Energy and Climate Change 5 (2024) 100134

Contents lists available at ScienceDirect

Energy and Climate Change

journal homepage: www.sciencedirect.com/journal/energy-and-climate-change

Full Length Article

Climate policy uncertainty, media coverage of climate change, and energy

markets: New evidence from time-varying causality analysis
Ecenur Uğurlu-Yıldırım a, *, Özge Dinç-Cavlak b
a
Department of Business Administration, Social Sciences University of Ankara, Hükümet Meydanı No: 2 06030, Altındağ, Ankara, Turkey
b
Department of Business Administration, Emniyet Mahallesi Muammer Bostancı Caddesi 06500, Ankara Hacı Bayram Veli University, Beşevler, Ankara, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: Climate change is one of the most critical issues in the last decade, making investigating climate change risks’
Climate policy uncertainty effects on the economy vital. Employing a novel time-varying Granger causality approach, we test causality from
Renewable energy climate policy uncertainty (CPU) to the index returns of clean energy and nonrenewable energy sectors between
Non-renewable energy
November 2009 and December 2021. Our analysis generally reveals a significant causality from CPU to S&P
Time-varying causality
clean energy sector index return throughout the sample period. In contrast, very limited significant causality is
Climate change
observed running from the CPU to the S&P nonrenewable energy index return. This result implies that in­
vestments in clean energy firms are affected by the CPU due to the cost of reversing the decisions in uncertain
environments. On the other hand, the U.S. newspaper media coverage of climate change has a significant impact
not only on the clean energy index returns but also on non-renewable energy index returns, implying an increase
in the media coverage regarding climate change influences the awareness of investors on climate change, which
affects their trading strategies.

1. Introduction with an increase in the ecological awareness of people in the last de­
cades, as investors start to consider the firms’ climate responsibility
Climate change is one of the most crucial environmental and human disclosures when they are making investment decisions [75].
issues, and globally, it is one of the leading systematic risks in this The severity and regularity of extreme climate incidents around the
century (see [2,38]). Moreover, it is now widely accepted that financial world have increased recently due to the rise in the intensity of green­
stability is threatened by climate change [15]. Climate change impacts house gases, which is caused by the massive misuse and employment of
the companies’ financial performances directly by altering the natural mineral resources (see [42,75]). Studies suggest that climate change risk
circumstances of the companies’ operation and production and indi­ caused 71 % of the direct economic costs among all-natural catastrophes
rectly by swapping working and competitive environments such as between 1990 and 2015 (see [56,75]). Investigating the effect of climate
climate policy and regulations [75]. However, firms’ sensitivity to change risks on the economy has become very important with the
climate change varies [51]. The energy industry is one of the most continued increase in global temperatures [21]. Climate change’s eco­
affected sectors by climate change and policies related to the climate. nomic impacts can be classified as physical risk, which is caused by
Greenhouse gas emission is one of the main causes of climate change, increasing temperatures, and transition risk that arises due to govern­
and it is impelled by non-renewable energy sources such as fossil fuels. ments’ opinion to apply policies to decrease emissions [66]. Climate
Hence, to convey the world to the low carbon economy, transforming change policies have various indirect impacts on the financial perfor­
from fossil fuels to clean energy is an essential policy instrument (see mances of companies. However, there is a lack of research investigating
[69]). Moreover, while the demand for energy is increasing because of how the uncertainty surrounding these policies affects the indices of the
industrial development and an increase in population, a shortage in energy sector. In this regard, the primary purpose of this study is to
fossil fuel sources makes the transformation to clean energy sources investigate the impact of climate policy uncertainty and the media
mandatory (see [33,49,67]). Yang and Wang [85] state that the financial coverage of climate change on the returns of clean energy sector index,
performance of companies is affected by environmental governance proxied by the S&P global clean energy index, and non-renewable

* Corresponding author.
E-mail address: ecenur.yildirim@asbu.edu.tr (E. Uğurlu-Yıldırım).

https://doi.org/10.1016/j.egycc.2024.100134

Available online 22 April 2024

2666-2787/© 2024 Elsevier Ltd. All rights reserved.
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

energy sector index, proxied by S&P 500 oil, gas, and consumable fuels transition to clean energy sources has recently started because of envi­
industry index, by employing the novel time-varying Granger causality ronmental concerns (see [20,82]). The consumption of clean energy
technique. No prior studies to our knowledge have investigated the rises with the changes in climate policies (see [49,54]). Since 2005, the
time-varying causality between climate policy uncertainty, climate consumption of clean energy has risen from 13.33 % to 18.86 %, and by
change awareness, and clean energy and non-renewable energy sector 2040, it is expected to increase to more than 50 % of investment in
index returns. global energy (see [33,49]). The implementation of renewable energy is
Studies examining the effect of CPU on the energy sector have been improved, and greenhouse gas emission is reduced with the application
conducted in the relevant literature. The importance of clean energy of climate-related policies (see [7,49,55]). The undesirable results of
consumption in mitigating climate change is underscored by its role in climate change on the economic system can be reduced with an increase
sustainable development and its attractiveness to environmentally in clean energy consumption (see [69]). Clean energy can be defined as
conscious investors (see [18,25,31,81]). However, the uncertainty sur­ the energy obtained from the natural and persistent stream of energy
rounding climate policies poses financial risks and affects investment that occurs in the environment, which helps to mitigate climate change
decisions, particularly in the energy sector (see [48,49,82]). Research (see [18,31]). Clean energy plays a crucial part in sustainable develop­
suggests that climate policy uncertainty dampens investment and in­ ment, as it helps to diminish carbon emissions and enhances the effi­
fluences organizational behavior, leading to a wait-and-see approach ciency of energy [25]. Therefore, clean energy consumption is vital in
among companies (see [34,45,68,72,77]). Based on robust theoretical mitigating climate change [81]. Attention to the companies operating in
foundations, this research seeks to enhance current understanding by the clean energy sector by environmentally responsible investors has
investigating how the effect of CPU on the returns of clean and nonre­ also increased not only because of the concerns about climate change
newable energy indices varies over time. Furthermore, another strand of but also because of their greater returns (see [43,47,57,82]). Liu and
the literature presents that the heightened environmental awareness Hamori [52] state that the number of investors who include clean energy
among consumers and investors has led to increased inspection of stocks in their portfolios has increased. As clean energy investment can
companies’ climate change mitigation efforts, affecting their brand offer energy demand and diminish climate problems, investigating the
value and financing capabilities (see [29,63,75]). However, research on link between the clean energy sector and climate change is crucial (see
the direct impact of climate change awareness on stock market returns is [18]). Existing empirical studies present that climate change affects
limited. This paper aims to fill this gap by investigating the dynamic clean energy firms, but the number of these studies is few. Studies
causality between media coverage of climate change and returns of including Hsu et al. [41], Nagar and Schoenfeld [55], and Bartram et al.
clean and non-renewable energy indices. [7] demonstrate that the stock prices of companies are influenced by
We add to the aforementioned literature by investigating the cau­ climate risk. While Hsu et al. [41] employ cross-section analysis to show
sality between CPU, media coverage of climate change, and returns of that climate risk is a factor in pricing stock returns, Bartram et al. [7]
clean and non-renewable energy indices. There are three major contri­ present changes in the climate policy that affect the volatility in the
butions. First, this study pioneers the use of time-varying Granger cau­ prices of firms in the energy sector.
sality analysis to examine the effect of CPU and media coverage of The climate policies underscore the importance of clean energy,
climate change on energy markets. This new time-varying Granger prompting leaders worldwide to prioritize its development in pursuit of
causality methodology not only permits the potential changes in the zero carbon emissions. Consequently, adjustments in climate policies of
causal direction but also enables us to detect the particular dates of the this nature are deemed beneficial for advancing the cause of clean en­
beginning and end dates of any period of causality. Moreover, this ergy [44,83]. Climate change policies indirectly influence companies’
method implements a robust econometric technique for integration and financial performances in several ways. Climate policy constraints on
cointegration features (see [40,70]). By employing this method, the the sectors with intensive energy reduce the market value of these in­
paper offers a new approach to understanding the dynamic relationships dustries. Although firms prefer to use intensive carbon energy because of
between the aforementioned factors and energy sector returns. Second, its economic benefits, governments legally make them diminish their
this study is the first one that uncovers a time-varying causality from carbon emissions to prevent the destruction of the ecological environ­
media coverage of climate change to both clean energy and nonrenew­ ment [8]. For instance, Chan et al. [16] state that the production costs of
able energy sector index returns. Finally, to the best of our knowledge, industries like electricity significantly increase with the reductions in
our paper is the first one that sheds light on the significant role of climate European emissions. Lemoine [46] presents that beliefs in upcoming
policy uncertainty in influencing stock returns within the clean and climate policy influence fossil capital investment, which changes the
non-renewable energy sectors by using the time-varying Granger cau­ fossil fuels demand.
sality methodology. The climate change economic models and climate forecasting are
Our paper proceeds as follows. Section 2 delineates the theoretical exposed to estimation bias if policy uncertainty is not considered [17].
foundations and the development of hypotheses. In the third section, the Climate policy uncertainty can be defined as a situation in which
data and methodology are discussed. The fourth section gives the whether the government will alter the existing climate policies or how
empirical results. Section five concludes the paper and presents policy and when they will alter these policies cannot be exactly anticipated.
implications. Uncertainty in economic policy results from various sources and reduces
the total investment [3]. Empirical studies support this view by showing
2. Theoretical framework and hypothesis that uncertainty in multiple fields, such as monetary, fiscal, and regu­
latory, has an adverse economic effect (see [10,27,65]). Fried et al. [29]
Our paper primarily covers two areas of research. The first explores examine the effects of the transition risk of climate policy in the United
the influence of CPU on the returns of both clean energy and non- States (US) and show that uncertainty in the possibility of the US
renewable energy indices, while the second investigates the impact of applying federal climate policy reduces investment, which reduces to­
climate change awareness on the returns of these same indices. day’s carbon emissions. Although there is no federal climate policy that
the US government has executed yet, companies have changed their
2.1. CPU and index returns investment decisions regarding the likelihood of the adaptation of such
policy in the near future [29]. In an uncertain environment, reversing
In a growing literature, studies have devoted significant effort to the decisions is costly; therefore, companies hesitate to make decisions
understanding the effect of climate change on the energy industry. In the and prefer to follow a wait-and-see strategy, which influences their
last century, the energy requirement of the world has been mainly ob­ performance and organizational behavior (see [34,45,68,72,77]). By
tained from burning fossil fuels, such as crude oil [19]. Nevertheless, the employing time-varying Granger causality analysis, Xi et al. [83]

2
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

reaffirm the notion that alterations in climate policies can variably investor attention generally corresponds to improved stock performance
impact the consumption of renewable energy sources over time. More [50]. Groß-Klußmann and Hautsch [35] state that investors start to
specifically, they show that while CPU does not consistently affect decide their trading strategies based on the news with the increase in
geothermal energy consumption throughout the entire duration of the globalization and an almost instant flow of information. In the last de­
study, it exhibits a sporadic influence on the consumption of other cades, the attention of ordinary consumers and investors on the climate
renewable energy sources and the overall consumption of renewable responsibility disclosures of the companies have been increased rapidly.
energy. Barnett [6] investigates the impact of climate policy uncertainty Stakeholders have started to evaluate the risk to the firm’s brand value
on asset prices in the oil market. By using the present carbon tax to by investigating the firm’s attitude to the risk of climate change. Com­
pursue a stochastic process as an uncertainty proxy, Bretschger and panies signal strong future development by actively dealing with climate
Soretz [14] examine the overall equilibrium effects of climate policy change and revealing emission information. Therefore, the financing
uncertainty. By constricting an uncertainty index, which includes US capability of the firm gets stronger as the acknowledgment of the
equity market volatility, uncertainties on global economic policy, investor increases. Recently, the environmental awareness of consumers
geopolitical risk, monetary policy, and US policy, Wang et al. [82] has increased, making green products sold above the market price [75].
demonstrate that clean energy volatility can be predicted by uncertainty Heightened public attention to climate change and pollution could
indices. positively impact the returns on sustainability indices for several rea­
Studies also show that the uncertainty index has a significant impact sons. Firstly, sustainable investors may expedite purchasing stocks of
not only on the renewable energy sector but also on the traditional en­ sustainable firms and divesting from conventional ones as environ­
ergy market (see Liang et al., 2020; [49,82]). Oil, coal, and natural gas, mental awareness rises. Secondly, increased public attention to envi­
which constitute traditional energy sources, have historically served as ronmental issues could raise environmental consciousness among
foundational pillars for national development, supplying fuel for ma­ traditional investors, leading them to develop preferences for sustain­
chinery and raw materials for production. However, their reliance on able stocks. Thirdly, anticipating increased demand for stocks of sus­
combustion processes results in substantial greenhouse gas emissions, tainable firms during periods of heightened public attention, investors
necessitating increased emphasis on transitioning to green and may engage in opportunistic profit-seeking strategies, further driving up
low-carbon alternatives to mitigate carbon output [64]. The effect of returns on sustainability indices [22,36,58]. Studies show that the ma­
policy uncertainties on the traditional energy market has been presented jority of US adults reinforce escalating energy prices to fight climate
in substantial empirical studies before. For instance, by employing the change [29]. For instance, air pollution serves as supplementary data
ARDL model, Shang et al. [69] present that while climate policy un­ that can increase the visibility and trading activity of both new energy
certainty reduces the non-renewable energy demand in the US, it in­ and polluting companies. Empirical research has consistently shown
creases the renewable energy demand. Bouri et al. [11] examine the that public attention plays a significant role in shaping the correlation
impact of the CPU on the returns on green and brown energy stocks and between environmental violations and the stock returns of implicated
demonstrate that, particularly during crisis periods, the CPU has a sig­ companies [50,84]. By utilizing Twitter-based sentiment, Reboredo and
nificant positive effect on green stock returns rather than brown stock Ugolini [63] demonstrate that sentiment does not significantly impact
returns. Similarly, Chen et al. [18] demonstrate that climate change renewable energy returns. Likewise, using the Google search volume
influences energy consumption, which affects the investment in clean index, Song et al. [73] empirically show that the effect of investor
energy. However, the magnitude of this effect changes significantly sentiment on clean energy markets is weak. In another study, Ouadghiri
among countries. et al. [58] empirically investigate the impact of public attention on
Following the examination outlined above, two hypotheses are climate change and pollution on the weekly returns of US sustainability
posited. stock indices, such as the DJSI US and the FTSE4Good USA Index,
H1a. There is a causality from CPU to S&P global clean energy index compared to their conventional parent indices like the S&P 500 Index
returns. and the FTSE USA. Analyzing data from 2004 to 2018, they find that
H1b. There is no causality from CPU to S&P 500 oil, gas, and heightened public attention to environmental issues positively in­
consumable fuels industry index returns. fluences the returns of US sustainability stock indices, potentially
We contribute to this literature strand by investigating the time- attracting various types of investors to favor stocks of sustainable firms.
varying causality from the climate policy uncertainty to the S&P However, the research on the impact of climate change awareness on the
global clean energy index and the S&P 500 oil, gas, and consumable companies’ stock market returns is scant. This paper adds to the litera­
fuels industry index. Our paper relates to Ren et al. [64] which examines ture by investigating and presenting the causality from the media
the reciprocal relationship between CPU and both traditional energy coverage of climate change and global warming to the clean energy
sectors (such as oil, coal, and natural gas) and green markets (including index returns and non-renewable energy index returns, by employing a
clean energy, green bonds, and carbon trading) by employing new time-varying causality approach. By utilizing this novel method­
time-varying Granger causality methodology. However, our study de­ ology, we apart from the previous studies as past empirical in­
viates from them as the public attention to climate change is considered vestigations into the correlation between public focus on environmental
in our analysis. concerns and stock performance primarily utilize the event study
approach (see [28,39]).
2.2. Public attention and index returns After the analysis provided earlier, two hypotheses are proposed to
be tested.
Another line of the literature that our paper contributes to is the H2a. There is a causality from US newspaper coverage of climate
literature on the effect of media coverage of climate change on the stock change or global warming to S&P global clean energy index returns.
market performances of energy firms. Investors face challenges in H2b. There is no causality from US newspaper coverage of climate
selecting stocks from a vast array of options due to limited rationality change or global warming to S&P 500 oil, gas, and consumable fuels
and cognitive capacity, leading to difficulties in processing extensive industry index returns.
information. To manage this search problem, investors often narrow
down their choices to stocks that capture their attention, potentially 3. Data and methodology
influencing purchase decisions and subsequent price increases [5,50].
Studies by Lou [53] and Bank et al. [4] indicate that factors such as 3.1. Data
advertising and search volume, acting as proxies for investor attention,
are correlated with stock price movements, suggesting that heightened In this study, we examine the causal relationship between climate

3
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

policy uncertainty, climate change media awareness, and the clean and affects the stock prices of energy (see [9,20]).
traditional energy stock indices returns between November 2009 and Table 1 provides the descriptive statistics of each variable, and Fig. 2
December 2021. Monthly data is employed. The beginning of the sample represents the graphical demonstration of each variable. We get the
period is determined by the availability of the S&P Global Clean Stock logarithmic transformation for all the variables to decrease the hetero­
Index data. scedasticity and non-normality. Fig. 2 depicts that only LNMEDIA has a
The natural logarithms of the Climate Policy Uncertainty index, trend.
hereafter LNCPU, is employed to measure climate policy uncertainty.
CPU index constructed by Gavriilidis [32] is a textual-based index that 3.2. Methodology
measures variations in environmental-related government policies. To
develop the Climate Policy Uncertainty (CPU) index, Gavriilidis [32] To ascertain the time-varying Granger causality relations between
adopts the methodology established by Baker et al. [3] for their Eco­ climate policy uncertainty, climate change awareness, and renewable
nomic Policy Uncertainty (EPU) index. This involves scouring articles in and non-renewable energy sector index returns, the procedure intro­
eight prominent US newspapers for terms related to climate policy. duced by Shi et al. [70,71] is employed by considering the time dy­
Subsequently, the number of relevant articles per month in each news­ namics to identify the instability periods in the causality [30]. The
paper is adjusted relative to the total number of articles, standardized lag-augmented VAR (LA-VAR) approach to performing Granger causal­
across the newspapers, and then averaged monthly, with the final series ity tests [26,79] is followed by three time-varying causality approaches:
normalized to have a mean value of 100 for the entire period. This index the forward recursive, rolling window, and recursive evolving algo­
serves as a valuable supplement to climate-risk assessment tools, rithms. The LA-VAR model is following;
capturing significant volatility events pertinent to climate policy [32,
∑
k+d ∑
k+d
64]. Fig. 1 displays the CPU levels throughout the years. As can be seen y1t = γ 10 + γ 11 t + ρ1i y1t− i + δ1i y2t− i + ε1t (1)
from the figure, it depicts peaks around main climate policy-related i=1 i=1
events like President Trump’s declaration to exit the Paris Agreement
in 2017. As in other policy uncertainties, uncertainty in climate policy Where y1 denotes clean energy and nonrenewable energy sector index
can influence companies by decreasing their economic output as they returns; y2 denotes climate policy uncertainty index, climate change
postpone their investments (see [69]). awareness index, and crude oil price. Also, d indicates the maximum
Climate change awareness is proxied by the LNMEDIA, which is the order of integration for yt. The null hypothesis of Granger non-causality
natural logarithms of the US newspaper coverage of climate change or from y2t to y1t is tested by employing Wald Test as follows;
global warming [13]. This measure is constructed by tracking the H0 = δ11 = … = δ1k = 0 (2)
climate change or global warming-related news coverage of the 5 US
newspapers (Washington Post, Wall Street Journal, New York Times, A real-time-varying causality test is employed by using subsamples of
USA Today, and Los Angeles Times) and counting the total number of the data based on supremum Wald statistics. The Wald statistic over [f1,
climate change or global warming-related news in each month. Despite f2] which are f1 and f2 starting and ending points, with a sample size
television’s widespread usage as an information source, studies con­ fraction of fw=f2-f1 ≥ f0 is specified by Wff12 , where T is the total number of
ducted in the United States indicate that newspapers contain a greater observations, and f0T is the minimum number of observations for esti­
number of stories on climate change compared to television (see [1,12, mation of the VAR system. and the supremum of Wald statistic is given
74]). Media coverage is a widely used proxy for public attention and an for the Granger non-causality inference [71];
increase in this variable indicates greater awareness about climate { }
sup
change (see [1,4,58]). By using this measure, we intend to find the SWf (f0 ) = Wf2 (f1 ) (3)
(f1 , f2 ) ∈ ̂0, f2 = f
impact of public climate change awareness on the energy market index
returns.
Return on the non-renewable energy market is proxied by the loga­ where 0̂ = {(f1 , f2 ) : 0 < f0 + f1 ≤ f2 ≤ 1 and 0 ≤ f1 ≤ 1 − f0 },

rithmic first difference of the S&P 500 oil, gas, and consumable fuels Although Shi et al. [71] propose that the recursive evolving window
industry index, hereafter DLNNONRENEWABLE. To measure the clean procedure [61,62] performs better than the forward expanding [78] and
energy market return, hereafter DLNCLEANENERGY, we use the loga­ the rolling window [76] procedures, our study takes into account all
rithmic first difference of the S&P global clean energy index to compute three procedure results based on the below test statistics also suggested
firms’ performances in the businesses associated with clean energy. The by Hammoudeh et al. [40];
data is obtained from the website investing.com. The test statistics are specified in Eq. (5) for the forward expanding
The logarithmic first difference of the WTI crude oil index (DLNWTI), window procedure;
the global benchmark for oil prices, is employed as a control variable.
Previous studies demonstrate that the World oil price significantly

Fig. 1. Time plots of the monthly CPU index. Source: Ren et al. [64].

4
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

Table 1
Descriptive statistics.
Variable Obs. Mean Std. Dev. Min Max Skewness Kurtosis

DLNNONRENEWABLE 146 0.001 0.073 − 0.407 0.259 − 0.784 9.752

DLNCLEANENERGY 145 0.001 0.080 − 0.354 0.187 − 0.788 5.532
LNCPU 146 4.687 0.684 2.466 6.444 − 0.207 3.113
LNMEDIA 146 5.721 0.533 4.836 6.956 0.350 2.006
DLNWTI 146 0.000 0.114 − 0.568 0.546 − 0.906 12.875

Notes: Table 1 displays the descriptive statistics of the variables in use. DLNNONRENEWABLE, DLNCLEANENERGY, LNCPU, LNMEDIA, and DLNWTI refer to the first
difference of the logarithm of S&P 500 oil, gas, and consumable fuels industry index prices, the first difference of the logarithm of S&P global clean energy index prices,
natural logarithms of Climate Policy Uncertainty index, natural logarithms of the media coverage on climate change and global warming, and first difference of the
logarithm of world crude oil prices, respectively.

Fig. 2. Variables. Notes: DLNNONRENEWABLE, DLNCLEANENERGY, LNCPU, LNMEDIA, and DLNWTI refer to the first difference of the logarithm of S&P 500 oil,
gas, and consumable fuels industry index prices, the first difference of the logarithm of S&P global clean energy index prices, natural logarithms of Climate Policy
Uncertainty index, natural logarithms of the media coverage on climate change and global warming, and the first difference of the logarithm of world crude oil prices,
respectively.

5
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

inf { } inf { } Table 3

̂f e = f : Wf (0) > cv and̂f f = f : Wf (0) < cv (4)
f ∈ [f0 , 1] ̂
f ∈ [ f e , 1] Unit-root tests (First differences).
Variable ADF PP
The test statistics are specified in Eq. (6) for the rolling window
procedure; DLNNONRENEWABLE Intercept − 20.612*** − 29.650***
DLNCLEANENERGY − 20.387 − 27.608
inf { } inf { } *** ***
̂f e = f : Wf (f − f0 ) > cv and̂f f = f : Wf (f − f0 ) < cv LNCPU − 19.324*** − 22.947***
f ∈ [f0 , 1] f ∈ [̂f e , 1]
LNMEDIA − 16.671 − 18.129
(5) *** ***
DLNWTI − 14.626 − 18.638***
The test statistics are specified in Eq. (7) for the recursive evolving ***
procedure;
DLNNONRENEWABLE Intercept − 20.539*** − 29.524***
inf { } inf { }
̂f e = f : SWf (f0 ) > scv and̂f f = f : SWf (f0 ) < scv DLNCLEANENERGY and Trend − 20.326*** − 27.556***
f ∈ [f0 , 1] f ∈ [̂f e , 1] LNCPU − 19.293 − 23.035***
***
(6)
LNMEDIA − 16.712*** − 18.356***
DLNWTI − 14.575*** − 18.547***
Where ̂f e and ̂f f are estimated as the first chronological observations
Notes: Table 3 displays the unit-root test results for the first differences.
whose test statistics respectively exceed or fall below the critical values
DLNNONRENEWABLE, DLNCLEANENERGY, LNCPU, LNMEDIA, and DLNWTI
for the origination and termination points of the causal relationship.
refer to the first difference of the logarithm of S&P 500 oil, gas, and consumable
Also, cv and scv are the critical values associated with the Wf and SWf fuels industry index prices, the first difference of the logarithm of S&P global
statistics, respectively [70]. clean energy index prices, natural logarithms of Climate Policy Uncertainty
index, natural logarithms of the media coverage on climate change and global
4. Results warming, and first difference of the logarithm of world crude oil prices,
respectively. ADF and PP indicate Dickey-Fuller and Phillips-Perron, respec­
First, we find the order of integration of each variable by employing tively. Superscripts *** and ** signify significance at 1 % and 5 %, respectively.
augmented Dickey-Fuller (ADF) [24] and Phillips-Perron (PP) unit-root
tests [60]. Unit-root test results are presented in Table 2 and Table 3 for
Table 4
levels and first differences, respectively. As seen from the tables, we can
Wald test results (Dependent variable: LNCLEANENERGY).
reject these tests’ null hypothesis, indicating that the variables are sta­
tionary. Therefore, the d values in the time-varying Granger causality Causality to: DLNCLEANENERGY
analysis are 0. Max Wald FE Max Wald RO Max Wald RE
After finding the level of integration for each variable, the number of Causality from:
lags to be included in the model, denoted by p, is set. To find the opti­ LNCPU 21.243*** 45.623*** 45.623***
mum lag for each model, we employ Akaike Information Criteria (AIC). 9.515 10.495 10.633
AIC suggests using three lags for model 1 and 2 lags for model 2. (12.220) (12.644) (12.921)
16.497 16.984 17.651
Therefore, the time-varying causality is obtained from the VAR model
LNMEDIA 28.454*** 63.708*** 64.704***
with p = 3 for the model in which the causality is to DLNCLEANE­ 6.795 8.130 8.260
NERGY, and p = 2 for the second model, in which the causality is to (9.841) (11.157) (11.864)
DLNNONRENEWABLE. 18.710 18.678 18.710
DLNWTI 19.942** 31.027*** 31.332***
In Tables 4 and 5, the Wald test results for the time-varying Granger
10.224 9.893 10.505
causality test based on the forward recursive (FE), rolling evolving (RO), (12.912) (14.227) (14.742)
20.369 22.683 22.683

Table 2 Notes: Table 4 displays the Wald test results for the analysis in which
Unit-root tests (Levels). DLNCLEANENERGY is the dependent variable. DLNCLEANENERGY, LNCPU,
LNMEDIA, and DLNWTI refer to the first difference of the logarithm of S&P
Variable ADF PP
global clean energy index prices, natural logarithms of Climate Policy Uncer­
DLNNONRENEWABLE Intercept − 12.299*** − 12.315*** tainty index, natural logarithms of the media coverage on climate change and
DLNCLEANENERGY − 10.692*** − 10.806 global warming, and first difference of the logarithm of world crude oil prices,
*** respectively. Lag lengths are determined by Akaike Information Criterion (AIC).
LNCPU − 5.850*** − 5.893***
Superscripts *** and ** signify significance at 1 % and 5 %, respectively.
LNMEDIA − 2.804** − 2.263
DLNWTI − 9.150*** − 8.839***
and recursive evolving (RE) techniques are presented. In Table 4, the
DLNNONRENEWABLE Intercept − 12.267 − 12.284*** results for the entire sample display that we reject the null hypothesis of
and Trend ***
no Granger causality from LNCPU to DLNCLEANENERGY at a 1 percent
DLNCLEANENERGY − 11.052*** − 11.107
*** significant level for all FE, RO, and RE algorithms, which indicates
LNCPU − 7.558*** − 7.727*** climate policy uncertainty Granger causes the return on the clean energy
LNMEDIA − 6.512 *** − 6.322*** stock index for the entire sample. We, therefore, support our hypothesis
DLNWTI − 9.122*** − 8.803***
of no causality from CPU to S&P 500 oil, gas, and consumable fuels
Notes: Table 2 displays the unit-root test results for the levels. DLNNONRE­ industry index returns. This finding is in line with Xi et al. [83] that the
NEWABLE, DLNCLEANENERGY, LNCPU, LNMEDIA, and DLNWTI refer to the CPU has an impact on the overall consumption of renewable energy.
first difference of the logarithm of S&P 500 oil, gas, and consumable fuels in­ Fig. 3 plots the time-varying causality from LNCPU to DLNCLEANE­
dustry index prices, the first difference of the logarithm of S&P global clean NERGY throughout the years. In Fig. 3, while the Wald statistic fails to
energy index prices, natural logarithms of Climate Policy Uncertainty index, exceed its critical value except at the beginning of the sample period for
natural logarithms of the media coverage on climate change and global warm­
FE estimation when RE and RO techniques are employed, we detect that
ing, and first difference of the logarithm of world crude oil prices, respectively.
climate policy uncertainty generally displays significant causal effects
ADF and PP indicate Dickey-Fuller and Phillips-Perron, respectively. Super­
scripts *** and ** signify significance at 1 % and 5 %, respectively. on the clean energy index returns for over all the period of study.

6
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

Table 5
Wald test results (Dependent variable: LNNONRENAWABLE).
Causality to: DLNNONRENEWABLE

Max Wald FE Max Wald RO Max Wald RE

Causality from:
LNCPU 6.803 10.947** 10.947**
7.431 8.863 9.040
(9.070) (10.856) (10.868)
16.899 16.597 16.899
LNMEDIA 13.078** 24.325*** 24.325***
7.421 7.837 8.131
(9.731) (9.964) (10.487)
14.596 13.726 14.596
LNWTI 11.708** 16.786*** 17.807***
6.207 6.588 6.636
(7.388) (8.736) (8.836)
12.107 11.766 12.107

Notes: Table 5 displays the Wald test results for the analysis in which
DLNNONRENEWABLE is the dependent variable. DLNNONRENEWABLE,
LNCPU, LNMEDIA, and LNWTI refer to the first difference of the logarithm of
S&P 500 oil, gas, and consumable fuels industry index prices, natural logarithms
of Climate Policy Uncertainty index, natural logarithms of the media coverage
on climate change and global warming, and first difference of the logarithm of
world crude oil prices, respectively. Lag lengths are determined by Akaike In­
formation Criterion (AIC). Superscripts *** and ** signify significance at 1 %
and 5 %, respectively.

Especially, the Granger curve exhibits a notable increase at the onset of

2017 and again at the beginning of 2018. A thorough examination of the
root causes can be attributed to VW’s admitted guilt to the EPA in 2017,
and President Trump’s declaration of the intention to withdraw from the
Paris Agreement towards the end of the same year. This finding supports
Shang et al. [69], which displays the CPU’s significant impact on
renewable energy demand, and Liang et al. [49], which shows a perfect
prediction impact of the CPU on the volatility of renewable energy.
However, Fig. 3 also indicates that during the breakout of the COVID-19
pandemic, there was no significant causality from the climate policy
uncertainty to the clean energy market.
Table 4 also shows that LNMEDIA Granger causes DLNCLEANE­
NERGY at a 1 percent significant level for the entire sample period.
Unlike Liu and Hamori [52], which present the effect of investor senti­
ment on clean energy stocks as weak, our finding indicates that a change
in public attention to climate change influences the clean energy index
returns. Our findings align with Ouadghiri et al. [58], demonstrating Fig. 3. Time-varying Causality from LNCPU to DLNCLEANENERGY. Notes: (a)
that increased public awareness of environmental concerns positively shows Forward recursive Wald test results for LNCLEANENERGY caused by
LNCPU, (b) shows the Rolling Wald test results for DLNCLEANENERGY caused
impacts the returns of US sustainability stock indices, potentially
by LNCPU, (c) shows the Recursive Expanding Wald test results DLNCLEANE­
attracting a diverse array of investors towards sustainable firms.
NERGY caused by LNCPU. (–) and (..) depict 90th and 95th percentiles of
Furthermore, our results partially corroborate previous research by Lou bootstrapped test statistics, respectively.
[53] and Bank et al. [4], which suggests that variables such as adver­
tising and search volume, serving as proxies for investor attention,
returns. Additionally, Fig. 4 shows the causality increases when the
exhibit correlations with fluctuations in stock prices, implying that
Covid-19 pandemic outbreak at the beginning of 2020, which is in line
heightened investor attention typically coincides with enhanced stock
with Wang et al. [82], which state in April 2020, throughout the
performance. The plot of the time-variant Granger causality from
Covid-19 pandemic, the IEA declared the clean energy sector as the best
LNMEDIA to DLNCLEANENERGY throughout the sample period is dis­
one to survive the risk of the economic lockdowns, which makes clean
played in Fig. 4. Fig. 4a, b, and c depict that for most of the sample
energy sector attractive to policymakers, portfolio managers and in­
period, there is causality from the media coverage of climate change and
vestors. Furthermore, it supports studies that present the share of clean
global warming to the return on the clean energy stock index, regardless
energy increased while the fossil fuels share decreased during the
of the employed algorithms. This suggests that when the media coverage
COVID-19 pandemic [37,80]. Another potential explanation for the
of climate change and global warming increases, people’s attention on
observed uptick in causality during this timeframe could be attributed to
climate increases, making them seek environmentally friendly firms to
President Trump’s rejection of the new emission rule in April 2020 [83].
invest in. Moreover, Fig. 4 displays the significance of causality in­
The results on the causality from DLNWTI to DLNCLEANENERGY
creases after the 2016 Paris Climate Accords, indicating the importance
depict that the crude oil prices Granger cause the clean energy market
of agreement on the clean energy market (see [49]). Moreover, as in the
significantly at 5 percent level when we employ the FE technique, which
impact of CPU on DLNCLEANENERGY, major events that affect the CPU,
contributes to the findings of Bondia et al. [9] that show there is cau­
such as VW’s admitted guilt to the EPA in 2017, and President Trump’s
sality from oil market to the renewable energy stock markets. Fig. 5 plots
declaration of the intention to withdraw from the Paris Agreement, in­
the causality for the sample period, and it shows the causality from the
crease the causality from climate risk awareness to clean energy index
oil prices to the clean energy market decreases throughout time, which

7
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

Fig. 4. Time-varying Causality from LNMEDIA to DLNCLEANENERGY. Notes:

(a) shows Forward recursive Wald test results for LNCLEANENERGY caused by Fig. 5. Time-varying Causality from DLNWTI to DLNCLEANENERGY. Notes:
LNMEDIA, (b) shows the Rolling Wald test results for DLNCLEANENERGY (a) shows Forward recursive Wald test results for LNCLEANENERGY caused by
caused by LNMEDIA, (c) shows the Recursive Expanding Wald test results LNWTI, (b) shows the Rolling Wald test results for DLNCLEANENERGY caused
DLNCLEANENERGY caused by LNMEDIA. (–) and (..) depict 90th and 95th by LNWTI, (c) shows the Recursive Expanding Wald test results LNCLEANE­
percentiles of bootstrapped test statistics, respectively. NERGY caused by LNWTI. (–) and (..) depict 90th and 95th percentiles of
bootstrapped test statistics, respectively.

supports Dawar et al. [20], which demonstrates that diminishing reli­

ance of clean energy stock returns on the returns of crude oil. enables us to ascertain whether significant causality emerges at any
Table 5 presents the time-varying Granger causality test results for juncture within this temporal scope. Therefore, we examine Fig. 6 and
the full sample, and findings show that, unlike the clean energy market, observe that climate policy uncertainty affects the non-renewable en­
we fail to reject the hypothesis of no Granger causality from the climate ergy market significantly during just 2 periods, including the second half
policy uncertainty index to the S&P 500 oil, gas, and consumable fuels of 2013 and the end of 2019. The observed escalation in this phenom­
industry index log return, when we employ the FE technique. For the RO enon could potentially stem from the widespread protests preceding the
and RE algorithms, we find significant causality from climate policy 2019 UN Climate Action summit, as well as the decision by the Trump
uncertainty to the non-renewable energy index returns; however, the administration to nullify the authority of US states to establish vehicle
significance is less than we see for the clean energy market. Therefore, emission standards. There is no causality from LNCPU to LNNONRE­
we support our hypothesis over the entirety of the sample period, no NEWABLE even after the Paris Agreement, which contradicts
discernible causality is observed from the CPU to the returns of the Diaz-Rainey et al. [23], which displays the significant effect of the Paris
nonrenewable energy index. This finding resonates well with Lopez et al. Agreement on the US oil and gas sector.
[86], which state climate policy uncertainty influences the decisions of The same table also shows that for the entire sample period, Wald
firms to reduce their carbon footprint directly, which indicates that if test results indicate a significant causality from LNMEDIA to
investors have fears about an increase in climate change, they convert DLNNONRENEWABLE. This finding suggests that the coverage of media
from brown to green companies, which causes the outperformance of on climate change affects the attention of investors on environmental
green stocks relative to brown ones (see [11,59]). Nonetheless, issues, which influences their investment decisions. Fig. 7 depicts the
employing an innovative time-varying Granger causality analysis time-varying causality from LNMEDIA to DLNNONRENEWABLE during

8
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

Fig. 6. Time-varying Causality from LNCPU to DLNNONRENEWABLE. Notes:

(a) shows Forward recursive Wald test results for LNNONRENEWABLE caused
by LNCPU, (b) shows the Rolling Wald test results for DLNNONRENEWABLE Fig. 7. Time-varying Causality from LNMEDIA to DLNNONRENEWABLE.
caused by LNCPU, (c) shows the Recursive Expanding Wald test results Notes: (a) shows Forward recursive Wald test results for LNNONRENEWABLE
DLNNONRENEWABLE caused by LNCPU. (–) and (..) depict 90th and 95th caused by LNMEDIA, (b) shows the Rolling Wald test results for DLNNONRE­
percentiles of bootstrapped test statistics, respectively. NEWABLE caused by LNMEDIA, (c) shows the Recursive Expanding Wald test
results DLNNONRENEWABLE caused by LNMEDIA. (–) and (..) depict 90th and
95th percentiles of bootstrapped test statistics, respectively.
the sample period. When we examine Fig. 7-c, the Recursive expanding
Wald test results, we observe that the periods with significant causality
2020 and increased the causality from media coverage of climate change
are the year 2013, the year 2018 except the second quarter and the
to the returns on the nonrenewable energy index.
period that covers years 2019 and 2020. These peaks can be related to
Finally, Table 5 presents that we reject the null hypothesis of no
some important events that affect CPU, such as the UN Climate Action
causality from DLNWTI to DLNNONRENEWABLE for the entire sample
Summit in 2019 and Trump’s rejection of the new emission rule in 2020.
period, suggesting causality from crude oil price to S&P 500 oil, gas, and
When we compare Figs. 4 and 7, we can see that the causality from
consumable fuels industry index for the period between November 2009
media coverage of climate change to the returns on the clean energy
and December 2021. Fig. 8 illustrates that the time-varying causality
index is stronger than the causality to the return on the nonrenewable
between oil prices and nonrenewable energy index return is strong and
energy index. This finding is complementary to the findings of Bouri
persists through the sample period. When we investigate Figs. 4 and 8
et al. [11], which show that the CPU has a significant positive effect on
together, we can infer that the impact of oil prices is more substantial on
green stock returns rather than brown stock returns. Fig. 7 also dem­
the nonrenewable energy market than the clean one. This finding is in
onstrates that the causality from LNMEDIA to DLNNONRENEWABLE is
line with Dawar et al. [20], which shows that decreasing dependence of
stronger at the beginning of the COVID-19 pandemic. This finding might
clean energy stock returns on crude oil.
suggest that with the lockdowns, people’s attention to climate change
increases, which affects their trading strategy toward cleaner products.
Moreover, the public attention on the nonrenewable energy sector might
be affected due to Trump’s rejection of the new emission rule in April

9
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

energy market is affected more than in the traditional energy market.

Another critical finding this study reveals is that there is a time-varying
causality from media coverage of climate change to the returns on both
the clean energy and nonrenewable energy sector indices throughout
the sample period. Media coverage of climate change also displays
strong causality with the returns of the global clean energy index,
particularly post-2016 Paris Climate Accords and during the COVID-19
pandemic, and with non-renewable energy returns, especially during
key events like the UN Climate Action Summit and Trump’s emission
rule rejection. These findings imply as people are exposed to climate
change news in the media, their awareness of the importance of climate
change increases, which changes their investment behavior.
Overall, our findings demonstrate the effect of CPU and media
coverage on the investors’ trading strategy for clean and traditional
energy stocks. These findings have numerous essential implications for
firms and investors. For firms, those operating in the clean energy sector
should recognize the influence of climate policy uncertainty on stock
returns. Monitoring the changes in climate policies and anticipating
their potential impacts on market dynamics is crucial for these firms to
ensure resilience and sustainable growth. Second, the significant cau­
sality observed from media coverage of climate change to both sectors’
returns emphasizes the importance of the public image of firms and their
communication strategies. Demonstrating the increase in the impact of
media coverage on the index returns during major events such as envi­
ronmental scandals (e.g., VW’s admission of guilt to the EPA) and policy
announcements (e.g., Trump’s rejection of the new emission rule) sug­
gests that firms should capitalize on such events by aligning their stra­
tegies with emerging trends and grasping growth opportunities.
Implications for investors can be listed as follows. First, investors should
consider incorporating climate policy uncertainty and media coverage
of climate change into their investment decision-making processes.
Second, in an uncertain environment, including both clean and tradi­
tional energy firms’ stocks can help investors to achieve more diversified
portfolios, as the traditional ones are not affected by the CPU as the
clean ones.
Finally, our paper also offers some policy implications. First,
reducing the uncertainties in climate policies is crucial to promote in­
vestment in clean energy because the clean energy sector is affected
more than the traditional energy sector by these uncertainties. Second,
policymakers should prioritize initiatives aimed at promoting environ­
mental awareness and sustainability among the public. Awareness of
Fig. 8. Time-varying Causality from DLNWTI to DLNNONRENEWABLE. Notes: climate change issues can be increased with effective communication
(a) shows Forward recursive Wald test results for DLNNONRENEWABLE caused strategies and educational campaigns, which encourage environmen­
by DLNWTI, (b) shows the Rolling Wald test results for DLNNONRENEWABLE
tally responsible behavior among consumers, investors, and businesses.
caused by LNWTI, (c) shows the Recursive Expanding Wald test results
Our study acknowledges certain limitations that provide avenues for
DLNNONRENEWABLE caused by DLNWTI. (–) and (..) depict 90th and 95th
percentiles of bootstrapped test statistics, respectively. future investigation. To advance research in this domain, it is recom­
mended to incorporate alternative indices as proxies for public atten­
tion. Additionally, expanding the sample size to include indices from
5. Conclusion and policy implications
diverse geographical regions would facilitate a more thorough and
comparative analysis. Furthermore, scholars could enrich the existing
In this study, we aim to investigate the time-varying causality from
literature by employing alternative methodologies, such as Dynamic
climate policy uncertainty and media attention on climate change on the
Conditional Correlation Generalized Autoregressive Conditional Heter­
clean and nonrenewable energy indices returns. To the best of our
oskedasticity (DCC-GARCH), to examine the interrelationships among
knowledge, this paper is the first to employ time-varying Granger cau­
climate policy uncertainty, media coverage of climate change, and
sality analysis to examine the impact of CPU and media coverage on
returns in the energy sector. These proposed avenues for future research
climate change on energy markets. Our findings highlight the impor­
offer opportunities to deepen understanding and enhance the robustness
tance of climate policy uncertainty for the firms’ stock returns in the
of findings in this field.
clean and non-renewable energy sectors.
The findings of this paper reveal significant causality from CPU to the
CRediT authorship contribution statement
clean energy index returns starting from the end of 2009 to the end of
202. Notably, peaks in causality accord with significant events such as
Ecenur Uğurlu-Yıldırım: Formal analysis, Data curation, Concep­
VW’s EPA scandal and Trump’s Paris Agreement withdrawal
tualization. Özge Dinç-Cavlak: Methodology.
announcement. On the contrary, our results indicate negligible causality
from CPU to the non-renewable energy index returns suggesting that in a
Declaration of competing interest
situation in which there is uncertainty about whether the government
will change the existing climate policies or not, investment in the clean
The authors declare that they have no known competing financial

10
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

interests or personal relationships that could have appeared to influence [33] D.E.H.J. Gernaat, H.S. De Boer, V. Daioglou, S.G. Yalew, C. Müller, D.P. Van
Vuuren, Climate change impacts on renewable energy supply, Nat. Clim. Chang. 11
the work reported in this paper.
(2021) 119–125.
[34] A. Greenland, M. Ion, J. Lopresti, Exports, investment and policy uncertainty, Can.
Data availability J. Econ. 52 (3) (2019) 1248–1288.
[35] A. Groß-Klußmann, N. Hautsch, When machines read the news: using automated
text analytics to quantify high frequency news-implied market reactions, J. Empir.
Data will be made available on request. Finance 18 (2011) 321–340.
[36] G. Gutsche, A. Ziegler, Which private investors are willing to pay for sustainable
investments? Empirical evidence from stated choice experiments, J. Bank. Financ.
102 (2019) 193–214.
References [37] S. Halbrügge, P. Schott, M. Weibelzahl, H.U. Buhl, G. Fridgen, M. Schöpf, How did
the German and other European electricity systems react to the COVID-19
[1] L. Antilla, Self-censorship and science: a geographical review of media coverage of pandemic? Appl. Energy 285 (2021) 116370.
climate tipping points, Public Understand. Sci. 19 (2) (2010) 240–256. [38] W.L. Hambira, J. Saarinen, O Moses, Climate change policy in a world of
[2] F. Atsu, S. Adams, Energy consumption, finance, and climate change: does policy uncertainty: changing environment, knowledge, and tourism in Botswana, Afr.
uncertainty matter? Econ. Anal. Policy. 70 (2021) 490–501. Geograph. Rev. 39 (2020) 1–15.
[3] S.R. Baker, N. Bloom, S.J. Davis, Measuring economic policy uncertainty, Q. J. [39] J.T. Hamilton, Pollution as news: media and stock market reactions to the toxics
Econ. 131 (4) (2016) 1593–1636. release inventory data, J. Environ. Econ. Manage 28 (1995) 98–113.
[4] M. Bank, M. Larch, G. Peter, Google search volume and its influence on liquidity [40] S. Hammoudeh, A.N. Ajmi, K. Mokni, Relationship between green bonds and
and returns of German stocks, Financ. Mark. Portf. Mang. 25 (2011) 239–264. financial and environmental variables: a novel time-varying causality, Energy
[5] B. Barber, T. Odean, All that glitters: the effect of attention and news on the buying Econ. 92 (2020) 104941.
behavior of individual and institutional investors, Rev. Financ. Stud. 21 (2008) [41] P.-H. Hsu, K. Li, C.Y. Tsou, The Pollution Premium, Working Paper, 2021.
785–818. [42] IMF, (2017). World economic outlook.
[6] M. Barnett, The Estimated Impacts of Climate Policy Risk, a, Working paper, 2020. [43] S. Kumar, S. Managi, A. Matsuda, Stock prices of clean energy firms, oil and carbon
[7] S.M. Bartram, K. Hou, S. Kim, Real effects of climate policy: financial constraints markets: a vector autoregressive analysis, Energy Econ. 34 (2012) 215e226.
and spillovers, J. Financ. Econ. 143 (2022) 668–696. [44] C. Kousky, S.H. Schneider, Global climate policy: will cities lead the way? Clim.
[8] A. Bondarev, A. Greiner, Environmental Pollution in a Growing Economy With Policy. 3 (2003) 359–372.
Endogenous Structural Change, Bielefeld Working Papers in Economics and [45] S.Y. Lee, R.D. Klassen, Firms’ response to climate change: the interplay of business
Management, 2014. uncertainty and organizational capabilities, Bus. Strat. Environ. 25 (8) (2016)
[9] R. Bondia, S. Ghosh, K. Kanjilal, International crude oil prices and the stock prices 577–592.
of clean energy and technology companies: evidence from non-linear cointegration [46] D. Lemoine, Green expectations: current effects of anticipated carbon pricing, Rev.
tests with unknown structural breaks, Energy 101 (2016) 558e565. Econ. Stat. 99 (2017) 499–513.
[10] B. Born, J. Pfeifer, Policy risk and the business cycle, J. Monet. Econ. 68 (2014) [47] T. Liang, J. Chai, Y.J. Zhang, Z.G. Zhang, Refined analysis and prediction of natural
68–85. gas consumption in China, J. Manag. Sci. Eng. 4 (2) (2019) 91–104.
[11] E. Bouri, N. Iqbal, T. Klein, Climate policy uncertainty and the price dynamics of [48] C. Liang, Y. Wei, X. Li, X. Zhang, Y. Zhang, Uncertainty and crude oil market
green and brown energy stocks, Financ. Res. Lett. 47 (2022) 102740. volatility: new evidence, Appl. Econ. 52 (2020) 2945–2959.
[12] M. Boykoff, J. Boykoff, Climate change and journalistic norms: a case study of US [49] C. Liang, M. Umar, F. Ma, T.L.D. Huynh, Climate policy uncertainty and world
mass-media coverage, Geoforum. 38 (2007) 1190–1204. renewable energy index volatility forecasting, Technol. Forecast. Soc. Change 182
[13] M. Boykoff, M. Daly, McNatt, A Nacu-Schmidt, United States newspaper coverage (2022) 121810.
of climate change or global warming, 2000-2022. Media and climate change [50] F. Liu, Y. Kang, K. Guo, X. Sun, The relationship between air pollution, investor
observatory data sets. Cooperative Institute for Research in Environmental attention and stock prices: evidence from new energy and polluting sectors, Energy
Sciences, University of Colorado, 2022. Policy 156 (2021) 112430.
[14] L. Bretschger, S. Soretz, Stranded Assets: How Policy Uncertainty Affects Capital, [51] R. Liu, H. Lingyun, X. Liang, X. Yang, Y. Xia, Is there any difference in the impact of
Growth, and the Environment, CER-ETH-Center of Economic Research (CER-ETH) economic policy uncertainty on the investment of traditional and renewable energy
at ETH Zurich, 2018. enterprises? -A comparative study based on regulatory effects, J. Clean. Prod. 255
[15] Carney, M. (2015). “Breaking the tragedy of the horizon—Climate change and (2020) 1–13.
financial stability.” Speech delivered at Lloyd’s of London, September 29. [52] T. Liu, S. Hamori, Does investor sentiment affect clean energy stock? Evidence from
[16] H.S. Chan, S. Li, F. Zhang, Firm competitiveness and the European Union emissions TVP-VAR-based connectedness approach, Energies 14 (2021) 3442.
trading scheme, Energy Policy 63 (2013) 1056–1064. [53] Lou, D. (2010). Maximizing short-term stock prices through advertising. SSRN
[17] Chari, V.V. (2018). The role of uncertainty and risk in climate change economics. Electronic Journal. Available at SSRN: https://ssrn.com/abstract=1571947.
Federal Reserve Bank of Minneapolis, No.576. [54] J.F. Mercure, P. Salas, P. Vercoulen, G. Semieniuk, A. Lam, H. Pollitt, P.B. Holden,
[18] X. Chen, Q. Fu, C.P. Chang, What are the shocks of climate change on clean energy N. Vakilifard, U. Chewpreecha, N.R. Edwards, J.E. Vinuales, Reframing - incentives
investment: a diversified exploration, Energy Econ. 95 (2021) 105136. for climate policy action, Nat. Energy 6 (12) (2021) 1133–1143.
[19] M. Child, O. Koskinen, L. Linnanen, C. Breyer, Sustainability guardrails for energy [55] Nagar, V., and Schoenfeld, J. (2021). Measuring weather exposure with annual
scenarios of the global energy transition, Renew. Sustain. Energy Rev. 91 (2018) reports. Tuck School of Business Working Paper no. 3438428. Available at SSRN:
321–334. https://ssrn.com/abstract=3438428.
[20] I. Dawar, A. Dutta, E. Bouri, T. Saeed, Crude oil prices and clean energy stock [56] National Climate Center of China, (2017). China climate index report (Beijing).
indices: lagged and asymmetric effects with quantile regression, Renew. Energy [57] B. Obama, The irreversible momentum of clean energy, Science (1979) 355 (2017)
163 (2021) 288–299. 126–129.
[21] M. Dell, B.F. Jones, B.A. Olken, What do we learn from the weather? The new [58] I.E. Ouadghiri, K. Guesmi, J. Peillex, A. Ziegler, Public attention to environmental
climate–economy literature, J. Econ. Lit. 52 (3) (2014) 740–798. issues and stock market returns, Ecol. Econ. 180 (2021) 106836.
[22] J. Derwall, N. Guenster, R. Bauer, K. Koedijk, The eco-efficiency premium puzzle, [59] L. Pástor, R.F. Stambaugh, L.A. Taylor, Sustainable investing in equilibrium,
Financ. Anal. J. 61 (2005) 51–63. J. Financ. Econ. 142 (2020) 550–571.
[23] J. Derwall, K. Koedijk, J. Ter Horst, A tale of values-driven and profit-seeking [60] P.C.B. Phillips, P. Perron, Testing for a unit root in time series regression,
investors, J. Bank. Finance. 35 (2011) 2137–2147. Biometrika 75 (2) (1988) 335–346.
[24] D.A. Dickey, W.A. Fuller, Distribution of the estimators for autoregressive time [61] P.C. Phillips, S. Shi, J Yu, Testing for multiple bubbles: limit theory of real-time
series with a unit root, J. Am. Stat. Assoc. 74 (1979) 427–431. detectors, Int. Econ. Rev. (Philadelphia) 56 (4) (2015) 1079–1134.
[25] I. Dincer, C Acar, A review on clean energy solutions for better sustainability, Int. J. [62] P.C. Phillips, S. Shi, J Yu, Testing for multiple bubbles: historical episodes of
Energy Res. 39 (5) (2015) 585–606. exuberance and collapse in the S&P 500, Int. Econ. Rev. (Philadelphia) 56 (4)
[26] J.J. Dolado, H. Lütkepohl, Making Wald tests work for cointegrated VAR systems, (2015) 1043–1078.
Econ. Rev. 15 (1996) 369–386. [63] J.C. Reboredo, A. Ugolini, Price connectedness between green bond and financial
[27] J. Fernández-Villaverde, P. Guerrón-Quintana, K. Kuester, J. Rubio-Ramirez,́ Fiscal markets, Econ. Model. 88 (2020) 25–38.
volatility shocks and economic activity, Am. Econ. Rev. 105 (2015) 3352–3384. [64] X. Ren, J. Li, F. He, B. Lucey, Impact of climate policy uncertainty on traditional
[28] K. Fisher-Vanden, K.S. Thorburn, Voluntary corporate environmental initiatives energy and green markets: evidence from time-varying granger tests, Renew.
and shareholder wealth, J. Environ. Econ. Manage 62 (2011) 430–445. Sustain. Energy Rev. 173 (2023) 113058.
[29] S. Fried, K. Novan, W.B. Peterman, The macro effects of climate policy uncertainty. [65] D. Rodrik, Policy uncertainty and private investment in developing countries,
Finance and Economics Discussion Series 2021-018, Board of Governors of the J. Dev. Econ. 36 (1991) 229–242.
Federal Reserve System (U.S., 2021. [66] G.D. Rudebusch, Climate Change Is a Source of Financial Risk," FRBSF Economic
[30] V. Fromentin, Time-varying causality between stock prices and macroeconomic Letter 2021-03, Federal Reserve Bank of San Francisco, San Francisco, 2021.
fundamentals: connection or disconnection? Financ. Res. Lett. 49 (2022) 103073. February.
[31] L. Gan, G.S. Eskeland, H.H. Kolshus, Green electricity market development: lessons [67] S.R. Santos da Silva, M.I. Hejazi, G. Iyer, T.B. Wild, M. Binsted, F. Miralles-
from Europe and the US, Energy Policy 35 (1) (2007) 144–155. Wilhelm, P. Patel, A.C. Snyder, C.R. Vernon, Power sector investment implications
[32] Gavriilidis, K. (2021). Measuring climate policy uncertainty. Available at SSRN. htt of technological forecasting and social change 182 (2022) 121810 climate impacts
ps://ssrn.com/abstract=3847388. on renewable resources in Latin America and the Caribbean, Nat. Commun 12
(2021) 1276.

11
E. Uğurlu-Yıldırım and Ö. Dinç-Cavlak Energy and Climate Change 5 (2024) 100134

[68] D. Schröder, Real options, ambiguity, and dynamic consistency—A technical note, [77] R. Tajaddini, H.F. Gholipour, Economic policy uncertainty, R & D expenditures and
Int. J. Prod. Econ. 229 (2020) 107772. innovation outputs, J. Econ. Stud. 48 (2) (2021) 413–427.
[69] Y. Shang, D. Han, G. Gozgor, M.K. Mahalik, B.K. Sahoo, The impact of climate [78] M.A. Thoma, Subsample instability and asymmetries in money-income causality,
policy uncertainty on renewable and non-renewable energy demand in the United J. Econom. 64 (1–2) (1994) 279–306.
States, Renew. Energy 197 (2022) 654–667. [79] H.Y. Toda, T. Yamamoto, Statistical inference in vector autoregressions with
[70] S. Shi, P.C. Phillips, S. Hurn, Change detection and the causal impact of the yield possibly integrated processes, J. Econom. 66 (1995) 225–250.
curve, J. Time Ser. Anal. 39 (2018) 966–987. [80] D. Wan, R. Xue, M. Linnenluecke, J. Tian, Y. Shan, The impact of investor attention
[71] S. Shi, S. Hurn, P.C. Phillips, Causal change detection in possibly integrated during COVID-19 on investment in clean energy versus fossil fuel firms, Financ.
systems: revisiting the money-income relationship, J. Financ. Econ. 18 (1) (2020) Res. Lett. 43 (2021) 101955.
158–180. [81] Y. Wang, Y. Wei, C. Wu, L Yin, Oil and the short-term predictability of stock return
[72] S.B. Sitkin, L.R. Weingart, Determinants of risky decision-making behavior: a test volatility, J. Empir. Finance 47 (2018) 90–104.
of the mediating role of risk perceptions and propensity, Acad. Manag. J. 38 (6) [82] J. Wang, F. Ma, E. Bouri, J. Zhong, Volatility of clean energy and natural gas,
(1995) 1573–1592. uncertainty indices, and global economic conditions, Energy Econ. 108 (2022)
[73] Y. Song, Q. Ji, Y.J. Du, J.B. Geng, The dynamic dependence of fossil energy, 105904.
investor sentiment and renewable energy stock markets, Energy Econ. 84 (2019) [83] Y. Xi, A.N.Q. Huynh, Y. Jiang, Y. Hong, Energy transition concern: time-varying
104564. effect of climate policy uncertainty on renewables consumption, Technol. Forecast.
[74] K. Stamm, F. Clark, P. Eblacas, Mass communication and public understanding of Soc. Change 192 (2023) 122551.
environmental problems: the case of global warming, Public Understand. Sci. 9 [84] X. Xu, S. Zeng, C. Tam, Stock market’s reaction to disclosure of environmental
(2000) 219–237. violations: evidence from China, J. Bus. Ethics 107 (2012) 227–237.
[75] Y. Sun, Y. Yang, N. Huang, X. Zou, The impacts of climate change risk on financial [85] X. Yang, L.J. Wang, Comparison of regional relationship between environmental
performance of mining industry: evidence from listed companies in China, Resour. performance and financial performance: evidence from Chinese listed companies in
Policy. 69 (2020) 1–9. heavy pollution industries, Hum. Geography 31 (2016) 155–160.
[76] N.R. Swanson, Money and output viewed through a rolling window, J. Monet. [86] J.M.R. Lopez, A. Sakhel, T. Busch, Corporate investments and environmental
Econ. 41 (3) (1998) 455–474. regulation: the role of regulatory uncertainty, regulation-induced uncertainty, and
investment history, Eur. Manag. J. 35 (2017) 91–101.

12