Avramov Etal21

Journal of Financial Economics 145 (2022) 642–664
Contents lists available at ScienceDirect
Journal of Financial Economics

journal homepage: www.elsevier.com/locate/jfec
Sustainable investing with ESG rating uncertaintyR

Doron Avramov a,∗, Si Cheng b, Abraham Lioui c, Andrea Tarelli d
a
Interdisciplinary Center (IDC), Herzliya, Israel
b
Chinese University of Hong Kong, Hong Kong
c
EDHEC Business School and EDHEC & Scientific Beta Research Chair, Nice, France
d
Catholic University of Milan, Milan, Italy
a r t i c l e i n f o a b s t r a c t
Article history: This paper analyzes the asset pricing and portfolio implications of an important barrier
Received 9 February 2021 to sustainable investing: uncertainty about the corporate ESG profile. In equilibrium, the
Revised 26 June 2021
market premium increases and demand for stocks declines under ESG uncertainty. In ad-
Accepted 25 July 2021
dition, the CAPM alpha and effective beta both rise with ESG uncertainty and the nega-
Available online 17 September 2021
tive ESG-alpha relation weakens. Employing the standard deviation of ESG ratings from six
JEL classification: major providers as a proxy for ESG uncertainty, we provide supporting evidence for the
G11 model predictions. Our findings help reconcile the mixed evidence on the cross-sectional
G12 ESG-alpha relation and suggest that ESG uncertainty affects the risk-return trade-off, social
G24 impact, and economic welfare.
M14
Q01
© 2021 Elsevier B.V. All rights reserved.
Keywords:
ESG
Rating uncertainty
Portfolio choice
Capital asset pricing model
1. Introduction nance (ESG) factors in portfolio selection and management.

Since the launch of the United Nations Principles for Re-
The global financial market has experienced exponen- sponsible Investment (PRI) in 2006, the number of signato-
tial growth in sustainable investing, an investment ap- ries has grown from 734 in 2010 to 1384 in 2015 and 3038
proach that considers environmental, social, and gover- in 2020, with total assets under management of US$21 tril-
lion in 2010, US$59 trillion in 2015, and US$103 trillion in
2020.1 In line with the increasing concerns about global
R
Bill Schwert was the editor for this article. We especially thank an
warming, BlackRock CEO Larry Fink wrote in a recent an-
anonymous referee for insightful comments and suggestions. We also nual letter that climate change will force businesses and
thank Bill Schwert, Yakov Amihud, Marcin Kacperczyk, Lubos Pástor, Lasse investors to shift their strategies, leading to a “fundamental
Heje Pedersen, Luke Taylor, seminar participants at Catholic University of reshaping of finance” and “significant reallocation of capi-
Milan, Ben-Gurion University of the Negev, Bar-Ilan University, Bocconi
tal.”2
University, and EDHEC Business School, and conference participants at the
2021 North American and European Summer Meetings of the Economet- As the ESG objective is becoming a primary focus in as-
ric Society for useful comments and discussions. We are solely responsi- set management, the reallocation of capital has major im-
ble for any remaining errors.
∗
Corresponding author.
1
E-mail addresses: [email protected] (D. Avramov), See, https://www.unpri.org/pri.
2
[email protected] (S. Cheng), [email protected] (A. Lioui), See, https://www.blackrock.com/corporate/investor-relations/larry-
[email protected] (A. Tarelli). fink- ceo- letter.
https://doi.org/10.1016/j.jfineco.2021.09.009
0304-405X/© 2021 Elsevier B.V. All rights reserved.
D. Avramov, S. Cheng, A. Lioui et al. Journal of Financial Economics 145 (2022) 642–664
plications for portfolio decisions and asset pricing. How- alpha relation is, hence, negative.4 Accounting for ESG un-
ever, ESG investors often confront a substantial amount of certainty, the equilibrium alpha increases with ESG uncer-
uncertainty about the true ESG profile of a firm. In the ab- tainty and the ESG-alpha relation weakens.
sence of a reliable measure of the true ESG performance, We move on to empirically test the model implications
any attempt to quantify it needs to cope with incomplete using U.S. common stocks from 2002 to 2019. We collect
and opaque ESG data and nonstructured methodologies. A ESG ratings from six major rating agencies, namely, As-
meaningful illustration of uncertainty about the ESG score set4 (Refinitiv), MSCI KLD, MSCI IVA, Bloomberg, Sustain-
is the pronounced divergence across ESG rating agencies.3 alytics, and RobecoSAM. We employ the average (standard
While such uncertainty could be an important barrier to deviation of) ESG ratings across rating agencies to proxy
sustainable investing, to date, little attention has been de- for the firm-level ESG rating (ESG uncertainty). Consistent
voted to the role of ESG uncertainty in portfolio decisions with existing studies, we confirm that there are substan-
and asset pricing. tial variations across different rating providers, while the
This paper aims to fill this gap by analyzing the equi- average rating correlation is 0.48. The variations are quite
librium implications of ESG uncertainty for both the ag- persistent throughout the entire sample period.
gregate market and the cross section. To pursue this We first examine how the ESG rating and uncertainty
task, we consider brown-averse agents who extract non- affect investor demand. To better capture the demand from
pecuniary benefits from holding green stocks, following ESG-sensitive investors, we consider three distinct types
Pástor et al. (2021a). We first study the aggregate market of institutions: norm-constrained institutions, hedge funds,
through a mean-variance setup that consists of the market and other institutions. Norm-constrained institutions, such
portfolio and a riskless asset. Due to uncertainty about the as pension funds as well as university and foundation en-
ESG profile, equities are perceived to be riskier. In addition, dowments, are more likely to make socially responsible
the demand for equities consists of two components: (1) investments compared to hedge funds or mutual funds
the usual demand when ESG preferences are muted and that are natural arbitrageurs (Hong and Kacperczyk, 2009).
(2) a demand for a pseudo-asset with a positive payoff for We first confirm that norm-constrained institutions display
a green market and a negative payoff for a brown market preferences for greener firms. Consistent with the model
as well as volatility that evolves from uncertainty about prediction, we find that in the presence of uncertainty
the market ESG score. Aggregating these components, we about the ESG profile, ESG-sensitive investors lower their
show that the overall demand for equities falls due to ESG demand for risky assets. For instance, among the high-ESG-
uncertainty, even when the market is green. rating portfolios, norm-constrained institutions hold 22.8%
We then formulate the market premium in equilibrium. of the low-uncertainty stocks but only 18.1% of the high-
While the higher risk due to ESG uncertainty essentially uncertainty stocks, indicating a 21% decline. The results are
commands a higher market premium, there is an offsetting particularly strong among high-ESG stocks, suggesting that
force when the market is green because ESG investors ex- rating uncertainty matters the most for ESG-sensitive in-
tract nonpecuniary benefits from holding green stocks. The vestors in their ESG investment. Notably, even with grow-
ultimate implications of ESG preferences with uncertainty ing ESG awareness, their demand for green assets has con-
for the market premium are thus inconclusive. When the tinued to diminish with rating uncertainty over the past
market is green neutral, however, the equity premium rises decade. In addition, while hedge funds invest more in low-
with ESG uncertainty. For perspective, when ESG uncer- ESG stocks, rating uncertainty plays a similar role in dis-
tainty is not accounted for and the market is green (green couraging stock investment.
neutral), the market risk does not change, the demand for We next examine the cross-sectional implications of
risky assets rises (does not change), and the market pre- ESG uncertainty. We first sort stocks into quintile portfolios
mium drops (does not change) relative to ESG indifference. based on their ESG uncertainty. Within each uncertainty
We further derive a CAPM representation in which both group, we further sort stocks into quintile portfolios ac-
alpha and the effective beta vary with firm-level ESG un- cording to their ESG ratings. We find that the ESG rating
certainty. The effective beta differs from the CAPM beta is negatively associated with future performance among
in the following way. While the CAPM beta is based on stocks with low ESG uncertainty, providing empirical sup-
the covariance and variance of actual returns, the effec- port for the predictions of Pástor et al. (2021a), who rely
tive beta reflects the notion that both the market and in- on deterministic ESG scores. For instance, brown stocks
dividual stock returns are augmented by a random ESG- outperform green stocks by 0.59% per month in raw re-
based component, which is positive for a green asset and turn and 0.40% per month in CAPM-adjusted return. How-
negative otherwise. Thus, the effective beta is based on ever, in the presence of ESG uncertainty, our model shows
the covariance and variance of ESG-adjusted returns. Re- that the ESG-alpha relation can be nonlinear and ambigu-
garding alpha, when ESG uncertainty is not accounted for, ous. Indeed, we demonstrate empirically that the negative
the CAPM alpha exclusively reflects the willingness to hold return predictability of ESG ratings does not hold for the
green stocks due to nonpecuniary benefits, and the ESG- remaining firms. The results are robust to adjusting returns
for alternative risk factors and controlling for firm charac-
teristics in Fama and MacBeth (1973) regressions.
3
Berg et al. (2020) report that the average correlation among six ma-
jor rating providers is only 0.54. They also find that, even when the cat-
egories of attributes considered for the evaluation of a firm’s ESG profile
are fixed, raters largely disagree on the measurement of these granular
4
characteristics. See, e.g., Heinkel et al. (2001) and Pástor et al. (2021a).
643
Finally, we calibrate the model for plausible values of To the extent that ESG uncertainty will decrease with a
market volatility and risk aversion. The investment uni- better understanding of a firm’s true ESG profile, our work
verse consists of a riskless asset and the market portfo- enriches academic and policy discussions in that context.
lio. Our calibration considers two types of agents who ob- Despite the rapid growth in the sustainable investing and
serve the returns on investable assets. One type of agents ESG data markets,5 the comparability of ESG information
accounts for ESG preferences with uncertainty in assess- remains a critical issue. Due to the lack of standards gov-
ing the risk-return profile of the optimal portfolio, while erning the reporting of ESG information, it is not a triv-
the other type is ESG indifferent. Accounting for ESG un- ial task to compare the ESG data of two different com-
certainty significantly reduces the demand for the mar- panies (Amel-Zadeh and Serafeim, 2018). In addition, the
ket portfolio and the certainty equivalent rate of return of construction of ESG ratings is nonregulated, and method-
ESG-sensitive agents. The calibration results reinforce the ologies can be opaque and proprietary, leading to sub-
notion that ESG uncertainty could negatively, and signif- stantial divergence across data providers (e.g., Mackintosh,
icantly, affect the risk-return trade-off, social impact, and 2018; Berg et al., 2020). Our findings imply that the lack
economic welfare. of consistency across ESG rating agencies makes sustain-
This paper contributes to several strands of the litera- able investing riskier and hence reduces investor participa-
ture. First, we explicitly account for uncertainty about the tion and potentially hurts economic welfare. This has im-
ESG profile in equilibrium asset pricing for both the ag- portant normative implications. For instance, it would be
gregate market and the cross section. Prior work has fo- useful for policy makers to establish a clear taxonomy of
cused on investors’ ESG preferences (e.g., Heinkel et al., ESG performance and unified disclosure standards for sus-
2001; Pástor et al., 2021a), while our model predictions tainability reporting. It would be especially instructive to
and calibration results highlight the importance of consid- identify which investments are really green. Doing so could
ering ESG uncertainty when analyzing sustainable invest- mitigate ESG uncertainty, thus reducing the cost of equity
ing. Specifically, the perceived equity risk increases with capital for green firms, leading to higher social impact.
ESG uncertainty, while the demand for equity falls. ESG Our study of the equilibrium implications of ESG un-
uncertainty also affects the market premium in aggregate, certainty owes a debt to the innovative setup developed
as well as the CAPM alpha and effective beta in the cross by Pástor et al. (2021a), although our focus is different.
section. Pástor et al. (2021a) comprehensively analyze the equilib-
Second, we contribute to the growing literature on rium implications of sustainable investing and conduct an
the cross-sectional return predictability of the ESG profile. analysis of welfare and social impact. They also account
Prior studies show weak return predictability of the over- for the possibility that ESG investors can disagree about a
all ESG rating (e.g., Pedersen et al., 2021) and mixed evi- firm’s ESG profile and analyze cases in which the market
dence based on different ESG proxies (e.g., Gompers et al., is green neutral or green. Notably, in their setup, the ESG
20 03; Hong and Kacperczyk, 20 09; Edmans, 2011; Bolton score is certain because investors are dogmatic about their
and Kacperczyk, 2020). Our contribution is to propose that ESG perceptions and can observe each other’s perceived
ESG uncertainty could tilt the ESG-performance relation ESG values. Relative to their important work, we study the
and serve as a potential mechanism to explain the op- implications of uncertainty about the corporate ESG pro-
posing findings. We show that ESG ratings are negatively file. In particular, the investors in our model agree that the
associated with future performance when there is little ESG scores are uncertain and they also agree on the un-
uncertainty and that the ESG-performance relation could derlying distribution of the uncertain scores. The empirical
be insignificant or positive when uncertainty increases. proxy for uncertainty is the dispersion, or disagreement,
Thus, the sin premium (Hong and Kacperczyk, 2009) and across raters. We show that ESG uncertainty affects the eq-
carbon premium (Bolton and Kacperczyk, 2020) could uity premium, investor’s demand for risky assets, economic
be attributed to the notion that sin stocks (i.e., com- welfare, and the alpha and beta components of stock re-
panies involved in producing alcohol, tobacco, and gam- turns.
ing) and carbon emissions are clearly defined and thus The remainder of this paper is organized as follows.
subject to minimal uncertainty among investors. On the Section 2 presents the model. Section 3 describes the
other hand, other ESG profiles could be more challeng- data and the main variables used. Section 4 empirically
ing to measure or rely on nonstandardized information examines how ESG ratings and uncertainty affect in-
and methodologies, thereby displaying more uncertainty vestor demand and cross-sectional return predictability.
and mixed evidence on return predictability. A recent work Section 5 calibrates the model and explores its quantita-
by Pástor et al. (2021b) further highlights the distinction tive implications. The conclusion follows in Section 6.
between ex ante expected returns and ex post realized
returns, and shows that U.S. green stocks outperformed 2. ESG and market equilibrium
brown stocks during the last decade, due to unexpect-
edly strong increases in environmental concerns. While our The theory section develops the economic setup. We
model is static in nature and formulates expected returns, start with a single risky asset, i.e., the market portfolio,
we also confirm that our findings are stronger in the pre- and a riskless asset. We derive the optimal portfolio and
2011 period. This suggests that the equilibrium outcome
over longer horizons could be even stronger than the full 5
The estimated spending on ESG data was US$617 million in 2019
sample evidence we report, due to the unexpected out- and could approach US$1 billion by 2021. See, http://www.opimas.com/
comes realized over the last decade. research/547/detail/.
644
discuss the implications of uncertainty about the ESG pro- be positive. In particular, if the agent learns that the mar-
file for the market premium and welfare. The single-asset ket ESG score is higher than previously thought (i.e., ˜g,M
setup is then extended to consider multiple risky assets. is positive), the price that he would be willing to pay for
We analyze the implications of ESG uncertainty for the de- the market will be revised upward (positive ˜M ), while a
mand of individual stocks, derive an asset pricing model downward price revision applies for a score lower than
for the cross section of stock returns, and discuss incre- previously thought.7
mental effects of ESG uncertainty on the alpha and beta Observe from Eq. (3) that the investment in the riskless
components of returns. asset does not contribute to the portfolio’s ESG profile, as
perceived by the agent. This is because the riskless asset
2.1. One risky asset is implicitly assumed to be green neutral. As ESG scores
are ordinal in nature, the choice of considering the riskless
Consider a single-period economy in which an optimiz- asset as a reference level does not imply loss of general-
ing agent trades at time 0 and liquidates the position at ity. In addition, to capture the ESG benefits and costs from
time 1. Let r˜M denote the random rate of return on the investing in the market, we allow the market portfolio to
market portfolio in excess of the riskless rate, r f , and let depart from green neutrality.
g˜M denote the true, but unobservable, ESG score of the The agent picks x, attempting to maximize the expected
market portfolio.6 We model the excess market return and value of preferences in Eq. (3). The first-order condition
the ESG score as suggests that the optimal portfolio in the presence of ESG
uncertainty is given by
r˜M = μM + ˜M , (1)
g˜M = μg,M + ˜g,M ,

1 μM + bμg,M
(2) x∗ = , (4)
γ σM,U
2
where E (r˜M ) = μM is the expected market excess return,
E (g˜M ) = μg,M is the expected value of the market ESG where b = AB , γ = AW0 stands for the relative risk aversion,
score, and ˜M and ˜g,M are zero-mean residuals. We as- and σM,U = σM2 + b2 σ 2 + 2bσ σ
M g,M ρg,M is the variance of
2
g,M
sume that the residuals obey a bivariate normal distribu-
return, as perceived by the agent. Henceforth, b is referred
tion with σM , σg,M , and ρg,M denoting the standard devia-
to as brown aversion for brevity. The ex ante market vari-
tion of return, the standard deviation of ESG score, and the
ance, σM,U
2 , is no longer equal to σ 2 because, with ESG un-
M
correlation between residuals, respectively.
certainty, the risky asset is perceived to be a package of
It is assumed that the agent knows the joint distribu-
two distinct securities. The first delivers the market excess
tion of return and the ESG score as well as the underly-
return r˜M , while the second reflects exposure to ESG uncer-
ing parameters. In the empirical analysis that follows, μg,M
tainty and yields bg˜M . The latter component can be inter-
and σg,M are proxied by the average and standard devia-
preted as investing b units in a pseudo-asset that pays g˜M
tion of ESG ratings across six major data vendors, respec-
per unit. As b increases, i.e., when the ratio between brown
tively. From an investor’s perspective, a higher σg,M in-
aversion and risk aversion increases, the ESG component
dicates more disagreement among ESG raters and hence
becomes more meaningful in investment decisions. A suf-
more uncertainty about the true ESG profile of the market.
ficient condition for σM,U 2 ≥ σM2 is that the brown aversion
Following Pástor et al. (2021a), we consider an optimiz-
and the correlation between market return and ESG score
ing agent who derives nonpecuniary benefits from hold-
are nonnegative (i.e., b ≥ 0 and ρg,M ≥ 0). As noted earlier,
ing stocks based on their ESG characteristics. Moreover,
these conditions are likely to be satisfied.
preferences are formulated through the exponential utility
In what follows, we consider a positive market pre-
(CARA) function
mium (i.e., μM > 0), which is plausible in the presence

˜ 1 , x = −e−AW˜ 1 −BW0 xg˜M ,
V W (3) of risk aversion. The brown-aversion assumption is sensi-
ble for ESG-perceptive investors. Additionally, to distill the
where W ˜ 1 = W0 1 + r f + xr˜M is the terminal wealth, W0 is incremental effects of ESG uncertainty, we consider two
the initial wealth, x is the fraction of wealth invested in benchmark cases. In the first, the agent is ESG indiffer-
the risky asset, A stands for the agent’s absolute risk aver- ent, and in the second, preference for ESG is accounted for,
sion, and B characterizes the nonpecuniary benefits that while the ESG profile is known for certain. The latter case
the agent derives from stock holdings. Positive (negative) is studied by Pástor et al. (2021a) in a multiple-security
B indicates that the agent extracts benefits from holding setup.
green (brown) stocks. Hence, B can be interpreted as the Equation (4) presents the optimal stock position in the
absolute brown aversion. In the following, we make the presence of uncertainty about the ESG profile. Stock invest-
sensible assumption of a nonnegative brown aversion (B ≥ ment is thus driven by the relative risk aversion, γ , and
0). Slightly departing from Pástor et al. (2021a), we formu- the price of risk of the portfolio that yields r˜M + bg˜M . To
late preferences for ESG to be wealth-dependent. Then, the give perspective on the optimal equity demand, consider
expression BW0 represents the relative brown aversion. the case that incorporates ESG preferences but excludes
In the presence of brown aversion, the correlation be- uncertainty. Then, the perceived volatility of the stock re-
tween residuals in Eqs. (1) and (2), ρg,M , is assumed to turn is still σM . Conforming to intuition, the demand for
6
Consistent with static setups, we do not formulate intertemporal pref-
7
erences; hence, the riskless rate is exogenously specified. We thank the referee for suggesting this avenue.
645
stocks rises as b rises and the market is green. Essentially, μUM = γ σM2 − bμg,M + γ (σM,U
2
− σM2 ). (8)
stocks are more attractive to a green-loving agent.
When ESG uncertainty is accounted for, however, this Retaining the assumptions of a green market and a
intuition is no longer binding. To illustrate, consider two brown-averse agent, the market premium diminishes rel-
limiting cases. In the first, b grows with no bound. The in- ative to Eq. (6), as captured by the second term in Eq. (7).
μM +bμg,M This is because, as implied by Pástor et al. (2021a) in
vestor then avoids equities, i.e., lim γ1 = 0. Simi-
b→∞ σM,U
2
a multi-asset context, an agent who extracts nonpecu-
larly, when ESG uncertainty rises with no bound, the de- niary benefits from holding green stocks is willing to com-
mand for stocks evaporates. Thus, both increasing brown promise on a lower risk premium relative to an ESG-
aversion and increasing uncertainty translate into increas- indifferent agent. If the market is green neutral, the second
ing equity risk. In the presence of ESG uncertainty, a term disappears; hence, the equity premium is unchanged
brown-averse agent could substantially reduce stock in- even when ESG preferences are accounted for.
vesting, even when the market is green, on average. Further accounting for uncertainty in Eq. (8), there are
Moving beyond the two limiting cases, we further ex- two conflicting forces. On the one hand, the agent extracts
amine portfolio tilts in the presence of ESG uncertainty. For nonpecuniary benefits from holding the green market, a
that purpose, we rewrite the optimal portfolio as force leading to diminished market premium. On the other
1 μM 1 μg,M hand, the market is perceived to be riskier; thus, it com-
x∗ = + b mands a higher market premium, as formulated in the
γ σM2 γ σM2
third term of Eq. (8). The overall effect is inconclusive. If
1 μM + bμg,M 2 σg,M
2
σM σg,M ρg,M the market is green neutral, the equity premium increases
− b + 2 b . (5)
γ σM2 σM,U
2 σM,U
2 relative to both benchmark cases due to the increasing risk
channel.
The first term on the right-hand side of Eq. (5) de- The same conflicting forces apply to the equilibrium
scribes the benchmark case of ESG indifference. Prefer- Sharpe ratio (the slope of the capital allocation line) when
ences for ESG generate the second and third terms. The accounting for ESG uncertainty, SRU , relative to ESG indif-
bμ
term γ1 g,M corresponds to the second benchmark case ference, SRI . Given market return volatility, σM , it follows
σM2
SRU σM,U
2
bμ
with ESG preferences when the ESG profile is known for that = − g,M . The first term is greater than one
SRI σM2 γ σM2
certain. It suggests that as b rises, the demand for the risky
and reflects the increase in perceived equity risk. The sec-
asset rises and portfolio tilt intensifies. The third term
ond captures the decrease in the market premium due to
purely reflects the incremental effect of ESG uncertainty.
the nonpecuniary benefits from ESG investing.
σg,M
2
The ratio stands for the contribution of ESG uncer- In the presence of ESG preferences, the market risk pre-
σM,U
2
mium thus incorporates an ESG incremental premium that
tainty to the total, ex ante, market variance. Additionally,
can be defined as
in the presence of a positive correlation between market
return and the ESG profile, the agent employs the market μNM − μIM = −bμg,M , (9)
portfolio to hedge against risk evolving from ESG uncer- 2
σ σ ρg,M
tainty, as captured by the hedge ratio M g,M . Hence, μUM − μIM = γ σM,U − σM2 − bμg,M . (10)
2 σM,U
the incremental effect of ESG uncertainty on stock invest- The no-uncertainty case is associated with a negative
ing (captured by the third term) is negative.8 ESG incremental premium when the market is green and
In addition, when the market is green neutral (i.e., the agent is brown-averse, while the incremental premium
μg,M = 0) and when the ESG profile is known for certain, is zero when the market is green neutral. In addition,
stock investing is unaffected relative to ESG indifference. In it is evident from Eq. (10) that the market premium in-
contrast, when the market is green neutral and ESG uncer- creases with ESG uncertainty. Collectively, with ESG uncer-
tainty is accounted for, participation in the equity market tainty, the incremental premium is positive when the mar-
is discouraged, relative to both benchmark cases. ket is green neutral. Otherwise, with a green market and a
We now turn to analyzing the equilibrium implications brown-averse agent, the sign of the incremental premium
of ESG preferences with uncertainty. It is assumed that, in is inconclusive due to the conflicting forces.
equilibrium, the representative agent’s wealth is fully in- The single-security economy establishes a solid bench-
vested in the market portfolio. Thus, equalizing the optimal mark in which to comprehend the more complex multi-
stock allocation in Eq. (4) to 1 yields the market premium. asset setup to be developed later in the text. While the
The market premiums for the cases of ESG indifference (I), cross-sectional ESG-alpha relation is negative when ESG
ESG preference with no uncertainty (N), and ESG prefer- uncertainty is not accounted for, the single-security case
ence with uncertainty (U) are given by provides the first clue that (1) the risk premium increases
with ESG uncertainty, and (2) the risk premium of a green
μIM = γ σM2 , (6) stock could exceed that of a brown stock in the presence
of ESG uncertainty. Taken together, the ESG-alpha relation
μNM = γ σM2 − bμg,M , (7) in the cross section can be subject to conflicting forces.
Up to this point, we have considered a single-agent
8
In the case where μM + bμg,M is negative, the ESG uncertainty effect
economy for ease of exposition. In what follows, to assess
on stock investing goes the opposite way. This requires the interaction of the welfare implications of ESG uncertainty in the aggre-
extreme brown aversion along with an extreme brown market. gate and to study the multi-asset economy, we extend the
646
framework to account for multiple heterogeneous agents. Proposition 1. The optimal portfolio strategy of investor i is
Thus, consider I agents indexed by i = 1, . . . , I, who differ given by
in their initial wealth Wi,0 , absolute risk aversion Ai , and 1
absolute brown aversion Bi . Market clearing requires that X ∗i = −1
i,U μr + bi μg , (14)
I ∗ Wi,0 γi
i=1 wi xi = 1, where wi = W0 is the fraction of agent i’s
initial wealth relative to aggregate wealth. With heteroge- where i,U = r + b2i g + 2bi rg is the covariance matrix of
neous agents, the market premium equivalent to Eq. (8) is r˜ + bi g˜ .
given by This portfolio strategy is the multi-asset version of
μUM = γM σM,U
2
− bM μg,M,U , (11) Eq. (4). It suggests that in the presence of ESG prefer-
ences, investors perceive asset excess returns to be the
where γM = I
1
−1 is the aggregate risk aversion, sum of (1) N stock excess returns r˜ and (2) N returns on
i=1 wi γi
I −1
i=1 wi γi bi
pseudo-assets yielding bi g˜ . Several implications are in or-
bM = is the aggregate brown aversion, σM,U
2 =
γM−1 der. First, infinitely brown-averse agents act as if they were
γM−1 infinitely risk averse, as, in the presence of ESG uncer-
I is the perceived aggregate variance, and
−1 −2
i=1 wi γi σi,U tainty, lim X ∗i = 0. Second, in another extreme case when
I −1 −2 bi →∞
i=1 wi bi γi σi,U
μg,M,U = bM γM−1 σ −2 μg,M is the perceived aggregate ESG uncertainty grows with no bound for all stocks, eco-
M,U
nomic agents avoid stocks altogether. Third, in intermedi-
ESG score. Online Appendix A.1 provides details.
ate cases, uncertainty about ESG profiles nonlinearly inter-
The changing cost of equity capital due to ESG prefer-
venes in formulating the optimal portfolio, through the in-
ences has implications for economic welfare and social im-
verse of i,U , and tends to reduce the demand for both
pact. For instance, Pástor et al. (2021a) show that when the
green and brown stocks.
market is green, the lower cost of equity capital could trig-
To highlight the incremental implications of ESG uncer-
ger increasing capital investment and social impact. In our
tainty for portfolio selection, we rewrite Eq. (14) as
setup, a green representative firm would be harmed by the
higher cost of equity capital induced by ESG uncertainty, 1 1
X ∗i = −1 μr + bi μg + i μr + bi μg , (15)
which could trigger adverse effects on capital investment γi r
γi
and social impact. In the calibration experiment described 2 −1
where i = −−1 r bi g + 2bi rg −1
r IN + b2i g + 2bi rg −1
r
in Section 5.1, we comprehensively analyze the utility loss
and IN stands for the N × N identity matrix.
attributable to ESG uncertainty. We also calibrate the mar-
The first term of the optimal portfolio coincides with
ket premium, as well as equity demand and welfare for
the strategy in Pástor et al. (2021a) (Eq. (4)). The second
two types of agents: the first is indifferent to ESG, while
term is exclusively attributable to ESG preferences with
the other is ESG perceptive and recognizes the uncertainty
uncertainty about the sustainability profile. Interestingly, in
about the sustainability profile.
the presence of heterogeneous agents, the ESG uncertainty
2.2. A multi-asset economy term precludes fund separation because the incremental
portfolio, evolving from ESG uncertainty, is agent specific.
We move on to formulate an economy populated with In particular, consider the alternative decomposition of
I optimizing agents, N risky assets, and a riskless asset. We the optimal portfolio:
aim to derive an asset pricing model for the cross section −1
λr −1 −1 r μr λg −1 r μg
of equity returns in the presence of ESG uncertainty, while X ∗i = i + b , (16)
we also extend the analysis of portfolio selection.
γi −1
1 r μ r γi i i −1
1 r μg
We model the excess returns and ESG scores on N as-
where λr = 1 −1
r μr , λg = 1 −1
r μg , and i = I N +
sets as
b2i −1
r g
−1
+ 2bi r rg .
r˜ = μr + ˜ r , (12) The decomposition shows that each optimizing agent
g˜ = μg + ˜ g , (13) holds three portfolios: (1) a riskless asset, (2) the maxi-
mum Sharpe ratio portfolio in the risk-return space, and
where μr is an N-vector of expected excess returns and μg (3) the maximum Sharpe ratio portfolio in the risk-ESG
is an N-vector of expected ESG scores. The residuals from space. Note that ESG uncertainty affects the demand for
both equations are assumed to obey a 2N-variate normal risky assets through the N × N matrix i , which enters
distribution. The N × N covariance matrices of returns and both risky asset portfolios. If all agents have the same bi ,
ESG ratings are denoted by r and g , respectively, while then the matrix i is common to all agents and, there-
rg is the N × N cross-covariance matrix between r˜ and g˜ fore, a three-fund separation results. Otherwise, the two
with diagonal elements that are assumed to be positive. risky portfolios are agent specific and, hence, fund sepa-
Similar to Eq. (3), the agent maximizes an exponen-
ration does not apply in the setup of heterogeneous agents
tial utility function, V W ˜ i,1 , X i = −e−AiW˜ i,1 −BiWi,0 X i g˜ , where with ESG uncertainty.

W˜ i,1 = Wi,0 1 + r f + X r˜ is the terminal wealth and X i is
i
the N-vector of portfolio weights per investor i. 2.3. CAPM with ESG uncertainty
Proposition 1 describes the optimal portfolio in the
presence of multiple risky assets. The proof is in Online The next two propositions illustrate the cross-sectional
Appendix A.2. asset pricing implications of ESG preferences, first exclud-
647
ing and then accounting for ESG uncertainty. The proofs Proposition 3. With ESG uncertainty, the equilibrium expected
are in Online Appendices A.3 and A.4. excess returns of the risky assets are formulated as

Proposition 2. Excluding ESG uncertainty, the equilibrium ex- μr = βμM + βe f f − β μM − bM μg,U − βe f f μg,M,U ,
pected excess returns of the risky assets are given by (18)
where μM = γM σM,U
2 −b μ
M g,M,U is the equilibrium market
μr = βμM − bM μg − βμg,M , (17) r X M
premium, β = is the N-vector of the equilibrium CAPM
σM2
where μM = γM σM
2 −b μ
M g,M is the equilibrium market pre- beta, βe f f =
M,U X M
is the N-vector of effective beta, and
2 = X X
mium, σM σM,U
2
M r M is the market return variance, β = I −1 −1
r X M i=1 wi γi i,U
σM2
is the N-vector of market beta, μg,M = X M μg is the ag- −1
M,U = I −1 is the inverse of the covariance ma-
I i=1 wi γi
gregate market greenness, X M = i=1 wi X i is the N-vector of B M μg
trix of ESG-adjusted perceived asset returns. μg,U = bM
is
aggregate market positions in risky assets, γM is the aggre-
the perceived aggregate ESG scores of individual assets, where
gate risk aversion, and bM is the aggregate brown aversion. I
BM = ( Ii=1 wi γi−1 −1
i,U )
−1 −1 −1
i=1 wi γi bi i,U , and μg,M,U =

X M μg,U is the perceived aggregate market ESG score. Online
In the absence of ESG uncertainty, the expected excess
Appendix A.4 displays simplified expressions for asset pricing
return expression in Eq. (17) is identical to that derived
with ESG uncertainty assuming homogeneous agents.
by Pástor et al. (2021a), with the slight modification that
the market can depart from green neutrality. Expected re- The expected excess return expression in Eq. (18) mod-
turns are affected by ESG preferences through (1) the mod- ifies the no-uncertainty case in Eq. (17) by replacing the
ified market premium and (2) the alpha component that market beta with the effective beta. Thus, it is the effective
stands for excess return unexplained by βμM . Alpha de- beta that is priced in the cross section of equity returns. To
pends on the effective ESG score, i.e., the difference be- give perspective on the notion of effective beta, note that
tween the firm’s own ESG score and the market ESG score the perceived return on an arbitrary asset still consists of
multiplied by the stock’s beta. A numerical example is use- two components: (1) the actual return and (2) b times the
ful to illustrate. Assume a stock with β = 1.2 and μg,M = 2. ESG score of that asset. Because ESG scores for the market
As long as the ESG score is below 2.4, the stock has a pos- and individual assets are random, both the covariance and
itive alpha even when the stock is green. The threshold variance terms, used to define beta, depart from the stan-
value 2.4 reflects zero alpha, while alpha turns negative if dard return-based counterparts. The effective beta is based
the ESG score goes above the threshold. For instance, if the on ESG-adjusted returns. Collectively, expected excess re-
ESG score is 3 (2), the effective ESG score is 0.6 (−0.4), and turns on N risky assets are formulated as the sum of three
alpha is negative (positive). Altogether, as long as the mar- terms. The first term reflects exposure to market risk, as in
ket is not green neutral, it is not the firm’s own ESG score the standard CAPM. Then, the difference between the ef-
that dictates the sign and magnitude of alpha. Instead, it is fective beta and the market beta gives rise to the second
the effective ESG score. term. The third term accounts for the uncertainty-adjusted
In the presence of ESG preferences and certainty about effective ESG scores, analogously to Eq. (17) but using the
the ESG profile, the beta measuring exposure to total mar- effective beta instead.
ket risk, β, coincides with the CAPM beta. This is because, To provide further intuition on the beta-pricing spec-
as noted earlier, the perceived return on any security is ification, we consider a simplified case in which agents
equal to the sum of (1) the actual return and (2) the have homogeneous preferences (γ and b are equal across
pseudo-asset return that is proportional to the ESG score, agents). The effective beta can then be represented as
while the ESG score is nonrandom. Thus, in the absence of
σM2 b2 σg,M
2
2bσrg,M
ESG uncertainty, the covariance and variance terms used to βe f f = β + βg + βrg , (19)
define beta are unchanged. With uncertainty, the ESG score σM,U
2 σM,U
2 σM,U
2
is random; hence, the resulting beta is no longer identical g X M rg X M

where βg = , βrg = σrg,M , and σM,U
2 = σM
2 + b2 σ 2 +
to the standard CAPM beta. σg,M
2 g,M
As proposed by Pástor et al. (2021a), in the absence of 2bσrg,M . The effective beta is a weighted average of (1) the
ESG uncertainty, equilibrium expected returns compensate CAPM beta, β; (2) the ESG uncertainty beta, βg ; and (3)
for exposure to (1) the market risk factor and (2) an ESG- the ESG-return cross-covariance beta, βrg . The ESG uncer-
based factor. When ESG uncertainty is in play, fund sepa- tainty beta represents the comovement between the as-
ration no longer results; thus, expected returns cannot be set’s own ESG uncertainty and the market ESG uncertainty.
represented through a multifactor model. Instead, we pro- The cross-covariance beta represents the asset’s contribu-
pose a CAPM-type representation, in which expected ex- tion to the aggregate ESG-return cross covariance, σrg,M .
cess returns are expressed as the sum of two components: The weights in Eq. (19) reflect the relative contributions to
the first reflects the exposure to the market factor, while the perceived market return variance, i.e., the actual return,
the second is a nonzero alpha that stands for (1) nonpe- the ESG component, and the cross-covariance component.
cuniary benefits from ESG investing and (2) an additional The asset’s effective beta coincides with its market beta
risk premium attributable to ESG uncertainty. The follow- if preferences for ESG are muted (b = 0) or if the market is
ing proposition explains the equilibrium expected returns not subject to ESG uncertainty (σg,M = σrg,M = 0). To pro-
with ESG uncertainty, which is the core of our analysis. vide more intuition about the dependence of the effective
648
beta on ESG uncertainty, consider the case in which the 2.4. Demand and expected return in a two-asset economy
covariance matrix of ESG uncertainty, g , is diagonal with
elements σg,2 j ( j = 1, . . . , N), while rg is diagonal with el- In particular, to gain additional intuition about the de-
ements σrg, j . The effective beta of asset j can be written as mand for multiple risky assets and their equilibrium ex-
pected returns, it is useful to consider a simplified econ-
σM2 b2 σ 2 X j σ 2 2bσrg,M X j σrg, j omy consisting of two risky assets (along with a riskless
βe f f , j = β j + 2g,M 2 g, j + . (20) asset), both green and brown. In that economy, expected
σM,U
2 σM,U σg,M σM,U
2 σrg,M
excess returns are denoted by μr,green for the green stock
and μr,brown for the brown, the corresponding ESG scores
Given positive market weights in equilibrium, X j > 0,
are μg > 0 and −μg , the variances of the ESG scores are
the effective beta increases with the asset’s own ESG un-
σg,2green and σg,2brown , and the correlation between the scores
certainty, σg,2 j , and with the covariance between firm’s ESG
is assumed to be zero. Asset returns are assumed to be un-
and return, σrg, j , while it does not depend on the mean
correlated with identical variance denoted by σr2 . Finally,
ESG score. Interestingly, as long as the aggregate ESG un-
ESG scores are assumed to be positively correlated with
certainty is nonzero, a positive market beta asset with
returns of the same asset, with covariances denoted by
certain ESG profile (σg, j = σrg, j = 0) has an effective beta
σM2
σrg,green and σrg,brown . The expressions below follow from
equal to 2 β j , which is lower than the market beta
σM,U
Propositions 1 and 3. Online Appendix A.5 provides further
β j . This is because the asset contributes to the aggre- details.
gate return-based risk, but not to the aggregate ESG un- The two-asset optimal strategy is formulated as
certainty. 1 μr,green + bi μg
Xi,∗green = , (23)
We next analyze alpha variation with ESG uncer- γi σr2 + b2i σg,2green + 2bi σrg,green
tainty in the case of homogeneous agents. Combining
Eqs. (18) and (19), we show in Online Appendix A.4 that
1 μr,brown − bi μg
the CAPM alpha can be expressed as Xi,∗brown = . (24)
γi σr2 + b2i σg,2brown + 2bi σrg,brown
σ
b 2 2bσrg,M
2
α= βg − β +
g,M
βrg − β The optimal portfolio illustrates that, for ESG-sensitive
σ 2
M,U
σM,U
2
agents (bi > 0), demand falls with higher ESG uncertainty

× μM + bM μg,M − bM μg − βμg,M . (21) but rises with higher ESG scores. The notion is that when
targeting an ESG level, uncertainty about the precise ESG
The second term on the right-hand side of Eq. (21) is iden- profile should be accounted for. As in the single-asset
tical to that in Eq. (17) when ESG uncertainty is excluded. setup, the effect of ESG uncertainty is amplified by the
The first term represents the incremental effect of ESG un- positive correlation between return and the ESG score. For
certainty and is further analyzed below. For ease of inter- ESG-indifferent agents (bi = 0), the demand for green and
pretation, we assume again that g and rg are diagonal. brown stocks is equal to the mean-variance demand when
Then, it follows that ESG preferences are excluded.
We next formulate expected excess returns in equilib-
b2 σg,M
2 X j σg,2 j 2bσrg,M X j σrg, j
αj = − βj + − βj rium. We denote the fraction and brown aversion of ESG-
σM,U
2 σg,M
2 σM,U
2 σrg,M sensitive investors by wESG and bESG > 0, while the corre-
sponding parameters of ESG-indifferent agents are wIND =
× μM + bM μg,M − bM μg, j − β j μg,M . (22)
1 − wESG and bIND = 0. Assuming that all agents have the
Given positive market portfolio weights in equilibrium, same relative risk aversion γ , expected excess returns on
X j > 0, the asset alpha increases with ESG uncertainty, σg,2 j . the green and brown assets are formulated as
Likewise, alpha increases with the asset ESG-return cross
σ2 σ
covariance, σrg, j . Additionally, in the presence of aggregate βgreen γ σM2 1 + b2ESG g,σgreen
2 + 2bESG rg,σgreen
2 − wESG bESG μg
μr,green =
r r
ESG uncertainty, a positive market beta asset with zero ef- ,
σ 2
σ
1 + (1 − wESG ) b2ESG g,σgreen + 2bESG rg,σgreen
fective ESG score (μg, j − β j μg,M = 0) and with certain ESG r
2
r
2
profile has a negative alpha because its effective beta in (25)

Eq. (20) is smaller than its market beta, as noted earlier.
We have shown that both alpha and the effective σg,2brown

σ
βbrown γ σM2 1 + b2ESG σr2
+ 2bESG rg,σbrown
2 + wESG bESG μg
beta rise with ESG uncertainty. The analysis is based on μr,brown =
r
,
σ2 σ
the simplifying assumption of homogeneous brown-averse 1 + (1 − wESG ) b2ESG g,σbrown
2 + 2bESG rg,σbrown
2
r r
agents. Relaxing the homogeneity assumption, alpha and
(26)
beta variations with ESG uncertainty appear quite complex
to analyze analytically. However, in the calibration devel- where βgreen and βbrown are the equilibrium CAPM betas.
oped in Section 5.2, we consider heterogeneous agents in In the limiting case where wESG = 0 or bESG = 0, all agents
a two-asset economy (both brown and green) and show are ESG indifferent and equilibrium expected excess re-
that, even then, alpha and the effective beta do increase turns boil down to βgreen γ σM
2 and β
brown γ σM . In the oppo-
2
with ESG uncertainty. Below, we provide further analytical site extreme, where wESG = 1, expected return diminishes
results for the two-asset economy for ease of interpreta- with the ESG score and rises with ESG uncertainty. The lat-
tion. ter force can magnify the required return to the extent that
649
a green asset could, possibly, deliver higher return than a 2019. Our sample begins in 2002, as we require ESG rat-
brown asset. ings from at least two data vendors.
Otherwise, in the intermediate case in which both ESG-
sensitive and ESG-indifferent agents populate the econ-
omy, the expected return difference between the brown 3.2. Main variables
and the green assets diminishes with ESG uncertainty. To
see why, consider two assets with identical beta (βgreen = We focus on the overall ESG rating from each data
βbrown = β ) and ESG uncertainty (σg,green = σg,brown = σg provider, i.e., “ESG Combined Score” from Asset4, “ESG
and σrg,green = σrg,brown = σrg ). The expected return gap Rating” from MSCI IVA, “ESG Disclosure Score” from
(also the alpha gap) is given by Bloomberg, “Sustainalytics Rank” from Sustainalytics, and
“RobecoSAM Total Sustainability Rank” from RobecoSAM.11
2wESG bESG μg
μr,brown − μr,green = . For MSCI KLD data, we construct an aggregate ESG rating
σ2 σ
1 + (1 − wESG ) b2ESG σg2 + 2bESG σrg2 by summing all strengths and subtracting all concerns (e.g.,
r r
Lins et al., 2017; Berg et al., 2020).
(27) ESG rating agencies can differ in terms of their sam-
When all agents are ESG sensitive (wESG = 1 and bESG > ple coverage and rating scale. In the Online Appendix, we
0), the difference in expected returns is independent of report the number of U.S. common stocks covered by each
ESG uncertainty and equal to 2bESG μg . In other words, con- data vendor over time. In addition, Asset4, Bloomberg, Sus-
trolling for ESG uncertainty and beta, the expected return tainalytics, and RobecoSAM apply a scale from 0 to 100,
gap between the brown and the green assets is fixed, re- MSCI IVA uses a seven-tier rating scale from the best (AAA)
flecting the nonpecuniary benefits from holding green as- to the worst (CCC), and the MSCI KLD rating ranges from
sets. The return gap is nonexistent when either bESG = 0 or −11 to +19 in our sample. Panel B further demonstrates
wESG = 0, as all agents are ESG indifferent. that requiring a common sample covered by all data ven-
Otherwise, when ESG preferences are heterogeneous, dors could significantly reduce the sample size and shorten
the expected return gap monotonically decreases with σg2 the sample period. Therefore, we focus on pairwise ESG
rating disagreement and then average across all rater pairs.
and σrg .9 This suggests that ESG uncertainty could weaken
Note that the ESG uncertainty in our model is motivated by
the negative ESG-performance relation, as the asset de-
the fundamental difficulty and lack of consensus in mea-
mand of ESG-sensitive agents diminishes, which, in turn,
suring and interpreting the true ESG profile. The disagree-
implies lower aggregate nonpecuniary benefits from ESG
ment among ESG raters is largely due to the lack of con-
investing. In the limit, when ESG uncertainty grows with
sensus on the scope and measurement of ESG performance
no bound, the expected return gap between green and
(Berg et al., 2020), and, as a result, investors cannot reli-
brown assets approaches zero.
ably observe the firm’s true ESG profile and are exposed
to uncertainty in their sustainable investment. Hence, we
3. Data employ the disagreement among ESG raters as a proxy
for uncertainty about a firm’s ESG profile and label such
3.1. Data sources disagreement ESG uncertainty to be consistent with the
model terminology.
Our sample consists of all NYSE/AMEX/Nasdaq common Specifically, we obtain 14 rater pairs from the six data
stocks with share codes 10 or 11; daily and monthly stock providers.12 To achieve comparability across rating agen-
data are obtained from the Center for Research in Secu- cies, we proceed as follows. For each rater pair-year, we
rity Prices (CRSP). We collect ESG rating data from six sort all stocks covered by both raters according to the
data vendors, including Asset4 (Refinitiv), MSCI KLD, MSCI original rating scale of the respective data provider and
IVA, Bloomberg, Sustainalytics, and RobecoSAM. These data calculate the percentile rank (normalized between zero
providers represent the major players in the ESG rating and one) for each stock-rater pair. Then, for each stock,
market, and their ratings are widely used by practitioners we compute the pairwise rating uncertainty as the sam-
as well as in a growing number of academic studies (e.g., ple standard deviation of the ranks provided by the two
Eccles and Stroehle, 2018; Berg et al., 2020; Gibson et al., raters in the pair. Specifically, let g j,t,A and g j,t,B denote
2021). the ESG rank for stock j in year t from raters A and B, re-
Quarterly and annual financial statement data come spectively. The pairwise rating uncertainty is calculated as
from the Compustat database. Analyst forecast data come
from the Institutional Brokers’ Estimate System (I/B/E/S).
We also acquire quarterly institutional equity holdings
agers with more than $100 million U.S. dollars under discretionary man-
from the Thomson-Reuters Institutional Holdings (13F) agement. All holdings worth more than $20 0,0 0 0 U.S. dollars or 10,0 0 0
database.10 The full sample period ranges from 2002 to shares are reported in the database.
11
Although the Bloomberg ESG disclosure score measures the extent of
disclosure of ESG-related data by a company, it is positively associated
9
The no-uncertainty case leads to μr,brown − μr,green = 2bM μg , where with ESG quality due to the largely voluntary nature of ESG disclosure
bM = wESG bESG . requirements (López-de-Silanes et al., 2020).
10 12
The institutional ownership data come from money managers’ quar- There are 14 (instead of 15) rater pairs because MSCI KLD data are
terly 13F filings with the U.S. Securities and Exchange Commission (SEC). only available until 2015, while RobecoSAM data start in 2016, as shown
The database contains the positions of all institutional investment man- in the Online Appendix.
650
|g j,t,A√−g j,t,B | .13 For perspective, a company that is ranked by equity returns, we construct 25 equity portfolios indepen-
2 dently sorted on the ESG rating and rating uncertainty, and
two data providers at the 33rd and 59th percentiles would
report the average ESG rating, ESG rating uncertainty, and
generate a rating uncertainty of 0.18.
monthly return.
Finally, we compute the firm-level ESG rating uncer-
In addition, we examine the market ESG uncertainty
tainty as the average pairwise rating uncertainty across
throughout the sample period, as well as the time trend in
all rater pairs. Similarly, we compute the pairwise aver-
ESG uncertainty at the market and individual stock level.
age rank and then average across all rater pairs to obtain
While ESG data vendors do not provide a direct assess-
the firm-level ESG rating. Notably, the pairwise measure
ment for the market ESG profile, we evaluate the value-
has the advantage of maximizing the use of available rat-
weighted ESG score of the U.S. market by using firm-level
ing information while still preserving comparability across
ratings per the different vendors. To preserve comparabil-
raters, despite the difference in their sample coverage.14
ity across data vendors, we rely on the same pairwise mea-
In addition, investors may not have access to all six data
sures used at the single-stock level.15 For each stock-rater-
vendors; therefore, the average pairwise rating level and
year, we average the percentile ranks corresponding to the
rating uncertainty provide an approximate assessment for
specific rater across all rater pairs covering this stock. For
the perceived ESG profile and rating uncertainty among in-
each rater-year, we then value-weight firms’ ESG average
vestors. As a robustness check, we also consider alternative
percentile ranks to obtain a rater-specific market-level ESG
proxies for ESG rating (ESGALL ) and rating uncertainty (ESG
rating. Finally, for each year, using all rater-specific market
UncertaintyALL ) using all ESG ratings from all raters (instead
ESG ratings, we evaluate the aggregate market-level ESG
of rater pairs), without requiring common coverage, at a
rating and rating uncertainty as the pairwise mean and
given point in time. The Online Appendix provides a de-
standard deviation across raters.
tailed definition for each variable.
In Fig. 1, the top graph plots the time-series of the mar-
In the Online Appendix, we present the pairwise ESG
ket ESG ratings corresponding to each data vendor, and we
uncertainty and correlation of ESG ratings. The average cor-
observe significant dispersion across vendors. The bottom
relation across all rater pairs is 0.48 and ranges from 0.25
graph shows the time-series of market ESG uncertainty, as
to 0.71. MSCI KLD and MSCI IVA exhibit the lowest correla-
well as the equal- and value-weighted average of stock-
tion and the highest rating disagreement with other raters,
level ESG uncertainty. Stock-level ESG uncertainty, on av-
and the average correlation is 0.38 and 0.34, respectively.
erage, diminishes during the first half of the sample, as
On the other hand, ratings provided by Sustainalytics and
the number of raters increases and their coverage widens.
RobecoSAM are more correlated with those of other raters,
Stock-level uncertainty remains stable in the second sub-
and the average correlation is 0.59 and 0.56, respectively.
period. Focusing on the market, as ESG ratings are cor-
Our findings are largely consistent with the existing litera-
related across firms and vendors, the evidence indicates
ture and echo the growing concerns related to the lack of
that the market ESG uncertainty does consistently prevail
agreement across ESG rating agencies (e.g., Chatterji et al.,
throughout the entire sample period. This further supports
2016; Amel-Zadeh and Serafeim, 2018; Berg et al., 2020;
our intuition that ESG uncertainty could play an important
Gibson et al., 2021).
role in asset pricing.
The Online Appendix also reports the summary statis-
tics for the stock-level data used in the paper. We report
the mean, standard deviation, median, and quantile distri- 4. Investor demand, stock return, and alpha
bution of the annual ESG rating and ESG rating uncertainty
and other stock characteristics. The average ESG rating is 4.1. Investor demand
0.46, and the ESG rating uncertainty is 0.18. In addition, to
study the demand for risky assets and the cross section of We start with the first testable hypothesis gener-
ated from the model, i.e., investor demand for risky
assets increases with the ESG score, consistent with
13
To illustrate, consider two ratings g1 and g2 . The pairwise rating un- Pástor et al. (2021a), while it diminishes with ESG
(g1 − g1 +2 g2 ) +(g2 − g1 +2 g2 ) =
2 2
|g1√−g2 | rating uncertainty, as formulated in Proposition 1 and
certainty is given by 2−1 2
.
14
Unlike standard economic measures that are cardinal and can be di- Eqs. (23) and (24). We rely on institutional ownership
rectly compared, ESG scores are ordinal in nature. Thus, ESG scores are as a proxy for the demand for ESG investment, as
sensitive to the sample coverage considered by the particular data ven-
Krueger et al. (2020) find that institutional investors incor-
dor. As shown in the Online Appendix, ESG rating agencies differ in
their sample coverage; the stand-alone rank (e.g., 90th percentile) pro- porate ESG when forming their portfolios. While retail in-
vided by one rater may not be directly comparable to the correspond- vestors could still have ESG preference, it is highly costly to
ing figure from another rater if, for instance, one rater covers, on aver- obtain and analyze the ESG information, especially when
age, more green firms. To ensure comparability across all vendors cov- even the most specialized raters do not agree, on aver-
ering a stock, a proper experiment for determining the stock-level aver-
age, on the firm ESG profile. Due to the complex nature
age ESG rating and rating uncertainty is to narrow down the focus to
only those stocks jointly covered by all vendors. This experiment, how- of ESG investment, retail investors often rely on financial
ever, could considerably shrink the sample, which reflects the coverage institutions to achieve their ESG target, thereby making in-
intersection of all vendors providing a rating for the stock. In contrast,
the pairwise measure requires only a minimal set of restrictions on com-
15
mon coverage and, hence, allows us to explore the richness in ESG ratings In unreported analysis, we confirm that the alternative measurement
provided by each data vendor, while still preserving comparability across method described above (ESGALL and ESG UncertaintyALL ) provides similar
vendors. results.
651
Fig. 1. Market ESG ratings and ESG uncertainty, and average stock-level ESG uncertainty.
The top graph shows the time-series of the market ESG score obtained from each data vendor, as well as the mean ESG rating across data vendors.
The bottom graph shows the time-series of market ESG uncertainty, as well as the equal- and value-weighted average of stock-level ESG uncertainty.
Section 3.2 provides details on the construction of the variables.
stitutional ownership a reasonable source for investigat- type 5), following Abarbanell et al. (2003).16 We follow
ing the ESG demand. For instance, Hartzmark and Suss- Hong and Kacperczyk (2009) to consider types 1, 2, and 5
man (2019) show that once Morningstar published sus- as norm-constrained institutions. Our data on hedge fund
tainability ratings for mutual funds, there was a massive holdings are constructed by matching the 13F institutional
shift of fund flows from low-sustainability funds to high- holdings with a manually collected list of the names of
sustainability ones. A recent study on Robinhood investors hedge fund companies.17 The remaining institutions are
also shows that retail investors do not respond to ESG dis- mostly mutual funds.18
closures (Moss et al., 2020). The analysis proceeds as follows. At the end of each
To test the model predictions based on ESG-sensitive year t, we independently sort stocks into quintile portfo-
investors, it is also critical to account for the heterogeneity lios based on their ESG rating and rating uncertainty to
among institutions, as they are subject to different social generate 25 (5 × 5) portfolios. The low- (high-) ESG-rating
norm pressures and apply various strategies to make so- and ESG-rating-uncertainty portfolios comprise the bottom
cially responsible investments. For instance, pension funds, (top) quintile of stocks based on the ESG rating and ESG
universities, religious organizations, banks, and insurance rating uncertainty, respectively. For each type of institu-
companies are more norm-constrained than hedge funds tion, we compute the average institutional ownership in
or mutual funds that are natural arbitrageurs (Hong and
Kacperczyk, 2009). We therefore consider three distinct
groups: norm-constrained institutions, hedge funds, and 16
We thank Brian Bushee for making the institutional investor classifi-
other institutions. Specifically, we disaggregate the 13F in- cation data available via his website: https://accounting-faculty.wharton.
stitutional holdings based on institution type, including upenn.edu/bushee/.
17
bank trust (type 1), insurance company (type 2), invest- We thank Vikas Agarwal for generously sharing the data. A detailed
ment company (type 3), independent investment advisor description of the hedge fund list is provided by Agarwal et al. (2013).
18
While mutual funds and hedge funds are increasingly subject to social
(which includes hedge funds, type 4), and others (includ-
norm pressures, as shown by the rapid growth of ESG investment, some
ing corporate/private pension funds, public pension funds, could still prioritize financial returns at the cost of lower ESG standard.
university and foundation endowments, and miscellaneous, However, this remains an empirical question that we directly test.
652
Table 1
Institutional ownership of portfolios sorted by ESG rating and uncertainty.
At the end of year t, stocks are independently sorted into quintiles according to their ESG ratings and ESG rating uncertainty to generate 25 (5 × 5)
portfolios. The low- (high)-ESG-rating and ESG-rating-uncertainty portfolios comprise the bottom (top) quintile of stocks based on the ESG rating and ESG
rating uncertainty, respectively. For each of the 25 portfolios, we compute the average institutional ownership in each quarter in year t + 1 and rebalance
the portfolios at the end of year t + 1. Panel A reports the time-series averages of quarterly institutional ownership of norm-constrained institutions for each
of the 25 portfolios and the average difference in institutional ownership between high- and low-ESG-rating portfolios (“HML-R”), as well as between high-
and low-ESG-rating-uncertainty portfolios (“HML-U”). Panels B and C report similar statistics for average ownership of hedge funds and other institutions,
respectively. The Online Appendix provides a detailed definition for each variable. Newey-West adjusted t-statistics are shown in parentheses. Numbers
with “∗ ”, “∗ ∗ ”, and “∗ ∗ ∗ ” are significant at the 10%, 5%, and 1% levels, respectively.
Panel A: Norm-constrained institutions
ESG rating ESG uncertainty
Low 2 3 4 High HML-U t-stat All
Low 0.170 0.183 0.187 0.178 0.179 0.009 (0.80) 0.177

2 0.185 0.192 0.207 0.209 0.184 −0.001 (−0.23) 0.195
3 0.189 0.215 0.210 0.212 0.191 0.002 (0.40) 0.200
4 0.211 0.211 0.211 0.215 0.211 0.000 (0.04) 0.211
High 0.228 0.236 0.238 0.225 0.181 −0.047∗ ∗ ∗ (−2.73) 0.230
HML-R 0.058∗ ∗ ∗ 0.053∗ ∗ ∗ 0.050∗ ∗ ∗ 0.047∗ ∗ ∗ 0.002 0.053∗ ∗ ∗

(10.21) (12.00) (8.33) (8.51) (0.08) (11.39)
Panel B: Hedge funds

∗∗∗
Low 0.157 0.157 0.160 0.156 0.130 −0.027 (−3.70) 0.157
2 0.143 0.147 0.155 0.153 0.149 0.006 (1.31) 0.149
3 0.153 0.144 0.144 0.149 0.153 −0.000 (−0.08) 0.150
4 0.148 0.144 0.140 0.142 0.141 −0.006∗ (−1.96) 0.142
High 0.127 0.124 0.128 0.128 0.119 −0.008 (−1.33) 0.127
HML-R −0.031∗ ∗ ∗ −0.033∗ ∗ ∗ −0.032∗ ∗ ∗ −0.029∗ ∗ ∗ −0.011 −0.030∗ ∗ ∗

(−6.14) (−8.15) (−6.30) (−5.57) (−1.25) (−8.06)
Panel C: Other institutions

∗∗
Low 0.347 0.367 0.357 0.363 0.317 −0.030 (−2.57) 0.356
2 0.343 0.374 0.387 0.390 0.354 0.010 (1.43) 0.370
3 0.370 0.373 0.371 0.384 0.360 −0.011 (−1.66) 0.368
4 0.382 0.375 0.378 0.369 0.360 −0.022∗ ∗ ∗ (−3.25) 0.370
High 0.363 0.368 0.363 0.357 0.328 −0.035 (−1.63) 0.363
HML-R 0.016 0.001 0.006 −0.005 0.011 0.007

(1.28) (0.13) (0.59) (−0.37) (0.35) (0.71)
each quarter in year t + 1 for each of the 25 portfolios, they hold 23.0% of the green stocks (i.e., stocks in the top
and rebalance the portfolios at the end of year t + 1. We ESG rating quintile), indicating a 30% increase. Second, the
report the time-series averages of quarterly institutional ownership gap between low- and high-ESG-rating portfo-
ownership for each of the 25 portfolios and the average lios attenuates when rating uncertainty increases. When
difference in institutional ownership between high- and uncertainty is low, green stocks display 5.8% higher insti-
low-ESG-rating portfolios (“HML-R”) as well as between tutional ownership than brown stocks, while the owner-
high- and low-ESG-rating-uncertainty portfolios (“HML- ship gap declines to an insignificant 0.2% when rating un-
U”). The standard errors in all estimations are corrected certainty is high. More importantly, this pattern is due to
for autocorrelation using the Newey and West (1987) a decline in the demand for green firms when ESG un-
method. certainty is high, and the difference is statistically signif-
We tabulate the results in Table 1, with Panel A for icant and economically meaningful. For instance, among
the stock ownership from norm-constrained institutions, the high-ESG-rating portfolios, norm-constrained institu-
Panel B for hedge funds, and Panel C for other institutions hold 22.8% of the low-uncertainty stocks but only
tions. Several findings are worth noting in Panel A. First, 18.1% of the high-uncertainty stocks, indicating a 21% de-
as expected, norm-constrained institutions are in favor of cline. In line with our working hypothesis, demand for
greener firms. For instance, they hold 17.7% of the brown green firms from norm-constrained institutions diminishes
stocks (i.e., stocks in the bottom ESG rating quintile), while with ESG rating uncertainty, suggesting that rating uncer-
653
tainty matters the most for ESG-sensitive investors in their We assess return predictability using a conventional
ESG investment (i.e., green stocks).19 portfolio sort. In particular, at the end of each year t,
Panel B reports similar statistics for hedge fund owner- we sort stocks into quintile portfolios based on their ESG
ship. Hedge funds invest more in brown stocks on average, rating uncertainty. Within each rating uncertainty group,
e.g., they hold 15.7% of the brown stocks, but only 12.7% of we further sort stocks into quintile portfolios according to
the green stocks.20 The ownership gap between low- and their ESG ratings and generate 25 (5 × 5) portfolios.21 The
high-ESG-rating portfolios tends to diminish as ESG rating low- (high)-ESG-rating and ESG-rating-uncertainty portfo-
uncertainty rises. For high-uncertainty stocks, the owner- lios comprise the bottom (top) quintile of stocks based on
ship gap is no longer significant. Unlike the case of norm- the ESG rating and ESG rating uncertainty, respectively. For
constrained institutions, rating uncertainty mostly affects each of the 25 portfolios, we compute the value-weighted
hedge fund holdings for brown stocks. For instance, within return in each month in year t + 1 and rebalance the port-
the lowest rating group, hedge funds hold 15.7% of the folios at the end of year t + 1. Within each quintile of port-
low-uncertainty stocks, but 13.0% of the high-uncertainty folios sorted by ESG rating uncertainty, we also implement
stocks, indicating a 17% decline. Despite the different in- the zero-cost trading strategy by taking long positions in
centives for hedge funds to implement sustainable invest- the bottom quintile of stocks (lowest ESG rating) and sell-
ment, we continue to find that the rating uncertainty mat- ing short stocks in the top quintile (highest ESG rating).
ters the most for investors in their preferred investment The payoff of the long-short investment strategy is com-
universe. puted as the low (bottom quintile) minus high (top quin-
As shown in Panel C, we do not find strong ESG pref- tile) portfolio return (“LMH-R”), indicating the return pre-
erence among other institutions. Conditional on the level dictability of ESG ratings after controlling for rating uncer-
of ESG rating, we find evidence that rating uncertainty re- tainty. We then report the time-series averages of monthly
duces investor demand, while the economic magnitude is returns for each of the 25 portfolios and the long-short
much smaller than in the previously discussed subsamples strategy.
for norm-constrained institutions and hedge funds. In addition to raw portfolio returns, we report risk-
Overall, our findings support the model prediction that, adjusted returns from (1) the CAPM, i.e., only adjusting
for ESG-sensitive investors, demand for risky assets infor the market factor (MKT, defined as the excess re-
creases with the ESG score but diminishes with ESG rat- turn on the value-weighted CRSP market index over the
ing uncertainty. Our findings suggest that, although in- one-month Treasury bill rate); (2) the Fama-French-Carhart
stitutional investors are likely to be more sophisticated four-factor model (FFC), consisting of the market factor
and have access to privileged information, the uncertainty (MKT), the size factor (SMB, defined as small minus big
about corporate ESG profile remains an important barrier firm return premium), the book-to-market factor (HML, de-
to their investment. This could further limit their capac- fined as the high book-to-market minus the low book-to-
ity to engage in ESG issues and improve the ESG perfor- market return premium) (Fama and French, 1993), and the
mance of the firm (e.g., Dimson et al., 2015; Dyck et al., Carhart (1997) momentum factor (MOM, defined as the
2019; Chen et al., 2020; Krueger et al., 2020). As more in- winner minus loser return premium); and (3) the Fama-
stitutions seek sustainable investing, it is likely that ESG- French six-factor model (FF6), consisting of the market fac-
induced investor demand will play an even more promi- tor (MKT), the size factor (SMB), the book-to-market fac-
nent role in the future. tor (HML), the profitability factor (RMW, defined as the ro-
bust minus weak return premium), the investment factor
4.2. Cross-sectional return predictability (CMA, defined as the conservative minus aggressive return
premium), and the momentum factor (MOM) (Fama and
In line with Pástor et al. (2021a), our model pre- French, 2018).22 The standard errors in all estimations
dicts a negative relation between the ESG rating and are corrected for autocorrelation using the Newey and
CAPM alpha when there is no uncertainty in ESG rat- West (1987) method.
ings (Proposition 2). Negative return predictability stems Table 2 reports the results, with Panel A for raw re-
from nonpecuniary benefits from holding green stocks. turn and Panel B for CAPM-adjusted return. In the interest
However, the ESG-alpha relation is less clear in the pres- of brevity, we tabulate the results of FFC-adjusted return
ence of ESG uncertainty due to the conflicting forces and FF6-adjusted return in the Online Appendix and only
of the uncertainty-adjusted stock beta and ESG rating discuss the main findings in this subsection. Several find-
(Proposition 3). ings are worth noting. First, the ESG rating is negatively as-
sociated with future performance among stocks with low
rating uncertainty, and the long-short portfolio return is
19
Perhaps not surprisingly, investor demand is less affected among significant at 0.59% per month. Brown stocks (i.e., stocks
other ESG rating groups, as such investment may not be entirely ESG-
in the bottom ESG rating quintile) continue to outperform
driven; hence, the rating uncertainty plays a lesser role in asset allocation
decisions.
20
Note that hedge funds can take both long and short positions, hence
21
the long position per se may not fully reflect the ESG preference of hedge We employ a conditional sort to better control for rating uncertainty,
funds. Unreported results examine the net hedge fund ownership, defined while an independent sort yields similar findings, as shown in the Online
as the hedge fund ownership minus the short interest, where the short Appendix.
22
interest is computed as the number of shares held short scaled by the We thank Kenneth French for making the common factor returns
number of shares outstanding (Jiao et al., 2016). The net hedge fund own- available via his website: https://mba.tuck.dartmouth.edu/pages/faculty/
ership is 10.3% for brown stocks and 9.4% for green stocks. ken.french/data_library.html.
654
Table 2
Performance of portfolios sorted by ESG rating and uncertainty.
At the end of year t, stocks are first sorted into quintiles according to their ESG rating uncertainty. Within each ESG rating uncertainty group, stocks are
further sorted into quintiles according to their ESG ratings to generate 25 (5×5) portfolios. The low- (high)-ESG-rating and ESG-rating-uncertainty portfolios
comprise the bottom (top) quintile of stocks based on the ESG rating and ESG rating uncertainty, respectively. For each of the 25 portfolios, we compute
the value-weighted return in each month in year t + 1 and rebalance the portfolios at the end of year t + 1. Panel A reports the time-series averages of
monthly returns for each of the 25 portfolios, as well as for the investment strategy of going long (short) the low- (high)-ESG-rating stocks (“LMH-R”). The
column “All” reports similar statistics for portfolios sorted by ESG ratings only. The row “All” reports returns for portfolios sorted by ESG uncertainty only,
as well as the investment strategy of going long (short) the high (low) ESG-uncertainty stocks (“HML-U”). In Panel B, portfolio returns are further adjusted
by the CAPM. The Online Appendix provides a detailed definition for each variable. Newey-West adjusted t-statistics are shown in parentheses. Numbers
with “∗ ”, “∗ ∗ ”, and “∗ ∗ ∗ ” are significant at the 10%, 5%, and 1% levels, respectively.
Panel A: Return Panel B: CAPM-adjusted return
ESG rating ESG uncertainty ESG uncertainty
Low 2 3 4 High All Low 2 3 4 High All

∗∗∗ ∗∗∗ ∗∗ ∗∗ ∗∗ ∗∗ ∗
Low 1.235 1.113 0.767 0.875 0.760 0.923 0.168 0.064 −0.311 −0.141 −0.101 −0.101
(2.95) (2.99) (1.98) (2.30) (2.32) (2.58) (0.93) (0.40) (−1.82) (−0.89) (−0.58) (−0.84)
2 1.245∗ ∗ ∗ 1.026∗ ∗ ∗ 1.093∗ ∗ ∗ 1.043∗ ∗ ∗ 1.095∗ ∗ ∗ 0.963∗ ∗ ∗ 0.187 0.076 0.115 0.042 0.151 −0.008
(3.36) (2.84) (3.30) (2.74) (2.91) (2.85) (1.16) (0.38) (0.77) (0.29) (0.77) (−0.07)
3 1.096∗ ∗ ∗ 0.965∗ ∗ ∗ 1.050∗ ∗ ∗ 1.104∗ ∗ ∗ 0.949∗ ∗ ∗ 1.021∗ ∗ ∗ 0.040 −0.031 0.002 0.064 0.079 0.053
(2.69) (2.83) (2.86) (2.89) (3.15) (3.11) (0.23) (−0.20) (0.02) (0.46) (0.42) (0.64)
4 0.730∗ ∗ 0.695∗ 1.105∗ ∗ ∗ 1.019∗ ∗ ∗ 0.990∗ ∗ ∗ 1.017∗ ∗ ∗ −0.192 −0.389∗ ∗ ∗ 0.108 0.040 0.006 0.095
(2.09) (1.81) (2.90) (2.96) (2.68) (3.42) (−1.24) (−3.28) (0.55) (0.34) (0.03) (1.32)
High 0.642∗ 0.842∗ ∗ 0.855∗ ∗ ∗ 1.184∗ ∗ ∗ 0.854∗ ∗ ∗ 0.805∗ ∗ −0.230∗ −0.063 −0.012 0.245∗ −0.001 −0.095
(1.97) (2.53) (3.06) (3.62) (2.81) (2.57) (−1.95) (−0.55) (−0.10) (1.83) (−0.01) (−1.49)
LMH-R 0.594∗ ∗ ∗ 0.271 −0.088 −0.309 −0.094 0.118 0.398∗ 0.128 −0.299 −0.387∗ −0.100 −0.006
(2.72) (1.30) (−0.39) (−1.43) (−0.42) (0.78) (1.86) (0.58) (−1.25) (−1.75) (−0.42) (−0.04)
Low 2 3 4 High HML-U Low 2 3 4 High HML-U

∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗ ∗
All 0.753 0.875 0.935 1.083 0.940 0.187 −0.155 −0.090 −0.003 0.120 0.071 0.226∗
(2.31) (2.61) (3.07) (3.28) (3.29) (1.40) (−1.98) (−1.20) (−0.04) (1.72) (0.84) (1.67)
green stocks (i.e., stocks in the top ESG rating quintile) af- balance information on ESG scores and uncertainty when
ter adjusting for risk exposures, i.e., the long-short port- making portfolio decisions.
folio yields a CAPM-adjusted (FFC-adjusted, FF6-adjusted) Additionally, we consider a univariate portfolio sort
return of 0.40% (0.46%, 0.50%) per month.23 based on ESG uncertainty and report the results in the
Second, the negative return predictability of ESG ratings row titled “All”. Consistent with the model prediction, as
no longer holds for the remaining firms and even turns shown in Eq. (22), we find that when ESG uncertainty is in
positive in some cases. For perspective, we also consider play at the market level, stocks with low ESG uncertainty
a univariate portfolio sort based on ESG ratings and report carry a negative and statistically significant CAPM alpha of
similar statistics in the column titled “All”. The ESG rating −0.16% per month. As shown in the Online Appendix, the
does not predict stock returns for the full sample, which is result is also robust to FFC-adjusted and FF6-adjusted re-
consistent with the existing literature showing weak return turns. Furthermore, returns are increasing in ESG uncer-
predictability of the overall ESG rating (e.g., Pedersen et al., tainty, although the patterns are not always monotonic.
2021) and mixed evidence based on different ESG proxies For instance, the high-minus-low ESG uncertainty portfo-
(e.g., Gompers et al., 2003; Hong and Kacperczyk, 2009; lio (“HML-U”) shows a monthly CAPM alpha of 0.23% that
Edmans, 2011; Bolton and Kacperczyk, 2020). The empiri- is statistically significant at the 10% level, supporting the
cal evidence that ESG uncertainty can nontrivially interact model prediction that CAPM alpha increases with ESG rat-
with the ESG-performance relation is also consistent with ing uncertainty. Collectively, our findings support the pre-
Eq. (27). Our results further highlight the importance of diction that brown stocks outperform green stocks only
rating uncertainty, as it not only affects investor demand in the absence of rating uncertainty, and ESG uncertainty
but also has meaningful asset pricing implications, i.e., the could tilt this relation via conflicting forces, as illustrated
negative ESG-alpha relation only exists among stocks with in Proposition 3.
low rating uncertainty. The lack of consistency across ESG As a robustness check, we perform regression analy-
rating agencies could be a barrier for investors who have to sis to further control for other firm characteristics. Specif-
ically, we estimate the following monthly Fama and Mac-
Beth (1973) regression:
Perfi,m = α0 + β1 ESGi,m−1 + β2 ESGi,m−1
23
As our model is derived in market equilibrium, it is based on one
market factor. However, the economic magnitude and statistical signifi-
× Low ESG Uncertaintyi,m−1
cance in FFC-adjusted and FF6-adjusted returns reinforce our conclusion + β3 Low ESG Uncertaintyi,m−1
that accounting for rating uncertainty can be useful even for investors
who use multiple investment factors in their portfolio decisions. + β4 M i,m−1 + ei,m , (28)
655
Table 3
ESG rating, uncertainty, and stock returns.
This table presents the results of the following monthly Fama-MacBeth regressions, as well as their corresponding Newey-West adjusted t-statistics:

Perfi,m = α0 + β1 ESGi,m−1 + β2 ESGi,m−1 × Low ESG Uncertaintyi,m−1 + β3 Low ESG Uncertaintyi,m−1 + β4 M i,m−1 + ei,m ,
where Perfi,m refers to the excess return (models 1 to 4) or CAPM-adjusted return (models 5 to 8) of stock i in month m, ESGi,m−1 refers to the ESG rating,
Low ESG Uncertaintyi,m−1 refers to a dummy variable that takes a value of one if the ESG rating uncertainty is in the bottom quintile across all stocks in
that month and zero otherwise. The vector M stacks all other control variables, including the Log(Size), Log(BM), 6M Momentum, Log(Illiquidity), Gross
Profitability, Corporate Investment, Leverage, Log(Analyst Coverage) and Analyst Dispersion. The Online Appendix provides a detailed definition for each
variable. Numbers with “∗ ”, “∗ ∗ ”, and “∗ ∗ ∗ ” are significant at the 10%, 5%, and 1% levels, respectively.
Stock returns regressed on lagged ESG rating and uncertainty
Excess return CAPM-adjusted return
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8
ESG 0.002 0.098 0.062 0.199 0.042 0.139 0.162 0.301

(0.01) (0.65) (0.33) (1.03) (0.23) (0.91) (0.77) (1.65)
ESG × Low ESG Uncertainty −0.163∗ −0.223∗ −0.254∗ ∗ −0.312∗ ∗
(−1.91) (−1.75) (−2.26) (−2.36)
Low ESG Uncertainty 0.114∗ 0.109 0.125∗ ∗ 0.114
(1.86) (1.38) (2.20) (1.61)
Log(Size) −0.100 −0.036 −0.101 −0.038 −0.044 0.111 −0.044 0.111
(−1.28) (−0.27) (−1.30) (−0.29) (−0.59) (0.77) (−0.60) (0.77)
Log(BM) 0.001 0.009 −0.001 0.008 −0.021 0.019 −0.024 0.017
(0.01) (0.14) (−0.01) (0.12) (−0.19) (0.18) (−0.21) (0.17)
6M Momentum 0.336 0.188 0.335 0.194 0.275 0.105 0.276 0.111
(0.70) (0.40) (0.69) (0.42) (0.50) (0.20) (0.50) (0.21)
Log(Illiquidity) 0.056 0.056 0.103∗ ∗ 0.103∗ ∗
(1.00) (1.03) (2.17) (2.15)
Gross Profitability 0.178 0.180 0.355∗ 0.359∗
(0.99) (1.00) (1.83) (1.85)
Corporate Investment 0.037 0.037 −0.005 −0.007
(0.49) (0.50) (−0.08) (−0.09)
Leverage −0.037 −0.037 −0.034 −0.034
(−0.78) (−0.79) (−0.73) (−0.73)
Log(Analyst Coverage) −0.019 −0.019 −0.174 −0.175
(−0.15) (−0.14) (−1.40) (−1.41)
Analyst Dispersion −0.536∗ ∗ ∗ −0.539∗ ∗ ∗ −0.828∗ ∗ ∗ −0.831∗ ∗ ∗
(−2.67) (−2.71) (−4.37) (−4.37)
Constant 2.309∗ 1.800 2.281∗ 1.775 0.591 −0.555 0.533 −0.614
(1.71) (1.09) (1.70) (1.09) (0.46) (−0.31) (0.42) (−0.34)
Obs 283,671 254,873 283,671 254,873 272,728 245,451 272,728 245,451

R-squared 0.045 0.080 0.048 0.082 0.043 0.076 0.045 0.078
where Perfi,m refers to the excess return or CAPM-adjusted profiles and disagreement on firm fundamentals, such as
return of stock i in month m, ESGi,m−1 refers to the ESG analyst dispersion. Overall, we confirm the early results in
rating, and Low ESG Uncertaintyi,m−1 refers to a dummy the portfolio sort and provide supporting evidence for the
variable that takes a value of one if the ESG rating un- ESG-augmented CAPM after considering rating uncertainty.
certainty is in the bottom quintile across all stocks in that
month and zero otherwise. The vector M stacks all other 4.3. Additional analysis and robustness checks
control variables, including Log(Size), Log(BM), 6M Momen-
tum, Log(Illiquidity), Gross Profitability, Corporate Investment, Given the rapid growth in sustainable investing during
Leverage, Log(Analyst Coverage) and Analyst Dispersion. The the last decade (e.g., GSIA, 2018; PRI, 2020), we next as-
parameter of interest is β2 . Since the model predicts a neg- sess how our findings evolve over time. We then conduct
ative ESG-performance relation when there is no rating un- robustness checks using an alternative proxy for ESG rating
certainty, we should see a negative value of β2 . The On- and rating uncertainty.
line Appendix provides a detailed definition for each vari- We divide the full sample into two subperiods, 2003–
able. We also report Newey and West (1987) adjusted t- 2010 and 2011–2019, and repeat the main analysis. Table 4
statistics. has a layout similar to Table 1, in which Panels A, B, and
We tabulate the results in Table 3, with models 1 to C show the results for the norm-constrained institutions,
4 for excess return and models 5 to 8 for CAPM-adjusted hedge funds, and other institutions, respectively. First,
return. As expected, the ESG rating does not predict stock we confirm that for all three types of institutions, their
returns for the full sample. More importantly, the ESG rat- preference for green assets increases over time. Norm-
ing is negatively associated with future stock performance constrained institutions hold 12.3% of the brown stocks
when rating uncertainty is low. This relation is signifi- (i.e., stocks in the bottom ESG rating quintile), while they
cant across all regression specifications after controlling for hold 19.2% of the green stocks (i.e., stocks in the top ESG
other potential sources of uncertainty about corporate ESG rating quintile) in the post-2011 period, indicating a 56%
656
D. Avramov, S. Cheng, A. Lioui et al.
Table 4
Institutional ownership of portfolios sorted by ESG rating and uncertainty: Subsample analysis.
At the end of year t, stocks are independently sorted into quintiles according to their ESG ratings and ESG rating uncertainty to generate 25 (5 × 5) portfolios. The low- (high)-ESG-rating and ESG-rating-
uncertainty portfolios comprise the bottom (top) quintile of stocks based on the ESG rating and ESG rating uncertainty, respectively. For each of the 25 portfolios, we compute the average institutional ownership
in each quarter in year t + 1 and rebalance the portfolios at the end of year t + 1. Panel A reports the time-series averages of quarterly institutional ownership of norm-constrained institutions for each of the 25
portfolios and the average difference in institutional ownership between high- and low-ESG-rating portfolios (“HML-R”), as well as between high- and low-ESG-rating-uncertainty portfolios (“HML-U”). We divide
the full sample into two subperiods, and report results for 2003–2010 on the left and 2011–2019 on the right. Panels B and C report similar statistics for average ownership of hedge funds and other institutions,
respectively. The Online Appendix provides a detailed definition for each variable. Newey-West adjusted t-statistics are shown in parentheses. Numbers with “∗ ”, “∗ ∗ ”, and “∗ ∗ ∗ ” are significant at the 10%, 5%, and
1% levels, respectively.
Panel A: Norm-constrained institutions
2003–2010 2011–2019
Low 2 3 4 High HML-U t-stat All Low 2 3 4 High HML-U t-stat All
∗
Low 0.234 0.239 0.243 0.238 0.262 0.028 (1.59) 0.239 0.114 0.133 0.137 0.124 0.105 −0.009 (−1.77) 0.123
2 0.238 0.250 0.257 0.264 0.244 0.006 (0.76) 0.251 0.138 0.140 0.163 0.160 0.130 −0.008 (−0.93) 0.145
3 0.244 0.266 0.258 0.261 0.251 0.007 (1.09) 0.255 0.140 0.170 0.168 0.167 0.137 −0.003 (−0.54) 0.151
4 0.262 0.265 0.255 0.265 0.269 0.007 (1.06) 0.264 0.166 0.163 0.172 0.170 0.161 −0.006 (−0.71) 0.165
High 0.271 0.276 0.277 0.266 0.195 −0.076∗ ∗ (−2.41) 0.273 0.190 0.200 0.203 0.188 0.168 −0.022∗ ∗ ∗ (−3.65) 0.192
HML-R 0.037∗ ∗ ∗ 0.037∗ ∗ ∗ 0.034∗ ∗ ∗ 0.028∗ ∗ ∗ −0.067 0.034∗ ∗ ∗ 0.076∗ ∗ ∗ 0.068∗ ∗ ∗ 0.065∗ ∗ ∗ 0.064∗ ∗ ∗ 0.063∗ ∗ ∗ 0.069∗ ∗ ∗
(8.58) (8.60) (6.87) (7.18) (−1.67) (12.78) (20.93) (21.05) (10.36) (11.55) (10.22) (27.30)
Panel B: Hedge funds
2003–2010 2011–2019
657
∗∗∗ ∗
Low 0.133 0.128 0.136 0.127 0.097 −0.036 (−4.08) 0.129 0.179 0.183 0.181 0.182 0.160 −0.019 (−1.97) 0.181
2 0.117 0.110 0.115 0.119 0.107 −0.010∗ ∗ (−2.39) 0.113 0.166 0.180 0.189 0.184 0.187 0.021∗ ∗ ∗ (5.08) 0.182
3 0.104 0.114 0.115 0.106 0.112 0.008 (1.55) 0.111 0.196 0.170 0.169 0.187 0.189 −0.008 (−1.65) 0.184
4 0.093 0.100 0.096 0.098 0.091 −0.002 (−1.00) 0.096 0.196 0.183 0.178 0.182 0.186 −0.009∗ (−1.81) 0.183
High 0.086 0.087 0.090 0.088 0.066 −0.020∗ ∗ (−2.13) 0.088 0.163 0.157 0.162 0.162 0.166 0.003 (0.78) 0.162
HML-R −0.047∗ ∗ ∗ −0.041∗ ∗ ∗ −0.046∗ ∗ ∗ −0.039∗ ∗ ∗ −0.030∗ ∗ ∗ −0.042∗ ∗ ∗ −0.016∗ ∗ ∗ −0.026∗ ∗ ∗ −0.019∗ ∗ ∗ −0.019∗ ∗ 0.006 −0.019∗ ∗ ∗

(−13.62) (−7.73) (−7.20) (−13.21) (−3.17) (−12.15) (−3.08) (−6.31) (−5.25) (−2.56) (0.51) (−5.68)
Panel C: Other institutions
2003–2010 2011–2019
∗∗
Low 0.385 0.389 0.384 0.414 0.356 −0.029 (−1.54) 0.390 0.314 0.347 0.333 0.317 0.283 −0.031 (−2.42) 0.327
2 0.387 0.409 0.401 0.407 0.379 −0.008 (−0.97) 0.396 0.304 0.343 0.374 0.374 0.331 0.027∗ ∗ ∗ (3.11) 0.347
3 0.394 0.385 0.377 0.384 0.363 −0.031∗ ∗ ∗ (−6.31) 0.376 0.349 0.362 0.365 0.385 0.356 0.007 (1.02) 0.361
4 0.375 0.383 0.388 0.367 0.351 −0.024∗ ∗ (−2.64) 0.370 0.388 0.369 0.368 0.372 0.369 −0.019∗ ∗ (−2.19) 0.370
High 0.357 0.367 0.361 0.353 0.291 −0.066 (−1.57) 0.360 0.369 0.370 0.365 0.361 0.361 −0.008 (−1.09) 0.366
HML-R −0.028∗ ∗ −0.023 −0.023∗ ∗ −0.061∗ ∗ ∗ −0.065 −0.029∗ ∗ ∗ 0.055∗ ∗ ∗ 0.023∗ ∗ ∗ 0.032∗ ∗ ∗ 0.044∗ ∗ ∗ 0.078∗ ∗ ∗ 0.039∗ ∗ ∗
(−2.13) (−1.67) (−2.23) (−5.46) (−1.36) (−3.34) (10.32) (3.35) (3.62) (4.56) (5.44) (9.84)
increase. For perspective, they hold 14% more green stocks ically, for each rater-year, we sort all stocks covered by this
than brown stocks in the pre-2011 period. While hedge rater according to the original rating scale and calculate
funds invest more in brown stocks during both periods, the percentile rank (normalized between zero and one) for
they hold 33% less green stocks in the pre-2011 period each stock. The firm-level ESG rating is defined as the aver-
and only 10% less green stocks in the post-2011 period. In- age rank across all raters (labelled ESGALL ), and the ESG rat-
terestingly, other institutions exhibit a shift in ESG pref- ing uncertainty is defined as the standard deviation of the
erence over time, i.e., from brown-loving to green-loving. ranks provided by all raters (labelled ESG UncertaintyALL ).
They hold 7% less green stocks in the pre-2011 period, but As noted earlier, this method can entail some bias due to
12% more green stocks in the post-2011 period. the lack of comparability across vendors.
Second, for norm-constrained institutions, demand for We repeat our main analysis using the alternative proxy
green firms diminishes with ESG rating uncertainty in both for ESG rating and rating uncertainty, and present the re-
periods, while the effect is stronger in the pre-2011 pe- sults in the Online Appendix. First, we confirm that norm-
riod. Among the green stocks, norm-constrained institu- constrained institutions have a strong preference for green
tions hold 27.1% (19.0%) of the low-uncertainty stocks, but assets in general, while they display a lower demand for
19.5% (16.8%) of the high-uncertainty stocks, in the pre- green firms when ESG uncertainty is high. For instance,
2011 (post-2011) period, indicating a 28% (12%) decline. among the high-ESG-rating portfolios, norm-constrained
It is possible that the rising popularity of sustainable in- institutions hold 23.4% of the low-uncertainty stocks, but
vesting also incentivizes institutional investors to invest in only 15.5% of the high-uncertainty stocks, indicating a 33%
ESG research and even create internal rating tools (e.g., decline. As a result, green stocks no longer attract more
Mooney, 2019), partially mitigating the negative effect of norm-constrained institutional investors than brown stocks
rating uncertainty. Overall, our findings confirm that even when rating uncertainty is high.
with growing ESG awareness, the demand for green assets Moving to cross-sectional stock returns, our findings are
diminishes with ESG rating uncertainty for ESG-sensitive largely consistent with the model prediction that the ESG
investors. rating is negatively associated with future performance
When there is no uncertainty in ESG ratings, our model among stocks with low rating uncertainty. The long-short
predicts a negative relation between the ESG rating and portfolio return (FFC-adjusted return, FF6-adjusted return)
expected CAPM alpha due to the nonpecuniary benefits of is significant at 0.52% (0.35%, 0.35%) per month. While the
holding green stocks. However, Pástor et al. (2021a,b) show CAPM-adjusted return is not statistically significant, the
that green assets have higher realized alphas when in- magnitude is sizable at 0.31% per month. Unreported re-
vestors’ tastes for green holdings shifted unexpectedly dur- sults show that the long-short portfolio yields a return of
ing the last decade. As a result, we expect our findings 1.05% per month and a CAPM-adjusted (FFC-adjusted, FF6-
to be stronger in the pre-2011 period, which provides a adjusted) return of 0.87% (0.75%, 0.73%) per month in the
cleaner setting in which to analyze the equilibrium ex- pre-2011 period, all statistically significant at the 5% or 1%
pected returns of stocks. level. We further confirm that ESG rating is negatively as-
Table 5 has a layout similar to Table 2, with Panel A sociated with CAPM-adjusted return when rating uncer-
for raw return and Panel B for CAPM-adjusted return. As tainty is low, after controlling for other firm characteris-
expected, the ESG rating is negatively associated with fu- tics. In short, our main results are robust to the alternative
ture performance among stocks with low rating uncer- definitions of ESG rating and rating uncertainty.
tainty in the pre-2011 period, yielding a significant long-
short portfolio return (“LMH-R”) of 1.12% (t-stat=3.06) per 5. Calibration
month and CAPM-adjusted return of 0.96% (t-stat=2.81)
per month. Consistent with our model prediction, the neg- As final experiments, we calibrate the model to study
ative ESG-CAPM alpha relation does not hold for the re- the general equilibrium implications of ESG rating uncer-
maining firms. A univariate portfolio sort based on ESG un- tainty for the market premium, the cross section of stock
certainty further confirms that CAPM alpha increases with returns, economic welfare, and equity demand. Following
ESG rating uncertainty, i.e., the high-minus-low ESG uncer- Pástor et al. (2021a), we consider ESG-indifferent (IND)
tainty portfolio (“HML-U”) shows a monthly CAPM alpha of and ESG-sensitive (ESG) agents. The former group does not
0.42% (t-stat=2.04) in the pre-2011 period. derive utility from ESG externalities (i.e., bIND = 0), while
In contrast, we do not find a negative return pre- the utility of the latter positively depends on the mar-
dictability of ESG ratings across all ESG-rating-uncertainty ket ESG score and negatively depends on rating uncer-
portfolios or a positive ESG uncertainty-CAPM alpha rela- tainty, through bESG > 0. Specifically, we assume that 20%
tion in the post-2011 period. Our findings in both subpe- of the agents have ESG preferences, while the remain-
riods remain unchanged for FFC- and FF6-adjusted return, ing fraction consists of ESG-indifferent agents. Hence, ESG-
as reported in the Online Appendix. Note that our results sensitive agents are not the vast majority in the economy,
should not be interpreted to mean that ESG rating uncer- yet they account for a substantial fraction.24
tainty no longer matters in the future. Instead, the equilib- The ESG parameters, bESG , μg,M , σg,M , ρg,M , and the
rium outcome over longer horizons could be even stronger stock-level counterparts of μg,M , σg,M , and ρg,M are un-
than the full sample evidence we report, due to the unex- known. In the data section above, we describe ways to map
pected outcomes realized over the last decade.
Next, we conduct robustness checks by using alterna- 24
In unreported results, we confirm that a larger fraction of ESG-
tive definitions of ESG rating and rating uncertainty. Specif- sensitive investors leads to stronger implications of ESG uncertainty.
658
Table 5
D. Avramov, S. Cheng, A. Lioui et al.

Performance of portfolios sorted by ESG rating and uncertainty: Subsample analysis.
At the end of year t, stocks are first sorted into quintiles according to their ESG rating uncertainty. Within each ESG rating uncertainty group, stocks are further sorted into quintiles according to their ESG ratings
to generate 25 (5×5) portfolios. The low- (high)-ESG-rating and ESG-rating-uncertainty portfolios comprise the bottom (top) quintile of stocks based on the ESG rating and ESG rating uncertainty, respectively.
For each of the 25 portfolios, we compute the value-weighted return in each month in year t + 1 and rebalance the portfolios at the end of year t + 1. Panel A reports the time-series averages of monthly returns
for each of the 25 portfolios, as well as for the investment strategy of going long (short) the low- (high)-ESG-rating stocks (“LMH-R”). The column “All” reports similar statistics for portfolios sorted by ESG
ratings only. The row “All” reports returns for portfolios sorted by ESG uncertainty only, as well as the investment strategy of going long (short) the high (low) ESG-uncertainty stocks (“HML-U”). We divide the
full sample into two subperiods, and report results for 2003–2010 on the left and 2011–2019 on the right. In Panel B, portfolio returns are further adjusted by the CAPM. The Online Appendix provides a detailed
definition for each variable. Newey-West adjusted t-statistics are shown in parentheses. Numbers with “∗ ”, “∗ ∗ ”, and “∗ ∗ ∗ ” are significant at the 10%, 5%, and 1% levels, respectively.
Panel A: Return
2003–2010 2011–2019
Low 1.427∗ 0.845 0.528 0.949 0.667 0.773 1.065∗ ∗ ∗ 1.351∗ ∗ ∗ 0.980∗ ∗ 0.809∗ ∗ 0.842∗ ∗ 1.056∗ ∗ ∗
(1.86) (1.35) (0.77) (1.43) (1.23) (1.23) (2.93) (3.25) (2.52) (2.00) (2.26) (2.92)
2 1.235∗ 0.973 0.955∗ 0.984 0.902 0.957 1.254∗ ∗ ∗ 1.073∗ ∗ ∗ 1.215∗ ∗ ∗ 1.096∗ ∗ ∗ 1.266∗ ∗ ∗ 0.968∗ ∗ ∗
(1.83) (1.44) (1.75) (1.53) (1.34) (1.64) (3.61) (3.42) (3.19) (2.68) (3.55) (2.79)
3 0.944 1.014∗ 0.919 1.157∗ 0.879∗ 0.764 1.231∗ ∗ ∗ 0.921∗ ∗ 1.166∗ ∗ ∗ 1.057∗ ∗ ∗ 1.011∗ ∗ ∗ 1.249∗ ∗ ∗
(1.26) (1.74) (1.43) (1.74) (1.70) (1.33) (3.53) (2.55) (3.20) (2.80) (2.94) (3.83)
4 0.497 0.502 0.928 0.763 1.108∗ 0.976∗ 0.937∗ ∗ 0.868∗ ∗ 1.262∗ ∗ ∗ 1.247∗ ∗ ∗ 0.884∗ 1.054∗ ∗ ∗
(0.86) (0.73) (1.29) (1.22) (1.91) (1.87) (2.52) (2.40) (4.15) (4.43) (1.92) (3.62)
High 0.309 0.346 0.524 1.205∗ ∗ 0.619 0.420 0.937∗ ∗ ∗ 1.283∗ ∗ ∗ 1.150∗ ∗ ∗ 1.166∗ ∗ ∗ 1.062∗ ∗ ∗ 1.147∗ ∗ ∗
(0.52) (0.57) (1.08) (2.05) (1.18) (0.75) (3.36) (5.14) (3.98) (3.78) (3.38) (4.26)
LMH-R 1.119∗ ∗ ∗ 0.499∗ 0.004 −0.256 0.048 0.353 0.127 0.068 −0.170 −0.357 −0.220 −0.091
(3.06) (1.78) (0.01) (−0.74) (0.12) (1.45) (0.59) (0.23) (−0.70) (−1.22) (−0.87) (−0.50)
659

All 0.482 0.533 0.666 1.011∗ 0.832∗ 0.350 0.994∗ ∗ ∗ 1.180∗ ∗ ∗ 1.174∗ ∗ ∗ 1.146∗ ∗ ∗ 1.037∗ ∗ ∗ 0.043
(0.81) (0.87) (1.25) (1.70) (1.71) (1.51) (3.50) (4.41) (3.97) (3.92) (3.35) (0.31)
Panel B: CAPM-adjusted return

2003–2010 2011–2019

Low 0.568∗ ∗ 0.058 −0.284 0.162 0.011 −0.006 −0.147 0.022 −0.376∗ −0.433∗ ∗ −0.262 −0.224
(1.99) (0.27) (−0.91) (0.68) (0.04) (−0.03) (−0.67) (0.10) (−1.95) (−2.14) (−1.17) (−1.52)
2 0.397∗ ∗ 0.172 0.236 0.222 0.175 0.218 0.023 0.086 −0.083 −0.162 0.098 −0.259∗ ∗
(2.00) (0.56) (0.98) (1.18) (0.55) (1.45) (0.09) (0.32) (−0.54) (−0.79) (0.43) (−2.06)
3 0.088 0.259 0.133 0.358∗ 0.237 0.034 0.061 −0.344∗ −0.169 −0.228 −0.158 0.019
(0.32) (1.14) (0.79) (1.89) (0.81) (0.25) (0.27) (−1.89) (−1.09) (−1.14) (−0.71) (0.17)
4 −0.192 −0.348∗ 0.109 −0.049 0.381∗ 0.243∗ ∗ −0.264 −0.408∗ ∗ ∗ 0.178 0.204 −0.418 −0.035
(−0.86) (−1.70) (0.26) (−0.25) (1.82) (2.34) (−1.22) (−2.94) (1.49) (1.27) (−1.23) (−0.37)
High −0.391∗ ∗ −0.403∗ ∗ −0.159 0.461∗ −0.030 −0.294∗ ∗ ∗ −0.070 0.310∗ ∗ ∗ 0.110 0.052 −0.037 0.093∗
(−2.06) (−2.29) (−0.80) (1.98) (−0.12) (−2.92) (−0.56) (3.02) (1.02) (0.32) (−0.23) (1.83)
LMH-R 0.959∗ ∗ ∗ 0.460 −0.126 −0.299 0.041 0.289 −0.077 −0.288 −0.486∗ −0.486 −0.225 −0.317∗
(2.81) (1.60) (−0.30) (−0.85) (0.10) (1.12) (−0.31) (−1.06) (−1.88) (−1.62) (−0.81) (−1.74)

∗ ∗∗ ∗ ∗∗ ∗
All −0.238 −0.255 −0.068 0.243 0.181 0.419 −0.077 0.117 0.048 0.027 −0.107 −0.029
(−1.87) (−2.35) (−0.45) (1.93) (1.33) (2.04) (−0.95) (1.74) (0.79) (0.32) (−1.07) (−0.19)
ESG ratings into scores for individual securities, and the i.e., the nonpecuniary benefits from holding the green mar-
market-level ESG rating follows through aggregation. The ket versus aversion to ESG uncertainty. For ESG-sensitive
resulting quantities are not on the scale of equity returns agents, we consider two values for brown aversion, namely,
and are ordinal in nature. In particular, a higher ESG rat- bESG is equal to 1 or 2. When the market is green, both
ing indicates a greener stock, while a higher standard de- cases generate an ESG return of 1% and 2% per year, re-
viation among raters amounts to greater ESG uncertainty. spectively. When the market is green neutral, brown aver-
Thus, stock-level and market-level ratings, as well as mea- sion is not mapped into the incremental expected return.
sures of rating uncertainty, can comfortably be used to as- We also consider two values for the correlation between
sess the model implications through cross-sectional regres- ESG and market return, ρg,M , namely, 0 and 0.5. The zero-
sions and portfolio sorts. In the calibration experiments correlation is a benchmark case that reflects the lower
that follow, we choose ESG parameters that conform to bound on the implications of ESG uncertainty. The positive
payoffs on pseudo-assets, as formulated in the theory sec- correlation is sensible, as described in the theory section.
tion.25 Further details are provided below. Finally, the market ESG uncertainty, σg,M , ranges between
0 and 0.04.26
5.1. Market premium, welfare, and equity demand Panel A of Fig. 2 describes the green-neutral market
case, with solid lines representing the case of ρg,M = 0
The analysis for the aggregate market is based on an and dashed lines corresponding to ρg,M = 0.5. The lim-
economy that consists of the market portfolio and a risk- iting case of bESG = 0 represents the departure point, at
less asset (in zero net supply). The market volatility pa- which all agents are indifferent to the market ESG pro-
rameter employed in the calibration is σM = 15.19%, which file. In that case, it follows that (1) the equilibrium market
is the annual estimate from monthly U.S. market returns, premium equals the ESG-indifference value, γ σM 2 = 6.50%,
spanning the period from July 1963 through December regardless of the level of ESG uncertainty; (2) both agent
2019. Then, employing the sample estimate for the equity types hold the market portfolio (x∗ESG = x∗IND = 1); and (3)
premium (6.5%), we obtain γ = 2.81, following Eq. (6). Two the agents perceive the same certainty equivalent excess
remarks are in order. First, while our sample for individual σ2
return (CE ESG = CE IND = γ 2M = 3.25%).
stocks starts in 2002, due to limited data for ESG ratings,
When bESG > 0, the ESG agents are sensitive to the mar-
the possibility of using longer return histories from the ag-
ket rating uncertainty. Then, the perceived market variance
gregate to sharpen estimates builds on Pástor and Stam-
σM,U
2 is higher than σM 2 .27 This force leads to an increasing
baugh (2002). In addition, expected market return is en-
equilibrium market premium, and more so for higher val-
dogenous in our setup, while the sample estimate is used
ues of bESG , σg,M , and ρg,M .
to set the risk aversion parameter.
As a result, the two types of agents have different cer-
We evaluate the equilibrium market premium on the
tainty equivalent return and demand for the market port-
basis of Eq. (11) for the multiple-agent case. The mar-
folio. On the one hand, the IND agents are not sensitive
ket demand and the certainty equivalent return from in-
to ESG uncertainty (σIND 2
,U = σM ). Thus, they benefit from
2
vestment differ across agent types. In particular, based on
the higher equilibrium market premium, which translates
Eq. (4), the optimal market demand for agent i is x∗i =
into a higher certainty equivalent return and a levered po-
1 μM +bi μg,M
γi , where σi,U
2 = σ 2 + b2 σ 2 + 2b σ σ
i M g,M ρg,M . In sition in the market portfolio (x∗IND > 1). On the other hand,
σi,U
2 M i g,M
addition, as derived in Online Appendix A.6, the certainty the ESG agents are more sensitive to ESG uncertainty than
equivalent excess return for agent i is given by CE i = the aggregate market (σESG 2
,U
> σM,U
2 ). Thus, their certainty
equivalent return and their demand for the market portfo-

1 μM +bi μg,M 2
2γi σi,U . Both the market demand and the cer- lio decline with increasing values of bESG , σg,M , and ρg,M .
tainty equivalent return increase in the perceived market We next quantitatively assess the economic cost of ESG
premium and diminish in the perceived market variance. uncertainty, as perceived by ESG agents. The cost is rep-
For ESG-sensitive agents, the perceived certainty equivalent resented by a diminishing certainty equivalent return rela-
return increases with the market ESG score, while the per- tive to σg,M = 0. When ρg,M = 0 and ESG uncertainty σg,M
ceived variance rises with ESG uncertainty and the corre- is set to 0.02 (0.04), the utility loss is 0.03% (0.13%) per
lation between the market ESG score and market return. year for bESG = 1 and 0.13% (0.47%) for bESG = 2. Consider-
The effect of ESG rating uncertainty is stronger for higher
values of bi and ρg,M .
26
To make the analysis sufficiently comprehensive, we Empirically, the magnitude of ESG uncertainty is comparable to the
scale of differences in ESG scores. For instance, considering the summary
run calibration experiments for multiple scenarios. First, statistics of our data set from the Online Appendix, the quartile devia-
we consider both green-neutral (μg,M = 0) and green tion of ESG ratings is 0.14. The values of ESG uncertainty are of the same
(μg,M = 0.01) markets. The ESG implications of the former order of magnitude as differences in ESG scores: the median ESG uncer-
case are exclusively attributed to ESG uncertainty. The lat- tainty is 0.16, while the 90th percentile is 0.33. Similarly, for calibration,
we consider values of ESG uncertainty that conform to ESG levels: a green
ter case involves the two conflicting forces noted earlier,
(brown) asset has a mean ESG score of 0.01 (−0.01), and ESG uncertainty
is of the order of 0.01 and multiples.
27
25
In our model, the g = 0 case reflects green neutrality. Having this ref- As we derive in Online Appendix A.1, the perceived aggregate mar-
erence point, all the model implications are invariant to a multiplicative ket variance, σM,U2
, is a harmonic weighted average of the market vari-
scaling of ESG ratings and rating uncertainty, as long as the brown aver- ances perceived by the agents, which in our example are σM, 2
IND = σM and
2
sion parameter is also scaled such that the pseudo return, bg, remains σM,2
ESG = σM + bESG σg,M + 2bESG σM σg,M ρg,M . It follows that σM,IND < σM,U <
2 2 2 2 2
unchanged. σM,
2
ESG .
660
Fig. 2. Equilibrium equity premium, certainty equivalent return, and market demand.
This figure shows the equilibrium market premium (μM ), the certainty equivalent excess return for ESG-sensitive (CE ESG ) and ESG-indifferent (CE IND ) agents,
the optimal market participation (x∗ESG and x∗IND ), and their variation with the market ESG uncertainty, σg,M . The relative risk aversion, γ , is 2.81, and the
market volatility, σM , is 15.19%. ESG-sensitive agents represent a fraction of wESG = 20% of the population and have a brown aversion bESG = {0, 1, 2}.
ESG-indifferent agents represent wIND = 80% of the population and have a brown aversion bIND = 0. The correlation between the market return and the
ESG score, ρg,M , is 0 (solid lines) or 0.5 (dashed lines). Panel A focuses on a green-neutral market (μg,M = 0), while Panel B describes a green market
(μg,M = 0.01). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
661
ing ρg,M = 0.5 instead, the corresponding figures are 0.26% sumed that βgreen = βbrown = 1, and the idiosyncratic an-
(0.55%) for bESG = 1 and 0.55% (1.08%) for bESG = 2. The cal- nualized return volatility is 20% for both assets. As σM =
ibrated utility loss accounts for a nontrivial proportion of 15.19%, the total stock return volatility is 25.12%.28 We
the overall certainty equivalent excess return when com- consider a positive correlation between return and ESG
pared to the benchmark case of no uncertainty, i.e., 3.25%. score for each asset, setting ρg,M = ρrg,green = ρrg,brown =
Therefore, from the perspective of ESG agents, ESG uncer- 0.5. The off-diagonal elements in g and rg are assumed
tainty leads to significant utility loss. to be zero.
When the market is green neutral, preferences for ESG Fig. 3 illustrates how the expected excess return, the
essentially reduce welfare because the only effect that CAPM alpha, and the effective beta vary with ESG uncer-
comes into play is aversion to ESG uncertainty. Depart- tainty for green and brown assets (σg,green and σg,brown ).
ing from a green-neutral market, the nonpecuniary bene- The solid lines represent the green asset while dashed
fits from holding green stocks intervene, and more so for lines represent the brown asset. We consider a market-
higher values of brown aversion and market ESG score. wide ESG uncertainty, σg,M , equal to 0.01 for the left
Panel B of Fig. 2 describes the green-market case, with graphs and 0.02 for the right graphs. Starting from the
solid lines corresponding to ρg,M = 0 and dashed lines to benchmark case of ESG indifference (bESG = 0), the ex-
ρg,M = 0.5. In the absence of ESG uncertainty (σg,M = 0) pected excess return for both assets is equal to the market
and when bESG > 0, the equilibrium market premium di- premium, 6.50%, while the alpha is zero and the effective
minishes with bESG . This translates into a lower certainty beta coincides with the unit market beta.
equivalent return and market demand for IND agents, who Considering ESG-sensitive agents (bESG > 0), the pos-
confront a lower market premium but do not extract non- itive ESG score of a green asset is associated with
pecuniary benefits from holding the green market. In con- lower expected return and alpha in equilibrium, as in
trast, ESG agents extract nonpecuniary benefits from the Pástor et al. (2021a). The effect is stronger for larger val-
positive market ESG tilt, which leads to a higher certainty ues of bESG . In addition, expected return rises with ESG un-
equivalent return and higher market demand for increasing certainty. Thus, in the presence of the conflicting forces of
values of bESG . ESG score (negative effect on alpha) and ESG uncertainty
As the parameter σg,M captures the trade-off between (positive effect on alpha), a green asset with high ESG
the two conflicting forces of ESG preferences, we derive a uncertainty could have higher expected return and alpha
break-even value of σg,M when the utility loss of ESG un- than a brown asset with low ESG uncertainty. For instance,
certainty entirely offsets the benefits from holding green when σg,M = 0.01, σg,green = 0.10, and bESG = 1 (bESG = 2),
stocks. When ρg,M = 0 and bESG is 1 (2), the welfare ben- the green asset displays an expected excess return of 6.78%
efits of a green market perceived by ESG agents vanish, (7.09%) and an alpha of 0.20% (0.42%). To compare, when
due to ESG uncertainty, for σg,M = 9.9% (7.2%), well above the ESG profile of the brown asset is known for certain, its
reasonable values. However, a positive correlation between expected excess return is 6.70% (6.90%) and alpha is 0.11%
market return and ESG rating amplifies the effects of ESG (0.23%).
uncertainty. When ρg,M = 0.5 and bESG is 1 (2), the thresh- The σg,green = 0 case merits further analysis. The zero-
old σg,M is much lower at 4.9% (4.3%). uncertainty asset does not contribute to the aggregate
The market premium is also subject to the two conflict- ESG uncertainty; thus, its effective beta is lower than the
ing forces, i.e., the negative ESG premium due to the green unit market beta, per Eq. (19), and the effect is stronger
market versus the positive contribution due to ESG uncer- when brown aversion and market-wide ESG uncertainty
tainty. When ρg,M = 0 and bESG is 1 (2), the two forces are are higher. For instance, when σg,M = 0.01, the effective
equal for σg,M = 6.0% (4.2%), while if ρg,M = 0.5 and bESG is beta is 0.987 (0.974) for bESG = 1 (bESG = 2). When σg,M =
1 (2), the threshold σg,M is at 2.1% (1.9%). 0.02, the effective beta is 0.974 (0.950) for bESG = 1 (bESG =
Overall, we reinforce the notion that ESG uncertainty 2). The diminished effective beta relative to the market
increases the market premium, as well as reduces the eco- beta induces a negative contribution to alpha and expected
nomic welfare for ESG-sensitive investors and discourages return.
their participation in the stock market. As demonstrated in Eq. (20), the effective beta does not
depend on the mean ESG score. Consequently, green and
5.2. Cross section of expected returns, alpha, and effective brown assets have the same effective beta for identical lev-
beta els of ESG uncertainty. The effective beta increases with
ESG uncertainty and can rise above the unit market beta,
We next calibrate the cross section of expected return, and the effect is stronger for higher values of brown aver-
the CAPM alpha, and the effective beta in equilibrium, all sion.
of which are formulated in Section 2.3. Finally, as long as the green and the brown assets have
To distill cross-sectional implications of ESG uncer- the same ESG uncertainty, the performance difference be-
tainty, we focus on the green-neutral market described tween brown and green assets (both expected return and
in Section 5.1. At the stock level, we consider green and
brown assets, with mean ESG scores μg,green = 0.01 and
28
The total return variance of the green asset, σgreen 2
, is given by
μg,brown = −0.01. Thus, for the green asset, ESG agents per- βgreen
2
σM2 + σid2 ,green , where σid,green is the idiosyncratic volatility. For βgreen =
ceive an incremental ESG return equal to 1% per year for 1, σM = 15.19%, and σid,green = 20%, it follows that σgreen = 25.12%. The
bESG = 1 and 2% per year for bESG = 2. The corresponding same applies to σbrown 2
. The covariance between returns is βgreen βbrown σM2 =
return figures are negative for the brown asset. It is as- (15.19% )2 , corresponding to a correlation ρgreen,brown = 36.59%.
662
Fig. 3. Two-asset pricing equilibrium: Expected stock return, alpha, and effective beta.
Considering the green-neutral market described in Fig. 2, Panel A, for green (solid lines) and brown (dashed lines) assets, this figure displays the equilibrium
expected excess stock return, (μr,green and μr,brown ), the CAPM alpha, (αgreen and αbrown ), the effective beta, (βe f f ,green and βe f f ,brown ), and their variation with
ESG uncertainty, σg,green , σg,brown . The mean ESG scores of the two assets are μg,green = 0.01 and μg,brown = −0.01. The market betas of the two assets
are βgreen = βbrown = 1, while their idiosyncratic return volatility is equal to 0.2. The correlation between return and the same-asset ESG score is ρg,M =
ρrg,green = ρrg,brown = 0.5. The graphs on the left describe a market-wide ESG uncertainty that is equal to σg,M = 0.01, while the right plots display results
for σg,M = 0.02. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
alpha) diminishes with increasing ESG uncertainty. Con- the effective beta increase with ESG uncertainty. Moreover,
sider, for instance, σg,M = 0.01. As the ESG uncertainty in- the alpha gap between brown and green assets diminishes
creases from 0 to 0.10, the difference in expected return with ESG uncertainty.
(μr,brown − μr,green ) decreases from 0.40% to 0.23% when
bESG = 1, and from 0.80% to 0.29% when bESG = 2. Simi-
6. Conclusion
lar patterns apply to alpha. Such calibration results follow
from Eq. (27).
We comprehensively analyze the equilibrium implica-
The overall evidence from the calibration indicates that
tions of ESG rating uncertainty for portfolio choice and as-
ESG uncertainty has meaningful implications for expected
set pricing. Starting with the market portfolio as the single
return, alpha, and effective beta. Notably, both alpha and
risky asset, we show that rating uncertainty leads to higher
663
perceived market risk, higher market premium, and lower Agarwal, V., Fos, V., Jiang, W., 2013. Inferring reporting-related biases in
investor demand. Next, we consider multiple risky assets hedge fund databases from hedge fund equity holdings. Manage. Sci.
59 (6), 1271–1289.
and heterogeneous economic agents and derive an ESG- Amel-Zadeh, A., Serafeim, G., 2018. Why and how investors use ESG infor-
augmented CAPM for the cross section of stock returns. In mation: evidence from a global survey. Financ. Anal. J. 74 (3), 87–103.
particular, we propose that ESG uncertainty could tilt the Berg, F., Kölbel, J.F., Rigobon, R., 2020. Aggregate confusion: the divergence
of esg ratings. SSRN Working Paper 3438533.
ESG-CAPM alpha relation and serve as a potential channel Bolton, P., Kacperczyk, M.T., 2020. Do investors care about carbon risk? J.
to explain the mixed evidence in prior studies. Financ. Econ.. Forthcoming
We empirically test the model implications and provide Carhart, M.M., 1997. On persistence in mutual fund performance. J. Fi-
nance 52 (1), 57–82.
supporting evidence. First, ESG rating uncertainty reduces
Chatterji, A.K., Durand, R., Levine, D.I., Touboul, S., 2016. Do ratings of
investor demand for stocks, especially for ESG-sensitive in- firms converge? Implications for managers, investors and strategy re-
vestors (i.e., norm-constrained institutions) in their ESG in- searchers. Strategic Manage. J. 37 (8), 1597–1614.
Chen, T., Dong, H., Lin, C., 2020. Institutional shareholders and corporate
vestment (i.e., green stocks). Second, brown stocks out-
social responsibility. J. Financ. Econ. 135 (2), 483–504.
perform green stocks only when rating uncertainty is low, Dimson, E., Karakaş, O., Li, X., 2015. Active ownership. Rev. Financ. Stud.
and the negative return predictability of ESG ratings does 28 (12), 3225–3268.
not hold for the remaining firms. We then calibrate the Dyck, A., Lins, K.V., Roth, L., Wagner, H.F., 2019. Do institutional investors
drive corporate social responsibility? International evidence. J. Financ.
model to assess its quantitative implications in the pres- Econ. 131 (3), 693–714.
ence of rating uncertainty. The analysis reinforces the no- Eccles, R.G., Stroehle, J.C., 2018. Exploring Social Origins in the Construc-
tion that ESG uncertainty could affect investors’ demand, tion of ESG Measures. SSRN Working Paper 3212685.
Edmans, A., 2011. Does the stock market fully value intangibles? Employee
the risk-return trade-off, and reduce economic welfare for satisfaction and equity prices. J. Financ. Econ. 101 (3), 621–640.
ESG-sensitive agents. Fama, E.F., French, K.R., 1993. Common risk factors in the returns on
Our findings echo the growing concerns regarding the stocks and bonds. J. Financ. Econ. 33 (1), 3–56.
Fama, E.F., French, K.R., 2018. Choosing factors. J. Financ. Econ. 128 (2),
lack of consistency of ESG information disclosure and rat- 234–252.
ings provided by different rating agencies. In the presence Fama, E.F., MacBeth, J.D., 1973. Risk, return, and equilibrium: empirical
of rating uncertainty, investors are less likely to make ESG tests. J. Polit. Economy 81 (3), 607–636.
Gibson Brandon, R., Krueger, P., Schmidt, P.S., 2021. ESG rating disagree-
investments and actively engage in corporate ESG issues.
ment and stock returns. Financ. Anal. J. Forthcoming.
This could increase the cost of capital for green firms and Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity
further limit their capacity to make socially responsible in- prices. Q. J. Econ. 118 (1), 107–156.
GSIA, 2018. Global sustainable investment review. Available at http://
vestments and generate real social impact. As the amount
www.gsi- alliance.org/trends- report- 2018/.
of sustainable investing is expected to keep growing, the Hartzmark, S.M., Sussman, A.B., 2019. Do investors value sustainability? a
overall impact will become even more striking. Viewed natural experiment examining ranking and fund flows. J. Finance 74
from this perspective, our results provide a conservative (6), 2789–2837.
Heinkel, R., Kraus, A., Zechner, J., 2001. The effect of green investment on
assessment of rating uncertainty. corporate behavior. J. Financ. Quant. Anal. 36 (4), 431–449.
Our evidence suggests that it would be useful for policy Hong, H., Kacperczyk, M., 2009. The price of sin: the effects of social
makers to establish a clear taxonomy of ESG performance norms on markets. J. Financ. Econ. 93 (1), 15–36.
Jiao, Y., Massa, M., Zhang, H., 2016. Short selling meets hedge fund 13F:
and unified disclosure standards for sustainability report- an anatomy of informed demand. J. Financ. Econ. 122 (3), 544–567.
ing. It would be especially instructive to identify which in- Krueger, P., Sautner, Z., Starks, L.T., 2020. The importance of climate risks
vestments are really green. Doing so could mitigate ESG for institutional investors. Rev. Financ. Stud. 33 (3), 1067–1111.
Lins, K.V., Servaes, H., Tamayo, A., 2017. Social capital, trust, and firm per-
uncertainty, thus reducing the cost of equity capital for formance: the value of corporate social responsibility during the fi-
green firms, leading to higher social impact. nancial crisis. J. Finance 72 (4), 1785–1824.
Our paper also suggests avenues for future research. López-de-Silanes, F., McCahery, J.A., Pudschedl, P.C., 2020. ESG perfor-
mance and disclosure: a cross-country analysis. Singapore J. Legal
While existing work studying equilibrium with ESG focuses
Stud. 217–241. (Mar 2020)
on a single-period environment, it would be natural to ex- Mackintosh, J., 2018. Is tesla or exxon more sustainable? It depends whom
tend ESG equilibrium to multiperiod dynamic setups. Then, you ask. Wall Street J.. September 17, 2018. Available at https://on.wsj.
com/2MQCC4m
the market ESG can display time variation, which would
Mooney, A., 2019. ESG investing sparks race in tech and hiring at asset
give rise to an incremental asset pricing factor. It would managers. Financ. Times. August 10, 2019. Available at https://www.
also be instructive to account for investors’ learning about ft.com/content/247f4034- 4280- 318a- 9900- 87608a575ede
the ESG profile of a firm. These and other topics in dy- Moss, A., Naughton, J.P., Wang, C., 2020. The irrelevance of ESG disclo-
sure to retail investors: evidence from Robinhood. SSRN Working Pa-
namic asset pricing are left for future research. per 3604847.
Newey, W.K., West, K.D., 1987. A simple positive-definite heteroskedas-
Supplementary material ticity and autocorrelation consistent covariance matrix. Econometrica
55, 703–708.
Pástor, L., Stambaugh, R.F., 2002. Investing in equity mutual funds. J. Fi-
Supplementary material associated with this article can nanc. Econ. 63 (3), 351–380.
be found, in the online version, at doi:10.1016/j.jfineco. Pástor, L., Stambaugh, R.F., Taylor, L.A., 2021. Sustainable investing in equi-
librium. J. Financ. Econ.. Forthcoming
2021.09.009. Pástor, L., Stambaugh, R.F., Taylor, L.A., 2021. Dissecting green returns.
SSRN Working Paper 3864502.
References Pedersen, L.H., Fitzgibbons, S., Pomorski, L., 2021. Responsible investing:
the ESG-efficient frontier. J. Financ. Econ.. Forthcoming
Abarbanell, J.S., Bushee, B.J., Smith Raedy, J., 2003. Institutional investor PRI, 2020. Principles for responsible investment. Available at https://www.
preferences and price pressure: the case of corporate spin-offs. J. Bus. unpri.org/download?ac=10948.
76 (2), 233–261.
664

Avramov Etal21

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Avramov Etal21

Uploaded by

Copyright:

Available Formats

Journal of Financial Economics 145 (2022) 642–664

Contents lists available at ScienceDirect

Journal of Financial Economics

Sustainable investing with ESG rating uncertaintyR

1. Introduction nance (ESG) factors in portfolio selection and management.

g˜M = μg,M + ˜g,M ,

is random; hence, the resulting beta is no longer identical g X M rg X M

proﬁle has a negative alpha because its effective beta in (25)

Panel A: Norm-constrained institutions

ESG rating ESG uncertainty

Low 2 3 4 High HML-U t-stat All

Low 0.170 0.183 0.187 0.178 0.179 0.009 (0.80) 0.177

HML-R 0.058∗ ∗ ∗ 0.053∗ ∗ ∗ 0.050∗ ∗ ∗ 0.047∗ ∗ ∗ 0.002 0.053∗ ∗ ∗

Panel B: Hedge funds

ESG rating ESG uncertainty

Low 2 3 4 High HML-U t-stat All

HML-R −0.031∗ ∗ ∗ −0.033∗ ∗ ∗ −0.032∗ ∗ ∗ −0.029∗ ∗ ∗ −0.011 −0.030∗ ∗ ∗

Panel C: Other institutions

ESG rating ESG uncertainty

Low 2 3 4 High HML-U t-stat All

HML-R 0.016 0.001 0.006 −0.005 0.011 0.007

Panel A: Return Panel B: CAPM-adjusted return

ESG rating ESG uncertainty ESG uncertainty

Low 2 3 4 High All Low 2 3 4 High All

ESG rating ESG uncertainty ESG uncertainty

Low 2 3 4 High HML-U Low 2 3 4 High HML-U

Stock returns regressed on lagged ESG rating and uncertainty

Excess return CAPM-adjusted return

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

ESG 0.002 0.098 0.062 0.199 0.042 0.139 0.162 0.301

Obs 283,671 254,873 283,671 254,873 272,728 245,451 272,728 245,451

Panel A: Norm-constrained institutions

ESG rating ESG uncertainty ESG uncertainty

Panel B: Hedge funds

ESG rating ESG uncertainty ESG uncertainty

Journal of Financial Economics 145 (2022) 642–664

Panel C: Other institutions

ESG rating ESG uncertainty ESG uncertainty

D. Avramov, S. Cheng, A. Lioui et al.

ESG rating ESG uncertainty ESG uncertainty

Panel B: CAPM-adjusted return

Journal of Financial Economics 145 (2022) 642–664

ESG rating ESG uncertainty ESG uncertainty

  equivalent return and their demand for the market portfo-

You might also like

equivalent return and their demand for the market portfo-