alternative equity portfolios; and employ a
different measure of innovation in the state variables.
Overall, the results are qualitatively similar to the benchmark results.
On the other hand, the relative risk aversion (RRA)

estimates are negative in the cases of the CAPM and most
ICAPM specications, with the exceptions of P (although
even there, the magnitude is lower than one) and KLVN
(around eight), both cases in the test with SM25. The
negative estimates associated with the market risk price
are likely a consequence of multicollinearity related to the
fact that the market covariances (betas) do not vary
signicantly across the test portfolios (Fama and French,
1993; Jagannathan and Wang, 2007). Regarding the factor
risk prices associated with the innovations in the state
variables, we have the same signs as in the benchmark
test without intercept. The sole exception is for gVS , which
is now estimated positively in the tests with SM25.
Thus, we conclude that the four models are rejected
due to inconsistency between the factor risk prices and
the predictive slopes, added to the negative estimates for
the coefcient of risk aversion.
For the four empirical-based multifactor models, the
point estimates of the intercept are economically signicant for most models, varying between 1% and 3% per
month. The sole exception is C in the test with SM25, with
a point estimate close to zero. Moreover, these point
estimates are statistically signicant in the cases of FF3,
PS, and FF5 for both tests. Hence, this represents evidence
that these three models are potentially misspecied.
On the other hand, the RRA estimates are negative in
the cases of FF3, PS, and FF5 for both sets of portfolios.
Regarding the risk price estimates for the hedging
factor, all four models pass the consistency restriction
with the slopes of the times-series regression in tests with
SBM25, although the RRA estimates are clearly negative in
the cases of FF3, PS, and FF5, as referred above. In tests
with SM25, only C satises the sign restrictions on the
hedging factor risk prices.
Overall, then, only the C model produces both reasonable RRA estimates and intertemporal risk prices that are
consistent with slopes from the time-series regressions.
5.2. Estimating by efcient GMM
5.1. Including an intercept in the cross-sectional tests

As a robustness check, we estimate the multifactor
models by including an intercept in the pricing equation.
If a given model is specied correctly, the intercept should
be indistinguishable from zero. Moreover, the factor risk
price estimates and cross-sectional R2 should not be too
different from the corresponding estimates in the benchmark or restricted specication without intercept.
Results available on the addendum to this paper show
that in tests with either set of portfolios, the CAPM
intercept is strongly statistically signicant at around 1%
per month (approximately 12% per year), which is economically signicant. These estimates represent evidence
that the CAPM is misspecied when tested with SBM25
or SM25.
For the ICAPM candidates, in most cases the intercept
is not statistically signicant. The exceptions are HL and
CV in the tests with SM25 with excess zero beta rates of
around 1% and 2% per month, respectively, which are
signicant at the 5% and 1% levels.
We reestimate the multifactor models analyzed in the

previous section by second-stage or efcient GMM, which
allows us to obtain the most efcient estimates of the
parameters, i.e., estimates with the lowest standard
errors. Thus, in the second step the moments are
weighted according to the inverse of the spectral density
matrix, which is equivalent to a generalized least squares
(GLS) cross-sectional regression of average excess returns
on factor covariances.
We compute the weighted least squares (WLS) coefcient of determination, which is equal to
R2WLS 1
1
0
a^ Scn N a^
1
0c
n
RS
N R
40
n
c
where a^ is the vector of demeaned pricing errors; S
N
contains the diagonal elements of the block of the spectral
density matrix associated with the N pricing errors; and R
is the vector of demeaned (average) excess returns. The
WLS R2 assigns less weight to the noisier pricing errors,

i.e., the pricing errors with greater variance (see Ferson

and Harvey, 1999; Shanken and Zhou, 2007).22
Lewellen, Nagel, and Shanken (2010) justify the use of
efcient GMM (or, equivalently, GLS/WLS cross-sectional
regressions of average returns on factor betas/covariances) over rst-stage GMM (or, equivalently, OLS
cross-sectional regressions) with the argument that it is
harder to obtain a high cross-sectional GLS/WLS R2 than a
high OLS R2, especially when pricing the SBM25 portfolios. Furthermore, the GLS/WLS R2 represents a measure
of how close the factor-mimicking portfolios are to the
meanvariance frontier constructed from the test assets.
Yet, Cochrane (1996, 2005) argues that in the efcient
GMM estimation, one is pricing repackaged portfolios,
which often represent extreme linear combinations of the
original portfolios. Thus, not only does one lose the
economic content of the original portfolios (e.g., explaining the size, value, and momentum anomalies), but these
efcient portfolios may also be uninteresting for the
average investor because of high transactions costs and
potential risk.23
Results available on the papers addendum indicate
that the CAPM cannot price the repackaged portfolios as
illustrated by the negative estimates of the WLS crosssectional coefcient of determination. At the same time,
the ICAPM specications signicantly improve the t
associated with the CAPM when the test portfolios are
SBM25, with R2WLS estimates varying between 54% (P) and
71% (CV). On the other hand, in the tests with SM25, the
t is quite low, with the exception of P (R2WLS 0:38). Most
models are, however, inconsistent with the ICAPM
because at least one hedging risk price has the wrong
sign. The sole exception is the KLVN model in the test
with SBM25, which meets the criteria on both the market
and state variable risk prices. The second-stage point
estimates for g and gCP in the KLVN model are in any case
largely insignicant.
As in the case of the ICAPM applications, all the four
empirical multifactor models present large explanatory
ratios (above 65%) in the test with SBM25. Yet, in the test
with SM25, only the C model delivers a positive R2WLS
estimate (66%), with the remaining three models presenting negative explanatory ratios. In the test with SBM25,
the sign consistency of the hedging risk prices with the
predictive slopes and the plausibility of the risk-aversion
estimates are met by the four models. The point estimates
for the risk prices in the PS and FF5 models are, however,
not signicant in most cases, suggesting the presence of
multicollinearity.
22
The WLS R2 should be more robust than the GLS R2 (that is based
on the full SN matrix) since the inverse of the spectral density matrix is
potentially misspecied when the number of moment conditions is
relatively large and the size of the time-series is not very large (see
Shanken and Zhou, 2007).
23
An alternative estimation procedure is to use the second-moment
matrix of returns from the test assets as weighting matrix (Hansen and
Jagannathan, 1997), also used by Hodrick and Zhang (2001), Jacobs and
Wang (2004), and Kan and Robotti (2008), among others. As Cochrane
(2005) points out, often the second-moment matrix of returns is closer
to being singular than the spectral density matrix, implying that the
resulting portfolios are even more extreme.
601
As in the rst-stage estimation, C is consistent with the

ICAPM criteria when tested on the SM25 portfolios. In
contrast with the rst-stage results, the FF3 model satises the sign restrictions on both the market and hedging
risk prices. However, the explanatory power of FF3 over
the SM25 portfolios is around zero (R2WLS 1%). In the
case of PS, the RRA estimate is negative while the
explanatory ratio is also negative ( 37%). Only the C
model satises the ICAPM restrictions in both tests while
still having positive explanatory power over the transformed portfolios, while the other three models meet
these criteria in the test with the SBM25 portfolios.
5.3. Pricing bond risk premia

We add excess bond returns to the equity portfolios to
assess whether the factor risk prices change in a signicant way by forcing each model to price jointly the crosssection of stock and bond returns. This becomes a more
demanding test for each model since we are pricing
simultaneously equity and bond risk premiums. We add
to each equity portfolio group the excess returns on seven
Treasury bonds with maturities of 1, 2, 5, 7, 10, 20, and 30
years, available from CRSP. Thus, we have a total of 33 test
assets in each asset pricing test (SBM25 or SM25).
Regarding the ICAPM specications, results available
in the addendum show that in the test with SBM25, the
only changes relative to the benchmark test are that gDY
becomes positive and the estimate for g in the KLVN
model is now positive (but smaller than one). Yet, both
estimates are largely insignicant. When the equity portfolios are SM25, there are a number of changes to the
equity-only test. Specically, the point estimates for gDEF ,
gDY , and gVS are now positive, while gCP is estimated
negatively. Moreover, the RRA parameter in the CV model
is now positive (3.47). All the point estimates for these
risk prices are, however, not signicant at the 10% level.
By comparing the risk price estimates with the corresponding slopes in the predictive regressions, in only one
case (KLVN tested on SBM25) is there consistency in signs.
However, the corresponding RRA estimate is implausibly
below one (0.51). In sum, all four models continue to be
inconsistent with the ICAPM, when we add bond risk
premiums to the asset pricing tests.
In the case of the empirical factor models tested on
SBM25, the point estimates for g in the PS and FF5 models
are now positive (4.13 and 2.08, respectively), and in the
rst model there is signicance at the 10% level. In these
two models, the risk price estimates for the liquidity and
DDEF factors ip signs relative to the benchmark test.
Thus, the PS model no longer meets the sign restriction in
the hedging factor risk prices, while FF5 meets these
criteria, however, both gL and gDEF are highly insignicant.
When the equity portfolios are SM25, the only change
relative to the benchmark test is that the risk price for
SMB becomes positive in the PS model. Yet, such a model
is still inconsistent with the ICAPM due to the negative
RRA estimate. Thus, when we add bond risk premiums to
the test with the size-momentum portfolios, it is still only
the C model that meets the ICAPM criteria.

602
Table 9
Beta factor risk premiums for ICAPM specications.
This table reports the estimation of the beta factor risk premiums from OLS cross-sectional regressions. The testing assets are the 25 size/book-tomarket portfolios (SBM25, Panel A) and 25 size/momentum portfolios (SM25, Panel B). lM represents the beta risk price for the market factor. lTERM , lDEF ,
lDY , lRF , lPE , lVS , lCP represent the beta risk prices associated with the term-structure spread, default spread, market dividend yield, one-month Treasury
bill rate, market priceearnings ratio, value spread, and the Cochrane-Piazzesi factor, respectively. The models are the baseline CAPM (CAPM) and the
multifactor models from Hahn and Lee (2006) (HL), Petkova (2006) (P), Campbell and Vuolteenaho (2004) (CV), and Koijen, Lustig, and Van Nieuwerburgh
(2010) (KLVN). The rst line associated with each model presents the beta risk price estimates (multiplied by 100), and the second line reports the
Shanken (1992) t-statistics (in parentheses). The column R2OLS denotes the OLS cross-sectional R2. The sample is 1963:072008:12. Italic, underlined, and
bold numbers denote statistical signicance at the 10%, 5%, and 1% levels, respectively.
Model
lM
lTERM
lDEF
lDY
lRF
lPE
lVS
lCP
R2OLS
Panel A: SBM25
CAPM
HL
P
CV
KLVN
0.55
(2.72)
0.44
2:01
0.41
2:06
0.41
2:07
0.41
1:97
0.42
0.50
2:43
0.47
(1.95)
0.32
(2.74)
0.39
(1.67)
0.02
(0.20)
0.06
(1.04)
0.74
0.35
( 0.30)
0.08
( 1.65)
0.77
0.74
(1.20)
0.96
( 1.44)
0.78
0.50
(0.63)
0.77
Panel B: SM25
CAPM
HL
P
CV
KLVN
0.46
2:29
0.57
(2.73)
0.52
2:48
0.52
2:38
0.55
2:20
0.09
0.45
2:51
0.34
( 1.28)
0.74
2:19
0.70
( 1.85)
0.10
( 1.59)
0.05
( 0.76)
0.50
0.95
( 1.03)
0.06
( 1.04)
0.67
2.04
(1.82)
0.82
( 0.47)
0.60
2.02
(1.37)
0.62
market beta risk prices24:
5.4. Estimating the models in expected return-beta

representation
g
We estimate the multifactor models in expected
return-beta form by using the time-series/cross-sectional
regression approach taken by Brennan, Wang, and Xia
(2004) and Cochrane (2005), among others. In the rst
step, we conduct time-series regressions to estimate the
factor loadings for each asset. For example, in the case of
FF3, we have
lM
VarRM t 1
43
42
The results for the ICAPM applications are displayed in

Table 9. The estimates for lM are positive in all four
multifactor models and in both tests. Thus, the implied
risk-aversion estimates are also positive in all cases, in
contrast to the benchmark test with GMM. Specically,
the implied estimates for g (not tabulated) vary between
2.07 and 2.23 in the tests with SBM25, and between 2.63
and 2.88 in the tests with SM25. For the hedging beta risk
prices, the signs are nearly the same as for the corresponding covariance risk prices in the benchmark
test with rst-stage GMM. The sole exception is HL
tested with SBM25, in which lDEF is now positive,
in contrast to the negative estimate for gDEF in the
benchmark test, which is in line with the positive slope
in the multiple forecasting regressions. Thus, in the test
with SBM25, HL satises the restrictions on the signs of
We use the same approach for the other multifactor

models. To compute the t-statistics associated with the
risk prices, we use the Shanken (1992) approach, which
accounts for the estimation error in the betas. The implied
relative risk aversion estimates can be obtained from the
24
This relation is only true when the cross-sectional regression is
estimated with single-regression betas instead of multiple-regression
betas. However, since in most of the multifactor models tested in the
paper the factors are only weakly correlated, this equation represents a
good approximation.
Ri,t 1 Rf ,t 1 di bi,M RM t 1 bi,SMB SMBt 1

bi,HML HMLt 1 ei,t ,
41
where the coefcients bi stand for the factor loadings. In

the second step, we conduct an OLS cross-sectional
regression of average excess returns on the factor loadings
to obtain estimates for the factor (beta) risk prices (l:
Ri Rf lM bi,M lSMB bi,SMB lHML bi,HML ai :

603
restrictions associated with the pricing equation for the

market return when pricing the other assets. Put differently, the GLS regression assigns a zero pricing error for
the excess market return and the market risk price is
numerically equal to the average excess market return
(0.38% per month in this sample).
Results available in the addendum show that in most
cases, the signs of the hedging factor risk prices are the
same as in the corresponding OLS cross-sectional regressions. Among the exceptions are the point estimates for
lDEF in the tests of HL and P with SM25, which are now
positive, although not statistically signicant. The risk
prices for HML are also now positive in the case of both
FF3 and PS tested on SM25 (contrary to the OLS regressions), and these point estimates are signicant at the 5%
and 10% levels, respectively. Thus, contrary to the OLS
cross-sectional regression, the FF3 and PS models are
consistent with the ICAPM when tested on the SM25
portfolios using GLS. However, the point estimate for lL is
not signicant at the 10% level.
the intertemporal risk prices, lTERM and lDEF , but lDEF is

insignicant (t-ratio 0.20). The KLVN model also satises
the ICAPM criteria on the risk prices (as in the secondstage GMM test) but the point estimate of lCP is largely
insignicant.
The results for the empirical models are displayed in
Table 10. As in the case of models HL, P, CV, and KLVN, the
market risk prices are positive for all four models. The
(untabulated) implied RRA estimates vary between 1.72
and 2.18 in the test with SBM25; in the test with SM25,
they range between 1.77 and 2.28. The signs of the
hedging risk prices are in most cases the same as the
covariance risk prices in the benchmark GMM tests. The
exceptions are lDEF , which changes its sign relative to gDEF
in the estimation with either SBM25 or SM25, and lSMB ,
which is estimated positively in PS, in the test with SM25.
Consequently, in addition to FF3 (in the test with
SBM25) and C (for both sets of portfolios), which satisfy
the risk price restrictions in the GMM benchmark test, the
risk price estimates for PS and FF5 in the test with SBM25
are also in line with the ICAPM criteria. The point estimate
for lDEF is strongly insignicant, however (t-ratio 0.66).
Overall, the qualitative results are very similar to the
benchmark test with rst-stage GMM.
We also estimate a GLS cross-sectional regression in
which the observations are weighted according to the
inverse of the covariance matrix of the residuals from the
time-series regressions (see Cochrane, 2005, Chapter 12).
When one of the factors is also a testing asset (as is the
case of the excess market return in all of the models
tested), this method allows us to incorporate fully the
5.5. Bootstrap simulation in the predictive regressions

We conduct a bootstrap simulation that produces an
empirical distribution that better approximates the nite
sample distribution of the coefcient estimates in the
predictive regressions in Section 3. In this simulation, the
market return and forecasting variables are simulated
(10,000 times) under the null of no predictability of the
market return and assuming that the predictor, zt, follows
Table 10
Beta factor risk premiums for empirical risk factors.
This table reports the estimation of the beta factor risk premiums from OLS cross-sectional regressions. The testing assets are the 25 size/book-tomarket portfolios (SBM25, Panel A) and 25 size/momentum portfolios (SM25, Panel B). lM represents the beta risk price for the market factor. lTERM , lDEF ,
lSMB , lHML , lUMD , lL represent the beta risk prices associated with the term-structure spread, default spread, size factor, value factor, momentum factor,
and liquidity factor, respectively. The multifactor models are Fama and French (1993) (FF3), Carhart (1997) (C), Pastor and Stambaugh (2003) (PS), and
the Fama and French (1993) ve-factor model (FF5). The rst line associated with each model presents the beta risk price estimates (multiplied by 100),
and the second line reports the Shanken (1992) t-statistics (in parentheses). The column R2OLS denotes the OLS cross-sectional R2. The sample is 1963:07
2008:12. Italic, underlined, and bold numbers denote statistical signicance at the 10%, 5%, and 1% levels, respectively.
Model
lM
lSMB
lHML
0.35
(1.82)
0.43
2:25
0.34
(1.75)
0.39
2:00
0.22
(1.58)
0.22
(1.58)
0.25
(1.78)
0.21
(1.51)
0.49
(3.86)
0.52
(4.02)
0.45
(3.55)
0.44
(3.39)
0.39
2:04
0.45
2:34
0.35
(1.68)
0.35
(1.65)
0.40
(2.69)
0.18
(1.20)
0.59
2:50
0.32
(1.61)
0.46
2:12
0.47
2:41
1.23
( 1.85)
0.28
(0.66)
lUMD
lL
lTERM
lDEF
R2OLS
Panel A: SBM25
FF3
C
PS
FF5
0.69
3.34
(3.47)
0.78
2.72
2:18
0.73
0.35
(2.81)
0.03
(0.66)
0.76
Panel B: SM25
FF3
C
PS
FF5
0.01
0.94
(5.38)
0.84
12.47
(2.80)
0.19
0.70
2:39
0.04
( 0.50)
0.73

604
r t,t q aq ut,t q ,
44
crucially on the way we construct the proxies for SMB and

HML in the predictive regressions.
zt 1 c fzt et 1 :
45
5.7. Alternative proxy for the expected market return
an AR(1) process:
This bootstrap procedure allows for the high persistence

of the forecasting variable and the cross-correlation
between the two residuals above, correcting for the
Stambaugh (1999) bias. Details of the bootstrap algorithm
are provided in Appendix C.
Results in the addendum show that all the slopes in
the predictive regressions (17)(20) are signicant at the
5% level for the 60-month horizon. In contrast, at the onemonth horizon, most of the coefcients are non-signicant at the 10% level according to the simulated p-values.
The sole exception is the coefcient associated with CP.
For the predictive regressions (24)(27), the results are
qualitatively similar. At the 60-month horizon, all the
slopes are signicant at the 1% level, with the exception of
DEF (p-value0.19) and CL (p-value0.22), while at the
one-month horizon, all the estimates are largely insignificant, with p-values well above 10%. These results complement the asymptotic t-statistics in that the point
estimates for the predictive slopes are more reliable at
long horizons (60 months).
5.6. Using different proxies for the state variables associated
with SMB and HML
We conduct the multiple long-horizon regressions by
using different proxies for the state variables associated
with SMB and HML. As in the proxy for CL, we construct
CSMB as
CSMBt
t
X
SMBs ,
46
s t59
and CHML is dened in an analogous way. Results available on the papers addendum show that both CSMB and
CHML forecast positive market returns at all horizons, but
both coefcients are not signicant at the 10% level in
most cases. The sole exception is the slope for CHML,
which is signicant in the regression for q12. Conditional on both CSMB and CHML, CUMD is negatively
correlated with expected market returns at q1 and is
positively correlated at q 12, 60, but all the coefcients
are highly insignicant. Regarding the liquidity factor, it
forecasts positive market returns at q1,12 and negative
returns for q60, conditional on both CSMB and CHML.
However, all the predictive slopes are largely insignificant. Finally, both TERM and DEF forecast positive market
returns at all horizons, conditional on CSMB and CHML,
although only DEF is a signicant predictor (at q 12,60).
When we compare the predictive slopes from the
regressions at q 60 (for which there is greater evidence
of predictability) with the factor risk price estimates in
the benchmark GMM test, only FF3 (in the test with
SBM25) and C (both tests) satisfy the sign restrictions.
However, there is no evidence that the state variables
associated with SMB and UMD forecast market returns at
any horizon. Hence, the results in the benchmark case
that the FF3 and C models satisfy the ICAPM criteria rely
There is a long literature discussing the biases and low

statistical power of predictive regressions of stock market
returns as those analyzed in Section 3.25 On the other
hand, Cochrane (2008) argues that predictive regressions
for stock returns have greater statistical power at very
long horizons and that the current log market dividend
yield (DY) is a very good proxy for expected long-run
market returns. By using a variance decomposition for DY,
Cochrane shows that nearly all the variation in DY is due
to the predictability of long-run returns by the dividend
yield. In other words, the long-run return coefcient of
expected returns on DY is close to one. Thus, we can use
DY as a proxy for long-run expected market returns in our
analysis.
We conduct the following contemporaneous multiple
regressions of DY on other state variables:
DY t Z0 Z1 TERM t Z2 DEF t vt ,
47
DY t Z0 Z1 TERM t Z2 PEt Z3 VSt vt ,
48
DY t Z0 Z1 TERM t Z2 CP t vt :
49
Results available in the addendum show that in all

three regressions, the coefcients associated with TERM
are negative. Conditional on TERM, DEF is positively
correlated with DY, while conditional on TERM, the slopes
for PE and VS are negative and positive, respectively.
Finally, conditional on TERM, CP forecasts positive market
returns. Based on heteroskedasticity-robust standard
errors (White, 1980), all the coefcient estimates are
signicant at the 1% level, with the sole exception of VS
which is not signicant at the 10% level. The implication
of these results is that in most cases, the consistency in
sign with the cross-sectional risk prices is not satised.
The exception is for KLVN tested on SM25, which now
meets all the ICAPM criteria.
We also conduct regressions of DY on the state variables associated with the empirical factors:
DY t Z0 Z1 SMBnt Z2 HMLnt vt ,
50
DY t Z0 Z1 SMBnt Z2 HMLnt Z3 CUMDt vt ,
51
DY t Z0 Z1 SMBnt Z2 HMLnt Z3 CLt vt ,
52
DY t Z0 Z1 SMBnt Z2 HMLnt Z3 TERM t Z4 DEF t vt :

53
The results reported in the papers addendum show
that the signs of the coefcients are basically the same as
the corresponding slopes in the forecasting regressions on
stock returns and most point estimates are strongly
signicant (1% or 5% levels). The sole exceptions are the
25
An incomplete list includes Nelson and Kim (1993), Stambaugh
(1999), Kilian (1999), Valkanov (2003), Lewellen (2004), Torous,
Valkanov, and Yan (2004), Campbell and Yogo (2006), Ang and Bekaert
(2007), and Boudoukh, Richardson, and Whitelaw (2008).

605
Table 11
Single predictive regressions for ICAPM state variables (SVAR).
This table reports the results for single long-horizon regressions for the stock market variance (SVAR), at horizons of 1, 3, 12, 24, 36, 48, and 60 months
ahead. The forecasting variables are the current values of the term-structure spread (TERM); default spread (DEF); market dividend yield (DY); one-month
Treasury bill rate (RF); market priceearnings ratio (PE); value spread (VS); and the Cochrane-Piazzesi factor (CP). The original sample is 1963:07
2008:12, and q observations are lost in each of the respective q-horizon regressions, for q 1, 3, 12, 24, 36, 48, 60. For each regression, in line 1 are
reported the slope estimates, and in lines 2 and 3 are reported Newey-West (in parentheses) and Hansen-Hodrick t-ratios (in brackets) computed with q
lags. Italic, underlined, and bold t-statistics denote statistical signicance at the 10%, 5%, and 1% levels, respectively. R2 (%) denotes the adjusted
coefcient of determination (in %).
Predictor
q 1
q 3
q 12
q 24
q 36
q 48
q 60
TERM
0.02
(1.05)
[1.34]
0.32
0.03
(0.62)
[0.55]
0.16
0.16
( 0.89)
[ 0.73]
0.93
0.50
( 1.68)
[ 1.51]
3.90
0.48
( 1.06)
[ 0.92]
2.19
0.28
( 0.48)
[ 0.41]
0.55
0.00
(0.00)
[0.00]
0.00
0.18
1:99
[2.62]
3.78
0.31
(1.83)
[1.63]
1.90
0.31
(0.82)
[0.67]
0.49
0.07
(0.09)
[0.07]
0.01
0.33
( 0.23)
[ 0.19]
0.15
0.68
( 0.36)
[ 0.30]
0.47
0.73
( 0.30)
[ 0.26]
0.42
0.00
( 4.08)
[ 4.91]
1.58
0.00
( 3.85)
[ 3.41]
3.68
0.02
( 3.30)
[ 2.66]
14.79
0.03
( 2.80)
2:23
22.97
0.05
2:51
2:11
27.33
0.06
2:46
2:25
29.31
0.07
( 2.58)
2:51
30.73
0.17
( 1.38)
[ 1.83]
0.84
0.35
( 0.99)
[ 0.88]
0.67
0.10
(0.14)
[0.11]
0.02
0.90
(0.73)
[0.64]
0.50
0.77
(0.46)
[0.42]
0.23
0.35
(0.15)
[0.16]
0.03
0.60
( 0.22)
[ 0.24]
0.07
0.00
(2.70)
[3.29]
0.54
0.00
(2.83)
2:51
1.50
0.01
2:30
[1.83]
6.94
0.02
(1.89)
[1.48]
11.68
0.03
(1.64)
[1.32]
14.51
0.04
(1.50)
[1.26]
15.60
0.04
(1.44)
[1.24]
15.90
0.00
(4.01)
[4.81]
1.12
0.01
(4.11)
[3.63]
2.72
0.03
2:19
[1.77]
6.77
0.04
(1.25)
[1.02]
5.02
0.04
(0.83)
[0.73]
3.34
0.03
(0.50)
[0.49]
1.13
0.03
(0.59)
[0.61]
0.96
0.03
( 1.84)
2:24
1.02
0.07
( 1.66)
[ 1.47]
1.39
0.19
( 1.61)
[ 1.36]
2.62
0.04
( 0.15)
[ 0.13]
0.06
0.12
(0.31)
[0.28]
0.25
0.28
(0.68)
[0.59]
0.98
0.46
(1.16)
[1.00]
2.15
R2 (%)
DEF
R2 (%)
DY
R2 (%)
RF
R2 (%)
PE
R2 (%)
VS
R2 (%)
CP
R2 (%)
coefcients associated with TERM and CL, which now

become negative, although the slope for CL is highly
insignicant. This implies that the FF5 model when tested
on the SM25 portfolios (but not on SBM25) turns out to be
consistent with the ICAPM. In contrast, the PS model no
longer satises the hedging risk price criteria in the test
with SBM25.
6. Higher order moments of the investment
opportunity set
6.1. The ICAPM with volatility forecasts
We have observed that a given risk factor is compatible with the ICAPM framework if its associated state
variable forecasts changes in either the rst or the second
moments of market returns. We now analyze whether the
candidate risk factors we have discussed forecast the
volatility of market returns and whether the predictive
slopes are consistent with the factor risk prices from all
the various cross-sectional tests.
To see if each of the candidate state variables individually forecasts the future volatility of aggregate stock
returns, we conduct single long-horizon regressions on

the realized stock market variance (SVAR):
SVARt,t q aq bq zt vt,t q ,
54
where SVARt,t q SVARt 1 SVARt q is the cumulative sum of SVAR over q periods, and vt,t q represents a
forecasting error with zero conditional mean.26 From (54)
it follows that Et SVARt,t q aq bq zt . If bq 4 0 (bq o 0),
the state variable is associated with positive (negative)
changes in the future volatility of aggregate returns.
The results reported in Table 11 show that both PE and
VS forecast positive variability in future market returns at
all horizons, but the slopes are statistically signicant
only at short horizons (from one to 12 months). The
slopes associated with DY are signicantly negative at
all horizons. In the case of TERM and DEF, the slopes are
positive at short horizons and become negative for horizons greater than three months (TERM) or 24 months
(DEF). However, these estimates are not statistically
26
SVAR is computed as the sum of squared daily returns on the S&P
500 index (Guo, 2006b; Goyal and Welch, 2008). The data on SVAR are
obtained from Amit Goyals Web page.

606
signicant, with the exception of DEF at very short

horizons. In the cases of RF and CP, we have the opposite
pattern, with negative coefcients at short horizons and
positive slopes for longer horizons, but there is statistical
signicance only at very short horizons.
Thus, if we restrict ourselves to the predictive slopes at
shorter horizons (for which there is greater statistical
signicance), the forecasting coefcients in the regressions for SVAR have opposite signs relative to the single
predictive regressions for expected market returns when
the state variables are DY, RF, PE, VS, and CP. Since a
positive slope in the predictive regressions of the market
volatility corresponds to a negative risk price in the crosssectional regression, as shown in Section 2, the slopes
associated with TERM, DEF, DY, PE, VS, and CP are consistent with the corresponding factor risk prices.27 However, the risk price estimates associated with DDY and
DPE (test with SM25) and DDEF (test with SBM25) are
highly insignicant. When we account for the restriction
of a risk-aversion coefcient (RRA) above one, the twofactor models based on DEF, VS, and CP satisfy the ICAPM
criteria in the tests with SBM25, while the specications
based on TERM, DY, and PE meet these criteria in the tests
with SM25.
When we compare the slopes from the single forecasting regressions for the market variance with the hedging
risk prices associated with the HL, P, CV, and KLVN models
in Section 4, we nd that the HL and KLVN models now
meet the sign restrictions in the hedging risk prices when
tested on the SM25 (but not on SBM25) portfolios.
However, as referred above, TERM is not a signicant
predictor of the volatility of aggregate returns.
We also conduct multiple long-horizon regressions for
the market variance with the variables associated with
models HL, P, CV, and KLVN:
are SM25. However, the RRA estimate is relatively large

(around 11) in the case of KLVN and there is no evidence
that TERM (conditional on DEF) forecasts SVAR at the onemonth horizon. In the case of the predictive regressions at
q 12, there is consistency in the hedging risk prices for
the HL and KLVN models tested on SBM25, but these
specications do not satisfy the restrictions on the RRA
parameter as discussed in Section 4.
Next, we conduct multiple long-horizon regressions
for the variance of returns corresponding to the four
empirical-based multifactor models, FF3, C, PS, and FF5:
SVARt,t q aq bq SMBnt cq HMLnt ut,t q ,
59
SVARt,t q aq bq SMBnt cq HMLnt dq CUMDt ut,t q ,

60
SVARt,t q aq bq SMBnt cq HMLnt dq CLt ut,t q ,
61
SVARt,t q aq bq SMBnt cq HMLnt dq TERM t

eq DEF t ut,t q :
62
The slope estimates are displayed in Table 12, which is

similar to Table 4. We can see that most of the slopes in
the four long-horizon regressions ip signs relative to the
predictive regressions for expected returns discussed in
Section 3. If we focus on the results at q60 (for which
the evidence of predictability is stronger, as indicated by
the adjusted R2 estimates), all four models do not satisfy
the consistency criteria on the hedging risk prices, since
there is at least one state variable in each model with the
wrong sign. Focusing on the one-month forecasting
regressions for market volatility, the predictive slopes in
the HL and KLVN models satisfy the sign consistency with
the corresponding risk prices when the testing portfolios
The results are presented in Table 13, which is similar

to Table 5. We can see that both SMBn and HMLn are
negatively correlated with the future market variance,
although SMBn is statistically signicant (10% level) only
at q12. Conditional on both SMBn and HMLn, CUMD
forecasts positive market volatility, while CL is negatively
correlated with future SVAR at q1 and positively correlated at longer horizons. However, the slope estimates
associated with both CUMD and CL are not statistically
signicant at any horizon. Conditional on both SMBn and
HMLn, TERM is negatively correlated with SVAR at q12
and q60, but this effect is statistically signicant only
for q12. On the other hand, the slopes associated with
DEF are positive at the three horizons, but there is
statistical signicance only for q1 and q12.
When we compare these predictive slopes with the
corresponding risk price estimates in Table 8, it follows
that the FF3 and FF5 models satisfy the sign restrictions
on the hedging risk prices from the test with SBM25.
However, the RRA estimate in the FF5 model is negative.
Thus, only the FF3 model satises the ICAPM criteria in
the test with SBM25. When the comparison is made
against the risk price estimates from the test with
SM25, none of the four multifactor models meets the
criteria.
Thus, when future investment opportunities are driven
by market volatility, only the FF3 model meets the ICAPM
criteria among the four empirical factor models (and only
in the test with the SBM25 portfolios). This means that
none of the other three models can be justied on the
argument that the respective state variables forecast
changes in future stock market volatility. Specically,
the Carhart (1997) model can be justied as an ICAPM
application only when we consider future expected market return as the single dimension of future investment
opportunities.28
27
However, in the case of VS and CP there is consistency only in the
tests with SBM25, while in the case of TERM, DY, and PE the consistency
only holds for the tests with the SM25 portfolios.
28
We conduct predictive regressions by using alternative proxies
for the stock market return variance. Results available on the addendum
to this paper show that the nding that the FF3 model is consistent with
SVARt,t q aq bq TERM t cq DEF t ut,t q ,
55
SVARt,t q aq bq TERM t cq DEF t dq DY t eq RF t ut,t q ,

56
SVARt,t q aq bq TERM t cq PEt dq VSt ut,t q ,
57
SVARt,t q aq bq TERM t cq CP t ut,t q :
58

607
Table 12
Multiple predictive regressions for ICAPM state variables (SVAR).
This table reports the results for multiple long-horizon regressions for the stock market variance (SVAR), at horizons of 1, 12, and 60 months ahead. The
forecasting variables are the current values of the term-structure spread (TERM); default spread (DEF); market dividend yield (DY); one-month Treasury
bill rate (RF); market priceearnings ratio (PE); value spread (VS); and the Cochrane-Piazzesi factor (CP). The original sample is 1963:072008:12, and q
observations are lost in each of the respective q-horizon regressions, for q 1, 12, 60. For each regression, in line 1 are reported the slope estimates, and in
lines 2 and 3 are reported Newey-West (in parentheses) and Hansen-Hodrick t-ratios (in brackets) computed with q lags. Italic, underlined, and bold
t-statistics denote statistical signicance at the 10%, 5%, and 1% levels, respectively. R2 (%) denotes the adjusted coefcient of determination (in %).
Row
TERM
DEF
0.01
(0.69)
[0.85]
0.07
( 1.82)
2:34
0.01
(0.69)
[0.87]
0.05
(1.59)
2:10
0.18
2:03
[2.68]
0.39
2:48
[3.29]
0.18
( 1.02)
[ 0.84]
0.22
( 1.27)
[ 1.10]
0.25
( 1.53)
[ 1.24]
0.03
( 0.12)
[ 0.10]
0.36
(1.02)
[0.82]
1.37
(2.87)
[2.58]
DY
RF
PE
VS
CP
R2 (%)
Panel A: q 1
1
3.68
0.00
( 4.07)
[ 5.01]
0.48
( 1.37)
[ 1.80]
11.31
0.00
(1.09)
[1.31]
0.00
(2.83)
[3.37]
0.97
0.04
( 1.90)
2:45
2.30
Panel B: q 12
1
1.41
0.03
( 6.53)
[ 5.57]
1.78
(1.78)
[1.58]
29.73
0.01
(1.94)
[1.64]
0.02
(1.78)
[1.50]
11.19
0.18
( 1.23)
[ 1.07]
2.46
Panel C: q 60
1
0.04
(0.06)
[0.06]
0.49
(0.53)
[0.53]
0.10
( 0.12)
[ 0.12]
0.47
( 0.76)
[ 0.80]
0.74
( 0.32)
[ 0.28]
0.96
(0.51)
[0.51]
0.22
0.11
( 4.20)
[ 4.19]
We also conduct the predictive regressions for SVAR by

using the alternative state variable proxies, CSMB and
CHML, in place of SMBn and HMLn, respectively. In results
available in the addendum to this paper, in most cases the
predictive slopes have the same sign as in the benchmark
regressions (59)(62). The only exception is the regression for the FF5 model at q60, in which the coefcients
associated with CSMB, TERM, and DEF ip sign relative to
(footnote continued)
the corresponding state variables forecasting SVAR does not generalize
to the alternative volatility measures.
11.41
2:10
2:10
46.30
0.06
(1.54)
[1.55]
0.06
( 1.42)
[ 1.41]
18.43
0.63
1:97
1:97
2.78
the benchmark regression. However, these estimates are

highly insignicant.
6.2. Simultaneous variation in the rst two moments of
aggregate returns
In the above analysis we test separately the consistency between the cross-sectional risk prices and the
predictive slopes for the rst two moments of aggregate
returns. In principle, a state variable can forecast both the
mean and variance of market returns in such a way that
the two cancel out maintaining a constant conditional
Sharpe ratio with no predictive power for the investment
opportunity set. According to the ICAPM, this variable

608
Table 13
Multiple predictive regressions for state variables constructed from
empirical factors (SVAR).
This table reports the results for multiple long-horizon regressions for
the stock market variance (SVAR), at horizons of 1, 12, and 60 months
ahead. The forecasting variables are the current values of the termstructure spread (TERM); default spread (DEF); size premium (SMBn);
value premium (HMLn); momentum factor (CUMD); and liquidity factor
(CL). The original sample is 1963:072008:12, and q observations are
lost in each of the respective q-horizon regressions, for q 1, 12, 60. For
each regression, in line 1 are reported the slope estimates, and in lines 2
and 3 are reported Newey-West (in parentheses) and Hansen-Hodrick tratios (in brackets) computed with q lags. Italic, underlined, and bold tstatistics denote statistical signicance at the 10%, 5%, and 1% levels,
respectively. R2 (%) denotes the adjusted coefcient of determination
(in %).
Row
SMBn
HMLn
CUMD
CL
TERM
DEF
R2 (%)
r t,t q a1q b1q TERM t c1q DEF t ut,t q :
Panel A: q 1
1
0.00
( 0.93)
[ 1.10]
0.00
( 0.96)
[ 1.12]
0.00
( 0.66)
[ 0.80]
0.00
( 1.84)
2:09
0.00
( 1.72)
2:11
0.00
0.00
( 1.60) (0.93)
[ 1.94] [1.09]
0.00
0.00
( 1.44)
( 0.54)
[ 1.72]
[ 0.60]
0.00
( 2.63)
[ 3.43]
Then, we compute the tted conditional variance,

d t,t q from the regression:
SVAR
0.71
SVARt,t q a2q b2q TERM t c2q DEF t vt,t q :
0.58
0.00
(0.06)
[0.07]
0.27
(2.62)
[3.46]
0.01
( 1.93)
[ 1.69]
0.01
( 1.92)
[ 1.69]
0.01
2:39
2:32
0.02
( 2.62)
2:31
0.00
2:33
1:98
0.00
0.00
2:27 (0.19)
[ 1.94] [0.17]
0.00
( 1.34)
[ 1.14]
0.01
( 3.38)
[ 2.96]
7.08
11.27
11.35
0.27
1.33
( 1.95) (3.79)
[ 1.63] [3.22]
19.54
Panel C: q 60
1
0.01
( 0.36)
[ 0.41]
0.01
( 0.42)
[ 0.42]
0.02
( 1.05)
[ 1.13]
0.01
( 0.69)
[ 0.69]
0.02
2:42
2:27
0.02
0.01
2:15 (0.36)
2:16 [0.36]
0.02
2:08
2:20
0.03
( 2.71)
[ 2.72]
23.34
23.33
0.02
(1.02)
[1.02]
25.36
0.26
2.14
( 0.37) (1.22)
[ 0.37] [1.22]
64
The pseudo-conditional Sharpe ratio is then equal to30:

b TERM t b
b1q b
b
a
c 1q DEF t
r t,t q
1q
:
SRt,t q q
b
b
b
d t,t q
a 2q b 2q TERM t c 2q DEF t 1=2
SVAR
65
11.36
0.00
(0.31)
[0.25]
63
0.71
Panel B: q 12
1
for the net change in the investment opportunity set.29 A

rise in this ratio signals an improvement in future investment opportunities (better mean and/or lower variance).
Following the discussion in Section 2, if a given state
variable is positively (negatively) correlated with the
conditional Sharpe ratio, the corresponding factor risk
price should be positive (negative). In other words, the
signs of the slopes are interpreted in the same way as the
coefcients in the regressions for the market return.
We follow Whitelaw (1994) and Tang and Whitelaw
(2011) in estimating the conditional Sharpe ratio for each
model and at each forecasting horizon. To exemplify our
approach, consider the HL model. First, we compute the
conditional expected return as the tted value, b
r t,t q from
the predictive regression on the market return:
25.96
should not be priced in the cross-section although the

comparison between the factor risk price and each of the
predictive slopes would point to an inconsistency with
the ICAPM.
To address this issue we measure the impact of each
state variable on a conditional Sharpe ratio, which proxies
Finally, we compute the time-series average of the partial

derivatives of SRt,t q with respect to each state variable in
the factor model. In the case of TERM, this partial
derivative is equal to
@SRt,t q b d 1=2 1 d 3=2 b
r t,t q :
b 1q SVAR t,t q SVAR t,t q b 2q b
2
@TERM t
66
Results available on the papers addendum show that

for the eight factor models, most implied slopes on the
pseudo-Sharpe ratio have the same signs as the corresponding slopes from the predictive regressions for the
market return in Section 3 at q60. The sole exception is
the coefcient associated with DEF in the FF5 model
which is now negative. Thus, these results show that the
effect of the state variables on the expected aggregate
return dominate the impact on the market volatility. This
implies that the qualitative conclusions regarding the
factor models satisfying the ICAPM criteria are mostly
the same as in Section 4. The exception is the case of the
FF5 model estimated with SBM25, which satises the sign
restrictions on the hedging risk prices. Yet, the corresponding negative estimate for the risk-aversion parameter invalidates this model as a plausible ICAPM
application.
29
Under some assumptions (meanvariance objective function and
that the state variables follow Markov processes), Brennan and Xia
(2006) and Nielsen and Vassalou (2006) show that the slope and the
intercept of the capital market line are two sufcient statistics for all
ICAPM state variables.
30
To obviate the problem of tted negative conditional variance due
to the linear specication of the predictive regression, we censorize the
tted variance at the minimum sample value of SVAR, that is,
b TERM t b
d t,t q maxa
b 2q b
SVAR
c 2q DEF t ,minSVARt .
2q

7. Power and size

In this section, we conduct different Monte Carlo
simulations to assess the power and size of the ICAPM
tests in the previous sections. In the rst simulation, we
want to check if the ICAPM criteria are satised for
articial return and factor data constructed under the
assumption that the ICAPM does not hold.
For that, we simulate 10,000 pseudo-samples of an
articial state variable according to an AR(1) process:
zbt 1 c fzbt ebz,t 1 ,
b 1, . . . ,10,000,
67
where ebz,t 1 denotes a simulated zero-mean normally

distributed error. In the simulation of ebz,t 1 , we use the
sample variance of the residuals from the AR(1) process
for TERM. We calibrate the autoregressive coefcient at
two possible values, 0.95 and 0.99, which are consistent
with the estimated coefcients obtained for our state
variables in Table 2. We x the drift, c, at the sample
mean of TERM. For each pseudo-sample, the initial realib
zation of zt is constructed as
!
c
1
b
,
,
68
z1 N
1f 1f2
and the innovation in the state variable is dened as
Dzbt 1 zbt 1 zbt .
The excess returns on the test assets are simulated
according to the following equation:
Ri,t 1 Rf ,t 1 b mi ebi,t 1 ,
b 1, . . . ,10,000,i 1, . . . ,N,
69
where mi represents the expected excess return for asset i,

which is assumed to be constant over time, and ebi,t 1
denotes a simulated zero-mean normal error for asset i.
These errors are simulated from the sample covariance
matrix of the excess returns of the SBM25 portfolios and
the market portfolio (the Nth asset in our cross-sectional
tests).
Under this data-generating process for the non-ICAPM
world, we assume that the CAPM holds, that is, the
individual expected excess returns are given by
mi g CovRi,t 1 Rf ,t 1 ,RMt 1 ,
empirical distribution of the hedging risk price (gz ) and

associated t-statistic.
To check whether the restriction on the hedging risk
price is satised, we compute the fraction of pseudosamples in which the signs of the predictive slope and risk
price are coincident, requiring also that both coefcients
be signicant at the 5% level in those samples (we call this
fraction the ICAPM acceptance rate). The results in
Table 14 (last two rows) show that for both values of f,
these acceptance rates are virtually zero when the comparison is made against both the one-month and
60-month predictive slopes. These low fractions are the
result of low individual statistical signicance for both the
predictive coefcient (fractions below 7% in the onemonth predictive regressions) and the hedging risk price
estimates (1%). These results show that in a world in
which the ICAPM does not hold, the sign restrictions on
the intertemporal risk price are almost never satised.
Next, we conduct an alternative Monte Carlo simulation in which we assume that the ICAPM does hold. The
question we want to address is how likely it is that the
restrictions on the ICAPM criteria are found signicant in
this ICAPM world.
The rst difference relative to the previous data-generating process is that the errors for the state variable,
ebz,t 1 , and the excess returns, ebi,t 1 , are simulated from
the sample augmented covariance matrix of the state
Table 14
Monte Carlo simulation.
This table reports results for a Monte Carlo simulation with 10,000
replications of an articial two-factor ICAPM. f, g, and gz denote the
calibrated values for the autoregressive coefcient of the state variable,
risk-aversion coefcient, and hedging risk price, respectively. tb1
(tb60 ) denotes the fraction of replications in which the predictive slope
in the one-month (60-month) regression is statistically signicant at the
5% level, while tgz denotes the percentage of replications in which the
risk price for the hedging factor is statistically signicant. Signb1
(Signb60 ) denotes the fraction of replications for which the predictive
slope in the one-month (60-month) regression shares the same sign as
the hedging risk price, and both coefcients are signicant at the 5%
level. The last two rows are associated with a simulation in which the
ICAPM is not true.
Parameters
tb1
tb60
tgz
Signb1
Signb60
f 0:95, g 5, gz 600
f 0:99, g 5, gz 600
f 0:95, g 4, gz 600
f 0:99, g 4, gz 600
f 0:95, g 3, gz 600
f 0:99, g 3, gz 600
f 0:95, g 5, gz 1200
f 0:99, g 5, gz 1200
f 0:95, g 4, gz 1200
f 0:99, g 4, gz 1200
f 0:95, g 3, gz 1200
f 0:99, g 3, gz 1200
f 0:95, g 5, gz 300
f 0:99, g 5, gz 300
f 0:95, g 4, gz 300
f 0:99, g 4, gz 300
f 0:95, g 3, gz 300
f 0:99, g 3, gz 300
f 0:95, g 2:80
f 0:99, g 2:80
0.94
0.99
0.93
0.98
0.89
0.97
0.95
0.99
0.93
0.99
0.91
0.98
0.94
0.99
0.93
0.99
0.90
0.98
0.07
0.05
0.94
0.99
0.93
0.99
0.91
0.98
0.94
0.99
0.93
0.99
0.91
0.99
0.94
0.99
0.92
0.99
0.91
0.99
0.49
0.31
0.99
1.00
0.98
1.00
0.97
1.00
0.99
1.00
0.99
1.00
0.98
1.00
0.98
1.00
0.98
1.00
0.96
0.97
0.01
0.01
0.91
0.96
0.90
0.96
0.86
0.95
0.91
0.96
0.89
0.96
0.87
0.95
0.92
0.96
0.90
0.96
0.86
0.92
0.00
0.00
0.91
0.96
0.89
0.96
0.85
0.95
0.91
0.96
0.88
0.96
0.86
0.95
0.91
0.96
0.89
0.96
0.85
0.92
0.00
0.00
70
where CovRi,t 1 Rf ,t 1 ,RM t 1 denotes the covariances

estimated with the original sample, and g is xed at 2.80,
which represents the RRA estimate obtained in the test of
the CAPM with SBM25 for the original sample.
The log market return used in the predictive regressions, r bt 1 ln1 Rbm,t 1 , is obtained from the simulated
simple market return:
Rbm,t 1 Rm,t 1 Rf ,t 1 b R f ,t 1 ,
609
71
where R f ,t 1 denotes the sample mean of the riskfree rate.

Armed with the simulated data, we run single predictive regressions (16) at the one-month and 60-month
horizons and obtain an empirical distribution of the
predictive slopes (b1 ,b60 ) and associated asymptotic
t-statistics. Similarly, we estimate the two-factor ICAPM
pricing Eq. (39) with the pseudo-data and obtain an

610
variable innovations and excess returns, that is, we allow

for the contemporaneous correlation between the state
variable and the test assets, including the market equity
premium.
The second major difference is that we allow expected
returns to be time-varying by assuming that the conditional covariances with the market factor are linear in the
lagged state variable:
Covt Ri,t 1 Rf ,t 1 ,RM t 1 ai bi zt ,
i 1, . . . ,N,
72
where the coefcients ai and bi are estimated from the

following time-series regressions for the original sample:
Ri,t 1 Rf ,t 1 RM t 1 ai bi TERM t ni,t 1 ,
i 1, . . . ,N:
73
Thus, the pseudo-return-generating process is given by

Ri,t 1 Rf ,t 1 b mbit ebi,t 1 ,
b 1, . . . ,10,000,i 1, . . . ,N,
74
mbit gai gbi zbt gz CovRi,t 1 Rf ,t 1 , Dzt 1 ,
75
where the covariances with the innovation in the

state variable, CovRi,t 1 Rf ,t 1 , Dzt 1 , are estimated
from the original sample for TERM. In the above equation,
we calibrate three alternative values for g (3, 4, and 5),
where three is close to the estimate obtained for the
CAPM. We also calibrate gz at three alternative values,
300, 600, and 1,200, where 600 is close to the estimate
obtained for the two-factor ICAPM with TERM in the
original sample. This gives a total of 18 alternative Monte
Carlo simulations.
The results in Table 14 show that the acceptance rates
of the ICAPM are relatively high, varying between 86% and
96% in the case of the one-month predictive regression,
while in the case of the 60-month regression, these rates
vary between 85% and 96%. These high acceptance rates
are the result of large individual statistical signicance for
the predictive slope (above 89% of the simulations) and
the factor risk price (over 96%), in conjunction with the
fact that the slope and risk price estimates in the pseudosamples have the same sign most of the time (over 91% in
untabulated results).
Overall, this simulation shows that in an ICAPM world,
there is a very high probability that the restrictions on the
intertemporal risk prices will be true in a sample similar
to ours.
We conduct a third Monte Carlo simulation to control
for the bias associated with a missing risk factor in the
ICAPM pricing equation.31 Specically, we assume that
the true ICAPM specication contains two hedging risk
factors: one that is observable, and hence used in the
empirical test, and another state variable that is not
observable to the econometrician. Results presented in
the addendum show that the acceptance rates, although
lower than in the benchmark simulation discussed above,
are still above 50%, thus showing that the ICAPM is
reasonably robust to the bias caused by a missing state
variable in the pricing equation.
31
We thank the referee for this suggestion.
8. Conclusion
Is the ICAPM a shing license for empirical multifactor models as Fama (1991) claims? We have studied
the restrictions associated with the ICAPM for a timeseries of the market return and a cross-section of portfolios. By using a simple version of the Merton (1973)
ICAPM, we identify three main conditions for a multifactor model to meet to be justiable by the ICAPM. First,
the candidates for ICAPM state variables must forecast the
rst or second moments of stock market returns. Second,
and most importantly, the state variables should forecast
changes in investment opportunities (expected market
return) with the same sign as its innovation prices the
cross-section. Specically, if a given state variable forecasts positive expected returns, it should earn a positive
risk price in the cross-sectional test of the respective
multifactor model. The third restriction associated with
the ICAPM is that the market (covariance) risk price
estimated from the cross-sectional tests must be economically plausible as an estimate of the coefcient of relative
risk aversion (RRA) of a representative investor.
We apply our ICAPM criteria to eight multifactor
models, tested over 25 portfolios sorted on size and
book-to-market (SBM25), and 25 portfolios sorted on size
and momentum (SM25). Our results show that only in
three out of 16 tests are factor risk prices consistent with
the ICAPM theory: the Fama and French (1993) threefactor model tested over SBM25, and the Carhart (1997)
model tested over SBM25 and SM25. Thus, these models
can be justied as empirical applications of the ICAPM.
When we consider changes in the investment opportunity
set driven by the second moment of aggregate returns,
only the Fama and French (1993) model tested with the
SBM25 portfolios satises the ICAPM criteria. In most
models, there are inconsistencies in both the risk price
estimates associated with the hedging risk factors and
the risk-aversion estimates. These ndings are robust to
many different specications of our tests.
Overall, the Fama and French (1993) three-factor
model performs the best in consistently meeting the
ICAPM restrictions when investment opportunities are
driven by the rst two moments of aggregate returns
when tested with the SBM25 portfolios. Apart from this
model and the Carhart (1997) model, the other models
cannot be justied with the ICAPM theory. The ICAPM is
not really a shing license after all.
Appendix A. GMM formulas
Following Cochrane (2005), the weighting matrix
associated with the GMM system (36) is given by
" n
#
W
0
W
,
A:1
0
IK 1
where Wn N N is the weighting matrix associated with
the rst N moments; 0 denotes a conformable matrix of
zeros; and IK 1 denotes a K 1-dimensional identity
matrix. In this specication, Wn is the weighting matrix
for the rst N moment conditions (corresponding to the N

pricing errors), while IK 1 is the weighting matrix associated with the last K 1 orthogonality conditions that
identify the factor means.
In the rst-step GMM (OLS cross-sectional regression),
Wn corresponds to the identity matrix, Wn IN , and in the
second-step GMM (GLS cross-sectional regression), Wn is
the inverse of the rst N N block of the spectral
density matrix, Wn S1
N .
The risk price estimates b^ have variance formulas
given by
^
Varb
1 0
0
0
^
d Wd1 d WSWdd
Wd1 ,
T
611
given by
g T b
8
Ri,t 1 Rf ,t 1 b gb Ri,t 1 Rf ,t 1 b RMbt 1 mbm
>
>
>
>
>
b b
b
b
>
>
> g1 Ri,t 1 Rf ,t 1 f 1,t 1 m1
>
>
>
b b
b
b
>
>

g
R
R
f

m
f ,t 1
>
2,t 1
2 i,t 1
2
>
>
b b
<
T 1 >
b
X
g R
R
f
mb
1
K
i,t 1
f ,t 1
K,t 1
T t 0>
>
RMbt 1 mbm
>
>
>
>
>fb
b
>
>
1,t 1 m1
>
>
>
>
^
>
>
>
>
:fb
b
K,t 1 mK
i 1, . . . ,N:
A:2
where d @g T b=@b represents the matrix of moments

sensitivities to the parameters; S^ is an estimator for the
spectral density matrix, S, derived under the heteroskedasticity-robust or White (1980) standard errors, that is,
no lags of the moment functions are considered in the
^ 32
computation of S.
0,
B:3
We save the t-statistics associated with the individual

bb1 , . . . ,tg
bbK leading to an
bb ,tg
factor risk prices, tg
empirical distribution of the t-statistics.
5. The empirical p-value associated with the market risk
price (for a two-sided test) is computed as
b
pg
8
< #ftg
bb Z tg
bg ftg
bb rtg
bg=10,000
b Z 0,
if g
: #ftg
bg ftg
bb Ztg
bg=10,000
bb r tg
b o 0,
if g
B:4
Appendix B. Bootstrap simulation for cross-sectional

test
The bootstrap algorithm used in the estimation of the
pricing equations in the cross-section of stock returns
consists of the following steps:
1. Each model is estimated by rst-stage GMM, and we
save the t-statistics associated with the market and
b,tg
b1 , . . . ,tg
bK .
hedging risk prices, tg
2. In each replication b 1, . . . ,10,000, we construct a
pseudo-sample of excess returns for each asset (of
size T) by drawing with replacement:
fRi,t 1 Rf ,t 1 b ,t sb1 ,sb2 , . . . ,sbT g,
i 1, . . . ,N,
B:1
where the time indices sb1 ,sb2 , . . . ,sbT are created randomly
from the original time sequence, 1, . . . ,T. Notice that all
excess returns have the same time sequence to preserve
the contemporaneous cross-correlation between asset
returns.
3. For each replication b 1, . . . ,10,000, we also construct an independent pseudo-sample of the factors:
b
fRMbt 1 ,f 1,t 1 ,
b
. . . ,f K,t 1 ,t
r b1 ,r b2 ,
. . . ,r bT g,
B:2
where the time sequence r b1 ,r b2 , . . . ,r bT is independent

from sb1 ,sb2 , . . . ,sbT . Notice that the time sequence is the
same for all factors to preserve their cross-correlations.
4. In each replication, we estimate the factor model by
rst-stage GMM, but using the articial data rather
than the original data. The moment conditions are
and similarly for the other factor risk prices. In the

bb Z tg
bg denotes the number
above expression, #ftg
of replications in which the pseudo-t-stats are greater
than or equal to the t-ratio from the original sample.
Appendix C. Bootstrap algorithm for predictive

regressions
The bootstrap algorithm associated with the longhorizon regressions consists of the following steps:
1. The long-horizon regression is estimated by OLS, and
cq :
we save the slope estimate, b
r t,t q aq bq zt ut,t q :
C:1
We also estimate by OLS the following system that

imposes the joint null of no predictability of returns
and a persistent predictor that follows an AR(1) process:
r t,t q aq ut,t q ,
C:2
zt 1 c fzt et 1 :
C:3
b t,t q and b
e t 1 , and the
The time-series of OLS residuals, u
b ,f
b are saved.
bq , c
OLS estimates, a
2. In each replication m 1, . . . ,10,000, we construct
pseudo-samples for the innovations in the market
return and the predictor by drawing with replacement
from the two residuals:
m
b t,t q g,
fu
fb
em
t 1 g,
m
m
t sm
1 ,s2 , . . . ,sT ,
m
m
t sm
1 ,s2 , . . . ,sT ,
C:5
m
sm
1 ,s2 ,
32
As stressed by Cochrane (2005), if the asset pricing model is true,
then the moments that dene the pricing errors will be orthogonal to all
past information, including the past pricing errors. This is equivalent to
using the Newey and West (1987) algorithm with zero lags.
C:4
. . . ,sm
T ,
where the time indices,

are created randomly from the original time sequence 1, . . . ,T. Notice
that the innovations in both the return and predictor
have the same time sequence to account for their
contemporaneous cross-correlation.
3. For each replication m 1, . . . ,10,000, we construct a
pseudo-sample of the market return and predictor, by

612
imposing the null:

rm
t,t q
bq u
bm
a
t,t q ,
m
b b m b
zm
t 1 c f zt e t 1 :
C:6
C:7
4. In each replication, we estimate the long-horizon

regression, but using the articial data rather than
the original data:
m
m
m
m
rm
t,t q aq bq zt vt,t q :
C:8
The initial value for zt, z0 is picked at random from one

of the observations of zt. In result, we have an empiricalm distribution of the regression slope estimates,
b g10,000 (as opposed to the asymptotic theoretical
fb
q m1
distribution).
5. The p-value associated with the slope estimate is
calculated as
8
>
bm Z b
b q g fb
b m r b
bq g=10,000 if b
bq Z0,
< #fb
q
q
b
pb q
m
m
b rb
>
b q g fb
b Z b
bq g=10,000 if b
bq o0,
: #fb
q
q
C:9
bm Z b
bq g denotes the number of bootstrapped
where #fb
q
slope estimates that are higher than the original slope
estimate.
References
Ang, A., Bekaert, G., 2007. Stock return predictability: Is it there? Review
of Financial Studies 20, 651708.
Ang, A., Hodrick, R., Xing, Y., Zhang, X., 2006. The cross-section of
volatility and expected returns. Journal of Finance 61, 259300.
Bali, T., 2008. The intertemporal relation between expected returns and
risk. Journal of Financial Economics 87, 101131.
Bali, T., Demirtas, K., Levy, H., 2009. Is there an intertemporal relation
between downside risk and expected returns? Journal of Financial
and Quantitative Analysis 44, 883909.
Bali, T., Engle, R., 2010. The intertemporal capital asset pricing model
with dynamic conditional correlations. Journal of Monetary Economics 57, 377390.
Boudoukh, J., Richardson, M., Whitelaw, R., 2008. The myth of longhorizon predictability. Review of Financial Studies 21, 15771605.
Brandt, M., Kang, Q., 2004. On the relationship between the conditional
mean and volatility of stock returns: a latent VAR approach. Journal
of Financial Economics 72, 217257.
Brennan, M., Wang, A., Xia, Y., 2004. Estimation and test of a simple
model of intertemporal capital asset pricing. Journal of Finance 59,
17431775.
Brennan, M., Xia, Y., 2006. Risk and valuation under an intertemporal
capital asset pricing model. Journal of Business 79, 135.
Campbell, J., 1987. Stock returns and the term structure. Journal of
Financial Economics 18, 373399.
Campbell, J., 1991. A variance decomposition for stock returns. Economic
Journal 101, 157179.
Campbell, J., 1993. Intertemporal asset pricing without consumption
data. American Economic Review 83, 487512.
Campbell, J., 1996. Understanding risk and return. Journal of Political
Economy 104, 298345.
Campbell, J., Shiller, R., 1988a. The dividend price ratio and expectations
of future dividends and discount factors. Review of Financial Studies
1, 195228.
Campbell, J., Shiller, R., 1988b. Stock prices, earnings, and expected
dividends. Journal of Finance 43, 661676.
Campbell, J., Thompson, S., 2008. Predicting excess stock returns out of
sample: Can anything beat the historical average? Review of
Financial Studies 21, 15091531.
Campbell, J., Vuolteenaho, T., 2004. Bad beta, good beta. American
Economic Review 94, 12491275.
Campbell, J., Yogo, M., 2006. Efcient tests of stock return predictability.
Journal of Financial Economics 81, 2760.
Carhart, M., 1997. On persistence in mutual fund performance. Journal of
Finance 52, 5782.
Chen, J., 2003. Intertemporal CAPM and the cross-section of stock
returns. Unpublished working paper. University of California, Davis.
Chen, L., Zhao, X., 2009. Return decomposition. Review of Financial
Studies 22, 52135249.
Cochrane, J., 1996. A cross sectional test of an investment-based asset
pricing model. Journal of Political Economy 104, 572621.
Cochrane, J., 2005. Asset Pricing, revised edition Princeton University
Press, Princeton, NJ.
Cochrane, J., 2007. Financial markets and the real economy. In: Mehra, R.
(Ed.), Handbook of the Equity Premium, Elsevier Science, North
Holland, pp. 237325.
Cochrane, J., 2008. The dog that did not bark: a defense of return
predictability. Review of Financial Studies 21, 15331575.
Cochrane, J., Piazzesi, M., 2005. Bond risk premia. American Economic
Review 95, 138160.
Da, Z., Schaumburg, E., 2011. The pricing of volatility risk across asset
classes. Unpublished working paper. University of Notre Dame and
Federal Reserve Bank of New York.
Epstein, L., Zin, S., 1991. Substitution, risk aversion, and the temporal
behavior of consumption and asset returns: an empirical analysis.
Journal of Political Economy 99, 263286.
Fama, E., 1991. Efcient capital markets: II. Journal of Finance 46,
15751617.
Fama, E., French, K., 1988. Dividend yields and expected stock returns.
Fama, E., French, K., 1989. Business conditions and expected returns on
stocks and bonds. Journal of Financial Economics 25, 2349.
Fama, E., French, K., 1992. The cross-section of expected stock returns.
Journal of Finance 47, 427465.
Fama, E., French, K., 1993. Common risk factors in the returns on stocks
and bonds. Journal of Financial Economics 33, 356.
Fama, E., French, K., 1996. Multifactor explanations of asset pricing
anomalies. Journal of Finance 51, 5584.
Fama, E., Schwert, G., 1977. Asset returns and ination. Journal of
Ferson, W., Harvey, C., 1999. Conditioning variables and the cross section
of stock returns. Journal of Finance 54, 13251360.
Gerard, B., Wu, G., 2006. How important is intertemporal risk for asset
allocation? Journal of Business 79, 22032241.
Ghysels, E., Santa-Clara, P., Valkanov, R., 2005. There is a risk-return
trade-off after all. Journal of Financial Economics 76, 509548.
Goyal, A., Welch, I., 2008. A comprehensive look at the empirical
performance of equity premium prediction. Review of Financial
Studies 21, 14551508.
Guo, H., 2006a. Time-varying risk premiums and the cross section of
stock returns. Journal of Banking and Finance 30, 20872107.
Guo, H., 2006b. On the out-of-sample predictability of stock returns.
Journal of Business 79, 645670.
Guo, H., Savickas, R., 2008. Average idiosyncratic volatility in G7
countries. Review of Financial Studies 21, 12591296.
Guo, H., Savickas, R., Wang, Z., Yang, J., 2009. Is the value premium a
proxy for time-varying investment opportunities? Some time-series
evidence. Journal of Financial and Quantitative Analysis 44, 133154.
Guo, H., Whitelaw, R., 2006. Uncovering the risk-return relation in the
stock market. Journal of Finance 61, 14331464.
Hahn, J., Lee, H., 2006. Yield spreads as alternative risk factors for size
and book-to-market. Journal of Financial and Quantitative Analysis
41, 245269.
Hansen, L., 1982. Large sample properties of generalized method of
moments estimators. Econometrica 50, 10291054.
Hansen, L., Hodrick, R., 1980. Forward exchange rates as optimal
predictors of future spot rates: an econometric analysis. Journal of
Political Economy 88, 829853.
Hansen, L., Jagannathan, R., 1997. Assessing specication errors in
stochastic discount factor models. Journal of Finance 52, 557590.
Hodrick, R., 1992. Dividend yields and expected stock returns: alternative procedures for inference and measurement. Review of Financial Studies 5, 357386.
Hodrick, R., Zhang, X., 2001. Evaluating the specication errors of asset
pricing models. Journal of Financial Economics 62, 327376.
Jacobs, K., Wang, K., 2004. Idiosyncratic consumption risk and the cross
section of asset returns. Journal of Finance 59, 22112252.
Jagannathan, R., Wang, Y., 2007. Lazy investors, discretionary consumption, and the cross-section of stock returns. Journal of Finance 62,
16231661.

Jagannathan, R., Wang, Z., 1996. The conditional CAPM and the crosssection of expected returns. Journal of Finance 51, 353.
Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling
losers: implications for stock market efciency. Journal of Finance
48, 6591.
Kan, R., Robotti, C., 2008. Model comparison using the Hansen-Jagannathan distance. Review of Financial Studies 22, 34503490.
Keim, D., Stambaugh, R., 1986. Predicting returns in the stock and bond
markets. Journal of Financial Economics 17, 357390.
Kilian, L., 1999. Exchange rates and monetary fundamentals: What do
we learn from long-horizon regressions? Journal of Applied Econometrics 14, 491510.
Koijen, R., Lustig, H., Van Nieuwerburgh, S., 2010. The cross-section and
time-series of stock and bond returns. Unpublished working paper.
University of Chicago, University of California at Los Angeles, New
York University.
Lettau, M., Ludvigson, S., 2001. Resurrecting the (C)CAPM: a crosssectional test when risk premia are time-varying. Journal of Political
Economy 109, 12381287.
Lewellen, J., 2004. Predicting returns with nancial ratios. Journal of
Lewellen, J., Nagel, S., 2006. The conditional CAPM does not explain assetpricing anomalies. Journal of Financial Economics 82, 289314.
Lewellen, J., Nagel, S., Shanken, J., 2010. A skeptical appraisal of assetpricing tests. Journal of Financial Economics 96, 175194.
Lintner, J., 1965. The valuation of risk assets and the selection of risky
investments in stock portfolios and capital budgets. Review of
Economics and Statistics 47, 1337.
Lo, A., Wang, J., 2006. Trading volume: implications for an intertemporal
capital asset pricing model. Journal of Finance 61, 28052840.
Lundblad, C., 2007. The risk return tradeoff in the long run: 18362003.
Maio, P., 2012. Bond risk premia, return decomposition and the Intertemporal CAPM. Unpublished working paper. Hanken School of
Economics.
Maio, P. Return decomposition and the Intertemporal CAPM. Hanken
School of Econmoics (Unpublished working paper).
Mehra, R., Prescott, E., 1985. The equity premium: a puzzle. Journal of
Monetary Economics 15, 145161.
Merton, R., 1973. An intertemporal capital asset pricing model. Econometrica 41, 867887.
Nelson, R., Kim, M., 1993. Predictable stock returns: the role of small
sample bias. Journal of Finance 48, 641661.
Newey, W., West, K., 1987. A simple, positive semi-denite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55, 703708.
613
Nielsen, L., Vassalou, M., 2006. The instantaneous capital market line.
Economic Theory 28, 651664.
Ozoguz, A., 2009. Good times or bad times? Investors uncertainty and
stock returns. Review of Financial Studies 22, 43774422.
Pastor, L., Sinha, M., Swaminathan, B., 2008. Estimating the intertemporal risk-return tradeoff using the implied cost of capital. Journal of
Finance 63, 28592897.
Pastor, L., Stambaugh, R., 2003. Liquidity risk and expected stock returns.
Journal of Political Economy 111, 642684.
Pennacchi, G., 2008. Theory of Asset Pricing. Pearson Addison Wesley,
Boston, MA.
Petkova, R., 2006. Do the Fama-French factors proxy for innovations in
predictive variables? Journal of Finance 61, 581612.
Scruggs, J., 1998. Resolving the puzzling intertemporal relation between
the market risk premium and conditional market variance: a twofactor approach. Journal of Finance 53, 575603.
Shanken, J., 1990. Intertemporal asset pricing: an empirical investigation. Journal of Econometrics 45, 99120.
Shanken, J., 1992. On the estimation of beta pricing models. Review of
Financial Studies 5, 134.
Shanken, J., Zhou, G., 2007. Estimating and testing beta pricing models:
alternative methods and their performance in simulations. Journal of
Sharpe, W., 1964. Capital asset prices: a theory of market equilibrium
under conditions of risk. Journal of Finance 19, 425442.
Stambaugh, R., 1999. Predictive regressions. Journal of Financial Economics 54, 375421.
Tang, Y., Whitelaw, R., 2011. Time-varying Sharpe ratios and market
timing. Quarterly Journal of Finance 1, 465493.
Torous, W., Valkanov, R., Yan, S., 2004. On predicting stock returns with
nearly integrated explanatory variables. Journal of Business 77,
937966.
Valkanov, R., 2003. Long-horizon regressions: theoretical results and
applications. Journal of Financial Economics 68, 201232.
Vassalou, M., 2003. News related to future GDP growth as a risk factor in
equity returns. Journal of Financial Economics 68, 4773.
White, H., 1980. A heteroskedasticity-consistent covariance matrix
estimator and a direct test for heteroskedasticity. Econometrica 48,
817838.
Whitelaw, R., 1994. Time variations and covariations in the expectation
and volatility of stock market returns. Journal of Finance 49,
515541.
Whitelaw, R., 2000. Stock market risk and return: an equilibrium
approach. Review of Financial Studies 13, 521547.
Yogo, M., 2006. A consumption-based explanation of expected stock
returns. Journal of Finance 61, 539580.

Multi Factor Models I Cap M

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Multi Factor Models I Cap M

Hochgeladen von

Copyright:

Verfügbare Formate

This article appeared in a journal published by Elsevier.

Author's personal copy

Contents lists available at SciVerse ScienceDirect

Journal of Financial Economics

Multifactor models and their consistency with the ICAPM$

Hanken School of Economics, Finland

Can any multifactor model be interpreted as a variant of the Intertemporal CAPM

Timo Korkeamaki, Anders Lound,

inability of the Sharpe (1964)Lintner (1965) CAPM to

Author's personal copy

Author's personal copy

P. Maio, P. Santa-Clara / Journal of Financial Economics 106 (2012) 586613

over SBM25 and SM25meet the ICAPM consistency

the ICAPM state variables with regards to the volatility of

where Si denotes the price of asset i; dxi is a Wiener

To simplify the exposition, we assume that the risk-free

where oi denotes the portfolio weight for asset i, and C

subject to the intertemporal budget constraint (4), where

For a textbook treatment, see Pennacchi (2008, Chapter 13).

Author's personal copy

where J W  denotes the marginal value of wealth; JWW 

where Ri,t 1 is the return on asset i between t and t 1;

asset that covaries positively with changes in the state

then the intertemporal risk price must be positive.

These two conditions imply that the return on asset i is

This last condition has an important economic content:

Author's personal copy

P. Maio, P. Santa-Clara / Journal of Financial Economics 106 (2012) 586613

associated with those factors is inconsistent with the

then the risk price of intertemporal risk must be negative.5

The economic implication of this last condition is that

3. Do state variables forecast future investment

candidates for state variables forecast future investment

Author's personal copy

computed as the log ratio of the sum of annual dividends to

These three factors are obtained from Kenneth Frenchs Web

r t,t q aq bq TERM t cq CP t ut,t q :

The second set of state variables is based on risk factors

We use the cumulative sum over the last 60 months since

MBSL MBSM MBSH MBBL MBBM MBBH

where MBSL, MBSM, MBSH, MBBL, MBBM, and MBBH denote

r t,t q aq bq TERM t cq DEF t dq DY t eq RF t ut,t q ,

MBSH MBBH MBSL MBBL

Since SMB C DSMBn and HML C DHMLn , this procedure

r t,t q aq bq SMBnt cq HMLnt dq CUMDt ut,t q ,

r t,t q aq bq SMBnt cq HMLnt dq CLt ut,t q :

Also we estimate a predictive regression associated with the

The value spread is calculated from six portfolios sorted on both

Author's personal copy

P. Maio, P. Santa-Clara / Journal of Financial Economics 106 (2012) 586613

Table 3 presents the results for the single long-horizon

at the 5% or 1% levels for q1 and q60. TERM forecasts

The second model is the ICAPM of Petkova (2006) that

The third model is an unrestricted version of the Campbell

Author's personal copy

opportunities are described by PE, TERM, and VS17:

The fourth model is the three-factor model from Koijen,

Next, we use multifactor models with less theoretical

SMB is used to explain the size premium, the positive

Author's personal copy

P. Maio, P. Santa-Clara / Journal of Financial Economics 106 (2012) 586613

Fama and French (1993) use this model to explain both

the momentum anomaly; that is, past short-term winners

where J W denotes the marginal value of wealth; JWW