Professional Documents
Culture Documents
SSRN Id4597677
SSRN Id4597677
SSRN Id4597677
Nhan Huynh2,*
2 Department of Applied Finance, Macquarie Business School, Macquarie University, Australia.
*Corresponding author – Email: david.huynh@mq.edu.au
Hoa Phan3
3School of Economics, Finance and Marketing, RMIT University, Australia
Email: thi.kieu.hoa.phan@rmit.edu.au
Abstract
This study delves into the relationship between firm-level climate risk and the
probability of default and how managerial ability adjusts this relationship. Using a broad
sample of U.S firms from 2002 to 2021, we confirm the positive impacts of firm-level exposure
to climate change risks on default risk. This effect is particularly pronounced within firms
grappling with greater financial constraints and distress. We unveil the moderating influence
of internal governance and information efficiency, which serve as powerful mechanisms for
mitigating the adverse impact of climate risk on a firm's probability of default. Furthermore,
we identify two other significant mechanisms at play: a firm's performance and profitability
volatility, both of which are substantially influenced by climate risk. Further, our findings
present compelling evidence that managerial ability acts as a safeguard against the
detrimental repercussions of climate risk on credit risk. Our findings remain consistent across
several robustness tests, additional and sensitivity analyses.
Keywords: climate change exposure, climate risk, default risk, managerial ability, financial
distress
Acknowledgement: The authors are indebted to Ling Nguyen (Macquarie University), An Pham &
Tran Nguyen Thai Ha (Van Lang University), Wutthikrai Torrobrum (Sheffield Hallam University), and
Chamaiporn Phokha (Thammasat University) for their valuable comments and suggestions. Dung Thi
Thuy Nguyen would also like to acknowledge the financial support from Van Lang University, Ho Chi
Minh City, Vietnam. The usual caveat applies.
Climate change has become a heated but unsettled debate among economists,
policymakers, and scholars over the past decades. In recent years, environmental quality and
climate change have been getting worse due to escalating greenhouse gas emissions1;
however, the current efforts are insufficient to tackle these challenges (Tran et al., 2022).
Under such consensus, the risks of environmental degradation and climate change pose for
the overall economies and businesses have been increasingly recognised. As such, a growing
body of literature has proposed theoretical and empirical evidence that climate risk directly
or indirectly affects firms regarding their capital structure (Nguyen & Phan, 2020; Jung et al.,
2018), dividend policy (Balachandran & Nguyen, 2018; Zhou & Wu, 2023), and financial
performance (Nguyen, 2018; Huang et al., 2018; Addoum et al., 2020). Given the
considerable impacts on corporate performance and decisions, climate change risk can also
exert certain pressures on firms’ financial status as well as their capacity to meet their
financial obligations. Several studies also consider the potential links between climate change
and corporate default risk by using carbon emissions as a proxy for climate risk (Kabir et al.,
2021; Capasso et al., 2020). However, firms disclose their emission levels according to their
past business strategies rather than their projected plans, which may result in inadequate
In addition, extensive prior research underscores the critical role of managerial ability
structural effectiveness (Demerjian et al., 2012; Demerjian et al., 2013). Firms headed over
understanding of risk (Gan, 2019; Wali Ullah et al., 2023). Essentially, managerial ability
1According to report of the World Meteorological Organization on Greenhouse Gas Bulletin, the rise in carbon
dioxide levels from 2020 to 2021 is significantly higher than the rate during the past ten years. See:
https://public.wmo.int/en/greenhouse-gas-bulletin
2023).
Although there is a rich strand of literature confirming the interplay between climate
risk and a firm's creditworthiness, the complicated relationship between firm-level risks,
firm-level climate risk and default risk, as well as the underlying managerial mechanisms
governing this relationship, remains basically unexplored area. Thus, this study endeavors to
fill this critical gap in the existing body of knowledge by addressing two principal questions:
(1) Does firm-level climate risk exert a discernible impact on their probability of default? and
(2) How does the managerial ability of firms shape and mediate this relationship? In this
study, we utilise the comprehensive data set on firm-level climate change exposure developed
by Sautner et al. (2023)2 to explore both internal mechanisms (financial performance) and
pressures from increasing stakeholders’ climate consciousness (Krueger et al., 2020; Bolton
& Kacperczyk, 2021). A comprehensive understanding of the intricate nexus between firm-
level climate risk, default risk, and the mediating factors is not only fundamental for shaping
corporate strategies but also holds profound implications for policymakers and stakeholders
Using quarterly data of the US firms covering the period 2002 to 2021, we provide
evidence that climate risk exposure exerts positive effects on default risk measured by
Distance to Default. Our results also uncover stronger impacts of firms’ exposure to climate
analyses confirm that the impacts are more pronounced for firms with higher financial
constraints and distress, and weaker for firms with internal corporate governance and
information efficiency. Furthermore, Moreover, our investigation reveals the operation of two
volatility, both of which bear visible impacts in firms with higher climate risk. Notably, our
2See Section 2 and Appendix A for detail descriptions of the Climate change exposure (CCE) indicators developed
by Sautner et al. (2023).
as an effective shield for firms from the deleterious consequences of climate risk on credit
risk. In addition, the effects of climate risk are more amplified in carbon-intensive firms and
industries, which are also driven by the significant tremors to climate risk exposure. Our
results remain robust to the introduction of a series of robustness tests and sensitivity
Our study contributes to the extant literature in the following ways. First, this paper
differs from prior studies in that we focus on the impacts of corporate-level climate risk
exposure on default risk, such as corporate decisions (Balachandran & Nguyen, 2018; Zhou
& Wu, 2023), financial performance (Nguyen, 2018; Zhang & Wang, 2014), or corporate risk
(Capasso et al., 2020; Kabir et al., 2021). Given the current literature on firm-level climate
risk, the carbon emission utilised as climate risk indicators can only capture the opportunity
and regulatory risks and ignore the physical factors of climate risk (Nguyen et al., 2023). In
addition, carbon footprints can only reflect self-reported past performance and past corporate
strategic settings, such as investment decisions on green or brown projects (Sautner et al.,
2023; Matsumura et al., 2014). As such, the utilization of carbon emissions, as in Capasso et
al. (2020) and Kabir et al. (2021), can provide biased results for sectors with lower direct
strategies, this study is the first to establish the impacts of firm-level climate risk exposure on
default risk, extracted from the conference calls using textual analyses. The series on this topic
is therefore supplemented with more comprehensive analyses of firm-level risk to the legal,
opportunistic, and physical dimensions of climate risk. One study of actual interest and
closely related to our paper is Nguyen et al. (2023), which confirms the climate risk proxied
by climate risk disclosures in annual filings and default risks of the S&P 500 non-financial
firms during 2010–2019. Our study can overcome the current limitations of Nguyen et al.
(2023) regarding climate risk measures. We can extend their findings with more robust
results with a quarterly dataset of 4,354 unique firms over a 20-year period, allowing us to
capture the impacts of several major climate treaties. Further, as stated by Nguyen et al.
regulatory risk and not opportunity and physical risk. In our study, we can provide more
comprehensive analyses by using the firm-level climate risk measures by Sautner et al.
(2023), which can capture the market's perception of a firm's exposure to several upside or
downside aspects associated with climate risk, namely, technological opportunities, physical
exploring the moderating effects of managerial ability on the relationship between firm-level
climate risk and the probability of default by utilizing the comprehensive measure of
managerial ability by Demerjian et al. (2012). Prior studies primarily focus on corporate
financial performance (Cheung et al., 2017; Fernando et al., 2020; Kumar & Zbib, 2022; Phan
et al., 2020), risk management (Bonsall et al., 2017; Cheng & Cheung, 2021; Abdesslem et al.,
2022), earnings quality (Demerjian et al., 2013), and corporate policies (Andreou et al., 2017;
Lee et al., 2023). By expressing the unexplored area of managerial ability's role in climate risk
mitigation, this study not only broadens the scope of analysis but also underscores the
implications of climate risk across various facets of corporate activities. In other words, by
successfully confirming the power of managerial in moderating the adverse effects of climate
risk on default risk, this study can extend the current strand of literature on the value of
managerial ability associated with the survival of firms at times of uncertainty. As such, we
can propose unique implications for stakeholders and firms in easing the impacts of climate
Third, from the series of additional analyses in this study, this study offers a vital
relationship between climate risk exposure and default risk. We confirm the plausible
information efficiency. Firms’ default risk is also driven by their future financial performance
study also adds value to the recent literature by considering the impacts of international
climate policies, which boosts the awareness of climate change and its damaging impacts on
corporate activities. Finally, our findings also propose critical implications for the academic
guidelines. The evidence from this study also raises practical propositions for businesses in
The remainder of the paper proceeds as follows. Section 2 offers the literature review,
relating to climate risk, default risk, and managerial ability. Section 3 describes the sample
and variable constructions, and Section 4 presents the main results. Section 5 reports the
In light of the adverse effects of climate risk on corporate operations, the existing
literature has explored various explanations for how climate change may affect their stability
and profitability, as well as elevate the potential default risk. Some studies have indicated that
countries with warmer temperatures tend to experience lower GDP growth rates or per capita
incomes compared to those in cooler climates (Burke et al., 2015; Bansal & Ochoa, 2012).
Additionally, economic downturns or recessions can negatively impact a firm's future cash
flows, thereby heightening default risk (Kabir et al., 2021). Furthermore, the instrumental
resulting in lower climate risk, can enhance firm value by fostering positive relationships with
various stakeholders, including suppliers, lenders, and investors (Kabir et al., 2021;
climate risk may have a diminished competitive advantage and reduced ability to sustain
consistent dividend payments, thereby increasing uncertainty regarding the company's future
cash flows due to both identifiable and unforeseen opportunities, regulatory changes, and
physical risks (Jung et al., 2018; Dewaelheyns et al., 2023). Firms experiencing cash flow
uncertainty may consequently find themselves lacking the capital to fulfil their financial
climate risk on firm performance and profit volatility, which can potentially lead to defaults
in firms with higher exposure to climate-related factors. First, prior studies have revealed that
climate risk can contribute to reduced firm performance (Krueger, 2015; Huang et al., 2017;
Nguyen, 2017; Busch & Lewandowski, 2017; Ferrat, 2021; Ding et al., 2021). Ding et al. (2021)
emphasize that climate-related natural disasters like tsunamis, floods, and wildfires compel
firms to accelerate the depreciation of their assets when reporting financial results. Other
research indicates that firms with greater climate risk may incur higher debt costs. Chava
may face elevated expenses in terms of bank financing and debt covenants, as banks may
scrutinize firms' environmental profiles to manage risk. Itzkowitz (2013) and Huang et al.
(2018) demonstrate that firms exposed to higher climate risks are more inclined to utilise
long-term debt and maintain larger cash reserves. Furthermore, credit ratings may be
downgraded in response to firms with heightened climate risk, resulting in increased debt
costs for such entities (Li et al., 2014; Barley, 2009; Graham et al., 2001). Consequently, a
Psillaki et al. (2010) and Kabir et al. (2021) strengthen the notion that a firm's performance
serves as a predictive indicator of the probability of default. Firms with superior performance
The second school of thought posits that climate risk contributes to the volatility of
firms' profits. According to Huang et al. (2017), climate risk, as assessed through extreme
weather events, can be a factor in asset volatility. This is because severe weather conditions
can lead to physical damage to a firm's fixed assets, resulting in a decrease in both the assets'
value and the income generated from them, potentially harming firms' profits. Utilizing a
panel dataset spanning from 2004 to 2018 across 42 nations, Kabir et al. (2021) demonstrate
that companies with elevated climate risk, as proxied by carbon emissions, experience greater
profit volatility, which in turn may elevate credit risk for these firms. The findings by Lemma
et al. (2019) and Zhu and Zhao (2022) also robustly confirm that firms' profit volatility serves
as a conduit through which climate risk impacts default risk. Based on the aforementioned
arguments on the impacts of climate risk on firm performance and stability, we theorize that
firms with higher climate risk may have higher default risk. Our first hypothesis, thus, is
established as follows:
been recognized and explored in prior research. Specifically, the effects of executive managers
on main corporate activities and performance have been regarded as crucial research topics
in the field of finance. the Hambrick and Mason (1984) upper echelons theory claims that
differences in managerial abilities influence how they make decisions and how well they can
explain relevant information, thereby affecting corporate strategy (Chen et al., 2023; Lee et
al., 2023). Indeed, firms with greater managers’ ability enhance the firm's information
environment (Baik et al., 2018), better comprehend market and industry trends, and
accurately forecast product demand (Demerjian et al., 2012; Phan et al., 2020).
In terms of operational activities, Koester et al. (2017) reveal that managers with higher
abilities are adept at minimizing income tax liabilities, engaging in strategic state tax
planning, and directing income towards international tax havens. They attribute this to the
which enables them to align firm decisions with effective tax strategies. Additionally, Biswas
et al. (2023) demonstrate that managerial competence negatively moderates the link between
product market competition and real activity manipulation (RAM). RAM encompasses
supply volume, and minimizing advertising costs. In the realm of investment activities,
previous research indicates that high-ability managers are associated with increased
corporate investment, greater spending on research and development (R&D), and enhanced
investment prospects (Andreou et al., 2017; Lee et al., 2023; Lee et al., 2018). However, this
relationship holds significant relevance in specific conditions, including factors such as lower
default risk, financial flexibility, competition within the industry, and certain managerial
characteristics like gender, age, stock options, and income levels. Moreover, Yung and Chen
(2018) reveal that highly competent managers tend to allocate resources to R&D expenses
rather than capital spending, resulting in improved firm value. In the context of financing
evidence suggesting that managers with greater abilities are positively linked to higher credit
ratings.
From the perspective of investors, Mishra (2014) demonstrates that shareholders tend
to demand higher returns from firms with more capable managers, indicating a positive
correlation between managerial ability and the cost of equity. Further, a highly competent
manager or good corporate governance reduces the information asymmetry and minimizes
the moral hazard issue where managers can act in their self-interest, which in turn decreases
the agency risk to equity market participants (Ali et al., 2018; Ashbaugh-Skaifea et al., 2006).
Equity investors, thus, can forecast a lower level of future cashflow volatility, thereby
decreasing the default risk of a firm, which ensures that the impact of reputation risk due to
performance, previous research consistently proves that high-ability managers are associated
with increased profitability (Demerjian et al., 2012; Kumar & Zbib, 2022; Koester et al., 2017).
Several studies have delved into the impact of managerial ability on various other facets,
including the firm's innovation output (Chen et al., 2015), information environment (Baik et
al., 2018), default risk (Ali et al., 2018; Chen et al., 2023; Chemmanur et al., 2009), as well as
the moderating role of managerial ability in mitigating the adverse effects of crude oil price
fluctuations on business performance (Phan et al., 2020) and the negative relationship
between equity financing and future abnormal returns (Demerjian et al., 2012).
Given the current literature, most previous studies have primarily focused on
reputation, educational background, skillset, and managerial style (Rajgopal et al., 2006;
Adams & Ferreira, 2009). Additionally, previous research has employed Data Envelopment
Analysis (DEA) to assess managerial talent or capabilities within specific industries (Murthi
et al., 1997; Leverty & Grace, 2012). However, relying solely on industry-specific
10
of managerial abilities. Demerjian et al. (2012) introduce a more precise approach to gauge
approach is applicable to a broad range of firms and outperforms existing methods for
With the aforementioned literature and the managerial ability by Demerjian et al.
(2012), we therefore hypothesize that a manager's high ability can weaken the positive impact
of climate risk on default risk. Climate risk can be considered a negative signal to evaluate the
firm’s solvency (Kabir et al., 2021); however, Bonsall et al. (2017) show that higher-ability
managers who more effectively increase revenues and financial stability, thereby possibility
mitigate the effect of climate risk on the probability of default. Further, competent managers
tend to implement more efficient risk management plans, which can diminish the negative
impact of climate risk on the probability of survival of a firm. As such, the second hypothesis
is proposed as follows:
Hypothesis 2 (H2): The positive effect of firm-level climate risk on default risk is weaker
11
3.1. Data
In this study, we utilise the sample of all U.S listed firms from Compustat, which
provides the quarterly financial data from 2002 to 2021. The firm-level climate change
exposure (CCE) measure is obtained from Sautner et al. (2023)3, which is standardised by
deducting the sample mean and dividing by the standard deviation. This firm-level climate
conference call transcripts, which can depict firms’ exposure to various facets of climate
change risks, including physical climate shocks, opportunity exposure, and regulatory risk.
Further, the measures of Sautner et al. (2023) can alleviate the issues of recognizing “niche
journalists, and financial market participants during the conference calls (Webersinke et al.,
2021).
We collect the firm-level default risk data from the Risk Management Institute - Credit
comprised two indicators of default risk, including Merton’s (1974) distance to default (DTD)
and Bharath and Shumway’s (2008) expected default frequency (EDF), which are widely
employed in the literature (Hillegeist et al., 2004; Bharath & Shumway, 2008; Nadarajah et
al., 2020)4. The higher the value of DTD, the lower the default risk, which is inverse for EDF.
utilities (SIC 4900–4999), and industries that are not clearly defined. We also drop the
quarter-firm observations with missing data for all variables used in the main models.
In this study, we also data on firm-specific variables that are standard for the default
12
Market to Book ratio, Leverage, Profitability, Cash holdings, Sales growth, Stock return,
Analyst following, Stock return volatility, Stock liquidity, Tangibility. All continuous
variables are winsorized at 1% level on both sides. Our final sample of non-missing
information of 126,208 firm-quarter observations from 4,354 unique firms. The descriptions
current literature. Prior studies utilise indirect proxies like CEO press visibility (Milbourn,
2003; Rajgopal et al., 2006) as an indicator of managerial ability. Using the manager fixed
effects to estimate the managerial ability, Bamber et al. (2010) assume the existence of
manager-specific effects when corporate strategies change, while the absence of changes
suggests reliance on previous policies without introducing new ones or personal touches.
Computing from the principal component analysis (PCA), a general ability measure by
Custódio et al. (2013), is derived from five CEO-specific variables, such as career diversity,
employment history, and experiences. Overall, these indirect proxies have faced criticism for
relying on comprehensive data about managers and firms over time, with the possibility of
To address those limitations of prior measures, we utilise the Managerial Ability (MA)
scores devised by Demerjian et al. (2012) in our analysis. These scores are instrumental in
allowing us to discern the potential ramifications of firm-level climate risk on the probability
of default. Demerjian et al. (2012) precisely construct the MA scores using an advanced two-
step methodology, which encompasses frontier analysis, data envelopment analysis (DEA),
5 For full details about this measure, please refer to Demerjian et al. (2012). The data for this variable is publicly
13
effectively a company generates revenue relative to industry peers using specified inputs,
including inventory costs, general and administrative expenses, net operating leases, fixed
assets, research and development expenses, purchased goodwill, and other intangible assets.
Firms achieving maximum efficiency receive a score of one, while lower scores indicate a
greater deviation from the efficiency frontier. In the second stage, a tobit regression is
conducted on the efficiency scores, accounting for various firm attributes such as size, age, cash
reserves, foreign operations, market share, and segment concentration. The residual obtained
from this regression denotes the firms’ managerial ability score, which is less susceptible to
measurement errors. As such, what sets this measure apart is its reliance on consistently
available financial data over time, resulting in reduced noise when compared to other
Given its credibility and reliability, this measure of managerial ability has been widely
embraced and highly regarded within the finance literature, as evidenced by its extensive
utilization in studies by notable scholars (Lee et al., 2018; Demerjian et al., 2013; Koester et
al., 2017; Kumar & Zbib, 2022; Phan et al., 2020; Andreou et al., 2017). As the data from
Demerjian's database is only available until 2020, we compute the managerial ability (MA)
score for all firms in our sample in 2021 by utilizing the simple moving average from MA during
the past three years. This approach can be rationalised by long-term managerial ability, as in
Doukas and Zhang (2021). Then, we create a dummy variable for firms with MA scores higher
than the average score of all firms over the sample period. In other words, the MA dummy in
our merged dataset equals to one for four quarters within a given financial year if a firm belongs
14
To estimate the impacts of firm-level climate change exposure on default risk, we use
where, DTD is our main variable of interest - the firm-level default risk proxied by
Distance to default6 and CCE is the climate change exposure of firm i in quarter t. 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 is
a vector of firm-specific characteristics that are specified in Section 3.1. Following the literature
on corporate default risk (Bharath & Shumway, 2008; Brogaard et al., 2017; Nadarajah et al.,
2021), we lag all our control variables by one year relative to our dependent variables to
mitigate the concern of reverse causality7. We also control for firm-fixed and time-fixed effects
in all equations. In addition, to correct for potential cross-sectional and serial correlation, all
standard errors are clustered at the firm level. The descriptions of all variables are reported in
Table 1. To consider the role of managerial ability in moderating the impacts of climate risk
where, MA is a dummy variable, which equals one if the MA score of firm i in year t is
higher than the mean of all firms over the sample period or equals zero otherwise. The set of
main variables, controls, and fixed effects are identical to Equation (1). Based on our second
hypothesis, firms with higher MA scores are less influenced by CCE than those with lower
scores. In other words, the estimated coefficient of the interaction term (𝛽3 ) is expected to be
6 The alternative default proxy - expected default frequency (EDF) is utilized in the robustness tests, which is
reported in Table 5.
7 Our results are qualitative unchanged when we perform contemporaneous regressions for both Equations (1) and
(2).
15
Table 2 summarizes key variables in our study, focusing on those included in our main
models. The mean distance to default (DTD) is 5.46 with a standard deviation of 2.648,
consistent with Nguyen et al.'s (2023) findings. Expected default frequency (EDF) has a mean
of 0.548 and a standard deviation of 0.157, aligning with prior U.S. default risk literature
(Brogaard et al., 2017; Nadarajah et al., 2021). Firm-level climate change exposure (CCE)
ranges from 0.000 to 0.208, with a mean of 0.093 and a standard deviation of 0.178. These
CCE values are slightly lower than those in Sautner et al. (2023), who examined data from
10,673 companies across 34 nations. Notably, CCOE has a higher mean of 0.036 compared to
CCRE and CCPE, which have average bigram counts of 0.007 and 0.003, respectively.
Regarding control variables, the Market to Book Ratio has a mean of 2.870, while Leverage
and Profitability have means of 0.224 and 0.031, respectively. Cash holdings range from 0.011
to 0.352, and Sales growth varies between -0.119 and 0.264. The average Stock return is 0.15,
with Stock volatility averaging 0.336. Stock liquidity in our sample has an average value of
We present the baseline results in Table 3 for the impacts of firm-level climate risk - CCE
and its components on default risk. We use ordinary least square (OLS) regression with firm
and time-fixed effects and clustered robust standard errors at the firm level. In column 1, the
indicating the negative correlation between CCE and Distance to default (DTD). In other
words, firms with higher idiosyncratic climate risk face a higher probability of default. In terms
of economic connotation, a one standard deviation rise in CCE results in 0.63%8 change in
default risk on a quarterly basis. To some extent, our findings strengthen the conclusion by
8We compute the estimated economic effect by multiplying the CCE coefficient (-0.195) by the standard deviation
of CCE (0.178) and dividing by it the mean of DTD (5.460).
16
default risk. Given the attention of climate change, higher CCE means that firms experience
both inherent expenses (reducing reputation and competitive advantage) and explicit expenses
(increased compliance costs, operational or interest expenses) (Zhou et al., 2020; Kabir et al.,
2021). As such, firms are more likely to suffer from profitability uncertainties and default on
climate risk would adversely impact the firm’s future cash flow forecasting, which is destructive
to their financing activities as well as default risk (Cadez et al., 2019; Li et al., 2022).
We further consider the impacts of three distinct components of CCE on default risk in
Columns (2)-(4). The estimated coefficient for CCOE is negative and statistically significant at
the 5% level. Regarding CCRE, we also obtain a negative and statistically significant coefficient
DTD. Our results suggest that firms exposed to higher risks, either climate regulatory issues or
climate opportunity risk tend to have a higher chance of default. Also, our findings not only
highlight but underscore a compelling connection for firms that tackle climate-related
default. This robust correlation highlights the profound impact that climate regulation has on
a firm's financial stability and resilience, imparting a sense of urgency to address these critical
issues. Conversely, when delving into the climate physical risk channel, we confront a complex
landscape marked by long-term risks meshed with multiple layers of uncertainty. The presence
challenge in accurately modelling physical risk. This intricate web of uncertainty ultimately
physical risk, utilizing asset-level data within forward-looking models appears as a promising
avenue. This approach has already spurred a competitive 'climate intelligence arms race'
within financial markets, as acknowledged by Keenan (2019) and Hain et al. (2022).
17
acute physical events solely to anthropogenic factors, known as extreme event attribution.
Given these challenges, the relatively muted relationship between physical risk and firm
4.2. Firm-level climate risk and default risk: The roles of managerial ability
Having established a relationship between a firm’s climate risk and probability of default,
we seek to examine the moderate impacts of managerial ability in this section. Using the mean
of firms’ MA scores to create a dummy for high-score firms, we utilise Equation (2) with the
with our prior results, the effect of CCE on DTD is negative and statistically significant across
all indicators. Given the interaction terms, all estimated coefficients are positively significant,
indicating that firms with higher MA scores can moderate the positive impact of CCE on
DTD. In other words, we can confirm that managers with more able managers can help lessen
the harmful impacts of idiosyncratic climate risk on their creditability, supporting our second
Hypothesis. Overall, our results are in line with prior studies on managerial ability that more
able managers are better at assessing and addressing risks, including navigating through
2013; Andreou et al., 2017). Further, better management teams are thus able to achieve higher
future earnings (Demerjian et al., 2012), lower variability of future performance, more
favourable credit ratings, and lower cost of debt (Bonsall IV et al., 2017), and lower credit risk
(Ali et al., 2018). When firms are more exposed to climate risk, higher managerial ability would
positively relate to the successful certification of firm’s performance to creditors and other
stakeholders in general (Chemmanur et al., 2009), thus reducing information asymmetry and
18
To strengthen our findings, we employ various robustness tests to validate our findings.
Initially, we reconfirm our findings by using the alternative dependent variables of Expected
default frequency (EDF) with the baseline model and report the results in Panel A of Table 5.
All coefficients of CCE and its elements are positively significant, indicating that firms with
higher CCE experience higher expected default frequency (EDF) or higher default risk.
Overall, our results remain consistent with the baseline findings that firms more exposed to
climate change experience a higher probability of default. In Panel B, we also utilise the EDF
to consider the moderate impacts of managerial ability on the nexus between CCE and default
risk by utilizing Equation (2). Overall, the estimated coefficients of the interaction terms are
all negatively significant across all CCE indicators, demonstrating the modified impacts of
managerial ability in reducing the impacts of CCE on firms’ credit risk. In Appendix A, we
also utilise an alternative sample of annual data to consider the relationship between CCE,
DTD and MA. Overall, our results are qualitatively unchanged when employing an annual
dataset. We also consider the impacts of an aggregate of the Climate Change News (CCN)
Index developed by Engle et al. (2020). In Appendix B, we include the CCN index as an
additional control variable to capture the public attention to climate risk. Overall, with the
In the next robustness check, we address these endogeneity concerns and verify the
reliability of the baseline finding by employing several approaches. First, we propose four
robustness checks using several alternative econometric approaches, which are reported in
Panel A of Table 6. For all robustness checks, we utilise our baseline model – Equation (1)
with the DTD as a dependent variable. In column (1), we utilise the random sampling method
19
sampling from an assumed infinite population. This method allows us to see whether our
baseline results still hold if we use a sample that is closer to the population. In column (2),
the positive impacts of CCE on default remained unchanged when we utilise the two-step
system GMM with the instrumental variable (IV) of the lagged value of CCE. There are well-
established tracks of studies confirming that the S-GMM estimator can produce reliable and
and dynamic sources of endogeneity. This robustness has been demonstrated in prior studies
(Schultz et al., 2010; Wintoki et al., 2012; Nguyen et al., 2023; Tran et al., 2022). In column
(3), we further consider an identification test of falsification test – the Placebo test. By
replacing the CCE scores of firms in a given quarter-year with a randomly drawn CCE score
from the sample in the Placebo test, the estimated coefficient is statistically insignificant. As
such, we can further reinforce the robustness of our baseline findings. In column (4), we adopt
an instrumental variable (IV) approach - the two-stage least square (2SLS) to address the
possible endogeneity. In these tests, we use the average CCE of the firms headquartered in
the same county as an IV. Overall, we also obtain the negatively significant coefficient,
There are two other possible sources of endogeneity, omitted variable bias and reverse
causality, which can bias our primary findings regarding how CCE affects default risk. As
Appendix C, we consider the impacts of CCE in previous periods on the current default risk
(DTD) by following the approach of Atif and Ali (2021). We report the results for Lag 1 and 2
in Panel A and B, respectively. Our results indicate that lagged CCE indicators are positively
correlated with DTD in the current period, suggesting that the route of causation runs from
CCE to default risk but not vice versa. In Appendix D, we utilise the Oster (2019) tests to
20
Overall, the results uphold that our results are highly suspect to be impaired by an omitted
variable bias.
In previous sections, we have empirically confirmed the nexus between climate risk and
default risk. In this section, we consider potential channels that may drive the relationship
between firms’ climate risk and default risk. The current literature indicates that climate risk
may lead to lower firm performance with higher profit volatility (Krueger, 2015; Huang et al.,
2017; Nguyen, 2017; Busch & Lewandowski, 2017; Ferrat, 2021; Ding et al., 2021). Ding et al.
(2021) explain that climate-related natural disasters require firms to increase the depreciation
of these assets when preparing financial reports. A significant increase in expenses will cause
a substantial decrease in the firm’s performance. Kabir et al. (2021) and Psillaki et al. (2010)
confirm that firm’s performance is a channel to predict the probability of default. In other
words, better performance means lower default risk and vice versa. Consequently, we argue
that profitability and performance volatility could be a channel through which idiosyncratic
climate risk affects firms' default risk. In this study, we employ two main variables of ROA and
SDROA to proxy for profitability and profitability volatility. SDROA is computed by taking the
rolling window covering quarters t-4 to t (Balachandran & Nguyen, 2018; Kabir et al., 2021).
To improve the robustness of our results, we also utilise industry-adjusted ROA and SDROA
in our analyses.
To confirm these channels, we conduct a two-stage regression. In the first stage, two
intermediary variables of profitability and profitability volatility (ROA and SDROA) are
9 We apply Oster's (2019) approach, which assumes that the coefficients' power related to the R-squares in
regressions with and without controls can be maintained to establish an identifiable range. Following the
assumptions proposed by Mian and Sufi (2014) in line with Oster (2019), we construct both the lower and upper
bounds of the identified range, setting 𝛿 = 1 and 𝑅𝑚𝑎𝑥 = min (2.2𝑅̃,1). We also consider the extreme values from
Oster's analysis, using 𝛿 = 1 and RMAX = 1.
21
In the second stage, we regress our primary dependent variable (DTD) on CCE and
mediator variables, in addition to other control variables. To better capture the effects, we also
introduce interaction terms between CCE and mediator variables in our expanded regression
models. We consider our mechanisms confirmed if the following conditions are met: a) CCE
shows a significant correlation with ROA and SDROA, and b) both mediator variables and
interaction terms are significantly associated with DTD. We construct the following models for
The results reported in Table 7, for the first stage regression in Panel A, we find that all
estimated coefficients of CCE are negatively (positively) for ROA (SDROA). This result
indicates that an increase in CCE decreases firms’ profitability and increases profitability
volatility. In Panel B, the results from the second stage reveal that ROA is positively associated
with DTD, demonstrating that an increase in ROA can reduce the firms’ default risk. The
interaction term (𝐶𝐶𝐸𝑖,𝑡 × 𝑅𝑂𝐴𝑖,𝑡 ) are all positive and significant, indicating that higher CCE
results in lower profitability, which could induce the default risk of firms. Taken together, our
results can confirm two potential mechanisms behind the negative relationship between CCE
22
risk in Table 8. To test this proposition, we interact CCE with the dummy variables identified
financial distress, including the KZ index10 (Phan et al., 2020; Chen & Wang, 2012), Altman
Z-score11 (Zhang et al., 2020; Chen & Wang, 2012), dependence on external finance - EFD12
(Rajan & Zingales, 1998) in Panel A. In Panel B, we utilise three internal governance
Ownership) (Ferreira et al., 2017), and information efficiency proxied by Analyst Coverage -
AC14 (Chang et al., 2006). For each factor, we create a dummy variable that equals one if the
value of firm i in given quarter is higher (lower) than the sample mean or equals zero
otherwise. Then, we interact CCE with the identified dummy variables and re-estimate the
baseline Equation (1). As supposed, all coefficients of the interaction terms in Panel A are
negative and statistically significant, indicating the influences of CCE on default risk are more
pronounced for firms confronting higher financial distress and external financing. In other
words, firms with potential cash flow problems or high leverage have higher default risk when
they become more exposed to climate change (Zhou & Wu, 2023; Kabir et al., 2021). In Panel
B, we obtain positively significant coefficients for the interaction terms between CCE and IO
and AC. For the HHI, the interaction coefficient enters positive but insignificant value.
Overall, as a proxy for good internal governance, firms with higher institutional ownership
exhibit lower positive impacts of CCE on DTD. With a lower significant level for coefficient of
AC, however, we can also conclude that the impacts of CCE on DTD can be degraded when
10 Following Baker et al. (2003), the KZ Index is computed as −1.002 × Cashflow − 39.368 × Dividends − 1.315 ×
Cash + 3.139 × Leverage.
11
Following Chen and Wang (2012), we compute the Altman Z-score as:
Z − score = 1.20X1 + 1.40X2 + 3.30X3 + 0.60X4 + 0.999X5, with X1 to X5 are ratio of EBIT to book value of
assets, retained earnings to book value of assets ratio, working capital to book value of assets ratio, ratio of market
value of equity to total liabilities, and ratio of net sales to book value of assets, respectively. We create a dummy
variable equals to 1 for firms if the value of the Altman Z-score is lower than 1.81.
12 In this study, we calculate the External Finance Dependence (EFD) using the approach outlined by Rajan and
Zingales (1998). EFD is derived as the difference between capital expenditures and cash flow from operations, scaled
by capital expenditures.
13 The HHI (Herfindahl–Hirschman Index) is calculated by summing the squares of the proportions of a firm's
shares held by its top five institutional investors. Higher HHI values signify concentrated ownership among a few
large institutional investors (Ferreira et al., 2017).
14 Prior research uses Analyst Coverage as a proxy for firm-level information efficiency, as it reflects the greater
information flow associated with an increased number of analysts (Chang et al., 2006; Nadarajah et al., 2021).
23
Table 9 presents the regression results models that examine the relation between climate
risk and default risk by considering the cross-sectoral heterogeneity. In Panel A of Table 9, we
investigate whether the impact of climate risk on default risk is different between firm types.
Prior works show that companies with higher carbon emissions tend to experience more
serious environmental issues which may negatively impact financial outcomes (Balachandran
& Nguyen, 2018; Nguyen, 2018; Safiullah et al., 2021; Kabir et al., 2021; Capasso et al., 2020).
Thus, to eliminate the impact of industries, in column (1) and column (2), we exclude firms in
the energy sector and those in the high-tech sector15. In column (3), to identify the impacts of
climate risk on the ‘financially prudent’ companies compared to their counterparts (Capasso
et al., 2020), we exclude firms with no ratings from our data sample.
In Panel B of Table 9, we further confirm that carbon-intensive levels can accentuate the
positive effects of CCE on default risk by using three different classifications of Cohen et al.
(2022), IIhan et al. (2021), and industry-average computed from the firm-level carbon
emissions. To rule out the confounding effect of the energy sector on our results, we exclude
all the fossil fuel and mining (extractive) companies, which is consistent with Capasso et al.
(2020). We find that all our results remain consistent even after such above groups of firms
5.6. Additional tests: CCE shocks and major climate policy events
To further confirm the impacts of climate risk on default risk, we rely on the effects of
15 We follow the method of Loughran and Ritter (2004) to classify high-tech and non-high-tech firms. High-tech
firms are classified as those in 4-digit SIC codes of 3571, 3572, 3575, 3577, 3578, 3661, 3663, 3669, 3671, 3672, 3674,
3675, 3677, 3678, 3679, 3812, 3823, 3825, 3826, 3827, 3829, 3841, 3845, 4812, 4813, 4899, 7371, 7372, 7373, 7374,
7375, 7378, and 7379.
24
classified based on the times higher (shocks) of ∆CCE compared to the sample mean in each
quarter. We find that the economic effect and the statistical significance increase with the
intensity of the shocks to CCE. In the last column, we deploy the Propensity Score Matching
approach (PSM) while accounting for significant shocks. This strategic choice is employed to
effectively mitigate the risk of reverse causality and self-selection bias stemming from potential
such, we can further confirm that our results are not determined by the divergences between
firms with and without large ∆CCE. The treated group includes firms with ∆CCE three times
higher than the sample mean in each quarter16. We employ the nearest neighbour matching
method to rigorously ensure that firms exhibiting high ∆CCE (treatment group) are indeed
sufficiently comparable to their matched counterparts with low ∆CCE (control group). In each
firm-quarter observation, we diligently pair firms from the treatment group with those in the
control group based on the closest propensity scores, all while guaranteeing that the most
significant disparity between the propensity scores of each firm-quarter and its matched peer
falls within an absolute value of 0.1%. Subsequently, we proceed to gauge the influence of CCE
on DTD within this matched sample, as presented in Column 4 of Panel A. Notably, the
coefficient associated with CCE in this matched sample exhibits statistical significance at the
5% level and is consistently negative. This reaffirms that the negative impact of CCE on DTD
persists even after accounting for the removal of firm-specific characteristics, thereby
substantiating and reinforcing our baseline findings using the PSM approach.
ramifications of major climate treaties on the intricate relationship between CCE and default
risks. In this study, we deliberate four key climate policy events, including 2007 EPA Emission
Legislation in, 2012 Doha Climate Summit, 2015 Paris Agreement in 2015, and Paris
16In unreported results, we perform the t-tests for statistical differences between the characteristics of treated and
control firms after the matching process. The results suggest that the comparable trend conjecture is not violated in
our analysis as all control variables are not statistically different between two groups.
25
in-differences (DID) analysis. We build a dummy - Post for pre- and post-period adjoining
those events, including Post takes the value of one if the observed year is in one year after the
event year period and zero otherwise. We then interact our independent variable CCR with this
dummy (CCR×Post) and re-estimate the baseline model as in the following model:
𝐷𝑇𝐷𝑖,𝑡 = 𝛼0 + 𝛽1 𝐶𝐶𝐸𝑖,𝑡 + 𝛽2 𝐶𝐶𝐸𝑖,𝑡 × (1 − 𝑃𝑜𝑠𝑡) + 𝛽3 𝐶𝐶𝐸𝑖,𝑡 × 𝑃𝑜𝑠𝑡 + 𝛽4 𝑃𝑜𝑠𝑡 + ∑ 𝛿𝑘 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑖,𝑡 + 𝜀𝑖,𝑡 (5)
Our empirical results convincingly illustrate that the apparent positive influence of CCE
on default risk becomes particularly more pronounced during the post-period encompassing
four important climate events spanning from 2007 to 2016. This empirically demonstrates the
notion that firms exhibiting higher exposure levels are fundamentally more responsive to
climate risk factors, intensely in the wake of heightened climate awareness, as confirmed by
26
This study offers evidence of climate risk's influence on firms' ability to meet their
financial obligations promptly. Our findings indicate a positive relationship between climate
risk and default risk, as measured by Distance to Default, suggesting that higher climate
exposure increases the likelihood of default. This effect is more pronounced for financially
constrained firms. However, firms with stronger internal governance, such as higher
institutional ownership and better information efficiency, tend to mitigate the impact of
climate risk on default risk. Furthermore, the effects of climate risk on default risk are more
prominent in carbon-intensive firms and industries. Importantly, our results highlight the role
of managerial ability as a protective factor, shielding firms from the adverse consequences of
climate risk on credit risk. Additionally, our study identifies two other significant mechanisms
at play: future performance and profit volatility, both of which have adverse effects on firms
Our findings have significant implications for key stakeholders in firms, including
lenders, policymakers, and regulators. Firm managers should recognise the adverse effects of
climate risk on a company's credit risk and adjust their financial decisions accordingly. For
instance, they can reduce leverage ratios to mitigate financial distress stemming from high
projects or products to minimize the impact of climate risk on default risk. Lenders, such as
banks and financial institutions, should incorporate climate risk into their credit risk
management models. Given that climate risk can affect the financial stability of both lenders
and borrowers, government oversight and regulations should mandate firms to disclose
27
28
29
30
This table reports the results from Oster (2019) tests to handle the potential omitted variable bias in our models.
We utilize the Mian and Sufi (2014) notions of Oster (2019) to construct the lower and upper bounds of the identified
set, with 𝛿 = 1 and 𝑅𝑚𝑎𝑥 = min (2.2𝑅̃,1) and the extreme ones of 𝛿 =1 and RMAX =1.
31
Abdesslem, R. B., Chkir, I., & Dabbou, H. (2022). Is managerial ability a moderator? The effect of
credit risk and liquidity risk on the likelihood of bank default. International Review of
Financial Analysis, 80, 102044.
Ali, S., Liu, B., & Su, J. J. (2018). Does corporate governance quality affect default risk? The role
of growth opportunities and stock liquidity. International Review of Economics & Finance,
58, 422-448.
Andreou, P. C., Karasamani, I., Louca, C., & Ehrlich, D. (2017). The impact of managerial ability
on crisis-period corporate investment. Journal of Business Research, 79, 107-122.
Ashbaugh-Skaife, H., Collins, D. W., & LaFond, R. (2006). The effects of corporate governance on
firms’ credit ratings. Journal 0f Accounting and Economics, 42(1-2), 203-243.
Atif, M., & Ali, S. (2021). Environmental, social and governance disclosure and default
risk. Business Strategy and the Environment, 30(8), 3937-3959.
Baik, B., Brockman, P. A., Farber, D. B., & Lee, S. (2018). Managerial ability and the quality of
firms’ information environment. Journal of Accounting, Auditing & Finance, 33(4), 506-527.
Baker, M., Stein, J. C., & Wurgler, J. (2003). When does the market matter? Stock prices and the
investment of equity-dependent firms. The Quarterly Journal of Economics, 118(3), 969-
1005.
Balachandran, B., & Nguyen, J. H. (2018). Does carbon risk matter in firm dividend policy?
Evidence from a quasi-natural experiment in an imputation environment. Journal of
Banking & Finance, 96, 249-267.
Bamber, L. S., Jiang, J., & Wang, I. Y. (2010). What’s my style? The influence of top managers on
voluntary corporate financial disclosure. The Accounting Review, 85(4), 1131-1162.
Bansal, R., & Ochoa, M. (2012). Temperature, aggregate risk, and expected returns. NBER
Working Paper, 17575, 10-11.
Bharath, S. T., & Shumway, T. (2008). Forecasting default with the Merton distance to default
model. The Review of Financial Studies, 21(3), 1339-1369.
Biswas, P. K., Ranasinghe, D., & Tan, E. K. (2023). Impact of product market competition on real
activity manipulation: Moderating role of managerial ability. Accounting & Finance, 63(1),
247-275.
Bolton, P., & Kacperczyk, M. (2021). Do investors care about carbon risk? Journal of Financial
Economics, 142(2), 517-549.
Bonsall IV, S. B., Holzman, E. R., & Miller, B. P. (2017). Managerial ability and credit risk
assessment. Management Science, 63(5), 1425-1449.
Brogaard, J., Li, D., & Xia, Y. (2017). Stock liquidity and default risk. Journal of Financial
Economics, 124(3), 486-502.
Burke, M., Hsiang, S. M., & Miguel, E. (2015). Global non-linear effect of temperature on economic
production. Nature, 527(7577), 235-239.
Busch, T., & Lewandowski, S. (2016). Corporate carbon and financial performance: A meta-
analysis. In Academy of Management Proceedings (Vol. 2016, No. 1, p. 11657). Briarcliff
Manor, NY 10510: Academy of Management.
Cadez, S., Czerny, A., & Letmathe, P. (2019). Stakeholder pressures and corporate climate change
mitigation strategies. Business Strategy and the Environment, 28(1), 1-14.
Capasso, G., Gianfrate, G., & Spinelli, M. (2020). Climate change and credit risk. Journal of
Cleaner Production, 266, 121634.
Chang, X., Dasgupta, S., & Hilary, G. (2006). Analyst coverage and financing decisions. The
Journal of Finance, 61(6), 3009-3048.
Chemmanur, T. J., Paeglis, I., & Simonyan, K. (2009). Management quality, financial and
32
33
34
35
Default risk
The annual average of distance to default for gauging how far a limited-liability
Distance to default (DTD) RMI-CRI, NUS
firm is away from default.
Expected default frequency The annual average of the expected default frequency is the substitution of DTD
RMI-CRI, NUS
(EDF) into a cumulative standard normal distribution.
Firm climate exposure
The firm-level exposure to climate change using word combinations for earnings
Climate change exposure conference calls. Climate change exposure bigrams are scaled by the total
Sautner et al. (2023)
(CCE) number of bigrams in the transcript of earnings conference calls, and it's
multiplied by 1,000
Components of climate
change exposure
Climate change The relative frequency with which bigrams that capture opportunities related to
opportunity exposure climate change occur in the transcripts of earnings Sautner et al. (2023)
(CCOE) conference calls.
The relative frequency with which bigrams that capture regulatory
Climate change regulatory
shocks related to climate change occur in the transcripts of earnings conference Sautner et al. (2023)
exposure (CCRE)
calls
The relative frequency with which bigrams that capture physical
Climate change physical
shocks related to climate change occur in the transcripts of earnings conference Sautner et al. (2023)
exposure (CCPE)
calls.
Managerial ability (MA)
The dummy variable equals one if the firms' managerial ability scores in year t are higher Demerjian's database
MA
than the whole sample mean over the examined period or equals zero. and authors’ calculations
Firm-level control
variables
Firm Size The natural logarithm of total assets. Compustat
The ratio of the market value of the common equity and its balance sheet value of
Market to Book ratio Compustat
the ordinary equity.
Leverage Leverage ratio measured as the ratio of total debts and total assets. Compustat
Cash holdings Measured by total cash plus marketable securities to lagged total assets. Compustat
Stock return The buy-and-hold stock return of the given quarter. Refinitiv
The annualised standard deviation of daily stock returns computed over a given
Stock return volatility Refinitiv
quarter of the financial year.
The natural logarithm of the quarterly Amihud illiquidity ratio over a quarter
multiplied by -1 by the average of the daily Amihud illiquidity ratio. The day
Stock liquidity Refinitiv
Amihud illiquidity ratio is the average of the ratio of the daily absolute stock
return to the trading volume on that day multiplied by 105
Tangibility The ratio of net property, plant, and equipment to total assets. Compustat
Governance indicators
Analyst following The number of analysts following during a given quarter. IBES
Institutional Ownership The ratio of institutional holding during a given quarter. Refinitiv
36
37
38
39
This table presents the regression results that investigate the relationship between managerial ability, climate
change exposures and default risk using the sample consisting of 4,354 U.S firms between 2002 and 2021. The
climate change exposure data is from Sautner et al. (2023). The alternative default risk is proxied by the Expected
default frequency (EDF). CCE is the measure that identifies the firm-level exposure to climate change using word
combinations for earnings conference calls. The firm-level Managerial Ability (MA) scores are computed by
Demerjian et al. (2012). We control firm and time-fixed effects in all specifications. Standard errors are clustered
at the firm level and are presented in parentheses under the associated coefficients. ∗∗∗, ∗∗, and ∗ indicate
significance at the 1%, 5%, and 10% levels, respectively. The descriptions of all variables are reported in Table 1.
40
This table presents the robustness checks for the relation between managerial ability climate change exposures
and default risk using the sample consisting of 4,354 U.S firms between 2002 and 2021. Panel A reports the
baseline results for four alternative econometric approaches, including Bootstrap, two-step system GMM, Placebo,
and two-stage least squares (2SLS). Panel B reports the result from S-GMM for the moderate impacts of MA. The
climate change exposure data is from Sautner et al. (2023). The default risk is proxied by the Distance to default
(DTD). CCE is the measure that identifies the firm-level exposure to climate change using word combinations for
earnings conference calls. The firm-level Managerial Ability (MA) scores are computed by Demerjian et al. (2012).
We control firm and time-fixed effects in all specifications. Standard errors are clustered at the firm level and are
presented in parentheses under the associated coefficients. ∗∗∗, ∗∗, and ∗ indicate significance at the 1%, 5%, and
10% levels, respectively. The descriptions of all variables are reported in Table 1.
41
Dependent variable: ROA & ROA Industry-adjusted ROA SDROA Industry-adjusted SDROA
SDROA (1) (2) (3) (4)
CCE -0.097*** -0.084** 0.031** 0.016**
(0.007) (0.024) (0.017) (0.018)
Controls Yes Yes Yes Yes
S.E. clustered by Firm Yes Yes Yes Yes
Firm FE Yes Yes Yes Yes
Time FE Yes Yes Yes Yes
Observation 126,208 126,208 120,654 120,654
Adjusted R-squared 0.561 0.609 0.446 0.338
42
43
44
45