Does renewable energy consumption add in ...

Renewable and Sustainable Energy Reviews 44 (2015) 576–585

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

Does renewable energy consumption add in economic growth? An application of auto-regressive distributed lag model in Pakistan Muhammad Shahbaz a,1, Nanthakumar Loganathan b, Mohammad Zeshan c, Khalid Zaman d,n a

Department of Management Sciences, COMSATS Institute of Information Technology, Lahore, Pakistan Faculty of Economics and Management Sciences, Universiti Sultan ZainalAbidin, 21300 Kuala Terengganu, Terengganu, Malaysia c Inha University, Incheon and Pukyong National University, Busan, Korea d Department of Management Sciences, COMSATS Institute of Information Technology, Abbottabad, Pakistan b

art ic l e i nf o

a b s t r a c t

Article history: Received 12 August 2014 Received in revised form 30 November 2014 Accepted 4 January 2015

The objective of the study is to examine the relationship between renewable energy consumption and economic growth by incorporating capital and labour as potential determinants of production function in case of Pakistan. This study used auto-regressive distributed lag (ARDL) model and rolling window approach (RWA) for cointegration in context of Pakistan. The study used quarterly data over the period of 1972Q1–2011Q4. The causality analysis applied through VECM Granger causality and innovative accounting approaches. The results reveal that all the variables in the study are cointegrated that shows the long run relationship between the variables. Furthermore, renewable energy consumption, capital and labour boost economic growth. The causality analysis shows the feedback effect between economic growth and renewable energy consumption. & 2015 Elsevier Ltd. All rights reserved.

Keywords: Renewable energy Economic growth Pakistan

Contents 1. 2. 3. 4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data source and methodological framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Empirical results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Results of rolling regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. The VECM Granger causality analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Conclusion and future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Developing and emerging economies face a two-fold energy challenge in 21st century i.e. meeting the needs of billions of people who still lack access to basic, modern energy services while

n

Corresponding author. Tel.: þ 92 334 8982744; fax: þ92 992 383441. E-mail addresses: [email protected] (M. Shahbaz), [email protected] (N. Loganathan), [email protected] (M. Zeshan), [email protected] (K. Zaman). URL: http://www.ciitlahore.edu.pk (M. Shahbaz). 1 UAN: 0092-42-111-001-007. http://dx.doi.org/10.1016/j.rser.2015.01.017 1364-0321/& 2015 Elsevier Ltd. All rights reserved.

576 577 578 580 581 581 584 584

simultaneously participating in a global transition to clean, lowcarbon energy systems. Both aspects of this challenge demand urgent attention [1]. Renewable Energy Systems, particularly hybrid systems are a cost-effective and sustainable technological solution for rural electrification. In most countries, energy prices are heavily regulated. Developed countries tend to levy taxes for environmental reasons on electricity, whereas developing countries heavily subsidize energy consumption for social reasons [2]. Energy is the key to economic development and social wellbeing, and renewable energy is the key to future without dangerous climate change [3]. The importance of energy has been widely recognized in relation to traditional factors of production. With

M. Shahbaz et al. / Renewable and Sustainable Energy Reviews 44 (2015) 576–585

modernization, production processes have become heavily dependent on energy, and sustainable economic growth cannot be achieved without sufficient and uninterrupted supply of energy [4]. The potential for renewable energy technologies to bridge the gap between energy supply and demand in Pakistan is significant. Masud [5] concluded that renewable energy projects have the potential to improve energy security, provide socioeconomic benefits, reduce local pollution and mitigate climate change. In addition, decentralized nature of renewable energy projects have the potential to provide electricity to remote and rural areas, thereby helping to alleviate poverty and reducing the need to collect and burn biomass fuel [6]. Pakistan is facing acute energy shortages. With a demand and supply gap of almost over 5000 mega watt (MW) and ever increasing prices of energy due to high fuel prices, demand of clean, cheap and sustainable energy is imperative for reducing dependence on imported energy resources. Developed countries and neighboring developing countries India and China have a considerable contribution of renewable energy in their energy mix. Hence, promotion of renewable energy technologies is inevitable to sustain economic growth of an economy [7]. The quest for energy independence, economic growth, and environmental sustainability increasingly suggests the importance of renewable energy sources. Renewable energy is gained by tapping into “existing flows of energy” and “natural processes” in ways that generate more usable energy than is expanded in the production process [8]. Like other developing countries in Pakistan, primary energy consumption has raised 80 per cent in preceding two decades. It was 34 million TOE in 1994–1995 and reached to 61 million TOE to just in 2009–2010. Indigenous natural gas, imported oil, hydel power generation, coal consumption, and nuclear power constitute 45 per cent, 35 per cent, 12 per cent, 6 per cent, and 2 per cent, respectively. Conventional energy is consumed mostly to fill the energy requirements in Pakistan. Pakistan has abundant potential for renewable energy. Solar energy is a cheaper source of energy as compared to fossil fuels. Solar energy program for 100 home has been initiated in Baluchistan which can enlighten 26,000 houses. In addition to this, the coastline zones in Pakistan are very rich for production of wind energy [9]. It is projected that it has likely to produce energy of 50,000 MW. The landscapes of Northern areas make it a suitable for wind energy. It is estimated that 5000 village can be electrified if this energy is made operational in Pakistan. It is more appropriate for micro hydro plants which can yield energy of 300 MW. Canal system in Pakistan arranges for a great prospect for renewable energy. Punjab has the potential to yield energy of 350 MW. Along with it, there are also the prospect for the micro plants which can provide energy of 3 MW to small households and business units. Pakistan is an agrarian economy; it would be able to make available cheaper energy to its rural sector. It can yield 17.25 million cubic meters energy in the form of biogas daily which is sufficient to meet the cooking obligation of fifty million folks.2 Hence, it is not the deficiency of resources but mismanagement which is making our lives difficult [10]. The main objective of present study is to investigate the relationship between renewable energy consumption capital, labour and economic growth in case of Pakistan by using Cobb– Douglas production function over the period of 1972Q1–2011Q4. In case of Pakistan, this study contributes to energy literature by five folds i.e., (i) using the ARDL bound testing approach for long run relationship; (ii) the rolling window approach (RWA) to examine robustness of the ARDL results; (iii) OLS and ECM for

2

We took the help of various reports, available on the official website of Alternative Energy Development Board Government of Pakistan, for this study.

577

long run and short run impacts of renewable energy consumption on economic growth; (iv) the VECM Granger causality approach examine causal relationship between the variables and (v) innovative accounting approach (IAA) test the robustness of the VECM Granger causality results. Our findings reveal that cointegration between renewable energy consumption, economic growth, capital and labour exists in case of Pakistan. In addition, our empirical evidence reports that renewable energy consumption has positive impact on economic growth. Capital and labour also adds in economic growth. Furthermore, estimated results indicated the bidirectional causality relationship between renewable energy consumption and economic growth.

2. Literature review Contemporaneous research on energy consumption provides a stream of information regarding the direction of causality between renewable energy consumption and economic growth. All the countries have different effects of renewable energy consumption; some countries report renewable energy consumption fetches handsome contribution in economic growth while for some countries this source of energy is not sufficient. For example, Ewing et al. [11] asserted that renewable energy consumption has minimal impact on economic growth in case of United States. In contrast, (Payne [12], United States) and (Tiwari [13], India) affirmed that renewable energy contributes much in economic growth and suggest that the share of renewable energy in total energy blend must rise over time. In the same spirit but with panel data,3 Tiwari [13] finds positive response of GDP in response to renewable energy consumption innovative shock. In addition, Magnani and Vaona [14] also share the same views for Italy and advised to discourage renewable energy conservation policies. Arifin and Syahruddin [15] reported that adoption of energy conservation policies would affect economic growth in Indonesia because causality is running from renewable energy consumption to economic growth. Khan et al. [16] examined the relationship between energy consumption, economic growth, FDI, relative prices, and financial development in low income, middle income, high income non-OECD, high income OECD, South Africa, Middle East and North Africa (MENA) and the aggregate data of the World over a period of 1975–2011. The results indicate that GDP per capita has a positive impact on energy consumption in low income, middle income, South Africa, MENA and aggregate data of the World. However, in high income OECD and non-OECD regions, there is no significant relationship been found in both regions. Mumtaz et al. [17] examined the direction of the causality between energy consumption and economic growth factors in case of Pakistan, over a period of 1975–2010. The results indicate that there is a unidirectional causality running from energy consumption to trade factors but not vice versa. Akhmat et al. [18] examined the relationship between energy sources and balance of payment factors in Pakistan over a period of 1975 to 2011. The results indicate that balance of payment factors induce an increase in energy consumption in Pakistan. The causality test indicates the unidirectional causality running from trade deficit to oil consumption and oil consumption to current account balance, while electricity consumption Grange causes exports and current account balance. In case of coal consumption, trade deficit Granger causes coal consumption while coal consumption Granger causes current account balance in Pakistan. Akhmat and Zaman [19] investigated the causal relationship among nuclear energy 3 Austria, Belgium and Luxembourg, Bulgaria, Finland, France, Germany, Greece, Republic of Ireland, Italy, Norway, Portugal, Spain, Sweden, Switzerland, Turkey and United Kingdom.

578


consumption, commercial energy consumption and economic growth in South Asian countries, over the period of 1975–2010. The results show that nuclear energy consumption Granger causes economic growth in Nepal and Pakistan; while, commercial energy consumption Granger causes economic growth in Bangladesh, Bhutan, Maldives, Nepal and Srilanka; gas consumption Granger causes economic growth in Bangladesh, Bhutan, India and Maldives; electricity consumption Granger causes economic growth in India and Srilanka, finally, coal consumption Granger causes economic growth in Bangladesh, Bhutan, Nepal and Sri Lanka. Other than growth hypothesis, empirical studies also extract conservation hypothesis. For instance, Sari et al. [20] concluded this view for United States while Sadorsky [21] for a panel of countries,4 both of the studies find that conservation policies are more suitable. On other hand, Apergis and Payne [22] worked with a panel of OECD countries5 and found the bidirectional causality between renewable energy consumption and economic growth. These finding are consistent with Apergis and Payne [23]; who worked with a panel of 13 Eurasian countries,6 Apergis and Payne [14]; for 6 Central American countries,7 Apergis and Payne [24]; for a panel of 80 countries.8 Recently, Apergis and Payne [25] investigated the causality between renewable electricity consumption and economic growth in case of Central American countries namely Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama. They applied panel cointegration developed by Larsson et al. [26] and panel VECM Granger causality approach. Their results indicated the cointegration between the variables. Renewable electricity consumption adds in economic growth and feedback hypothesis exists between renewable electricity consumption and economic growth in long run and renewable electricity consumption Granger causes economic growth in short run. In contrast, some studies find no causal relationship between renewable energy consumption and economic growth. Payne [27] reports no causal relationship between renewable energy consumption and economic growth for United States. A same inference is drawn by Menegaki [28] for a panel of 27 European countries.9 Mahmoodi and Mahmoodi [29] tested the direction of causal relation between renewable energy consumption and economic growth in 7 Asian developing economies.10 Their findings validated the feedback hypothesis for Bangladesh, conservation hypothesis for India, Iran, Pakistan, and Syrian Arab Republic and neutral hypothesis between both variables is confirmed for

4 Argentina, Brazil, Chile, China, Colombia, Czech Republic, Hungary, India, Indonesia, Korea, Mexico, Peru, Philippines, Poland, Portugal, Russia, Thailand, Turkey. 5 Australia, Austria, Belgium, Canada, Denmark, France, Germany, Iceland, Italy, Japan, Luxembourg, The Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States. 6 Armenia, Azerbaijan, Belarus, Estonia, Georgia, Kazakhstan, Kyrgyzstan, Latvia, Moldova, Russia, Tajikistan, Ukraine, Uzbekistan. 7 Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, Panama. 8 Algeria, Argentina, Australia, Austria, Bangladesh, Belgium, Bolivia, Brazil, Bulgaria, Canada, Cameron, Chile, China, Comoros, Costa Rica, Denmark, Dominican Republic, Ecuador, Egypt, El Salvador, Ethiopia, Finland, France, Gabon, Germany, Ghana, Greece, Guatemala, Guinea, Honduras, Hungary, Iceland, India, Indonesia, Iran, Ireland, Italy, Japan, Jordan, Kenya, Korea, Luxembourg, Madagascar, Malawi, Malaysia, Mali, Mauritius, Mexico, Morocco, Mozambique, The Netherlands, New Zealand, Nicaragua, Norway, Pakistan, Panama, Paraguay, Peru, Philippines, Poland, Portugal, Romania, Senegal, South Africa, Spain, Sri Lanka, Sudan, Swaziland, Sweden, Switzerland, Syria, Thailand, Tunisia, Turkey, Uganda, United Kingdom, United States, Uruguay, Venezuela, Zambia. 9 Belgium, Bulgaria, Czech Rep., Denmark, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Latvia, Lithuania, Luxemburg, Hungary, Netherland, Austria, Poland, Portugal, Romania, Slovenia, Slovakia, Finland, Sweden, UK, Norway. 10 India, Iran, Pakistan, Syrian Arab Republic, Bangladesh, Jordan and Sri Lanka.

Sri Lanka. Table 1 presents the summary of empirical studies regarding relationship between renewable energy consumption and economic growth.

3. Data source and methodological framework The purpose of present investigation is to expose the relationship between renewable energy consumption and economic growth in case of Pakistan using quarterly data over the period of 1972Q1–2011Q4. This study employed Cobb–Douglus production function to analyze the correlation between renewable energy consumption and economic growth including capital and labour as additional factors of production. Commonly, the equation of production function is as follows: Y ¼ ARα1 K α2 Lα3 eu

ð1Þ

where Y is domestic output in real terms; R, K and L indicate renewable energy consumption, real capital and labour, respectively. A shows the level of technology to be utilized in the country and e is the error term supposed to be identically, independently and normally distributed. The returns to scale is associated with renewable energy consumption, capital and labour is shown by α1 ; α2 and α3 , respectively. We have converted all the series into logarithms in order to linearize the form of nonlinear Cobb– Douglas production. The main reason is that linear specification does not seem to provide consistent results and not helpful for policy making purpose (Shahbaz et al. [35]; Shahbaz and Feridun [36]). To cover this problem, we use log-linear specification to investigate relationship between renewable energy consumption and economic growth in case of Pakistan. Ehrlich [37,38], Cameron [39] and Layson [40] recommended to pertain log-linear modeling in attaining better, consistent and efficient empirical results.11 The log-linear functional form of Cobb–Douglus production function is modeled as follows: log Y t ¼ log A þ α1 log R t þ α2 log K t þ α3 log Lt þ ut

ð2Þ

The empirical equation to investigate the relationship between renewable energy consumption and economic growth is modeled keeping technology constant. The log-linear specification to assess the relationship between renewable energy consumption and economic growth is as follows: ln Y t ¼ α0 þ α1 ln Rt þ α3 ln K t þ α4 ln Lt þ ut

ð3Þ

where ln Y t , ln Rt , ln K t and ln Lt is the logarithm of per capita real GDP, renewable energy consumption (kg of oil equivalent per capita) per capita, capital use per capita and labour per capita, respectively. The long run association among renewable energy consumption and economic growth in case of Pakistan is investigated by applying the ARDL bounds testing approach presents by Pesaran et al. [41]. The empirical literature indicates the various cointegration approaches in order to test cointegration. But, the ARDL bounds testing approach is preferable due to its advantages over other cointegration techniques. For instance, order of integration of the series does not matter for applying the ARDL bounds testing if no variable is found to be stationary at I(2). This approach is more appropriate as compared to conventional cointegration techniques for small sample (Haug [42]). Within the general-tospecific framework, unrestricted version of the ARDL chooses proper lag order to capture the data generating procedure (see Shahbaz and Lean [43] for more details). Appropriate specification of the ARDL model is sufficient to simultaneously correct for residual serial correlation and endogeneity problems (Pesaran and Shin [44]). The equation of unrestricted error correction 11

See Shahbaz et al. [32] for more details.


579

Table 1 Summary of studies on renewable energy consumption-growth nexus. Nos.

Feedback hypothesis

Conservation hypothesis

Growth hypothesis

Neutrality hypothesis

1. 2. 3. 4. 5. 6. 8. 9.

Ewing et al. [11] Tiwari [13] Apergis and Payne [22] Apergis and Payne [23] Apergis and Payne [24] Apergis and Payne [25] Apergis and Payne [30] Hamdi and Sbia [31]

Sari et al. [20] Sadorsky [21] Abdullah et al. [32]

Payne [12] Bowden [33] Magnani and Vaona [14] Arifin and Syahruddin [15] Zeshan and Ahmad [34]

Payne [27] Menegaki [28] Mahmoodi and Mahmoodi [29]

Note: Growth hypothesis represents unidirectional causality running from renewable energy consumption to economic growth; conservation hypothesis represents unidirectional causality running from economic growth to renewable energy consumption; feedback hypothesis represents bi-directional causality; neutrality hypothesis represents no causality.

model (UECM) to investigate the long-and-short runs relations between the series is following: Δ ln Y t ¼ ϑ1 þ ϑT T þ ϑY ln Y t 1 þ ϑR ln Rt 1 þ ϑK ln K t 1 þ ϑL ln Lt 1 þ

p X

ϑi Δ ln Y t i þ

i¼1

þ

q X

ϑj Δ ln Rt j þ

j¼0

t X

s X

ϑl Δ ln K t l

l¼0

ϑm Δ ln Lt m þ μt

ð4Þ

m¼0

Δ ln Rt ¼ α1 þ αT T þ αY ln Y t 1 þ αR ln Rt 1 þ αK ln K t 1 þ αL ln Lt 1 þ

p X

αi Δ ln Rt i þ

i¼1

þ

q X

αj Δ ln Y t j þ

j¼0

t X

s X

αl Δ ln K t l

l¼0

αm Δ ln Lt m þ μt

ð5Þ

m¼0

Δ ln K t ¼ ρ1 þ ρT T þ ρY ln Y t 1 þ ρR ln Rt 1 þ ρK ln K t 1 þ ρL ln Lt 1 þ

p X

ρi Δ ln K t i þ

i¼1

þ

q X

ρj Δ ln Y t j þ

j¼0

t X

r X

ρk Δ ln Rt k

k¼0

ρm Δ ln Lt m þ μt

ð6Þ

m¼0

Δ ln Lt ¼ σ 1 þ σ T T þ σ Y ln Y t 1 þ σ R ln Rt 1 þ σ K ln K t 1 þ σ L ln Lt 1 þ þ

p X

σ i Δ ln Lt i þ

q X

i¼1

j¼0

r X

t X

k¼0

σ k Δ ln Rt k

σ j Δ ln Y t j σ m Δ ln K t m þ μt

ð7Þ

m¼0

where Δ is the differenced operator and μt is residual term in period t. The akaike information criterion (AIC) is applied to decide suitable lag length of first differenced variables following Lütkepohl [45]. The proper calculated F-statistic depends upon the appropriate lag order selection of the series to be included in the model.12 The overall significance of the coefficients of lagged variables is investigated by applying F-test advanced by Pesaran et al. [41]. The null hypothesis of no long run relationship between the variables in Eq. (3) is H 0 : ϑY ¼ ϑR ¼ ϑK ¼ ϑL ¼ 0 against the alternate hypothesis of long run relationship i.e. H 0 : ϑY a ϑR a ϑK a ϑL a0. Two asymptotic critical values have been generated by Pesaran et al. [41]. These bounds are upper critical bound (UCB) and lower critical bound (LCB) are used to decide whether variables are cointegrated for long run relationship or not. If all the variables are stationary at I(0) then we use LCB to test cointegration between the series. We use UCB to 12

For details see Shahbaz et al. [32].

examine long run relationship between the series if the variables are integrated at I(1) or I(0) or I(1)/I(0). We compute the value of Ftest applying following models such as F Y ðY=R; K; LÞ, F R ðR=Y; K; LÞ, F K ðK=Y; R; LÞ and F L ðL=Y; R; KÞ for Eqs. (4)–(7), respectively. The decision of cointegration is taken with the help of following rules: if upper critical bound (UCB) is less than our computed F-statistic then we conclude cointegration. If computed F-statistic does not exceed lower critical bound then no cointegration among the variables. The decision about cointegration between the series is questionable if computed F-statistic is found between LCB and UCB.13 Our decision regarding cointegration is inconclusive if calculated F-statistic falls between LCB and UCB. In such an environment, error correction method is an easy and suitable way to investigate cointegration between the variables. After the confirmation of long run relationship among the variables then in next step we investigate the causal relation between the series. Granger [46] stated that once the variables are integrated at I(1) then vector error correction method (VECM) is suitable approach to test the direction of causal link among the variables. Relatively, the VECM is restricted form of unrestricted VAR (vector autoregressive) and restriction is levied on the presence of long run relationship between the series. All the series are endogenously used in the system of error correction model (ECM). This shows that in such situation, response variable is explained both by its own lags and lags of independent variables as well as the error correction term and by residual term. The VECM in five variables case can be written as follows: l X

Δ ln Y t ¼ α 3 1 þ

m X

α11 Δ ln Y t i þ

i¼1

þ

o X

p X

α44 Δ lnK t r þ l X

β11 Δ ln Rt i þ

i¼1

þ

α55 Δ lnLt s þ η1 ECT t 1 þ μ1i

o X

Δ ln K t ¼ ϕ 3 1 þ

m X

p X

β55 Δ lnK t s þ η2 ECT t 1 þ μ2i

k¼1

ð9Þ

s¼1 l X

m X

ϕ11 Δ ln K t i þ

i¼1 n X

β22 Δ ln Y t j

j¼1

β44 Δ lnK t r þ

r¼1

þ

ð8Þ

s¼1

r¼1

Δ ln Rt ¼ β 3 1 þ

α22 Δ ln Rt j

j¼1

ϕ33 Δ ln Rt k þ

ϕ22 Δ ln Y t j

j¼1 p X

ϕ55 Δ lnLt s þ η4 ECT t 1 þ μ4i

s¼1

ð10Þ 13 If the variables are integrated at I(0) then F-statistic should be greater than lower critical bound for the existence of cointegration between the series.

580


Δ ln Lt ¼ δ 3 1 þ

l X

δ11 Δ ln Lt i þ

i¼1

þ

n X

δ33 Δ ln Rt k þ

m X

The data span of present study is 1972Q1–2011Q4. The data on renewable energy consumption is collected from GoP [49]. We have used world development indicators (CD-ROM) published by World Bank [50] to collect on data on real GDP, real capital and labour, respectively.

δ22 Δ ln Y t j

j¼1 p X

δ55 Δ lnK t s þ η4 ECT t 1 þ μ4i

s¼1

k¼1

ð11Þ where Δ indicates differenced operator and uit are residual terms and assumed to be identically, independently and normally distributed. The statistical significance of lagged error term i.e. ECT t 1 further validates the established long run relationship between the variables. The estimates of ECT t 1 also shows the speed of convergence rate from short run towards long run equilibrium path in all models. The VECM is superior to test the causal relation once series are cointegrated and causality must be found at least from one direction. Further, the VECM helps to distinguish between short-and-long runs causal relationships. The statistical significance of estimate of lagged error term i.e. ECT t 1 with negative sign confirms the existence of long run causal relation using t-statistic. Short run causality is indicated by the joint χ 2 statistical significance of the estimates of first difference lagged independent variables. For example, the significance of α22;i a 0 8 i implies that renewable energy consumption Granger-causes economic growth and causality runs from economic growth to renewable energy consumption can be indicated by the significance of β22;i a 0 8 i . Finally, we use Wald or F-test to test the joint significance of estimates of lagged terms of independent variables and error correction term. This further confirms the existence of short-and-long run causality relations (Shahbaz et al. [47]) and known as measure of strong Granger-causality (Oh and Lee [48]). Table 2 Ng–Perron unit root test analysis. Level Variables ln Y t ln Rt ln K t ln Lt 1st Difference ln Y t ln Rt ln K t ln Lt n

MZa

MZt

MSB

MPT

2.4251 0.9137 0.6889 1.0930

1.0907 0.6512 0.5238 0.7322

0.4497 0.7127 0.7603 0.6699

37.1408 93.6113 108.301 82.0658

27.6202n 30.3374n 29.1947n 26.5620

3.7074 3.8864 3.8123 3.6354

0.1342 0.1281 0.1305 0.1368

3.3513 3.0519 3.1704 3.4840

Indicates the significance at the 1% level.

4. Empirical results and discussions This study applies Ng–Perron unit root test in order to test the order of integration. This test is superior and more powerful as compared to traditional unit root tests such ADF, DF-GLS, KPPS etc. Baum [51] stated that it is necessary condition to test the integrating order of the variables before applying the ARDL bounds testing approach to cointegration relationship between the series. The assumption of the ARDL bounds testing is that the variables should be integrated at I(0) or I(1) or I(0)/I(1) and no series is stationary at I(2). If any variable is integrated at I(2) then the computation of the ARDL F-statistic becomes invalid. The results of Ng–Perron unit root test are shown in Table 2. The empirical results indicate that all the series are non-stationary at level and stationary at 1st difference. So, all the variables are indicated order one i.e. I(1). The ARDL approach to cointegration checks the presence of long run link among variables. The lag selection is very important in case of the ARDL approach to cointegration. Hence this study uses akaike information criterion (AIC) to choose suitable lag length that helps us in capturing the dynamic relationship to select the best ARDL model to estimate. The results of lag length are reported in Table 3 which indicates that lag 5 is appropriate. The results of the ARDL bounds testing testing approach are reported in Table 4. The empirical evidence indicates that our computed F-statistics for F Y ðY=R; K; LÞ, F R ðR=Y; K; LÞ and F L ðL=Y; R; KÞ are 8.776, 4.061 and 4.249 for economic growth, renewable energy consumption and labour equations, respectively. These F-statistics are greater that upper critical bounds developed by Pesaran et al. [41] at 1per cent, 10 per cent and 5 per cent levels of significance. This confirms the presence of cointegration between economic growth, renewable energy consumption, capital and labour in case of Pakistan. This implies that there is a long run relationship between the variables over the period of 1972Q1–2011Q4. The robustness of the ARDL bounds testing approach is examined by applying Johansen multivariate cointegration approach. The results are reported in Table 5. We can infer that there are two cointegrating vectors which validate the presence of long run relationship between the variables. This entails that the ARDL cointegration analysis is reliable and robust.

Table 3 Lag selection criteria. VAR lag order selection criteria Lag

LogL

LR

FPE

AIC

SC

HQ

0 1 2 3 4 5 6 7 8

129.9537 905.2496 946.4360 1112.641 1575.482 1655.641 1661.778 1672.323 1682.308

NA 1499.585 77.49561 303.9806 822.1521 138.1672a 10.25650 17.06517 15.63497

2.24e 06 1.03e 10 7.37e 11 1.02e 11 2.87e 14 1.24e 14a 1.41e 14 1.53e 14 1.67e 14

1.65728 11.6480 11.9794 13.9558 19.8353 20.6794a 20.5497 20.4779 20.3987

1.5777 11.2501 11.2632 12.9213 18.4825 19.0083a 18.5603 18.1702 17.7727

1.6249 11.4863 11.6884 13.5355 19.2855 20.0006a 19.7415 19.5404 19.3320

LR: sequential modified LR test statistic (each test at 5% level), FPE: Final prediction error, AIC: Akaike information criterion, SC: Schwarz information criterion, HQ: HannanQuinn information criterion. a

Indicates lag order selected by the criterion.


Table 4 The ARDL cointegration analysis. Variable

ln Y t

ln Rt

ln K t

ln Lt

F-statistics Critical values Lower bounds Upper bounds Diagnostic tests Durbin–Watson

8.776n 1% level 5.23 3.93

4.061nnn 5% level 3.12 4.25

1.949 10% level 2.75 3.79

4.249nnn

2.0906

2.0009

2.0476

2.0907

n

Note:

and

nnn

show the significance at 1% and 10% level, respectively.

Table 5 The Johansen cointegration analysis. Hypothesis R¼0 Rr1 Rr2 Rr3 n

Trace statistic

Maximum eigen value

n

89.6299n 39.2551n 10.7930 2.3744

142.0526 52.4226n 13.1674 2.3744

581

1 per cent level of significance. There is 0.4001 per cent of economic growth is stimulated by 1 per cent increase in labour, all else is same. The lower segment of Table 6 reports the results of short dynamics of renewable energy consumption, capita and labour on economic growth. In short span of time, renewable energy consumption, lagged of renewable energy consumption, capital and labour contribute economic growth significantly. The negative and statistically significant estimate of ECM t 1 corroborates the established long run relationship between renewable energy consumption, capital, labour and economic growth in case of Pakistan. The results indicate that estimate of ECM t 1 is 0.0341 and statistically significant at 1 per cent level. This implies that 3.41 per cent changes in economic growth are corrected by deviations in short run towards long run equilibrium path in each quarter. In this model, short run deviations in economic growth take 29 years and 6 month to converge to long run equilibrium path. The short run diagnostic tests show that no serial correlation is found and same interpretation can be drawn for ARCH test. Our empirical exercise indicates that there is no problem of heterogeneity and error term has homogenous variance. The Ramsey reset test shows that functional for model is well specified.

Indicates significance at 1% level.

4.1. Results of rolling regression Table 6 Long-and-short run analysis. Dependent variable ¼ ln Y t Variable

Coefficient

Std. error

Prob. value

Long run results Constant ln Rt ln K t ln Lt

5.5906n 0.6103n 0.1357nn 0.4001n

0.0602 0.0213 0.0068 0.0236

0.0000 0.0000 0.0489 0.0000

Short run results Constant ln Rt ln Rt 1 ln K t ln Lt ECM t 1

0.0045n 0.0738n 0.0012nn 0.0873nnn 0.0847n 0.0341n

0.0007 0.0234 0.0005 0.0500 0.0495 0.0093

0.0000 0.0019 0.0348 0.0827 0.0000 0.0004

Diagnostic tests Test χ 2 SERIAL χ 2 ARCH χ 2 WHITE χ 2 REMSAY

F-statistic 1.5420 1.3191 4.0318 0.0046

Prob. value 0.1686 0.2525 0.0001 0.9455

n

Denote the significant at 1% level. Denote the significant at 5% level. Denote the significant at 10% level.

nn

The rolling regression model is used to evaluate the stability of coefficient of the ARDL model in the sample size. Other estimation methods assume that the coefficients of variables remain constant over the sample size. But in reality the economic condition cannot remains constant and as results the economic indicators are fluctuated over time and their coefficients cannot remains same (Pesaran and Timmermann [52]). With the help of rolling regression approach, we can estimate the coefficient of each observation of the sample by setting the rolling window size. If the economic indicators are changed overtime so this approach captures this instability. Figs. 1–4 show the rolling window results. The black tick line represents the coefficients and light black upper and lower band represents the coefficients’ statistical level of significance (at 5 per cent) Fig. 1 shows the intercept which remains positive over the sample size. Fig. 2 shows the graph of lnRt coefficients. It shows negative relation with economic growth in 1978Q4 and 1999Q3 to 2006Q2. In the remaining sample, it is positively related to economic growth. Fig. 3 represents the capital coefficients. It shows that capital is negatively related to economic growth in 1976Q4 to 1977Q4, 1997Q4 to 1998Q4, and 2005Q1 to 2006Q4. Fig. 4 represents the labour force coefficients. It shows that labour is negatively related to economic growth in 1982Q2–1984Q2, 1989Q3– 1990Q2, and 1991Q2–1993Q1.14

nnn

The investigation of long run relationship between the variables leads us to examine the marginal impacts of renewable energy consumption, capital and labour on economic growth in long run as well as in short run. Table 6 deals with long run marginal impact of determinants of economic growth. The results shown in Table 6 reveal that positive relationship found from renewable energy consumption to economic growth and statistically significance level is 1 per cent. All else is constant, 1 per cent rise in renewable energy consumption spurs economic growth by 0.6103 per cent. The impact of capital on economic growth is positive and is statistically significant at 5 per cent level of significant. Keeping the other things constant, 1 per cent increase in capital use enhances domestic production and hence economic growth by 0.1357 per cent. The relationship between labour and economic growth is positive and it is statistically significant at

4.2. The VECM Granger causality analysis The direction of causal relationship between these variables is investigated by applying the VECM Granger causality approach. The appropriate environmental and energy policy to sustain economic growth is dependent upon the nature of causal relation between the series. In doing so, we applied the VECM Granger causality approach to detect the causality between renewable energy consumption, capital, labour and economic growth which would help policy makers in formulating comprehensive energy policy to accelerate economic growth for long run. Table 7 presents the empirical evidence of long run and short run causality relationships. The results suggest that feedback hypothesis 14 We are grateful to Qazi Muhammad Adnan Hye for his help in applying rolling window approach.

582


50

40

30

20

10

0

-10 1976Q4

1979Q2

1981Q4

1984Q2

1986Q4

1989Q2

1991Q4

1994Q2

1996Q4

1999Q2

2001Q4

2004Q2

2006Q4

2009Q2

2011Q4

Window size 20

Fig. 1. Coefficient of INPT and its two S.E. bands based on rolling OLS. 1.5

1.0

0.5

0.0

-0.5 1976Q4

1979Q2

1981Q4

1984Q2

1986Q4

1989Q2

1991Q4

1994Q2

1996Q4

1999Q2

2001Q4

2004Q2

2006Q4

2009Q2

2011Q4

Window size 20

Fig. 2. Coefficient of lnR and its two S.E. bands based on rolling OLS.

between renewable energy consumption and economic growth, renewable energy consumption and labour, labour and economic growth, in case of Pakistan for long run. The results indicate that causality running from renewable energy consumption to economic growth is stronger compared to causal relationship from economic growth to renewable energy consumption. This shows that government must pay her attention to launch comprehensive energy policy in exploring new sources of renewable energy to sustain economic growth. The R&D activities should be encouraged in energy sector. To overcome energy crisis in the country, government must give incentive to foreign investors to investment in energy sector of Pakistan. The unidirectional causality exists running from capital to renewable energy consumption, economic growth and labour. The feedback effect is also found between economic growth and labour and, labour and renewable energy consumption. In short run, bidirectional causal relationship is found between renewable energy consumption and economic growth. The feedback hypothesis also exists between capital and economic growth. Economic growth and labour Granger cause each other. The unidirectional causality is found running from labour to renewable energy

consumption. The statistically significance of joint long-and-short run causality corroborates our long run and short run causal relationships between the series over the study period of 1972Q1–2011Q4. Existing energy literature reveals that Granger causality approaches such the VECM Granger causality test has some limitations. The causality test cannot capture the relative strength of causal relation between the variable beyond the selected time period. This weakens the reliability of causality results by the VECM Granger approach. To overcome this problem, we applied innovative accounting approach (IAA). The IAA is combination of variance decomposition method (VDM) and impulse response function (IRF). The variance decomposition approach (VDM) determines the response of the dependent actor to shocks stemming from independent actors. The IRF is an alternate of VDM. Table 7 shows the results of VDM.15 The variance decomposition approach indicates the magnitude of the predicted error variance

15 The results of impulse response function are available from authors upon request.


583

4

2

0

-2

-4

-6

-8 1976Q4

1979Q2

1981Q4

1984Q2

1986Q4

1989Q2

1991Q4

1994Q2

1996Q4

1999Q2

2001Q4

2004Q2

2006Q4

2009Q2

2011Q4

2004Q2

2006Q4

2009Q2

2011Q4

Window size 20

Fig. 3. Coefficient of lnK and its two S.E. bands based on rolling OLS. 10

8

6

4

2

0

-2 1976Q4

1979Q2

1981Q4

1984Q2

1986Q4

1989Q2

1991Q4

1994Q2

1996Q4

1999Q2

2001Q4

Window size 20

Fig. 4. Coefficient of lnL and its two S.E. bands based on rolling OLS. Table 7 VECM Granger causality analysis. Dependent variables

Direction of causality Short run ln Y t

ln Y t

–

ln Rt

8.3149n [0.0000] 3.8681n [0.0107] 3.1994n [0.0078]

ln K t ln Lt

n

ln Rt

ln K t n

6.6272 [0.0003] – 0.9990 [0.3949] 1.8315 [0.1442]

ln Lt n

4.1369 [0.0075] 1.0695 [0.3641] – 0.6327 [0.5950]

nn

3.3038 [0.0400] 3.0175nn [0.0320] 0.3270 [0.8058] –

Long run

Joint long-and-short run causality

ECT t 1

ln Y t ; ECT t 1 n

0.0455 [ 4.1305] 0.1614n [ 3.5750] –

–

0.0379n [ 3.8132]

2.4401nn [0.0484]

7.3173n [0.0000] –

ln Rt ; ECT t 1 n

ln K t ; ECT t 1

ln Lt ; ECT t 1

–

7.9252 [0.0000] 3.7904n [0.0059] –

2.4920nn [0.0500] 3.9814nn [0.0043] –

4.5172n [0.0018]

6.6326n [0.0012]

3.6494n [0.0073]

8.6325 [0.0000] –

n

Show significant at 1%. Show significant at 5%.

nn

for a series accounted for by innovations from each of the independent variable over different time-horizons beyond the selected time period.

Table 8 reports that economic growth is explained 40.61 per cent by innovative shocks of renewable energy consumption. The contribution of capital and labour contribute to economic growth

584


Table 8 Variance decomposition method. Period

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Variance decomposition of ln Y t

Variance decomposition of ln Rt

Variance decomposition of ln K t

Variance decomposition of ln Lt

ln Y t

ln Rt

ln K t

ln Lt

ln Y t

ln Rt

ln K t

ln Lt

ln Y t

ln Rt

ln K t

ln Lt

ln Y t

ln Rt

ln K t

ln Lt

100.000 98.868 96.950 93.782 88.797 85.626 81.871 77.774 72.732 69.819 67.293 64.960 62.002 60.337 58.964 57.751 55.999 55.086 54.341 53.683

0.000 0.931 2.374 4.945 8.082 11.539 15.505 19.549 23.181 25.951 28.640 31.195 32.962 34.471 36.036 37.557 38.231 38.918 39.748 40.614

0.000 0.001 0.001 0.002 0.005 0.153 0.392 0.648 0.755 0.988 1.161 1.263 1.219 1.214 1.175 1.114 1.001 0.922 0.854 0.799

0.000 0.199 0.674 1.269 3.115 2.680 2.230 2.027 3.330 3.240 2.903 2.580 3.816 3.976 3.823 3.576 4.767 5.072 5.055 4.902

13.244 13.863 14.324 15.113 15.736 16.118 16.240 16.308 16.671 16.982 17.171 17.306 17.606 17.875 18.064 18.205 18.452 18.671 18.828 18.948

86.755 85.972 85.217 84.038 82.361 80.746 79.291 77.947 76.220 74.804 73.711 72.893 72.106 71.575 71.264 71.110 70.946 70.885 70.910 70.981

0.000 0.017 0.065 0.099 1.224 2.449 3.739 4.942 6.3734 7.505 8.402 9.056 9.585 9.878 10.016 10.034 9.956 9.817 9.653 9.480

0.000 0.146 0.393 0.748 0.677 0.685 0.729 0.801 0.734 0.707 0.714 0.743 0.702 0.671 0.656 0.649 0.644 0.626 0.6070 0.590

4.348 5.314 5.320 5.575 6.380 7.021 7.314 7.564 8.318 8.913 9.225 9.371 9.754 9.954 9.980 9.928 10.056 10.129 10.146 10.146

1.212 1.301 1.763 2.234 1.967 2.197 2.844 3.784 4.929 6.420 8.224 10.164 12.290 14.308 16.228 17.972 19.735 21.214 22.514 23.643

94.439 93.105 92.296 90.983 90.503 89.267 87.790 85.882 83.895 81.450 78.850 76.205 73.715 71.407 69.312 67.420 65.653 64.163 62.852 61.678

0.000 0.279 0.619 1.205 1.148 1.513 2.050 2.768 2.857 3.215 3.699 4.258 4.239 4.329 4.481 4.678 4.554 4.491 4.486 4.531

1.151 1.710 1.661 1.527 3.154 3.706 3.670 3.513 3.912 4.090 3.987 3.833 4.023 4.113 4.025 3.902 4.040 4.113 4.050 3.957

0.992 2.208 3.096 3.929 2.203 2.023 2.296 2.815 2.249 2.473 2.984 3.660 3.346 3.674 4.233 4.911 4.682 4.999 5.505 6.101

0.739 1.426 2.086 2.824 3.041 3.943 4.878 5.881 5.691 6.278 6.987 7.769 7.529 7.932 8.473 9.089 8.864 9.172 9.615 10.132

97.116 94.653 93.155 91.717 91.601 90.326 89.154 87.789 88.147 87.157 86.040 84.736 85.101 84.278 83.268 82.096 82.412 81.714 80.828 79.808

is minimal i.e. 0.799 per cent and 4.90 per cent, respectively. A 53.68 per cent of economic growth is contributed by factors outside the model such as technological advancements. Renewable energy consumption is contributed 70.98 per cent by its own shocks and 18.94 per cent by innovations stemming in economic growth. Capital and labour explain renewable energy consumption by 9.48 per cent and 0.59 per cent their innovative shocks. The contribution of renewable energy consumption is greater than economic growth to capital. A 61.67 per cent capital is contributed by its own innovations. The share of labour is negligible. Finally, economic growth, renewable energy consumption and capital do not seem to contrite much to labour through their innovative shocks. Almost, 80 per cent of labour is explained by its own innovations. Overall, results indicate bidirectional causality between economic growth and renewable energy consumption. The causal relationship is stronger from renewable energy consumption to economic growth. These findings are consistent with the VECM Granger causality analysis. This entails that causality results are robust and reliable for policy making purpose.

5. Conclusion and future research The present study investigated the relationship between renewable energy consumption and economic growth using Cobb–Douglus production function in case of Pakistan. The autoregressive distributed lag model or the ARDL bounds testing approach to cointegration applied to test the existence of long run relationship between renewable energy consumption, capital, labour and economic growth. The VECM Granger causality approach is used to examine the direction of causal relationship between these series and innovative accounting approach is used to test the robustness of the causality results. Our empirical results confirmed that the variables are cointegrated for long run relationship over the study period of 1972Q1– 2011Q4. The results indicated that renewable energy consumption raises economic growth. Capital and labour are also important factors of economic growth contributing to domestic production in the country. The rolling window results explain that renewable energy consumption, capital and labour positively impact on economic except few quarters. The causality analysis reveals that

feedback effect exists between renewable energy consumption and economic growth and same inference can be drawn for nonrenewable energy consumption. The current study can be augmented to investigate the relationship between renewable energy consumption and economic growth at provincial level. Sectoral analysis also can be conducted in key sectors such as agriculture, industrial and services. This may help the policy makers to formulating comprehensive energy policy for sustainable economic growth at provincial as well as sectoral levels.

References [1] TWAS. Sustainable energy for developing countries. Trieste, Italy: The Academy of Sciences for the Developing World (TWAS); 2008. [2] UNDP. Alliance for rural electrification: shinning a light for progress. Brussels: Renewable Energy House; 2009. [3] UN . A global green new deal for climate, energy, and development. Technical note, United Nations Department of Economic and Social Affairs; 2009. Available at: 〈http://sustainabledevelopment.un.org/content/documents/ cc_global_green_new_deal.pdf〉 (accessed on 10th February, 2014). [4] Dar M, Azeem R, Ramzan, MM. Impact of Energy Consumption on Pakistan’s Economic Growth. Int J Humanit Soc Sci Invent 2013;2(6):51–60. [5] Masud J. Clean energy development in Pakistan. ADB climate change and energy workshop. Bangkok: Hagler Bailly Pakistan; 2009. [6] Yazdanie, M. Renewable energy in Pakistan: policy strengths, challenges & the path forward. ETH Zurich; 2010. Available at: 〈http://www.cepe.ethz.ch/educa tion/termpapers/Yazdanie.pdf〉 (accessed on 8th February, 2014). [7] Technology Times. Energy crisis and the role of renewable energy. Pakistan’s first newspaper on S&T; 2013. Available at: 〈http://technologytimes.pk/eng lish-news.php?title=Energy%20crisis%20and%20the%20role%20of%20renew able%20energy〉 (accessed on 8th February, 2014). [8] Levin Institute. Renewable and alternative energy sources. The Levin Institute —The State University of New York; 2013. Available at: 〈http://www.globaliza tion101.org/renewable-and-alternative-energy-sources/〉 (accessed on 10th February, 2014). [9] Sheikh MA. Energy and renewable energy scenario of Pakistan. Renewable Sustainable Energy Rev 2010;14:354–63. [10] Khalil MS, Khan NA, Mirza IA. Renewable energy in Pakistan: status and trends. Pakistan Alternative Energy Development Board; 2005. [11] Ewing BT, Sari R, Soytas U. Disaggregate energy consumption and industrial output in the United States. Energy Policy 2007;35:1274–81. [12] Payne JE. On biomass energy consumption and real output in the US. Energy Sources Part B: Econ Plann Policy 2011;6:47–52. [13] Tiwari AK. A structural VAR analysis of renewable energy consumption, real GDP and CO2 emissions: evidence from India. Econ Bull 2011;31:1793–806. [14] Magnani N, Vaona A. Regional spillover effects of renewable energy generation in Italy. Working paper series, Department of Economics, University of Verona, WP Number: 12; 2011.


[15] Arifin J, Syahruddin N. Causality relationship between renewable and nonrenewable energy consumption and GDP in Indonesia. Econ. Finance Indone 2011;59:1–18. [16] Khan MA, Khan MZ, Zaman K, Arif M. Global estimates of energy-growth nexus: application of seemingly unrelated regressions. Renewable Sustainable Energy Rev 2014;29(1):63–71. [17] Mumtaz R, Zaman K, Sajjad F, Lodhi M, Irfan S, Khan M, et al. I. Modeling the causal relationship between energy and growth factors: journey towards sustainable development. Renewable Energy 2014;63:353–65. [18] Akhmat G, Zaman K. Nuclear energy consumption, commercial energy consumption and economic growth in South Asia: bootstrap panel causality test. Renewable Sustainable Energy Rev 2013;25:552–9. [19] Akhmat G, Zaman K, Shukui T. Balance of payment prospects and policy implications in light of energy sources in Pakistan. Energy Educ Sci Technol Part B: Soc Educ Stud 2013;5(4):655–86. [20] Sari R, Ewing BT, Soytas U. The relationship between disaggregate energy consumption and industrial production in the United States: an ARDL approach. Energy Econ 2008;30:2302–13. [21] Sadorsky P. Renewable energy consumption and income in emerging economies. Energy Policy 2009;37:4021–8. [22] Apergis N, Payne JE. Renewable energy consumption and economic growth: evidence from a panel of OECD countries. Energy Policy 2010;38:656–60 (2010a). [23] Apergis N, Payne JE. Renewable energy consumption and growth in Eurasia. Energy Econ 2010;32:1392–7. [24] Apergis N, Payne JE. Renewable and non-renewable energy consumptiongrowth nexus: evidence from a panel error correction model. Energy Econ 2012;34:733–8. [25] Apergis N, Payne JE. The electricity consumption-growth nexus: renewable versus non-renewable electricity in Central America. Energy Sources Part B: Econ Plann Policy 2012;7:423–31. [26] Larsson R, Lyhagen J, Lothgren M. Likelihood-based cointegration tests in heterogeneous panels. Econometr J 2001;4:109–42. [27] Payne JE. On the dynamics of energy consumption and output in the US. Appl Energy 2009;86:575–7. [28] Menegaki AN. Growth and renewable energy in Europe: a random effect model with evidence for neutrality hypothesis. Energy Econ 2011;33:257–63. [29] Mahmoodi M, Mahmoodi E. Renewable energy consumption and economic growth: the case of 7 Asian developing countries. Am J Sci Res 2011;35:146–52. [30] Apergis N, Payne JE. The renewable energy consumption-growth nexus in Central America. Appl Energy 2011;88:343–7. [31] Hamdi H, Sbia R. A note on electricity-growth nexus in Bahrain. Bull Energy Econ 2013;1(4):28–34. [32] Abdullah MI, Wei L, Anwar W, Bhutta US. Energy crisis and performance of industry of Pakistan: an empirical study. Bull Energy Econ 2013;1:21–7 (2013).

585

[33] Bowden N, Payne JE. Sectoral analysis of the causal relationship between renewable and non-renewable energy consumption and real output in the US. Energy Sources Part B: Econ Plann Policy 2010;5:400–8. [34] Zeshan M, Ahmad V. Energy consumption and economic growth in Pakistan. Bull Energy Econ 2013;1:8–20. [35] Shahbaz M, Zeshan M, Afza T. Is energy consumption effective to spur economic growth in Pakistan? New evidence from bounds test to level relationships and Granger causality tests Econ Modell 2012;29:2310–9. [36] Shahbaz M, Feridun M. Electricity consumption and economic growth empirical evidence from Pakistan. Qual Quant 2012;46:1583–99. [37] Ehrlich I. The deterrent effect of capital punishment: reply. Am Econ Rev 1997;67:452–8. [38] Ehrlich I. Crime, punishment and the market for offences. J Econ Perspect 1996;10:43–67. [39] Cameron S. A review of the econometric evidence on the effects of capital punishment. J Socio-econ 1994;23:197–214. [40] Layson S. Homicide and deterrence: another view of the Canadian time series evidence. Can J Econ 1983;16:52–73. [41] Pesaran MH, Shin Y, Smith R. Bounds testing approaches to the analysis of level relationships. J Appl Econometr 2001;16:289–326. [42] Haug A. Temporal aggregation and the power of cointegration tests: a Monte Carlo study. Oxford Bull Econ Stat 2002;64:399–412. [43] Shahbaz M, Lean HH. The dynamics of electricity consumption and economic growth: a revisit study of their causality in Pakistan. Energy 2012;39:146–53. [44] Pesaran MH, Shin Y. An autoregressive distributed lag modeling approach to cointegration analysis. In: Strom S, editor. Chapter 11 in econometrics and economic theory in the 20th century: the Ragnar Frisch Centennial symposium. Cambridge: Cambridge University Press; 1999. [45] Lütkepohl H. Structural vector autoregressive analysis for cointegrated variables. AStA Adv Stat Anal 2006;90:75–88. [46] Granger CWJ. Investigating causal relations by econometric models and crossspectral methods. Econometrica 1969;37:424–38. [47] Shahbaz M, Tang CF, Shabbir MS. Electricity consumption and economic growth nexus in Portugal using cointegration and causality approaches. Energy Policy 2011;39:3529–36. [48] Oh W, Lee K. Causal relationship between energy consumption and GDP: the case of Korea 1970–1999. Energy Econ 2004;26:51–9. [49] GoP. Economic survey of Pakistan. Islamabad, Pakistan: Planning commission wing; 2011–12. [50] World Bank. World development indicator (CD-ROM), Washington, DC; 2012. [51] Baum CF. A review of Stata 8.1 and its time series capabilities. Int J Forecast 2004;20:151–61. [52] Pesaran MH, Timmermann A. Market timing and return prediction under model instability. J Empirical Finance 2012;9:495–510.