Energy 238 (2022) 122018

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Renewable energy and economic growth: New insight from country

risks
Qiang Wang a, b, *, Zequn Dong a, b, Rongrong Li a, b, Lili Wang a, b
a
School of Economics and Management, China University of Petroleum (East China), Qingdao, 266580, People's Republic of China
b
Institute for Energy Economics and Policy, China University of Petroleum (East China), Qingdao, 266580, People's Republic of China

a r t i c l e i n f o a b s t r a c t

Article history: This study explored the relationship between renewable energy consumption and economic growth
Received 13 March 2021 from new risk-based perspectives, including political risks, financial risks, economic risks and composite
Received in revised form risks. The study uses a panel threshold model to empirically analyze panel data for the Organization for
2 September 2021
Economic Cooperation and Development (OECD) countries from 1997 to 2015. The results show when
Accepted 7 September 2021
Available online 9 September 2021
composite risks and political risks are used as threshold variables, there is a single threshold between
renewable energy consumption and economic growth. When that threshold is exceeded, the positive
effect of renewable energy on economic development increases. When economic risks and financial risks
Keywords:
Country risks
are used as threshold variables, there is a double threshold between renewable energy consumption and
Renewable energy economic growth. When the first threshold value is exceeded, but not the second, renewable energy
Economic development positively impacts economic development. However, when economic risk and financial risk do not lie
Panel data between the two threshold values, there is an insignificant negative correlation between renewable
energy consumption and economic growth.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction studies have found that renewable energy positively impacts eco-
nomic development [8e10]; other studies have found that
Energy is an indispensable source of power for economic renewable energy inhibits it [11e13].
development. Mainstream research has concluded that energy can However, previous studies on the relationship between the two
be divided into two types: renewable and non-renewable. Research have not considered country-level risks. More specifically, few
analyzing the impact of non-renewable energy on economic studies have assessed the impact of renewable energy consumption
development is very comprehensive. Many studies have shown on economic development at different levels of country risks.
that the use of traditional fossil energy can promote economic Country risks refer to the probability of loss or the degree of
growth [1e5]. However, economic growth is not the only goal, and instability faced by investors or traders within a country; these
the use of non-renewable energy has been widely criticized for its risks mainly include political risk, financial risk, and economic risk.
unsustainability and significant carbon emissions. In recent de- First, with respect to political risk, a stable political environment
cades, as climate change has become more significant and there has helps to promote the implementation of renewable energy policies.
been an increased awareness of environmental protection, the In the early stage of a country's development, renewable energy
importance of developing renewable energy has gradually development is often driven by policies, rather than being demand-
emerged; particularly in recent years, significant attention has been driven. A changing political environment may increase the eco-
paid to renewable energy consumption [6]. With the increase in nomic cost of developing renewable energy, which does not sup-
this consumption, more researchers have begun to study renewable port economic development. One study noted that improving
energy [7], focusing particularly the relationship between renew- political stability and the quality of bureaucracy can increase gov-
able energy consumption and economic development. Some ernment sensitivity to the environment and encourage the estab-
lishment of incentives to increase the use of renewable energy [14].
Second, with respect to financial risk, financing is a consistent
* Corresponding author. School of Economics and Management, China University problem facing the renewable energy industry. Renewable energy
of Petroleum (East China), Qingdao, 266580, People's Republic of China. projects require high upfront costs and long capital payback
E-mail address: [email protected] (Q. Wang).

https://doi.org/10.1016/j.energy.2021.122018
0360-5442/© 2021 Elsevier Ltd. All rights reserved.
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

periods [15]. These challenges are coupled with the uncertainty of renewable energy consumption and economic growth. This can
new technologies that require long innovation cycles, and a strong serve as a reference for other countries' renewable energy
dependence on infrastructure; clean energy requires more patience industries.
and high-risk funds [16]. Without a stable financial environment, The rest of this article is organized as follows. The next section
these problems are further magnified. Therefore, a higher level of provides a literature review related to the research topic. Section 3
financial risk will increase the economic cost of renewable energy introduces the data and methods. Section 4 presents the data
consumption, which does not support economic development. One processing, display, and analysis of results. Section 5 presents the
study noted that eliminating financial risk may help release the conclusions and recommendations.
potential of renewable energy, as in a hypothetical “low risk” sce-
nario, the advantages of renewable energy are more clear [17]. This 2. Literature review
scenario shows the highest installed capacity and significantly
reduced total system cost. 2.1. The relationship between renewable energy consumption and
Finally, economic risk mainly emerges from an unstable eco- economic development
nomic environment. An unstable economic environment may
greatly reduce energy demand, especially for renewable energy. In the past ten years, the research on the relationship between
Renewable energy has higher development costs and more infra- renewable energy consumption and economic growth has attracted
structure construction demands compared to non-renewable en- widespread attention. Similar to broader research on energy con-
ergy. This leads to a significant increase in the economic cost of sumption and economic development, most studies focus on
renewable energy development and consumption. A stable eco- analyzing the causal relationship between the two. Empirical
nomic environment provides for better technology, capital, and studies have proposed four testable hypotheses in the context of
talent conditions, encouraging reductions in the economic cost of the renewable energy-growth relationship: growth hypothesis,
renewable energy consumption. conservative hypothesis, feedback hypothesis, and neutrality hy-
In summary, political, financial, and economic country risks pothesis. First, the growth hypothesis explains the unidirectional
directly or indirectly affect renewable energy and economic causal relationship between renewable energy consumption and
development, and therefore play an important role in the rela- economic growth. This shows that an increase in renewable energy
tionship between renewable energy consumption and economic consumption will promote economic growth. Second, the conser-
growth. This study explores the impact of different levels of country vative assumption explains the one-way relationship from eco-
risks on the relationship between the two, in addition to consid- nomic growth to renewable energy consumption, which indicates
ering the role of composite factors. The analysis above and the that economic growth determines energy use. Further, when there
literature review that follows reveal the following research gaps. is a two-way causal relationship between renewable energy con-
(1) Few studies have used empirical analysis to determine whether sumption and economic growth, the feedback hypothesis is valid.
country risks impact on relationship between renewable energy This hypothesis explains that renewable energy affect economic
consumption and economic development. (2) Previous studies growth; the reverse is also true. Fourth, the neutrality assumption
have reached conflicting conclusions concerning the linear rela- means that there is no causational link between renewable energy
tionship between renewable energy consumption and economic consumption and economic growth.
development. This highlights the need to consider the non-linear All four hypotheses have been confirmed by empirical literature
impact of country risks on the relationship between renewable using different samples. Inglesi found renewable energy con-
energy consumption and economic growth. sumption positively impacts economic growth. Encouraging
The innovations and contributions of this article are as follows. renewable energy benefits the environment and the economic
First, to better explain the impact of country risk on the relationship development of individual countries [18]. Lau & Lu used panel data
between renewable energy consumption and economic growth, we of 29 OECD countries from 1990 to 2013 to analyze the impact of
apply multifaceted risk indicators (composite risk, political risk, renewable energy and non-renewable energy consumption on
financial risk, economic risk) to explore the impact of country risks economic growth. Renewable energy consumption was found to
on this relationship. Second, to focus on previously conflicting positively impact economic growth, showing the validity of the
conclusions about the linear relationship between renewable en- growth hypothesis [19]. This is also consistent with the findings of
ergy consumption and economic development, this study applies a Ben and Ben [20].
non-linear approach. The study summarizes country risk channels Bulut and Muratoglu used data from 1990 to 2015, and applied
that affect renewable energy consumption and economic devel- co-integration and causality tests to study the relationship between
opment. Based on this, we use a threshold regression model to renewable energy consumption and GDP in Turkey. They found that
measure the nonlinear impact of renewable energy consumption the causal relationship between renewable energy consumption
on economic development under different risk levels. and GDP is insignificant. This may be because the proportion of
We select the Organization for Economic Cooperation and renewable energy is too small to have a significant impact on GDP
Development (OECD) countries as the sample because their [21]. Ocal and Aslan also studied Turkey's renewable energy and
appropriate characteristics for the research topic. Renewable en- economic development, but reached different conclusions. That
ergy consumption has become an important driving force for the article applied the Toda-Yamamoto causality test and ARDL
economic development of OECD countries. Fossil fuel use in OECD methods to conduct the research; the results verify that the pro-
countries is gradually decreasing, and the proportion of renewable tection hypothesis applies to Turkey's renewable energy con-
energy power is gradually increasing. A 2019 International Energy sumption and economic growth [12].
Agency (IEA) report stated that the proportion of renewable energy Other studies supported the feedback hypothesis [22,23]. Lin
reached 10.8% of the total primary energy supply of the OECD; this and Moubarak found a long-term two-way causal relationship
new high was significantly higher than other organizations. The IEA between renewable energy consumption and economic growth.
also predicts that by 2035, renewable energy will provide one-third This finding shows that China's economic growth supports the
of OECD countries’ total power generation. These countries have a development of the renewable energy industry, helping promote
first-mover advantage in the development and use of renewable economic growth [22]. Rafindadi and Ozturk used Germany's
energy, and have rich experience in developing and coordinating studied quarterly time series data from 1971 to 2013; causality
2
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

analysis revealed the feedback effect between renewable energy 3. Data and methodology
consumption and economic growth [23].
The literature discussed above indicates that research on the 3.1. Variable description
linear relationship between renewable energy consumption and
economic development has achieved fruitful results. However, The sample data for this study includes balanced panel data
some scholars have found that renewable energy consumption has from 1997 to 2015 for 34 OECD countries. The explanatory variable
a non-linear impact on economic development. For example, Tugcu is renewable energy consumption (renewable energy consumption
and Topcu used a nonlinear autoregressive distributed lag model, as a percentage of energy consumption); the explained variable is
and found a nonlinear cointegration relationship between energy economic development (real per capita GDP, in 2010 constant U.S.
consumption and economic growth, for all energy sources dollars) [43]; and the controlled variables include trade opening
(renewable/non-renewable energy) [24]. Luqman and Ahmad also (the ratio of total import and export to GDP) and the level of ur-
applied a nonlinear autoregressive distribution lag model and used banization (ratio of urban population to total population). To study
Pakistan's annual data from 1990 to 2016 to show that renewable the potential different effects of renewable energy consumption on
energy consumption has an asymmetric positive impact on eco- economic development at different country risk levels, we use 4
nomic growth [25]. different risk indexes as threshold variables: the composite risk
index, economic risk index, political risk index, and financial risk
index.
2.2. The impact of country risks on renewable energy consumption The country risk index was created from data in The Interna-
and economic development tional Country Risk Guide (ICRG). ICRG is considered an authori-
tative risk rating agency, with many studies using its data to
When studying renewable energy consumption, some scholars measure country risk levels [44e46]. The guide predicts and ana-
have found that traditional research methods do not consider po- lyzes political, financial, and economic risk for 140 countries and
litical risk factors, which can lead to biased model estimates. regions. The rating results are considered, “the standard that other
Brunnschweiler reported that renewable energy projects, like other ratings can refer to.” Of the guide's risks, the political risk index
types of investment projects, benefit from overall political stability, includes 12 components, with a total score range of 0e100. The
a sound regulatory framework, effective governance, and secure economic risk and financial risk indexes include 5 components
property rights [26]. This is similar to research done by Wu and (details on the components are in the Appendix), with a total score
Broadstock [27]. In addition, studies [28e30] have found that cor- range of 0e50. The composite risk score is generated by dividing
ruption and political instability will reduce the rigidity of envi- the total score of the three indexes (political risk, financial risk, and
ronmental policies and do not support renewable energy economic risk) by two, with a total score range of 0e100. The
investments. Unfair competition caused by corruption may also be composite risk is selected because it is an indispensable part of
associated with unfair and illegal treatment of foreign investment ICRG's national risk evaluation system. This risk is used to measure
companies, which may lead investors to hesitate when making the overall national risk, to avoid the disadvantaged of only
investment-related decisions. Therefore, controlling corruption has studying a single factor, and to provide a more extensive evaluation.
a positive impact on renewable energy consumption. A lower risk index score is associated with a higher risk level; a
Financial risk and economic risk also affect the relationship higher risk score is associated with a lower risk level. Table 1 shows
between renewable energy consumption and economic develop- the symbols, definitions, and data sources of the variables used in
ment. First, both types of risk may directly affect economic devel- the study.
opment. Many studies have shown that the inability to obtain
financing leads to economic gaps and poverty traps [31e33], and 3.2. Methodology
the expansion of financial risk can further increase the difficulty of
financing, which does not support economic development. Ibrahim This study applies advanced econometric methods to study the
and Alagidede conducted empirical research based on sample data nonlinear impact of country risk on the relationship between
from sub-Saharan Africa, reporting a nonlinear relationship be- renewable energy consumption and economic growth. The method
tween finance and economic development [34]. Financial devel- includes the following steps. 1) LLC [47], Fisher-ADF [48], and
opment and economic growth are significantly positively Fisher-PP [49] are applied to test the stationarity of the selected
correlated; however, below a certain estimated threshold, finance variables. 2) If these variables are confirmed to be non-stationary,
is not sensitive to economic growth. The influencing factors of the Kao panel cointegration test [50] method is applied to test
economic risk, such as inflation, can directly affect economic whether there is a cointegration relationship. 3) After confirming
development [34]. Many studies [35e37] have shown that mod- the presence or absence of a co-integration relationship between
erate and stable inflation makes it easier for companies to make variables, the panel threshold model is used to confirm the
investment decisions, promoting economic development. In addi- threshold value and model regression.
tion, financial risk and economic risk can indirectly affect economic
development by affecting renewable energy consumption. For
example, when the financial development system continues to 3.2.1. Panel unit root tests
improve, financial risk is reduced, and it becomes more convenient To ensure the reliability of the panel threshold effect regression
to obtain credit. This not only facilitates the ability to obtain in- and prevent spurious regression, we first apply the panel unit root
vestment opportunities and encourages entrepreneurship [38,39], test to test whether the data set is a stationary series, before the
but also increases investments in renewable energy and promotes empirical research. The LLC test is based on the following equation:
the implementation of green technologies [15,40]. During eco-
X
kl
nomic downturns, conventional economic activities are damaged, Dduv ¼ au Yuv1 þ bug DYuvL þ cgu dgv þ euv ，q ¼ 1; 2; 3 (1)
unemployment generally increases, and per capita income declines. L¼1
These events do not support the development of capital-intensive
renewable energy projects and adversely affects renewable en- where au ; cgu ; dgv ; euv represent the autoregression coefficients of
ergy consumption [41,42]. the model; and the corresponding vectors of the regression
3
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

Table 1
Symbols, definitions and data sources of research variables.

Symbol variable Definition data

source

GDP Economic development GDP per capita (constant 2010 US$) WDI
database
RE Renewable energy Renewable energy consumption (% of total final energy consumption) WDI
consumption database
COM Composite risk Composite assessment of political risk, financial risk and economic risk ICRG
POL Political risk the score of government's stability, social economy, internal and external conflicts, corruption, religious conflicts, and ICRG
bureaucratic quality and other components.
FIN Financial risk the score of foreign debt, current account, international liquidity, exchange rate and other components. ICRG
ECO Economic risk the score of GDP, inflation, and national budget and other components. ICRG
OPEN Trade openness Trade (% of GDP) WDI
database
CO2 carbon emission CO2 emissions (metric tons per capita) WDI
database
UL Urbanization level Urban population (% of total population) WDI
database

parameters are q ¼ 1,2,3. The null hypothesis of this test is au ¼ 0; nonlinear equation when studying the nonlinear relationship be-
when P ¼ 0, the variable has a unit root, and if the null hypothesis is tween the independent and dependent variable. Instead, the
rejected, the variable is stationary. number of thresholds and threshold values are determined by
The Fisher-ADF unit root test is shown in Equation (2): sample data. This approach avoids errors caused by artificially
dividing samples, and the difference in regression coefficients can
X
p
be compared after endogenous grouping, according to the
Fisher  ADF ¼ 2 logðGmÞ/P (2)
threshold value division interval.
m
The general process of threshold regression estimation is as
The Fisher-PP unit root test is shown in Equation (3): follows: randomly select any one of the threshold variables as the
threshold value; divide the data into two intervals; and then use
1 X
K
the OLS method to estimate the parameter values of these two
Choi  ADF ¼ pffiffiffiffiffiffiffiffiffiffiffi g1 ðGmÞ/Kð0; 1Þ (3)
Tm1 m1 intervals. The next steps are to calculate the total residual of the two
intervals’ sum of squares; record the residual sum of squares; select
In these expressions, m, g1 represent the reciprocal of the different threshold values; and record the residual sum of squares
normal distribution function; and Gm represents the P value of the corresponding to the threshold value. This operation is repeated
ADF unit root test. The null hypothesis is that ai ¼ 0 has a unit root; continuously, comparing the residual sum of squares. The threshold
if ai <0, there is no unit root. value corresponding to the smallest residual sum of squares is
theoretically the optimal threshold value estimate. Based on this,
3.2.2. Panel cointegration test the presence and authenticity of the threshold value must be tested
A time series analysis generally requires that the time series be using hypothesis testing, and the threshold effect of this model can
stable; if the time series is not stable, it can lead to a pseudo be determined after the test is passed.
regression problem. However, the time series in a real economy is The specific steps are as follows:
usually non-stationary. The time series can be differentiated to First, a panel threshold model is established to confirm the non-
make it stable, but this can lead a loss in the total amount of long- linear impact of country risks on the relationship between
term information, which is needed to analyze the problem. renewable energy and economic development. The basic model is
Therefore, cointegration is used to address this problem. If the as follows:
cointegration test is passed, the combination of a set of non-
stationary series has a stable equilibrium, and the original equa- lnYit ¼ alnC þ lnXit þ mi þ εit (4)
tion can be directly regressed based on the original data. With this
Based on formula (4), it is first assumed that there is a single
outcome, the regression result at this time is accurate, and there is
threshold effect to establish a single threshold model (5); this is
no pseudo regression problem. Because the time series of panel
then extended to a double threshold model (6).
data is short, this study applies the Kao panel cointegration test as
the co-integration test method.
lnYit ¼ alnC þ b1 lnXit $ Iðqit  gÞ þ b2 lnXit $ Iðqit ＞gÞ þ mi þ εit
3.2.3. Panel threshold regression model (5)
There are many structural mutation problems in economic ac-
tivities, such as the nonlinear relationship between financial con- lnYit ¼ alnC þ b1 lnXit $ Iðqit  g1 Þ þ b2 lnXit $ Iðg1 ＜qit  g2 Þ
straints and investment decisions. The typical way to address these þ b3 lnXit $ Iðqit ＞g2 Þ þ mi þ εit (6)
problems is to add the quadratic term of the explanatory variable,
add dummy variables and interaction terms, or artificially divide In the basic equation, Yit is the explained variable; C is the
the group into groups for regression. Taking these actions, however, controlled variable; a is the coefficient of the controlled variable;
causes the explanatory variable to be highly collinear with the Xit is the core explanatory variable; qit is the threshold variable; g is
quadratic term. Dividing group boundaries also leads to biased the threshold value; and Ið*Þ is the indicator function. Next, the
regression results. To study this structural mutation phenomenon, dummy variable is set as IitðbÞ ¼ fqit  gg, when qit  g, I ¼ 1;
Hansen proposed a non-dynamic panel threshold model [51]. The otherwise, I ¼ 0. The variables b1 ; b2 ; b3 are the influence co-
advantage of this model is that the user does not need to provide a efficients of the explanatory variables on the explained variables
4
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

when the threshold variables are in different threshold intervals.

þ b2 lnREit $ Iðqit ＞g1 Þ þ mi þ εit (7)
The variable mi represents constant terms; and εit is random error
terms.
b (g) and the lnGDPit ¼ a1 lnOPEN þ a2 lnUL þ a3 lnCO2 þ b1 lnREit
For a given threshold g, the estimated value b of b
corresponding residual sum of squares can be obtained after esti- $ Iðqit  g1 Þ þ b2 lnREit $ Iðg1 ＜qit  g2 Þ þ b3 lnREit $ Iðqit ＞g2 Þ þ mi
mating the following model: þ εit
(8)
e ðgÞ0 b
S n ð gÞ ¼ b e ðgÞ
In these expressions, GDPit is the explained variable, repre-
The variable g b corresponds to the smallest residual sum of senting the economic development of country i in year t; REit is the
squares; Sn ðgÞ is the optimal threshold. After determining the explanatory variable, representing the renewable energy con-
estimated threshold value, the corresponding parameter values of sumption of country i in year t; and OPEN; UL; CO2 are the control
the model are determined. After determining the parameter values, variables, representing trade opening, urbanization, and carbon
the threshold effect is further tested to assess the significance of the emissions, respectively. The variable qit is the threshold variable;
threshold effect and the authenticity of the threshold estimated g1 ; g2 represents the threshold value of different levels; a1 ; a2 ; a3 is
value. the coefficient of the control variable; and b1 ; b2 ; b3 represents the
To test the significance of the threshold effect, the hypothesis of coefficient of the core explanatory variable in different intervals.
the model test is H0 ∶ b1 ¼ b2 ; H1 ∶b1 ¼ b, and the following LM The variable Ið*Þ is the indicator function; mi is the constant term;
statistics are constructed to test the null hypothesis: and εit is the random error term.

b
S0  S1 g 4. Empirical findings
F1 ðgÞ ¼
c
s2
4.1. Panel unit root test
In this expression, S0 and S1 g b are the null hypothesis (no
threshold effect) and the residual sum of squares under the con-
This article applies three unit root test methods: LLC, Fisher-
c2 is the variance of
dition of the threshold effect, respectively; and s ADF, and Fisher-PP. The panel unit root test is used to test
the threshold regression residuals. Because the threshold value is whether the variable is stable [52]. Table 2 shows the test results,
not identifiable under the null hypothesis, F1 does not obey the and indicates that when only the LLC method is used, the overall
standard asymptotic distribution, and the critical value cannot be risk and political risk are stable at level. The other variables are all
obtained. non-stationary series at level and cannot be directly regressed.
The Hansen test involves applying the self-sampling method However, after the first-order difference, the results obtained are
(Bootstrap) to obtain the first-order asymptotic distribution, and very significant, indicating that the null hypothesis has been
then generate the P value. If the P value is less than the significance rejected. After the first difference, each group of variables has no
level, the null hypothesis can be rejected, and the threshold effect is unit root. The results of the four test methods are consistent.
significant at the significance level. In contrast, if the P value is Therefore, the selected variables can be considered to be stable
greater than the significance level, it support that the threshold after the first-order difference.
effect is not significant at this level.
Then, the authenticity of the threshold estimate is tested, and 4.2. Panel cointegration test
the corresponding null hypothesis is: H0 ∶ g b ¼ b0 . The specific
statistics are as follows: The Kao panel cointegration test is a reliable method of testing
whether variables have a long-term stable coordination relation-
b
S1 g  S1 g ship [53]. It is applied in this study to assess the long-term rela-
LRg ¼
c
s2 tionship between economic development, renewable energy
consumption, national risks, trade openness, urbanization and
S1 g is the unconstrained residual sum of squares. At the a level of carbon emissions. Table 3 shows the results of the panel cointe-
significance, when: gration test, which includes all variables. The figure shows that the
null hypothesis is rejected; in other words, in 34 OECD countries, a
h pffiffiffiffiffiffiffiffiffiffiffiffi i long-term cointegration relationship has been established between
LRg  C a ¼  2ln 1  1  a
all variables since 1997. This indicates that the regression residuals
The null hypothesis is accepted under the 95% confidence level, of the equation are stable, and the regression results are accurate.
with C a ¼ 7.35. After the first true threshold is obtained, to deter- The next step is to estimate the panel threshold model.
mine whether there are double thresholds or other thresholds, the
existence of two or more thresholds can be tested in turn until the 4.3. Threshold effect test
null hypothesis cannot be rejected. The specific thresholds are then
determined. According to one study's non-dynamic threshold Before estimating the specific threshold of the model, it is
regression model [51], and based on the assumption that there is a necessary to ensure that the sample data has a threshold effect. The
threshold effect, this study constructs the following threshold threshold model is used to measure the specific threshold. The
regression model to examine the single threshold effect and the statistical package Stata16.0 is used to perform a threshold effect
double threshold effect of renewable energy consumption on eco- test on the sample data (results in Table 4; Table 5). The per capita
nomic development under different country risk levels. The specific gross national product (LNGDP) indicates economic development;
models are presented in model (7) and model (8): and composite risk, political risk, economic risk, and financial risk
are used as threshold variables for empirical testing. Stata16.0
lnGDPit ¼ a1 lnOPEN þ a2 lnUL þ a3 lnCO2 þ b1 lnREit $ Iðqit  g1 Þ generated specific F statistics and P values through 300 repeated
Bootstrap sampling events (bootstrap method) to determine the
5
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

Table 2
Panel unit root test results.

Variable Test method At level At 1st difference Order of integration

t-statistic Prob. t-statistic Prob.

LNGDP LLC 14.4442 1.0000 12.6702*** 0.0000 I (1)

Fisher-ADF 3.11177 1.0000 212.208*** 0.0000 I (1)
Fisher-PP 2.12549 1.0000 217.276*** 0.0000 I (1)
LNRE LLC 4.80711 1.0000 15.4496*** 0.0000 I (1)
Fisher-ADF 24.7123 1.0000 349.851*** 0.0000 I (1)
Fisher-PP 26.3544 1.0000 766.351*** 0.0000 I (1)
LNCOM LLC 2.29770** 0.0108 23.1220*** 0.0000 I (0)
Fisher-ADF 52.8249 0.9122 486.082*** 0.0000 I (1)
Fisher-PP 50.9827 0.9387 802.277*** 0.0000 I (1)
LNPOL LLC 2.80742*** 0.0025 12.3993*** 0.0000 I (0)
Fisher-ADF 61.8681 0.6861 276.647*** 0.0000 I (1)
Fisher-PP 72.4215 0.3343 343.403*** 0.0000 I (1)
LNFIN LLC 0.42100 0.6631 19.0683*** 0.0000 I (1)
Fisher-ADF 67.6613 0.4888 374.730*** 0.0000 I (1)
Fisher-PP 70.1988 0.4038 497.731*** 0.0000 I (1)
LNECO LLC 1.88391 0.9702 21.2724*** 0.0000 I (1)
Fisher-ADF 19.1599 1.0000 410.460*** 0.0000 I (1)
Fisher-PP 16.1839 1.0000 1206.92*** 0.0000 I (1)
LNOPEN LLC 18.0873 1.0000 21.5188*** 0.0000 I (1)
Fisher-ADF 21.3753 1.0000 150.052*** 0.0000 I (1)
Fisher-PP 54.9960 0.8723 102.635*** 0.0042 I (1)
LNUL LLC 5.11864 1.0000 20.7158*** 0.0000 I (1)
Fisher-ADF 11.8874 1.0000 353.165*** 0.0000 I (1)
Fisher-PP 11.3200 1.0000 570.368*** 0.0000 I (1)
LNCO2 LLC 3.2133 1.0000 18.1031*** 0.0000 I (1)
Fisher-ADF 36.4857 0.9994 371.087*** 0.0000 I (1)
Fisher-PP 31.9192 0.9999 702.997*** 0.0000 I (1)

Note: *, **, *** represent significant at 1%, 5%, and 10% inspection levels, respectively. I (0), I (1) respectively represent in levels, first differences stationary of the variables.

threshold value and the number of thresholds for each threshold

Table 3 variable. The F statistic is used to determine whether there is a
Kao panel cointegration tests results. threshold effect. When the F statistic is significant, the impact of
Statistic P-value renewable energy consumption on economic development is
considered to have a threshold effect. First, the existence of
ADF 2.135955** 0.0163
Residual variance 0.002372 threshold is tested; this includes testing whether there is a signif-
HAC variance 0.001884 icant threshold effect when different variables are used as
Note: *, **, *** represent significant at 1%, 5%, and 10% inspection levels,
threshold variables. If there is a single threshold then check the
respectively. double threshold effect, and if there is no single threshold effect, no
threshold regression analysis is required. The resulting threshold
existence results are shown in Table 4.

Table 4
Threshold effect test results 1: Existence Test results.

Threshold variable Threshold test F P-value 1% critical value 5% critical value 10% critical value

LNCOM Single threshold effect 32.930*** 0.007 29.172 18.015 11.528

Double threshold effect 9.292 0.120 27.383 13.692 10.305
LNPOL Single threshold effect 23.516** 0.020 29.188 15.811 11.283
Double threshold effect 8.115 0.127 25.850 11.495 9.027
LNECO Single threshold effect 15.581* 0.060 30.841 16.743 11.667
Double threshold effect 11.930** 0.010 12.255 3.620 0.258
LNFIN Single threshold effect 11.835* 0.083 27.904 17.699 10.648
Double threshold effect 13.273** 0.043 18.425 11.911 9.154

Note: *, **, *** represent significant at 1%, 5%, and 10% inspection levels, respectively.

Table 5
Threshold effect test results 2: Authenticity Test.

Threshold variable Threshold number Threshold value 95% confidence interval

LNCOM Single threshold 4.109 [ 4.109, 4.117 ]

LNPOL Single threshold 4.205 [ 4.204, 4.205 ]
LNECO Single threshold 3.656 [ 3.302, 3.670 ]
Double threshold 3.702 [ 3.418, 3.768 ]
LNFIN Single threshold 3.480 [ 3.401, 3.670 ]
Double threshold 3.552 [ 3.546, 3.560 ]

6
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

The test results show that when renewable energy consumption estimated coefficient of the impact of renewable energy con-
is used as the core explanatory variable and composite risk is used sumption on economic development is 0.0863. This indicates that,
as the threshold variable, the single threshold F statistic of LNCOM when the country's overall risk is stable, the role of renewable
is 32.930, which is greater than the 1% significance level critical energy consumption in promoting economic development in-
value of 29.172. The single-threshold test is significant at the 1% creases. In other words, the impact of renewable energy con-
level; that is, it passes the single-threshold test with a threshold sumption on economic development is affected by the overall
value of 4.109. The double-threshold F statistic is 9.292, which is stability of the country. Countries with lower overall risk have a
less than the 10% significance level critical value of 10.305. This more stable national environment, where more renewable energy
means the double-threshold effect test failed. Similarly, when po- consumption will promote economic development. The impact of
litical risk is used as a threshold variable, the LNPOL single renewable energy consumption on economic development is not
threshold F statistic is 23.516, which is greater than the critical significant when the overall risk of the country is high. This may be
value of 15.811 at the 5% significance level. This means it passes the due to the fact that when country risk is high, economic and social
single threshold test with a threshold value of 4.205. The double- instability causes little willingness for industrial production, and
threshold F statistic is 8.115, which is less than the critical value the control of risk may impose additional costs and regulatory
of 9.027 at the 10% significance level. As such, it does not pass the burdens on the government, the higher the economic cost required
double-threshold test. to develop renewable energy. As the risk decreases, business ex-
When economic risk is used as a threshold variable, the LNECO pectations of the aftermath increase and national policies become
single threshold F statistic is 15.581, which is greater than the more willing to shift to the green economy, the renewable energy
critical value of 11.677 at the 10% significance level; that is, it passes sector will attract an inflow of capital and technology, which will
the single threshold test. The double threshold F statistic is 11.930, lead to a further reduction in the cost of renewable energy and
which is greater than the critical value of 3.620 at a 5% significance significantly increase the support of renewable energy for eco-
level. As such, it passes the double threshold test, the two thresh- nomic development.
olds are 3.656 and 3.702, respectively. Similarly, when financial risk Second, when political risk (LNPOL) is used as the threshold
is used as a threshold variable, it also passes the double threshold variable, the regression results show that the impact of renewable
test with thresholds of 3.656 and 3.702, respectively. energy consumption on economic development is consistently
positive during the sample period. However, the positive impact is
4.4. Analysis of threshold effect regression results greater when the political risk is higher than the threshold. When
the political risk is lower than the threshold of 4.205, a 1% increase
After obtaining the threshold value of each variable, we further in renewable energy consumption promotes economic growth by
analyze the threshold effect of renewable energy consumption on 0.0204%. When the economic risk exceeds the threshold, a 1% in-
economic development using the risk threshold interval of crease in renewable energy leads to a 0.0892% increase in economic
different countries. Table 6 shows the specific regression results. development. This shows that improving the political risk situation
The first step in the analysis is to consider the parameter esti- stimulates an increase in renewable energy consumption, reducing
mation result of composite risk (LNCOM), which reflects the overall the economic cost of renewable energy development. This view is
risk setting using country risks as the threshold variable. The consistent with a previous study [54], who argue that a 1%
LNCOM column in Table 6 shows the test result of the single improvement in the political risk profile increases the consumption
threshold effect model. When the composite risk of the sample is of renewable energy by 0.025%. The improvement in political sta-
less than the threshold value of 4.109, the estimated coefficient of bility and institutional quality will eventually increase the use of
renewable energy consumption on economic development is renewable energy, further leading to an increase in economic effi-
0.0115; the overall risk of the country is high. Renewable energy ciency. A stable political environment contributes to the continuity
consumption promotes economic development, but this effect is of environmental protection policies; the policy-driven nature of
not significant. When the threshold value is exceeded, the the renewable energy industry causes this effect to deepen. A more
stable political environment also supports the voice of democracy,
and the government's initiative to support the renewable industry
Table 6 can be enhanced by citizens freely expressing their environmental
Threshold effect regression results. expectations. Corruption, on the other hand, undermines the
Variable Threshold variable establishment of strong laws that prevent projects having a nega-
tive impact on the environment. In corrupt environments, these
LNCOM LNPOL LNECO LNFIN
projects crowd out the renewable energy development market
LNRE(qit <g1) 0.0115 0.0204 0.0193*** 0.0183***
[29]. This increases the economic cost of developing renewable
(0.56) (0.96) (-4.42) (-3.19)
LNRE(qit g1) 0.0863*** 0.0892***
energy. Moreover, corruption will directly threaten economic
(5.45) (5.59) development, by exploiting public finances and reducing market
LNRE(g1<qit g2) 0.106*** 0.105*** efficiency.
(6.47) (6.41) Third, the two thresholds of economic risk (LNECO), 3.656 and
LNRE(qit >g2) 0.0223*** 0.0242***
3.702, are used to divide the sample countries into three levels:
(-4.51) (-4.78)
LNOPEN 0.464*** 0.472*** 0.460*** 0.465*** high risk, medium risk, and low risk. Under different economic risk
(13.12) (13.23) (12.74) (12.99) levels, the impact of renewable energy consumption on economic
LNUL 0.145 0.377** 0.397** 0.315* development is significantly different. Below the first threshold, the
(0.85) (2.25) (2.36) (1.87)
estimated coefficient of renewable energy consumption on eco-
LNCO2 0.300*** 0.333*** 0.356*** 0.351***
(6.63) (7.38) (7.74) (7.73)
nomic development is 0.0193. In other words, when the economic
constant 6.851*** 5.742*** 5.652*** 6.007*** risk is high, renewable energy consumption inhibits economic
(9.53) (8.21) (8.09) (8.57) development. When the economic risk is higher than the first
R2 0.439 0.430 0.434 0.432 threshold, but lower than the second threshold of 3.702, the effect
Note: *, **, *** represent significant at 1%, 5%, and 10% inspection levels, coefficient of renewable energy consumption on economic devel-
respectively. opment is 0.106. This indicates that the economic risk is at a
7
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

medium level. In this scenario, the impact of renewable energy 5. Conclusions and recommendations
consumption on economic development has changed from being
inhibited to being promoted. When the economic risk exceeds the With the increasing severity of global climate change and an
second threshold, the estimated coefficient of the impact of increasing awareness about environmental protection, renewable
renewable energy consumption on economic development energy has the potential to gradually replace traditional energy. An
is 0.0223. This result shows that the consumption of renewable increasing number of studies have focused on the relationship
energy positively impacts economic development only when the between renewable energy consumption and economic develop-
economic risk is moderate. If the economic risk is too high or too ment. However, past studies on the relationship between the two
low, the risk will have a negative impact. First of all, high economic have not considered country-level risk factors. To fill this gap, this
risk does not support investments in renewable energy, because study applied a panel threshold model to analyze the panel data
few investors are willing to invest in capital-intensive industries sets of OECD countries from 1997 to 2015.
when the risks are higher than expected. An unstable economic The results provide evidence of the nonlinear impact of
environment will also cause low social productivity. These do not renewable energy consumption on economic development under
support economic development. When risks are too low, invest- different country risks (composite risk, political risk, financial risk,
ment opportunities will decrease in an overly stable economic and economic risk). The impact of renewable energy consumption
environment, and the potential for economic development is on economic development differs within different threshold in-
limited. Moderate economic risk generally occurs during the rising tervals of different country risks. In general, renewable energy
period of economic development, which provides a relatively stable consumption promotes economic development. According to the
economic environment, and fully releases economic vitality. This composite risk indicator, measuring overall country risks, the
helps maximize the role of renewable energy consumption in impact of renewable energy consumption on economic develop-
promoting economic development. ment is affected by the country's composite risk. Countries with a
Finally, when financial risk (LNFIN) is used as the threshold lower composite risk have a more stable environment, and
variable, the parameter estimation results show that the sample is renewable energy consumption has a greater promotional effect on
also divided into three levels. When the financial risk is below the economic development.
first threshold, the consumption of renewable energy has a Similarly, a stable political environment helps renewable energy
restraining effect on economic development. When the financial consumption play a larger role in promoting economic develop-
risk exceeds the first threshold and is lower than the second ment. In countries with higher political risk, every 1% increase in
threshold, the impact of renewable energy consumption on eco- renewable energy consumption promotes economic growth by
nomic development is positive; for every 1% increase in renewable 0.0204%. In countries with low political risk, a 1% increase in
energy consumption, economic development increases by 0.105%. renewable energy leads to a 0.0892% increase in economic devel-
When the financial risk exceeds the second threshold, the impact opment. In contrast with the single threshold effect associated with
coefficient of renewable energy consumption on economic devel- the first two risks, both financial risk and economic risk pass the
opment becomes negative again, with an elasticity coefficient dual threshold test. The two thresholds divide the risk levels into
of 0.0242. Similar to economic risk, moderate financial risk helps high risk, medium risk, and low risk. The empirical results of the
ensure moderate financial stability and makes the financial system panel threshold model show that when economic and financial risk
more flexible. In this scenario, renewable energy consumption can thresholds are used, the impact of renewable energy consumption
promote economic development; otherwise, it has a negative ef- on economic growth exhibits an inverted U shape: when the risk is
fect. In summary, when the level of risk is at different levels, the low or high, renewable energy has a negative impact on economic
countermeasures needed are different. When the financial risk is growth; when the risk is moderate, renewable energy has a positive
high, measures should be taken to reduce the financial risk. When effect on economic growth.
the financial risk is low, corresponding policies should be adjusted; This conclusion indicates that the positive impact of renewable
the rigidity of policies should be changed; and the flexibility of risk energy consumption on economic growth is conditional. To play an
control measures should be improved. When the financial risk is important role of renewable energy consumption in the economic
between the first threshold and the second threshold, the financial growth, economic and financial risks must be held at a medium-
risk is a medium level. This facilitates the positive impact of risk level. The specific countermeasure is that policymakers
renewable energy consumption on economic growth. The intensity should formulate policies consistent with the development stage of
of risk control policies can be maintained or minor adjustments can national risks based on ICRG economic risk and financial risk data,
be made. Finance, whether it is financial stability or financial and fully consider the complexity of the impact of economic risks
development, is critical to developing renewable energy. This is and financial risks on the relationship between renewable energy
because the financial market plays a vital role in the financing of and economic growth. Risk control measures need to be both soft
renewable energy projects. As such, the quality of financial devel- and hard to avoid results that conflict with the policy starting point,
opment has a profound impact on the renewable energy industry; a due to the adoption of a single policy. It is also important to
stable financial system is a prerequisite for renewable energy in- appropriately adjust risk control measures to control risk indicators
vestment. The development of renewable energy is inseparable within the medium risk threshold.
from financial support. Renewable energy projects have high The conclusions above lead to the following recommendations.
upfront capital costs and require a significant initial investment to Our results show that in general, renewable energy consumption
start production [55]. In terms of continuous innovation and positively impacts economic development. It is important that the
research investment, continuous capital support is also needed to government plan an energy transition strategy, because developing
enable clean energy technology for long-term development and renewable energy under appropriate conditions can improve
substantive breakthroughs. This contributes to national economic environmental conditions and benefit the macro economy. In
development. addition, country risks play an important role in the relationship

8
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

between renewable energy and economic development, and the Author contribution statement
direction and extent of the impact of renewable energy on eco-
nomic development depends on the level of different types of Qiang Wang: Conceptualization, Methodology, Software, Data
country risks, so policy makers should consider country risk factors curation, Writing e original draft, Supervision, Writing- Reviewing
when developing renewable energy policies. It is also important to and Editing. Zequn Dong: Methodology, Software, Data curation,
improve composite risks and the political risk index because Investigation Writing- Original draft, Writing- Reviewing and
countries with more stable social order and low levels of corruption Editing. Rongrong Li: Conceptualization, Methodology, Software,
are more likely to attract investments by high-quality multinational Data curation, Writing e original draft. Lili Wang: Methodology,
companies. The overflow effect of foreign capital can advance Investigation Writing- Original draft, and Editing.
technical progress and upgrade industrial structures, helping to
increase renewable energy consumption and encourage green
Declaration of competing interest
economic growth. Therefore, it is necessary to establish a long-term
corruption prevention and monitoring mechanism, improve gov-
The authors declare that they have no known competing
ernment credibility and ensure policy implementation, which is
financial interests or personal relationships that could have
crucial for the development of the new energy industry. Further-
appeared to influence the work reported in this paper.
more, the government should focus on stimulating market vitality,
while maintaining the smooth operation of the economy and
financial stability. When the risk is too high, the policy authorities Acknowledgement
should adjust corresponding currency and fiscal policies according
to the situation, control the inflation rate, and prevent harmful The authors would like to thank the editor and these anony-
levels of inflation. It is also important to improve the construction mous reviewers for their helpful and constructive comments that
of financial market systems, advance the financial marketization greatly contributed to improving the final version of the manu-
process, and improve the efficiency of financial market operations. script. This work is supported by National Natural Science Foun-
When the risk is too low, the government should adopt cuts in the dation of China (Grant No. 71874203).
reserve requirement ratio (RRR) or interest rates to fully release
liquidity, improve the quality of foreign debt, and encourage banks
Appendix. Risk components and scores of ICRG risk rating
to develop green credit and protect the responsible financing needs
of renewable energy companies.

Country risks and its components Points（max.）

(I) Political risk 100

Government Stability 12
Socioeconomic Conditions 12
Investment Profile 12
Internal Conflict 12
External Conflict 12
Corruption 6
Military in Politics 6
Religious Tensions 6
Law and Order 6
Ethnic Tensions 6
Democratic Accountability 6
Bureaucracy Quality 4
(II) Financial risk 50
Foreign Debt as a Percentage of GDP 10
Exchange Rate Stability 10
Foreign Debt Service as a Percentage of Exports of Goods and Services 10
Current Account as a Percentage of Exports of Goods and Services 15
Net International Liquidity as Months of Import Cover 5
(III) Economic risk 50
GDP per Head 5
Real GDP growth 10
Annual Inflation Rate GDP 10
Budget Balance as a Percentage of GDP 10
Current Account as a Percentage of GDP 15

9
Q. Wang, Z. Dong, R. Li et al. Energy 238 (2022) 122018

References developing and transition economies. CER-ETH Economics working paper

series 2009;15(3):241e74.
[27] Wu L, Broadstock DC. Does economic, financial and institutional development
[1] Ivanovski K, Hailemariam A, Smyth R. The effect of renewable and non-
matter for renewable energy consumption? Evidence from emerging econo-
renewable energy consumption on economic growth: non-parametric evi-
mies. Int J Econ Pol Emerg Econ 2015;8(1):20.
dence. J Clean Prod 2020.
[28] Fredriksson PG, Svensson J. Political instability, corruption and policy for-
[2] Mohammadi H, Parvaresh S. Energy consumption and output: evidence from a
mation: the case of environmental policy 2003;87(7e8):1383e405.
panel of 14 oil-exporting countries. Energy Econ 2014;41(jan):41e6.
[29] Junxia L. Investments in the energy sector of Central Asia: corruption risk and
[3] Lee CC. Energy consumption and GDP in developing countries: a cointegrated
policy implications. Energy Pol 2019;133:110912.
panel analysis. Energy Econ 2005;27(3):415e27.
[30] Uzar U. Is income inequality a driver for renewable energy consumption?
[4] Narayan PK, Smyth R. Energy consumption and real GDP in G7 countries: new
J Clean Prod 2020;255:120287.
evidence from panel cointegration with structural breaks. Energy Econ
[31] Oded G, Joseph Z. Income distribution and macroeconomics. Rev Econ Stud
2008;30(5):2331e41.
1993;(1):35e52.
[5] Zafar MW, Shahbaz M, Hou F, Sinha A. From nonrenewable to renewable
[32] Beck T, Demirgü-Kunt A, Levine R Finance. Inequality and the poor. J Econ
energy and its impact on economic growth: the role of research & develop-
Growth 2007;12(1):27e49.
ment expenditures in asia-pacific economic cooperation countries. J Clean
[33] Sehrawat M, Giri AK. The impact of financial development, economic growth,
Prod 2018.
income inequality on poverty: evidence from India. Empir Econ 2018;55.
[6] Wang Q, Li S, Pisarenko Z. Heterogeneous effects of energy efficiency, oil price,
[34] Ibrahim M, Alagidede P. Nonlinearities in financial developmenteeconomic
environmental pressure, R&D investment, and policy on renewable energy –
growth nexus: evidence from sub-Saharan Africa. Res Int Bus Finance
evidence from the G20 countries. Energy 2020;209:118322.
2018;46:95e104.
[7] Wang Q, Guo J, Dong Z. The positive impact of official development assistance € Bayrak M. Inflation and economic growth: a dynamic panel
[35] Aydın C, Esen O,
(ODA) on renewable energy development: evidence from 34 Sub-Saharan
threshold analysis for Turkish republics in transition process. Procedia - Social
Africa Countries. Sustainable Production and Consumption 2021;28:532e42.
and Behavioral Sciences 2016;229:196e205.
[8] Ohler A, Fetters I. The causal relationship between renewable electricity
[36] Boukraine W. Short and long-run determinants of inflation in Tunisia. MPRA
generation and GDP growth: a study of energy sources. Energy Econ
Paper; 2020.
2014;43(may):125e39.
[37] Jalil A, Tariq R, Bibi N. Fiscal deficit and inflation: new evidences from Pakistan
[9] Fang Y. Economic welfare impacts from renewable energy consumption: the
using a bounds testing approach. Econ Modell 2014;37(feb):120e6.
China experience. Renew Sustain Energy Rev 2011;15(9):5120e8.
[38] Klapper L, Laeven L, Rajan R. Entry regulation as a barrier to entrepreneurship.
[10] Wang Q, Wang L. Renewable energy consumption and economic growth in
J Financ Econ 2006;82(3):591e629.
OECD countries: a nonlinear panel data analysis. Energy 2020;207:118200.
[39] Munemo J. Foreign direct investment and business start-up in developing
[11] Marques AC, Fuinhas JA. Is renewable energy effective in promoting growth?
countries: the role of financial market development. Q Rev Econ Finance
Energy Pol 2012;46:434e42.
2017;65:97e106.
[12] Ocal O, Aslan A. Renewable energy consumptioneeconomic growth nexus in
[40] Anton SG, Afloarei Nucu AE. The effect of financial development on renewable
Turkey. Renew Sustain Energy Rev 2013;28:494e9.
energy consumption. A panel data approach. Renew Energy 2020;147:330e8.
[13] Bhattacharya M, Paramati SR, Ozturk I, Bhattacharya S. The effect of renewable
[41] Chang T-H, Huang C-M, Lee M-C. Threshold effect of the economic growth rate
energy consumption on economic growth: evidence from top 38 countries.
on the renewable energy development from a change in energy price: evi-
Appl Energy 2016;162:733e41.
dence from OECD countries. Energy Pol 2009;37(12):5796e802.
[14] Uzar U. Political economy of renewable energy: does institutional quality
[42] Salim RA, Rafiq S. Why do some emerging economies proactively accelerate
make a difference in renewable energy consumption? Renew Energy
the adoption of renewable energy? Energy Econ 2012;34(4):1051e7.
2020;155:591e603.
[43] Wang Q, Su M, Li R, Ponce P. The effects of energy prices, urbanization and
[15] Kim J, Park K. Financial development and deployment of renewable energy
economic growth on energy consumption per capita in 186 countries. J Clean
technologies. Energy Econ 2016;59:238e50.
Prod 2019;225:1017e32.
[16] Zhang F. Leaders and followers in finance mobilization for renewable energy
[44] Shareef R, Hoti S. Small island tourism economies and country risk ratings.
in Germany and China. Environmental Innovation and Societal Transitions
Math Comput Simulat 2005;68(5):553e66.
2020;37:203e24.
[45] Aziz OG. Institutional quality and FDI inflows in Arab economies. Finance Res
[17] Sweerts B, Longa FD, van der Zwaan B. Financial de-risking to unlock Africa's
Lett 2018;25:111e23.
renewable energy potential. Renew Sustain Energy Rev 2019;102:75e82.
[46] Chiu Y-B, Lee C-C. Effects of financial development on energy consumption:
[18] Inglesi-Lotz R. The impact of renewable energy consumption to economic
the role of country risks. Energy Econ 2020;90:104833.
growth: a panel data application. Energy Econ 2016;53:58e63.
[47] Levin A, Lin C-F, James Chu C-S. Unit root tests in panel data: asymptotic and
[19] Gozgor G, Lau CKM, Lu Z. Energy consumption and economic growth: new
finite-sample properties. J Econom 2002;108(1):1e24.
evidence from the OECD countries. Energy 2018;153:27e34.
[48] Maddala GS, Wu S. A comparative study of unit root tests with panel data and
[20] Ben Jebli M, Ben Youssef S. Output, renewable and non-renewable energy
a new simple test. Oxf Bull Econ Stat 1999;61(S1):631e52.
consumption and international trade: evidence from a panel of 69 countries.
[49] Ppc B, Pierre P. Testing for a unit root in time series regression. Biometrika
Renew Energy 2015;83:799e808.
1988;(2):335e46.
[21] Bulut U, Muratoglu G. Renewable energy in Turkey: great potential, low but
[50] Kao C. Spurious regression and residual-based tests for cointegration in panel
increasing utilization, and an empirical analysis on renewable energy-growth
data. J Econom 1999;90(1):1e44.
nexus. Energy Pol 2018;123:240e50.
[51] Hansen BE. Threshold effects in non-dynamic panels: estimation, testing, and
[22] Lin B, Moubarak M. Renewable energy consumption e economic growth
inference. J Econom 1999;93(2):345e68.
nexus for China. Renew Sustain Energy Rev 2014;40:111e7.
[52] Li R, Wang Q, Liu Y, Jiang R. Per-capita carbon emissions in 147 countries: the
[23] Rafindadi AA, Ozturk I. Impacts of renewable energy consumption on the
effect of economic, energy, social, and trade structural changes. Sustainable
German economic growth: evidence from combined cointegration test.
Production and Consumption 2021;27:1149e64.
Renew Sustain Energy Rev 2017;75:1130e41.
[53] Wang Q, Zhang F. The effects of trade openness on decoupling carbon emis-
[24] Tugcu CT, Topcu M. Total, renewable and non-renewable energy consumption
sions from economic growth e evidence from 182 countries. J Clean Prod
and economic growth: revisiting the issue with an asymmetric point of view.
2021;279:123838.
Energy 2018;152:64e74.
[54] Su C-W, Umar M, Khan Z. Does fiscal decentralization and eco-innovation
[25] Luqman M, Ahmad N, Bakhsh K. Nuclear energy, renewable energy and
promote renewable energy consumption? Analyzing the role of political
economic growth in Pakistan: evidence from non-linear autoregressive
risk. Sci Total Environ 2021;751:142220.
distributed lag model. Renew Energy 2019;139:1299e309.
[55] Patel S. Climate finance: engaging the private sector. World Bank; 2010.
[26] Brunnschweiler CN. Finance for renewable energy: an empirical analysis of

10