1 s2.0 S0960148122006061 Main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Renewable Energy 193 (2022) 434e447

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Climate change caused by renewable and non-renewable energy


consumption and economic growth: A time series ARDL analysis for
Turkey
lu a, *, Mustafa Güllü b
Hakan Acarog
a
Department of Economics, Faculty of Economics and Administrative Sciences, Eskisehir Osmangazi University, 26480, Eskisehir, Turkey
b
Republic of Turkey Ministry of National Education, 06830, Ankara, Turkey

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

Article history: This paper focuses on climate change in Turkey caused by energy consumption using an Autoregressive
Received 8 October 2021 Distributed Lag and Toda-Yamamoto causality analysis. The motivation and aim are: Finding evidence of
Received in revised form causality for the relationship between energy consumption, growing economies and climate change
24 March 2022
depending on parameters that vary over time, which are observed and argued through political impli-
Accepted 27 April 2022
Available online 4 May 2022
cations. Temperature and precipitation are the dependent variables for climate change; energy types and
Gross Domestic Product per capita are the independent variables for economic determiners. Data was
collected annually from various institutions between 1980 and 2019. According to the Toda-Yamamoto
Keywords:
Climate change
test, a negative relationship is determined between renewable energy consumption and temperature
Renewable energy consumption in both the short and long term. The results reveal that a 1% increase in renewable energy reduces the
Non-renewable energy consumption temperature by 0.031%. The increase of renewable energy may help in decreasing temperature. Pre-
GDP per Capita cipitation and non-renewable energy consumption have a positive relationship in both the short and
Precipitation long term, with a 1% increase in non-renewable energy consumption causing a 0.175% increase in pre-
Temperature cipitation, indicating a negative effect on climate change. Encouraging renewable energy consumption
through government incentives can be a powerful solution to decrease the negative effects of climate
change in Turkey.
© 2022 Elsevier Ltd. All rights reserved.

1. Introduction emissions keeps in its current vicious-cycle form, the natural re-
sources for developing countries may not suffice within the frame
Renewable and non-renewable energy consumption-climate of sustainable economic growth in the near future for new gener-
scenarios depend on socio-economic factors, energy technologies, ations [4,5]. While, renewable energy use is potentially viable so-
and policy design, that will produce specific emission profiles and lution for Turkey [6,7], it has many advantages over conventional
energy mixes, and this will irreversibly have an effect on global energy including the ecological balance [8,9]. The fact that Turkey is
climate conditions [1]. Over the last few decades, environmental a Mediterranean country and is geographically located within a
degeneration caused by global warming has become one of the humid and temperate climate creates important potential for
major concerns among policymakers and researchers [2]. With the renewable energy. Due to the high current account deficit caused
current behaviour of consumption and production of the fast-paced by foreign dependency in energy, the use of renewable energy
industrialized global world, climate change has a negative impact instead of conventional energy becomes mandatory [10]. Although
on ecosystems, species, and humanity [3]. If the environmental renewable energy systems for energy production are initially costly
deterioration caused by the consumption of conventional energy [11,12], once in place, climate change can be sorted and controlled
use (i.e., fossil fuels or non-renewable energy) through carbon by consuming clean and sustainable energy in both the short and
long run [13]. However, initial barriers disproportionately affect the
developing and less developed countries.
Climate change, defined as a significant temporal change (i.e.,
* Corresponding author..
precipitation and temperature) in the related aspects and patterns
lu), mustafagullu@hotmail.
E-mail addresses: hacaroglu@ogu.edu.tr (H. Acarog
com (M. Güllü). of the global climate system [14], has been one of the most

https://doi.org/10.1016/j.renene.2022.04.138
0960-1481/© 2022 Elsevier Ltd. All rights reserved.
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Nomenclature NREC Non-renewable energy consumption


PP Phillips Perron
PREC Precipitation
Abbreviations Prob Probability
ADF Augmented Dickey Fuller REC Renewable energy consumption
Adj. R-sq Adjusted R square SC Schwarz information criterion
AIC Akaike information criterion St.e. Standard error
ARDL Autoregressive Distributed Lag TEMP Temperature
CUSUM Cumulative sum of recursive residuals t-St T-statistic
CUSUMQ Cumulative sum of squares of recursive residuals VAR Vector Autoregressive
ECM Error correction model WDI World Development Indicators
EIA Energy Information Administration
F-st F statistics Symbols
GDP Gross Domestic Product b; d; r; t; a; Estimated coefficients
HQ Hannan-Quinn s; g; u; 4
IEA International Energy Agency D First difference operator
IAMC Integrated Assessment Modeling Consortium ECTt1 Error-correction term
IRENA International Renewable Energy Agency εit White noise term
IPCC Intergovernmental Panel on Climate Change wi Coefficients of Error-correction
L Natural logarithms

prominent issues over the last decade. Greenhouse gas emissions, The energy transition is required to protect the environment
deforestation, soil erosion, and ocean acidification are some of the during climate change [25]. According to Ref. [26], a significant
factors contributing to climate change [15]. It causes global energy transition is required to achieve the goal of climate-neutral
warming by increasing average temperatures and decreasing energy systems. Renewable energy utilization appears to be an
average precipitation levels all over the world. Turkey is sur- alternative to conventional energy in this concept. Turkey has
rounded by the Mediterranean Sea on its eastern and western valuable renewable energy potential, i.e., solar, wind, geothermal,
borders, where the evidence of global warming is easily observed. biomass, and hydraulic energy. The energy produced and
Long-term downward trends in precipitation and arid conditions consumed is contingent on domestic resources. By 2020, as a result
from the 1970s to the present have been strongly affected by of the development of new technologies, the decrease in costs and
drought in most of the Mediterranean region, particularly in the increase in investments, the global renewable energy genera-
Turkey. The decrease in precipitation and predicted drought con- tion capacity had reached 2799 Giga Watt (GW) with an annual
ditions as a result of global climate change are likely to affect growth of 10.2% and an annual capacity increase of 261 GW [27].
Turkey, which has a semi-arid climate now, and makes it vulnerable The electricity generated from the total renewable energy power
to increased droughts [16]. In the summer, there is a noticeable plants in the world and the development of the total renewable
decrease in precipitation, and rising temperatures are expected as a energy installed capacity over the years are given in Fig. 1. Ac-
result of climate change [17]. Furthermore, the magnitude and cording to Fig. 1, while the total electricity generated from renew-
frequency of heatwaves [18], as well as the daily temperature limits, able energy in the world was 3,725,192 Giga Watt hour (GWh) in
are increasing as a result of projection estimates [19]. A change is 2008, it increased to 6,963,450 GWh in 2019. While the total
expected in winter precipitation [20] and daily temperatures, renewable energy power plant in the world had an installation of
together with a decreasing trend in the number of frosty days [21]. 1,060,668 Mega Watt (MW) in 2008, it increased to 2,802,004 MW
Enhancing heat waves and the fresh water drought, along with in 2020.
economic growth and increase of population, are anticipated to The worldwide increased interest in renewable energy, the
have corollaries for economic activity, energy consumption, and transformation to renewable energy alternatives in Turkey is
human health [22]. More negative involvements contain infrequent increasing day by day. While there were 11,288 MW of renewable
scarcities, poor air nature and quality, diminishing crop efficiency, energy in Turkey in 2000, it increased to 49,368 MW in 2020 [28].
and a raised factor of vegetation and forest fires [23]. According to Turkey's National Renewable Energy Action Plan,
The Intergovernmental Panel on Climate Change (IPCC), the 61,000 MW of renewable energy is expected to generate about 159
Integrated Assessment Model Consortium (IAMC), the International Tera Watt-hour (TWh) of electricity in 2023 [29]. Fig. 2 depicts the
Renewable Energy Agency (IRENA), the International Energy generation of electricity from the renewable energy power plants in
Agency (IEA), the United States Energy Information Administration Turkey, as well as the evolution of total renewable energy installed
(EIA), and energy companies such as ExxonMobil Shell, and BP to power over time.
create scenarios to better understand the effects of climate change According to Fig. 2, total electricity generation from renewable
and produce reports on the future of energy system development energy in Turkey was 44,538 GWh in 2006 and increased to
[1]. According to these projections, the Paris Agreement has 132,264 GWh in 2019. While the total renewable energy in Turkey
decided to change the energy system and shift to a low-carbon in 2006 had an installed power of 13,185 MW, it had an installation
economy, making both climate and energy projections more of 49,368 MW in 2020. This potential and ability can make a rapid
complicated. In this regard, energy-economy models are becoming increase in renewable energy generation and consumption, instead
a crucial role in designing climate policy decisions [24]. Considering of consuming fossil fuels. At the same time, there has been
all these phenomena, quantitative research needs to be conducted remarkable population growth in Turkey, which means there is a
in order to describe the convenient production and consumption growing economy with increasing GDP per capita values. Therefore,
patterns with new challenges. the ecological balance is influenced by the trade-off between

435
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Fig. 1. Electricity generation (GWh) and installed capacity (MW) of total renewable power plants around the world over time.
Source: Adapted from Ref. [27].

Fig. 2. Development of electricity generation (GWh) and installed power (MW) of total renewable power plants in Turkey over time.
Source: Adapted from Ref. [27].

consuming renewable and non-renewable energy. Turkish society parameters on the relationship between energy, economy, and
experiences this transformation through the effects of critical climate in Turkey. Therefore, the link between those terms should
climate change. be statistically examined. It is indicated prior to the analysis how
The visualization below shows the issue with the projected this study defines climate change in terms of parameters. For this
numbers for Turkey: Turkey's installed capacity data by primary purpose, this research gathers the evidence of climate change
sources for July 2021 are given in Fig. 3. According to the installed impact on Turkish economy through energy consumption. Because
capacity data, the highest installed capacity is natural gas with 348 of low-carbon emissions, there has recently been a shift away from
power plants and 25,542.9 MW installed power. According to Fig. 3, the use of fossil fuels and toward the use of renewable energy. This
the capacity by primary sources in Turkey are as follows: Dam and observation may assist in providing an economic policy implication
normal streams total 31,436.3 MW; wind energy 9929.3 MW; solar regarding the decision of which type of energy consumption may
power 7324.1 MW; geothermal 1650.2 MW; biomass 1415.4 MW. be convenient and necessary in the near future. This study tries to
With this data, it is understood that more than half of the installed understand the above-mentioned link excluding the carbon-
capacity is from conventional energy in Turkey. If these ratios were emission issues by conducting time series econometric tech-
to be reversed, in other words, if the installed capacity from niques (i.e., Autoregressive Distributed Lag (ARDL) Model).
renewable energy resources increases in the following years, there The literature on the relationship between carbon emissions,
is a possibility that climate change and environmental issues may energy consumption and economic growth is presented and dis-
be affected positively in Turkey. This can be achieved through cussed systematically in the literature section. Unlike previous
sustainable economic growth. But the relationship between the research, this study focuses on global warming, which is a critical
above-mentioned issues needs to be clarified, and it begins with indicator of climate change and is linked to the use of renewable
some analytical reasoning questions and after interrogating those energy. Because there are many studies in the literature that
relationships with causality analysis. examine the subject in terms of carbon emissions, examining the
Based on the above-mentioned energy overview of Turkey, this issue in terms of global warming will add to the literature from a
study aims to find causality evidence based on time-varying different perspective. According to Ref. [31], renewable energy

436
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Fig. 3. Installed capacity by primary sources in Turkey (MW) (July 2021).


Source: Adapted from Ref. [30].

consumption has a negative impact on CO2 emissions in the Middle and non-renewable energy consumption (see for instance
East and the North African region between 1980 and 2015. From Refs. [14,33e36], or separately i.e., renewable energy or non-
1975 to 2018, Ref. [32] investigated the level of environmental renewable energy consumption and environmental factors such
degradation in Pakistan caused by CO2 emissions, and they as emissions (see for instance Refs. [37e41]) in the concept of panel
discovered that CO2 emissions from urbanization and energy data (i.e., panel ARDL; panel cointegration) and time series tech-
consumption were a significant threat to the environment. How- niques (i.e., dynamic ARDL; non-linear ARDL; ARDL and Granger
ever, this study's approach differs from the literature of environ- causality), by using one of those methodologies. While Ref. [39]
mental Kuznets curve studies in that it uses climate change indicates that the number of studies on the relationship between
variables as dependent variables in the models rather than carbon climate change and renewable energy is increasing with the
emission variables. availability of more data on the subject and the development of the
The climate change variable is determined by the temperature methodology, Ref. [37] mentions that the use of fossil fuels in en-
and the precipitation. The temperature and precipitation data are ergy production causes global warming, and the way to overcome
obtained from the Turkish State Meteorological Service; renewable this problem is through the increase of renewable energy projects,
energy and non-renewable energy consumption data are obtained and according to Ref. [38], rising in environmental sustainability
from the IEA, and GDP per capita data are obtained from World will result in an increase in renewable energy consumption and it is
Development Indicators (WDI) for the period 1980e2019. The found to be both in the long and short-term relationship between
dependent variable that stays on the left side of Eq. (1) is the CC economic growth, environmental sustainability, and renewable
variable that represents climate change with the factors of annual energy consumption. However, Ref. [34] also included oil price, real
average precipitation and annual average temperature. The depen- GDP, CO2 emissions variables; Ref. [33] included trade, inflation,
dent variables used for the model include average precipitation in and terrorism variables in the analysis, Ref. [36] included CO2 and
depth (PREC) and annual average temperature (TEMP), as repre- GDP; unlike these variables, Ref. [35] included income, domestic
sentations for climate change. On the other hand, the independent and foreign direct investment, institutional quality, physical infra-
variables that stay on the right side of Eq. (1) are REC; NREC; and GDP, structure and urbanization variables in the analysis. They specif-
respectively. The REC variable represents the renewable energy ically used real GDP per capita, consumer price index, and financial
consumption. The NREC variable represents the non-renewable development variables. In Ref. [38], per-capita renewable energy
energy consumption. The GDP variable represents the real Gross consumption and environmental sustainability variables are used
Domestic Product per capita. The long and the short-term rela- in the analysis. In this study, precipitation and temperature are
tionship are examined with the ARDL bounds test, and the causality taken as dependent variables in order to measure climate change,
relationship is examined with the Toda Yamamoto causality test. similar to Refs. [14], (i.e., panel data analysis for a subset of African
countries), and [40] (i.e., statistical analysis for Poland but only used
air temperature for a climate change proxy) but in this case, unlike
CC ¼ ffREC; NREC; GDPg (1) others, a time series and causality methodology for a specific
Energy consumption is represented by both REC and NREC. A country was used. On the other hand, little effort has been made to
negative relationship is determined between REC and TEMP in the directly address climate change and its drivers in both renewable
long-term and short-term. A positive relationship is determined and non-renewable energy consumption scenarios, as well as in
between PREC and NREC in the long-term and short-term. When developing economies. Ref. [40] mentions that, there are many
the causality directions are examined, no causality is found be- research gaps for the impact of climate change on renewable en-
tween TEMP and energy consumption (REC and NREC), but a bi- ergy in specific geographical regions. This study aims to fill a gap in
directional causality relationship is found between NREC and the case of Turkey by employing a time series ARDL methodology
PREC. Furthermore, a unidirectional relationship from PREC to REC based on the Toda-Yamamoto causality tests. Turkey was chosen
has been determined. because there has been very little research based on case studies of
Recently, most existing scientific research focus on the re- the country, and Turkey's location is critical because it is located in
lationships between either the combination of renewable energy both Asia and Europe. As a result, Turkey has the potential to serve

437
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

as a model for other countries along this climate change axis.


Furthermore, once this methodology is applied to the Asian and
European regions, this case study can be used as a reference for
both.
The concept of the paper is shown in Fig. 4 which explains both
short-term and long-term and causality relationships between the
indicated variables. In Fig. 4, short-term relationships are repre-
sented by wavy thin lines, while long-term relationships are rep-
resented by wavy thick lines. Positive relationships are represented
by orange, while negative relationships are represented by blue.
Finally, the direction of causality is represented by one-way and
two-way straight and green arrows.
The novelties presented in this study are summarized as
follows:

 This work predicts a time series ARDL model throughout the


Toda-Yamamoto causality analysis for the relationship between
climate change-renewable and non-renewable energy
consumption-GDP per capita in a case study for Turkey.
 Precipitation and temperature variables are separately consid-
ered in the econometric models as the determiners of climate
change.
Fig. 4. The paper's thesis statement. Note: The concept was created by the authors.
 A 1% increase in renewable energy consumption reduces tem-
perature by 0.031% which can be accepted as a positive effect.
 A 1% increase in non-renewable energy consumption increases problems [14]. However, in the literature, Ref. [52] finds that energy
precipitation by 0.175% which can be accepted as a negative consumption has a detrimental effect on the environment in most
effect. Arctic countries. Contrary to general energy consumption, there are
 As the research paper's novel argument, energy consumption opinions that renewable energy is beneficial for climate change.
through renewable energy consumption can help to reduce According to Ref. [46] renewable electricity generation in Turkey
global warming, and the findings of this case study can serve as will contribute to environmental improvement and play a signifi-
a scientific guideline for policymakers. cant role in reducing GHG emissions. Ref. [54] also evaluates
renewable energy consumption as a way to reduce CO2 emissions
The rest of the paper is structured as follows: The methodology in Europe and North America. Ref. [14], who examines the re-
is given through the literature; the conceptual framework; data and lationships between renewable and non-renewable energy con-
variables; and the econometric model in section 2. Section 3 pre- sumption, economic growth and climate change for 16 African
sents the results and discussion. Finally, conclusions and future countries, determines that renewable energy consumption has a
works are presented in section 4. positive effect on climate change, while non-renewable energy
consumption has a negative effect. Ref. [55]shows that for OECD
2. Methodology countries, non-renewable energy consumption increases CO2
emissions, while renewable energy consumption decreases it.
2.1. Literature Ref. [50] finds that renewable energy has a significant negative
impact on CO2 emissions. According to Ref. [53] renewable energy
There is a vast amount of literature pertaining to the different consumption in European Union countries contributes less to
aspects of the relationship between renewable energy, non- greenhouse gas emissions than fossil energy consumption. Some
renewable energy, Gross Domestic Product (GDP), and climate studies in the literature show that renewable energy has a negative
change. Some of these studies have dealt with the relations as a effect on the climate. Ref. [51] investigates the effects of agriculture,
single country such as: Refs. [42e51], and some of them have dealt renewable energy production, and globalization on CO2 emissions
with a country group such as: Refs. [14, 52e55], for panel data in Turkey and concludes that agriculture, renewable energy pro-
studies or comparison of countries. These studies are grouped into duction, and economic globalization all contribute to environ-
three sub-groups in terms of their approaches to the subject. In the mental pollution. Ref. [48], who investigates the relationships
first group, studies that compare climate change with energy between CO2 emissions, GDP, renewable and non-renewable en-
(renewable and non-renewable) consumption are discussed. In the ergy consumption for Tunisia, shows that non-renewable energy
second group, studies that examine the relationships between GDP has a positive effect on CO2 emissions, while renewable energy has
and climate change are investigated. In the third group, studies that a weak and negative effect. The findings of this study support the
deal with the subject within the scope of the environmental Kuz- aforementioned ones (i.e., REC positively affects climate change)
nets curve are included. The condensed literature review can be but the approach in this study is different from the previous liter-
found in Table A1 in Appendix A. Table A1 is designed by author/ ature by considering climate change as a dependent variable for an
year, data/period, applied method(s), conducted models, and ARDL time series analysis. Furthermore, because there has been no
findings. Moreover, the abbreviations for the common terminology study on climate change in terms of temperature and precipitation
are given at the bottom of Table A1. (i.e., accepting those variables as dependent in the model) in
Turkey, this article fills a gap in the literature. Specifically, the
2.1.1. Renewable - non-renewable energy consumption and climate estimation of the vector autoregressive (VAR) model for the cau-
change sality estimations and the related findings will provide the privi-
Renewable energy is one of the best options for dealing with lege, and allow for a better understanding of the mentioned
climate change, which is one of the humanity's most pressing relationships from a single location: Turkey.
438
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

2.1.2. GDP and climate change independent variables. The flow chart of the applied methodology
There are efforts in the developed world to reduce CO2 emis- with the indicated dependent and independent variables is given
sions by adopting renewable energy policies and reducing reliance with Fig. 5 as follows:
on coal and gas [54]. The ways of accessing energy also affect
economic growth in different ways. In the literature, this effect has
2.3. Data and variables
been discussed with different variables. Ref. [42] investigates the
relationship between electricity consumption and prices, industrial
The data was collected annually for the period between 1980
value-added, urbanization, population growth, CO2 emissions, and
and 2019. The PREC and TEMP data are obtained from the Turkish
GDP in Pakistan. According to the results of the study, it has been
State Meteorological Service; REC and NREC data are obtained from
determined that electricity consumption has a positive and signif-
the IEA, and GDP data are obtained from World Development In-
icant effect on economic growth. There is also an inconsistency in
dicators (WDI). PREC represents Turkey's total precipitation in
the literature on the relationship between renewable energy and
millimeters (mm); TEMP represents the maximum temperature of
economic growth. Ref. [47] discovers a unidirectional causality
Turkey in Centigrade degrees (oC); REC and NREC represent Tur-
relationship between renewable energy consumption and eco-
key's net electricity consumption of renewable and non-renewable
nomic growth while researching the relationships between
energy in billion-kilowatt-hours (b. kWh), and GDP represents real
renewable energy consumption, international trade, oil price, and
Gross Domestic Product per capita in constant 2010 United States
economic growth for Tunisia. Similarly, Ref. [44] investigates the
Dollars (US$). Variables were included in the analysis by taking
relationship between renewable electricity generation and eco-
their natural logarithms (L). The effect of REC and NREC on climate
nomic growth (i.e., running from renewable electricity generation
change has been estimated by ARDL and Toda Yamamoto Causality
to economic growth), in the United States and China. However, one
methods. Descriptive statistics for TEMP, PREC, REC, NREC, and GDP
of the studies examining the relationship between renewable en-
variables for the period 1980e2019 are given in Table 1 as follows:
ergy consumption and economic growth in Turkey, Ref. [49], con-
According to descriptive statistics, the highest standard devia-
cludes that renewable energy consumption has a negative effect on
tion is in the LNREC variable, and the lowest standard deviation is in
economic growth. This study differs from previous ones in that it
focuses directly on climate change rather than environmental fac-
tors, such as harmful emissions or ecological footprint, and uses
energy consumption and GDP per capita as explanatory variables
for the short and long terms of causality relationships.

2.1.3. Environmental Kuznets curve (EKC) and climate change


Examining the EKC hypothesis, which is an inverted U-shaped
relationship between income per capita and some types of pollu-
tion, such as SO2 and CO2 emissions, in environmental economics
is widely seen in the literature [52]. There is a close relationship
between carbon emissions and climate change. There are studies in
the literature in which the EKC hypothesis is valid. Ref. [46], tests
the validity of the EKC hypothesis in Turkey, showing that there is a
U-shaped EKC relationship between greenhouse gas and income.
Likewise, Ref. [50] investigates the EKC for India and found evi-
dence of an inverted U-shaped EKC with a turnaround point of
2937.77 USD. There are also studies in the literature that question
the validity of the EKC hypothesis. Ref. [52], who investigates the
EKC hypothesis for Arctic countries, finds little evidence for the
existence of the EKC hypothesis. There are studies in the literature
that conclude that the EKC hypothesis is not valid in some studies.
Investigating the existence of the EKC hypothesis in Vietnam,
Ref. [43] concludes that the EKC hypothesis did not exist. Ref. [53]
investigates the EKC hypothesis on 16 European Union countries by
separating the final amount of energy consumption as renewable
and fossil fuel energy consumption, concluding that the EKC hy-
pothesis is not valid. Lastly, in his study for Tunisia, Ref. [48] con-
cludes that the inverted U-shaped EKC hypothesis does not support
the long-run relationship graphically and analytically. This study,
however, differs from the others in that it focuses on climate change
in a specific growing economy using TEMP and PREC data rather
than emissions. As a result, it is directly based on climate change
throughout its determiners, which is correlated with global
warming.

2.2. Conceptual framework

This study examines the effects of REC and NREC on climate


change in Turkey. As climate change determiners (i.e., PREC and
TEMP data) are taken as dependent variables; as economic de-
terminers, REC, NREC, and GDP per capita data are taken as Fig. 5. Methodology flow chart.

439
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

the LTEMP variable. The annual maximum average TEMP for Turkey The standard log-linear functional characteristic of the long-run
was 21.1  C in 2019, and the highest average annual precipitation relationship between TEMP, PREC, and REC, NREC, GDP per capita
was 793.8 mm in 2009. can be represented by the ARDL bounds approach as follows:

2.4. Econometric model DLTEMPit ¼ d0i þ d1i LTEMPi;t1 þ d2i LRECi;t1 þ d3i LNRECi;t1
X
p

In this study, two-time series models are used for the following þ d4i LGDPi;t1 þ rij DLTEMPi;tj
j¼1
variables. Temperature (TEMP) and precipitation (PREC) are used as
dependent variables to measure climate change, while renewable X
q1 X
q2
energy consumption (REC), non-renewable energy consumption þ b1ij DLRECi;tj þ b2ij DLNRECi;tj
(NREC) and Gross Domestic Product (GDP) per capita are inde- j¼0 j¼1

pendent variables. Following Refs. [14,56], two implicit forms of X


q3
equations for causality and cointegration relations between the þ b3ij DLGDPi;tj þ mi þ εit
variables are as follow in Eq. (2) and Eq. (3): j¼1

(6)
LTEMPit ¼ f(LRECit, LNRECit, LGDPit) (2)
DLPRECit ¼ d0i þ d1i LPRECi;t1 þ d2i LRECi;t1 þ d3i LNRECi;t1
LPRECit ¼ f(LRECit, LNRECit, LGDPit) (3)
X
p

Expressing the parameters, then the equation form is con-


þ d4i LGDPi;t1 þ rij DLPRECi;tj
j¼1
structed as follows: In Eq. (4) and Eq. (5) while b0 indicates the
constant term; b1, b2, and b3 represent the coefficients of the in- X
q1 X
q2

dependent variables. þ b1ij DLRECi;tj þ b2ij DLNRECi;tj


j¼0 j¼1

LTEMPit ¼ b0 þ b1 LRECit þ b2 LNRECit þ b3 LGDPit þ ε (4) X


q3
þ b3ij DLGDPi;tj þ mi þ εit
j¼1
LPRECit ¼ b0 þ b1 LRECit þ b2 LNRECit þ b3 LGDPit þ ε (5)
(7)
The concept of cointegration is a method that makes regressions
involving I(1) variables potentially meaningful and presents the where i ¼ 1, 2, ….,N and t ¼ 1,2, …, T. In the above-mentioned
main problems and methods used in applications as a complete equations, Eq. (6) and Eq. (7), εit denotes the white noise term
cointegration treatment mathematically. Assuming {yt: t¼0,1, …} and D is the first difference operator. In addition, long-term re-
and {xt: t¼0,1, …} are two I(1) processes, yt - bxt will generally be an lationships are represented by dki ðk ¼ 1; 2; 3; 4Þ , while short-term
I(1) process for any number b. However, for some b ¼ 0 it is possible summation symbols represent the error correction term.
for yt - bxt to be an I(0) process and meaning that it has a constant The F-statistic (Wald test) is used in the ARDL bounds test
mean, constant variance, and is asymptotically uncorrelated. If cointegration method to detect the presence of long-term cointe-
there is a cointegration parameter b with these properties, y and x gration between variables. Ref. [61] determines two limits as lower
are said to be cointegrated. If y and x are not cointegrated, it in- and upper limits based on the F-test statistics. If the estimated
dicates that the regression of y on x is spurious, that there is no value of the F-test statistic is less than the lower limit, it is inter-
significant relationship, and that there is no long-run relationship preted as there is no long-term relationship between the variables.
between y and x [57]. Nonetheless, when the estimated value of the F-test statistic is
In the cointegration analysis, the long or short-term relation- higher than the upper limit, it can be understood as a long-term
ships between variables are examined, and there are limitations in relationship between the variables. Yet, if the calculated F-test
many methods used in the literature. For example, in Ref. [58], statistical value is between the lower limit and the upper limit, the
more than two variables are not permitted and in Ref. [59], the result is interpreted as indecisive. An ARDL test-based error
variables must be integrated to the same degree. However, correction model (ECM) is formulated to predict the short-run
Refs. [60,61]'s ARDL model examines the dynamic relationship relationship between variables in Eqs. 8e15 as follows:
structure between variables and includes a cointegration test that
examines whether there is cointegration between variables even if X
p X
q1
they become stationary at different levels. DLTEMPit ¼ w1 ECTt1 þ rij DLTEMPi;tj þ b1ij DLRECi;tj
j¼1 j¼0

Table 1
X
q2 X
q3

Descriptive statistics and correlation of variables.


þ b2ij DLNRECi;tj þ b3ij DLGDPi;tj þ m1 þ ε1t
j¼0 j¼0
description LTEMP LPREC LREC LNREC LGDP
(8)
mean 2.952 6.435 3.506 4.145 8.351
standard dev. 0.045 0.112 0.643 0.911 0.787
maximum 3.049 6.676 4.869 5.281 9.442 X
p X
q1
minimum 2.844 6.200 2.409 2.432 7.128 DLRECit ¼ w2 ECTt1 þ tij DLRECi;tj þ a1ij DLTEMPi;tj
correlation coefficients j¼1 j¼0
LTEMP 1.000
LPREC 0.029 1.000 X
q2 X
q3

LREC 0.617 0.081 1.000


þ a2ij DLNRECi;tj þ a3ij DLGDPi;tj þ m2 þ ε2t
LNREC 0.672 0.012 0.881 1.000 j¼0 j¼0
LGDP 0.662 0.006 0.879 0.957 1.000 (9)

440
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Ref. [63]'s method can estimate a suitable VAR model or perform


X
p X
q1 a causality test, even if it has unit-roots or is stationary. In this
DLNRECit ¼ w3 ECTt1 þ sij DLNRECi;tj þ g1ij DLRECi;tj model, the steps can be followed without looking at the relation-
j¼1 j¼0 ships between the variables and the degree of integration. First, an
X
q2 X
q3 appropriate k delay in the VAR analysis is required to determine the
þ g2ij DLTEMPi;tj þ g3ij DLGDPi;tj þ m3 þ ε3t dmax value, which is the highest level of stationarity. In the
j¼0 j¼0 following step it is necessary to estimate the model with the
(10) (k þ dmax) lag. Finally, the Wald test is applied to the coefficients
obtained from the VAR model. As a finding, if there is causality, its
directions are determined. The VAR models used in the article are
X
p X
q1
DLGDPit ¼ w4 ECTt1 þ uij DLGDPi;tj þ 41ij DLRECi;tj modelled in the studies of [63e65] in Eq. 16 and 17 as follows:
j¼1 j¼0 2 3 0 2 31
TEMPt TEMPti
X
q2 X
q3 6 RECt 7 Xk B 6 RECti 7 C
42ij DLNRECi;tj þ 43ij DLTEMPi;tj þ m4 þ ε4t 6 7 Bbi 6 7C
þ 4 NRECt 5 ¼ b0 þ @ 4 NRECti 5 A
j¼0 j¼0 i¼1
GDPt GDPti
(11) 0 2 31 2 3
TEMPtj εTEMP
X
dmax B 6 7C 6 7
þ Bbj 6 RECtj 7 C þ 6 εREC 7 (16)
X
p X
q1 @ 4 NRECtj 5 A 4 εNREC 5
DLPRECit ¼ w5 ECTt1 þ rij DLPRECi;tj þ b1ij DLRECi;tj j¼kþ1
GDPtj εGDP
j¼1 j¼0
2 3 0 2 31
X
q2 X
q3
PRECt PRECti
þ b2ij DLNRECi;tj þ b3ij DLGDPi;tj þ m5 þ ε5t 6 RECt 7 Xk B 6 7C
6 7 Bbi 6 RECti 7 C
j¼0 j¼0 4 NRECt 5 ¼ b0 þ @ 4 NRECti 5 A
i¼1
(12) GDPt GDPti
0 2 31 2 3
PRECtj εPREC
X
dmax B 6 7 C 6 7
X X Bbj 6 RECtj 7 C þ 6 εREC 7
p q1
DLRECit ¼ w6 ECTt1 þ tij DLRECi;tj þ a1ij DLPRECi;tj þ @ 4 NRECtj 5 A 4 εNREC 5 (17)
j¼kþ1
j¼1 j¼0 GDPtj εGDP
X
q2 X
q3
þ a2ij DLNRECi;tj þ a3ij DLGDPi;tj þ m6 þ ε6t Where TEMP is maximum temperature, PREC is precipitation, REC
j¼0 j¼0 and NREC are net electricity consumption of renewable and non-
(13) renewable energy and GDP is real Gross Domestic Product per
capita.

X
p X
q1
DLNRECit ¼ w7 ECTt1 þ sij DLNRECi;tj þ g1ij DLRECi;tj 3. Results and discussion
j¼1 j¼0

X
q2 X
q3 3.1. Empirical results
þ g2ij DLPRECi;tj þ g3ij DLGDPi;tj þ m7 þ ε7t
j¼0 j¼0 Before starting the analysis, it is necessary to check the statio-
(14) narity levels of the series. In fact, the ARDL test does not care
whether the series are of I(0) or I(1) order, or a combination of the
two. In both cases, the existence of cointegration can be tested. But
X
p X
q1
DLGDPit ¼ w8 ECTt1 þ uij DLGDPi;tj þ 41ij DLRECi;tj if the series is of I(2) degree, the ARDL methodology of Refs. [60,61]
j¼1 j¼0 will not be valid. The presence of a unit root indicates that the effect
of a shock is long-lasting. Therefore, it is very important to deter-
X
q2 X
q3
þ 42ij DLNRECi;tj þ 43ij DLPRECi;tj þ m8 þ ε8t mine whether the process has a unit root [57]. For these purposes,
j¼0 j¼0 tests in Refs. [66,67] are applied to the series to determine the
stationarity levels of the series. Test results are shown in Table 2.
(15)
The ADF and PP unit root test results in Table 2 show that the
After the ARDL test is applied, CUSUM and CUSUMQ tests are LTEMP and LPREC series are stationary at the level, while the LREC,
applied to check the robustness of the results and the stability of LNREC, and LGDP series are stationary at the first difference. ARDL
the model. These tests generate a graph by recursively calculating bounds test is used instead of the Johansen Cointegration test to
the regression coefficients and residuals. For the stability of the investigate the cointegration for the long-term relationship at
estimated results, the graph should remain between the critical different stationary levels. Therefore, two different models have
values [62]. been established. In one of the models, the temperature is
In the ARDL method, the existence of short and long-term re- considered the dependent variable, and in the other model, the
lationships between climate variables and REC, NREC, and GDP are precipitation is considered the dependent variable. Table 3 shows
tested, but the direction of causality is not investigated. As the se- the ARDL lag length, bounds test, and diagnostic test results for two
ries are not cointegrated at the same level, Ref. [63]'s technique is models.
used as the causality test (Please keep in mind that this is not the TEMP ARDL (2,0,1,4) and PREC ARDL (1,2,1,0) models have been
traditional Granger causality test.). Thereby, firstly, a VAR model selected by estimating the appropriate lag length. The ARDL bounds
with appropriate lag length should be set up and Granger causality test and diagnostic test results of the established models are given
should be tested. in Table 3, and when the diagnostic test results are examined, it is
441
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Table 2
ADF and PP unit root test results for the variables.

variables ADF unit root test results PP unit root test results

level first difference level first difference

t-St. prob. t-St. prob. t-St. prob. t-St. prob. result

LTEMP 6.101 (0) 0.000 6.339 (3) 0.000 8.072 (16) 0.000 19.026 (14) 0.000 I(0)
LPREC 6.371 (0) 0.000 7.461 (1) 0.000 6.980 (7) 0.000 30.358 (36) 0.000 I(0)
LREC 2.984 (0) 0.149 7.661 (0) 0.000 2.943 (1) 0.160 8.611 (5) 0.000 I(1)
LNREC 0.312 (2) 0.987 6.588 (0) 0.000 0.196 (9) 0.990 13.111 (37) 0.000 I(1)
LGDP 1.827 (0) 0.672 6.111 (0) 0.000 2.178 (3) 0.488 6.111 (1) 0.000 I(1)

Note: The lag lengths for the ADF test in parenthesis has been defined according to Schwarz information criterion and critical values have been taken from Ref. [68]. The
maximum lag length has been assigned as 8. The values in parenthesis are ADF lag values. The lag lengths for the PP test in parenthesis have been defined as minimum lags
where autocorrelation doesn't exist according to Newey-West criterion.

Table 3
ARDL model, bounds test, and diagnostic test results.

model LogL AIC* BIC HQ Adj. R-sq choice

TEMP 366 83.561013 ¡4.031167 3.547314 3.862290 0.613716 ARDL(2, 0, 1, 4)


PREC 445 38.664913 ¡1.703606 1.351713 1.580786 0.294741 ARDL(1, 2, 1, 0)

F Statistics Breusch-Godfrey- Breusch-Pagan-Godfrey Ramsey RESET test normality test


Serial Corr. LM test Heterosk.test
F-st p-val F-st p-val F-st p-val Jarque B. p-val

TEMP 5.834 2.322 0.464 6.728 0.818 0.056 0.814 7.691 0.213
PREC 13.582 0.365 0.873 5.511 0.650 0.308 0.582 0.568 0.752

Note: The most suitable lag according to AIC has been shown as *: Akaike information criterion, HQ: Hannan-Quinn information criterion, BIC: Bayes information criterion. For
ARDL bounds test F statistics; the upper and lower limits of 10% significance are: 2.37e3.2; the upper and lower limits of 5% significance are: 2.79e3.67; the upper and lower
limits of 1% significance are: 3.65e4.66. The result of Breusch-Godfrey Serial Correlation LM test and Breusch- Pagan Godfrey Heteroskedasticity test is F-st value, obser-
vation*R-is square value, the first number in Ramsey RESET test is F statistics, Normality test is Jarque Bera value and p-val is the probability value.

seen that the model does not have heteroskedasticity and serial and REC. A positive and significant relationship is found between
correlation. In addition, the Jarque-Bera test for normality and the PREC and NREC. According to the ARDL long-term coefficients, a 1%
Ramsey RESET test for specification errors are applied and any increase in REC results in a 0.031% decrease in TEMP at the 10%
problem is detected. After determining that the model is appro- significance level, while a 1% increase in NREC results in a 0.175%
priate and consistent according to the diagnostic test results, ARDL increase in PREC in Turkey.
bounds test is applied for the existence of a long-term relationship CUSUM and CUSUMQ tests are used to measure the stability of
between the variables. the coefficients of the variables and to determine whether there is a
The F-statistic results of the models in Table 3 confirm that there structural break in the ARDL model. The test results are presented
is a long-term cointegration relationship between the variables. in Fig. 6.
Subject to the results of the ARDL bounds test, the F-statistical CUSUM and CUSUMQ statistics are plotted at the 5% significance
values for the TEMP and PREC models are found to be 5.834 and level and if these statistics remain within the critical limit, the co-
13.582, respectively. At the 5% level of significance, these F-statis- efficients in the ARDL model are considered stable. According to
tics values are greater than the upper bound critical value. As a Fig. 6, there is no structural break because no value exceeds the
result, it has been demonstrated that the variables have a long-term critical limits at the 5% significance level. Thereby, concluding that
relationship, and the parameters of the ARDL (2,0,1,4) and ARDL the long-term coefficients calculated by the ARDL bounds test are
(1,2,1,0) models demonstrate that the long-term and short-term stable.
relationships have been estimated. The short-term and long-term VAR analysis is applied to determine the suitable lag length for
estimation results of the models are shown in Table 4 as follows: TEMP, PREC, REC, NREC, and GDP data. So, two different VAR
At the 10% significance level, the long-term results of the ARDL models are established and the suitable lag length is investigated.
test show a negative and significant relationship between TEMP Appropriate lag length is determined according to Akaike (AIC),

Table 4
ARDL long-term and short-term coefficients.

ARDL long-term coefficients ARDL short-term coefficients

coefficients St.e. t-St. prob. coefficients St.e. t-St. prob.

TEMP ARDL(2, 0, 1, 4)
LREC 0.031 0.017 1.811 0.0820 0.074 0.037 1.970 0.0600
LNREC 0.010 0.013 0.760 0.4538 0.077 0.059 1.311 0.2016
LGDP 0.143 0.053 2.673 0.0130 0.227 0.131 1.733 0.0953
ECT(-1) 1.523
PREC ARDL(1, 2, 1, 0)
LREC 0.068 0.061 1.124 0.2698 0.060 0.088 0.679 0.5018
LNREC 0.175 0.078 2.219 0.0341 0.208 0.142 1.463 0.1537
LGDP 0.144 0.073 1.960 0.0592 0.214 0.105 2.031 0.0511
ECT(-1) 1.140

442
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Fig. 6. Plot of Cusum and CusumQ test for TEMP and PREC models.

Schwarz (SC) and Hannan-Quinn (HQ) information criteria as it is cannot be found. Whilst, bidirectional causality is found between
shown in Table 5. PREC and NREC, and unidirectional causality is found from PREC to
The established VAR models were used to determine whether REC.
the inverse roots of the autoregressive characteristic polynomial
regarding series stationarity are within the unit circle. Because the 3.2. Discussion
inverse roots of the characteristic polynomial are inside the circle, it
has been determined that there is no stationary problem in the Climate change, which can be defined by key indicators such as
model. The conclusion is that there is no autocorrelation problem. rising global temperatures, the effect of greenhouse gases, a lack of
As a result, Toda-Yamamoto causality analysis is used. Table 6 precipitation, and drought, has a daily negative impact on the
summarizes the findings. world. In the last few decades, climate change has become a threat
In Ref. [63]'s causality analysis, the H0 hypothesis states that to plants, animals and people, in short, the entire ecosystem. With
there is no causal relationship between the variables, while the H1 the increase in energy demand due to industrial viability, increase
hypothesis suggests that there is a causal relationship between the in population, economic growth, the effects of energy obtained
variables. Toda Yamamoto causality results reveal that the null from fossil fuels and renewable energy on climate change have
hypothesis cannot be rejected because the probability values be- become important. In line with those global facts and trends,
tween REC, NREC, and TEMP are greater than the critical value of Turkey faces many environmental, ecological, and economic con-
0.05. Nonetheless, the causality between REC, NREC, and TEMP cerns. As shown in this study, conventional energy use is still the
443
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Table 5
Detection of the suitable lag length of variables for the VAR model.

lag LogL LR FPE AIC SC HQ

TEMP 1 167.2991 230.7926* 4.12e-09* ¡7.962113* 7.091347* 7.655127*


PREC 1 124.3779 238.5687* 4.19e-08* ¡5.642050* 4.771284* 5.335064*

Note: The most suitable lag according to AIC has been shown as * and, LR: Log likelihood ratio; FPE: Final Prediction Error; AIC: Akaike information criterion; SC: Schwarz
information criterion; and HQ: Hannan-Quinn information criterion. Since there is not enough space in the table not all the lag values are presented, and only the lag value-
related variable is given.

Table 6
Causality results for variables.

H0 hypothesis value prob. decision result

REC does not Granger cause of TEMP 1.402 0.495 do not reject H0 no causality
NREC does not Granger cause of TEMP 3.338 0.188 do not reject H0 no causality
TEMP does not Granger cause of REC 1.345 0.510 do not reject H0 no causality
TEMP does not Granger cause of NREC 1.671 0.433 do not reject H0 no causality
REC does not Granger cause of PREC 2.870 0.238 do not reject H0 no causality
NREC does not Granger cause of PREC 7.299 0.026 reject H0 NREC / PREC
PREC does not Granger cause of REC 6.303 0.042 reject H0 PREC / REC
PREC does not Granger cause of NREC 7.178 0.027 reject H0 PREC / NREC

main input of the production and consumption behavior of Turkey. temperature (but no causality is found); non-renewable energy
However, there has been a significant increase in the number of consumption and precipitation (two-way causality is found);
flood disasters and forest fires in recent years. Therefore, the additionally, one-way causality is found from precipitation to
Turkish Government should take some precautions and should renewable energy consumption. Furthermore, long-term and the
abide by a sustainable growth strategy. This is not only important short-term relationships between climate change, economic
for the country itself, but also on a global scale. Since Turkey is a growth, and energy types have been identified and demonstrated.
small open emerging market economy with significant trade and The relationship between precipitation and temperature as a
tourism interactions with other countries. proxy for climate change and renewable energy, non-renewable
In the literature, there are studies that investigate the relation- energy and economic growth is important for realizing policies in
ship between renewable energy, non-renewable energy and this section. Suggestions are made for a more effective, more result-
climate change with different models. Examining the relationship oriented and more rational policy-making process while devel-
between renewable energy and climate change in China with the oping policies related to climate change in Turkey. Ref. [71], who
Gray prediction model, Ref. [13] took hydroelectric generation and establish a panel threshold model specific to China in order to
precipitation as a representation for renewable energy and climate measure the impact of renewable energy on climate change and
factor, respectively. As a result, it has been determined that the CO2 emissions, conclude that because of the increase in high
effects of climate change on hydroelectricity vary depending on the technological innovations, renewable energy has a greater effect on
state, and that in the next two decades, climate change will cause reducing CO2 emissions, and emissions decrease as coal-based
direct losses of around 2% in hydroelectricity. In this study, a pos- energy consumption decreases. According to the findings of this
itive relationship is found between precipitation and non- study, TEMP values decrease as REC increases, and that viable
renewable energy consumption, and it is found that the increase renewable energy utilization may play a critical role in a struggling
in non-renewable energy consumption causes an increase in pre- global warming problem by maintaining ecological balance. The
cipitation. The fact that non-renewable energy causes excessive increase in REC and the decrease in TEMP in Turkey reveal the
precipitation might also have a negative consequence for both negative relationship between REC and global warming, and it will
agriculture and the general economy. Ref. [69] conducted a study be a source of motivation for countries other than Turkey to turn to
on the effects of floods and landslides caused by climate change and renewable energy sources from fossil fuel sources. As a result,
increased precipitation in the USA. According to the results of the cleaner and less expensive energy will be used, which will
research, it has been determined that the total annual precipitation contribute to the fight against and prevention of global warming.
and the number of extreme weather events in the USA and other Nonetheless, limitations of the research are related with the
parts of the world have increased in recent years, and this increase gathered convenient data belonging to all variables. In time series
has damaged crop production as well as the total economy. research a necessary time length of data for variables should be
Ref. [70] examined the effects of trade, economic growth, provided for statistical concerns. Because all variables in time series
fertility rate, renewable and non-renewable energy consumption econometric models must cover at least the same period, the lim-
on Europe's ecological footprint and conclude that a 1% increase in itation period of data for one variable can affect the validity of the
real GDP reduces the increase in environmental degradation both model. The longer the data set, the more confident the findings.
in the short and in the long terms. According to the long-term However, developing countries are unlikely to focus heavily on the
estimation of non-renewable energy consumption, it is concluded preservation or provision of data in their national government
that a 1% increase in fossil fuel consumption increases environ- services. As a result, having scientific proof for critical issues is
mental degradation by 98% and increases global GHG emissions. linked to having strong and qualified institutions.
The increase in renewable energy has been found to increase
environmental degradation by a small percentage of about 0.04% in 4. Conclusions and future works
the long run. In this study, environmental degradation is discussed
within the framework of climate change and the relationships are This paper investigates climate change caused by renewable and
determined between renewable energy consumption and non-renewable energy consumption throughout a growing
444
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

economy, in Turkey. A time series ARDL analysis is conducted for particular, may encourage both consumers and investors to prefer
this purpose in the period of 1980e2019. The long and short-term renewable energy sources to produce on-site electricity over con-
relationships are determined with the ARDL bounds test, and ventional energy. In the long run, these policies have the potential
causality relationships are displayed with the Toda Yamamoto and viability of renewable energy applications, and the tempera-
causality test. Firstly, the results of the tests reveal that a 1% in- ture can be reduced or at least maintained at steady-state level if all
crease in renewable energy consumption reduced the temperature developing countries implement them.
by 0.031%. This finding supports the increase of renewable energy In order to struggle with the negative effects of climate change
use can help mitigating global warming by reducing greenhouse (i.e., the global warming caused by energy use and produced
emissions, and the negative effects of climate change can be emissions), which is observed by temperature and precipitation
decreased. Secondly, a 1% increase in non-renewable energy con- levels, it is necessary to reduce food waste, reuse recyclable ma-
sumption causes an increase of 0.175% in precipitation. This finding terials, construct new buildings that produce their own energy
has been evidence of sudden weather changes in Turkey due to from renewable energy sources and implement energy efficiency
meteorological events in recent years. The change in precipitation and effectively. Nonetheless, the present emission routes and sce-
regime, which may be influenced by fossil fuel consumption, causes narios suggest global warming of approximately 3 C by 2100 with a
heavy rain, hail, and flooding, causing damage to buildings, vehi- steady-state increase in temperature eventually. Despite significant
cles, and crops in plantations. Thirdly, there is no evidence of a link efforts to accelerate the transition to renewable energy, progress
between temperature and energy consumption (both renewable has been slow. For future works, it is necessary to increase the
and non-renewable). However, evidence for a bi-directional cau- number of country-specific and quantitative analyses (i.e., time
sality relationship between non-renewable energy consumption series econometrics) related to the relationship between climate
and precipitation has been discovered. In addition, a unidirectional change, energy and economy by considering more details. For
relationship has been determined from precipitation to renewable instance, carbon emissions and local specific parameters can be
energy consumption. Referencing the bi-directional relationship, added to the analysis through energy supply and demand, energy
the increase in non-renewable energy consumption potentially has infrastructure, energy endowments, and energy transportation.
an adverse effect on precipitation, and precautions on this issue are Developing countries can be targeted for a sample research area.
better taken by decision-makers. Attention should be given to Nevertheless, the content included in this study is a global energy
supplying electricity needs from renewable energy sources, which concern related to economics and specifically with Gross Domestic
will lead to using energy more efficiently, and avoiding fossil fuels Product per capita. The markers gathered from all countries sam-
and petroleum-based products. The unidirectional relationship, on ples can raise our awareness by gaining scientific knowledge and
the other hand, highlights evidence of a recent mankind life-cycle taking precautions to prevent countries from an impending natural
scenario in line with the technological advancements covering disaster. This quantitative research is one of them but the number
increasing renewable energy utilization from time to time. More should be increased with more economic indicators.
specifically, it describes recent consumption patterns that have
favored increasing renewable energy utilization demand. While CRediT authorship contribution statement
commenting on this issue without taking into account the fact that
global warming is increasing is open to misinterpretation. Hakan Acarog  lu: Conceptualization, Investigation, Methodol-
The discovery of a bi-directional causality relationship between ogy, Data curation, Software, Writing e original draft, Preparation,
non-renewable energy consumption and precipitation in a specific Writing e review & editing, Visualization, Validation, Supervision.
growing economy can be explained by the fact that this economy's Mustafa Güllü: Investigation, Methodology, Data curation, Soft-
energy demand is heavily based on conventional energy in terms of ware, Formal analysis, Writing e original draft, Preparation,
production and consumption patterns. However, high participation Writing e review & editing, Visualization, Validation.
levels can be very harmful with strong floods and its effects are
starting to be seen frequently in Turkey. In addition to floods, the Declaration of competing interest
number of forest fires has increased in recent years in Turkey. Those
disasters are the main indicators or effects of climate change The authors declare that they have no known competing
through global warming. As a result, the government would have financial interests or personal relationships that could have
done better to take some precautions. For instance, the construc- appeared to influence the work reported in this paper.
tions and homes can be built in safer places, stronger dams can be
constructed, and it is necessary to have intermediate zones so that
Appendix A
the regions will not be flooded and destroyed. Promotions of
microgrid applications based on solar and wind energy, in

Table A1
The literature review for selected papers.

Author/Year Data/Period Method(s) Model Findings


€lük & Mert,
Bo WDI/1961e2010 ARDL CO2 ¼ f fGDP; EPRg Renewable electricity generation contributes to environmental improvement.
(2015) [46], annual
Brini et al. WDI/1980e2011 ARDL GDP ¼ There is a two-way relationship between renewable energy consumption and
(2017) [47], annual f fREC; PO; TRADEg international trade, and a one-way relationship between oil prices.
Ocal & Aslan, WDI/1980e2011 ARDL and Toda GDP ¼ f fREC; K; Lg The result of causality between renewable energy consumption and economic
(2013) [49], annual Yamamoto growth is inconsistent.
Ameway, WDI/1990e2015 ARDL REG ¼ There is a unidirectional causality running from renewables to economic
(2021) [44], annual f fGDPPC; L; TOg growth.
Abbasi, (2021) WDI/1972e2018 Dynamic ARDL GDP ¼ f fEC; EP; Energy consumption, industrial growth, urbanization and CO2 emissions
[42], annual IVA; UPG; CO2g positively affect economic growth.
(continued on next page)

445
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Table A1 (continued )

Author/Year Data/Period Method(s) Model Findings

Brini et al. WDI, EIA, NOAA/1980 Panel ARDL CC ¼ f fREC; NREC; There is bidirectional causality between non-renewable energy and climate
(2021) [14], e2014 annual GDPg change, and unidirectional causality from climate change to renewable energy
consumption.
Al Mulali, Euromonitor database/ ARDL CO2 ¼ f fGDP; K; L; While fossil fuel energy consumption increases pollution, renewable energy
(2015) [43], 1982e2011 annual EX; IM; ELF; ELR; consumption does not have a significant effect on reducing pollution.
NRECg
Amoako & WDI/1981e2014 FMOLS and CCR EC ¼ f fFDI; FD; PX; While FDI and industry value added are positively related to energy
Insaidoo, annual TO; IVAg consumption, financial development and energy price are inversely related.
(2021) [45],
Baek, (2015) WDI/1960e2010 ARDL CO2 ¼ f fGDP; ENg Energy consumption has a detrimental effect on the environment for most
[52], annual countries.
Bo€lük & Mert, WDI/1990e2018 ARDL CO2 ¼ f fGDPPC; Renewable energy consumption contributes approximately 1/2 less than fossil
(2014) [53], annual REN; FOSSg energy consumption in terms of greenhouse gas emissions.
Jebli & Youssef, WDI, EIA, PENN WORLD ARDL, VECM and Granger E ¼ f fGDP; GDP 2 ; Non-renewable energy and trade have a positive impact on CO2 emissions, but
(2015) [48], TABLES/1980e2009 RE; NRE; TRADEg renewable energy has a weak and negative impact.
annual
Njoh, (2021) WDI GLM CO2 ¼ f fREN; UPG; Energy consumption is inversely proportional to CO2 emissions.
[54], GDPg
Shafiei & WDI, EIA/1980e2011 STIRPAT for EKC CO2 ¼ f fP; GDP; Non-renewable energy consumption increases CO2 emissions and renewable
Salim, annual REC; NRECg energy consumption reduces.
(2014) [55],
Sinha & WDI, PENN WORLD ARDL for EKC CO2 ¼ f fGDP; Renewable energy has a significant negative impact on CO2 emissions.
Shahbaz, TABLES/1971e2015 GDP 2 ; REN; TRADE;
(2018) [50], TFPg
Yurtkuran, WDI, OECD and KOF/ Gregory-Hansen CO2 ¼ f fREN; AGR; Agriculture, economic globalization and renewable energy production increase
(2021) [51]. 1970e2017s cointegration test and KOFEg environmental pollution.
Bootstrap ARDL

CC: Climate change factors, EX: Export, NREC: Non-renewable energy consumption,
CCR: Canonical Cointegration Regression, FD: Financial Development, P: Population,
CO2: Carbon dioxide emissions, FDI: Foreign Direct Investment, PO: Price oil,
EC: Energy consumption, FMOLS: Fully Modified Ordinary Least Squares, PX: Energy Price,
ELF: Electricity consumption from fossil fuels sources, FOSS: Fossil fuel energy consumption, REG: Renewable electricity generation,
ELR: Electricity consumption from renewable sources, GDP: Real Gross Domestic Product, REN: Renewable energy consumption,
EN: Energy consumption, GDPPC: Per capita Gross Domestic Product, TFP: Total factor productivity,
EP: Electricity prices, GLM: General linear model, TO: Trade openness,
EPR: Electricity production from renewable sources, IM: Import, TRADE: Foreign trade,
REC: Renewable energy consumption. IVA: Industrial value-added, URB: Urbanization
K: Gross fixed capital formation, VECM: Vector error correction model.
KOFE: Economic KOF Globalization Index,
L: Total labour force.

References [12] M. Ahmad, Z. Ahmed, X. Yang, N. Hussain, A. Sinha, Financial development


and environmental degradation: do human capital and institutional quality
make a difference? Gondwana Res. 105 (2022) 299e310.
[1] S. Paltsev, Projecting energy and climate for the 21st century, Economic En-
[13] B. Wang, X-j Liang, H. Zhang, L. Wang, Y.-M. Wei, Vulnerability of hydropower
ergy Environ. Pol. 9 (2020) 43þ.
generation to climate change in China: results based on Grey forecasting
[2] Z. Langnel, G.B. Amegavi, Globalization, electricity consumption and ecological
model, Energy Pol. 65 (2014) 701e707.
footprint: an autoregressive distributive lag (ARDL) approach, Sustain. Cities
[14] R. Brini, Renewable and non-renewable electricity consumption, economic
Soc. 63 (2020) 102482.
growth and climate change: evidence from a panel of selected African
[3] S. Zaidi, K. Saidi, Environmental pollution, health expenditure and economic
countries, Energy 223 (2021) 120064.
growth in the Sub-Saharan Africa countries: panel ARDL approach, Sustain.
[15] W.T.L. Yong, V.Y. Thien, R. Rupert, K.F. Rodrigues, Seaweed: a potential climate
Cities Soc. 41 (2018) 833e840.
change solution, Renew. Sustain. Energy Rev. 159 (2022) 112222.
[4] A. Sharif, O. Baris-Tuzemen, G. Uzuner, I. Ozturk, A. Sinha, Revisiting the role of
[16] M. Turkes, M.T. Turp, N. An, T. Ozturk, M.L. Kurnaz, Impacts of climate change
renewable and non-renewable energy consumption on Turkey's ecological
on precipitation climatology and variability in Turkey, in: N.B. Harmancioglu,
footprint: evidence from Quantile ARDL approach, Sustain. Cities Soc. 57
D. Altinbilek (Eds.), Water Resources of Turkey, Springer International Pub-
(2020) 102138.
lishing, Cham, 2020, pp. 467e491.
[5] S. Shan, S.Y. Genç, H.W. Kamran, G. Dinca, Role of green technology innovation
[17] F. Giorgi, P. Lionello, Climate change projections for the Mediterranean region,
and renewable energy in carbon neutrality: a sustainable investigation from
Global Planet. Change 63 (2008) 90e104.
Turkey, J. Environ. Manag. 294 (2021) 113004.
[18] F.G. Kuglitsch, A. Toreti, E. Xoplaki, P.M. Della-Marta, C.S. Zerefos, M. Türkeş, et
[6] E. Kılıç, R. Puig, G. Zengin, C.A. Zengin, I.P.P. Fullana, Corporate carbon foot-
al., Heat wave changes in the eastern Mediterranean since 1960, Geophys.
print for country Climate Change mitigation: a case study of a tannery in
Res. Lett. 37 (2010).
Turkey, Sci. Total Environ. 635 (2018) 60e69.
lu, F.P. García Ma
rquez, High voltage direct current systems through [19] E. Erlat, M. Türkeş, Observed changes and trends in numbers of summer and
[7] H. Acarog
tropical days, and the 2010 hot summer in Turkey, Int. J. Climatol. 33 (2013)
submarine cables for offshore wind farms: a life-cycle cost analysis with
1898e1908.
voltage source converters for bulk power transmission, Energy (2022) 123713.
[20] M. Türkeş, T. Koç, F. Sariş, Spatiotemporal variability of precipitation total
[8] A. Marra, E. Colantonio, The path to renewable energy consumption in the
series over Turkey, Int. J. Climatol. 29 (2009) 1056e1074.
European Union through drivers and barriers: a panel vector autoregressive
[21] E. Erlat, M. Türkeş, Analysis of observed variability and trends in numbers of
approach, Soc. Econ. Plann. Sci. 76 (2021) 100958.
lu, M.C. Baykul, Economic analysis of flat-plate solar collectors frost days in Turkey for the period 1950e2010, Int. J. Climatol. 32 (2012)
[9] H. Acarog
1889e1898.
(FPSCs): a solution to the unemployment problem in the city of Eskisehir,
[22] J. Lelieveld, P. Hadjinicolaou, E. Kostopoulou, J. Chenoweth, M. El Maayar,
Renew. Sustain. Energy Rev. 64 (2016) 607e617.
C. Giannakopoulos, et al., Climate change and impacts in the eastern medi-
[10] U. Bulut, G. Muratoglu, Renewable energy in Turkey: great potential, low but
terranean and the Middle East, Climatic Change 114 (2012) 667e687.
increasing utilization, and an empirical analysis on renewable energy-growth
[23] I.G. Dino, C. Meral Akgül, Impact of climate change on the existing residential
nexus, Energy Pol. 123 (2018) 240e250.
building stock in Turkey: an analysis on energy use, greenhouse gas emissions
[11] C. Xue, M. Shahbaz, Z. Ahmed, M. Ahmad, A. Sinha, Clean energy consumption,
and occupant comfort, Renew. Energy 141 (2019) 828e846.
economic growth, and environmental sustainability: what is the role of eco-
[24] E. Rhodes, A. Hoyle, M. McPherson, K. Craig, Understanding climate policy
nomic policy uncertainty? Renew. Energy 184 (2022) 899e907.
projections: a scoping review of energy-economy models in Canada, Renew.

446
H. Acaroglu and M. Güllü Renewable Energy 193 (2022) 434e447

Sustain. Energy Rev. 153 (2022) 111739. 587e595.


[25] M. Tutak, J. Brodny, Renewable energy consumption in economic sectors in [47] R. Brini, M. Amara, H. Jemmali, Renewable energy consumption, International
the EU-27. The impact on economics, environment and conventional energy trade, oil price and economic growth inter-linkages: the case of Tunisia,
sources. A 20-year perspective, J. Clean. Prod. 345 (2022) 131076. Renew. Sustain. Energy Rev. 76 (2017) 620e627.
[26] F. Wiese, J. Thema, L. Cordroch, Strategies for climate neutrality. Lessons from [48] M. Ben Jebli, S. Ben Youssef, The environmental Kuznets curve, economic
a meta-analysis of German energy scenarios, Renewable. Sustain. Energy. growth, renewable and non-renewable energy, and trade in Tunisia, Renew.
Trans. 2 (2022) 100015. Sustain. Energy Rev. 47 (2015) 173e185.
[27] IRENA, Renewable Capacity Statistics 2021 Abu Dhabi, International Renew- [49] O. Ocal, A. Aslan, Renewable energy consumptioneeconomic growth nexus in
able Energy Agency (IRENA), 2021. Turkey, Renew. Sustain. Energy Rev. 28 (2013) 494e499.
[28] IRENA, Renewable Energy Statistics Abu Dhabi, The International Renewable [50] A. Sinha, M. Shahbaz, Estimation of environmental Kuznets curve for CO2
Energy Agency, 2018. emission: role of renewable energy generation in India, Renew. Energy 119
[29] ETKB, National Renewable Energy Action Plan for Turkey, Republic of Turkey (2018) 703e711.
Ministry of Energy Natural Resources, Ankara, 2014. [51] S. Yurtkuran, The effect of agriculture, renewable energy production, and
[30] TEIAS, Turkish Electricity Transmission Company, 2021. globalization on CO2 emissions in Turkey: a bootstrap ARDL approach, Renew.
[31] L. Charfeddine, M. Kahia, Impact of renewable energy consumption and Energy 171 (2021) 1236e1245.
financial development on CO2 emissions and economic growth in the MENA [52] J. Baek, Environmental Kuznets curve for CO2 emissions: the case of Arctic
region: a panel vector autoregressive (PVAR) analysis, Renew. Energy 139 countries, Energy Econ. 50 (2015) 13e17.
(2019) 198e213. [53] G. Bo€lük, M. Mert, Fossil & renewable energy consumption, GHGs (greenhouse
[32] K. Sufyanullah, K.A. Ahmad, M.A. Sufyan Ali, Does emission of carbon dioxide gases) and economic growth: evidence from a panel of EU (European Union)
is impacted by urbanization? An empirical study of urbanization, energy countries, Energy 74 (2014) 439e446.
consumption, economic growth and carbon emissions - using ARDL bound [54] A.J. Njoh, Renewable energy as a determinant of inter-country differentials in
testing approach, Energy Pol. 164 (2022) 112908. CO2 emissions in Africa, Renew. Energy 172 (2021) 1225e1232.
[33] K.R. Abbasi, M. Shahbaz, J. Zhang, M. Irfan, R. Alvarado, Analyze the envi- [55] S. Shafiei, R.A. Salim, Non-renewable and renewable energy consumption and
ronmental sustainability factors of China: the role of fossil fuel energy and CO2 emissions in OECD countries: a comparative analysis, Energy Pol. 66
renewable energy, Renew. Energy 187 (2022) 390e402. (2014), 547e56.
[34] K.H. Nguyen, M. Kakinaka, Renewable energy consumption, carbon emissions, [56] S.S. Mahjabeen, S. Chughtai, B. Simonetti, Renewable energy, institutional
and development stages: some evidence from panel cointegration analysis, stability, environment and economic growth nexus of D-8 countries, Energy
Renew. Energy 132 (2019) 1049e1057. Strategy Rev. 29 (2020) 100484.
[35] M.M. Islam, M. Irfan, M. Shahbaz, X.V. Vo, Renewable and non-renewable [57] J.M.A. Wooldridge, Introductory Econometrics : a Modern Approach/Jeffrey M.
energy consumption in Bangladesh: the relative influencing profiles of eco- Wooldridge., Cengage Learning, Boston, 2018.
nomic factors, urbanization, physical infrastructure and institutional quality, [58] R.F. Engle, C.W.J. Granger, Co-integration and error correction: representation,
Renew. Energy 184 (2022) 1130e1149. estimation, and testing, Econometrica 55 (1987) 251e276.
[36] K.R. Abbasi, K. Hussain, M. Radulescu, I. Ozturk, Asymmetric impact of [59] S. Johansen, Estimation and hypothesis testing of cointegration vectors in
renewable and non-renewable energy on the industrial sector in Pakistan: Gaussian vector autoregressive models, Econometrica 59 (1991) 1551e1580.
fresh evidence from Bayesian and non-linear ARDL, Renew. Energy 187 (2022) [60] M.H. Pesaran, Y. Shin, An Autoregressive Distributed-Lag Modelling Approach
944e957. to Cointegration Analysis, CAMBRIDGE UNIVERSITY PRESS, Great Britain,
[37] S. Mukhtarov, S. Yüksel, H. Dinçer, The impact of financial development on 1998, p. 371.
renewable energy consumption: evidence from Turkey, Renew. Energy 187 [61] M.H. Pesaran, Y. Shin, R.J. Smith, Bounds testing approaches to the analysis of
(2022) 169e176. level relationships, J. Appl. Econom. 16 (2001) 289e326.
[38] N. Das, P. Bera, D. Panda, Can economic development & environmental sus- [62] R.L. Brown, J. Durbin, J.M. Evans, Techniques for testing the constancy of
tainability promote renewable energy consumption in India?? Findings from regression relationships over time, J. Roy. Stat. Soc. B 37 (1975) 149e192.
novel dynamic ARDL simulations approach, Renew. Energy 189 (2022) [63] H.Y. Toda, T. Yamamoto, Statistical inference in vector autoregressions with
221e230. possibly integrated processes, J. Econom. 66 (1995) 225e250.
[39] K. Solaun, E. Cerda, Climate change impacts on renewable energy generation. [64] H. Yamada, H.Y. Toda, Inference in possibly integrated vector autoregressive
A review of quantitative projections, Renew. Sustain. Energy Rev. 116 (2019) models: some finite sample evidence, J. Econom. 86 (1998) 55e95.
109415. [65] M. Irandoust, The renewable energy-growth nexus with carbon emissions and
[40] F.A. Canales, P. Jadwiszczak, J. Jurasz, M. Wdowikowski, B. Ciapała, technological innovation: evidence from the Nordic countries, Ecol. Indicat. 69
B. Ka zmierczak, The impact of long-term changes in air temperature on (2016) 118e125.
renewable energy in Poland, Sci. Total Environ. 729 (2020) 138965. [66] D.A. Dickey, W.A. Fuller, Distribution of the estimators for autoregressive time
[41] J. Gyimah, X. Yao, M.A. Tachega, I. Sam Hayford, E. Opoku-Mensah, Renewable series with a unit root, J. Am. Stat. Assoc. 74 (1979) 427e431.
energy consumption and economic growth: new evidence from Ghana, En- [67] C.B.P. Peter, P. Perron, Testing for a unit root in time series regression, Bio-
ergy 248 (2022) 123559. metrika 75 (1988) 335e346.
[42] K.R. Abbasi, M. Shahbaz, Z. Jiao, M. Tufail, How energy consumption, industrial [68] J.G. MacKinnon, Numerical distribution functions for unit root and cointe-
growth, urbanization, and CO2 emissions affect economic growth in Pakistan? gration tests, J. Appl. Econom. 11 (1996) 601e618.
A novel dynamic ARDL simulations approach, Energy 221 (2021) 119793. [69] C. Rosenzweig, F.N. Tubiello, R. Goldberg, E. Mills, J. Bloomfield, Increased Crop
[43] U. Al-Mulali, B. Saboori, I. Ozturk, Investigating the environmental Kuznets Damage in the US from Excess Precipitation under Climate Change, Elsevier
curve hypothesis in Vietnam, Energy Pol. 76 (2015) 123e131. Science B.V., Amsterdam., Great Britain, 2002, p. 197.
[44] B. Ameyaw, Y. Li, Y. Ma, J.K. Agyeman, J. Appiah-Kubi, A. Annan, Renewable [70] A.A. Alola, F.V. Bekun, S.A. Sarkodie, Dynamic impact of trade policy, economic
electricity generation proposed pathways for the US and China, Renew. En- growth, fertility rate, renewable and non-renewable energy consumption on
ergy 170 (2021) 212e223. ecological footprint in Europe, Sci. Total Environ. 685 (2019) 702e709.
[45] S. Amoako, M. Insaidoo, Symmetric impact of FDI on energy consumption: [71] B. Lin, J. Zhu, The role of renewable energy technological innovation on
evidence from Ghana, Energy 223 (2021) 120005. climate change: empirical evidence from China, Sci. Total Environ. 659 (2019)
[46] G. Bo€ lük, M. Mert, The renewable energy, growth and environmental Kuznets 1505e1512.
curve in Turkey: an ARDL approach, Renew. Sustain. Energy Rev. 52 (2015)

447

You might also like