The Effect of Renewable and Non-Renewable Energy on Economic Growth: A Panel Cointegration Analysis for the Top Renewable Energy Consumers (1970–2023)

Abstract

The relationship between renewable (REN) and non-renewable (NREN) energy and economic growth plays a fundamental role in sustainable development. The number of studies on this relationship in countries with the highest REN consumption is limited. This study analyzes the effects of REN and NREN on economic growth between 1970 and 2023, focusing on the ten leading countries in REN consumption. These countries constitute an appropriate sample for analysis, not only due to their high REN capacity but also because they represent diverse levels of economic development. For this purpose, second-generation panel data methods were employed to investigate the long-run effects, taking into account cross-sectional dependence and heterogeneity in the dataset. The CADF unit root test developed by Pesaran indicated that all variables are stationary at their first differences. The Westerlund panel cointegration test confirmed the existence of a long-run relationship among the variables. Long-run coefficients were estimated using the Common Correlated Effects Mean Group (CCE) approach developed by Pesaran and the Augmented Mean Group (AMG) estimators proposed by Bond & Eberhardt and Eberhardt & Teal. The results revealed that renewable energy consumption has a positive and significant effect on economic growth, while fossil fuel consumption continues to have a favorable effect on growth. However, the negative and significant effect of primary renewable energy production suggests that technological deficiencies and efficiency problems in current production structures may play a role. Overall, this study highlights the necessity of aligning energy policies with both environmental sustainability and economic growth objectives.

1. Introduction

The rapid growth seen in the global population and advances in civilization resulted in a rise in energy demand [1]. Over the past centuries, the transition from agrarian economies to industrial societies has significantly elevated energy consumption, accelerating urbanization and technological progress, but also contributing to environmental degradation [2]. In the early periods of development, many countries tend to overlook the risk of environmental pollution. Accordingly, the higher the level of economic growth, the greater the environmental degradation [3]. Thus, environmental degradation has become one of the problems that needs to be solved as soon as possible. In this context, scholars, policymakers, and governments are making significant efforts to construct environmentally conscious economies [2].

Considering the environmental implications of economic growth, energy is a fundamental pillar across all sectors of modern economies and lays the foundation for almost all economic activities [4]. Energy has an undeniable importance for growth. Similarly, the increasing environmental damage caused by greenhouse gas emissions from traditional NREN consumption necessitates a higher level of attention. Therefore, despite energy being necessary for development and social welfare, the solutions to climate change within the sustainable development agenda rely on the increased use of REN [3]. In this regard, the development of REN sources and the supply of critical minerals are central elements of contemporary energy policies.

It is widely recognized that energy is a fundamental production factor, and that energy demand keeps increasing steadily on a global level [5]. As stated by British Petroleum [6], global primary energy consumption rose by approximately 2.2% in 2017, the most rapid rise since 2013. Furthermore, critical minerals (essential raw materials used in the production of REN technologies such as electric vehicles and power grids) have gained increased importance. According to the REN21 [7] report, growth in the REN sector in 2023, driven by the expanding share of clean energy applications, significantly increased the demand for critical minerals. For instance, lithium demand tripled, while demand for cobalt and nickel increased by 70% and 40%, respectively. These trends highlight that growing energy consumption is closely tied to pressing issues such as environmental degradation [8]. Comparing fuels, natural gas showed the highest increase in consumption, followed by REN and oil, respectively. Despite the increasing significance and growing consumption, REN still has a small portion in the global energy portfolio in comparison to NREN sources [6].

REN plays a significant role in sustainable growth in economy, and the drive to increase its use is a central policy objective for many developing countries. Globally, REN has various sources, including biomass, direct solar energy, wind energy, hydropower (for heating and electricity generation), ocean energy (wave and tidal), and geothermal energy [9]. One of the primary factors laying the foundation of the development of REN is climate change. Scientists widely acknowledge that fossil fuel use significantly contributes to greenhouse gas emissions, making it a major source of global warming [10]. Therefore, effectively implemented REN policies not only help reduce emissions but also slow climate change down and contribute to economic growth. The increase in REN adoption has been supported in policies of many governments, including REN generation tax credits, installation subsidies, renewable portfolio standards, and the establishment of markets for REN sources [11]. Moreover, advancements in technology that reduce the total installation costs of REN facilities have been another driving force behind this trend. The substantial fluctuations in oil prices over the past decade have also played a significant role in encouraging the shift to REN sources [12]. As a result, strategic investments in REN sources have strengthened energy security in both industrialized and developing countries by reducing dependence on fuel imports, improving access to energy, and contributing to poverty alleviation. This, in turn, yields a multitude of benefits, including low-carbon development and job creation [13].

Despite the increasing share of REN sources, NREN has remained a dominant component of energy use in production since the Industrial Revolution. Nowadays, NREN is not only the most utilized energy source, but it also constitutes more than 87% of global primary energy consumption [14]. This widespread use of NREN not only facilitates output production but also is a major source of CO₂ emissions [15]. The rapid increase in energy consumption, its effects on the environment, and the global pursuit of alternative energy sources have created intense debate. However, the exclusion of NREN from the equation of economic growth introduces concerns regarding social welfare. REN stands out as the most viable substitute for non-renewable sources due to its potential to reduce carbon emissions. However, when not used efficiently, it may also negatively affect productivity [14]. Replacing NREN with REN alternatives processes increases the energy security [16]. Increasing energy efficiency is widely considered the most cost-effective approach to reducing emissions, enhancing energy security, and boosting competitiveness. This can be achieved by increasing both the generation and consumption of REN [17]. The widespread use of fossil fuels clearly elevates greenhouse gas (GHG) and CO₂ emissions, thereby exacerbating environmental degradation. Climate change leads to long-term fluctuations in key components of the global climate system, including precipitation and temperature [18]. Ultimately, although NREN sources may continue to provide access to technological innovation in the short term, they are approaching the final stage of their life cycle [19].

The REN storage capacity is limited in comparison to fossil fuels, which may result in supply shortages during periods of high demand [20]. Although the initial investment costs of REN are higher than those of NREN sources, the costs of solar and wind technologies have significantly declined in recent years. This trend is expected to lead to a gradual decrease in the overall cost of REN production. Consequently, the average cost of producing REN is likely to decrease as its usage becomes more widespread [21]. Many studies reported that energy consumption contributes to economic growth. However, while energy usage promotes growth, the usage of fossil fuels contributes to elevating CO₂ emissions and environmental degradation. These negative outcomes impact both public health and national economy; low air quality harms human health, reduces labor productivity, and imposes economic burdens [22]. These negative effects underscore the growing importance of investing in REN as an alternative to fossil fuels [23].

Aiming to address certain gaps observed in the existing literature, this study provides three major contributions.

• While most studies in the literature consider total energy consumption as a single variable, this study differentiates between REN and NREN, thereby conducting a comparative analysis of their long-run effects on growth. This approach allows for more concrete inferences regarding which type of energy should be promoted under specific conditions in the context of sustainable development policies.
• The inclusion of countries with both high levels of REN consumption and diverse socioeconomic structures allows for the observation of how the energy-growth nexus varies across structural differences. The scarcity of studies addressing these countries within a long-term and comparative framework enhances the originality of this research.
• By employing second-generation panel data techniques (CCE, AMG) that consider cross-sectional dependence and heterogeneity, this study provides more reliable long-run coefficient estimates. In addition, the long-run equilibrium among the variables was confirmed through the Westerlund (2007) [24] panel cointegration test, while the stationarity of the series was verified using Pesaran’s CADF test.

This study introduces a new perspective to the literature on the relationship between energy consumption and economic growth at both theoretical and empirical levels. By simultaneously addressing the environmental and economic dimensions of energy transition, the study offers policymakers an evidence-based and practical reference framework.

2. Empirical Literature Review

In recent years, more studies comprehensively investigated the association of economic growth with both REN and NREN. However, empirical studies reported mixed and sometimes contradictory findings, largely due to variations in methodologies, country samples, and specific empirical models employed. One of the key motivations for focusing on this topic is the promising potential of energy consumption to enhance our understanding of the dynamics of economic growth. More recently, climate change has become a global threat to all nations, and the driving factor in this change is carbon emissions. As a result, countries have increasingly turned to both REN and NREN sources to meet increasing energy demands. This study primarily characterizes the relationship between energy usage and economic growth and provides practical policy insights for decision-makers and researchers.

There are 4 popular hypotheses aiming to clarify the association between energy and economic growth [11,25,26,27,28]. Growth Hypothesis posits a unidirectional causality from energy consumption to economic growth, stating that energy consumption, together with labor and capital, plays a significant role in promoting economic growth [29]. Therefore, policies reducing energy consumption may decrease production and thereby reduce economic performance [30]. Conservation Hypothesis, which holds in case of causality from economic growth to energy consumption, suggests that economic growth drives energy consumption [31]. It claims that energy-saving policies designed to decrease energy demand would not reduce economic performance. Therefore, measures aiming to reduce GHG emissions, increase energy efficiency, or manage energy demand would have only a negligible effect on economic growth since the economy is relatively less energy-dependent [32]. Feedback Hypothesis posits a bidirectional causal relationship, emphasizing that energy-saving policies and energy supply shocks would negatively influence economic growth, and that these adverse effects, in turn, are reflected in energy consumption patterns [23]. The Neutrality Hypothesis, which applies in case of no causality between energy consumption and economic growth, states that there is no significant relationship between them [33].

The relevant is summarized in

Table A1. Summary of Empirical Studies on the Relationship between Energy Consumption and Economic Growth. Panel A: Studies Reporting Positive Effects, (REN or NREN positively affects economic growth); Panel B: Studies Reporting Negative Effects, (REN or NREN negatively affects economic growth); Panel C: Studies Reporting No Significant Effect; Panel D: Studies Reporting Mixed Results; Panel E: Studies Reporting Bidirectional Causality.

Panel A
Author(s)	Year	Countries	Period	Methodology	Result
Valadkhani & Nguyen [57]	2019	79 Countries	1965–2017	Panel Data	Positive
Lin & Moubarak [56]	2014	China	1977–2011	ARDL/ECM-Granger	Positive
Pao & Fu [61]	2013	Brazil	1980–2010	VECM Granger	Positive
Doğan [36]	2016	Türkiye	1988–2012	ARDL	Positive
Zhang [62]	2011	Russia	1970–2008	Cointegration/Granger Causality	Positive
Chien & Hu [53]	2007	45 Economies	2001–2002	Data Envelopment Analysis DEA	Positive
Payne [63]	2011	US	1949–2007	Toda–Yamamoto Procedure	Positive
Shahbaz et al. [64]	2012	Pakistan	1972–2011	VECM	Positive
Canh [65]	2010	Vietnam	1975–2010	Johansen Granger Causality/VAR	Positive
Hamit-Haggar [66]	2016	11 SSA	1971–2007	Panel Cointegration/Bootstrap Corrected Granger Causality	Positive
Lee & Chang [67]	2008	16 Asian	1971–2002	FMOLS and Causality	Positive
War & Ayres [68]	2010	US	1946–2000	Granger/VECM	Positive
Pao & Tsai [69]	2011	BRIC	1980–2007	Gray Prediction/VECM	Positive
Yiping [70]	2011	China	1978–2008	OLS	Positive
Apergis & Payne [52]	2011	6 Central America	1980–2006	FMOLS	Positive
Al-mulali et al. [71]	2014	18 LAC	1990–2011	DOLS	Positive
Caraiani et al. [72]	2014	Emerging Europe	1980–2013	VECM	Positive
Öztürk & Bilgili [73]	2015	51 SSA	1980–2009	Dynamic Panel OLS	Positive
Aslan & Ocal [17]	2016	8 EU	1990–2009	ARDL Asymmetric	Positive
Adams et al. [74]	2018	30 SSA	1980–2012	FMOLS/DOLS	Positive
Afonso et al. [75]	2017	28 Countries	1995–2013	ARDL (PMG and MG)	Positive
Gökçeli [76]	2023	BRICS	1990–2019	GMM	Positive
Caner & Yaşar [77]	2024	Turkey	1990–2019	ARDL Toda–Yamamoto	Positive
Kavaz & Kaya [78]	2023	Turkey	1982–2021	ARDL	Positive
Bloch et al. [51]	2015	China	1965–2013	ARDL VECM Granger Causality	Positive
Panel B
Author(s)	Year	Countries	Period	Methodology	Result
Alshehry & Belloumi [59]	2015	Saudi Arabia	1971–2010	Cointegration/VAR	Long-run: Negative, Short-run: Positive
Alam et al. [58]	2012	Bangladesh	1972–2006	ARDL/VECM	Negative
Jaforullah & King [79]	2015	USA	1965–2012	VECM Granger	Negative
Apergis & Payne [80]	2010	15 Emerging Markets	1980–2006	FMOLS/Panel Causality	Negative
Öztürk & Acaravcı [81]	2011	11 MENA	1971–2006	ARDL Bounds	Negative
Razmi et al. [60]	2019	Iran	1990–2014	ARDL	Long-run Negative, Short-run Positive
Panel C
Author(s)	Year	Countries	Period	Methodology	Result
Ozcan & Ozturk [54]	2019	17 Emerging	1990–2016	Bootstrap Panel Causality	No Significant Effect
Özer [55]	2023	Denmark	1990–2018	Fourier ADF/Fourier ADL	No Significant Effect
Panel D
Author(s)	Year	Countries	Period	Methodology	Result
Wolde-Rufael [82]	2010	6 Coal Consuming Countries	1965–2005	VAR Granger	Mixed
Tuna & Tuna [23]	2019	5 ASEAN	1980–2015	Hacker & Hatemi-J	Mixed
Tuğcu & Topçu [83]	2018	G7	1980–2014	NARDL/Granger	Mixed
Destek & Aslan [84]	2017	17 Emerging Economies	1980–2012	Panel Granger	Mixed
Bildirici & Bakirtas [85]	2014	BRICTs	1980–2011	ARDL	Oil Positive, NG and Coal Mixed
Lei & Pan [86]	2014	Biggest Coal Consumers	2000–2010	Panel Causality	Mixed
Luqman et al. [87]	2019	Pakistan	1990–2016	NARDL	REN Positive and Negative Shocks; Nuclear Positive
Panel E
Author(s)	Year	Countries	Period	Methodology	Result
Apergis & Payne [88]	2011	16 Emerging Economies	1990–2007	Panel Granger	Bidirectional Causality
Aydın [89]	2019	26 OECD	1980–2015	Panel Frequency Causality	Bidirectional Causality
Kahia et al. [90]	2016	13 MENA Net Oil Exporters	1980–2012	FMOLS/Panel Granger	Bidirectional Causality

, which includes the methodologies and results of various empirical studies. Researchers achieved remarkable results by addressing the relationships between economic growth, energy use, and environmental degradation [34,35,36,37]. These studies reveal that increasing energy security, the depletion risk of conventional energy resources, GHG emissions, and other environmental challenges have necessitated a shift from conventional to REN sources. Therefore, it is very important to comprehend the relationships between REN and NREN consumption, CO₂ emissions, and economic growth. It offers insight into the extent of the economy’s dependency on energy and supports the effective design of energy policies [38].

In summary, even though previous studies provided extensive evidence on the relationship between energy consumption and economic growth, a considerable portion of the studies treat energy consumption as a single aggregate variable and do not examine in detail the differentiated effects of renewable (REN) and non-renewable (NREN) energy. Moreover, the number of studies that evaluate countries ranking among the top in renewable energy consumption worldwide within a long-term and comparative analytical framework is limited. This study seeks to fill this gap in the literature by applying second-generation panel data techniques to the ten countries with the highest levels of REN consumption. In doing so, it identifies the long-run effects of different types of energy on economic growth both at the panel level and at the individual country level. The findings allow for the development of evidence-based strategic recommendations that can be tailored to country-specific conditions by policymakers. In particular, the panel cointegration and causality tests employed in this study, while consistent with methods used in similar studies in the literature, provide an extended framework in terms of both scope and methodology. This approach enables a more robust theoretical and empirical analysis of the energy transition-economic growth nexus, as emphasized in the introduction.

3. Materials and Methods

3.1. Theoretical Framework and Model Specification

This study investigates the long-term effects of REN and NREN consumption on economic growth by focusing on ten leading countries in renewable energy use. The selected countries (Brazil, China, France, Germany, India, Italy, Japan, Spain, the UK, and the USA) draw attention not only for their high renewable energy capacity but also for representing varying levels of economic development. This selection also allows for comparing the differing effects of renewable energy policies across developed and developing countries. The analysis covers the period 1970–2023. The reason for selecting this period has two aspects: on the one hand, the availability of a long-term and reliable dataset, and on the other, the opportunity to examine the energy-growth relationship across different periods. For instance, the oil crises of the 1970s, the growing awareness of climate change in the 1990s, and the acceleration of renewable energy policies after 2000 are all critical processes encompassed within this timeframe. This study primarily aims to quantitatively examine the effect of renewable energy consumption on economic growth.

The following model was estimated for the analysis conducted using panel data:

$${\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{G}\mathrm{D}\mathrm{P}\mathrm{P}\mathrm{C}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{R}\mathrm{E}\mathrm{N}\mathrm{E}\mathrm{W}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{F}\mathrm{O}\mathrm{S}\mathrm{S}\mathrm{I}\mathrm{L}}_{\mathrm{i}\mathrm{t}}+{{\mathsf{\beta }}_3\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{D}\mathrm{E}\mathrm{F}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{P}\mathrm{O}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{{\mathsf{\beta }}_5\mathrm{L}\mathrm{O}\mathrm{G}\mathrm{E}\mathrm{X}\mathrm{P}}_{\mathrm{i}\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}\mathrm{t}}$$

(1)

Each parameter β in the model denotes the coefficient of the estimated variables, with i representing each panel (country) and t denoting each time series. The parameter ${\mathsf{\beta }}_0$ indicates the constant term, whereas ${\mathrm{u}}_{\mathrm{i}\mathrm{t}}$ refers to the error term, capturing the difference between the actual and predicted values of the model.

The logarithmic conversion of variables not expressed in percentages is very important in the macroeconomic literature to ensure consistency and interpretability of results. Moreover, it significantly reduces data heterogeneity [39]. Thanks to the log transformation, coefficients can be interpreted approximately as percentage changes. All variables employed in this study and their respective data sources were obtained from the World Bank (WDI) [40] database and the Energy Institute—Statistical Review of World Energy [41] reports, and they are presented in detail in

Table A2. Data Description.

Variable	Description	Source
GDPGR	Gross Domestic Product Annual Growth Rate	World Bank
LOGRENEW	Per capita energy consumption from renewables (kWh)	Energy Institute—Statistical Review of World Energy
LOGFOSSIL	Fossil fuels per capita (kWh)	Energy Institute—Statistical Review of World Energy
INF	Inflation, GDP deflator (annual %)	World Bank
POPGR	Annual population growth rate	World Bank
LOGEXP	Exports of goods and services (current US $)	World Bank
LOGCO2	Annual CO2 emissions (per capita)	Energy Institute—Statistical Review of World Energy
LOGPRIMREN	Primary energy consumption from renewables (terawatt-hours)	Energy Institute—Statistical Review of World Energy

.

3.2. Econometric Approach

3.2.1. Cross-Sectional Dependence and Multicollinearity

The issue of multicollinearity arises when the correlation among the variables included in the model is considerably high. Such a situation may cause biased and inconsistent coefficient estimates. Therefore, the correlation levels among the variables were analyzed and reported in

Table A3. Correlation Table.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1) GDPGR	1.000
(2) LOGFOSSIL	−0.359	1.000
(3) LOGRENEW	−0.373	0.479	1.000
(4) LOGEXP	−0.271	0.608	0.501	1.000
(5) INF	−0.044	−0.150	0.091	−0.133	1.000
(6)LOGPRIMREN	0.096	−0.045	0.556	0.414	0.089	1.000
(7) LOGCO2	−0.140	0.607	0.496	0.425	0.021	0.119	1.000
(8) POPGR	0.364	−0.698	−0.338	−0.616	0.168	0.135	−0.528	1.000

. In the literature, correlation coefficients higher than 0.70 are generally considered indicative of a statistically significant correlation, which in turn implies a potential risk of multicollinearity [42].

A correlation analysis was conducted to analyze the linear relationships among the variables included in the model, and the results are presented in Table A3. As indicated by the findings, a moderate correlation was found between the variables LOGEXP and LOGFOSSIL, as well as between POPGR and LOGEXP. Therefore, a Variance Inflation Factor (VIF) analysis was performed to ensure the absence of cross-sectional dependence in the model, and the results are reported in

Table A4. Variance Increase Factor Test.

	VIF	1/VIF
LOGEXP	3.28	0.304
LOGFOSSIL	3.17	0.315
LOGPRIMEN	3.02	0.331
POPGR	2.99	0.335
LOGRENEW	2.58	0.387
LOGCO2	1.85	0.541
INF	1.09	0.915
Mean VIF	2.57

.

The threshold value widely accepted in the literature suggests that a VIF coefficient higher than 5 indicates the presence of multicollinearity. Given the results achieved, the VIF values of the variables included in the model range between 1.09 and 3.28, with an average VIF of 2.57. The fact that all variables remain below the threshold demonstrates that there is no serious multicollinearity problem in the model. This result not only supports the findings of the correlation analysis but also provides an important basis for the reliability of the estimated coefficients.

3.2.2. Descriptive Statistics, Cross-Section Dependence, and Unit Root Tests

The descriptive statistics of the variables included in the model are presented in

Table A5. Descriptive Statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
GDPGR	540	3.317	3.58	−10.94	19.3
LOGFOSSIL	540	9.936	0.991	7.113	11.4
LOGRENEW	540	7.425	1.173	4.396	9.078
LOGEXP	540	26.011	1.587	21.545	28.944
INF	540	26.702	190.572	−2.61	2736.97
LOGPRIMREN	540	5.489	1.24	2.237	8.886
LOGCO2	540	15.256	4.441	−1.1	18.714
POPGR	540	0.763	0.71	−1.854	2.762

. Accordingly, when examining the descriptive statistics, it was observed that the standard deviation values of the variables are relatively low. This indicates that the deviations of the variables from their means are limited and that the dataset exhibits a stable distribution. The fact that there are a total of 540 observations for each variable demonstrates that the panel dataset has a balanced structure. Therefore, the subsequent analyses were conducted within the framework of a balanced panel dataset.

Table A6. Cross-Section Dependency Test Result.

Test	Statistics	Probability Value
LM	460.6	0.000
LM adj.	157.3	0.000
LM CD	17.61	0.000

presents the results of the Breusch-Pagan [43] LM, Lagrange Multiplier, and Pesaran, Ullah, and Yamagata (2008) [44] cross-sectional dependence tests. The null hypothesis of “no cross-sectional dependence” was rejected in all tests, indicating the existence of cross-sectional dependence among the units that constitute the panel. This finding suggests that economic or structural shocks that may arise in one of the countries included in the panel could also have effects on other countries. The identification of cross-sectional dependence requires the use of second-generation unit root tests in the subsequent stages of the analysis.

After determining the presence of cross-sectional dependence among the units through the Pesaran CD test, the Pesaran (2007) [45] CADF unit root test, which is one of the second-generation unit root tests, was applied to the series. The hypotheses of the unit root analysis are presented as follows:

H0: 

The series contains a unit root (The series is non-stationary).

H1: 

The series does not contain a unit root (The series is stationary).

The results of the unit root analysis are presented in

Table A7. Unit Root Test Results.

Fixed		CADF		Trend + Fixed	CADF
		t-Stat	Prob		t-Stat	Prob
GDPGR	Düzey	−2.156	0.098	Düzey	−1.88	0.958
	Birinci Fark	−3.288	0.000	Birinci Fark	−3.413	0.000
LOGRENEW	Düzey	−2.21	0.118	Düzey	−2.658	0.129
	Birinci Fark	−4.437	0.000	Birinci Fark	−4.483	0.000
LOGFOSSIL	Düzey	−1.064	0.993	Düzey	−2.181	0.733
	Birinci Fark	−3.6	0.000	Birinci Fark	−3.653	0.000
INF	Düzey	−1.907	0.331	Düzey	−2.554	0.226
	Birinci Fark	−4.576	0.000	Birinci Fark	−4.647	0.000
POPGR	Düzey	−1.967	0.26	Düzey	−2.721	0.086
	Birinci Fark	−3.206	0.000	Birinci Fark	−3.319	0.000
LOGEXP	Düzey	−1.933	0.3	Düzey	−1.708	0.991
	Birinci Fark	−4.128	0.000	Birinci Fark	−4.356	0.000
LOGPRIMREN	Düzey	−2.174	0.09	Düzey	−2.343	0.511
	Birinci Fark	−3.154	0.000	Birinci Fark	−3.237	0.001
LOGCO2	Düzey	−1.944	0.286	Düzey	−2.502	0.288
	Birinci Fark	−3.401	0.000	Birinci Fark	−3.384	0.000

. The probability values of the variables at their level are statistically higher than the critical value of 0.05, both in the constant and in the constant plus trend specifications. On the other hand, the probability values obtained from the unit root tests of the first differences were found to be lower than the critical value of 0.05. Accordingly, it was observed that all series contain a unit root at level in both the constant and constant + trend specifications but become stationary once their first differences are taken.

The results obtained from the Pesaran and Yamagata (2008) [46] homogeneity test are presented in

Table A8. Homogeneity Test Results.

Tests	Test Statistic	Probability Value
Δ	4.465	0.000
${\mathsf{\Delta }}_{\mathrm{a}\mathrm{d}\mathrm{j}}$	4.957	0.000

. Given the findings, the null hypothesis that the intercept and slope coefficients in the model are homogeneous was rejected at the 5% significance level. This result reveals that the intercept and slope coefficients in the model are heterogeneous. Therefore, conducting the panel cointegration analysis under the heterogeneous slope assumption and employing heterogeneous coefficient estimators consistent with this assumption would be more appropriate.

3.2.3. Panel Cointegration Analysis (Westerlund Test)

In the model, the heterogeneous coefficient structure was taken into account, and the Westerlund (2007) [24] panel cointegration test was applied. Accordingly, the results of the Westerlund (2007) [24] panel cointegration test are presented in

Table A9. Westerlund (2007) [24] Panel Cointegration Test Results.

Statistics	Test Statistic	Z-Value	p-Value	Robust p-Value
${\mathrm{G}}_{\mathsf{\tau }}$	−5.030	−8.072	0.000	0.000
${\mathrm{G}}_{\mathrm{a}}$	−10.912	1.117	0.868	0.10
${\mathrm{P}}_{\mathsf{\tau }}$	−20.337	−11.408	0.000	0.000
${\mathrm{P}}_{\mathrm{a}}$	−16.069	−2.189	0.014	0.000

. The null hypothesis of the test is defined as “no cointegration.” Since heterogeneous slope coefficients were identified in the model, the ${\mathrm{G}}_{\mathsf{\tau }}$ and ${\mathrm{G}}_{\mathrm{a}}$ statistics were adopted [47]. Given the findings, both of these test statistics reject the null hypothesis at the minimum 5% significance level, indicating the existence of a cointegration relationship among the variables. Given the presence of cross-sectional dependence in the model, robust p-values obtained through the bootstrap method were considered in assessing significance.

Furthermore, the probability value obtained for the ${\mathrm{P}}_{\mathrm{a}}$ test statistic was also found to be statistically significant at the 5% level. This result supports the existence of a long-run relationship among the variables. Therefore, the null hypothesis was rejected, and the alternative hypothesis, “there is cointegration among the series,” was accepted.

3.2.4. Long- and Short-Run Coefficient Estimations (CCE and AMG)

To estimate the long-run coefficients for the entire panel in the presence of heterogeneous slope coefficients, the Common Correlated Effects Mean Group (CCE) estimator developed by Pesaran (2006) [48] was employed (see

Table A10. Pesaran CCE Estimator Results.

GDPGR	Coefficient	Std.	err.	z
LGDPGR	−0.043	0.088	−0.490	0.623
LOGRENEW	61.593	26.528	2.320	0.020
LOGFOSSIL	8.371	3.498	2.390	0.017
LOGEXP	4.079	1.097	3.720	0.000
LOGPRIMREN	−62.178	24.477	−2.540	0.011
INF	−0.155	0.097	−1.600	0.109
POPGR	−2.739	1.882	−1.460	0.146
LOGCO2	0.034	0.040	0.860	0.391
M_GDPGR	0.970	0.092	10.580	0.000
L_LGDPGR	0.047	0.090	0.530	0.597
L_LOGRENEW	−74.759	50.920	−1.470	0.142
L_LOGFOSSIL	−6.427	8.182	−0.790	0.432
L_LOGEXP	−4.482	1.955	−2.290	0.022
L_LOGPRIMREN	79.050	48.313	1.640	0.102
L_INF	0.033	0.034	0.990	0.321
L_POPGR	1.846	1.813	1.020	0.309
L_LOGCO2	0.072	0.274	0.260	0.794
_cons	6.581	119.312	0.060	0.956
Possibility	0.000
Total Observations	530
Wald/Chi2	114.27

). In addition, to address potential endogeneity concerns, a one-period lagged value of the dependent variable was included in the model. This specification accounts for unobserved common shocks and cross-sectional dependence across countries while allowing for heterogeneity in slope coefficients.

The results obtained using the CCE estimator developed by Pesaran (2006) [48] effectively control for unobserved common shocks and cross-sectional dependence, while simultaneously accommodating cross country heterogeneity in coefficients. As indicated by the empirical findings: (1) The coefficient of renewable energy consumption (LOGRENEW) is positive (61.59) and significant at the 5% level. In logarithmic terms, this coefficient implies that a 1% increase in renewable energy consumption leads to approximately a 0.62-point increase in GDP growth, thereby supporting the argument that renewable energy use contributes to economic growth. (2) The coefficient of fossil fuel consumption (LOGFOSSIL) is also positive (8.37) and significant at the 5% level, suggesting that a 1% increase in fossil fuel consumption raises economic growth by about 0.08 points. This finding indicates that fossil fuels continued to play an important role in growth during the sample period. (3) The coefficient of exports (LOGEXP) is positive (4.08) and significant at the 1% level, implying that a 1% increase in exports contributes to a roughly 0.04-point increase in growth. This finding lends support to the export-led growth hypothesis. (4) In contrast, the coefficient of primary renewable energy production (LOGPRIMREN) is negative (−62.17) and significant at the 5% level, suggesting that technological deficiencies or efficiency problems in cur-rent production structures may impose constraints on growth. (5) The coefficients of inflation (INF), population growth (POPGR), and carbon dioxide emissions (LOGCO2) were found to be statistically insignificant, indicating that these variables do not exert a notable short-term effect on economic growth.

Among the lagged variables, only the one-period lagged value of LOGEXP was negative (−4.48) and significant at the 5% level. This result suggests that while an increase in exports supports growth in the short run, it may lead to an adverse balancing effect in the subsequent period.

The Augmented Mean Group (AMG) estimator, developed by Eberhardt and Bond (2009) [49] and Eberhardt and Teal (2010) [50], incorporates the dynamic component of common shocks across countries to estimate long-run relationships. The estimation results obtained to analyze the short-run dynamics of the model are presented in

Table A11. The Pesaran Augmented Mean Group (AMG) Forecaster Results.

Dependent Variable: GDPGR	Coef.	Std. Err.	z	P > z
LGDPGR	−0.112	0.060	−1.870	0.062
LOGRENEW	16.109	7.411	2.170	0.030
LOGFOSSIL	10.536	1.893	5.570	0.000
LOGEXP	−0.069	0.912	−0.080	0.939
LOGPRIMREN	−16.822	6.915	−2.430	0.015
INF	−0.080	0.076	−1.050	0.295
POPGR	−0.136	1.404	−0.100	0.923
LOGCO2	−0.013	0.088	−0.150	0.880
_CONS	−129.040	42.002	−3.070	0.002
Possibility	0.000
Total Observations	530
Wald/Chi2	13,670.14

.

Given the analysis results: (1) The coefficient of renewable energy consumption (LOGRENEW) remains positive (16.10) and is statistically significant at the 5% level. Although its magnitude is smaller compared to the CCE estimation, the direction of the relationship remains unchanged. (2) The coefficient of fossil fuel consumption (LOGFOSSIL) is strongly positive (10.53) and significant at the 1% level, indicating that fossil fuels exert a substantial short-run effect on economic growth. (3) The coefficient of primary renewable energy production (LOGPRIMREN) is negative (−16.82) and significant at the 5% level, confirming the finding obtained through the CCE method. (4) By contrast, the coefficients of exports (LOGEXP), inflation (INF), population growth (POPGR), and carbon dioxide emissions (LOGCO2) are statistically insignificant. (5) The coefficient of the lagged dependent variable (common dynamic com-ponent) is high and significant, demonstrating that the growth process exhibits a pronounced persistence (inertia).

Both estimation techniques (CCE and AMG) consistently revealed that renewable energy consumption has a positive effect, primary renewable energy production has a negative effect, and fossil fuel consumption has a positive effect on economic growth. This consistency strengthens the robustness of the findings, showing that they are not method-dependent. Even though the AMG estimates yield differences in coefficient magnitudes, the preserved signs indicate that the overall pattern remains unchanged.

These results suggest that, while renewable energy consumption supports economic growth in the course of energy transition, the current production structure continues to face efficiency challenges, and fossil fuels remain an influential factor in economic performance.

4. Results and Discussion

The results obtained using second-generation panel data methods (CCE and AMG) demonstrated that renewable energy consumption has a positive and significant effect on economic growth. This result aligns with the literature reporting a positive nexus between renewable energy and growth. Studies such as those carried out by Bloch et al. (2015) [51], Apergis and Payne (2010) [52], and Chien & Hu (2007) [53] also showed that renewable energy use supports long-term economic growth. The results achieved in the present study reinforce this strand of literature, highlighting that renewable energy consumption is not only a cornerstone of environmental sustainability but also a critical determinant of macroeconomic performance.

The evidence further revealed that fossil fuel consumption continues to have a positive influence on economic growth. This underscores the fact that fossil fuels still play an important role in the economic structures of the sampled countries and suggests that the transition to renewable energy may impose short-term growth costs. Similarly, studies such as those carried out by Ozcan & Ozturk (2019) [54] and Özer (2023) [55] emphasized the pivotal role of fossil fuels in sustaining economic growth. This result indicates that energy policies should not only focus on expanding renewable energy sources but also manage a gradual and well-structured transition away from fossil fuels.

Interestingly, the study identifies a negative effect of primary renewable energy production on economic growth. This finding may be attributable to technological deficiencies or efficiency problems inherent in current production structures. Moreover, in developing economies, the immaturity of renewable energy technologies may generate productivity shortfalls, thereby imposing constraints on economic performance.

The findings also showed that exports promote long-run economic growth, thereby validating the export-led growth hypothesis. In contrast, inflation, population growth, and CO₂ emissions are found to have no significant impact on economic growth. This suggests that the dynamics of growth are more sensitive to structural factors such as energy consumption and trade rather than to demographic or environmental variables in the long run.

To test the robustness of the findings achieved in this study, multiple methodological strategies were applied. First, the signs and significance levels of the coefficients obtained from the CCE and AMG estimators yielded consistent results, indicating that the findings are not biased by the estimation method used. Second, even when control variables such as exports, inflation, and population growth were included in the model, the positive and significant effect of renewable energy consumption remained robust. Third, the stationarity of the series and heterogeneity within the panel were confirmed through various tests; in particular, the Westerlund (2007) [24] cointegration test validated the presence of long-run relationships. These multi-faceted robustness checks enhance the reliability of the results and provide a solid foundation for policy implications.

The results achieved in this study are consistent with those indicated in studies reporting a positive relationship in Panel A of Table A1 (e.g., Apergis & Payne, 2010 [52]; Bloch et al., 2015 [51]). However, as demonstrated by the studies in Panel B that found a neutral effect, the energy-growth nexus can vary depending on country-specific and temporal differences. Furthermore, consistent with the evidence presented by studies in Panel C that report a negative effect, our finding that efficiency problems in renewable energy investments may constrain growth reinforces this stream of the literature. Therefore, the present study not only corroborates the positive effect but also provides a broader context that helps explain negative findings in the literature.

From a policy perspective, the results demonstrate that enhancing renewable energy consumption is indispensable for sustainable growth, while also highlighting that the energy transition must be supported by complementary investments in technological innovation, grid modernization, and improvements in production efficiency. In this regard, this study not only aligns with the literature but also contributes by offering an integrated framework that reconciles different strands of empirical evidence.

A more detailed comparison with the empirical literature further reinforces the robustness and originality of the current findings. For example, studies conducted in a wide range of contexts such as by Lin and Moubarak (2014) [56] for China, Doğan (2016) [36] for Turkey, and Valadkhani and Nguyen (2019) [57] across 79 countries all reported a positive relationship between renewable energy consumption and economic growth. These results resonate with our study, particularly considering similar periods of energy policy shifts and structural transformation. Conversely, studies such as those by Alam et al. (2012) [58] for Bangladesh and Alshehry and Belloumi (2015) [59] for Saudi Arabia found negative long-run effects, which may stem from technological inefficiencies or underdeveloped energy infrastructures—an explanation consistent with our finding regarding the negative impact of primary renewable energy production. Moreover, mixed or insignificant results reported by Razmi et al. (2019) [60], Ozcan and Ozturk (2019) [54], and Özer (2023) [55] emphasize the importance of country-specific characteristics and methodological choices. The inclusion of countries with diverse socioeconomic backgrounds and the use of second-generation panel methods (CCE and AMG) in our study offer a comprehensive framework that reconciles these inconsistencies. In this way, our work extends the literature by demonstrating how heterogeneous national conditions and energy structures shape the impact of both renewable and non-renewable energy on growth.

5. Conclusions

This study examined the long-run effects of renewable and non-renewable energy consumption on economic growth for the ten countries with the highest renewable energy (REN) consumption over the period 1970–2023. The analysis employed second-generation panel data techniques such as CCE and AMG, taking into account cross-sectional dependence and heterogeneity. The findings indicate that REN consumption has a positive and significant impact on economic growth, whereas primary renewable energy production negatively affects growth, pointing to the constraining role of technological efficiency issues. In addition, fossil fuel consumption continued to contribute to growth during the sample period.

For policymakers, the energy transition should not be driven solely by an increase in REN consumption but must also be supported through modernization of production technologies, strengthening of grid infrastructure, and expanded investments in energy efficiency. Developing efficiency-oriented strategies is crucial to ensure that REN investments contribute to long-term economic growth. Furthermore, since the phase-out of fossil fuels may entail short-term economic costs, it is recommended that this transition be managed gradually and in a balanced manner.

Future studies can be extended by incorporating models that account for technological differences in renewable energy production, sectoral heterogeneity, and institutional quality indicators. Moreover, the effects of variables such as energy prices, foreign investment, and R&D expenditures could be analyzed in a more comprehensive framework.