[Link]

1017/S0022109020000757 Published online by Cambridge University Press

JOURNAL OF FINANCIAL AND QUANTITATIVE ANALYSIS Vol. 56, No. 6, Sept. 2021, pp. 1985–2009
© THE AUTHOR(S), 2020. PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS ON BEHALF OF THE MICHAEL G. FOSTER
SCHOOL OF BUSINESS, UNIVERSITY OF WASHINGTON
doi:10.1017/S0022109020000757

Climate Change News Risk and Corporate

Bond Returns

Thanh D. Huynh
Monash University, Monash Business School, Department of Banking and Finance
[Link]@[Link]

Ying Xia
Monash University, Monash Business School, Department of Banking and Finance
[Link]@[Link] (corresponding author)

Abstract
We examine whether climate change news risk is priced in corporate bonds. We estimate
bond covariance with a climate change news index and find that bonds with a higher climate
change news beta earn lower future returns, consistent with the asset pricing implications of
demand for bonds with high potential to hedge against climate risk. Moreover, when
investors are concerned about climate risk, they are willing to pay higher prices for bonds
issued by firms with better environmental performance. Our findings suggest that corporate
policies aimed at improving environmental performance pay off when the market is con-
cerned about climate change risk.

I. Introduction
The scientific literature shows that climate change has become a prominent
risk that will potentially create substantial costs to the economy (Burke, Hsiang,
and Miguel (2015), Dietz, Bowen, Dixon, and Gradwell (2016), and Lesk, Rowhani,
and Ramankutty (2016)). The influential review by Stern (2007), for example,
estimates that the overall cost of climate change will be equivalent to at least
5% of the global gross domestic product each year. Given the well-documented
evidence of the impact of climate change on consumption and investment
opportunities,1 investors may increasingly find it desirable to hedge themselves
against climate change risk today (Engle, Giglio, Kelly, Lee, and Stroebel (2020)
(EGKLS hereafter)). In this study, we show that investors’ demand for corporate
bonds with high potential to hedge against climate change risk can have an impact
on the cross section of corporate bond returns.

We are grateful to Michael Barnett (the referee) and Hendrik Bessembinder (the editor) for their
extremely helpful comments and suggestions. We also thank Stephen Brown, Roger Edelen, Bart Frijns,
Neal Galpin, Michael Gofman, Zhiguo He, Juhani Linnainmaa, Christian Lundblad, Lyndon Moore, and
Phong Ngo for their constructive comments. We thank Stefano Giglio and Johannes Stroebel for making
the climate change news index data available. All errors are our own.
1
Burke et al. (2015), Dell, Jones, and Olken (2012), and Hsiang (2010) provide macro-level evidence
that a country’s economic activities significantly decline as the climate becomes warmer. Deryugina and
Hsiang (2014) find that the U.S. economy has not fully implemented adaptation measures to offset the
impact of warming temperatures.

1985
 [Link] Published online by Cambridge University Press
1986 Journal of Financial and Quantitative Analysis

We measure a bond’s potential to hedge against climate change risk, denoted

by βCCN, by estimating the bond’s covariance with the climate change news index,
which is a news-based measure of climate change risk developed by EGKLS.
Specifically, βCCN is estimated from 60-month rolling regressions of bond excess
returns on innovations in the monthly climate change news index. By construction,
a higher βCCN indicates higher bond returns as the climate change news index
increases. Examining the out-of-sample performance of the monthly βCCN in
predicting cross-sectional variations in future bond returns, we find that bonds
with a higher βCCN are significantly associated with lower future returns. A
one-standard-deviation increase in βCCN is associated with a decrease in future
excess returns of 6.29 basis points (bps) per month, representing a 12.60% reduc-
tion relative to the sample mean. We can also interpret this estimate in terms of the
dollar cost of debt financing by assuming the issuance of a new bond with the same
characteristics as the average bond in our sample, but with a higher βCCN. Given that
an average bond trades at $102.50 for an offering amount of 558,723 bonds, an
increase in bond price of 6.29 bps represents an estimated saving of $3.60 million in
the cost of debt financing for a representative firm in our sample.
Our findings are consistent with the intertemporal hedging hypothesis, which
posits that investors demand bonds with strong hedging potential (Merton (1973),
Campbell (1993), (1996), and Bali, Brown, and Tang (2017)). Specifically, since
climate change affects future investment and consumption opportunities, investors
prefer to hold bonds that perform better as the climate change news index increases.
This is because the increase in the value of bonds with a high βCCN can offset the loss
in consumption and investment opportunities. Investors thus demand more of
these bonds and are willing to accept lower future returns (Addoum, Delikouras,
Korniotis, and Kumar (2019)).
We conduct further analyses to test the intertemporal hedging hypothesis.
First, we find that, during times of heightened climate change news risk, a
one-standard-deviation increase in βCCN is associated with a total reduction of
22.0 bps in future monthly bond returns. This finding is consistent with the pre-
diction that demand for bonds with strong hedging ability increases during periods
of high climate change risk. Second, we find that the effect of climate change news
risk on long-term corporate bonds is about 2 times the effect on short-term bonds,
suggesting that investors are concerned about the potential impact of climate
change risk being more severe in the distant future, raising the hedging demand
for long-term bonds.
To explore plausible environmentally related determinants of a bond’s βCCN,
we examine the environmental performance of bond issuers, using two alternate
environmental scores (ESCORE) obtained from two environmental, social, and
governance (ESG) rating agencies; Sustainalytics and MSCI. Consistent with a
bond’s covariance with the time-variance of climate change news risk, we find that
βCCN is positively associated with the issuer’s ESCORE when climate change news
risk is high. In a similar vein, we find that, during times of intensified climate
change news risk, βCCN is significantly higher among bonds whose issuers are
preferred by environmentally responsible institutions, or have low exposure to the
political risk associated with the environment. Taken together, these findings are
consistent with the intertemporal hedging hypothesis, whereby investors prefer
 [Link] Published online by Cambridge University Press
Huynh and Xia 1987

bonds with a high βCCN because they are issued by firms with better environmental
performance, and thus offer strong potential to hedge against climate change
news risk.
We next investigate the robustness of our findings. First, we confirm that our
results are robust in subsamples of investment-grade and noninvestment–grade
bonds, suggesting that our findings are not specific to a category of credit ratings
such as noninvestment grade bonds, which are less liquid (Chen, Lesmond, and Wei
(2007)). Second, we show that bonds with a high βCCN exhibit lower future yield
spreads, supporting our conclusion that investors are willing to accept a lower
yield on these bonds with strong hedging potential. Third, we find an insignificant
relation between firm-level βCCN and either the issuer’s expected default risk
or future cash flows. This result indicates that the effect of βCCN is driven by
investors’ perceptions about a bond’s exposure to climate change risk, but
the firm’s fundamentals are not necessarily affected. Fourth, we confirm that our
results remain robust when we use an alternative climate change news index, the
Crimson Hexagon (CH) negative climate change news index. Finally, we perform
“placebo” tests using a sample of U.S. government bonds and find an insignificant
relation between climate change news risk and future Treasury bond returns,
suggesting that the effect of βCCN on corporate bond returns is not a mechanical
result.
Our study contributes to the contemporary climate finance literature that
examines the effect of climate change risk on financial markets and firms. Seltzer,
Starks, and Zhu (2020) find that bond yields and credit ratings are jointly deter-
mined by the firm’s environmental profile and its regulatory risk exposure. Ilhan,
Sautner, and Vilkov (2021) find that uncertainty about climate policy is priced in the
option market. Uncertainty about climate change impacts and climate model uncer-
tainty can also affect the social cost of carbon (Barnett, Brock, and Hansen (2020)).
Other studies show the effect of temperatures on the equity risk premium (Bansal,
Ochoa, and Kiku (2016)), stock prices of carbon-intensive firms (Choi, Gao, and
Jiang (2020)), and establishment productivity (Addoum, Ng, and Ortiz-Bobea
(2020)). In addition, climate risk also affects the prices of municipal bonds
(Goldsmith-Pinkham, Gustafson, Lewis, and Schwert (2020) and Painter (2020))
and house prices (Giglio, Maggiori, Rao, Stroebel, and Weber (2018), Bernstein,
Gustafson, and Lewis (2019), and Baldauf, Garlappi, and Yannelis (2020)).
Our study differs from existing work in both its theoretical and empirical
contributions. Theoretically, we are motivated by the well-established asset pricing
theory on intertemporal hedging demand (Merton (1973) and EGKLS) to examine
the effect of climate change news risk on bond returns. Empirically, our measure of
a bond’s exposure to climate change risk, βCCN, is distinct from other environmental
risk measures in that it can be estimated for a large cross section of bonds.
By investigating the effect of a corporate bond’s exposure to climate change
news risk on future returns, our study offers a direct response to the call of EGKLS
for research on this important yet unexplored question. Moreover, EGKLS invite
future research to examine whether the news-based climate change index is subject
to data mining. Our study helps address these data mining concerns.
We also contribute to the literature examining the cross-sectional determinants
of corporate bond returns by showing that climate change news risk is a novel factor
 [Link] Published online by Cambridge University Press
1988 Journal of Financial and Quantitative Analysis

affecting corporate bonds’ future returns.2 Last, our study has implications for both
business managers and policymakers. Specifically, the fact that investors prefer the
bonds of environmentally friendly issuers speaks to the financial benefits of invest-
ing in environmental policy at the corporate level. Our findings suggest that these
firms can enjoy considerably lower costs of debt, especially when investors’
concerns about climate change risk are elevated.

II. Hypothesis Development

This study examines the asset pricing implications of demands for hedging
against climate change risk on corporate bonds. The empirical test is motivated by
the model of Merton (1973), who shows that, in a multi-period economy, investors
prefer to hedge against future shocks to consumption and investment opportunity
sets. The model predicts that state variables that are correlated with changes
in investment opportunities are priced in capital markets. In particular, an asset’s
covariance with such a state variable is related to its expected returns (Bali (2008)).
Provided that the climate change news index is a state variable affecting
investors’ investment opportunity set (EGKLS), it can affect future returns on
corporate bonds via intertemporal hedging demands (Merton (1973), Bali et al.
(2017), and Bali et al. (2019)). Specifically, while an increase in climate change risk
reduces investors’ consumption and future investment opportunities, they can
compensate for this loss through an increase in the returns on bonds that have
greater covariance with the climate change news index. Therefore, on account of
intertemporal hedging demand, investors prefer to hold bonds that have higher
covariance with the climate change news index, and they are willing to pay higher
prices and accept lower future returns for these bonds.3
To test this prediction, we estimate monthly rolling window regressions of
individual bond returns on the climate change news index controlling for common
determinants of corporate bond returns. The monthly coefficient on the climate
change index from these rolling regressions, denoted βCCN, represents a corporate
bond’s covariance with the climate change news index. Since bonds with a high
βCCN provide higher returns as the climate change news risk increases, they serve as
good assets to hedge against climate change risk. Therefore, investors are willing to
pay higher prices and accept lower future returns on bonds with a higher βCCN. We
thus propose the following intertemporal hedging hypothesis:

Hypothesis 1. There is a negative relation between a bond’s βCCN and future

returns.

Asset pricing theory suggests that, if the negative returns on high-βCCN bonds
are due to hedging demands, the effect should be time-varying (Bloom (2009),
2
Studies on the cross-sectional determinants of corporate bond returns include Gebhardt, Hvidkjaer,
and Swaminathan (2005), Bessembinder, Kahle, Maxwell, and Xu (2008), Lin, Wang, and Wu (2011),
Feldhütter (2012), Jostova, Nikolova, Philipov, and Stahel (2013), Bai, Bali, and Wen (2019), Bali,
Subrahmanyam, and Wen (2019), Massa and Zhang (2021), among others.
3
This notion is consistent with the evidence found in green bonds. Baker, Bergstresser, Serafeim, and
Wurgler (2018) find that investors prefer green bonds, whose proceeds are used for environmentally
sensitive purposes. They bid up green bond prices and are willing to accept lower future returns.
 [Link] Published online by Cambridge University Press
Huynh and Xia 1989

Bekaert, Hoerova, and Lo Duca (2013), and Bali et al. (2017)). Specifically, since
fears about potential losses in consumption and investment opportunities increase
during times of heightened climate change news risk, investors’ intertemporal
hedging demand is expected to become more pronounced. Thus, these investors
are willing to accept even lower expected returns from bonds with a higher βCCN for
hedging purposes in times of high climate change news index. We thus propose the
second hypothesis, as follows:

Hypothesis 2. The negative relation between a bond’s βCCN and future returns is
stronger in times of heightened climate change news risk.

We next examine whether the pricing of climate change risk depends on a

bond’s term structure. On the one hand, the prices of long-term bonds could be more
sensitive to climate change risk than short-term bonds, because the consequences of
climate change risk will be more intense and will affect a firm’s resiliency in the long
run. On the other hand, realizations of climate change risk such as natural disasters,
regulatory changes, and technological advances in a transitioning economy can
affect the current investment opportunity set (BlackRock (2016)). Hedging against
climate risk is valuable in the short run because businesses can adapt to climate
change in the long run (Giglio et al. (2018)). Therefore, short-term bonds could also
be affected. Whether climate change news risk affects long-term or short-term bonds
is ultimately an empirical question that has not been examined in the corporate bond
market. In practice, institutional investors believe that climate risks have already
materialized and it is difficult to fully hedge against the risk in the long run as climate
change is worsening with time (Ilhan, Krueger, Sautner, and Starks (2019) and
Krüger, Sautner, and Starks (2020)). Thus, we state the next hypothesis, as follows:

Hypothesis 3. The negative relation between βCCN and future returns is stronger for
long-term bonds.

The previous hypotheses suggest that bonds with a high βCCN have low
exposure to climate change news risk. An interesting follow-up question is why
investors prefer these bonds when climate change news risk is high. In our context,
a directly relevant characteristic is a firm’s environmental performance. EGKLS
show that a firm’s environmental performance is useful in constructing strategies
to hedge against climate change news risk. Seltzer et al. (2020) find that, in states
with high environmental enforcement rates, firms with a lower ESCORE have
lower credit ratings and higher yields. Other studies documenting the effect of
ESG factors on firm values include those of Heinkel, Kraus, and Zechner (2001),
Krüger (2015), Ferrell, Liang, and Renneboog (2016), Lins, Servaes, and Tamayo
(2017), and Albuquerque, Koskinen, and Zhang (2019). These studies together
indicate that firms with good environmental performance have low exposure to
climate risk, and thus, their bonds provide investors with strong hedging potential,
especially when climate change news risk is high. We therefore propose the
following testable hypothesis:

Hypothesis 4. In times of high climate change news risk, a bond’s βCCN is higher
when its issuer has better environmental performance.
 [Link] Published online by Cambridge University Press
1990 Journal of Financial and Quantitative Analysis

III. Data and Variable Construction

A. Corporate Bond Data

We obtain corporate bond transaction records from the Financial Industry

Regulatory Authority’s Trade Reporting and Compliance Engine (TRACE)
Enhanced database for the period from July 2002 to Dec. 2016. We merge this
database with the Mergent Fixed Income Securities Database (FISD) for informa-
tion on the characteristics of corporate bond issues and issuers. The procedure to
clean the TRACE Enhanced data closely follows that of Dick-Nielsen (2009),
(2014).4 Following Bao, Pan, and Wang (2011), we require the bonds in our sample
to trade on at least 75% of their trading days over the sample period. This restriction
ensures that the liquidity measure of Bao et al. can be reliably estimated using
transaction data. Following Bai et al. (2019), we further restrict our sample of
corporate bonds to those listed and traded in the U.S. public market, eliminating
bonds that: i) are convertible; ii) have a trading price below $5 or above $1,000;
iii) have a floating coupon rate; or iv) have less than 1 year to maturity. Following
Lin et al. (2011) and others, we calculate the monthly corporate bond returns as of
month t as
ðPi,t þ AIi,t Þ þ C i,t  ðPi,t1 þ AIi,t1 Þ
(1) ri,t ¼ ,
ðPi,t1 þ AIi,t1 Þ

where Pi,t is the last transaction price at which bond i was traded in month t, which is
computed as the trading volume-weighted average of all intraday transaction prices
(Bessembinder et al. (2008)),5 AIi,t is accrued interest, and Ci,t is the coupon
payment, if any, in month t. We then compute the monthly bond excess return as
EXCESS_RETURNi,t = ri,trf,t, where rf,t is the monthly risk-free rate proxied by
the 1-month Treasury bill rate obtained from the Federal Reserve Board of Gover-
nors (FRB).
We also calculate aggregate bond market risk factors, that is, the excess
corporate bond market return, the term spread, the default spread, the TED spread,
and bond market illiquidity. Specifically, monthly excess corporate bond market
returns (EXCESS_BMKT_RETURN) are computed as monthly returns on the
Barclays Capital Corporate Bond Index over the monthly risk-free rate, where
the bond index data are obtained from Bloomberg. The term spread
(TERM_SPREAD) is the difference between the monthly return on the Ibbotson
U.S. long-term government bond index and the 1-month Treasury bill return. The
default spread (DEFAULT_SPREAD) is measured as the difference between BAA-
and AAA-rated corporate bond monthly yields. To proxy for general funding
constraints in the market, we use the TED spread (TED_SPREAD) obtained from
the Federal Reserve Bank of St. Louis, which is the difference between the 3-month

4
We start the sample in 2002 because this is when the TRACE Enhanced database begins. To clean
the data, we follow the sample code available from the Wharton Research Data Service (WRDS) Clean
TRACE Enhanced File.
5
We obtain similar results when using a bond’s last valid daily price in the last 5 days of the month to
calculate its monthly return.
 [Link] Published online by Cambridge University Press
Huynh and Xia 1991

London Interbank Offered Rate (LIBOR), based on U.S. dollars, and the 3-month
Treasury bill rate. Bond market illiquidity (MARKET_ILLIQUIDITY) is the mar-
ket illiquidity for all U.S.-listed corporate bonds, constructed by Dick-Nielsen,
Feldhütter, and Lando (2012).6

B. Climate Change News Beta

The climate change news index constructed by EGKLS is a marketwide index

reflecting climate change risk.7 Based on the idea that climate change attracts wide
attention from the media in times of elevated concerns about climate change risk,
the climate change news index captures the intensity of discussions about climate
change in The Wall Street Journal (WSJ). Specifically, the index is measured as the
correlation between texts in the WSJ and the climate change vocabulary (CCV),
where the CCV is constructed using a comprehensive search of various authorita-
tive reports published by different governmental and research organizations.
EGKLS conduct a variety of validation tests and show that this index reasonably
captures the aggregate negative view among investors about climate change risk at a
given point in time. Following their seminal study, our analysis uses innovations in
the climate change news index, which are the residuals from the first-order auto-
regressive model.
For each bond i in each month t, we estimate the climate change news beta
(βCCN) from the monthly rolling regression of bond excess returns on innovations in
the monthly climate change news index over a 60-month window with a minimum
of 30 valid monthly return observations.8 We control for excess market returns,
bond market illiquidity, the term spread, the default spread, and the TED spread,
as follows:

(2) EXCESS_RETURNi,t ¼ αi,t þ βCCN CCNt

þ βMARKET EXCESS_MKT_RETURNt
þ βILLIQ MARKET_ILLIQUIDITYt
þ βTERM TERM_SPREADt
þ βDEFAULT DEFAULT_SPREADt
þ βTED TED_SPREADt þ εi,t ,

6
The data on the market illiquidity index are available on Peter Feldhütter’s website at http://
[Link]. The data to compute the term spread and the default spread, used by Welch and Goyal
(2007), are available on Amit Goyal’s website at [Link]
7
The climate change news index data are available on both Stefano Giglio’s website at [Link]
[Link]/view/stefanogiglio and Johannes Stroebel’s website at [Link]
8
We confirm that our findings remain robust when using 36-month (or 24 months) rolling windows,
with at least 18 months (12 months) of valid returns over each window. These results are reported in
Supplementary Material Table A8. These robustness tests are relevant in the context of our study, given
the short time dimension in the data and the potential likelihood that concerns about climate change news
may be changing rapidly, as political actions like the Paris Climate Accord, climate activism, and
enhanced climate damages (such as the increased intensity of sea-level rise, wildfires, and hurricanes)
appear to be accelerating in recent times. We thank the referee for suggesting these implications.
 [Link] Published online by Cambridge University Press
1992 Journal of Financial and Quantitative Analysis

where CCN represents innovations in the monthly climate change news index
and EXCESS_MKT_RETURN is the excess equity market return obtained from
Kenneth French’s data library ([Link]
french/[Link]). The βCCN estimated from equation (2) captures a bond’s covari-
ance with innovations in the climate change news index. By construction, a greater
βCCN indicates an increase in bond value as innovation in the climate change news
index increases.9

C. Other Control Variables

We employ a comprehensive list of variables to control for the effects of
known determinants of corporate bond returns. Specifically, we measure the down-
side risk (DOWNSIDE_RISK) of a bond as the average of the four lowest monthly
return observations over the past 36 months beyond the 10% value-at-risk (VaR)
threshold. We multiply this measure by 1 to obtain an intuitive interpretation
where higher downside risk is associated with higher expected returns.10 We control
for corporate bond illiquidity (ILLIQUIDITY) using the measure of Bao et al.
(2011), which is also employed by Bai et al. (2019). Using transaction-based data
from TRACE, ILLIQUIDITY
is computed at the end of each month t for each bond
i as covt Δpi,t,τ ,Δpi,t,τþ1 , where Δpi,t,τ is the change in the natural logarithm of
the price on day τ of month t.11 In addition, we control for the bond’s credit risk,
using its Standard & Poor’s (S&P) historical credit ratings, obtained from Mergent
FISD. We require individual bonds to have valid ratings information, and we
convert the letter ratings to numerical scores, where one refers to an AAA rating
and 22 refers to a D rating. Bonds with ratings from AAA to BBB are defined
as investment-grade bonds, and noninvestment-grade bonds have ratings below
BBB.12
Our regressions also include the bond’s market risk using its market
beta, βBOND_MARKET, which is estimated from the monthly rolling regressions
of bond excess returns on the excess return of the Barclays Capital Corporate
Bond Index (EXCESS_BMKT_RETURN), using data over a 60-month window.
To control for a bond’s exposure to economic conditions, we include βTERM,
βDEFAULT, and βTED, estimated from equation (2). Other bond characteristics, that
is, the natural logarithm of the time to maturity and the natural logarithm of the
amount outstanding, are also included in our regression models. We also add return

9
We obtain qualitatively similar results when using alternative combinations of risk factors in
equation (2) and additional factors, such as Fama and French’s (1993) size factor (SMB) and book-
to-market factor (HML).
10
Bai et al. (2019) suggest that this expected shortfall measure is a more robust measure of downside
risk than the 5% VaR. By construction, the expected shortfall has a smaller magnitude than the 5% VaR.
Untabulated results confirm the robustness of our results when using the 5% VaR as an alternative
measure.
11
We confirm the robustness of our results when using Roll’s (1984) measure of illiquidity.
12
We do not consider junk bonds with “substantial risks” according to S&P ratings (i.e., a rating of
CCC+ or below) to avoid the contamination of these relatively illiquid and risky bonds on our results.
Nevertheless, we show in Table A9 of the Supplementary Material that our results still hold if we include
bonds with all ratings, that is, S&P ratings from AAA to D.
 [Link] Published online by Cambridge University Press
Huynh and Xia 1993

reversal (REVERSAL), calculated as the bond’s excess return in the previous

month.
We further control for differences in firm characteristics and risk.13 We include
idiosyncratic risk (IDIO_RISK), which is computed as the standard deviation of the
residuals from the monthly regression of daily stock returns on Fama and French’s
(1993) three risk factors (Ang, Hodrick, Xing, and Zhang (2006) and Campbell and
Taksler (2003)). Following Greenwood and Hanson (2013), we control for the
leverage ratio (LEVERAGE), which is the sum of long-term debt, short-term debt,
minority interest, and preferred stock, scaled by total assets. We also include other
standard firm-level determinants of risk, such as firm size (ln(MARKET_CAP)),
defined as the natural logarithm of the market value of the issuer’s common equity
and return on equity (ROE), which accounts for the cross-sectional differences in
issuers’ cash flows and is computed as income before extraordinary items divided
by the book value of common equity.

D. Sample and Summary Statistics

Our sample is the intersection of corporate bond data, stock data from the Center
for Research in Security Prices (CRSP), and accounting data from the Compustat
Annual Fundamentals file. To mitigate the impact of outliers, we winsorize all
continuous variables at their 1st and 99th percentiles. After applying the aforemen-
tioned restrictions on the bond data and requiring bonds to have sufficient data to
estimate betas, our main sample contains 239,164 bond-month observations from
Jan. 2005 to Dec. 2016 covering 8,231 unique corporate bonds.14 Table 1 reports
summary statistics for the key variables used in the baseline analysis.

TABLE 1
Summary Statistics

Table 1 reports summary statistics for bond-month observations over the sample period from Jan. 2005 to Dec. 2016. The
descriptive statistics include the sample mean, 25th percentile, median, 75th percentile, and standard deviation of the key
variables used in this study. Variables are defined in the Appendix.
Variable N Mean 25th Median 75th Std. Dev.

EXCESS_RETURN 481,549 0.499 0.523 0.337 1.489 2.167

βCCN 247,350 0.405 1.417 0.355 0.624 2.095
DOWNSIDE_RISK 389,084 3.445 1.508 2.584 4.374 3.014
MATURITY 492,481 9.443 3.594 6.386 10.053 8.943
RATING 488,127 8.186 6.000 8.000 10.000 3.231
ln(AMOUNT_OUT) 492,481 19.906 19.337 20.026 20.436 0.885
ILLIQUIDITY 492,481 0.959 0.098 0.289 0.783 3.032
βBOND_MARKET 287,725 1.025 0.567 0.908 1.341 0.678
βTERM 247,350 2.328 7.012 1.168 3.415 11.826
βDEFAULT 247,350 14.534 8.371 12.131 35.725 45.478
βTED 247,350 0.063 2.038 0.014 2.271 5.303
IDIO_RISK 478,596 1.249 0.690 0.965 1.438 1.075
LEVERAGE 478,558 0.354 0.228 0.332 0.460 0.195
ln(MARKET_CAP) 492,467 16.727 15.784 16.807 17.829 1.586
ROE 478,547 0.188 0.067 0.124 0.194 3.614

13
While we aim to be conservative by using a comprehensive set of control variables, we confirm that
the results do not qualitatively change if we do not control for firm-level characteristics.
14
The sample period starts in Jan. 2005, instead of at the beginning of the TRACE Enhanced database
(July 2002), because we use a 60-month rolling window to estimate the climate change beta and require
at least 30 available observations.
 [Link] Published online by Cambridge University Press
1994 Journal of Financial and Quantitative Analysis

Over the sample period, the average monthly excess return on corporate bonds
is 0.50%, with a standard deviation of 2.17. The representative corporate bond has a
climate change news beta of 0:41 and a standard deviation of 2.10. Consistent
with Bao et al. (2011), the average bond in our sample is investment grade, with an
average rating score of 8.19 (BBB+). The average bond has an illiquidity measure
of 0.96, downside risk of 3.45%, 9.44 years to maturity, and $441.64 million
outstanding. Other estimates of beta, that is, the bond market, the default spread,
the term spread, and TED spread betas, have mean values of 1.03, 14.53, 2:33,
and 0:06, respectively. The average firm in our sample has an idiosyncratic risk
value of 1.25, a leverage ratio of 0.35, a market value of equity of $18.38 million,
and an annual return on equity of 0.19.

IV. Empirical Results

A. Climate Change News Beta and Expected Corporate Bond Returns

In this section, we examine the relation between a bond’s climate change news
beta and future returns at the bond-month level. Following prior literature on
corporate bond returns (Bessembinder et al. (2008), Lin et al. (2011), and Bai
et al. (2019)), we include various bond- and firm-level variables in the baseline
specification, as follows:

(3) EXCESS_RETURNi,tþ1 ¼ α þ γβCCN

i,t þ δ0 X i,t þ λ0 Y j,t þ θi þ φ þ τ þ μ þ εi,t ,

where, for bond i, month t, and firm j, Xi,t represents a vector of bond-level control
variables (i.e., downside risk, maturity, ratings, amount outstanding, illiquidity,
short-term reversal, bond market beta, term spread beta, default spread beta, and
TED spread beta); Yj,t is a set of firm-level control variables (i.e., idiosyncratic risk,
leverage ratio, market capitalization, and return on equity); and, finally, θi, φ, τ, and
μ are vectors for the bond/firm, year, month, and the firm’s headquarter state fixed
effects, respectively. We obtain the historical headquarters locations from the data
library of Bill McDonald, who, in turn, obtained the information from the firms’
historical 10-K filings on the U.S. Securities and Exchange Commission’s
EDGAR.15 For ease of interpretation of the coefficient estimates, we multiply the
returns by 100.
Following contemporary asset pricing research (e.g., Patton and Verardo
(2012) and Ben-Rephael, Carlin, Da, and Israelsen (2021)), we estimate equation
(3) using panel regression analysis and controlling for bond/firm, year-month, and
headquarters’ state fixed effects. The inclusion of these fixed effects accounts for
unobserved heterogeneity across firms/bonds, macroeconomic trends, inter-year
seasonality effects, and time-invariant state factors. As Patton and Verardo point
out, the use of fixed effects allows for cross-sectional differences in betas and,
importantly, captures low-frequency changes in betas over time. Moreover,
Petersen (2009) shows that panel regressions with fixed effects allow researchers
to improve the efficiency of estimates and straightforwardly compute standard
15
We thank Bill McDonald for making the data available on his website. Loughran and McDonald
(2016) provide an excellent survey on textual analysis in accounting and finance.
 [Link] Published online by Cambridge University Press
Huynh and Xia 1995

errors clustered at the firm level, making the estimated coefficients and standard
errors more robust than those obtained from the traditional Fama–MacBeth
framework.16
Table 2 reports the results from the regression of 1-month-ahead corporate
bond excess returns on βCCN. Columns 1 and 2 report the results of regressions
without control variables. Columns 3 and 4 present the results of baseline regres-
sions with both bond- and firm-level control variables. In columns 1 and 3, we
add firm, state, year, and month fixed effects. Bond, state, year, and month fixed
effects are included in the specifications of columns 2 and 4. Across all model
specifications, the point estimates on βCCN are negative and statistically significant
at the 1% level, indicating a negative relation between the climate change news beta
and future bond returns.
The coefficients are also economically significant. For example, in column 4
of Table 2, the coefficient estimate on βCCN of 0:03 indicates that a one-standard-
deviation increase in the climate change news beta is associated with a drop of
6.29 bps (= 0:030  2:095) in the next month’s bond excess return, which is
equivalent to a decrease of 12.60% relative to the sample mean of excess returns.
To elaborate on the economic significance of our results, we compare the effect
of βCCN to that of DOWNSIDE_RISK, which Bai et al. (2019) show to be a
strong predictor of future corporate bond returns. The estimated coefficient on
DOWNSIDE_RISK is 0.09, meaning that a one-standard-deviation increase in
DOWNSIDE_RISK is associated with an increase of 25.62 bps (= 0:085  3:014) in
the next month’s bond excess return, which is equivalent to an increase of 51.34%
relative to the sample mean. The effect of βCCN on future bond returns therefore
appears to be smaller than the effect of DOWNSIDE_RISK. The relatively smaller
effect of βCCN is also consistent with Goldsmith-Pinkham et al.’s (2020) conclusion
that the market has not fully incorporated climate risk into asset prices.
We can also interpret this estimate on βCCN in terms of the dollar cost of debt
financing, if we assume that an average firm issues a new corporate bond with the
same characteristics as the average bond in our sample. Given that the average bond
price is $102.5, with an offer of 558,723 bonds, a decrease in excess bond returns by
6.29 bps means that the new bond is expected to be issued at a higher price, which
translates into savings of $3.60 million in the cost of debt financing for the
average firm.
Our results are consistent with Hypothesis 1, as well as the prediction of
EGKLS on the asset pricing implications of hedging against climate change news
risk. Since bonds with a higher βCCN perform better as the climate change news
index increases, investors demand more of these bonds to hedge against climate
change news risk. Therefore, investors are willing to pay higher prices for bonds
with a higher βCCN and to accept lower future returns on these bonds.
The coefficients on the other control variables are also consistent with prior
literature. For example, columns 3 and 4 of Table 2 show that bonds with greater
downside risk, longer time to maturity, a larger size, lower past month returns, lower

16
We confirm that our conclusions do not change if we cluster standard errors by bond and year.
 [Link] Published online by Cambridge University Press
1996 Journal of Financial and Quantitative Analysis

TABLE 2
Climate Change News Beta and Expected Returns on Corporate Bonds

Table 2 reports the results from the panel regressions of 1-month-ahead bond excess returns (EXCESS_RETURN) on βCCN
over the sample period from Jan. 2005 to Dec. 2016. Columns 1 and 2 report the regression results with βCCN in month t as
independent variable and without other control variables. Columns 3 and 4 report the regression results with both βCCN and
control variables. t-statistics computed using clustered standard error at the issuer level are presented in parentheses. *, **,
and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively. Variables are defined in the
Appendix.
Dependent Variable: Future EXCESS_RETURN

Variable 1 2 3 4

βCCN 0.014*** 0.031*** 0.014*** 0.030***

(-4.003) (-5.069) (-3.664) (-4.347)
DOWNSIDE_RISK 0.064*** 0.085***
(10.232) (9.634)
ln(MATURITY) 0.111*** 0.383***
(10.016) (11.667)
ln(1+RATING) 0.129** 0.328***
(2.004) (2.756)
ln(AMOUNT_OUT) 0.018** 0.410***
(2.531) (3.290)
REVERSAL 0.063*** 0.075***
(11.156) (13.746)
ILLIQUIDITY 0.031*** 0.033***
(5.349) (5.428)
β BOND_MARKET
0.051** 0.040
(2.097) (0.909)
β TERM
0.006*** 0.009***
(6.801) (7.450)
βDEFAULT 0.001*** 0.001***
(4.210) (4.117)
βTED 0.003 0.006*
(1.277) (1.882)
IDIO_RISK 0.150*** 0.168***
(7.846) (8.893)
LEVERAGE 0.064 0.194
(0.290) (0.643)
LN(MARKET_CAP) 0.047 0.098***
(1.349) (2.677)
ROE 0.002 0.002
(1.423) (1.530)
Firm fixed effects Yes No Yes No
Bond fixed effects No Yes No Yes
State fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Month fixed effects Yes Yes Yes Yes
No. of obs. 240,749 240,749 239,164 239,164
Adj. R2 0.064 0.062 0.085 0.084

liquidity, a higher term spread beta, a lower default spread beta, a smaller TED
spread beta, and higher equity idiosyncratic risk have higher future returns.17
We also examine the cross-sectional relation between βCCN and future bond
returns, using Fama and MacBeth (1973) regression approach. The estimation
results reported in Table A1 of the Supplementary Material are consistent with
our baseline conclusions.

17
In Table A10 of the Supplementary Material, we show that the effect of the bond-level climate
change news beta is not confounded by the stock-level climate change news beta.
 [Link] Published online by Cambridge University Press
Huynh and Xia 1997

B. Nonlinearity of the Climate Change News Premium

We next examine Hypothesis 2, which posits that the effect of the climate
change news beta on future bond returns changes over time and is more pronounced
during times of heightened climate change news risk. We define a month as having a
high climate change news index if the innovation in the monthly climate change
news index is greater than the median value of the index; otherwise, the month has
a low climate change news index. Accordingly, we construct the dummy variable
HIGH_CCN, which is equal to 1 for months with a high climate change news index,
and 0 otherwise. We interact HIGH_CCN with βCCN and reexamine our baseline
regressions by including both the interaction term HIGH_CCN  βCCN and
HIGH_CCN. Table 3 reports the estimation results.
Consistent with the prediction of Hypothesis 2, Table 3 shows that the coef-
ficient on the interaction term between HIGH_CCN and βCCN is negative and
significant at the 1% level, indicating that, during times when the climate change
news index is high, the impact of the climate change news beta on future bond
returns is significantly greater compared to low climate change news index periods.
The total premium of βCCN is also economically large. For example, the coefficient
estimates in column 2 suggest that, during times of a high climate change news
index (i.e., HIGH_CCN = 1), a one-standard-deviation increase in βCCN is associ-
ated with a total reduction of 22.0 bps in future bond excess returns.18

TABLE 3
Marketwide Climate Change News Risk and the Climate Change News Premium

Table 3 reports the results from the regression tests of the nonlinear effect of climate change news beta on future excess return
on corporate bonds, conditional on high and low periods of the climate change news index. HIGH_CCN is a dummy variable
equal to 1 if the monthly climate change news index is greater than the median value of the monthly climate change news
index, and 0 otherwise. HIGH_CCN  βCCN is an interaction term between HIGH_CCN and βCCN. Standard errors are clustered
at the firm level in all regressions. t-statistics are presented in parentheses. *, **, and *** indicate statistical significance at the
10%, 5%, and 1% level, respectively. Variables are defined in the Appendix.
Dependent Variable: Future EXCESS_RETURN

Variable 1 2

HIGH_CCN  βCCN 0.075*** 0.076***

(5.362) (5.303)
βCCN 0.014*** 0.029***
(5.604) (6.983)
HIGH_CCN 0.014 0.007
(0.550) (0.276)
Control variables Yes Yes
Firm fixed effects Yes No
Bond fixed effects No Yes
State fixed effects Yes Yes
Year fixed effects Yes Yes
Month fixed effects Yes Yes
No. of obs. 239,164 239,164
Adj. R2 0.087 0.086

18
This is computed as ð0:076  0:029Þ  2:095, where 0:076 and 0:029 are the coefficients on
β CCN
and HIGH_CCN  βCCN, respectively, and 2.095 is the standard deviation of βCCN.
 [Link] Published online by Cambridge University Press
1998 Journal of Financial and Quantitative Analysis

C. Time to Maturity and the Climate Change News Premium

To empirically test Hypothesis 3, we reestimate our baseline regressions
separately for the subsamples of long-term and short-term bonds, respectively,
where long-term bonds are those with a time to maturity longer than 20 years.
Table 4 reports the estimation results. Columns 1 and 3 present the results for the
short-term bond subsample, while columns 2 and 4 report the results for the long-
term bond subsample. The results show that the coefficient estimate on the climate
change news beta of long-term bonds is significantly larger than that of short-term
bonds, although the effect remains significant among short-term bonds. Moreover,
the Z-statistic, which tests the statistical significance of the difference between the
coefficients on βCCN for the two subsamples, suggests that the difference is signif-
icantly different from each other. These results indicate that the effect of βCCN on
future bond returns is more pronounced for long-term bonds.
Our findings that long-term corporate bonds are more sensitive to hedging
demands are consistent with those of Painter (2020). At the same time, the prices of
short-term bonds are also affected by hedging demands, which is consistent with
the results of Goldsmith-Pinkham et al. (2020). Whereas our study investigates the
corporate bond market, both Painter and Goldsmith-Pinkham et al. examine the
municipal bond market. The municipal bond market is highly segmented because
the tax incentives given to local investors cause municipal bonds to be tightly held
by local investors (Schultz (2012) and Schwert (2017)). In contrast, corporate bonds
are predominantly held by institutional investors (Bai et al. (2019)), who are likely
to pay attention to climate risk in all its forms (Krüeger et al. (2020)). A priori, this
difference in clientele between the two markets could lead to differing pricing
effects.

TABLE 4
Time-to-Maturity and Climate Change News Premium

Table 4 reports the results for the regressions in subsamples partitioned based on the time-to-maturity of corporate bonds.
Columns 1 and 3 report the results for the subsample of bonds with time-to-maturity less than or equal to 20 years. Columns 2
and 4 report the results for the subsample of bonds with time-to-maturity greater than 20 years. Z-statistics are for the statistical
test of the difference between the coefficient estimate on βCCN in the short time-to-maturity subsample and the coefficient
estimate on βCCN in the long time-to-maturity subsample. Standard errors are clustered at firm level in all regressions.
t-statistics are presented in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level,
respectively. Variables are defined in the Appendix.
Dependent Variable: Future EXCESS_RETURN

MATURITY ≤20 MATURITY >20 MATURITY≤20 MATURITY >20

Variable 1 2 3 4

βCCN 0.009** 0.032*** 0.026*** 0.061***

(2.013) (4.914) (2.965) (6.059)
Z-statistics for the difference in the 2.963*** 2.550***
coefficients on βCCN
Control variables Yes Yes Yes Yes
Firm fixed effects Yes Yes No No
Bond fixed effects No No Yes Yes
State fixed effects Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes
Month fixed effects Yes Yes Yes Yes
No. of obs. 199,342 39,822 199,342 39,822
Adj. R2 0.086 0.121 0.084 0.123
 [Link] Published online by Cambridge University Press
Huynh and Xia 1999

D. Environmental Determinants of the Climate Change News Beta

In this section, we empirically test Hypothesis 4 by exploring possible envi-
ronmentally related determinants of the climate change news beta. We posit that,
when there are enhanced concerns about climate change in the market, investors
will value bonds issued by firms that either adopt positive environmental corporate
practices or implement progressive transition policies, which provide more effec-
tive hedges against future realizations of climate change risk.19
We employ two alternative proxies for a firm’s environmental performance,
provided by two different ESG rating agencies, namely, MSCI and Sustainalytics.
First, we obtain annual firm-level environmental scores from the MSCI ESG
database, which provides evaluations of a firm’s scores for various aspects of
environmental performance, such as the adoption of waste management and greater
use of renewable energy. Section B in the Supplementary Material details the
subcategories of the environmental scores. Following Hong and Kostovetsky
(2012) and EGKLS, we calculate a net environmental score (ESCORE) for each
firm by subtracting the total score of all negative environmental categories (con-
cerns) from the total score of all positive environmental categories (strengths).
Since the MSCI ESCORE data are available at an annual frequency, we lag the
variable before matching the data with our baseline sample to allow ample time for
the market to incorporate the information into bond prices.
We estimate the regression of βCCN on the lagged values of ESCORE,
HIGH_CCN, an interaction term between HIGH_CCN and lagged ESCORE,
and the same set of control variables as for our baseline regressions. Column 1 of
Table 5 reports the estimation results. The estimated coefficient on the interaction
term HIGH_CCN  ESCORE is positive and significant at the 1% level. These
results indicate that a firm’s good environmental performance reduces its bond
exposure to climate change news risk, which in turn improves its bond value during
high climate change news index periods, when being environmentally friendly is
more valuable.20
An equivalent approach is to examine whether issuers of high-βCCN bonds are
preferred by environmentally responsible institutional investors. To identify these
institutions, we follow Cao, Titman, Zhan, and Zhang (2019) and sort financial
institutions in the Thomson Reuters 13F filing database into three groups, based on
the average ESCORE of firms in their portfolio holdings, where the top tercile
contains institutions with the highest average portfolio ESCORE and the bottom
tercile consists of institutions with the lowest average portfolio ESCORE. Institu-
tions in the top tercile are deemed environmentally responsible.
For each firm, we calculate the fraction of environmentally responsible insti-
tutions (ERIO) as the number of environmentally responsible institutions investing
in the firm, divided by the total number of institutions holding the firm’s equity.

19
This prediction is also consistent with recent studies (e.g., Starks, Venkat, and Zhu (2018) and
Dyck, Lins, Roth, and Wagner (2019)) showing that institutional investors align their investments with
ESG criteria to manage portfolio risk.
20
The stronger effect of ESCORE on the climate change news beta during uncertain times is consistent
with the argument of Lins et al. (2017), who show that firms of strong corporate social responsibility
build strong reputations and trust between themselves and both stakeholders and investors.
 [Link] Published online by Cambridge University Press
2000 Journal of Financial and Quantitative Analysis

TABLE 5
Environmental Determinants of the Climate Change News Beta

Table 5 reports the results for the regression tests of the environmental determinants of the climate change news beta. The
dependent variable is βCCN. All independent variables are lagged. The environmental score (ESCORE) of a firm in column 1 is
described in Section B of the Supplementary Material using data from MSCI ESG Research. In column 2, ERIO is the
environmentally responsible institutional ownership measured as the number of environmentally responsible institutions
divided by the total number of institutions holding a firm’s shares. In columns 3 and 4, ESCORE_SUS and ERIO_SUS are,
respectively, the environmental score and environmentally responsible institutional ownership computed using Sustainalytics
data. In column 5, ENV_RISK is the share of the transcript of the conference call that focuses on political risk related to the
environment. t-statistics computed using clustered standard error at the issuer level are presented in parentheses. *, **, and
*** indicate statistical significance at the 10%, 5%, and 1% level, respectively. Variables are defined in the Appendix.
Dependent Variable: Future βCCN

Variable 1 2 3 4 5

HIGH_CCN  ESCORE 0.064***

(4.965)
ESCORE 0.035*
(1.835)
HIGH_CCN  ERIO 1.032***
(5.836)
ERIO 2.699***
(3.945)
HIGH_CCN  ESCORE_SUS 0.127***
(9.389)
ESCORE_SUS 0.021
(0.698)
HIGH_CCN  ERIO_SUS 0.608***
(4.877)
ERIO_SUS 0.237
(0.736)
HIGH_CCN  ENV_RISK 0.021**
(2.533)
ENV_RISK 0.000
(0.047)
HIGH_CCN 0.227*** 0.364*** 0.162*** 0.235*** 0.022
(9.962) (8.947) (10.868) (7.636) (0.347)
DOWNSIDE_RISK 0.043*** 0.036* 0.008 0.007 0.031**
(3.699) (1.730) (0.961) (0.713) (2.199)
ln(MATURITY) 0.827*** 1.028*** 0.485*** 0.693*** 0.688***
(6.252) (5.665) (3.633) (5.246) (4.535)
ln(1+RATING) 0.216 0.064 1.265*** 0.978*** 0.376
(0.700) (0.128) (3.541) (2.977) (1.048)
ln(AMOUNT OUT) 0.126** 0.109 0.220*** 0.155*** 0.179***
(2.127) (1.324) (4.053) (2.840) (2.724)
REVERSAL 0.025*** 0.041*** 0.019*** 0.029*** 0.031***
(15.310) (14.999) (8.796) (16.599) (14.714)
ILLIQUIDITY 0.001 0.011 0.002 0.007 0.005
(0.074) (0.657) (0.192) (0.617) (0.404)
βBOND_MARKET 0.197*** 0.474*** 0.298*** 0.289*** 0.217***
(2.870) (3.861) (3.610) (3.749) (2.606)
βTERM 0.027*** 0.055*** 0.012*** 0.005* 0.034***
(11.359) (12.291) (4.066) (1.840) (11.642)
β DEFAULT
0.005*** 0.014*** 0.008*** 0.004*** 0.006***
(10.152) (11.165) (11.018) (6.206) (8.583)
β TED
0.021*** 0.073*** 0.026*** 0.014*** 0.029***
(9.186) (11.659) (10.580) (6.241) (8.384)
IDIO_RISK 0.039*** 0.009 0.032** 0.013 0.009
(2.913) (0.531) (2.261) (1.050) (0.673)
LEVERAGE 0.089 0.457 0.479 0.774** 0.225
(0.235) (0.982) (1.195) (2.330) (0.554)
ln(MARKET_CAP) 0.031 0.146* 0.002 0.045 0.014
(0.479) (1.661) (0.036) (0.871) (0.196)
ROE 0.008*** 0.008*** 0.005** 0.005** 0.008***
(3.508) (2.715) (2.017) (2.354) (2.771)
(continued on next page)
 [Link] Published online by Cambridge University Press
Huynh and Xia 2001

TABLE 5 (continued)
Environmental Determinants of the Climate Change News Beta

Dependent Variable: Future βCCN

Variable 1 2 3 4 5

Bond fixed effect Yes Yes Yes Yes Yes

State fixed effects Yes Yes Yes Yes Yes
Year fixed effects Yes Yes Yes Yes Yes
Month fixed effects Yes Yes Yes Yes Yes
No. of obs. 189,918 223,282 140,435 158,306 196,077
Adj. R2 0.151 0.183 0.107 0.091 0.139

Similar to the above analysis, we lag ERIO by 6 months and regress βCCN on the
lagged ERIO value, HIGH_CCN, an interaction term between lagged ERIO and
HIGH_CCN, and controls. The results are reported in column 2 of Table 5. We find
that the coefficients on ERIO and the interaction term between ERIO and
HIGH_CCN are both positive and significant. These results suggest that bonds
whose issuers are preferred by environmentally responsible institutions have a
higher βCCN, and the effect of ERIO on βCCN is stronger when the climate change
news index is high.
Our second source of data is Sustainalytics, which provides firms’ ESG scores
at a monthly frequency starting from Aug. 2009. For each firm in a given month, the
Sustainalytics ESCORE is a weighted average of 57 environmental indicators,
where the weights are proprietary and assigned to an industry depending on the
industry’s exposure to a risk indicator. The score ranges from 0 to 100, with
100 representing the strongest environmental performance. Following Seltzer
et al. (2020), we standardize the Sustainalytics ESCORE by subtracting the mean
and dividing it by the standard deviation. We denote this measure ESCORE_SUS.
We estimate the regression of βCCN on the 1-month-lagged ESCORE_SUS,
HIGH_CCN, an interaction term between the lagged ESCORE_SUS and
HIGH_CCN, and other control variables used in the baseline regression. The
estimation results are reported in column 3 of Table 5. The estimated coefficient
on the interaction term HIGH_CCN  ESCORE_SUS is positive and significant at
the 1% level, which is consistent with our previous finding that, during periods of
high climate change news risk, the bonds of firms with better environmental
performance exhibit a higher βCCN.
We also use the Sustainalytics ESCORE to recalculate the fraction of envi-
ronmentally responsible institutions (ERIO_SUS) and then regress βCCN on the
lagged ERIO_SUS, HIGH_CCN, an interaction term between lagged ERIO_SUS
and HIGH_CCN, and controls. The results are presented in column 4 of Table 5.
Consistent with our previous results, the significant and positive coefficient on
HIGH_CCN  ERIO_SUS suggests that, when climate change news risk is high,
the bonds of firms preferred by environmentally responsible institutions have a
higher βCCN.
To provide further evidence that the climate change news beta reflects inves-
tors’ perception of a bond’s exposure to climate change risk, we employ a text-
based measure constructed by Hassan, Hollander, van Lent, and Tahoun (2019) that
captures analysts’ concerns about a firm’s exposure to political risk associated with
the environment that are raised during conference calls. Specifically, the measure,
 [Link] Published online by Cambridge University Press
2002 Journal of Financial and Quantitative Analysis

denoted ENV_RISK, is computed as the share of the transcript of the conference

call that focuses on political risk related to environment. To the extent that analysts’
questions reflect market participants’ concerns, this measure represents investors’
perception of firm-level environmental policy risk (Ilhan et al. (2021)).
As before, we estimate the regressions of βCCN on the lagged ENV_RISK,
HIGH_CCN, and the interaction term between the lagged ENV_RISK and
HIGH_CCN. In column 5 of Table 5, we find a negative and significant coefficient
on the interaction term, indicating that, during periods of high climate change risk, a
bond’s βCCN is significantly lower since analysts raise more concerns about the
issuer’s exposure to political risk associated with the environment. This result is
consistent with the notion that a low βCCN represents an increase in climate change
risk exposure as perceived by market participants.

E. Environmental Profile and the Climate Change News Premium

The previous section shows that, when marketwide concern about climate
change risk is elevated, the bonds of issuers with stronger environmental perfor-
mance have a higher βCCN. That is, investors prefer high-βCCN bonds because they
have strong potential to hedge against climate risk when the market is most con-
cerned about climate change. In this section, we examine whether the effect of βCCN
on bond returns is, on average, more pronounced among firms that offer strong
hedging potential.
We first follow Ilhan et al. (2021) and Seltzer et al. (2020) and classify
industries into the top polluting industries and non-top polluting industries using
the 2-digit Standard Industrial Classification (SIC). The top polluting industries are
petroleum and coal products; primary metal industries; electric, gas, and sanitary
services; air transportation; trucking and warehousing; water transportation; oil and
gas extraction; railroad transportation; stone, clay, and glass products; paper and
allied products; metal mining; nonclassifiable establishments; chemical and allied
products; general merchandise stores; and textile mill products.
Since the bonds of nonpolluting issuers arguably have better potential to hedge
against climate risk than those of polluters, we expect the negative relation between
βCCN and future returns to be more pronounced in the sample of nonpolluting firms’
bonds than in the sample of polluting firms’ bonds. To test this prediction, we
estimate our baseline model for each subsample and report the results in Table 6.
We find that the coefficient on βCCN is negative and statistically significant in
the subsample of non-top polluters (column 1), while the effect is insignificant in
the subsamples of top polluters (column 2). This finding is consistent with the idea
of stranded assets, where investors divest or avoid these assets altogether because
they face high environmental regulatory risk and do not provide investors with
hedging capacity (Hong, Li, and Xu (2019) and Hong et al. (2020)). Thus, investors
might not distinguish between high- and low-βCCN bonds among polluters’ stranded
assets.
While the industry-level analysis is an intuitive additional test for the inter-
temporal hedging hypothesis, it is a broad classification and does not allow one firm
to exhibit better environmental performance than another firm in the same industry.
We thus conduct another subsample analysis in which we split firms in our sample
 [Link] Published online by Cambridge University Press
Huynh and Xia 2003

TABLE 6
Environmental Profile, Climate Change News Beta, and Future Bond Returns

Table 6 reports the effect of βCCN on future bond returns in various subsamples split based on firms’ environmental profiles.
Columns 1 and 2 report the results of the subsample analysis based on whether the firm is in the top polluting industry.
Columns 3 and 4 report the results of the subsample analysis based on whether the firm’s MSCI ESCORE is above the sample
median value. Columns 5 and 6 report the results of the subsample analysis based on whether the firm’s Sustainalytics
ESCORE _SUS is above the sample median. Standard errors are clustered at the firm level in all regressions. t-statistics are
presented in parentheses. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level, respectively. Variables
are defined in the Appendix.
Dependent Variable: Future EXCESS_RETURN

Top Non-Top Low High Low ESCORE High ESCORE

Polluters Polluters ESCORE ESCORE _SUS _SUS
Variable 1 2 3 4 5 6

βCCN 0.008 0.043*** 0.011 0.053*** 0.018 0.038***

(0.928) (8.684) (0.878) (6.500) (1.275) (4.023)
Control Yes Yes Yes Yes Yes Yes
variables
Bond fixed Yes Yes Yes Yes Yes Yes
effects
State fixed Yes Yes Yes Yes Yes Yes
effects
Year fixed Yes Yes Yes Yes Yes Yes
effects
Month fixed Yes Yes Yes Yes Yes Yes
effects
No. of obs. 47,400 191,764 99,022 93,132 71,077 74,804
Adj. R2 0.094 0.088 0.085 0.092 0.114 0.103

into high- and low-ESCORE subsamples, based on the median value. As before,
we employ two alternate measures of ESCORE, namely, those of MSCI and
Sustainalytics. Columns 3 and 4 of Table 6 report the results for subsample analysis
using the MSCI ESCORE values, while columns 5 and 6 present the results of the
subsample analysis using the Sustainalytics environmental score (ESCORE_ SUS).
We find that the effect of βCCN is stronger and more robust among high-ESCORE
firms. These results indicate that allowing for heterogeneity in environmental
performance among firms does not override the insignificant effect in the subsam-
ple of firms with a low ESCORE, which is again consistent with the concept of
stranded assets. However, among firms with good environmental performance,
investors appear to prefer the bonds of issuers with higher ESCORE values.

V. Robustness Tests
In this section, we conduct a battery of tests to check the robustness of our
findings. First, since Table 2 shows a significant negative effect of the climate
change news beta on future bond returns, a natural question arises as to whether
βCCN has long-term predictive power for future bond returns. We test this question
by regressing future excess bond returns from month t + 2 to month t + 12 on βCCN
measured in month t. The results reported in Panel A of Table A2 in the Supple-
mentary Material indicate that the predictive power of βCCN remains significant
when predicting future 2- to 8-month returns. The predictability, however, becomes
insignificant from month t + 9 onward and does not reverse. The fact that the effect
remains significant beyond month t + 1 suggests that our results are not driven by a
 [Link] Published online by Cambridge University Press
2004 Journal of Financial and Quantitative Analysis

short-run reversal effect or a mechanical effect arising from bid–ask bounce

(Jegadeesh (1990) and Lehman (1990)).21
Second, since a factor affecting bond returns might not necessarily drive yields
(Campbell (1995)), it would be useful to examine whether our baseline findings
hold for bond yields. We follow Nanda, Wu, and Zhou (2019) to compute yield
spreads as the difference between volume-weighted yields on corporate bonds and
the estimated yield on government bonds for the period from Jan. 2005 to Dec.
2016. The monthly trading yield on a corporate bond is the volume-weighted
average of the yield to maturity across intraday transactions at the end of each
month. We report the estimation results in Table A3 in the Supplementary Material.
The coefficient on βCCN is negative and statistically significant at the 1% level,
indicating that bonds with higher climate change news betas have lower future yield
spreads. These results suggest that investors perceive a bond with a higher βCCN to
be less risky, consistent with our central hypothesis.
Third, we employ an alternative climate change news index provided by
EGKLS: the Crimson Hexagon (CH) negative climate change news index. The
CH index is obtained from the data analytics vendor Crimson Hexagon and is only
available from June 2008. It is calculated as the share of all news articles that focus
on climate change and which have been categorized by Crimson Hexagon as news
with negative sentiment. The estimation results reported in Supplementary Material
Table A4 show that the coefficient on βCH remains negative and statistically
significant at the 1% level.
Fourth, a potential concern regarding our results is that the pricing of βCCN could
be specific to a category of credit ratings such as noninvestment-grade bonds, which
are riskier and less liquid (Chen et al. (2007)). We thus partition the sample into two
groups, based on whether the bond’s credit rating is investment grade (i.e., an S&P
credit rating from BBB to AAA) or noninvestment grade (i.e., an S&P credit rating
from B to BB+). As shown in Table A5 of the Supplementary Material, the
coefficient on βCCN is negative and statistically significant in both subsamples.
Moreover, a Z-statistics test shows that the estimated coefficients in the two sub-
samples are not statistically different from each other, suggesting that the effect of
βCCN is equally strong among both investment-grade and noninvestment-grade bonds.
Fifth, we investigate whether the climate change news premium comes from
changes in the cash flow risk. We do so by examining the effects of the issuer’s
overall climate change news beta aggregated across the issuer’ bond betas on firm-
level expected default risk and future cash flows. Following prior studies (Bharath
and Shumway (2008) and Brogaard, Li, and Xia (2017)), we measure default risk as
the expected default frequency (EDF), which captures the probability of the firm’s
cash flows not meeting its debt obligations. A firm’s cash flow is computed as
operating income before depreciation, scaled by the book value of total assets. The
estimation results reported in Table A6 of the Supplementary Material show an
insignificant relation between the firm-level climate change news beta and either

21
To feasibly hedge against climate change risk in the bond market, investors should be able to infer
a bond’s future βCCN from its past βCCN. We examine this question by regressing future βCCN from month
t + 1 to month t + 8 on βCCN in month t. The results reported in Panel B of Table A2 in the Supplementary
Material show that the coefficient on βCCN in month t is large, positive, and statistically significant.
 [Link] Published online by Cambridge University Press
Huynh and Xia 2005

the expected default risk or future cash flows. These results provide suggestive
evidence that the effect of βCCN is potentially driven by investors’ perceptions about
a bond’s exposure to climate change risk, but the firm’s fundamentals are not
necessarily affected by changes in βCCN.
Finally, we address a potential concern that our findings could be driven by the
potential systematic measurement error arising from the method to estimate the
climate change news beta. To mitigate this concern, we conduct a placebo test by
repeating the same analysis on a sample of U.S. government bonds. Table A7 of the
Supplementary Material reports the results of these tests. We find that the relation
between future government bond returns and climate change news (or its beta) is
statistically insignificant. These findings indicate that the negative relation for
corporate bonds is not a mechanical result.

VI. Conclusion
In this study, we examine the effect of climate change news risk on individual
corporate bond returns. We construct a climate change news beta, βCCN, that captures
a bond’s covariance with the climate change news risk index. We show that bonds
with a higher βCCN are associated with lower future returns and the effect of βCCN is
more pronounced during periods of high climate change news risk. Further analysis
of issuers’ environmental profiles suggests that, when marketwide concern about
climate change risk is elevated, the bonds of issuers with stronger environmental
performance have a higher βCCN. These results are consistent with the hypothesis of
intertemporal hedging demand, which posits that investors are willing to pay higher
prices for (and accept lower future returns on) bonds with a higher βCCN, since these
bonds offer better potential to hedge against climate change risk.
Our study’s findings also suggest that a firm’s investment in improving its
environmental performance will help lower its cost of debt financing, especially
when the market is most concerned about climate change risk. These findings have
important implications for business managers and regulators when attempting to
emphasize the important roles of climate change risk, as well as socially responsible
investment.

Appendix. Variable Definitions

Variable Definition Source

EXCESS_RETURN A bond’s monthly return in excess of the monthly risk-free rate, TRACE, Mergent FISD,
measured as a percentage. A bond’s monthly return is calculated as FRB
in equation (1). The risk-free rate is proxied by the 1-month Treasury
bill rate.
βCCN The climate change news beta is estimated from the monthly rolling TRACE, EGKLS
regressions of individual bond excess returns on innovations in the
monthly climate change news index over a 60-month window, after
controlling for the excess market return, bond market illiquidity, the
term spread, the default spread, and the TED spread. At least 30
observations are required in the window to estimate beta.
DOWNSIDE_RISK The average of the four lowest monthly return observations over the TRACE, Mergent FISD
past 36 months (beyond the 10% VaR threshold), multiplied by –1
and measured as a percentage (Bai et al. (2019)).
MATURITY A bond’s time to maturity, measured in years. Mergent FISD
(continued on next page)
 [Link] Published online by Cambridge University Press
2006 Journal of Financial and Quantitative Analysis

Variable Definition Source

ln(MATURITY) The natural logarithm of a bond’s time to maturity. Mergent FISD

RATING A bond’s credit rating as a numerical score, where 1 refers to an AAA Mergent FISD
rating and 22 refers to a D rating.
ln(AMOUNT_OUT) The natural logarithm of a bond’s amount outstanding. Mergent FISD
ILLIQUIDITY Bond illiquidity is computed
at the end of each month t for each bond i TRACE, Mergent FISD
as cov t Δp i,t,τ ,Δp i,t,τþ1 , where Δp i,t,τ is the change in the natural
logarithm of the price on day τ of month t (Bao et al. (2011)).
REVERSAL Excess returns on corporate bonds in the prior month. TRACE, Mergent FISD
βBOND_MARKET The bond market beta is estimated from the monthly rolling TRACE, Mergent FISD,
regressions of individual bond excess returns on the excess bond Bloomberg
market return over a 60-month window. The excess bond market
return is the monthly return on the Barclays Capital Corporate Bond
Index over the monthly risk-free rate.
βTERM The term spread beta is estimated from equation (2). The term spread TRACE, Mergent FISD,
is computed as the difference between the monthly return on the Welch and Goyal (2014)
Ibbotson U.S. long-term government bond index and the 1-month
T-bill return.
βDEFAULT The default spread beta is estimated from equation (2). The default TRACE, Mergent FISD,
spread is computed as the difference between BAA-rated and AAA- Welch and Goyal (2014)
rated corporate bond monthly yields.
βTED The TED spread beta is estimated from equation (2). The TED spread TRACE, Mergent FISD,
is computed as the difference between the 3-month LIBOR, based on FRED
U.S. dollars and the 3-month Treasury Bill rate.
IDIO_RISK The standard deviation of the residuals from the regression of daily CRSP
stock returns on Fama and French’s (1993) 3 daily risk factors over a
month.
LEVERAGE The sum of long-term debt (DLTT), short-term debt (DLC), minority Compustat
interests (MIBT), and preferred stock (PSTK), divided by total assets
(AT).
ln(MARKET_CAP) The natural logarithm of the market value of a firm’s common equity CRSP
(PRC  SHROUT) at the end of each month. The market value of
equity is measured in thousands.
ROE Income before extraordinary items (IB) divided by the book value of Compustat
common equity (CEQ).
β CH
The Crimson Hexagon (CH) climate change news beta is estimated TRACE & EGKLS
from the monthly rolling regressions of individual bond excess returns
on the CH negative climate change news index over a 60-month
window, after controlling for excess market return, bond market
illiquidity, the term spread, the default spread, and the TED spread. At
least 30 observations are required per window.
ESCORE The sum of the net scores (strengths minus concerns) of the MSCI ESG (formerly KLD)
environmental performance subcategories.
ESCORE_SUS Firm environment score provided by Sustainalytics. Sustainalytics
HIGH_CCN A dummy variable equal to 1 when innovations in the monthly climate EGKLS
change news index is greater than the median value, and 0
otherwise.
ERIO Environmentally responsible institutional ownership is measured as MSCI ESG KLD
the number of environmentally responsible institutions divided by the (Sustainalytics),
(ERIO_SUS)
total number of institutions holding a firm’s shares. Following Cao et
Thomson Reuters, WRDS
al. (2019), we define institutions as environmentally responsible
13F holdings database
institutions if their portfolios’ quarterly average MSCI (Sustainalytics)
environment scores are in the top tercile of all institutions.
ENV_RISK The share of the transcript of the conference call that focuses on Hassan et al. (2019)
political risk related to environment

Supplementary Material
To view supplementary material for this article, please visit [Link]
10.1017/S0022109020000757.
 [Link] Published online by Cambridge University Press
Huynh and Xia 2007

References
Addoum, J. M.; S. Delikouras; G. M. Korniotis; and A. Kumar. “Income Hedging, Dynamic Style
Preferences, and Return Predictability.” Journal of Finance, 74 (2019), 2055–2106.
Addoum, J. M.; D. T. Ng; and A. Ortiz-Bobea. “Temperature Shocks and Establishment Sales.” Review
of Financial Studies, 33 (2020), 1331–1366.
Albuquerque, R.; Y. Koskinen; and C. Zhang. “Corporate Social Responsibility and Firm Risk: Theory
and Empirical Evidence.” Management Science, 65 (2019), 4451–4469.
Ang, A.; R. J. Hodrick; Y. Xing; and X. Zhang. “The Cross-Section of Volatility and Expected Returns.”
Journal of Finance, 61 (2006), 259–299.
Bai, J.; Bali, T. G.; and Q. Wen. “Common Risk Factors in the Cross-Section of Corporate Bond
Returns.” Journal of Financial Economics, 131 (2019), 619–642.
Baker, M.; D. Bergstresser; G. Serafeim; and J. Wurgler. “Financing the Response to Climate Change:
The Pricing and Ownership of US Green Bonds.” Working Paper, Harvard University (2018).
Baldauf, M.; L. Garlappi; and C. Yannelis. “Does Climate Change Affect Real Estate Prices? Only If You
Believe in It.” Review of Financial Studies, 33 (2020), 1256–1295.
Bali, T. G. “The Intertemporal Relation Between Expected Returns and Risk.” Journal of Financial
Economics, 87 (2008), 101–131.
Bali, T. G.; S. J. Brown; and Y. Tang. “Is Economic Uncertainty Priced in the Cross-Section of Stock
Returns?” Journal of Financial Economics, 126 (2017), 471–489.
Bali, T. G.; A. Subrahmanyam; and Q. Wen. “The Economic Uncertainty Premium in Corporate Bond
Returns: An Empirical Investigation.” Working Paper, Georgetown University (2019).
Bansal, R.; M. Ochoa; and D. Kiku. “Climate Change and Growth Risks.” Working Paper, Duke
University (2016).
Bao, J.; J. Pan; and J. Wang. “The Illiquidity of Corporate Bonds.” Journal of Finance, 66 (2011),
911–946.
Barnett, M.; W. Brock; and L. P. Hansen. “Pricing Uncertainty Induced by Climate Change.” Review of
Financial Studies, 33 (2020), 1024–1066.
Bekaert, G.; M. Hoerova; and M. Lo Duca. “Risk, Uncertainty and Monetary Policy.” Journal of
Monetary Economics, 60 (2013), 771–788.
Ben-Rephael, A.; B. I. Carlin; Z. Da; and R. D. Israelsen. “Information Consumption and Asset Pricing.”
Journal of Finance, forthcoming (2021).
Bernstein, A.; M. T. Gustafson; and R. Lewis. “Disaster on the Horizon: The Price Effect of Sea Level
Rise.” Journal of Financial Economics, 134 (2019), 253–272.
Bessembinder, H.; K. M. Kahle; W. F. Maxwell; and D. Xu. “Measuring Abnormal Bond Performance.”
Review of Financial Studies, 22 (2008), 4219–4258.
Bharath, S. T., and T. Shumway. “Forecasting Default with the Merton Distance to Default Model.”
Review of Financial Studies, 21 (2008), 1339–1369.
Black Rock. “Adapting Portfolios to Climate Change.” Available at [Link]
rate/literature/whitepaper/[Link] (2016)
Bloom, N. “The Impact of Uncertainty Shocks.” Econometrica, 77 (2009), 623–685.
Brogaard, J.; D. Li; and Y. Xia. “Stock Liquidity and Default Risk.” Journal of Financial Economics,
124 (2017), 486–502.
Burke, M.; S. M. Hsiang; and E. Miguel. “Global Non-Linear Effect of Temperature on Economic
Production.” Nature, 527 (2015), 235–239.
Campbell, J. Y. “Intertemporal Asset Pricing Without Consumption Data.” American Economic Review,
83 (1993), 487–512.
Campbell, J. Y. “Some Lessons from the Yield Curve.” Journal of Economic Perspectives, 9 (1995),
129–152.
Campbell, J. Y. “Understanding Risk and Return.” Journal of Political Economy, 104 (1996), 298–345.
Campbell, J. Y., and G. B. Taksler. “Equity Volatility and Corporate Bond Yields.” Journal of Finance,
58 (2003), 2321–2350.
Cao, J.; S. Titman; X. Zhan; and W. E. Zhang. “ESG Preference and Market Efficiency: Evidence from
Mispricing and Institutional Trading.” Working Paper, Chinese University of Hong Kong (2019).
Chen, L.; D. A. Lesmond; and J. Wei. “Corporate Yield Spreads and Bond Liquidity.” Journal of
Finance, 62 (2007), 119–149.
Choi, D.; Z. Gao; and W. Jiang. “Attention to Global Warming.” Review of Financial Studies, 33 (2020),
1112–1145.
Dell, M.; B. F. Jones; and B. A. Olken. “Temperature Shocks and Economic Growth: Evidence from the
Last Half Century.” American Economic Journal: Macroeconomics, 4 (2012), 66–95.
Deryugina, T., and S. M. Hsiang. “Does the Environment Still Matter? Daily Temperature and Income in
the United States.” Working Paper, University of Illinois at Urbana-Champaign (2014).
 [Link] Published online by Cambridge University Press
2008 Journal of Financial and Quantitative Analysis

Dick-Nielsen, J. “Liquidity Biases in TRACE.” Journal of Fixed Income, 19 (2009), 43–55.

Dick-Nielsen, J. “How to Clean Enhanced TRACE Data.” Working Paper, Copenhagen Business School
(2014).
Dick-Nielsen, J.; P. Feldhütter; and D. Lando. “Corporate Bond Liquidity Before and After the Onset of
the Subprime Crisis.” Journal of Financial Economics, 103 (2012), 471–492.
Dietz, S.; A. Bowen; C. Dixon; and P. Gradwell. “’Climate Value at Risk’ of Global Financial
Assets.” Nature Climate Change, 6 (2016), 676–679.
Dyck, A.; K. V. Lins; L. Roth; and H. F. Wagner. “Do Institutional Investors Drive Corporate Social
Responsibility? International Evidence.” Journal of Financial Economics, 131 (2019), 693–714.
Engle, R. F.; S. Giglio; B. T. Kelly; H. Lee; and J. Stroebel. “Hedging Climate Change News.” Review of
Financial Studies, 33 (2020), 1184–1216.
Fama, E. F., and K. R. French. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of
Financial Economics, 33 (1993), 3–56.
Fama, E. F., and J. D. MacBeth. “Risk, Return, and Equilibrium: Empirical Tests.” Journal of Political
Economy, 81 (1973), 607–636.
Feldhütter, P. “The Same Bond at Different Prices: Identifying Search Frictions and Selling Pressures.”
Review of Financial Studies, 25 (2012), 1155–1206.
Ferrell, A.; H. Liang; and L. Renneboog. “Socially Responsible Firms.” Journal of Financial Econom-
ics, 122 (2016), 585–606.
Gebhardt, W. R.; S. Hvidkjaer; and B. Swaminathan. “The Cross-Section of Expected Corporate Bond
Returns: Betas or Characteristics?” Journal of Financial Economics, 75 (2005), 85–114.
Giglio, S.; M. Maggiori; K. Rao; J. Stroebel; and A. Weber. “Climate Change and Long-Run Discount
Rates: Evidence from Real Estate.” Working Paper, Yale University (2018).
Goldsmith-Pinkham, P. S.; M. Gustafson; R. Lewis; and M. Schwert. “Sea Level Rise and Municipal
Bond Yields.” Working Paper, Yale University (2020).
Greenwood, R., and S. G. Hanson. “Issuer Quality and Corporate Bond Returns.” Review of Financial
Studies, 26 (2013), 1483–1525.
Hassan, T. A.; S. Hollander; L. van Lent; and A. Tahoun. “Firm-Level Political Risk: Measurement and
Effects.” Quarterly Journal of Economics, 134 (2019), 2135–2202.
Heinkel, R.; A. Kraus; and J. Zechner. “The Effect of Green Investment on Corporate Behavior.” Journal
of Financial and Quantitative Analysis, 36 (2001), 431–449.
Hong, H.; G. A. Karolyi; and J. Scheinkman. “Climate Finance.” Review of Financial Studies, 33 (2020),
1011–1023.
Hong, H., and L. Kostovetsky. “Red and Blue Investing: Values and Finance.” Journal of Financial
Economics, 103 (2012), 1–19.
Hong, H.; F. W. Li; and J. Xu. “Climate Risks and Market Efficiency.” Journal of Econometrics, 208
(2019), 265–281.
Hsiang, S. M. “Temperatures and Cyclones Strongly Associated with Economic Production in the
Caribbean and Central America.” Proceedings of the National Academy of Sciences, 107 (2010),
15367–15372.
Ilhan, E.; P. Krueger; Z. Sautner; and L. T. Starks. “Institutional Investors’ Views and Preferences on
Climate Risk Disclosure.” Working Paper, Frankfurt School of Finance and Management (2019).
Ilhan, E.; Z. Sautner; and G. Vilkov. “Carbon Tail Risk.” Review of Financial Studies, forthcoming
(2021).
Jegadeesh, N. “Evidence of Predictable Behavior of Security Returns.” Journal of Finance, 45 (1990),
881–898.
Jostova, G.; S. Nikolova; A. Philipov; and C. W. Stahel. “Momentum in Corporate Bond Returns.”
Review of Financial Studies, 26 (2013), 1649–1693.
Krüger, P. “Corporate Goodness and Shareholder Wealth.” Journal of Financial Economics, 115 (2015),
304–329.
Krüger, P.; Z. Sautner; and L. T. Starks. “The Importance of Climate Risks for Institutional Investors.”
Review of Financial Studies, 33 (2020), 1067–1111.
Lehmann, B. N. “Fads, Martingales, and Market Efficiency.” Quarterly Journal of Economics, 105
(1990), 1–28.
Lesk, C.; P. Rowhani; and N. Ramankutty. “Influence of Extreme Weather Disasters on Global Crop
Production.” Nature, 529 (2016), 84–98.
Lin, H.; J. Wang; and C. Wu. “Liquidity Risk and Expected Corporate Bond Returns.” Journal of
Financial Economics, 99 (2011), 628–650.
Lins, K. V.; H. Servaes; and A. Tamayo. “Social Capital, Trust, and Firm Performance: The Value of
Corporate Social Responsibility During the Financial Crisis.” Journal of Finance, 72 (2017),
1785–1824.
 [Link] Published online by Cambridge University Press
Huynh and Xia 2009

Loughran, T., and B. McDonald. “Textual Analysis in Accounting and Finance: A Survey.” Journal of
Accounting Research, 54 (2016),1187–1230.
Massa, M., and L. Zhang. “The Spillover Effects of Hurricane Katrina on Corporate Bonds and the
Choice Between Bank and Bond Financing.” Journal of Financial and Quantitative Analysis,
forthcoming (2021).
Merton, R. C. “An Intertemporal Capital Asset Pricing Model.” Econometrica, 41 (1973), 867–887.
Nanda, V.; W. Wu; and X. A. Zhou. “Investment Commonality Across Insurance Companies: Fire Sale
Risk and Corporate Yield Spreads.” Journal of Financial and Quantitative Analysis, 54 (2019),
2543–2574.
Painter, M. “An Inconvenient Cost: The Effects of Climate Change on Municipal Bonds.” Journal of
Financial Economics, 135 (2020), 468–482.
Patton, A. J., and M. Verardo. “Does Beta Move with News? Firm-Specific Information Flows and
Learning About Profitability.” Review of Financial Studies, 25 (2012), 2789–2839.
Petersen, M. A. “Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches.”
Review of Financial Studies, 22 (2009), 435–480.
Roll, R. “A Simple Implicit Measure of the Effective Bid–Ask Spread in an Efficient Market.” Journal of
Finance, 39 (1984), 1127–1139.
Schultz, P. “The Market for New Issues of Municipal Bonds: The Roles of Transparency and Limited
Access to Retail Investors.” Journal of Financial Economics, 106 (2012), 492–512.
Schwert, M. “Municipal Bond Liquidity and Default Risk.” Journal of Finance, 72 (2017), 1683–1722.
Seltzer, L.; L. T. Starks; and Q. Zhu. “Climate Regulatory Risk and Corporate Bonds.” Working Paper,
University of Texas at Austin (2020).
Starks, L. T.; P. Venkat; and Q. Zhu. “Corporate ESG Profiles and Investor Horizons.” Working Paper,
University of Texas at Austin (2018).
Stern, N. “The Economics of Climate Change: The Stern Review.” Cambridge University Press,
Cambridge (2007).
Welch, I., and A. Goyal. “A Comprehensive Look at the Empirical Performance of Equity Premium
Prediction.” Review of Financial Studies, 21 (2007), 1455–1508.