What impacts of renewable energy consumption on CO2 emissions ...

What impacts of renewable energy consumption on C02 emissions and the economic and financial development? A Panel data Vector AutoRegressive (PVAR) approach Montassar Kahia

Mohamed Kadria

Mohamed Sa/ouane Ben Aissa

LAREQUAD & FSEGT, University of



Tunis EI Manar, Tunisia.



[email protected]

[email protected]

[email protected]

Abstract This study employed a PVAR approach due its numerous advantages to investigate the relative effects of renewable energy consumption on C02 emissions and the economic and financial development in 24 MENA countries from 1980 to 2012. Results highlight that the degree of effect of renewable energy use is not very pronounced, which implies that the renewable energy sector in these economies is in its immaturity. This study shows that the investigated countries should improve the renewable energy sector through providing banking loans for investment on green energy projects not only to reduce environmental damage but also to promote economic growth simultaneously. Keyword: Renewable energy consumption, C02 emissions, financial development, economic growth, PVAR.

I. INTRODUCTION Recently, the most serious environmental threat to human society is the potential for considerable changes in the global climate. Climate changes have the potential to disrupt seriously business activity (Sadorsky, [1]). The risk of such changes is extensively approved to be related to emissions of greenhouse gases (OHO), principally carbon dioxide (C02) but also methane and nitrous oxide. The primary source of carbon dioxide emissions is the burning of fossil fuels, which accounted for 87% of the global energy supply in 2012 (Jaforullah and King, [2]). Sathaye et ai. [3] pointed out that the efforts made by the United Nations Framework Convention on Climate Change (UNFCCC) and the Kyoto Protocol are considered insufficient to address the climate change issue. Hence, the challenge is to provide reliable and affordable sources of energy while simultaneously reducing OHO emissions. The promotion of renewable energy is widely advocated as an effective solution to 978-1-4673-9768-1/16/$31.00 ©2016

IEEE

the mitigation of C02 emissions (Tiwari, [4]; Shafiei and Salim, [5]). Tn addition, renewable energy improves the macroeconomic efficiency and, therefore, stimulates economic growth due to the expansion of business and new employment opportunities brought by renewable energy industries (Chien and Hu, [6]). Further than the abovementioned factor, the financial sector may also control energy emissions through stimulating technological progress in the energy sector aimed to reduce emissions (Jensen, [7]). This implies that financial development, which illustrates the actual level of financial resources available for productive purpose and channels funds to projects through banks and stock markets Sadorsky, [8]), can play positive and vital role in combating environmental degradation primarily by curbing CO2 emissions (Shahbaz et aI., [9]). In this context, Tamazian et ai. [10], Tamazian and Rao [11] and Boutabba [12] confinned that environmental pollution decreases with the financial development. Tn addition, renewable energy consumption is shown to be positively related with financial development and revealed bidirectional causality, supporting the feedback hypothesis (Al-mulali et aI., [13]). This indicates that the use of renewable energy sources is expected to enhance the development of the fmancial sector. On the other hand, a good financial development makes it possible to provide greater funding for environmentally friendly projects, which related to cleaner energy at lesser charges. Further, Kumbaroglu et al. [14] confirmed that financial assistance and its consequent technological investments are necessary for steady evolution of the energy sector. Financial development may generally attract Foreign Direct Investment (FDl) and improve research and development (R&D) activities and consecutively accelerate economic activities, and consequently,

influence environmental quality through investment in the eco-friendly related projects (Tamazian and Rao, [11]; Mahdi Ziaei [15]). Briefly, a developed financial sector decreases the cost of borrowing and promotes investment activities (Shahbaz et al., [16]) and also lowers energy emissions by increasing efficiency in the energy sector (Tamazian and Rao, [11]; Tamazian et al., [10]). For all these reasons, exclusion of financial development in the growth- energy consumption-emissions nexus may lead to omission of an important variable in the model. Therefore, in this study, we consider the financial development as an important additional factor in the regression. This can provide more informative policy implications. Further, studies related to the nexus between economic growth, energy consumption, financial development, and CO2 emissions are still in the early stages (for instance, Tamazian et al., [10]; Mahdi Ziaei, [15] and so forth). In this context, Mahdi Ziaei [15] confirmed that the pairwise linkage shows no definitive conclusion for any pairs. However, as far as we know, literature in the field of renewable energy consumption (in disaggregated framework) is relatively less

researched and can be considered as nascent. This article differs from the earlier studies in a number of ways. First, it investigates the relative effect of renewable energy consumption on CO2 emissions and financial and economic development. Second, the existing studies have analyzed the dynamics among the test variables in VAR and/or VECM framework, and group specific of them have performed the panel data methods by employing fixed and/or random effect and/or GMM approach and/or panel cointegration and Granger-causality investigation. Therefore, this research uses a panel data vector autoregression methodology. The main advantage of this method is that it combines the traditional VAR approach, which treats all the variables in the system as endogenous, with the panel data approach, which allows for unobserved individual heterogeneity. At last, this analysis has not been previously performed for the MENA countries. The remainder of the paper is organized as follows. Section 2 will describe the methods. Section 3 will discuss the empirical results. Finally, conclusions and policy implications will be provided in Section 4.

II. METHODS II.1. PVAR methodology: This investigation employs panel data vector autoregressive method developed by Love and Ziccino [17] to examine the effect of renewable energy consumption on CO2 emissions and economic and financial development by modeling the endogenous behaviour between renewable energy consumption, CO2 emission and, economic and financial development as well as determining economically interpretable disturbances. The panel VAR model generally takes the following specification:

Z it = α i + Γ ( L ) Z it + μi + d c ,t + ε it (1) Where Zit represents a vector of endogenous variable (REC, GDP, CO2, FD). REC denotes the natural logarithm of the weight of consumption from renewable energy sources on total energy consumption, GDP is the natural logarithm of real per capita GDP in constant 2000 US dollars, which is the primary and most used growth indicator (Silva et al., [18]; Arouri et al., [19]), CO2 is our proxy for the natural logarithm of carbon dioxide emissions per capita in metric tons (Hossain, [20]). FD means the natural logarithm of the total value of domestic credit to private sector

as a share of GDP, which is used as a proxy for financial development (Charfeddine and Ben Khediri, [21]). α i denotes a matrix of countryspecific fixed effects, subscripts i and t refer country and time, respectively. Γ ( L ) refers to the matrix polynomial in the lag

μi

denotes a

vector of country-specific effects captured in this model, d c ,t refers to the country specific time dummy and ε it is a vector of residuals. Furthermore, Schwarz Information Criterion is used to select the optimal autoregressive order in this model. In applying the VAR model to the panel data framework, restriction requires to be imposed to make sure that the underlying structure is the same for each cross-sectional unit. In order to overcome the restriction on parameters due to its violation in practice, fixed effects denoted by μi are introduced in the model to allow for individual heterogeneity in the levels of all the variables. Nevertheless, conventional mean-differencing procedure that is frequently used to remove the fixed effects might create biased coefficients because the fixed effects are correlated with the regressors due to lags of the dependent variables. One way to avoid this problem is the use of forward

mean-differencing or namely “Helmert procedure” (Arellano and Bover, [22]). Besides, this technique allows the use of the lagged values of regressors as instruments and estimates the coefficients more consistently by system GMM. Further, Sims [23] suggests a contemporaneous recursive causal ordering of variables in the VAR based on their degree of exogeneity. This technique is based on Cholesky decomposition of a variancecovariance matrix of residuals to make sure the orthogonalization of shocks. In our specification, we assume that REC is placed first in the VAR ordering since its current shocks have an effect on the contemporaneous value of GDP, CO2 emissions and FD while GDP, CO2 emissions and FD have an impact on the REC with a lag. GDP is placed second

in the ordering. CO2 emissions and FD are placed last (Silva et al., [18]; Tiwari, [24]). To analyze the impulse response functions (IRFs), the estimation of the confidence intervals for the IRFs is necessary. Since the IRFs are derived from the estimated VAR coefficients and their standard errors, Monte Carlo simulations are used to generate the confidence intervals based on the estimated coefficients and the standard errors. Further, 5th and 95th percentiles of the distribution of the generated coefficients from 1000 bootstraps are employed as the confidence interval for the IRFs, following Love and Zicchino [17] and Boubtane et al. [25]. Additionally, variance decompositions are also presented to demonstrate how essential a shock is in specifying the deviations of variables in the

panel VAR specification by providing the percent of the movement in one variable that is explained by the shock to another variable II.2. Data and econometric analysis

accumulated over time. We report the overall effect accumulated over the 10 years.

Annual data from 1980 to 2012 were obtained from the U.S. Energy Information Administration (EIA) and the World Bank Development Indicators online databases. The MENA countries included in this analysis are Algeria, Lebanon, Israel, Jordan, Bahrain, Egypt, Iran, Malta, Morocco, Tunisia, Turkey, Iraq, Kuwait, Libya, Oman, Qatar, Saudi Arabia, Armenia, Cyprus, Georgia, Mauritania, Syria, United Arab Emirates, and Yemen1. Before going ahead with PVAR model, the first step of the analysis is to look at the data properties of the respective variables. In fact, unit root tests for panel data consist of two generations: First the one which acknowledges the presence of cross-sectional independence and second the one which recognizes the presence of cross-sectional dependence (Barbieri, [27]). Among the first generation are Levin et al. [28], Im et al. [29] and Maddala and Wu [30] while among the second are Pesaran [31] , Smith et al. [32], and Carrion-iSilvestre at al. [33]. However, Panel unit root tests of the first-generation can lead to spurious results (because of size distortions) if significant degrees of positive residual crosssection dependence exist and are ignored (Maddala and Wu, [30]). Hence, testing for the cross-sectional dependence in a panel causality study is crucial for selecting the appropriate estimator (Boubtane et al., [25]). Following Salim and Rafiq [34], to investigate the existence of cross-sectional dependence we employ the cross-sectional dependence test statistic of Pesaran [35]. The results 1 REN21[26]

consistently reject the null hypothesis of crosssection independence, providing evidence of cross-sectional dependence in the panel at 5% level of significance2. Considering the presence of cross-sectional dependence in the panel, we perform second generation panel unit root test of Pesaran [31], Smith et al. [32] and Carrion-i-Silvestre et al. [33] to determine the degree of integration in the respective variables. In addition, in most of the cases, researchers employ panel cointegration procedure derived by Pedroni ([36], [37]. However, the cross-sectional independence assumption proposed by Pedroni ([36], [37]) is highly restrictive (Banarjee and Carrion -iSilvestre [38]). Given the presence of crosssection dependence, the cointegration technique suggested by Westerlund [39] is performed. III. RESULTS AND DISCUSSION Results from the panel unit root tests indicate that all variables involved in this study are stationary in first differences, i.e., they are I (1) variables, while they are non-stationary at levels3. Since the integration order of the variables is the same, we perform in this case the panel cointegration test. The results prove that the null hypothesis of no-cointegration cannot be rejected by all the four tests4. Hence, the empirical properties of the variables need to perform the VAR in first differences, since no cointegration relationships exist between 2

The results are available upon request from authors. The results are available upon request from authors for providing spaces. 4 The results are available upon request from authors. 3

the (non-stationary) respective variables (in level). TABLE 1 below reports the estimated coefficients by system GMM after the fixed effects and the country dummy variables have been removed. In fact, it is obvious from the TABLE 1 that the effect of one and two-year lagged value of REC, GDP, CO2 emissions and FD on the REC is not significant, exception for the two-year lagged value of FD, which significantly has a positive effect on the current value of the REC. Further, evidence proves that only the two-year lagged value of FD has a negative and significant effect on the existing value of the GDP. In addition, the results show that just the one year lagged value of REC and CO2 emissions each has a negative and significant effect on the present value of the CO2 emissions. Last of all, we find that response of FD to one and two years lagged value of FD is positive and significant. TABLE 2. Results of panel VAR model. Response of

Response to

(

D REC(t −1)

( ) D (CO ( ) )

( )

)

D GDP(t −1) 2 t −1

( ) D( REC ) D FD(t −1)

(

(t −2)

D GDP(t − 2 )

(

D CO2(t −2)

(

)

)

( )

DCO2t( )

-0.035 (-0.977)

0.001 (0.581)

-0.021 (-2.802)***

0.006 (0.156)

-0.032 (-0.105)

0.045 (0.422)

0.109 (0.980)

0.056 (0.851)

-0.064 (-0.398)

-0.001 (-0.021)

-0.258 (-2.668)***

0.039 (0.779)

-0.063 (-0.394)

0.055 (1.442)

0.029 (0.552)

0.131 (3.123)***

0.049 (1.476)

-0.002 (-0.110)

-0.012 (-1.346)

0.005 (1.002)

0.093 (0.260)

0.054 (0.776)

0.201 (1.607)

0.017 (0.229)

-0.068 (-0.474)

-0.010 (-0.330)

-0.037 (-0.869)

0.057 (1.057)

DREC(t)

)

DGDP(t)

( )

( )

D FD(t)

0.348 -0.083 -0.021 0.076 (2.060)** (-1.951)* (-0.521) (1.862)* Note: four-variable VAR model is estimated by GMM. Country-time and fixed effects are removed prior to estimation. Reported numbers show the coefficients of regressing the column variables on lags of the row variables. Heteroskedasticity adjusted t-statistics are in parentheses. ***, ** and * indicates significance at 1%, 5% and 10% with critical values of 2.576, 1.96 and 1.645, respectively.

D FD(t −2)

Then, in order to evaluate the two-way effects among renewable energy consumption, real and financial emissions GDP, CO2 development, we implement impulse-response functions of the panel VAR model. Fig.1 plots impulse-response functions together with 5 percent errors bands generated through Monte Carlo simulations with 1000 repetitions. The first row of Fig.1 confirms the finding that the responses of renewable energy consumption to a one standard deviation shock in the remaining variables of the panel VAR (rescaled in terms of shocks of one standard deviation). The figure proves that the response of REC in one standard deviation shock in GDP is negligible whereas the response of REC in one standard deviation shock in CO2 emissions and FD is negative and positive,

respectively. In terms of second row, the response of GDP in one standard deviation shock in REC is positive and has a declining trend. This finding indicates that the renewable energy sector in the MENA countries is in its immaturity. The third row of the figure displays that the response of CO2 emissions in one standard deviation shock in REC and vice versa is negative. More precisely, this implies that increasing use of renewable energy sources is effective to reduce carbon dioxide (CO2) emissions. This finding is consistent with those of Tiwari [24] who confirms the crucial role of renewable energy in curbing CO2 emissions. On the other hand, our results depart sharply from those exposed by Menyah and Wolde-Rufael [40] and Tiwari [4] who conclude that renewable energy consumption does not contribute to reductions in emissions. The findings in the last row confirm that the response of FD in one standard deviation shock in REC and CO2 emissions is mostly positive whereas the response of FD in one standard deviation shock in GDP is negligible. The result of our particular interest involves that the increasing consumption of renewable energy sources lead to improvements in financial development. This helps the developed financial institutions and the capital markets to give an opportunity to lend capital to the renewable energy sector, and also to provide debt and equity financing in funding the green renewable energy projects. This result is consistent with the finding of Al mulali et al. [13] who found bidirectional causality between renewable energy consumption and financial development in both short- and long-run, supporting the feedback hypothesis. Fig. 1. Variable impulse-responses of panel VAR model Impulse-responses for 2 lag VAR of REC GDP CO2 FD (p 5) REC (p 95) REC

REC

(p 5) GDP (p 95) GDP

1.0082

GDP

0.0509

-0.0873

-0.0548 0

(p 5) REC (p 95) REC


REC

(p 5) CO2 (p 95) CO2

GDP

6

s


6

s


0.0090

-0.0315



REC

s

6

-0.0154 0

(p 5) CO2 (p 95) CO2

s

6

6

s

response of CO2 to FD shock (p 5) FD (p 95) FD

CO2

FD

0.1670

-0.0104 0

0

6

s

0.0178

response of FD to GDP shock

FD

0.0199

GDP

-0.0372 0

response of FD to REC shock

6

s

(p 5) FD (p 95) FD

response of CO2 to CO2 shock

0.0136

-0.0164

0

response of GDP to FD shock

CO2

-0.0576 6

s

response of CO2 to GDP shock

0.0115

6

s

(p 5) CO2 (p 95) CO2 0.1435

0

6

s

-0.0187 0

response of GDP to CO2 shock

-0.0133 0

response of CO2 to REC shock

FD

0.0084

GDP

0.0385

6

s

(p 5) FD (p 95) FD

-0.0106 0

response of GDP to GDP shock

REC

0

response of REC to FD shock

CO2

0.0110

-0.0110 0

response of GDP to REC shock

FD

-0.0636 6

s

response of REC to CO2 shock

0.0931

-0.0027

(p 5) FD (p 95) FD 0.1122

0

6

s

response of REC to GDP shock

0.0096

CO2

-0.0449 0

6

s

response of REC to REC shock

(p 5) CO2 (p 95) CO2 0.0282

-0.0006 0

s

response of FD to CO2 shock

6

0

s

response of FD to FD shock

Errors are 5% on each side generated by Monte-Carlo with 1000 reps

To evaluate the relative importance of different

6

structural shocks to endogenous variables by measuring the contributions of shocks on the variance changes of variables, we compute variance decomposition. TABLE 2 below exposes the variance decomposition analysis derived from the orthogonalized impulseresponse coefficient matrices. The variance decomposition detects that REC explains 0.253% of the changes in GDP, 1.97% of the fluctuations of CO2 emissions, and 0.183% of the variation in FD. These variations show that the magnitude of the effect of renewable energy consumption is rather small. This confirms that the renewable energy sector in the MENA countries is in its infancy, and there has been an intense reliance on non-renewable energy sources to meet the rising demand. Further, GDP, CO2 emissions and FD each has a small explanatory power in the model. TABLE 2. Variance decomposition of the panel VAR

D ( REC )

D ( REC )

D (GDP )

D (CO2 )

D ( FD)

0.99552

0.00001

0.00018

0.00429

D (G D P

)

0.00253

0.98814

0.00015

0.00918

D (C O

)

0.01970

0.04659

0.93260

0.00111

0.00183

0.02545

0.00223

0.97049

2

D ( FD

)

Note: The results are based on the orthogonalized impulse-responses. Percent of variation in the row variable (10 periods ahead) explained by column variable.

CONCLUSION IV. IMPLICATIONS

AND

POLICY

This study examines the effect of renewable energy consumption on CO2 emissions and economic and financial development using PVAR approach for 24 MENA countries over the period 1980-2012. The panel VAR investigation employed in this study exposes some attractive findings concerning the dynamic interaction between them. The response of GDP in one standard deviation shock in REC is positive and has a declining REFERENCES

trend. This result generally indicates that the renewable energy sector in the MENA countries is in its first stage of development and requires to reconsider its market structure, which has been stifled by a grouping of constraints, such as the absence of relative legal and policy framework, the high initial capital costs, the rising energy needs and the lack of commercial skills and information. Unfortunately, these barriers have an adverse impact on economic growth in the short-run. Further, the response of CO2 emissions in one standard deviation shock in REC and vice versa is negative. This finding implies that increasing use of renewable energy sources is an effective solution to curb carbon dioxide (CO2) emissions. Ultimately, the response of FD in one standard deviation shock in REC and vice versa is positive. This implies that the renewable energy sector is regarded with increasing interest as an effective means to accelerate the recovery from any financial crisis, while expansion of the financial sector will be likely to offer greater funding and to confer superior financial services at lesser charges for investment in the eco-friendly related projects. Hence, to attain stable and sustainable growth in renewable energy use, governments should design and implement effective support policies to boost investment in new clean energy technologies and to build energy expertise so as to reduce carbon emissions and to promote economic growth simultaneously. Fortunately, the financial system is crucial in this regard. This implies that financial institutions can address the challenges of environment degradation through investing in research and development on innovative technologies relating to cleaner energy. The policy recommendation is hence that financial system should consider the environment characteristic in their current operations. [1] Sadorsky, P., 2011. Some future scenarios for renewable energy. Futures 43, 1091–1104.

[2] Jaforullah, M., King, A., 2015. Does the use of renewable energy sources mitigate CO2 emissions? A reassessment of the US evidence. Energy Economics 49, 711–717.

[5] Shafiei, S., Salim, R.A., Non renewable and renewable energy consumption and CO2 emissions in OECD countries: A comparative analysis. Energy Policy 66, 547– 556.

[3] Sathaye, J., Shukla, P.R., Ravindranath , N.H., 2006. Climate Change, Sustainable Development and India: Global and National Concerns. Current Science India 90, 314–325.

[6] Chien, T., Hu, J.L., 2007. Renewable Energy and Macroeconomic Efficiency of OECD and non-OECD Economies. Energy Policy 35, 3606–3615.

[4] Tiwari, A.K., 2011. Comparative performance of renewable and non renewable energy source on economic growth and CO2 emissions of Europe and Eurasian countries: a PVAR approach. Econ. Bull. 31 (3), 2356– 2372.

[7] Jensen, A.L., 1996. Beverton and Holt life history invariants result from optimal trade off of reproduction and survival. Canadian Journal of Fisheries and Aquatic Sciences 53, 820–822. [8] Sadorsky, P., 2010. The impact of financial development on energy consumption in emerging economies. Energy Policy 38, 2528–2535.

[9] Shahbaz, M., Solarin, S.A., Mahmood, H., Arouri, M., 2013. Does financial development reduce CO2 emissions in Malaysian economy? A time series analysis. Economic Modelling 35, 145–152.

[24] Tiwari, A.K., 2011b. A structural VAR analysis of renewable energy consumption, real GDP and CO2 emissions: evidence from India. Econ. Bull. 31 (2), 1793– 1806.

[10] Tamazian, A., Chousa, J.P., Vadlamannati, K.C., 2009. Does higher economic and financial development lead to environmental degradation: evidence from BRIC countries. Energy Policy 37, 246–253.

[25] Boubtane, E., Coulibaly, D., Rault, C., 2013. Immigration, Growth, and Unemployment: Panel VAR Evidence from OECD Countries. LABOUR, CEIS 27(4), 399-420.

[11] Tamazian, A., Rao, B.B., 2010. Do economic, financial and institutional developments matter for environmental degradation? Evidence from transitional economies. Energy Economics 32 (1), 137–145.

[26] REN21, 2013. Renewable Energy Policy Network for the 21st Century. Renewables 2013 Global Status Report.

[12] Boutabba, M.A., 2014. The impact of financial development, income, energy and trade on carbon emissions: Evidence from the Indian economy. Economic Modelling 40, 33–41.

[27] Barbieri, L., 2010. Panel unit root tests: a review, Università Cattolica del Sacro Cuore, Dipartimento di Scienze Economiche e Sociali-Piacenza Available from: http://www3.unicatt.it/unicattolica/dipartimenti/DISES/alle gati/LBarbie ri43.pdf.

[13] Al-mulali, U., Tang, C.F., Ozturk, I., 2015. Estimating the Environment Kuznets Curve hypothesis: Evidence from Latin America and the Caribbean countries. Renewable and Sustainable Energy Reviews 50, 918–924.

[28] Levin, A., Lin, C.F., Chu, C.S.J., 2002. Unit root tests in panel data: asymptotic and finite sample properties. Journal of Econometrics 108, 1–24.

[14] Kumbaroglu, G., Karali, N., Ar.kan, Y., 2008. CO2, GDP and RET: an aggregate economic equilibrium analysis for Turkey. Energy Policy 36 (7), 2694–2708. [15] Mahdi Ziaei, S., 2014. Effects of financial development indicators on energy consumption and CO2 emission of European, East Asian and Oceania countries. Renewable and Sustainable Energy Reviews 42, 752–759. [16] Shahbaz, M., Shamim, A., Naveed, A., 2010. Macroeconomic environment and financial sector's performance: econometric evidence from three traditional approaches. The IUP Journal of Financial Economics 1, 103–123. [17] Love, Inessa, and L. Zicchino, 2006. Financial Development and Dynamic Investment Behavior: Evidence from Panel VAR. The Quarterly Review of Economics and Finance, 46, 190-210. [18] Silva, S., Soares, I., Pinho, C., 2011. The impact of renewable energy sources on economic growth and CO2 emissions - a SVAR approach. Universidade do Porto, Faculdade de Economia do Porto, 407. [19] Arouri, M.-E., Youssef, A., M.henni, H., Rault, C., 2012. Energy consumption, economicgrowth and CO2 emissions in Middle East and North African countries. Energy Policy 45, 342–349. [20] Hossain, Md.S., 2011. Panel estimation for CO2 emissions, energy consumption, economic growth, trade openness and urbanization of newly industrialized countries. Energy Policy 39, 6991–6999. [21] Charfeddine, L., Ben Khediri, K., 2015. Financial development and environmental quality in UAE: Cointegration with structural breaks. Renewable and Sustainable Energy Reviews , http://dx.doi.org/10.1016/j.rser.2015.07.059i.

[29] Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels. Journal of Econometrics 115, 53–74. [30] Maddala, G., Wu, S., 1999. A comparative study of unit root tests and a new simple test. Oxford Bulletin of Economics and Statistics 61, 631–652. [31] Pesaran, M.H., 2007. A simple panel unit root test in the presence of cross section dependence. Journal of Applied Econometrics 22, 265–312. [32] Smith, V., S. Leybourne, and T.-H. Kim 2004. More powerful panel unit root tests with an application to the mean reversion in real exchange rates. Journal of Applied Econometrics, 19, 147–170. [33] Carrion-i-Silvestre, J.L., del Barrio-Castro, T., LopezBazo, E., 2005. Breaking the panels: an application to GDP per capita. Econometrics Journal 8, 159–175. [34] Salim, R.A., Rafiq, S., 2012. Why do some emerging economies proactively accelerate the adoption of renewable energy? Energy Economics 34, 1051–1057. [35] Pesaran, M.H., 2004. General Diagnostic Tests for Cross Section Dependence in Panels; CESifo Working Paper No. 1229. Cambridge University, United Kingdom. [36] Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and Statistics 61, 653–670. [37] Pedroni, P., 2004. Panel cointegration: asymptotic and finite sample properties of pooled time series tests with an application to the ppp hypothesis: new results. Econometric Theory 20, 597–627. [38] Banarjee, A., Carrion -i- Silvestre, J., 2006. Cointegration in panel data and with breaks and crosssection dependence. European Central Bank. Working paper no. 591.

[22] Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of error component models. Journal of Econometrics 68, 29–51.

[39] Westerlund, Joakim, 2007. Testing for panel cointegration with multiple structural breaks. Oxf. Bull. Econ. Stat. 68, 101–132.

[23] Sims, C.A., 1980. Macroeconomics and Reality. Econometrica 48, 1-48.

[40] Wolde-Rufael, Y., Menyah, K., 2010. Nuclear energy consumption and economic growth in nine developed countries. Energy Econ. 32, 550–556.