Renewable Energy, Urbanization, and CO₂ Emissions: A Global Test

Abstract

A fixed effects regression and two-step system generalized method of moments (GMM) is used to analyze secondary data from the World Bank, covering 163 countries over the period from 2000 to 2016. The study tests the relationship between renewable energy, urbanization, and CO₂ emissions. The empirical results show that urbanization has an inverted U-shaped relationship with CO₂ emissions, while renewable energy consumption mitigates CO₂ emissions. If causal, a 1% increase in renewable energy use leads to a 1.2% decrease in CO₂ emissions. The results also show that the GDP per capita has an inverted U-shaped relationship with CO₂ emissions, confirming the environmental Kuznets curve (EKC). We also found that innovation, proxied by residents’ patents, has a non-linear effect on CO₂ emissions. As a policy implication, developing countries should increase the share of renewable energy in their total energy use, and promote innovative activities by increasing government spending on R&D.

1. Introduction

Over the past decade, research on the correlates and causes of CO₂ emissions has proliferated [1,2]. One can identify two strands of these studies. The first explored the regional drivers of CO₂ emissions; for example, the authors focused on the former Soviet Union countries [3], EU member states [4], developed nations [5], African countries [6], and ASEAN nations [7]. The second strand of research investigated socio-economic predictors of CO₂ emissions for multi- and single-country studies. This stream of research stems from the EKC theory, which postulates that “the process of economic growth is expected eventually to limit the environmental degradation created in the early stages of development” [8], p. 1392.

Scholars suggest that other variables beyond income explain cross-national variations in CO₂ emissions. For example, ample studies show that energy (fossil fuel and renewable) consumption is a paramount antecedent of CO₂ emissions [9,10,11]. While energy is considered to be one of the most significant predictors of CO₂ emissions, “this impact is more severe when accompanied by demographic growth and rural migration into cities, given that population increases and urban development lead to increases in energy consumption and, consequently, to greater atmospheric pollution” Martínez-Zarzoso and Maruotti [12], p. 1344.

Indeed, urbanization intensifies economic activity, which, in turn, affects CO₂ emissions. For example, large-scale movements of populations from rural areas to cities promote the growth of the transportation sector, which is also associated with a larger carbon footprint [13]. A striking example is presented by Maiti and Agrawal [14], p. 277, “due to uncontrolled urbanization in India, environmental degradation has been occurring very rapidly … worsening water quality, excessive air pollution, noise, dust and heat, and the problems of the disposal of solid and hazardous wastes”. Conversely, urbanization leads to energy efficiency in some economic sectors, while increasing energy usage in others. For example, Bilgili et al. [15], using data from 10 Asian countries from 1990 to 2014, explored the link between GDP, urbanization, and energy intensity. The cointegration method and Granger causality test results suggest that urbanization reduces energy intensity and fosters energy productivity. Moreover, urbanization is associated with a greater density of economic agents, which eases the enforcement of environmental regulations, and encourages the use of public transport instead of private vehicles [16].

The aim of this study is to contribute to the extant research on the antecedent of CO₂ emissions by focusing on urbanization and renewable energy consumption. This study extends nascent research by using data from 163 countries from 2000 to 2016. Therefore, our study maximizes the sample size that is available to explore this topic. Our results show that urbanization leads to greater environmental degradation, while renewable energy consumption mitigates CO₂ emissions. We also found an inverted U-shaped relationship between innovation, GDP, and CO₂ emissions.

The rest of the paper is structured as follows: Section 2 reviews the empirical literature, Section 3 discusses the data and model, Section 4 presents the main empirical results, and, finally, Section 5 concludes the study.

2. Literature Review

Several earlier studies used panel and time-series data to analyze the links between CO₂ emissions, economic growth, and renewable energy usage across different countries. However, the obtained results are inconclusive. This chapter mainly reviews only the most relevant studies related to the urbanization–CO₂ emissions nexus, the trade–CO₂ emissions nexus, and the energy consumption–CO₂ emissions nexus.

2.1. Urbanization–CO₂ Emissions Nexus

Worldwide, countries are facing the increasingly negative impacts of climate change and urbanization. Due to uncontrolled and poorly managed city planning, urbanization is the main cause of the degradation of environmental quality. There is extensive literature exploring the link between urbanization and CO₂ emissions in the context of developing and developed countries, by applying multiple cointegration techniques. For instance, in a sample from rapidly urbanized China, Liu [17] studies the relationship between energy use and demographics from 1978 to 2008, using the autoregressive distributed lag (ARDL) cointegration system. The empirical estimates offer evidence that GDP, energy use, and urbanization have significant long-term relationships. A recent study by Hashmi et al. [18] assesses the non-linear association between urbanization paths and CO₂ emissions in selected Asian countries from 1971 to 2014, using conventional stochastic impacts by regression on population, affluence, and technology (STIRPAT) framework. To estimate the long-term impact, robust methods of dynamic panel data methods were applied. The empirical findings show that there is a non-monotonic link between urbanization and CO₂ emissions in the largest cities. It was identified that, in the long term, the demographic transition from rural to urban economies mitigates CO₂ emissions. However, an over-concentration of modernization in the largest cities severely affects the environmental quality. Furthermore, energy intensity and economic expansion have a positive influence on CO₂ emissions, but trade openness has a negative impact. The study suggests the adoption of several eco-friendly policies to improve environmental quality.

In a panel of global regions, Al-mulali et al. [19] analyzed the long-term evidence between urbanization, energy consumption, and greenhouse gas emissions spanning the period from 1980 to 2008. The fully modified ordinary least squares (FMOLS) regression was applied to complete the investigation. According to the findings, while 84% of countries have a favorable long-term link between urbanization, energy use, and carbon emissions, just 16% have inconclusive outcomes. Moreover, some countries have a negative long-term association, whereas others, particularly low-income countries, demonstrate no link between urbanization, energy use, and CO₂ emissions. In another study, Zhang et al. [20] examine the impact of urbanization on environmental degradation, using a sample of more than 140 nations from 1961 to 2011 within the STIRPAT framework. A non-linear relationship was found between urbanization and CO₂ emissions, implying that urbanization initially increased CO₂ emissions, and that innovation and demographic transitions improved environmental quality.

The world’s most populated country, India, was empirically investigated by Franco et al. [21] by testing interrelationships between urbanization, energy use, and emissions, using census data from 1901 to 2011. Urbanization is rapidly increasing in India, along with the consumption of energy and emissions. The research states that India should implement reliable and affordable energy sources to achieve sustainable development. A recent study by Rahman and Alam [22] assessed the causal links between green energy, population, economic growth, and emissions in Bangladesh, using data from 1973 to 2014. The empirical findings state that clean energy consumption improves the quality of the environment, and that urbanization and economic expansion are harmful to the environment.

Recently, Liu et al. [23] researched whether urbanization, renewable energy, and economic progress could make the environment eco-friendlier in selected Asian countries. The research applied panel methodologies to examine links between the variables from 1995 to 2014. According to the long-term estimations, energy consumption and urbanization processes increase CO₂ emissions, while economic growth reduces them. It was highly recommended by the authors that a smart urban system be implemented in the panel of countries, making renewable energy sources more popular. In the case of the EU, Destek et al. [24] assessed the relationship between CO₂ emissions, economic growth, energy use, trade, and urbanization in 10 countries from 1991 to 2011. According to the FMOLS findings, a 1% increase in energy consumption results in a 1.0863% rise in CO₂ emissions, whereas a 1% increase in trade openness results in a 0.0686% decrease in CO₂ emissions. Pata [25] investigated the causal interlinks between GDP, CO₂ emissions, the financial sector, green energy use, and urbanization, using the ARDL estimator for Turkey. According to the ARDL and FMOLS findings, economic expansion, financial development, and urbanization enhance environmental degradation, but overall, renewable energy consumption and alternative energy use have no influence on CO₂ emissions. The results reveal that the major contributors to CO₂ emissions are economic growth, urbanization, and financial development. In the case of newly industrialized countries, Sharif Hossain [26] similarly focused on CO₂, energy use, urbanization, and economic growth from 1971 to 2007. The study applied multiple techniques to conduct the research, and there was a unidirectional, short-term causal relationship between urbanization and economic growth, and between trade openness and urbanization. This suggests that when energy consumption increases in newly industrialized countries, greenhouse gas emissions increase, and the environment becomes more polluted. However, in terms of economic growth and urbanization, environmental quality has been shown to be excellent in the long term.

2.2. Trade–CO₂ Emissions Nexus

Multiple empirical estimates suggest that liberalizing trade openness has a significantly positive effect on CO₂ emissions, but several literature sources also prove negative effects. The positive side is mainly conferred through free trade, in which a country receives increased opportunities to join international markets and import cleaner technologies to reduce carbon emissions. The argument for a negative effect is that trade boosts industrial activity, which eventually raises CO₂ emissions and harms environmental quality. Kasman and Duman [27] assessed the causality between trade openness and CO₂ emissions in new EU states and candidate countries, using panel data analysis from 1992 to 2010. The findings indicated short-term unidirectional panel causation running from energy consumption and trade to the environment. A recent study by Chen et al. [28] investigates the effect of trade openness on greenhouse gas (GHG) emissions, using data covering more than 60 economies located in the Belt and Road regions from 2001 to 2019. The results suggest that greater trade liberalization increases environmental degradation. Furthermore, the indirect effect of trade openness on CO₂ emissions is positive, due to the economic effect, but the indirect effect due to energy use and technological effects is negative. As a result, green energy use must be increased, the energy intensity must be lowered, and regulations must be developed to mitigate climate change effects in accordance with local demands.

Alola et al. [29] find that green energy use improves environmental quality, but that non-renewable energy consumption leads to environmental degradation, in the context of 16 EU member states from 1997 to 2014. In a panel of emerging markets, Hu et al. [30] examine the relationships between green energy, GDP, trade, and CO₂ emissions in 25 economies from 1996 to 2012. The study reports bi-directional causality between renewable energy use, trade, and emissions in the long term. Several other studies also assess the role of trade openness in renewable energy and CO₂ emission interlinks for the top renewable energy countries [31], BRICS [32], China [33], Australia, Canada [34], and Pakistan [35,36].

2.3. Energy–CO₂ Emissions Nexus

Numerous scholars have tested the interlinks between renewable energy, greenhouse gas emissions, and other related variables within the mixed estimation techniques [37,38,39,40]. The link between non-renewable energy and CO₂ emissions was also analyzed in a panel of various countries [41,42,43,44].

Saidi and Omri [45] examine the effect of nuclear energy consumption on CO₂ emissions in a sample of Organisation for Economic Co-operation and Development (OECD) countries from 1990 to 2018. According to the results of time-series regression methods, nuclear energy clearly reduces GHG in the long term. In a panel of major nuclear energy-consuming countries, Al-Mulali [46] investigated the impact of energy use on economic growth and the environment for 30 countries from 1990 to 2010. Based on the findings, energy usage has a favorable effect on GDP growth, but no long-term effect on CO₂ emissions. Moreover, fossil fuels enhance GDP growth, but the advantage of depending on nuclear energy is that there is less damage to the environment. Lee et al. [47] also analyzed the effect of energy demand on CO₂ emissions using data from 18 selected countries. The findings show that a 1% increase in nuclear power over time resulted in a 0.26–0.32% decrease in CO₂ emissions per capita.

Apergis et al. [48] assessed the causal effects between green energy use, environmental degradation, GDP, and other variables in a sample of 19 economies. The conclusions from a Granger test show that renewable energy is a vital variable for achieving sustainable development. Moreover, these findings are also confirmed for a sample of African countries [49,50].

Several of the most recent studies explore the interrelations between energy, CO₂ emissions, and other economic variables. For example, Islam [51] tests the role of institutions in the effect of energy consumption and foreign direct investment (FDI) on CO₂ emissions. Using the ARDL estimator for Bangladesh from 1972 to 2016, the study reports that FDI and innovation mitigate environmental degradation, while economic progress and energy demands have a positive impact on CO₂ emissions. Surprisingly, institutions have a positive effect on CO₂ emissions in both the short- and long term. In a different study, Haldar and Sethi [52] assessed the effect of institutions, FDI, financial development, and other economic variables on CO₂ emissions in a sample of 39 developing economies from 1995 to 2017. The authors used a large set of empirical methods, such as GMM, FMOLS, and Common Correlated Effects Mean Group (CCEMG), to test the robustness of the main results. The regressions show that renewable energy and the improvement in the quality of institutions mitigate CO₂ emissions. Rahman et al. [53] used data for newly industrialized states from 1979 to 2017 to assess the relationship between human capital, energy, and CO₂ emissions. The FMOLS and pooled mean group (PMG) results suggest that economic growth and the accumulation of human capital improve environmental outcomes, while energy consumption increases CO₂ emissions.

3. Materials and Methods

To explore the relationship between renewable energy, urbanization, and CO₂ emissions, we start with the EKC hypothesis. The EKC theory posits that there is a quadratic (inverted U-shaped) relationship between GDP per capita and CO₂ emissions:

$${\mathrm{CO}}_2={\mathrm{a}}_0+{\mathrm{a}}_1\mathrm{GDP}+{\mathrm{a}}_2{\mathrm{GDP}}^2+\mathsf{\epsilon }\mathrm{c}$$

(1)

Next, we augment this model with our variables of interest; thus, the extended specification can be expressed as:

$${\mathrm{CO}}_2={\mathrm{a}}_0+{\mathrm{a}}_1\mathrm{GDP}+{\mathrm{a}}_2{\mathrm{GDP}}^2+{\mathrm{a}}_3\mathrm{RENEWABLES}+{\mathrm{a}}_4\mathrm{URBANIZATION}+\mathsf{\epsilon }$$

(2)

Finally, to mitigate the problem of omitted variable bias, we include a set of control variables comprising trade openness, FDI, population growth, and tourism receipts. The final model can now be expressed as:

$${\mathrm{CO}}_2={\mathrm{a}}_0+{\mathrm{a}}_1\mathrm{GDP}+{\mathrm{a}}_2{\mathrm{GDP}}^2+{\mathrm{a}}_3\mathrm{RENEWABLES}+{\mathrm{a}}_4\mathrm{URBANIZATION}+\mathrm{bX}+\mathsf{\epsilon }$$

(3)

where CO₂ stands for CO₂ emissions per capita, GDP is GDP per capita (a proxy for affluence), RENEWABLES is renewable energy consumption as a percentage of GDP, X is a set of controls, and ε is an error term. We estimate Equation (3) using fixed effects regression and two-step system GMM estimators. The two-step system GMM estimator requires that the number of instruments should not be higher than the corresponding number of countries. In our study, the number of countries is 163, and the number of instruments is 48. Indeed, calculating the coefficients for Equation (3) may be influenced by common panel data issues, including endogeneity, omitted variable bias, and simultaneity. Using the fixed effects method partially considers the issue of the omitted variable bias. More significantly, extant research suggests that the two-step system GMM estimator is a more effective method in cases when N > t, and where there is a need to include a lagged dependent variable. Equation (3) under the two-step GMM model can be presented as:

$${\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}}={\mathsf{\sigma }}_0+{\mathsf{\sigma }}_1{\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}+{\mathsf{\sigma }}_2{\mathrm{R}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\sigma }}_3{\mathrm{U}}_{\mathrm{i},\mathrm{t}}+{\displaystyle \sum }_{\mathrm{h}=1}^{\mathrm{k}}{\mathsf{\gamma }}_{\mathrm{h}}{\mathrm{Z}}_{\mathrm{h},\mathrm{i},\mathrm{t}-\mathsf{\tau }}+{\mathsf{\upsilon }}_{\mathrm{i},\mathrm{t}}$$

(4)

$$\begin{gathered} {\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}}-{\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}={\mathsf{\sigma }}_1\left({\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}-{\mathrm{CO}}_2{}_{\mathrm{i},\mathrm{t}-2\mathsf{\tau }}\right)+{\mathsf{\sigma }}_2\left({\mathrm{R}}_{\mathrm{i},\mathrm{t}}-{\mathrm{R}}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}\right) \\ +{ \mathsf{\sigma }}_3\left({\mathrm{U}}_{\mathrm{i},\mathrm{t}}-{\mathrm{U}}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}\right)+{\displaystyle \sum }_{\mathrm{h}=1}^{\mathrm{k}}{\mathsf{\delta }}_{\mathrm{h}}\left({\mathrm{Z}}_{\mathrm{h},\mathrm{i},\mathrm{t}-\mathsf{\tau }}-{\mathrm{Z}}_{\mathrm{h},\mathrm{i},\mathrm{t}-2\mathsf{\tau }}\right)+\left({\mathsf{\upsilon }}_{\mathrm{i},\mathrm{t}}-{\mathsf{\upsilon }}_{\mathrm{i},\mathrm{t}-\mathsf{\tau }}\right) \end{gathered}$$

(5)

where σ is the parameters to be estimated, τ is the coefficient of auto-regression, U is urbanization, R is renewable energy, Z is the vector of control variables, and υ is the two-way disturbance term. The two-step system GMM is widely used in empirical research to assess the drivers of CO₂ emissions at a cross-country level [54,55,56,57]. The descriptive statistics are reported in

Table 1. Summary statistics.

Variables	Description	Mean	Std. Dev.	Min	Max
CO2	CO2 emissions (metric tons per capita)	4.15	5.27	0	44.53
GDP	GDP per capita, PPP (constant 2017 international currency)	14.94	17.88	0.46	114.21
Urbanization	Urban population (% of total population)	56.81	23.24	8.46	100
Renewables	Renewable energy consumption (% of total final energy consumption)	30.29	29.05	0	98.18
Trade	Trade as % of GDP	76.10	47.64	0.22	538.56
Population	Population growth (annual %)	1.55	1.43	−5.08	16.03
Industry	Industry (including construction), value added (% of GDP)	27.04	11.93	3.65	82.86
FDI	Foreign direct investment, net inflows (% of GDP)	9.20	51.62	−8.96	952.57
Tourism	Tourism receipts as % of GDP	6.26	10.54	0.01	82.71

. The data come from the World Bank.

Figure 1 and Figure 2 plot the visual correlation between renewable energy, urbanization, and CO₂ emissions for 163 countries. As expected, renewable energy is negatively correlated with CO₂ emissions, while urbanization has a positive link with environmental degradation.

4. Results

The results from the fixed effects regression estimator are reported in

Table 2. Main results: Fixed effects regression.

	I	II	III	IV
CO2t−1	0.8014	0.6983	0.6771	0.6517
	(70.40) ***	(53.68) ***	(44.09) ***	(41.30) ***
GDP	0.0037	0.0040	0.0049	0.0060
	(2.74) ***	(2.98) ***	(3.31) ***	(4.06) ***
GDP2	−0.0000	−0.0000	−0.0000	−0.0000
	(3.14) ***	(2.99) ***	(2.86) ***	(3.62) ***
URBANIZATION		0.0026	0.0048	0.0275
		(2.43) **	(3.94) ***	(7.02) ***
URBANIZATION2				−0.0194
				(6.09) ***
RENEWABLES		−0.0084	−0.0116	−0.0110
		(15.02) ***	(17.48) ***	(16.60) ***
TRADE			0.0002	0.0002
			(1.28)	(1.01)
POPULATION			−0.0006	0.0033
			(0.14)	(0.72)
INDUSTRY			0.0030	0.0026
			(4.27) ***	(3.63) ***
FDI			−0.0000	0.0000
			(0.00)	(0.09)
TOURISM			0.0001	0.0002
			(0.25)	(0.45)
Constant	0.0929	0.2789	0.1558	−0.4148
	(4.94) ***	(4.67) ***	(2.03) **	(3.44) ***
R2	0.67	0.70	0.73	0.73
N	2940	2932	2167	2167

** p < 0.05; *** p < 0.01.

. We use the fixed effects regression model to mitigate the impact of the time-invariant omitted variable bias that exists in panel data studies. Model 1 presents the findings from estimating the quadratic relationship between GDP per capita and CO₂ emissions within our global sample. As expected, we observe an inverted U-shaped relationship between GDP and environmental degradation. The turning point is $74,000 internationally. The R-squared suggests that economic development and lagged CO₂ emissions explain 67% of global variations in CO₂ emissions, and the coefficients for GDP per capita and GDP per capita-squared are significant at the 1% level. Figure 3 plots the visual association between GDP per capita and the logged CO₂ emissions.

Model 2 now includes renewable energy and urbanization. The findings show that renewable energy mitigates CO₂ emissions, while urbanization is positively associated with environmental degradation. Model 3 includes the remaining control variables. The coefficient for urbanization is positive and significant, i.e., a 10% increase in urbanization leads to an 3% rise in CO₂ emissions. At the same time, Ahmed et al. [58] suggests an inverted U-shaped relationship between urbanization and CO₂ emissions. Therefore, we included the urbanization-squared term in Model 4. Our results also show that urbanization has a quadratic relationship with CO₂ emissions. This is in line with the modernization theory, which posits high rates of urbanization leading to an improvement in environmental quality, due to structural transformation, population awareness, and innovation. In addition, in line with related research [59], renewable energy mitigates CO₂ emissions, even after considering a rich set of control variables. For example, in the final model, a 1% increase in renewable energy consumption is associated with a 1.1% decrease in CO₂ emissions per capita.

Overall, the results in Table 2 suggest that renewable energy consumption and demographic transformation, proxied by urbanization, are significant predictors of CO₂ emissions.

Table 3. Main results: Two-step GMM results.

	I	II
CO2t−1	0.7160	0.7221
	(19.21) ***	(22.75) ***
GDP	0.0051	0.0084
	(1.73) *	(3.05) ***
GDP2	−0.0000	−0.0000
	(2.35) **	(3.03) ***
URBANIZATION	0.0029	0.0155
	(1.96) *	(3.47) ***
URBANIZATION2		−0.0133
		(3.54) ***
RENEWABLES	−0.0136	−0.0118
	(9.15) ***	(9.53) ***
TRADE	0.0002	0.0004
	(0.73)	(1.75) *
POPULATION	0.0127	0.0008
	(1.48)	(0.10)
INDUSTRY	0.0023	−0.0014
	(1.78) *	(1.15)
FDI	−0.0006	−0.0004
	(4.75) ***	(3.64) ***
TOURISM	−0.0032	−0.0032
	(3.42) ***	(4.51) ***
Constant	0.3631	0.1222
	(3.17) ***	(0.86)
AR 1	0.000	0.000
AR 2	0.316	0.311
Hansen test	0.327	0.088
Fisher statistic		1394.81
Number of countries	163	163
Number of instruments	48	48
N	2167	2167

* p < 0.1; ** p < 0.05; *** p < 0.01.

reports the results from estimating a two-step GMM estimator. The results are in line with the fixed effects regression findings: GDP per capita has an inverted U-shaped relationship with CO₂ emissions. Urbanization is non-linearly related to climate change, and renewable energy reduces CO₂ emissions. The Fisher statistics confirm that, overall, the model is significant. If causal, a 1% increase in renewable energy use leads to a 1.2% decrease in CO₂ emissions. Again, the results further confirm that renewable energy is an important tool for decreasing CO₂ emissions globally.

In

Table 4. Robustness test.

	I	II	III
CO2t−1	0.8159	0.6161	0.8217
	(26.22) ***	(18.66) ***	(31.87) ***
GDP	0.0074	0.0271	0.0134
	(1.69) *	(8.44) ***	(3.26) ***
GDP2	−0.0001	−0.0002	−0.0002
	(2.58) **	(5.21) ***	(3.89) ***
URBANIZATION	0.0060	0.0049	0.0029
	(3.34) ***	(2.71) ***	(1.71) *
RENEWABLES	−0.0063	−0.0101	−0.0051
	(4.94) ***	(6.87) ***	(4.49) ***
TRADE	0.0025	−0.0021	0.0022
	(4.68) ***	(5.58) ***	(4.28) ***
POPULATION	0.0245	0.0300	0.0190
	(2.79) ***	(2.05) **	(2.52) **
INDUSTRY	−0.0019	0.0037	0.0026
	(1.12)	(1.75) *	(1.30)
FDI	−0.0024	0.0007	−0.0019
	(8.30) ***	(2.61) **	(6.73) ***
TOURISM	0.0001	0.0120	0.0026
	(0.02)	(2.09) **	(0.51)
INTERNET	−0.0007
	(1.32)
PATENTS		0.1950
		(5.48) ***
PATENTS2		−0.0128
		(4.99) ***
FREEDOM			0.0170
			(0.60)
Constant	−0.1884	−0.6242	−0.3872
	(1.68) *	(3.63) ***	(1.49)
AR (1)	0.002	0.089	0.002
AR (2)	0.222	0.615	0.493
Hansen p-value	0.051	0.214	0.187
Fisher statistic	2173.59	2351.01	1705.44
Number of countries	185	122	143
Number of instruments	47	51	47
N	825	505	601

* p < 0.1; ** p < 0.05; *** p < 0.01.

, we further test whether the effects of renewable energy and urbanization on CO₂ emissions can withstand the inclusion of additional controls. In Model 1, we include internet users as a percentage of the population, as extant research suggests that ICT is an important antecedent of environmental quality [60]. Following Li et al. [61], we include a squared patent application term to capture the potential non-linear relationship between innovative indicators and CO₂ emissions in Model 2. The results suggest that there is an EKC-type relationship between patents and CO₂ emissions. Overall, the results show that urbanization, innovation, and renewable energy are significantly linked to CO₂ emissions. Our results may also suggest that at lower levels of innovative development, patenting activity may be taking place at industrial levels in energy-intensive economic sectors. However, once an economy reaches a certain level of innovative capacity, further patenting activity leads to energy efficiency and lower levels of CO₂ emissions.

In Model 3, we add the economic freedom index from the Fraser Institute, as a number of studies test the empirical relationship between economic freedom and CO₂ emissions [62,63]. For example, Adesina and Mwamba [64] using data for 24 countries in Africa, show that economic freedom has mixed effects on CO₂ emissions, depending on the level of economic development. In our modeling, economic freedom does not significantly predict CO₂ emissions.

5. Conclusions

This paper employs the dynamic panel of 163 countries from 2000 to 2016, to estimate the link between urbanization, renewable energy, and CO₂ emissions. Using a two-step system GMM estimator and fixed effects regression, we find that renewable energy has a negative effect on GHG, while urbanization increases CO₂ emissions. Moreover, we confirm the existence of the EKC in our sample, and present evidence that innovation is an important antecedent of environmental quality. The validity of our results is confirmed using Fisher statistics, an AR 2 test, and Hansen p-values.

These results have important policy implications. First, it is essential to reconsider the energy mix used by countries to foster a rapid switch to green energy. For example, tools such as grants, low-interest loans, and tax cuts or subsidies may be used to promote the adoption of renewable energy technologies by the private sector and households. The people-private-public partnership is the key element for achieving this mission. The benefits of this partnership are that citizens will be better initiated with the advantages of clean energy solutions, and can make decisions to change their dependency on non-renewable energy sources to renewable ones.

Second, structural transformation reforms are needed by developing countries to foster innovative activities, and to increase GDP per capita beyond the turning point to ensure sustainable development. It is important to increase R&D spending, and to set up educational institutions to increase the number of R&D personnel. Finally, the process of urbanization should be followed by carefully designed urban policies that may decrease the carbon footprint produced by demographic transformations.

Prospective studies should explore the effects of renewable energy, urbanization, and CO₂ emissions for sub-regions, and separately for countries with different levels of economic development. It is also important to assess the role of other variables that may influence the relationship between urbanization and CO₂ emissions, such as institutions or human capital. In addition, a limitation of the study is the exclusion of CO₂ emission mitigation technologies and their contribution to reducing CO₂ emissions. While urbanization is associated with negative environmental outcomes such as CO₂ emissions, it also motivates policymakers to adopt mitigation approach strategies such as carbon storage or efficient use of economic resources.