[email protected]

Energy Economics 43 (2014) 125–139

Contents lists available at ScienceDirect

Energy Economics

j ourna l homepage: www.e lsev ie r .com/ locate /eneco

The causal relationship between renewable electricity generation andGDP growth: A study of energy sources☆

Adrienne Ohler ⁎, Ian FettersDepartment of Economics, Illinois State University, Normal, IL 61790-4200, USA

☆ The authors would like to thank James Payne, NMohammadi for valuable comments and suggestions whAdditionally, we would like to thank Josep Carrion-i-Silvesing the GAUSS code to run the panel stationarity test allow⁎ Corresponding author.

E-mail addresses: [email protected] (A. Ohler), icfette@

http://dx.doi.org/10.1016/j.eneco.2014.02.0090140-9883/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t
a r t i c l e i n f o
Article history:Received 25 February 2013Received in revised form 3 October 2013Accepted 19 February 2014Available online 1 March 2014

JEL classification:C3O5Q2Q3Q4

Keywords:Renewable energyElectricity generationPanelGranger-causalityBiomassWaste energyCross-sectional dependence

This paper examines the causal relationship between economic growth and electricity generation fromrenewable sources (biomass, geothermal, hydroelectric, solar, waste, and wind) across 20 OECD countries over1990 to 2008. The results from a commonly used panel error correction model find (a) a bidirectional relation-ship between aggregate renewable generation and real GDP, (b) biomass, hydroelectricity, waste, and windenergy exhibit a positive long-run relationship with GDP, (c) hydroelectricity and waste generation exhibit ashort-run positive bidirectional relationship with GDP growth, and (d) biomass, hydroelectric, and wasteelectricity generation have the largest impact on real GDP in the long-run. We extend the analysis to considerthe possibility of structural breaks and cross-sectional dependence. Accounting for cross-sectional dependence,we find that in the short-run, increases in biomass and waste generation negatively affect GDP, while aggregaterenewable and hydroelectricity increase GDP. Energy conservation policies will positively impact GDP, if thepolicies cause decreases in biomass or waste energy but increase hydroelectricity and wind energy.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

As energy costs have risen, more scrutiny has been placed on thepotential negative consequences of expanded energy use; however, areduction in energy usage could have unintended consequences foreconomic growth. In order to determine the impact of energy use oneconomic growth, a plethora of literature has looked at the relationshipbetween energy consumption and economic growth. Payne (2010b)

icholas Apergis, and Hassanich helped improve the paper.tre and Ruhul Salim for provid-ing for multiple breaks.

ilstu.edu (I. Fetters).

provides an extensive overviewof this literature, examining 101 studiesover the period 1978 to 2008, but no clear consensus has been found onthe causal nature of this relationship.1

From this literature, a much smaller body of work has emergedexamining a possible relationship between renewable energy and eco-nomic growth. Empirical evidence on the relationship is mixed. Severalstudies find a bidirectional relationship between renewable energyconsumption and economic growth (Apergis and Payne, 2010a,b,2011a,b,d, 2012a,b; Apergis et al., 2010). Sadorsky (2009, 2009b)reports no evidence of a bidirectional relationship in the short-run butfinds a relationship in the long run from real GDP to renewable energyconsumption. Menegaki (2011) fails to find a bidirectional relationship,examining 27 European countries. Payne (2009, 2012) also fails to find

1 Four primary econometric approaches are used to analyze the causal relationship:Granger–Sims causality testing, Engle–Granger/Johanssen–Juselius cointegration anderror-correction modeling, Toda–Yamamoto long-run causality testing, and panelcointegration error correctionmodeling. Of the studies examined, 23.1% showed unidirec-tional causality from energy consumption to GDP growth, 19.5% found causality fromGDPgrowth to energy consumption, 28.2% show a bidirectional relationship, and 29.2% showno relationship.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.eneco.2014.02.009&domain=pdf

http://dx.doi.org/10.1016/j.eneco.2014.02.009

mailto:[email protected]

mailto:[email protected]

Unlabelled image

http://dx.doi.org/10.1016/j.eneco.2014.02.009

Unlabelled image

http://www.sciencedirect.com/science/journal/01409883

3 Net neutral carbon emitters release CO2 as part of the natural carbon cycle of the earth.Biomass products extract CO2 from the air as they grow, and release CO2 when burned.

4 Biomass includes solid biofuels, biogasoline, biodiesels, biogases, and other liquidbiofuels. Municipal waste is defined as waste collected from the residential, commercialand public service sectors, used for the production of heat and power in a central location.

126 A. Ohler, I. Fetters / Energy Economics 43 (2014) 125–139

evidence of causality in theUS. Examining energy use by sector, Bowdenand Payne (2010) find only unidirectional causality from renewableenergy consumption to real GDP, similar to Pirlogea and Cicea (2012)who reports that Romanian renewable energy consumption Granger-causes output.

Furthermore, the role of individual sources is important givencountries' current challenges in determining the optimal mix of en-ergy. Almost no published research exists on the nexus between in-dividual sources of renewable energy and GDP growth. Payne(2011) examines the relationship between biomass consumptionand GDP in the US, and finds a positive unidirectional relationshipfrom biomass to GDP. Ewing et al. (2008) analyze the impact of in-dustrial production and employment on hydroelectricity, solar,wind, wood, and waste energy.

This paper seeks to contribute to the literature on the nexus betweenrenewable energy and GDP growth by examining individual renewablesources, including biomass, geothermal, hydroelectric, solar, waste,and wind. Using electricity generation data on 20 OECD countriesfrom 1990 to 2008, we implement a panel error correction model(ECM) to analyze the causal relationship between real GDP and eachindividual renewable energy source. Following Apergis and Payne(2012b), we utilize a production model framework accounting forcapital and labor. The results find evidence of a bidirectional short-runrelationship between aggregate renewable electricity generation andGDP. We further test for structural breaks in the data and examine thepossibility of cross-sectional dependence (CSD). Extending the panelECM to control for CSD, we find that renewable energy positivelyimpacts GDP, but changes in GDP negatively impacts renewable energy.We implement a similar analysis for each individual source of energy,but an analysis of the interactions between sources is beyond thescope of this paper.

Our results contribute to the literature in several important ways.First, we add to the energy-growth literature by making a distinctionbetween electricity generation, electricity consumption, and energyconsumption.2 We examine renewable electricity generation measuredin MWh rather than energy consumption, because consumption refersto energy delivered to end-use sectors, and because consumption mayor may not include wind and hydroelectric power depending on thedata source and their measurement unit of energy. Electricity genera-tion also differs from electricity consumption by measuring thefirm's production rather than the end user's consumption. Examiningelectricity generation, allows us to disaggregate the data by energysource, and still follow the production framework model. The produc-tion model allows us to overcome some omitted variable bias whileavoiding using ad hoc control variables. Including capital and labor inthe analysis is common in the energy-economic growth and renewableenergy-economic growth literature (Payne, 2010a,b).

We extend the renewable energy-growth literature by examiningindividual sources of energy. Biomass, hydroelectricity, waste, andwind energy sources exhibit a positive long-run equilibrium relationshipwith GDP growth. In the short-run, hydroelectricity exhibits the largestpositive Granger-causing impact on GDP growth; and GDP growth hasthe largest positive impact on biomass, solar, and waste energy.

We further extend the analysis to examine the possibility of struc-tural breaks and account for CSD. The results highlight the importanceof controlling for CSD. While the estimates for biomass, geothermal,and solar are similar to the previous results, the results for aggregaterenewable, hydroelectricity, waste, and wind change. The estimatesfrom the CSD corrected panel ECM find that in the short-run, wasteenergy has a negative impact on GDP growth, and hydroelectricityhas a positive impact. GDP growth still exhibits a positive impact on bio-mass and waste energy, but a negative impact on hydroelectricity.Wind exhibits a positive bidirectional relationship with GDP. Finally,

2 Several studies have examined theelectricity consumption–GDP relationship (Apergisand Payne, 2011; Ozturk, 2010; Payne, 2010).

geothermal exhibits a negative bidirectional relationship with GDPand solar shows a unidirectional relationship from GDP to solarenergy.

The results highlight the need to distinguish between renewablesources. Biomass and waste generation are important drivers in therenewable energy–GDP relationship, but the environmental impactsbetween sources vary. For example, biomass and waste generationemit CO2, nitrogen, and sulfur into the atmosphere, despite beinglabeled as net neutral CO2 emitters.3 According to the EPA, US renew-ables averaged 1.22 lbs of SO2 per MWh and 0.06 lbs of NOx per MWhdespite hydroelectricity, solar, and wind emitting negligible amountsof either. Comparatively, biomass and waste generation emit morecarbon than solar, wind, and geothermal generation. Table 1 comparesthe average annual CO2 emissions of US power plants by primary fuelsources, including natural gas, geothermal generation, the four largestbiomass fuel sources, and three types of coal generation. The averagepower plant emissions are reported in lbs per MWh, and demonstratethat CO2 may increase as biomass generation increases, depending onthe type of biomass used and the source replaced. Municipal solidwaste (MSW) averaged 2993 lbs of CO2 per MWh, emitting more CO2

than natural gas, subbituminous coal, and lignite coal. If MSW energyreplaces coal, lignite, or subbituminous coal, CO2 emissions wouldincrease.4

The remainder of the paper is organized as follows: Section 2presents the unit root test results, examines cointegration betweenthe variables from a production model framework, and providesestimates from a fully modified OLS model to examine the long-runrelationship between renewable energy and GDP. Section 3 presentsthe results of a panel ECM for aggregate renewable electricity gener-ation to test for a causal relationship between renewable energy andGDP.We then implement the same econometric model to analyze in-dividual sources of electricity generation. Section 4 extends the pre-vious analysis by examining the data for structural breaks and CSD.We then implement the panel ECM controlling for CSD to analyzethe individual sources. Section 5 concludes with a discussion of pol-icy implications.

2. Data, unit roots, and tests for cointegration

Data were collected from the International Energy Agency's dataseton world renewable and waste energy statistics. We examine grosselectricity production (GWh) by energy source for 20 countries from1990 to 2008.5 Real GDP, gross fixed capital formation, and size of thelabor force were collected from the OECD.

Renewable electricity generation includes biomass, hydroelectric,geothermal, solar, waste, andwind. Table 2 presents the average annualgrowth rate in generation for each country. Only six countries utilizegeothermal and solar energy over the entire time period considered.Most countries utilize biomass, hydro, and waste energy, with biomassandwaste energy contributing tomost of the renewable energy growth.When utilized, wind and solar energy exhibit the largest growth rates.Comparing the GDP growth rate to the growth in renewable energy,no discernible trend appears. The top 5 countries in aggregate renew-able energy growth, Denmark, Netherlands, Belgium, Portugal, andGermany, are ranked 7th, 11th, 5th, 16th, and 3rd in terms of real GDPgrowth.

We examine only the renewable fraction of municipal waste.5 Countries included: Australia, Austria, Belgium, Canada, Denmark, France, Germany,

Iceland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain,Sweden, Switzerland, United Kingdom, and United States.

Table 1Carbon emission statistics from US power plants by primary fuel source (lbs. per MWh).Source: EPA eGRID2012 Version 1.0 Plant File (Year 2009 Data) available at http://www.epa.gov/cleanenergy/energy-resources/egrid/faq.html.

Coal (bit.) Coal (lig.) Coal(subbit.)

Coal–biomass(bit.)

Biomass(black liquor)

Biomass(landfill gas)

Biomass(municipal solid)

Biomass (woodand wood waste)

Geothermal Naturalgas

Total net generation(GWh in 2009)

861,743 79,112 830,321 1731 23,668 7606 14,450 16,389 15,009 20,284

Mean 3339.09 2423.93 2312.50 566.69 184.55 13.05 2992.68 112.18 38.53 900.35Median 2094.15 2358.53 2268.68 560.60 134.92 0.00 3320.02 0.00 60.00 423.40St. dev. 18,130.53 211.16 872.79 45.76 162.96 64.13 1095.38 278.01 35.62 1050.67Minimum 686.84 2093.42 880.37 504.50 0.00 0.00 730.80 0.00 0.00 29.00Maximum 323,729.63 2919.20 10,014.27 611.53 851.53 490.16 5688.32 2029.73 88.80 3559.61

Hydro, solar, nuclear, and wind have been excluded from comparison. Their CO2 emissions output are negligible. As an extreme outlier, SouthernMinnesota Beet Sugar power plant wasexcluded to maintain consistent and comparable results. For coal–biomass generation, the percent of generation from biomass is approximately 50% for power plants in this category.

127A. Ohler, I. Fetters / Energy Economics 43 (2014) 125–139

Four potential hypotheses for the energy–GDP relationship havebeen outlined in the literature as the growth, conservation, neutrality,and feedback hypotheses. We apply these hypotheses to renewableelectricity generation such that the growth hypothesis implies causalityfrom renewable electricity generation to economic growth, and issupported if an increase in generation positively impacts real GDP. Theconservation hypothesis suggests that causality occurs from growth toelectricity generation, and an increase in economic growth positivelyaffects generation. If no relationship is found, then the neutralityhypothesis is supported. Finally, the feedback hypothesis states thatrenewable generation and economic growth are complementary andinterdependent. This hypothesis is supported if the data confirm a bidi-rectional positive relationship between GDP growth and renewablegeneration.

Applying a production function framework to the renewableenergy–GDP relationship, let Yit = f(Kit, Lit, REit, NREit) where Y, K, andL represent output, capital, and labor for country i in year t. Capitaland labor are included in themodel as ameans for controlling for poten-tial omitted variable bias (Lutkepohl, 1982). This variant of the Solowgrowth model considers electricity as a factor of production, separatedinto renewable (RE) and nonrenewable (NRE) generation. We includenonrenewable generation as a means to differentiate the impact differ-ent energy sources have on GDP. Accounting for unobserved variationand measurement errors, the linear econometric model becomes

Yit ¼ β1i þ β2itþ β3REit þ β4NREit þ β5Kit þ β6Lit þ εit ð1Þ

Table 2Average annual growth rate from 1990 to 2008.

Real GDP Total RE Biomass Ge

Australia 0.063 1.355 10.997 –

Austria 0.042 1.920 8.896 –

Belgium 0.041 13.421 24.854 –

Canada 0.049 1.442 4.950 –

Denmark 0.042 15.702 17.070 –

France 0.035 2.163 3.341 –

Germany 0.032 9.620 27.303 –

Iceland 0.081 7.824 – 17Italy 0.046 3.385 41.152 3.Japan 0.006 0.388 2.913 3.Luxembourg 0.074 8.711 – –

Netherlands 0.051 15.254 24.624 –

New Zealand 0.052 0.760 3.407 3.Norway 0.073 1.463 2.452 –

Portugal 0.066 10.271 4.897 54Spain 0.071 8.441 11.106 –

Sweden 0.045 1.611 10.144 –

Switzerland 0.029 1.719 5.650 –

United Kingdom 0.053 8.557 18.285 –

United States 0.052 0.808 −0.709 0.

where each variable is taken in logarithmic form, and the estimatedcoefficients represent elasticities. Output, Yit, measured as real GDP, isa linear function of renewable and nonrenewable electricity generation,real gross fixed capital formation, and the labor force. Country fixedeffects and a time trend are included in the model with the parametersβ1i and β2i.

We extend the model to analyze six individual sources of renewableelectricity generation, utilizing the production model proposed inEq. (1). RE is replaced with each individual source, and NRE with thecomplement to that source. For example, the long-run econometricmodel for biomass is written as

Yit ¼ β1i þ β2itþ β3REit þ β4NREit þ β5Kit þ β6Lit þ εit ð2Þ

where non-biomass generation is calculated as total generation minusbiomass generation. We include the complement variable as a meansto differentiate the impact different energy sources have on GDP,while minimizing the loss of degrees of freedom in the estimatedmodel. The five other renewable sources follow a similar model.

2.1. Unit root tests

We implement several unit root tests for each of the variables.Tables 3–5 present the results for six different tests: Levin et al. (2002),Breitung (2001), Breitung andDas (2005), Im et al. (2003), two differentFisher-type tests with an augmented Dickey–Fuller specification and aPhillips–Perron specification (Choi, 2001; Hadri, 2000). For all tests,

othermal Hydro Solar Waste Wind

−0.823 – – –

1.419 – 32.929 –

4.281 – 5.717 32.1561.340 – 2.121 –

1.593 – 25.755 15.5881.498 – 13.474 –

2.249 71.202 7.980 48.781.980 6.658 – – –

163 2.072 38.051 27.067 70.135350 −0.067 – 6.020 –

2.294 – 4.270 –

3.336 – 5.623 28.383853 0.160 – – –

1.459 – 8.058 –

.204 9.786 36.389 – 72.3594.431 63.332 18.143 73.6480.934 – 27.857 41.5341.612 21.877 6.258 –

2.518 – 14.424 63.5730190 0.586 3.185 3.442 18.966

http://www.epa.gov/cleanenergy/energy-resources/egrid/faq.html

Table 3Panel unit root test results with aggregate renewable measure.

Levin–Lin–Chu adjusted t Breitung λ Im–Pesaran–Shin Fisher–ADF inverse Fisher–PP inverse χ2 — p Hadri LM z

Y 1.437 3.837 1.444 98.854a 63.957a 17.048a

(0.925) (0.999) (0.926) (0.000) (0.009) (0.000)ΔY −4.433a −0.638 −4.874a 153.903a 139.741a 2.429a

(0.000) (0.262) (0.000) (0.000) (0.000) (0.008)RE 2.414 −0.010 −0.772 70.944a 153.525a 7.274a

(0.992) (0.4960) (0.220) (0.002) (0.000) (0.000)ΔRE −5.249a −6.3610a −6.416a 77.688a 530.891a −2.297

(0.000) (0.000) (0.000) (0.000) (0.000) (0.989)NRE 2.067 −0.5630 0.672 36.392 71.208a 10.205a

(0.981) (0.2870) (0.749) (0.633) (0.002) (0.000)ΔNRE −2.835a −8.763a −4.848a 139.296a 371.079a −0.597

(0.002) (0.000) (0.000) (0.000) (0.000) (0.725)K 2.383 1.9150 −1.135 60.416b 70.890a 13.832a

(0.991) (0.972) (0.128) (0.020) (0.002) (0.000)ΔK −3.379a −1.725b −2.452a 84.715a 156.340a 3.792a

(0.000) (0.042) (0.007) (0.000) (0.000) (0.000)L 5.025 4.256 0.887c 46.782 111.578a 19.275a

(1.000) (1.000) (0.813) (0.214) (0.000) (0.000)ΔL 5.810 −3.0750a −2.487a 56.199b 270.269a 5.672a

(1.000) (0.001) (0.006) (0.046) (0.000) (0.001)

a Statistical significance at .01 level.b Statistical significance at .05 level.c Statistical significance at .10 level.


the null hypothesis is non-stationarity, except the Hadri LagrangeMultiplier test, which assumes that all panels are trend stationary.Each test has advantages. The Levin–Lin–Chu and Breitung tests assumea common root across the countries, and Breitung has the highestpower and smallest size distortion. The Im–Pesaran–Shin test allowsfor an individual root for each country. The null hypothesis for thistest is that all panels have a unit root, and the alternative is that a frac-tion of the panels are stationary. The Fisher tests also allow for as muchheterogeneity across countries as possible, conducting unit root tests for

Table 4Unit root test results by renewable source.

Levin–Lin–Chuadjusted t

Breitung λ Im–Pesaran–Shin Fisher–ADF inverse Fisher–PInverse χ

Biomass 3.897(1.000)

1.742(0.959)

1.031(0.849)

27.932(0.830)

99.628a

(0.000)ΔBio 0.859

(0.805)−5.388a

(0.000)−3.413a

(0.000)81.384a

(0.000)327.972a

(0.000)Geotherm −6.415a

(0.000)0.685(0.753)

−1.560b

(0.059)10.907(0.537)

6.591(0.883)

ΔGeo −3.022a

(0.001)−2.960a

(0.002)−2.879a

(0.002)33.201a

(0.001)41.567a

(0.000)Hydro 0.595

(0.724)−1.723c

(0.042)−1.511b

(0.065)48.333(0.172)

102.735a

(0.000)ΔHydro −5.405a

(0.000)−5.644a

(0.000)−5.616a

(0.000)63.853a

(0.010)510.312a

(0.000)Solar 4.314

(1.000)4.410(1.000)

3.580(1.000)

1.993(0.999)

79.188a

(0.000)ΔSolar −4.511a

(0.000)0.810(0.791)

−2.486a

(0.007)5.597(0.935)

96.456a

(0.000)Waste −3.192a

(0.001)2.227(0.987)

−0.962(0.168)

25.718(0.776)

104.974a

(0.000)ΔWaste −1.101

(0.136)−3.079a

(0.001)−6.515a

(0.000)100.663a

(0.000)270.049a

(0.000)Wind 1.968

(0.975)2.803(0.998)

1.643(0.950)

12.514(0.897)

39.756a

(0.005)ΔWind −3.176a

(0.001)−4.925a

(0.000)−3.611a

(0.000)63.578a

(0.000)137.613a

(0.000)


each country and combining the test results. Finally, Hadri conducts atest with the null hypothesis that the data contain a unit root. For alltests, we include a country intercept and a trend.

Table 3 reports the unit root test results for real GDP, capital, labor,renewable energy, and nonrenewable energy. Most of the tests confirmthat all variables are non-stationary in level and stationary in first differ-ence. Most notably, the Im–Pesaran–Shin test result finds evidence thateach variable is integrated of order one, which is the test commonlyfound in the renewable energy–GDP literature.

P2 — p

Hadri LM z Countries included

13.231a

(0.000)Australia, Austria, Belgium, Canada, Denmark, France, Germany,Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain,Sweden, Switzerland, United Kingdom, United States.0.381

(0.352)8.849a

(0.000)Iceland, Italy, Japan, New Zealand, Portugal, United States.

0.371(0.356)5.940a

(0.000)All

−2.568(0.995)8.845a

(0.000)Germany, Italy, Portugal, Spain, Switzerland, United States.

3.233a

(0.001)16.782a

(0.000)Austria, Belgium, Canada, Denmark, France, Germany, Italy, Japan,Luxembourg, Netherlands, Norway, Spain, Sweden, Switzerland,United Kingdom, United States.2.114c

(0.017)17.493a

(0.000)Belgium, Denmark, Germany, Italy, Netherlands, Portugal, Spain,Sweden, United Kingdom, United States.

2.574a

(0.005)

Table 5Unit root test results for renewable complement variables by source.

Levin–Lin–Chu adjusted t Breitung λ Im–Pesaran–Shin Fisher–ADF inverse Fisher–PP inverse χ2 — p Hadri LM z

Non-biomass 5.551 0.730 2.252 21.060 63.031a 13.167a

(1.000) (0.7673) (0.988) (0.9776) (0.0035) (0.000)ΔNon-bio 13.167a −7.587a −4.972a 55.642b 388.594a 0.245

(0.000) (0.000) (0.000) (0.0194) (0.000) (0.403)Non-geotherm 4.628 3.181 2.189 2.094 4.952 7.779a

(1.000) (0.999) (0.986) (0.999) (0.960) (0.000)ΔNon-geo −1.608c 0.761 −1.673b 24.135b 98.0151a 0.068

(0.054) (0.777) (0.0047) (0.020) (0.000) (0.473)Non-hydro 0.1165 −0.993 −0.297 53.767c 84.184a 10.749a

(0.546) (0.161) (0.383) (0.072) (0.000) (0.000)ΔNon-hydro −2.99a −7.538a −4.619a 60.507b 384.212a −0.219

(0.001) (0.000) (0.000) (0.019) (0.000) (0.587)Non-solar 0.3203 0.053 0.642 1.594 20.561c 8.425a

(0.626) (0.521) (0.740) (0.999) (0.057) (0.000)ΔNon-solar −2.410a −4.929a −1.295c 3.948 126.645a 1.569c

(0.008) (0.000) (0.098) (0.984) (0.000) (0.058)Non-waste 1.541 2.44 −2.215 27.492 70.988a 11.495a

(0.938) (0.596) (0.013) (0.694) (0.000) (0.000)ΔNon-waste −4.454a −6.714a −7.431a 117.310a( 300.679a 0.104

(0.000) (0.001) (0.000) 0.000) (0.000) (0.458)Non-wind 1.385 0.386 0.833 6.604 47.494a 9.289a

(0.917) (0.650) (0.798) (0.998) (0.001) (0.000)ΔNon-wind −10.045a −5.917a −2.482a 20.636 228.598a 1.139

(0.001) (0.000) (0.007) (0.419) (0.000) (0.127)

These variables are calculated in a manner similar to nonrenewable generation. For example, non-biomass generation is calculated as total generation minus biomass generation.a Statistical significance at .01 level.b Statistical significance at .05 level.c Statistical significance at .10 level.


Table 4 presents the unit root tests for the individual sources ofenergy. Depending on the test, a case could be made that each of thesources are integrated of order one. Biomass passes 4 tests, Breitung,Im–Pesaran–Shin, Fisher–ADF, and Hadri. Geothermal and wind energyalso pass 4 tests, including Breitung and Fisher–ADF. Hydro passes 3tests, Levin–Lin–Chu, Fisher–ADF, and Hadri. Solar and waste pass twoand three tests, respectively. The Im–Pesaran–Shin test suggests thatall sources are integrated of order onewith the exception of geothermaland hydroelectricity. The Hadri test suggests that these two sources areintegrated of order one.

We construct controls for all other electricity generation. Non-biomass, non-geothermal, non-hydro, non-solar, non-waste, andnon-wind generation account for causality from other sources. Thesecomplement variables are calculated in a manner similar to non-renewable generation, but also contain other renewable sources. TheLevin–Lin–Chu and Im–Pesaran–Shin test results demonstrate that thecomplement variables are unambiguously integrated of order one, aspresented in Table 5.

Table 6Panel cointegration tests by source.

Aggregate renewable Biomass Geotherm

Within dimensionPanel v-stat 10.805a 15.30a 3.871a

Panel rho-stat 3.847 3.027 1.995Panel pp-stat −0.343 −2.270b 0.125Panel adf-stat −1.874b −2.746a 0.049

Between dimensionGroup rho-stat 4.820 4.041 2.310Group pp-stat −2.830a −4.438a −1.149Group adf-stat −3.151a −3.434a −1.035c

Tests assume an individual intercept and individual trend. Onlyweight statistics are reported forCriteria. The null hypothesis is no cointegration.


2.2. Tests for cointegration

To test for cointegration, we implement seven panel cointegrationtests proposed by Pedroni (1999, 2004) which account for heteroge-neous panels. The tests check for a unit root process on the residualssuch that εit = ρiεit − 1 + wit in Eq. (1). The null hypothesis of nocointegration, ρi = 1, is rejected if the cointegration term is statisticallysignificant. Table 6 presents the results for both panel and group tests.Panel tests require a common autoregressive coefficient, and grouptests allow for individual autoregressive coefficients.

Depending on the test, a case could be made that a cointegratingrelationship exists between each renewable energy source, real GDP,capital, and labor. For aggregate renewable electricity generation, fourof the seven tests reject the null hypothesis of no cointegration at the5% significance level, indicating a long-run relationship between output,capital, labor, and renewable and nonrenewable generation. In column3, five of the seven tests indicate a long-run relationship between out-put, capital, labor, biomass, and non-biomass generation. Biomass,

al Hydro Solar Waste Wind

11.886a 6.640a 10.264a 14.051a

3.748 0.878 3.240 1.349−0.235 −1.673b −0.745a −3.855a

−1.095b 0.578 −2.734a −2.014b

4.763 2.063 4.161 2.457−3.453a −0.953 −6.739a −6.376a

−2.692a −0.799 −4.235a −3.369a

thewithin-dimension tests. Automatic lag length determined by the Schwarz Information

Table 7Dynamic OLS and fully modified OLS long-run estimates.

DOLS FMOLS

Log(Renew) 0.012 0.149a

(1.06) (9.97)b

Log(Non-renew) 0.013c 0.329a

(1.33) (10.41)Log(Capital) 0.186a 0.382a

(11.00) (14.60)Log(Labor) 0.288a 1.10a

(5.11) (12.55)

Long-run estimates from Eqs. (3a)–(3e) are provided in this table. t-statistic reported inthe parenthesis. N = 380.



hydro, waste, and wind energy, all indicate cointergration at the 5%significance level for the tests that require a common root across coun-tries, and for the tests that allow countries to have an individualautoregressive process. Geothermal only rejects the null hypothesisfor one test, and solar energy rejects the null hypothesis of nocointegration for two within tests. Thus, we proceed to examine thelong-run equilibrium relationship for all sources, but note that geother-mal and solar energy only exhibit a weak cointegrating relationshipwith real GDP.

2.3. Long-run equilibrium parameters

To examine the parameters of the long-run relationship betweenrenewable generation and economic growth, we estimate Eq. (1)using two methods: dynamic OLS, and fully modified OLS. DOLS worksto reduce the bias from endogeneity and serial correlation found inOLS estimations. Kao and Chiang (2000) also found that DOLS outper-forms OLS estimators for panel data. The fully modified OLS (FMOLS)technique proposed by Pedroni (2001) is reported for case comparisonwith previous studies.

Estimation results for Eq. (1) are displayed in Table 7. All coefficientsare positive and statistically significant, andmost notably the coefficientfor renewable electricity. The DOLS results predict that a 1% increasein renewable generation increases real GDP by 0.012%. The FMOLSestimates an elasticity for renewable generation of 0.149. Apergis andPayne (2011b) report slightly higher elasticities for developed countries0.265 and developing countries 0.429. Apergis and Payne (2011c)report an elasticity of 0.074 for emerging markets.6

The FMOLS estimates are similar to those previously found in theliterature modeling the renewable energy-economic growth nexus ina production framework. Our estimated elasticity for renewable energyof 0.149 ismuch lower thanApergis and Payne (2012b) at 0.371, but thenon-renewable estimates are comparable at 0.329 and 0.384. The esti-mates for capital elasticity are also similar at 0.382 and 0.388, and thelabor elasticities differ at 1.10 and 0.812. Our estimates are also compa-rable to Apergis and Payne (2011b) who report elasticities of 0.265 forRE, 0.294 for NRE, 0.327 for capital, and 0.361 for labor, for developedcountries.

The long-run relationship modeled in Eq. (2) is estimated using theFMOLS technique. Table 8 presents the results for each source bycolumn: (1) biomass, (2) geothermal, (3) hydroelectricity, (4) solar,(5) waste, and (6) wind. Noteworthy patterns emerge: a) across allrenewable energy sources the estimated coefficients are positive andstatistically significant; b) a 1% increase in biomass increases real GDPby 0.129%; c) a similar increase in hydroelectricity andwaste generationincreases real GDP by 0.114%, and 0.096%; d) geothermal, solar, andwind have the smallest impact with estimated elasticities of 0.085,0.055, and 0.053; and e) the elasticities for non-biomass (0.41), non-geothermal (0.53), non-hydro (0.38), non-solar (0.48), and non-waste(0.53) variables are comparable to thenon-renewablemeasure reportedin Table 7 of 0.33. Interestingly, non-wind generation has an elasticity of3.86.

In the next sections, we proceed to test for causality between eachsource and real GDP.

3. Results for the panel error correction models

Using a panel ECM, we examine the causal relationship betweenrenewable electricity and economic activity. The first difference ofeach variable is modeled as a function of the lagged difference of eachexplanatory variable and an error correction term (ECT) based on the

6 Apergis and Payne (2010b) report an elasticity of 0.20 for Eurasian countries withRussia and 0.07 without. Apergis and Payne (2010) reports 0.76 for OECD countries, andApergis and Payne (2011d) estimates 0.24 for Central American countries. These papersexclude NRE as a control.

lagged residuals from Eq. (1). One lag-length is utilized, such that themodel becomes

ΔYit ¼ α1i þ θ11iΔYit−1 þ θ12iΔREit−1 þ θ13iΔNREit−1 þ θ14iΔKit−1þ θ15iΔLit−1 þ λ1iε1it−1 þ u1it ð3aÞ

ΔREit ¼ α2i þ θ21iΔYit−1 þ θ22iΔREit−1 þ θ23iΔNREit−1þ θ24iΔKit−1 þ θ25iΔLit−1 þ λ2iε2it−1 þ u2it ð3bÞ

ΔNREit ¼ α3i þ θ31iΔYit−1 þ θ32iΔREit−1 þ θ23iΔNREit−1þ θ34iΔKit−1 þ θ35iΔLit−1 þ λ3iε3it−1 þ u3it ð3cÞ

ΔKit ¼ α4i þ θ41iΔYit−1 þ θ42iΔREit−1 þ θ43iΔNREit−1 þ θ44iΔKit−1þ θ45iΔLit−1 þ λ4iε4it−1 þ u4it ð3dÞ

ΔLit ¼ α5i þ θ51iΔYit−1 þ θ52iΔREit−1 þ θ53iΔNREit−1 þ θ54iΔKit−1þ θ55iΔLit−1 þ λ5iε5it−1 þ u5it ð3eÞ

where the ECTs are modeled as the lagged residuals from Eq. (1). Anegative and statistically significant coefficient for the ECT, λ, impliesthe variables are cointegrated and shocks to the model generate move-ments back toward equilibrium. A positive (negative) and statisticallysignificant result for each θkj coefficient indicates that the variable hasa positive (negative) short-run causal impact on the dependentvariable. Our panel ECM is estimated using the method developed byPesaran et al. (1999), which accounts for hetergoneous panels. Wetest for efficiency using a Hausman test, which indicates the mean-group estimator is preferred over the pooled mean group and dynamicfixed-effects estimators. Themean-group estimator allows the long-runestimates to vary by country, and the results presented are of theunweighted means of the coefficients.

The results for thepanel ECM for renewable generation are presentedin Table 9. Each row represents an equation from the model with thecolumns capturing the different sources of causation in the short-runby the independent variables and the long-run by the ECT. We providetest statistics for a test of inclusion for each variable. In parenthesis, wereport the estimated coefficient for each variable and below in brackets,we report the p-value from a likelihood ratio test with a null hypothesisof H0: θkj = 0.

All variables have a short-runGranger-causing impact on the depen-dent variable. For real GDP in row 1, the results imply that renewableelectricity, non-renewable electricity, capital, and labor are significantsources of causation in the short-run. For renewable generation in

Table 8Fully modified OLS for each renewable sources.

Log(GDP) Log(GDP) Log(GDP) Log(GDP) Log(GDP) Log(GDP)

(1) (2) (3) (4) (5) (6)

Log(Bio) 0.129a – – – – –

(15.21) – – – – –

Log(Non-bio) 0.412a – – – – –

(10.77) – – – – –

Log(Geo) – 0.085a – – – –

– (6.43) – – – –

Log(Non-geo) – 0.533a – – – –

– (7.48) – – – –

Log(Hydro) – – 0.114a – – –

– – (5.96) – – –

Log(Non-hydro) – – 0.384a – – –

– – (11.63) – – –

Log(Solar) – – – 0.055a – –

– – – (15.20) – –

Log(Non-solar) – – – 0.478a – –

– – – (3.90) – –

Log(Waste) – – – – 0.096a –

– – – – (12.19) –

Log(Non-waste) – – – – 0.528a –

– – – – (10.80) –

Log(Wind) – – – – – 0.053a

– – – – – (23.13)Log(Non-wind) – – – – – 3.864a

– – – – – (6.64)Log(Capital) 0.325a 0.251a 0.378a 0.294a 0.296a 0.321a

(17.15) (8.15) (17.99) (9.25) (10.94) (11.73)Log(Labor) 0.903a 1.184a 1.220a −0.168b 1.156a 0.822a

(11.39) (8.73) (11.07) (−2.04) (13.07) (9.91)c

N 342 114 380 114 304 190

Estimates for capital and labor are excluded for brevity. t-statistic reported in the parenthesis.a Statistical significance at .01 level.b Statistical significance at .05 level.c Statistical significance at .10 level.


row 2, the results suggest that GDP, non-renewable electricity, capital,and labor are significant sources. These results provide evidence ofbidirectional causality between renewable generation and real GDP,supporting the feedback hypothesis.

Given the nature of our estimation, the ECT coefficient λ should benegative for all equations, indicating that deviations from equilibriumgenerate movements back toward the long-run equilibrium. In Table 9,the ECT is statistically significant for all equations suggesting thateconomic growth, renewable and nonrenewable generation, capital,

Table 9Panel causality tests for aggregate renewable measure.

Sources of Causation (independent variables)

Short-run

Dependent variable ΔGDP ΔRE ΔNR

(3a) GDP – 89.74 (0.0385) 64.5– [0.00]a [0.00

(3b) ΔRE 56.80 (0.848) – 71.2[0.00]a – [0.00

(3c) ΔNRE 71.63 (0.209) 44.86 (−0.228) –

[0.00]a [0.00]a –

(3d) ΔK 167.34 (−0.075) 70.09 (0.050)b 67.4[0.00]a [0.00]a [0.00

(3e) ΔL 111.23 (−0.117) 65.38 (0.020) 75.0[0.00]a [0.00]a [0.00

This table reports the partial F-statistics with respect to short-run changes in the independent vacriterion, and the coefficients are reported in parenthesis. p-Values are reported in brackets belocausality.


and labor are cointegrated. The speed of adjustment for GDP is10.6 months, and 8.4 months for renewable energy, which is muchquicker than that suggested by Apergis and Payne (2011b, 2012).Thus, the data provide evidence of both a short- and long-run relation-ship between renewable generation and economic growth, supportingmuch of the literature that includes both renewable and non-renewable energy, capital, and labor.

Next, we examine the dynamics of GDP growth with individualsources using the corresponding panel ECM. In a manner similar to the

Long-run

E ΔK ΔL ECT

6 (−0.013) 106.42 (0.007) 102.29 (−0.141) −1.129]a [0.00]a [0.00]a [0.000]a

2 (0.119) 55.45 (0.066) 88.36 (2.045) −1.429]a [0.00]a [0.00]a [0.000]a

52.08 (0.298) 86.68 (0.744) −1.053[0.00]a [0.00]a [0.000]a

3 (−0.096)c – 121.74 (0.183) −0.487]a – [0.00]a [0.026]a

3 (−0.028) 130.79 (0.054) – −0.665]a [0.00]a – [0.000]a

riables from a likelihood ratio test. The lag length is one based on the Schwarz informationw for the likelihood ratio test. For the ECT, we report theWald test to determine long-run

Table 10aPanel causality test for biomass.

ΔGDP ΔBio ΔNon-biomass ΔK ΔL EC

(4a) ΔGDP – 112.23 (−0.014) 109.38 (0.145) 141.84 (0.031) 155.03 (0.145) −1.032– [0.00]a [0.00]a [0.00]a [0.00]a [0.000]a

(4b) ΔBio 102.77 (1.901) – 90.78 (0.695) 83.21 (0.015) 70.50 (3.760) −1.053[0.00]a – [0.00]a [0.00]a [0.00]a [0.000]a

(4c) ΔNon-biomass 78.76 (−0.396) 102.75 (−0.074) – 103.49 (0.080) 69.79 (0.486) −1.369[0.00]a [0.00]a – [0.00]a [0.00]a [0.000]a

(4d) ΔK 149.79 (−0.374) 95.75 (−0.082) 107.97 (0.408) – 105.01 (0.894) −0.460[0.00]a [0.00]a [0.00]a – [0.00]a [0.014]b

(4e) ΔL –148.91 (0.074)c 152.56 (0.003) 71.11 (0.026) 127.99 (0.022) – −0.603[0.00]a [0.00]a [0.00]a [0.00]a – [0.000]a

This table reports the partial F-statistics with respect to short-run changes in the independent variables from a likelihood ratio test. The lag length is one based on the Schwarz informationcriterion, and the coefficients are reported in parenthesis. p-Values are reported in brackets below for the likelihood ratio test. For the ECT, we report theWald test to determine long-runcausality.


Table 10bPanel causality test for geothermal.

ΔGDP ΔGeothermal ΔNon-geothermal ΔK ΔL EC

(5a) ΔGDP – 23.69 (−0.074) 15.77 (−0.030) 79.09 (0.110) 40.84 (−0.059) −1.562– [0.01]a [0.15] [0.00]b [0.00]b [0.000]b

(5b) ΔGeo 26.41 (−3.124) – 10.42 (−1.288) 20.61 (1.416) 20.67 (5.503) −1.270[0.006]b – [0.49] [0.04]a [0.04]a [0.000]b

(5c) ΔNon-geothermal 30.61 (−0.339) 7.81 (−0.036) – 42.49 (0.238) 20.34 (0.223) −1.230[0.00]b [0.73]c – [0.00]b [0.04]a [0.000]b

(5d) ΔK 63.56 (−1.005) 28.64 (−0.234) 11.36 (0.178) – 43.86 (0.726) 0.387[0.00]b [0.00]b [0.41] – [0.00]b [0.204]

(5e) ΔL 59.12 (−0.056) 33.08 (−0.027) 4.22 (−0.083) 51.73 (−0.010) – −0.696[0.00]b [0.00]b [0.96] [0.00]b – [0.000]b




aggregate renewable energy specification, the model for biomass be-comes

ΔYit ¼ α1i þ θ11iΔYit−1 þ θ12iΔBIOit−1 þ θ13iΔNonBIOit−1þ θ14iΔKit−1 þ θ15iΔLit−1 þ λ1iε1it−1 þ u1it ð4aÞ

ΔBIOit ¼ α2i þ θ21iΔYit−1 þ θ22iΔBIOit−1 þ θ23iΔNonBIOit−1þ θ24iΔKit−1 þ θ25iΔLit−1 þ λ2iε2it−1 þ u2it ð4bÞ

ΔNonBIOit ¼ α3i þ θ31iΔYit−1 þ θ32iΔBIOit−1 þ θ23iΔNonBIOit−1þ θ ΔK þ θ ΔL þ λ ε þ u ð4cÞ

7 This result opposes the findings of Payne (2011), which found evidence of a unidirec-tional relationship, where increases in biomass consumption positively affect real GDP forthe United States.

34i it−1 35i it−1 3i 3it−1 3it

ΔKit ¼ α4i þ θ41iΔYit−1 þ θ42iΔBIOit−1 þ θ43iΔNonBIOit−1þ θ44iΔKit−1 þ θ45iΔLit−1 þ λ4iε4it−1 þ u4it ð4dÞ

ΔLit ¼ α5i þ θ51iΔYit−1 þ θ52iΔBIOit−1 þ θ53iΔNonBIOit−1þ θ54iΔKit−1 þ θ55iΔLit−1 þ λ5iε5it−1 þ u5it: ð4eÞ

where the ECT is the lagged residuals from Eq. (2). For the other fivesources, we use a similar model, and notate the equations as 5a–5e forgeothermal, 6a–6e for hydroelectric, 7a–7e for solar, 8a–8e for waste,and 9a–9e for wind.

Table 10a–10f reports the panel causality results by source, whereeach panel (a–f) represents a new source. The columns display theshort and long-run sources of causation for each equation. Examiningthe partial-F statistics in row 1 for each panel, we note the short-run

impact of each energy source on GDP. Biomass, geothermal, hydroelec-tricity, waste and wind exhibit a statistically significant short-run causalrelationship with GDP.

In Table 10a, the estimated coefficient of −0.014 from the GDPestimation (4a) indicates that short-run increases in biomass decreasethe growth of real GDP. Eq. (4b) for biomass reports that real GDP hasa positive short-run impact on biomass. Thus, a bidirectional relation-ship exists between biomass and GDP.7

For hydroelectricity and waste energy in Tables 10c and 10e, theestimated coefficients for each renewable source in Eqs. (6a) and (8a)indicate a positive short-run impact on real GDP. Eqs. (6b) and (8b)report a positive short-run causal relationship from real GDP to therenewable energy source. These results indicate a positive bidirectionalrelationship, supporting the feedback hypothesis between hydroelec-tricity and GDP, and between waste energy and GDP. In Tables 10band 10f, the results for geothermal and wind report short-run causality,but both exhibit negative bidirectional causality with real GDP, as notedby Eqs. (5a) and (5b) for geothermal, and Eqs. (9a) and (9b) for windenergy. Finally, solar energy in Table 10d exhibits only unidirectionalcausality fromGDP to solar electricity generation. Thus, hydroelectricityand waste energy are two important factors in the renewable energy–GDP nexus.

Table 10cPanel causality test for hydro.

ΔGDP ΔHydro ΔNon-hydro ΔK ΔL EC

(6a) ΔGDP – 68.84 (0.062) 65.32 (−0.001) 98.08 (−0.009) 104.93 (0.057) −1.154– [0.00]a [0.00]a [0.00]a [0.00]a [0.000]a

(6b) ΔHydro 34.95 (0.524) – 64.01 (0.224) 55.09 (0.038) 81.02 (0.590) −1.488[0.00]a – [0.00]a [0.00]a [0.00]a [0.000]a

(6c) ΔNon-hydro 57.32 (0.282) 66.15 (−0.147) – 64.29 (0.260) 71.83 (0.350) −1.142[0.00]a [0.00]a – [0.00]a [0.00]a [0.000]a

(6d) ΔK 147.95 (−0.207) 57.27 (0.148) 74.83 (−0.083) – 105.29 (0.229) −0.47[0.00]a [0.00]a [0.00]a – [0.00]a [0.022]b

(6e) ΔL 119.56 (−0.136) 77.51 (0.007) 65.09 (−0.030) 140.62 (0.018)c – −0.613[0.00]a [0.00]a [0.00]a [0.00]a – [0.000]a



Table 10dPanel causality test for solar.

ΔGDP ΔSolar ΔNon-solar ΔK ΔL EC

(7a) ΔGDP – 8.79 (0.007) 26.61 (−0.073) 22.25 (0.030) 37.13 (0.028) −1.222– [0.64] [0.005]a [0.02]b [0.00]a [0.000]a

(7b) ΔSolar 27.55 (18.783) – 3.02 (1.922) 24.07 (−4.545) 23.35 (5.850) −1.391[0.00]a – [0.99] [0.01]b [0.02]b [0.000]a

(7c) ΔNon-solar 34.21 (0.574) 42.42 (−0.138) – 9.68 (−0.365) 35.52 (0.318) −1.240[0.00]a [0.00]a – [0.56] [0.00]a [0.000]a

(7d) ΔK 34.5 (−0.058) 24.10 (−0.033) 32.74 (−0.435) – 28.02 (−0.557) −0.237[0.00]a [0.01]b [0.00]a – [0.00]a [0.173]

(7e) ΔL 7.20 (−0.186) 3.00 (0.002) 16.70 (−0.011) 34.34 (0.068) – −0.477[0.78] [0.99] [0.12] [0.00]a – [0.056]c

This table reports the partial F-statistics with respect to short-run changes in the independent variables from a likelihood ratio test. The lag length is one based on the Schwarz information cri-terion, and the coefficients are reported in parenthesis. p-Values are reported in brackets below for the likelihood ratio test. For the ECT,we report theWald test to determine long-run causality.



Analysis of the ECT, λ, also supports the idea that biomass, hydroelec-tricity, andwaste energy are driving factors in the renewable energy–GDPrelationship. The estimated coefficient for λ is negative and statisticallysignificant across all equations (a–e) for several energy sources, includ-ing biomass (Eqs. (4a)–(4e)), hydroelectricity (Eqs. (6a)–(6e)), andwaste energy (Eqs. (8a)–(8e)), consistentwith a long-run cointegratingrelationship between real GDP, capital, labor, and the energy source.Geothermal, solar, and wind all have a positive and/or insignificantECT for capital in Eqs. (5d), (7d) and (9d). As noted before, geothermaland solar may not exhibit a cointegrating relationship with GDP.

These results confirm the findings of a bidirectional relationship andprovide support for the feedback hypothesis between renewable energy

Table 10ePanel causality test for waste.

ΔGDP ΔWaste ΔNon

(8a) ΔGDP – 42.64 (0.029) 70.22– [0.00]a [0.00

(8b) ΔWaste 52.75 (2.706) – 54.42[0.00]a – [0.00

(8c) ΔNon-waste 57.44 (−0.896) 67.05 (0.021) –

[0.00]a [0.00]a –

(8d) ΔK 158.06 (0.752) 87.14 (0.054) 69.06[0.00]a [0.00]a [0.00

(8e) ΔL 92.44 (−0.012) 68.47 (−0.0001) 48.08[0.00]a [0.00]a [0.00

This table reports the partial F-statistics with respect to short-run changes in the independent variterion, and the coefficients are reported in parenthesis. p-Values are reported in brackets below for


and real GDP. All sources do exhibit a long-run relationship with GDP,but only hydroelectricity andwaste generation exhibit a strong positivebidirectional relationship that supports the feedback hypothesis.Furthermore, biomass, hydroelectricity, and waste energy exhibit thelargest long-run impact on real GDP.

4. Additional analysis for structural breaks and cross-sectionaldependence

Extending the previous renewable energy–economic growth litera-ture, we test for unit roots with structural breaks, and we examine themodel accounting for cross-sectional dependence (CSD). Structural

-waste ΔK ΔL EC

(0.061) 118.19 (−0.091) 69.62 (0.146) −1.438]a [0.00]a [0.00]a [0.000]a

(0.504)b 63.70 (0.042) 89.38 (1.327) −0.887]a [0.00]a [0.00]a [0.000]a

85.19 (0.196) 36.30 (−0.331) −1.360[0.00]a [0.00]a [0.000]a

(0.013) – 93.85 (−0.384) −0.414]a – [0.00]a [0.070]c

(0.004) 68.47 (0.009) – −0.632]a [0.00]a – [0.000]a

ables from a likelihood ratio test. The lag length is one based on the Schwarz information cri-the likelihood ratio test. For the ECT,we report theWald test to determine long-run causality.

Table 10fPanel causality test for wind.

ΔGDP ΔWind ΔNon-wind ΔK ΔL EC

(9a) ΔGDP – 43.80 (−0.003) 43.96 (−0.847) 66.37 (−0.002) 70.25 (−0.027) −1.464– [0.00]a [0.00]a [0.00]a [0.00]a [0.000]a

(9b) ΔWind 50.15 (−4.487) – 34.41 (5.576) 40.85 (1.058) 61.46 (−1.565) −0.577[0.00]a – [0.00]a [0.00]a [0.00]a [0.003]a

(9c) ΔNon-wind 46.35 (−0.078) 29.90 (0.001)b – 39.62 (0.030) 57.52 (0.024) −1.357[0.00]a [0.00]a – [0.00]a [0.00]a [0.000]a

(9d) ΔK 94.15 (0.828) 39.46 (−0.023) 51.67 (−5.736) – 88.34 (−0.547) −0.358[0.00]a [0.00]a [0.00]a – [0.00]a [0.101]

(9e) ΔL 45.75 (−0.147) 30.71 (0.004) 18.57 (−0.513) 34.34 (0.073) – −0.519[0.00]a [0.00]a [0.07]c [0.00]a – [0.000]a



Table 11Structural breaks for aggregate renewable energy.

AO 1 break IO 1 break AO 2 breaks IO 2 breaks


breaks occur when shocks to the series cause a permanent change. CSDoccurs when renewable electricity generation is not independent be-tween countries. This may likely be the case for countries where the elec-tricity grid is interconnected, and policies in one country may affect theelectricity generation in another country. Thus, we examine the panelECM controlling for CSD, but due to the limited amount of data, weforgo the production model framework and exclude capital and laborfrom the analysis.

Australia 2000 2004 1996, 2003a 1995, 2004a

(0.06) (0.07) (0.00), (0.04) (0.01), (0.01)Austria 1994 1996 1998, 2000 1997, 2002

(0.00) (0.11) (0.01), (0.26) (0.06), (0.77)Belgium 2001 2003 1998, 2004 1996, 2003

(0.00) (0.07) (0.00), (0.00) (0.01), (0.59)Canada 2005 2003 1993, 2004a 1994, 2003

(0.02) (0.05) (0.00), (0.00) (0.03), (0.01)Denmark 1999 1995a 1994, 1999 1995, 2001

(0.00) (0.00) (0.00), (0.00) (0.00), (0.06)France 1999a 2000 1999, 2003 1993, 2000

(0.59) (0.13) (0.88), (0.37) (0.62), (0.20)Germany 2001 1998 1999, 2004 1998, 2005

(0.00) (0.07) (0.00), (0.00) (0.04), (0.23)Iceland 2001 2005 1998, 2003 1995, 2005

(0.00) (0.01) (0.00), (0.01) (0.21), (0.01)Italy 1997a 1997a 1997, 1999a 1998, 2002a

(0.00) (0.00) (0.15), (0.56) (0.01), (0.88)Japan 1992 1996a 1991, 1996a 1992, 1996a

(0.93) (0.07) (0.10), (0.02) (0.04), (0.00)Luxembourg 1999 1996 1998, 2004 1998, 2004

(0.00) (0.04) (0.00), (0.00) (0.01), (0.01)Netherlands 2001 1994 1997, 2003 1994, 2002

(0.02) (0.39) (0.00), (0.00) (1.00), (0.02)New Zealand 1999a 1992a 1998, 2001 1992, 2000

(0.19) (1.00) (0.70), (0.11) (1.00), (0.01)Norway 2006a 1998a 1998, 2006a 1996, 2003

(0.02) (0.01) (0.09), (0.06) (0.10), (0.06)Portugal 1994a 1994a 1994, 2001a 1994, 1997a

(0.02) (0.00) (0.05), (0.83) (0.00), (0.99)Spain 1997 1999 1994, 2001 1994, 2001

(0.00) (0.02) (0.02), (0.01) (1.00), (0.01)Sweden 1994a 1995a 1994, 1998 1995, 2006

(0.66) (0.26) (0.66), (0.22) (0.38), (0.51)a a

4.1. Structural breaks

Using the time series tests developed by Perron and Vogelsang (1992)and Clemente et al. (1998), we test each country's renewable generationfor a unit root with one and two structural breaks. The additive outlier(AO) and innovational outlier (IO) are two approaches to modeling thestructural break,where the AO testmodels the break as an abrupt change,such as a policy change, and the IO test models the break as a gradualchange, such as a change in technology with slow adoption.

Table 11 reports the optimal structural break for each country. Inparenthesis is the p-value for that break's estimated coefficient, andthe superscript notates the unit root test result. For most countries, wefail to reject the presence of a unit root, and most structural breaks arestatistically significant, occurring between 1997 and 2002. For severalcountries not in the EU (Canada, Japan, New Zealand, United States),the structural break falls outside these dates and/or is not statisticallysignificant. We note that in 1997 the EU signed the Amsterdam Treaty,which promoted sustainable development and encouraged renewablegeneration. Thus, it is likely that the structural break is caused by a com-mon factor, and we consider the possibility of CSD.

Given that several countries have structural breaks, we test for panelstationarity allowing for multiple unknown breaks. Table 12 reports theresults of a test proposed by Carrion-i-Silvestre et al. (2005), whichprovides flexibility in the number of breaks, as well as allowing thebreak points to differ across countries. Our results indicate the presenceof a unit root, even when allowing for structural breaks. Thus, the vari-ables of interest remain non-stationary at levels.

Switzerland 1999 2000 1999, 2003 1997, 2003(0.35) (0.68) (0.11), (0.16) (0.01), (0.02)

United Kingdom 2001 1997 1998, 2003 1996, 2003(0.00) (0.23) (0.00), (0.00) (0.00), (0.01)

United States 1999 2000 1992, 1999 1993, 2000(0.34) (0.40) (0.33), (0.21) (0.16), (0.89)

This table reports the optimal break dates for the series. In parenthesis is the p-value forthe estimated coefficient of each break. The additive outlier (AO)model assumes changestake place rapidly. The innovational outlier (IO) model assumes changes take placegradually.

a Indicates thenull hypothesis of a unit root is rejected because the t-statistic is less thanthe critical value:−3.560 for AOwith 1 break,−4.270 for IOwith 1 break, and−5.49 forAO and IO with 2 breaks. The test is with a 5% significance level.

4.2. Cross-sectional dependence

We test for CSD using themethoddeveloped in Pesaran (2004). Eachof the energy sources are tested for CSD and the average correlationacross countries is reported in Table 13. The results indicate CSD for allrenewable energy sources, non-renewable energy, and GDP.

Next, we implement the Pesaran (2007) a unit root test thatcontrols for CSD. Table 14 reports the results that GDP, capital, andlabor are trend stationary when accounting for CSD. Only renewable,

non-renewable, non-biomass, hydro, non-hydro, solar, non-solar, non-waste, and non-wind are integrated of order one.

Table 13Cross-sectional dependence test statistic and average correlation.

Pesaran CSD test Average correlation

Log(GDP) 56.88 (0.00) 0.947Log(Renewable) 24.85 (0.00) 0.414Log(Non-renewable) 31.80 (0.00) 0.529Log(Biomass) 40.62 (0.00) 0.753Log(Geothermal) 6.60 (0.00) 0.391Log(Hydro-electric) 7.08 (0.00) 0.118Log(Solar) 8.75 (0.00) 0.518Log(Waste) 39.19 (0.00) 0.821Log(Wind) 26.97 (0.00) 0.922

p-Values are reported in parenthesis. The null hypothesis is cross-sectional independence.

Table 12Panel unit root tests with structural breaks.

Bartlett kernel Quadratic kernel

Homogeneous breaks Heterogeneous breaks Homogeneous breaks Heterogeneous breaks

Log(GDP) 42.813 264.153 43.476 267.844(0.000) (0.000) (0.000) (0.000)

Log(Renewable) 18.107 104.895 18.279 99.315(0.000) (0.000) (0.000) (0.000)

Log(Non-renewable) 28.811 156.125 29.201 176.583(0.000) (0.000) (0.000) (0.000)

Log(Biomass) 17.133 301.957 16.814 288.526(0.000) (0.000) (0.000) (0.000)

Log(Geothermal) 7.206 183.966 7.902 181.291(0.000) (0.000) (0.000) (0.000)

Log(Hydro-electric) 12.686 49.173 12.596 46.047(0.000) (0.000) (0.000) (0.000)

Log(Solar) 41.207 60.642 45.126 61.642(0.000) (0.000) (0.000) (0.000)

Log(Waste) 32.102 282.286 31.095 271.380(0.000) (0.000) (0.000) (0.000)

Log(Wind) 42.004 122.868 41.839 119.541(0.000) (0.000) (0.000) (0.000)

p-Values are reported in parenthesis. The tests allow up to 5 structural breaks. The null hypothesis is panel stationarity.


Westerlund and Edgerton (2007) propose four cointegration tests,and control for CSD using a bootstrapping method for the reportp-values. Table 15 reports the results that each energy source does not ex-hibit a cointegrating relationship between GDP, the renewable source,and the source's complement. Thus, using this approach to testing forcointegration, each of the renewable sources do not appear to becointegrated with GDP.

Thus, we further test for cointegration using the panel ECM whilecontrolling for CSD.

4.3. Mean group panel error correction model corrected for cross-sectionaldependence

The mean group panel ECM is used to further test for cointegration,and examinepossible short run causality. Following themodel proposedby Pesaran (2006) and Binder and Offermanns (2007), we account forunobserved common factors by including the cross-country average ofeach variable.8 The model becomes

ΔYit ¼ α1i þ θ11iΔYit−1 þ θ12iΔREit−1 þ θ13iΔNREit−1 þ v11iΔYþ v12iΔREit

þ v13iΔNREit þ λ1i þ εCSD1it−1 þ u1it

ð10aÞ

8 Due to the limited number of observation and estimated parameters, capital and laborare excluded.

ΔREit ¼ α2i þ θ21iΔYit−1 þ θ22iΔREit−1θ23iΔNREit−1 þ v21iΔYit þ v22iΔREit

þ v23iΔNREit þ λ2iεCSD2it−1 þ u2it

ð10bÞ

ΔNREit ¼ α3i þ θ31iΔYit−1 þ θ32iΔREit−1 þ θ33iΔNREit−1 þ v31iΔYit þ v32iΔREit

þ v33iΔNREit þ λ3iεCSD3it−1 þ u1it

ð10cÞ

where the ECT is modeled as the lagged residuals of the long-runequation that accounts for CSD

εCSDit ¼ Yit− β1i þ β2itþ β3iREit þ β4iNREit þ ζ1iYþ ζ2iREþ ζ3iNRE� �

:

For each individual source, we use a similar model, and notate theequations as 11a–11c for biomass, 12a–12c for geothermal, 13a–13cfor hydroelectric, 14a–14c for solar, 15a–15c for waste, and 16a–16cfor wind.

Tables 16a and 16g present the results from the panel ECM. Examin-ing the partial-F statistics in row 1 for each panel, we note the short-runimpact of each energy source on GDP. All sources exhibit a statisticallysignificant short-run causal relationship with GDP, except solar energy.In Table 16a, aggregate renewable energy exhibits a bidirectionalrelationship as before, but the estimated coefficients indicate a changein sign. Increases in renewable energy increase GDP, but increases inGDP decrease renewable generation. The results for biomass, geother-mal, and solar energy remain unchanged. The estimated coefficient of−0.018 in Table 16b Eq. (11a) indicates that short-run increases inbiomass decrease the growth of real GDP. Eq. (4b) for biomass reportsthat real GDP has a positive short-run impact on biomass. Table 16creports geothermal exhibits a short-run negative bidirectional causalrelationship with GDP. Finally, solar energy in Table 16e exhibits onlyunidirectional causality from GDP to solar electricity generation.

For hydroelectricity and waste energy in Tables 16d and 16f, we stillfind bidirectional causality, but the estimated coefficients for eachsource changes. Increases in hydroelectricity increase GDP, butincreases in GDP decrease hydroelectricity. The reverse is the case for

Table 16bPanel causality test for biomass without capital and labor.

ΔGDP ΔBio ΔNon-biomass EC

(11a) ΔGDP – 134.45 (−0.018) 119.35 (−0.016) −0.895– [0.00]a [0.00]a [0.000]a

(11b) ΔBio 168.27 (0.401) – 181.54 (0.730) −1.486[0.00]a – [0.00]a [0.000]a

(11c)ΔNon-biomass

127.03 (0.151) 167.98 (0.070) – −1.288[0.00]a [0.00]a – [0.000]a

This table reports the partial F-statistics with respect to short-run changes in theindependent variables from a likelihood ratio test. The lag length is one based on theSchwarz information criterion, and the coefficients are reported in parenthesis. P-valuesare reported in brackets below for the likelihood ratio test. For the ECT, we report theWald test to determine long-run causality.

b Statistical significance at .05 level.c Statistical significance at .10 level.

a Statistical significance at .01 level.

Table 16cPanel causality test for geothermal without capital and labor.

ΔGDP ΔGeothermal ΔNon-geothermal

EC

(12a) ΔGDP – 64.79 (−0.046) 34.80 (0.205) −0.229– [0.00]a [0.00]a [0.092]c

(12b) ΔGeo 41.79 (−1.618) – 36.37 (0.959) −0.869[0.00]a – [0.00]a [0.005]a

(12c)ΔNon-geothermal

51.18 (0.230) 19.22 (−0.021) – −1.304[0.00]a [0.08]c – [0.001]a

This table reports the partial F-statistics with respect to short-run changes in theindependent variables from a likelihood ratio test. The lag length is one based on theSchwarz information criterion, and the coefficients are reported in parenthesis. P-valuesare reported in brackets below for the likelihood ratio test. For the ECT, we report theWald test to determine long-run causality.a Statistical significance at .01 level.

Table 14Unit root test with cross-sectional dependence and trend.

GDP Renewable Non-renew

Lag(0) 0.660 −4.375 −3.225(0.745) (0.000) (0.001)

Lag(1) 2.34 −0.721 0.641(0.990) (0.235) (0.739)

GDP Biomass Non-bio

Lag(0) 0.098 0.089 −4.128(0.539) (0.535) (0.000)

Lag(1) 2.544 0.926 −1.321(0.995) (0.823) (0.093)

GDP Geothermal Non-geo

Lag(0) −1.281 −0.589 1.829(0.100) (0.278) (0.966)

Lag(1) 1.374 0.008 2.659(0.915) (0.503) (0.996)

GDP Hydro Non-hydro

Lag(0) 0.660 −3.519 −2.435(0.745) (0.000) (0.007)

Lag(1) 2.34 −0.631 0.327(0.990) (0.264) (0.628)

GDP Solar Non-solar

Lag(0) 2.183 −2.284 −1.529(0.985) (0.011) (0.063)

Lag(1) 1.613 −0.256 −0.748(0.947) (0.399) (0.227)

GDP Waste Non-waste

Lag(0) 0.217 1.040 −2.471(0.586) (0.851) (0.007)

Lag(1) 1.809 1.068 1.224(0.965) (0.857) (0.890)

GDP Wind Non-wind

Lag(0) −1.576 1.678 −2.828(0.057) (0.953) (0.002)

Lag(1) −1.171 1.042 −2.006(0.121) (0.851) (0.022)

This table reports the Z t-bar statistics for the Pesaran (2007) unit root test controlling forcross-sectional dependence, including a trend. p-Values are reported in parenthesis. Allvariables are log-transformed.

Table 16aPanel causality test for renewable without capital and labor.

ΔGDP ΔRenew ΔNon-renew EC

(10a) ΔGDP – 185.65 (0.042) 111.79 (0.085) −0.612– [0.00]a [0.00]a [0.000]a

(10b) ΔRenew 104.18 (−2.233) – 97.92 (0.367) −1.313[0.00]a – [0.00]a [0.000]a

(10c)ΔNon-renew

166.35 (−0.144) 121.66 (−0.218) – −0.813[0.00]a [0.00]a – [0.000]a




Table 15Westerlund error correction model panel cointegration test, robust p-values.

Aggregate renew Biomass Geothermal Hydro Solar Waste Wind

Gt 0.724 0.812 0.670 0.872 0.628 0.908 0.418Ga 0.776 0.408 0.802 0.968 0.632 0.754 0.242Pt 0.698 0.564 0.620 0.742 0.506 0.838 0.212Pa 0.710 0.606 0.610 0.758 0.502 0.830 0.138

Bootstrapped p-values at 500.


waste generation. Increases inwaste decreaseGDP, but increases in GDPcause increases in waste.9

Analysis of the ECT, λ, supports the idea of a cointegrating relation-ship between renewable energy and GDP when accounting for CSD.The estimated coefficient for λ is negative and statistically significantacross all equations (a–c) for all energy sources in Tables 16a–16g,consistent with a long-run cointegrating relationship between realGDP, the energy source, and all other sources. The speed of adjustmentfor GDP is 13.4 months for biomass, 52 for geothermal, 20 for hydro, 24for solar, 15 for waste, and 14 for wind, which is much slower than theresults from Section 3, but perhaps somewhat more realistic.

9 The appendix includes the time and fixed effects coefficients for this model.

These results highlight the complicated nature of the relationship ofrenewable energy and GDP, and the importance of disaggregatingsources to examine the individual impacts. Future researchmay examinethe impact of CSD in a production model framework.

5. Conclusions

This study contributes to previous research on the causal relationshipbetween renewable energy and economic growth by examining electric-ity generation by each renewable source: biomass, geothermal, hydro-electricity, solar, waste, and wind energy. Using data from 20 OECDcountries from 1990 to 2008, we implement several panel cointegrationtests. Similar to previous studies, we report a long-run relationship be-tween real GDP, renewable generation, non-renewable energy, real

b Statistical significance at .05 level.cStatistical significance at .10 level.

Table 16dPanel causality test for hydro without capital and labor.

ΔGDP ΔHydro ΔNon-hydro EC

(13a) ΔGDP – 177.79 (0.064) 152.03 (−0.033) −0.611– [0.00]a [0.00]a [0.000]a

(13b) ΔHydro 115.26 (−1.878) – 176.56 (−0.522) −1.613[0.00]a – [0.00]a [0.000]a

(13c)ΔNon-hydro

108 (−0.023) 100.08 (−0.051) – −0.752[0.00]a [0.00]a – [0.007]a




Table 16ePanel causality test for solar without capital and labor.

ΔGDP ΔSolar ΔNon-solar EC

(14a) ΔGDP – 15.21 (0.002) 35.12 (0.137) −0.507– [0.23] [0.00]a [0.031]b

(14b) ΔSolar 117.79 (7.595) – 49.14 (−1.686) −1.275[0.00]a – [0.00]a [0.003]a

(14c) ΔNon-solar 17.10 (−0.025) 38.50 (−0.022) – −2.149[0.15] [0.00]a – [0.000]a


c Statistical significance at .10 level.

a Statistical significance at .01 level.b Statistical significance at .05 level.

Table 16gPanel causality test for wind without capital and labor.

ΔGDP ΔWind ΔNon-wind EC

(16a) ΔGDP – 70.36 (0.010) 51.34 (1.113) −0.855– [0.00]a [0.00]a [0.005]a

(16b) ΔWind 77.06 (3.390) – 85.33 (6.729)b −0.847c

[0.00]a – [0.00]a [0.007]a

(16c) ΔNon-wind 91.83 (0.013) 178.77 (−0.007) – −1.641[0.00]a [0.00]a – [0.000]a




gross fixed capital, and the labor force. The long-run elasticities from afully modified OLS model are positive and statistically significant. Usinga panel ECM to test for causality, we find short-run bidirectional causalitybetween renewable energy and real GDP, supporting the feedback hy-potheses. For the individual sources of renewable energy, we utilize thesame cointegration tests and FMOLS estimation method. All sources ex-hibit a long-run relationship with real GDP, capital, and labor; but, wefind onlyweak evidence of a long-run cointegrating relationship betweengeothermal energy and GDP, and solar energy and GDP. The elasticitiesestimated indicate that biomass, hydroelectricity, and waste energyhave the largest impact on real GDP with elasticities of 0.129, 0.114, and0.096. Geothermal, solar, and wind exhibit the smallest impact with esti-mated elasticities of 0.085, 0.055 and 0.053.

Utilizing the panel ECM for each individual source, we find thathydroelectricity and waste energy are the only two sources to exhibit

Table 16fPanel causality test for waste without capital and labor.

ΔGDP ΔWaste ΔNon-waste EC

(15a) ΔGDP – 80.0 (−0.015) 162.36 (0.080) −0.776– [0.00]a [0.00]a [0.000]a

(15b) ΔWaste 62.30 (2.193) – 78.19 (−1.343) −0.576[0.00]a – [0.00]a [0.004]a

(15c)ΔNon-waste

146.16 (−0.152) 159.67 (−0.158) – −1.336[0.00]a [0.00]a – [0.000]a




short-run bidirectional causality with real GDP, which supports thefeedback hypothesis found in the aggregate measure of renewableenergy. The results for biomass report short-run causality, butwe reporta negative causal effect from biomass to GDP and a positive impactfrom GDP to biomass. Geothermal and wind both exhibit negativebidirectional causality. Finally, solar energy exhibits only unidirectionalcausality from GDP to solar electricity generation.

We extend our analysis with the panel ECM to control for cross-sectional dependences, but without the production model framework.The results for biomass, geothermal, and solar energy remainunchanged. For hydroelectricity and waste energy, we still find bidirec-tional causality, but the estimated coefficients for each renewableenergy source changes. Increases in hydroelectric generation increaseGDP, but increases in GDP decrease hydroelectric generation. Thereverse is the case for waste generation.

The growing attention over renewable energy is due to concernover environmental quality standards, greenhouse gas emissions, andglobal climate change; yet, the contribution of renewable energy toenvironmental quality is questionable. Salim and Rafiq (2012) reportbidirectional causality between pollution and renewable energy,and Apergis et al. (2010) find that renewable energy consumption has apositive impact on CO2 emissions, suggesting that renewable energygrowth does not contribute to reductions in emissions. They attributethese findings to inadequate storage and intermittency issues ofrenewables, suggesting that renewable sources increase the needfor backup power from fossil fuels (Marques and Fuinhas, 2012).The results presented in our paper suggest that increasesin renewable energy may not necessarily lead to decreases in CO2,because biomass and waste generation are important componentsin the renewable energy–GDP nexus. Biomass and waste generationemit CO2, and by only examining the aggregatemeasure of renewable en-ergy, researchers can reach misleading interpretations that renewableenergy sources are not environmentally friendly and do not reduce CO2

emissions. Such an interpretation does not consider that sustainable bio-mass and waste energy are net neutral, releasing carbon that had beenpreviously extracted from the atmosphere rather than the ground. Addi-tionally, the distinction between carbon-neutral and non-carbon neutralbiomass and waste energy production may provide further insight intorenewables' impact on CO2 emissions.

The results of this paper are important because they illustratethat individual sources of renewable energy differ in impact oneconomic growth, as well as their impact on the environment.When developing environmentally friendly or economicallybeneficially energy policies, it is important that policymakers take bothinto consideration. Energy conservation policies may negatively impactGDP, if the policies cause decreases in hydroelectricity energy, but notnecessarily for biomass, geothermal, solar, or waste energy.

Future research should examine the interaction between the sourcesof energy, as well as their relationship with GDP.


Appendix A

Table A1
Time effects from the ECM accounting for cross-sectional dependence.
Aggregate renewable Biomass Geothermal Hydro Solar Waste Wind

Australia −0.0340 0.117 – 0.0120 – – –

(0.105) (0.141) – (0.013) – – –

Austria 0.0722 −1.309 – 0.0171 – 2.813 –

(0.833) (1.573) – (0.101) – (2.654) –

Belgium −1.026 0.203 – −0.0102 – 0.183 1.457c

(6.097) (0.206) – (0.034) – (0.166) (0.753)Canada 0.0447 0.131a – 0.00870 – 0.0720b –

(0.089) (0.015) – (0.010) – (0.037) –

Denmark −5.880 0.568b – −0.0108 – −1.742 0.207(30.468) (0.283) – (0.212) – (3.986) (0.332)

France 0.0239 0.0971 – 0.0287 – 1.068 –

(0.359) (0.130) – (0.071) 0.0744 (1.890) –

Germany 0.290c 0.0880 – 0.929 (0.235) 0.131 −0.239(0.151) (0.057) – (7.363) – (0.248) (0.826)

Iceland 0.127 – 0.0486 0.0368 – – –

(0.140) – (0.217) (0.091) – – –

Italy 0.0416 −0.0802 −0.00301 −3.610 −0.0989 −0.112 −0.844a

(0.272) (0.208) (0.007) (17.010) (0.329) (0.370) (0.252)Japan −0.146 −0.288a 0.000533 0.210 – −0.284 –

(0.245) (0.060) (0.034) (0.609) – (0.351) –

Luxembourg −0.125 – – −0.113 – −0.0152 –

(0.081) – – (0.720) – (0.050) –

Netherlands −0.375 0.167 – −0.648c – 1.738 0.120(0.659) (0.685) – (0.361) – (9.344) (0.127)

New Zealand 0.0336a −0.215a 0.0347b 0.0400a – – –

(0.006) (0.041) (0.017) (0.009) – – –

Norway −0.135a −0.0108 – −0.110a – 4.389 –

(0.033) (0.163) – (0.020) – (12.164) –

Portugal −0.318 −0.0278b −0.292 −0.128 0.123a – 0.0362(0.202) (0.013) (0.178) (0.347) (0.039) – (0.089)

Spain −0.222 0.377 – 0.161 −0.333a 0.667 2.375(0.176) (0.517) – (0.216) (0.081) (0.734) (1.777)

Sweden −0.127 −0.0707 – −0.172b – 8.974 −5.083(0.166) (0.087) – (0.067) – (68.328) (6.506)

Switzerland −0.141 −0.109 – 0.131 −0.0960 −0.101 –

(0.119) (0.393) – (0.235) (0.657) (1.101) –

United Kingdom 0.252 0.100 – −0.186b – 0.406 0.207(4.031) (0.469) – (0.082) – (0.399) (0.309)

United States 0.0861 −0.0185 −0.0119 0.152 −6.127 0.144a 0.362(0.215) (0.036) (0.041) (0.132) (53.244) (0.041) (0.404)

Group mean −0.378 −0.0156 −0.0373 −0.163 −1.076 1.146c −0.140(0.296) (0.090) (0.052) (0.192) (1.012) (0.628) (0.618)

a Statistical significance at .01 level.b Statistical significance at .05 levelc Statistical significance at .10 level.

Table A2Fixed effects from the ECM accounting for cross-sectional dependence.


Australia 3.296 −16.19 – 13.01b – – –

(8.83) (95.64) – (5.27) – – –

Austria 7.891 −31.03 – 8.450 – 169.0 –

(26.29) (28.34) – (23.70) – (153.71) –

Belgium −7.099 280.0 – −12.35 – −77.74b 165.9(61.07) (204.52) – (16.08) – (31.57) (212.89)

Canada 13.12 −18.93a – 11.45a – 12.73c –

(10.90) (5.32) – (3.17) – (7.57) –

Denmark −38.30 5.087 – −34.75 – −40.78 23.79(24.75) (19.68) – (65.11) – (35.44) (41.34)

France −11.75 86.86b – 6.510 – 78.30 –

(50.94) (41.71) – (19.22) – (60.11) –

Germany 185.8 121.0a – 37.69 242.0a 1.617 64.94(274.02) (31.51) – (55.64) (93.34) (68.90) (160.88)

Iceland 9.014 – −2.512 1.176 – – –

(9.05) – (12.45) (25.77) – – –

Italy −15.32 −73.91c −14.87b −7.570 16.30 −72.14 248.7(18.63) (43.22) (7.12) (7.89) (256.94) (90.59) (425.71)

Japan −23.38 −37.48a −7.940 −16.25 – −104.9a –

(15.97) (8.61) (15.50) (33.53) – (27.83) –

Table A2 (continued)


Luxembourg −44.55b – – 1.531 – −4.011 –

(17.40) – – (48.80) – (45.52) –

Netherlands −121.9 168.8 – −253.5 – −78.37 −12.81(152.09) (295.38) – (291.04) – (82.15) (17.84)

New Zealand 47.37a −33.88b 29.88b 30.98b – – –

(16.64) (17.12) (12.45) (13.72) – – –

Norway −7.477 7.580 – −14.43c – 192.7c –

(23.00) (15.99) – (8.27) – (106.43) –

Portugal −85.85 59.20a −22.09a −54.10 48.49 – 93.67(75.38) (22.93) (5.88) (107.18) (39.30) – (65.43)

Spain −51.62 21.75 – 57.53 −90.02a 46.55 −60.18(55.25) (109.09) – (74.01) (8.82) (100.77) (47.91)

Sweden −47.04 30.37 – −64.52c – 24.46 49.95a

(30.96) (20.45) – (34.07) – (107.51) (13.68)Switzerland −58.64 −32.73 – −5.779 −15.85 4.770 –

(53.82) (38.52) – (31.51) (23.32) (9.20) –

United Kingdom −12.89 −5.714 – −11.36 – 37.80 358.2b

(28.75) (15.59) – (23.20) – (31.12) (172.69)United States 54.55 19.27b 45.97 35.27 −15.08 20.29c 34.67

(38.20) (9.28) (30.47) (31.51) (11.73) (11.16) (265.69)Group Mean −10.24 30.55 4.739 −13.55 30.98 13.15 96.68b

(13.87) (20.48) (11.03) (14.24) (46.19) (20.81) (40.17)



References

Apergis, N., Payne, J.E., 2010a. Renewable energy consumption and economic growth:evidence from a panel of OECD countries. Energy Policy 38, 656–660.

Apergis, N., Payne, J.E., 2010b. Renewable energy consumption and growth in Eurasia.Energy Econ. 32, 1392–1397.

Apergis, N., Payne, J.E., 2011a. A dynamic panel study of economic development and theelectricity consumption–growth nexus. Energy Econ. 33, 770–781.

Apergis, N., Payne, J.E., 2011b. On the casual dynamics between renewable and non-renewable energy consumption and economic growth in developed and developingcountries. Energy Syst. 1–14.

Apergis, N., Payne, J.E., 2011c. Renewable and non-renewable electricity consumption–growth nexus: evidence from emerging market economies. Appl. Energy 88,5226–5230.

Apergis, N., Payne, J.E., 2011d. The renewable energy consumption–growth nexus inCentral America. Appl. Energy 88, 343–347.

Apergis, N., Payne, J.E., 2012a. A global perspective on the renewable energyconsumption–growth nexus. Energy Sources Part B 314–322.

Apergis, N., Payne, J.E., 2012b. Renewable and non-renewable energy consumption–growth nexus: evidence from a panel error correction model. Energy Econ. 34,733–738.

Apergis, N., Payne, J.E., Menyah, K., Wolde-Rufael, Y., 2010. On the causal dynamicsbetween emissions, nuclear energy, renewable energy, and. Ecol. Econ. 69, 2255–2260.

Binder, M., Offermanns, C.J., 2007. International Investment Positions and Exchange RateDyanmics: A Dynamic Panel AnalysisCenter for Financial Studies.

Bowden,N., Payne, J.E., 2010. Sectoral analysis of the causal relationship between renewableand non-renewable energy consumption and real output in theUS. EnergySources PartB 400–408.

Breitung, J., 2001. The Local Power of Some Unit Root Tests for Panel Data, vol. 15.Emerald Group Publishing Limited.

Breitung, J., Das, S., 2005. Panel unit root tests under cross-sectional dependence.Statistica Neerlandica 59 (4), 414–433.

Carrion-i-Silvestre, J.L., Barrio-Castro, T.D., Lopez-Bazo, E., 2005. Breaking the panels: anapplication to the GDP per capita. Econ. J. 8, 159–175.

Choi, I., 2001. Unit root tests for panel data. J. Int. Money Financ. 20 (2), 249–272.Clemente, J., Montanes, A., Reyes, M., 1998. Testing for a unit root in variables with a

double change in the mean. Econ. Lett. 59, 175–182.Ewing, B.T., Sari, R., Soytas, U., 2008. The relationship between disaggregate energy

consumption and industrial production in the United States: an ARDL approach.Energy Econ. 30, 2302–2313.

Hadri, K., 2000. Testing for stationarity in heterogeneous panel data. Econ. J. 3 (1),148–161.

Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels.J. Econom. 115 (1), 53–74.

Kao, C., Chiang, M.H., 2000. On the estimation and inference of a cointegrated regressionin panel data. Center for Policy Research, Paper 145.

Levin, A., Lin, C.-F., Chu, C.-S.J., 2002. Unit root tests in panel data: asymptotic and finite-sample properties. J. Econom. 108 (1), 1–24.

Lutkepohl, H., 1982. Non-causality due to omitted variables. J. Econ. 19, 267–378.Marques, A.C., Fuinhas, J.A., 2012. Is renewable energy effective in promoting growth?

Energy Policy 46, 434–442.Menegaki, A.N., 2011. Growth and renewable energy in Europe: a random effect model

with evidence for neutrality hypothesis. Energy Econ. 33, 257–263.Ozturk, I., 2010. A literature survey on energy–growth nexus. Energy Policy 38, 340–349.Payne, J.E., 2009. On the dynamics of energy consumption and output in the US. Appl.

Energy 86, 575–577.Payne, J.E., 2010a. A survey of the electricity consumption–growth literature. Appl. Energy

87, 723–731.Payne, J.E., 2010b. Survey of the international evidence on the casual relationship

between energy consumption and growth. J. Econ. Stud. 37 (1), 53–95.Payne, J.E., 2011. On biomass energy consumption and real output in the US. Energy

Sources Part B 6, 47–52.Payne, J.E., 2012. The casual dynamics between US renewabl energy consumption, output,

emissions, and oil prices. Energy Sources Part B 7 (4), 323–330.Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels with

multiple regressors. Oxf. Bull. Econ. Stat. 61 (S1), 653–670.Pedroni, P., 2001. Fully modified OLS for heterogeneous cointegrated panels. Adv. Econ.

15, 93–130.Pedroni, P., 2004. Panel cointegration: asymptotic and finite sample properties of pooled

time series tests with an application to the PPP hypothesis. Econometric Theory597–625.

Perron, P., Vogelsang, T., 1992. Nonstationarity and level shifts with an application topurchasing power parity. J. Bus. Econ. Stat. 10, 301–320.

Pesaran, M.H., 2004. General Diagnostic Tests for Cross Section Dependence inPanelsInstitute for the Study of Labor (IZA).

Pesaran, M.H., 2006. Estimation and inference in large heterogeneous panels with amultifactor error structure. Econometrica 74 (4), 967–1012.

Pesaran, M.H., 2007. A simple panel unit root test in the presence of cross‐sectiondependence. J. Appl. Econ. 22 (2), 265–312.

Pesaran, M.H., Shin, Y., Smith, R.P., 1999. Pooled mean group estimation of dynamicheterogeneous panels. J. Am. Stat. Assoc. 94, 621–634.

Pirlogea, C., Cicea, C., 2012. Econometric perspective of the energy consumption andeconomic growth relation in European Union. Renew. Sust. Energ. Rev. 16,5718–5726.

Sadorsky, P., 2009a. Renewable energy consumption and income in emerging economies.Energy Policy 37, 4021–4028.

Sadorsky, P., 2009b. Renewable energy consumption, CO2 emissions and oil prices in theG7 countries. Energy Econ. 31, 456–462.

Salim, R.A., Rafiq, S., 2012. Why do some emerging economies proactively accelerate theadoption of renewable energy? Energy Econ. 34 (4), 1051–1057.

Westerlund, J., Edgerton, D.L., 2007. A panel bootstrap cointegration test. Econ. Lett. 97(3), 185–190.

http://refhub.elsevier.com/S0140-9883(14)00036-X/rf0005



























































































[email protected]

Documents

Transcript of [email protected]