Renewable energy, non-renewable energy, carbon dioxide emissions and economic growth in selected Mediterranean countries

Abstract

The aim of this paper is to analyze the causality relationship between economic growth, CO₂ emissions and renewable and non-renewable electricity consumption, for a panel of 14 Mediterranean countries over the period 1980–2011. We use a generalized method of moments dynamic model to assess the impact of these three variables on growth and a panel vector error correction model to infer the direction of the causal relationship among the different variables. Empirical evidence shows that there is a bidirectional causal relationship between growth and electricity consumption (renewable and non-renewable) in the short-run and a unidirectional relationship running from renewable electricity consumption and economic growth in the long run. Renewable electricity consumption causes CO₂ emissions in the long-run term and conversely in the short run. The variance decomposition result shown that the renewable electricity consumption is the most important factor to explain economic growth. The finding implies the importance of developing renewable energy in Mediterranean countries to enhance growth and limit pollutant emissions.

1 Introduction

Climate change and global warming may be one of the greatest threats facing the planet. The main factor is the greenhouse gas generated by human activity due to an increase in energy consumption and especially fossils fuels (coal, natural gas, and oil). About 66.5 % of all primary energy and around 66 % of electricity consumption is derived from fossils fuels.¹ Their combustion represents 99.4 % of CO₂ emissions in 2013.

The exponential increase on energy consumption and polluting emissions having adverse effects on environment, we can mention: rising temperatures, drought, floods and altered ecosystems…

Along with the rest of the world, the Mediterranean region is facing several energy and environmental challenges such as a high demand of energy, a large dependence on fossil fuel, an increased volatility of oil prices and the severe negative effects of carbon emissions on the environment. On the one hand, according to the Energy World Organization, energy consumption in the Mediterranean region will be based on oil in 2030; on the other hand, the south shore of the Mediterranean is the most vulnerable regarding global warming due to its dependence on climate sensitive economic sectors (like agriculture, tourism…). So it is crucial to find solutions to deal with this issue. During the last decades, efforts have been undertaken and the Mediterranean countries tried to implement different actions and strategies to resolve the energy and environmental problems and develop renewable energy and energy efficiency. We can mention the Mediterranean Strategy for Sustainable Development (MSSD), the Mediterranean Solar Plan (MSP) and recently the Renewable Energy Solution for the Mediterranean (RES4MED).

Undeniably, the Mediterranean region has a strong production capacity of renewable energy, especially solar and wind. However, electricity generation mix is still predominated by fossil fuels.

The economic debate over the role of energy consumption and economic growth on environment is diversified and we can categorize the studies into three strands: the relationship between pollutants and economic growth, the relationship between energy consumption and economic growth and a last one that focuses on the relationship between energy consumption, polluting emissions and economic growth.

The first one focuses on the relationship between economic growth and polluting emissions. Regarding the empirical studies of Grossman and Krueger (1991), Dinda (2004) and Tsurumani and Managi (2010) which expose the three effects: scale, technique and composition effects. The scale effect has a negative impact on the environment, because it reflects an increase on emissions driven by an increase in production. The technique effect reflects the impact of economic growth on the environment, it can be positive or negative, it depends on the degree of the stringent environmental regulations, which promote a new production method that respects the environment further. The third effect is the composition effect and it is the consequence of the structural transformation of an economy when income grows. The impact can be positive or negative, it depend of the degree of transformation from energy-intensive economy to technology-intensive economy with cleaner activities. The clear effect of these three effects generates the environmental Kuznets curve (EKC).

The second stand of research focuses on the nexus between energy consumption and economic growth. Since the pioneering study by Kraft and Kraft (1978), empirical findings are mixed and sometimes controversial for the same countries (Liz and Van Montfort 2007; Belloumi 2009; Tsani 2010; Omri 2013). The energy policy implications may be significant, depending on the kind of causal relationship between variables. Finally, based on the two types of previous analysis, a new line of research emerged has focused on the relationship between economic growth, CO₂ emissions and energy consumption (Ang 2008; Halicioglu 2009; Arouri et al. 2012). There is no consensus on empirical results. Some authors have also sought to define the role of renewable energy in this relationship, their main objective is to analyze the role of renewable energy to reduce pollutant emissions and enhance growth (Menyah and Wolde-Rufael 2010; Apergis and Payne 2009b; Salim and Rafiq 2012; Shafiei and Salim 2014).

The purpose of this study is to analyze the relationship and the direction of causality between economic growth, carbon emissions, renewable and non-renewable electricity consumption for 14 Mediterranean countries, over the period between 1980 and 2011. The goal is to analyze the effect of renewable and non-renewable electricity consumption and CO₂ emissions on growth using Generalized Method of Moments and the Vectoriel Error Correction Model. The empirical finding can conduct policy decision makers in term of growth development while limiting CO₂ emissions by exploiting the potential on renewable energy in the Mediterranean region.

We organize the rest of the paper as follows. Section 2 describes Materials and methods. Section 3 reports the empirical results. In Sect. 4, we discuss our results. Finally, we conclude by some policy recommendations in Sect. 5.

2 Materials and methods

2.1 Concise literature review

A large number of studies have investigated the causal relationship between energy consumption and economic growth firstly, then that between energy consumption, economic growth and pollutant emissions. A deeper analysis of the first strand allows us to categorize the different results into four hypotheses: growth hypothesis, conservation hypothesis, feedback hypothesis and neutrality hypothesis. The growth hypothesis asserts that the energy consumption can affect the growth process. This hypothesis is supported whether there is a unidirectional causality running from energy consumption to economic growth (Belloumi 2009; Tsani 2010; Ang 2007). In this case, energy plays a vital role on economic growth; and energy conservation policies can have negative impacts. The conservation hypothesis claims that energy conservation policies may not have an adverse impact on economic growth. It is confirmed if there is unidirectional causality running from economic growth to energy consumption (Kraft and Kraft 1978; Liz and Van Montfort 2007). The feedback hypothesis—arguing that energy conservation policies designed to reduce energy consumption may decrease economic growth performance—is confirmed if there is bidirectional causality between energy consumption and growth, which means that these variables are interdependent (Belloumi 2009). Finally, the neutrality hypothesis supports the absence of causal relationship between energy consumption and economic growth. In this case, energy conservation policies will not have any negative impact on economic growth and we can reduce dependency by reducing energy consumption (Soytas et al. 2007).

Table 1 sums up these results.

Table 1 Summary of literature review for energy consumption and economic growth

Recently, growth and renewable energy using nexus is considered as an important search field of research. In the latest decades, several papers have addressed this topic. To do so, they covered many geographic locations and used different econometric tools (see Table 2). There is no clear consensus among researchers regarding the existence of a relationship between economic growth and renewable energy. While some authors confirm the feedback hypothesis (Apergis and Payne 2010a, b, 2011b; Beldirici 2013; Sebri and Ben-Salha 2014), others validate a conservation hypothesis (Menyah and Wolde-Rufael 2010; Apergis and Payne 2011a, b; Ocal and Aslan 2013). The neutrality hypothesis was found by Menegaki (2011) and Tugcu et al. (2012).

Table 2 Summary of literature review for renewable energy consumption and economic growth

The new trend in literature is to divide the effects of renewable and non-renewable energy consumption on economic growth. Soytas et al. (2007) initiated this strand of research. They examined the impact of different sources of energy consumption on industrial output of USA during 2001–2005. They concluded that the non-renewable energy sources are much stronger in explaining the variation of industrial production than the renewable energy sources.

Apergis and Payne (2012) analyzed the relationship between renewable and non-renewable energy and economic growth in 80 countries between 1990 and 2007 within a multivariate panel framework. Their results confirm the existence of a feedback hypothesis for both types of energy. Tugcu et al. (2012) investigated the long-run causal relationship between renewable and non-renewable energy consumption and economic growth in the G7 countries for the period between 1980 and 2009. They used a new causality test developed by Hatemi-J (2012) and found bidirectional causality for all countries.

More studies confirm that energy consumption is a key determinant of CO₂ emissions and the nexus between energy consumption, pollutant emissions and economic growth have been the subject matter of considerable academic research over the past few decades. Ang (2007) initiated this strand of research. There are different and various results (see Table 3). Some studies argue that CO₂ emissions cause growth. For instance, Soytas and Sari (2009), Halicioglu (2009) and Ozturk and Acaravci (2011) who highlighted the existence of such relationship for Turkey. Sebri and Ben-Salha (2014) confirmed a causal relationship for BRICS countries between 1971 and 2010. These results imply that pollutant emissions lead to economic growth and environmental policies can have a negative impact on growth. Other studies found that CO₂ emissions bring about energy consumption, like Halicioglu (2009) for Turkey, Lean and Smyth (2009) for five Asian countries between 1980 and 2006.

Table 3 Summary of literature review for energy consumption, CO₂ emissions and economic growth

Another type of causality exists running energy consumption to CO₂ emissions. In this case, a conservation energy policy can have a positive impact on environment. This relationship was found by Ang (2008) for Malaysia, Arouri et al. (2012) and Omri (2013) for the MENA region countries. Finally, some studies showed that GDP causes CO₂ emissions like Apergis and Payne (2009a) for 6 countries from Central America and Lotfalipour et al. (2010) for Iran.

According to all these studies, we can confirm the non consensual results regarding the sense of causality between the different variables. The difference can be attributed to the differences on period, countries or econometric approach.

Given the importance of energy and environmental policy choices in the Mediterranean region, it is important to try to define the causal relationship among variables and their impact on economic growth in this area.

2.2 Data

Our empirical analysis is based on annual series time data for the period 1980–2011 for 14² Mediterranean countries. Data on electricity renewable and non-renewable consumption (Billion kilowatt hours) are obtained from the International Energy Agency (IEA 2012). The data on real GDP (constant 2005 US$) and CO₂ emissions (metric tons) are taken from World Development Indicator for the World Bank (WDI 2015).

The estimated model is as follows:

$$Y_{it} = \alpha_{i} + \sigma Y_{it - 1} + \beta_{1} {\text{CO}}_{2it} + \beta_{2} {\text{REC}}_{it} + \beta_{3} {\text{NREC}}_{it} + \tau X_{it} + u_{it}$$

(1)

The subscript i = 1…, N denotes the country and t = 1…, T denotes the time period. Y _it is the real GDP, CO_2it denotes CO₂ emissions, REC _it represent the renewable electricity consumption, NREC _it the non-renewable electricity consumption. The coefficients σ, β ₁, β ₂ represent the long-run elasticity for GDP, carbon emissions and renewable and non-renewable electricity consumption, respectively.

We have converted all the series into logarithms. The presence of the lagged dependant variable (Y _it−1) in the model indicates the dynamic nature of real GDP, which explains the interdependent economic growth across period. In this paper, we elaborate two empirical investigations. The first one uses Generalized Method of Moments dynamic (GMM). The second one uses the Vector Error Correction Model (VECM).

2.3 Methods

2.3.1 GMM system

In this paper, we use a Generalized Method of Moments dynamic. GMM System is developed by Arellano and Bond (1991) and Blundell and Bond (1998). Bond et al. (2001) argue this method is able to correct unobserved country heterogeneity, omitted variable bias, measurement error and endogeneity. The GMM system combines the relevant regressions in first and level differences. First differencing correct endogeneity bias via instrumenting the explanatory variables. Instruments for differenced equations are obtained from values of explanatory variables lagged at least twice, and instruments for level equations are lagged differences of the variable. It also checks for unobserved heterogeneity and omitted variable bias. Blundell and Bond (1998) showed that the instruments proposed by Arellano and Bond (1991) are weak; this weakness is due to a low correlation between the proposed instruments and model variables (regressors) in first differences, they call this problem “weak instruments”. To overtake, the authors propose an estimator that combines both, a model level and a model in first differences, as shown in the system below:

$$Y_{it} = \sigma Y_{it - 1} + \beta X_{it} + \alpha_{i} + \varepsilon_{it}$$

$$Y_{it} - Y_{it - 1} = \sigma (Y_{it - 1} - Y_{it - 2} ) + + \beta \left( {X_{it} - X_{it - 1} } \right) + \left( {\varepsilon_{it} - \varepsilon_{it - 1} } \right)$$

The GMM estimators are consistent if the instruments are valid. As suggested by Arellano and Bond (1991) and Blundell and Bond (1998), two specification tests are used: Sargan/Hansen test and AR2 test. Firstly, Sargan/Hansen test is fit for an overall validity of the instruments. The second test examines the null hypothesis that error term ɛ _it of the differenced equation is not serially correlated particularly at the second order (AR2). The results will show what the impact of these variables is on growth.

2.3.2 Panel vector error correction model

The subsequent analysis will be based on a three-stage procedure: integration analysis, cointegration analysis and causality analysis. A first step is to analyze the integration order of variables. Stock and Watson (1989) show that causality tests are very sensitive to the stationarity of series that must be integrated in the same order. In this study, we use four unit root tests: the augmented Dickey–Fuller test (1979) (ADF), the Philips–Perron test (1988) (PP), the Im, Pesaran and Shin test (2003) and Levin, Lin and Chu test (2002).

Once the order of the integration of variables is found to be homogeneous, transition to the second step is possible. In this one, we will examine the presence of cointegration. There are many possible tests of cointegration. However, the most commonly used method is the Johansen cointegration test. This test is based on the autoregressive representation discussed by Johansen (1988) and Johansen and Juselius (1990). This test determines the number of cointegrating equations. Two statistics are proposed here. The first one is based on the trace statistic. The second one is based on the maximum eigenvalue. At this level, we have to find at least one cointegration relation to justify the move to the third stage. The presence of cointegration implies that causality exists between the series, but it does not indicate the direction of causal relationship.

Since the variables are cointegrated, a VECM is set up for investigating short- and long-run causality. We follow the famous procedure of Engle and Granger (1987) to examine the short-run as well as the long-run causal dynamics between GDP, CO₂ emissions, renewable and non-renewable electricity consumption. A Vector Error Correction Model can be written as follows:

$$\Delta {\text{GDP}}_{t} = \alpha_{1} + \mathop \sum \limits_{i = 1}^{p1} \theta_{1i} \Delta {\text{GDP}}_{t - i} + \mathop \sum \limits_{i = 1}^{q1} \varphi_{1i} \Delta {\text{CO}}_{2t - i} + \mathop \sum \limits_{i = 1}^{r1} \sigma_{1i} \Delta {\text{RE}}_{t - i} + \mathop \sum \limits_{i = 1}^{s1} \delta_{1i} \Delta {\text{NRE}}_{t - i} + \omega_{1} {\text{ECT}}_{t - 1} + \varepsilon_{1t}$$

(2)

$$\Delta {\text{CO}}_{2t} = \alpha_{2} + \mathop \sum \limits_{i = 1}^{p2} \theta_{2i} \Delta {\text{GDP}}_{t - 1} + \mathop \sum \limits_{i = 1}^{q2} \varphi_{2i} \Delta {\text{CO}}_{2t - i} + \mathop \sum \limits_{i = 1}^{r2} \sigma_{2i} \Delta {\text{RE}}_{t - i} + \mathop \sum \limits_{i = 1}^{s2} \delta_{2i} \Delta {\text{NRE}}_{t - i} + \omega_{2} {\text{ECT}}_{t - 1} + \varepsilon_{2t}$$

(3)

$$\Delta {\text{RE}}_{t} = \alpha_{3} + \mathop \sum \limits_{i = 1}^{p3} \theta_{3i} \Delta {\text{GDP}}_{t - i} + \mathop \sum \limits_{i = 1}^{q3} \varphi_{3i} \Delta {\text{CO}}_{2t - i} + \mathop \sum \limits_{i = 1}^{r3} \sigma_{3i} \Delta {\text{RE}}_{t - i} + \mathop \sum \limits_{i = 1}^{s3} \delta_{3i} \Delta {\text{NRE}}_{t - i} + \omega_{3} {\text{ECT}}_{t - 1} + \varepsilon_{3t}$$

(4)

$$\Delta {\text{NRE}}_{t} = \alpha_{4} + \mathop \sum \limits_{i = 1}^{p4} \theta_{4i} \Delta {\text{GDP}}_{t - i} + \mathop \sum \limits_{i = 1}^{q4} \varphi_{4i} \Delta {\text{CO}}_{2t - i} + \mathop \sum \limits_{i = 1}^{r4} \sigma_{4i} \Delta {\text{RE}}_{t - i} + \mathop \sum \limits_{i = 1}^{s4} \delta_{4i} \Delta {\text{NRE}}_{t - i} + \omega_{4} {\text{ECT}}_{t - 1} + \varepsilon_{4t}$$

(5)

where α _j , θ _j φ _j σ _j δ _j (j = 1, 2, 3, 4) are parameters to be estimated: ɛ _jt (j = 1, 2, 3, 4) are white noise error terms, ECT is the error correction term derived from the corresponding long-run equilibrium relationship. The coefficient ω _j (j = 1, 2, 3, 4) represents the deviation of the dependent variables from the long-run equilibrium.

Long-run relationships are determined by the coefficients of ECT. The short-run causality is determined by implementing a Wald test of the significance of the lags of each of the explanatory variables in each equation of the VECM.

However, the Granger causality tests do not indicate the interaction between variables. They indicate only the existence of causality among the variables. We use the variance decomposition analysis by applying the Cholesky decomposition technique in VECM like Alshehry and Belloumi (2015). We use also the impulse response function to trace the effects of a shock of one endogenous variable on the other variables.

3 Empirical results

3.1 GMM results

The results of the GMM estimation are reported in Table 4. All variables (renewable electricity consumption, non-renewable electricity consumption and CO₂ emissions) have positive impacts on economic growth.

Table 4 GMM system estimation results

The coefficient of renewable electricity consumption is positive and statistically significant, implying that an increase of 1 % on renewable electricity consumption accelerates economic growth by 0.032 %. The coefficient of non-renewable electricity consumption is also positive and has a statistically significant meaning that 1 % increase on non-renewable electricity consumption enhances economic growth by 0.155 %. Likewise, we find a similar positive impact of CO₂ emissions on economic growth. The coefficient is positive and statistically significant. It indicates that an increase by 1 % on CO₂ emission leads growth by 0.109 %.

3.2 VECM results

Results of the different steps for the VECM estimation are reported in Tables 5, 6 and 7.

Table 5 Panel unit root test

Table 6 Cointegration test

Table 7 VECM Granger causality analysis

Before conducting cointegration and causality analysis, we must test the integration order of variables in question. As we show in Table 5, the four tested uses confirm that all the variables are integrated in first order (I(1)).

The next step is to apply cointegration analysis to examine whether a long-run cointegration relationship exists among these variables. The results are reported in Table 6 and we find two cointegration equations. The existence of the cointegration between series confirms the causal relationship between the variables, but it fails to give the direction. Hence, we follow the procedure of Engle and Granger (1987) to examine the short-run as well as the long-run causal dynamics between the competing variables.

For the long-run causality, we test the significance of the error correction term. We can confirm the causality relationship if the ECT coefficient is negative and statistically significant. The short-run Granger causality is investigated by testing the significance of the sum of lagged of explanatory variables by using the partial F-statistic. The results are reported in Table 7.

Regarding the long-run causality, ECT’s coefficients are negative and statistically significant for the Eqs. (2), (3) and (5) where GDP, CO₂ emissions and non-renewable energy are the dependant variables. These corroborate the established long-run equilibrium relationship for the three variables. These results imply that there is a bidirectional long-run causality between CO₂ emissions and growth, bidirectional long-run causality between non-renewable electricity consumption and growth and a bidirectional long-run causality between carbon emissions and non-renewable electricity consumption. We can also conclude that there is a unidirectional long-run causality running from renewable electricity consumption to growth, to carbon emissions and to non-renewable electricity consumption. For Eq. (4), the ECT coefficient is negative but not significant. So there is no causality running from economic growth, CO₂ emissions and non-renewable electricity consumption to renewable electricity consumption. Figure 1 schematizes the long-run causal relationship between the four series for the selected Mediterranean countries.

Fig. 1

Interaction between variables in long run

The results of short-run Granger causality test presented in Table 7 indicates the existence of a short term bidirectional causality running from CO₂ emissions, renewable and non-renewable electricity consumption to growth. For pollutant emissions, there is unidirectional causality running from CO₂ emissions to renewable and non-renewable electricity consumption. There is no causality in the two directions between renewable and non-renewable electricity consumption. Figure 2 recapitulates the short-run causal relationship between the four series for the panel.

Fig. 2

Interaction between variables in short run

3.3 Results of variance decomposition and impulse response

The results of variance decomposition approach are shown in Table 8. It is shown that 91.36 % of economic growth is explained by its own innovative shocks. The contribution of renewable electricity consumption is equal to 7.38 %. The share of carbon emissions and non-renewable electricity consumption to economic growth are very negligible, 0.189 and 1.068 %, respectively. One standard deviation shock in carbon emissions explains 45.263 % by its own innovative shocks. The contributions of economic growth, renewable and non-renewable electricity consumption to CO₂ emissions are 44.72, 7.895 and 2.114 %. The results also show that 85.161 % of renewable electricity are explained by its own innovative shock whereas the contribution of economic growth, CO₂ emissions and non-renewable electricity consumption to renewable electricity consumption are equal to 11.559, 1.175 and 2.103 %, respectively. The share of economic growth to contribute in non-renewable electricity consumption is important. It represents 54.938 %. A 16.919 % of renewable electricity consumption is explained by one standard deviation shock in non-renewable electricity consumption. A 21.02 % of non-renewable electricity consumption is contributed by its own standard shocks. The share of CO₂ emissions contributes by 7.139 % of non-renewable electricity consumption.

Table 8 Variance decomposition of the four variables

We turn now to the impulse response function. The results are shown in Fig. 3. The response of economic growth and CO₂ emissions increases due to shock on renewable electricity consumption. The response in economic growth and non-renewable electricity consumption increases due to shock on CO₂ emissions. The response on CO₂ emissions and non-renewable electricity consumption increases first then goes down due to shocks stemming in economic growth.

Fig. 3

Responses of variables to generalized one SD innovations

4 Discussion

The obtained empirical results of GMM estimation confirm the role of electricity consumption in the economic process. The non-renewable electricity consumption has greater impact on growth compared to renewable electricity consumption. This is due to the low share of renewable energy; it is around 20 % on the global energy mix for this region. The positive impact of carbon emissions on economic growth can be explained by the importance of the industrial sector of this panel.

From the result of long-run causality, our results are consistent compared to the findings of the different studies in literature. The feedback hypothesis between non-renewable energy and economic growth was found by Tugcu et al. (2012) for G7 countries. This result implies that energy conservation may have a negative impact on economic growth. However, the results of variance decomposition indicate that the share of non-renewable electricity consumption explaining economic growth is minimal. The bidirectional causality between CO₂ emissions and economic growth was found by Salim and Rafiq (2012) and Omri (2013). We can deduce that reducing pollutant emissions can have a negative impact on economic growth in the Mediterranean countries. Also, economic growth can lead CO₂ emissions. However, according to the results of response function, we expect a decreasing linear relationship between GDP and CO₂ emissions. The same result was found by Alshehry and Belloumi (2015) for Saudi Arabia.

Growth hypothesis is also verified and renewable electricity consumption could be a driver for long term economic growth. According to the results of variance decomposition, it seems that renewable electricity consumption is the most important variable in explaining economic growth. The result of response function confirms the positive impact of this variable on growth.

In the short run, the feedback hypothesis finding between renewable energy and economic growth was found by Apergis and Payne (2010a, b), Tugcu et al. (2012), Beldirici (2013) and Sebri and Ben-Salha (2014). This result stipulates that renewable energy is a determinant of growth. The unidirectional causality running from CO₂ emissions to renewable energy was found by Menyah and Wolde-Rufael (2010). We can argue that the global warming caused by CO₂ urges countries to develop renewable energy. In the short run, reducing CO₂ emissions or non-renewable electricity consumption may hamper economic growth in the Mediterranean countries.

5 Conclusion and policy implications

The key focus of this study is to examine the equilibrium and the direction of a causal relationship between economic growth, CO₂ emissions, renewable and non-renewable electricity consumption for 14 Mediterranean countries over 1980–2011. We began by investigating the short-run equilibrium relation. We used GMM system technique proposed by Blundell and Bond (1998). The empirical results of our estimations indicate the existence of a short-run relationship between growth, CO₂ emissions, renewable and non-renewable electricity consumption. The results imply that 1 % increase of CO₂ emissions enhances economic growth by 0.109 %; 1 % increase of non-renewable electricity consumption accelerates economic growth by 0.155 % and finally 1 % increase of renewable electricity consumption extends economic growth by 0.032 %.

It, therefore, appears that renewable energy has a positive impact on economic growth. This one is lower than non-renewable energy. This is mainly due to the low share of renewable energy in the energy mix in the countries studied.

Next, we have applied a panel vector error correction model to analyze the causal dynamic relationship between economic growth, CO₂ emissions, renewable and non-renewable electricity consumption. Our results suggest that electricity consumption enhance economic growth. We have found a feedback hypothesis, in the short run, between economic growth and renewable and non-renewable electricity consumption.

We can, thereby, conclude that electricity consumption is a determinant factor of the GDP growth in these countries. It’s important to take into account the negative effect of energy conservation policies on economic growth. The high level on electricity demand resulted in a high level of economic growth and vice versa.

Analysis of the bidirectional causality between renewable energy and economic growth has generated a very interesting result for this study, indicating that an increase in income could be a core factor driving the development of the renewable energy sector.

The empirical evidence shows bidirectional causality between CO₂ emissions and economic growth. This result implies that the degradation of the environment has an impact on economic growth and a high leveling of economic growth thus leading to polluting emissions. The evidence seems to suggest that to reduce polluting emissions it may sacrifice its economic growth and to develop an alternative option to increase the use of renewable energy.

Our results also show that in the short run, CO₂ cause renewable energy. As a result, an increase in CO₂ emissions, which is the mean cause of global warming, boosts policy makers in order to take measures of scaling down fossil energy consumption and developing more renewable sources. In the long run, we find the opposite result and we can think that the components of renewable energy are still based pollutants.

Some policy implications can emerge from this study. First, these countries should adopt climate policy to reduce CO₂ emissions, such as carbon tax (Zhang 2015) and develop renewable energy to lower energy dependency. Second, efforts must be made to encourage industries to adopt new technologies, offer subsidies to develop electric installations based on renewable energy and reduce subsidies to conventional energy in the southern region of the Mediterranean.

It’s also important to reduce energy demand by adopting energy efficiency policy.