International Journal of Energy Economics and Policy Vol. 1, No. 4, 2011, pp.140-149 ISSN: 2146-4553 www.econjournals.com Energy-Growth Causality: Asian Countries Revisited Evan Lau, corresponding author Department of Economics, Faculty of Economics and Business, Universiti Malaysia Sarawak, 94300 Kota Samarahan Sarawak, Malaysia. Tel: +6082-582430. Email: lphevan@feb.unimas.my Xiao-Hui Chye Department of Economics, Faculty of Economics and Business, Universiti Malaysia Sarawak, 94300 Kota Samarahan Sarawak, Malaysia. Email: celinechye@yahoo.com Chee-Keong Choong Centre for Economic Studies, Faculty of Business and Finance, Universiti Tunku Abdul Rahman (Perak Campus), Jalan Universiti, Bandar Barat, 31900 Kampar, Perak, Malaysia. E-mail: choongck@utar.edu.my ABSTRACT: Understanding the impact of energy consumption on economic growth is an important consideration in the formulation of both energy and environmental policies. Motivated by this development, this paper empirically re-examines the direction of causality and the sign (in the panel sense) between energy consumption (EC) and the gross-domestic product (GDP) for seventeen selected Asian countries. Results reveal long-run stable equilibriums in these countries, while the EC brings about a positive impact on GDP. Causality runs from EC to GDP in the short-run, while the long-run causal linkage exists from GDP to EC. This indicates that energy is a force for economic growth in the short-run, but in the long-run, the EC is fundamentally driven by economic growth. Efficient coordination and cooperation towards the implementation of energy conservation policies to support sustainable economic development should be in the regional agenda. Keywords: Energy consumption; Panel analysis; Economic growth; Asian countries JEL classifications: Q43, C32 1. Introduction Energy consumption has steadily increased over the past few decades in Asian countries due to the population increment and industrial expansion1. Energy consumption is expected to increase to 159.3 quadrillion BTU in 2015, 187.8 quadrillion BTU in 2020, 217.0 quadrillion BTU in 2025, 246.9 quadrillion BTU in 2030 and 277.3 quadrillion BTU in 2035. The average annual percentage change from 2007 to 2035 in Asia is 2.8 percent, which is higher than other regions, such as the Middle Eastern countries (2.2 percent), Central and South America (1.8 percent), and Africa (1.8 percent) (EIA, 2010; Table 1, 9). The major users of energy were China and India, who continue to lead the world in relation to economic growth and energy demand growth. Together, China and India accounted for about 10 percent of the world's total energy consumption in 1990 and 20 percent in 2007 (EIA, 2010). China and India’s other significant increases include a fast-paced growth in population, rapid economic growth and industrial expansion into other areas of the Asian region. 1 According to EIA (2008), Asia’s total primary energy consumption in 1990 was 47.4 quadrillion British Thermal Units (BTU). This number doubled to about 127.1 quadrillion BTU in 2007. Energy-Growth Causality: Asian Countries Revisited 141 The episodic energy crisis, coupled with depleting energy sources, environmental costs and high- energy consumption, has forced governments around the globe to more intently monitor and manage energy markets (ECSSR, 2004). Growing concerns had attracted the interest of the government in Asian countries. These measures include cooperation for energy conservation and the efficient usage of energy policies. In this context, the long-run relationship between energy consumption and economic growth has been a lively topic of empirical assessment. In the energy economics literature, the direction of causality as to whether the adoption of energy savings inhibits or stimulates economic growth has been a much debated matter2. Understanding the impact and causality patterns of energy consumption on economic growth is an important consideration in the formulation of both energy and environmental policies. Accordingly, Squalli (2007, 1193-4), Payne (2010a, 54 – 55) and Ozturk (2010, 340-341) provide excellent descriptions of four hypotheses related to the relationship between energy consumption and economic growth. Briefly, the four patterns include: (1) The “growth” hypothesis, where the causality runs from energy consumption to growth. This pattern exists in energy dependent countries (Yu and Choi, 1985 for the Philippines, Masih and Masih, 1996 for India, Asafu-Adjaye, 2000 for India and Indonesia, Soytas and Sari, 2003 for Turkey, France, Japan and Germany, Lee, 2005 for a panel of eighteen developing countries and Tsani, 2010 for Greece); (2) The “conservation” hypothesis, where GDP Granger-causes energy consumption (Kraft and Kraft (1978) for the United States (US), Abosedra and Baghestani (1989) for the US, Cheng and Lai (1997) for Taiwan, Cheng (1999) for India, Ang (2008) for Malaysia and Zhang and Cheng (2009) for China). Recently, Phung (2011) found positive unidirectional causality running from GDP to energy consumption in Vietnam. For this purpose, policies such as the reduction in greenhouse emissions designed to reduce energy consumption and waste may not adversely affect real GDP. (3) The “neutrality” hypothesis views the absence of Granger-causality between energy consumption with GDP (Yu and Hwang (1984); Altinay and Karagol (2004); Halicioglu (2009) and Payne (2010a). (4) The “feedback” hypothesis suggests that energy consumption and GDP are interdependent and support the existence of bi-directional causality (Hwang and Gum, 1991; Yang, 2000; Oh and Lee, 2004; Climent and Pardo, 2007; Apergis and Payne, 2009; Ozturk and Acaravci 2010). The literature has not come to a general agreement on the nature of causal relationships between energy consumption and economic growth. In this context, policies aiming at the gradual curtailing of energy need to consider the potential causal linkages between economic growth and energy consumption. Motivated by this development, the goal of this study is to empirically re-examine the direction of causality and sign (in the panel sense) between energy consumption (EC) and real GDP for seventeen Asian countries. Once the causality is ascertained, appropriate energy development policies in these countries can be adopted. As such, the structure of the rest of this paper is as follows. A brief and intuitive account of the econometric methodology employed is provided in Section 2, before discussing the results in detail in Section 3. Some policy implications and conclusions are made in Section 4. 2. Econometric Modeling 2.1 Panel Unit Root and Stationary Tests The first step in the estimation of dynamic panels is to test whether the variables at hand contain unit roots. Studies that have used joint panel unit root tests include Maddala and Wu (1999, MW), Hadri (2000, HADRI), Levin et al., (2002, LLC) and Im et al. (2003, IPS). The null hypothesis in all joint panel unit root tests, except the HADRI test, is that the panel series has a unit root (non- stationary). Unlike the augmented Dickey Fuller (ADF) test, the HADRI test is similar to the Kwiatkowski, Phillips, Schmidt, and Shin (KPSS – based LM) statistic, which has a null hypothesis of 2 Literature about the energy-growth causality was coined from the seminal work of Kraft and Kraft (1978). Since then, impressive volumes of papers were dedicated to this genre. Ozturk (2010) and Payne (2010a) conducted an excellent survey, while Payne (2010b) and Narayan et al. (2010) investigated the electrical consumption and growth literature. On the Asian side, Yu and Choi (1985), Masih and Masih (1996, 1998), Asafu-Adjaye (2000) Soytas and Sari (2003) and Lee and Chang (2008) were among the champions. International Journal of Energy Economics and Policy, Vol. 1, No. 4, 2011, pp.140-149 142 level (trend) stationarity and an alternative of difference stationarity in the panel. A comparison of the results obtained from the wide range of panel unit root tests can provide some insight into the stationarity properties of the data. If both procedures fail to reject the null hypothesis (or if both reject), we have mixed results and can only conclude that the data are not informative enough. On the other hand, if an ADF type panel unit root test rejects the null and the KPSS type test fails to reject it, we have greater confidence that the series under consideration is in fact stationary. As these panel- based unit root tests are becoming common in the literature, interested readers may refer to their original articles for a more comprehensive discussion. 2.2 Panel Cointegration We then proceed to examine whether there exists any long-run equilibrium relationship between the variables under investigation. We resort to Pedroni (1999, 2001, 2004) and Kao (1999) panel cointegration tests. Pedroni considers seven different statistics, four of which are based on pooling the residuals of the regression along the within-dimension (panel test) of the panel. The other three are based on pooling the residuals of the regression along the between-dimension (group test) of the panel. The within-dimension tests take into account common time factors and allow for heterogeneity across countries. The between-dimension tests are the group-mean cointegration tests, which allow for the heterogeneity of parameters across countries. Kao (1999) proposed Dickey Fuller (DF) and ADF-type tests for it , where the null is specified as no cointegration. In this study, we only report the ADF-type test. The details of these tests are discussed in Appendix 1. 2.3 Panel Fully Modified OLS (FMOLS) Estimates To obtain the long-run estimates for the cointegrating relationship (the coefficients of EC), we adopt the panel group mean Fully Modified OLS (FMOLS), following the work by Pedroni (2000). The FMOLS procedure accommodates the heterogeneity that is typically present, both in the transitional serial correlation dynamics, and in the long-run cointegrating relationships. The FMOLS estimator is described in detail in Appendix 1. 2.4 Granger Causality Tests To test for panel causality, we estimate a panel based vector error correction model (VECM) with a dynamic error correction term based on the analysis in Holtz-Eakin et al. (1988, 1989). The empirical models are as follows: ititipit m p ip m p pitipjit ECTECGDPGDP 111 1 12 1 111       (1a) ititipit m p ip m p pitipjit ECTGDPECEC 212 1 22 1 212       (1b) where:  is the lag operator and p denotes the lag length. The specification in Equation 1 allows for testing the causality direction. For example, in the short-run, the EC does not Granger cause GDP where :0H ip12 0 for all i and p , while i1 0 in Equation (1a) 3. The rejection implies that EC  GDP, supporting the growth hypothesis. Similar analogous restrictions and testing procedures can be applied in testing the hypothesis that GDP does not Granger cause movement in EC, where the null hypothesis H0: 022 ip for all i and p , while i2 0 in Equation (1b). 3 The F-test or Wald 2 of the explanatory variables (in the first differences) indicates the short-run causal effects ( 012 ip for all i and p ), while the long-run causal ( i1 =0) relationship is implied through the significance of the lagged ECT, which contains the long-run information. Energy-Growth Causality: Asian Countries Revisited 143 3. Empirical Results 3.1 Data sources Annual data from 1980 to 2006 for the 17 Asian countries were utilized for the study4. Per capita total primary energy consumption (EC) data were obtained from the International Energy Annual 2006 of Energy Information Administration (EIA). Real GDP data were obtained from the World Development Indicators (WDI) 2008 of the World Bank. All variables were transformed into the logarithmic form. 3.2 Panel Unit Root and Stationary Results The results, made available upon request, illustrate that the series of the variables are of an I(1) process, as the pooled data are stationary in their first differences. These results enable us to test the cointegration among EC and GDP. 3.3 Panel Cointegration Results From the panel cointegration results in Table 1, we find strong evidence to reject the null hypothesis of no cointegration for all seven statistics provided by Pedroni (1999, 2001, 2004). Similarly, we reject the null hypothesis of no cointegration using the ADF-type statistics from the Kao (1999) panel cointegration tests, suggesting that that the two-dimensional model for the Asian countries is cointegrated and moves together in the long-run. Thus, we find that GDP and EC are cointegrated in the multi-country panel setting for the sample period. Table 1. Panel Cointegration Tests Results A: Pedroni Residual Cointegration test Panel cointegration statistics (within-dimension) Panel v-statistic 4.246 (0.000) Panel PP type -statistic -2.212 (0.035) Panel PP type t-statistic -2.318 (0.027) Panel ADF type t-statistic -4.525 (0.000) Group mean panel cointegration statistics (between-dimension) Group PP type -statistic 2.187 (0.037) Group PP type t –statistic 4.122 (0.000) Group ADF type t- statistic 2.706 (0.010) B: Kao Residual Cointegration test ADF 2.513 (0.006) Notes: The number of lag truncations used in the calculation of the seven Pedroni statistics is lag 5. Probability values are in parenthesis. 3.4 Panel FMOLS Estimates Having established cointegration in the long-run, we estimate the long-run parameters of the model by using the FMOLS technique. The FMOLS corrects the standard OLS for bias induced by the endogeneity and serial correlation of the regressors (Lee, 2005). The elasticity of energy consumption is important for understanding the past and assessing future economic dynamics. It represents the weights with which the marginal relative changes of the energy consumption contributes to the relative change of output (Lee et al., 2008). Table 2 reports the results of the long-run estimates for seventeen Asian countries and the panel estimates based on Pedroni’s group mean FMOLS estimator. The panel results of the regression equation with GDP as the dependent variable illustrate that the coefficient of the EC is positive and 4 The Asian countries included Bangladesh, Bhutan, Brunei Darussalam, China, Hong Kong, India, Indonesia, Japan, Korea, Malaysia, Maldives, Nepal, Pakistan, Philippines, Singapore, Sri Lanka, and Thailand. International Journal of Energy Economics and Policy, Vol. 1, No. 4, 2011, pp.140-149 144 statistically significant at the 5 percent significance level. A one percent increase in energy consumption leads to a 0.21 percent increase in GDP for these seventeen Asian countries. This positive coefficient on EC implies that more energy results in greater outputs, as suggested by Lee, (2005), Narayan and Smyth (2008), Lee and Chang, (2008) and Ozturk, (2010). Turning to the country specific evidence, the results also indicate a positive and significant relationship between EC and GDP for all countries. The elasticity estimates range from 0.10 (Hong Kong) to 0.94 (Philippines). The results suggest that the EC contributes most to the Philippines’ output, whereas it contributes least to Hong Kong’s output. Having inelastic coefficients on EC suggests that the vulnerability of energy prices would not have a significant impact on the consumption patterns in these countries, as it would be considered necessities for the society as a whole. Table 2. FMOLS Estimates Countries Energy Consumption Bangladesh 0.920 (10.660)* Bhutan 0.220 (33.840)* Brunei 0.330 (9.305)* China 0.370 (2.080)* Hong Kong 0.100 (27.710)* India 0.140 (10.250)* Indonesia 0.660 (5.790)* Japan 0.390 (3.220)* Korea 0.160 (9.410)* Malaysia 0.170 (11.100)* Maldives 0.380 (1.780)* Nepal 0.150 (4.790)* Pakistan 0.230 (38.580)* Philippines 0.940 (12.200)* Singapore 0.340 (7.750)* Sri Lanka 0.180 (14.620)* Thailand 0.200 (5.630)* Panel estimates 0.210 (44.330)* Notes: The values in parentheses are the t-statistics. Asterisk (*) shows significance at 5 percent level. All variables are transformed into logarithm form prior to estimation. 3.5 Panel Granger Causality Test Results Once the long-run estimates have been determined, we turn to the causality linkages. The empirical results presented in Table 3 illustrate that the coefficient of the error correction term (ECT) is not statistically significant in the GDP equation, indicating the absence of a long-run causality relationship running from EC to GDP. However, we note the existence of a significant short-run causal relationship running from EC to GDP, since the estimated coefficients of the explanatory variables are statistically significant. The short-run results are supported by Asafu-Adjaye (2000), Soytas and Sari (2003), Lee (2005), Narayan and Smyth (2008) and Tsani (2010), who established evidence of a short-run Granger causality running from EC to GDP. Table 3. Panel Granger Causality Test Results Dependent Variables ΔGDP ΔEC ECT 2-statistics (p-value) Coefficient t-ratio ΔGDP - 19.661 (0.001) 0.001 0.614 ΔEC 2.108 (0.715) - -0.002 -4.056 Notes: Parenthesized values are the probability of rejection of Granger non-causality. Δ is the first different operator. Estimations are based on the pooled data for 1980-2006 and 17 Asia Countries (N=17, T=27) with four lags. All variables are transformed into logarithm form prior to estimation. Energy-Growth Causality: Asian Countries Revisited 145 On the other hand, we find evidence of the existence of a long-run relationship running from GDP to EC, in which the coefficient of the error correction term (ECT) is statistically significant in the EC equation. This result illustrates that energy consumption is determined by economic growth; supporting the conservation hypothesis. This pattern is similar to results from developing countries (Cheng and Lai, 1997; Cheng, 1999; Mahadevan and Asafu-Adjaye, 2007; Ang, 2008 and Ozturk, 2010). 4. Concluding remarks Using panel estimation for seventeen Asian countries, this paper empirically examines the relationship between energy consumption and the gross domestic product (GDP). We find that the variables were in a stationary fashion in their first differences or were in an I(1) process. The panel cointegration results reveal a long-run equilibrium relationship among the two variables. The results of the FMOLS show that the energy consumption variable has a positive sign. This indicates that an increase in GDP would lead to a greater use of energy. From the Granger causality test, there is a short-run unidirectional causal relationship running from energy consumption to GDP. This implies that in the short-run, energy consumption leads to economic growth, since the economies in these 17 Asian countries are energy-dependent economies. Additionally, in the long-run, GDP Granger causes energy consumption for the panel. This provides additional evidence in support of the proposition that energy consumption is a result of economic activity, rather than being an essential input to production. In the short-run, the implementation of energy conservation policies might lead to a significant, but temporary, negative impact on economic growth in these Asian countries. However, economic development in the Asian countries is less dependent on energy in the long-run. Cooperation for energy conservation policies among the Asian countries would be an imperative move that would not harm GDP. Proactive agendas of research and development on renewable technologies in response to depleting supplies of energy sources would be another avenue that could be used to improve energy transportation facilities and infrastructure development to improve delivery efficiency. Niu et al. (2011)5 argued that developing countries may benefit from their developed nations counterparts, where they may fetch advanced technology and capital to facilitate efficient energy use, while reducing energy consumption and carbon emissions. Efforts have also been made in pursuit of more environmentally-friendly and resource-saving societies to promote energy efficiency in the face of concern about the effects of global warming for the Asian region (Chang, 2010; Lean and Smyth, 2010 and Li et al., 2011). With the recent experience of unprecedented high levels of energy prices, depleting energy sources and international initiatives such as Kyoto protocol, the commitment needs to be established to facilitate successful energy conservation policies. Acknowledgement The authors are grateful to Universiti Malaysia Sarawak (UNIMAS) for the supporting this research through grant No. FPI (F01)/01/2011(01). We are thankful for the comments and suggestions of the participants at the International Conference on Applied Economics (ICOAE 2011), Perugia, Italy, 25 - 27 August 2011. All remaining flaws are responsibilities of the authors. References Abosedra, S., Baghestani, H., 1989. New Evidence on the Causal Relationship between United States Energy Consumption and Gross National Product. Journal of Energy Development 14, 285 – 292. Altinay, G., Karagol, E., 2004. Structural break, unit root, and the causality between energy consumption and GDP in Turkey. Energy Economics 26, 985-94. Ang, J., 2008. Economic Development, Pollutant Emissions and Energy Consumption in Malaysia. Journal of Policy Modeling 30, 271-278. Apergis, N., Payne, J.E., 2009. Energy consumption and economic growth in Central America: evidence from a panel cointegration and error correction model. Energy Economics 31, 211-216. 5 Accordingly, as per Table 2 (pp. 2123) in the paper, four developed countries (Australia, New Zealand, Japan and South Korea) have the technological advancement therefore, their energy efficiency is higher comparatively to the four developing countries (China, India, Indonesia and Thailand). International Journal of Energy Economics and Policy, Vol. 1, No. 4, 2011, pp.140-149 146 Asafu-Adjaye, J., 2000. The relationship between energy consumption, energy prices, and economic growth: time series evidence from Asian developing countries. Energy Economics 22, 615-625. Chang, C. C., 2010. A multivariate causality test of carbon dioxide emissions, energy consumption and economic growth in China. Applied Energy 87, 3533–3537 Cheng, B.S., 1999. Causality between energy consumption and economic growth in India: an application of cointegration and error correction modeling. Indian Economic Review 34, 39-49. Cheng, B.S., Lai, T.W., 1997. An investigation of cointegration and causality between energy consumption and economic activity in Taiwan. Energy Economics 19, 435-444. Climent, F., Pardo, A., 2007. Decoupling factors on the energy-output linkage: the Spanish case. Energy Policy 35, 522-528. Emirates Center for Strategic Studies and Research (ECSSR), 2004. Asian Energy Markets: Dynamics and Trend, Emirates Center for Strategic Studies and Research. Energy Information Administration (EIA), 2008 International Energy Outlook 2008. Washington, D.C.: Energy Information Administration. Energy Information Administration (EIA), 2010. International Energy Outlook 2010. Washington, D.C.: Energy Information Administration. Hadri, K., 2000. Testing for stationarity in heterogeneous panel data. Econometrics Journal 3, 148- 161. Halicioglu, F., 2009. An Econometric Study of CO2 Emissions, Energy Consumption, Income and Foreign Trade in Turkey. Energy Policy 37, 1156–1164. Holtz-Eakin, D., Newey, W., Rosen, H., 1988. Estimating vector autoregressions with panel data. Econometrica 56, 1371-1395. Holtz-Eakin, D., Newey, W., Rosen, H., 1989. The revenues-expenditure nexus: evidence from local government data. International Economic Review 30, 415-429. Hwang, D.B.K., Gum, B., 1991. The Causal Relationship between Energy and GNP: The Case of Taiwan Province of China. Journal of Energy and Development 16, 219-226. Im, K.S., Pesaran, M.H., Shin, Y., 2003. Testing for unit roots in heterogeneous panels. Journal of Econometrics 115, 53-74. Kao, C., 1999. Spurious Regression and Residual-Based Tests for Cointegration in Panel Data. Journal of Econometrics 90, 1-44. Kraft, J., Kraft, A., 1978. On the relationship between energy and GNP. Journal of Energy and Development 3, 401-413. Lean, H.H., Smyth, R., 2010. CO2 emissions, electricity consumption and output in ASEAN. Applied Energy 87, 1858–1864 Lee, C.C., 2005. Energy consumption and GDP in developing countries: a cointegrated panel analysis. Energy Economics 27, 415-427. Lee, C.C., Chang, C.P., 2008. Energy consumption and economic growth in Asian economies: a more comprehensive analysis using panel data. Resource and Energy Economics 30, 50-65. Lee, C.C., Chang, C.P., Chen, P.F., 2008. Energy-income causality in OECD countries revisited: the key role of capital stock. Energy Economics 30, 2359-2373. Levin, A., Lin, C.F., Chu, C.S.J., 2002. Unit root tests in panel data: Asymptotic and finite sample properties. Journal of Econometrics 108, 1-24. Li, F., Dong, S.C., Li, X., Liang, Q., Yang, W.Z., 2011. Energy consumption-economic growth relationship and carbon dioxide emissions in China. Energy Policy 39, 568–574 Maddala, G.S. and Wu, S., 1999. A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics 61, 631-652. Mahadevan, R., Asafu-Adjaye, J., 2007. Energy consumption, economic growth and prices: a reassessment using panel VECM for developed and developing countries. Energy Policy 35, 2481-2490. Masih, A.M.M., Masih, R., 1996. Energy consumption, real income and temporal causality: results from a multi-country study based on cointegration and error-correction modeling techniques. Energy Economics 18, 165-183. Masih, A.M.M., Masih, R., 1998. A multivariate cointegrated modeling approach in testing temporal causality between energy consumption, real income, and prices with an application to two Asian LDCs. Applied Economics 30, 1287-98. Energy-Growth Causality: Asian Countries Revisited 147 Narayan, P.K., Smyth, R., 2008. Energy consumption and real GDP in G7 countries: New evidence from panel cointegration with structural breaks. Energy Economics 30, 2331 - 2341. Narayan, P.K., Narayan, S., Popp, S., 2010. Does Electrical Consumption Panel Granger Cause GDP? A New Global Evidence. Applied Energy 87, 3294–3298. Niu, S., Ding, Y., Niu, Y., Li,Y., Luo, G., 2011. Economic growth, energy conservation and emissions reduction: A comparative analysis based on panel data for 8 Asian-Pacific countries. Energy Policy 39, 2121–2131. Oh, W., Lee, K., 2004. Causal relationship between energy consumption and GDP revisited: the case of Korea 1970-1999. Energy Economics, 26, 51-59. Ozturk, I., 2010. A Literature Survey on Energy-Growth Nexus. Energy Policy, 38: 340 – 349. Ozturk, I., Acaravci, A., 2010 The causal relationship between energy consumption and GDP in Albania, Bulgaria, Hungary and Romania: Evidence from ARDL bound testing approach. Applied Energy 87, 1938–1943. Payne, J.E., 2009. On the dynamics of energy consumption and output in the US. Applied Energy 86, 575-577. Payne, J.E., 2010a. Survey of the international evidence on the causal relationship between energy consumption and growth. Journal of Economic Studies 37, 53-95. Payne, J.E., 2010b. A Survey of the Electrical Consumption-Growth Literature. Applied Energy 87, 723-731. Pedroni, P., 1999. Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and Statistics 61, 653-670. Pedroni, P., 2000. Fully modified OLS for heterogeneous cointegrated panels. Advances in Econometrics 15, 93-130. Pedroni, P., 2001. Purchasing power parity tests in cointegrated panels. The Review of Economics and Statistics 83, 727-731. Pedroni, P., 2004. Panel cointegration: Asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econometric Theory 20, 597-625. Phung T.B., 2011. Energy Consumption and Economic Growth in Vietnam: Threshold Cointegration and Causality Analysis, International Journal of Energy Economics and Policy, 1, 1–17. Soytas, U., Sari, R., 2003. Energy consumption and GDP: causality relationship in G-7 and emerging markets. Energy Economics 25, 33-37. Squalli, J., 2007. Electricity consumption and economic growth: bounds and causality analyses of OPEC countries. Energy Economics 29, 1192-1205. Tsani, S.Z., 2010. Energy consumption and economic growth: A causality analysis for Greece. Energy Economics 32, 582–590. Yang, H.Y., 2000. A note on the causal relationship between energy and GDP in Taiwan. Energy Economics, 22, 309-317. Yu, E.S.H., Choi, J.Y., 1985. The causal relationship between energy and GNP: an international comparison. Journal of Energy and Development 10, 249-272. Yu, E.S.H., Hwang, B., 1984. The relationship between energy and GNP: further results. Energy Economics 6, 186-90. Zhang, X.P., Cheng, X.M., 2009. Energy Consumption, Carbon Emissions and Economic Growth in China. Ecological Economics 68, 2706–2712. International Journal of Energy Economics and Policy, Vol. 1, No. 4, 2011, pp.140-149 148 Appendix Panel Cointegration and Fully Modified OLS Pedroni panel cointegration test There are in all seven panel cointegration tests. Detailed description of the formulae for the seven panel cointegration statistics, are given in Pedroni (1999: 660-661). A. Within-dimension (panel tests): a) Panel -Statistic b) Panel Phillip-Perron (PP) type  -Statistics c) Panel Phillips-Perron (PP) t -Statistic (non-parametric) d) Panel Augmented Dickey Fuller (ADF) t -Statistic (parametric) B. Between-dimension (group tests): e) Group Phillip-Perron (PP) type  -Statistics f) Group Phillips-Perron (PP) t -Statistic (non-parametric) g) Group Augmented Dickey Fuller (ADF) t -Statistic (parametric) These seven statistics are based on the estimated panel cointegration regression residuals of the likely cointegrating vector titiiiti ECtGDP ,,1,   (A.1) varying across countries, thus permitting full heterogeneity ( i ), fixed effects ( i ) and individual specific deterministic trends ( ti ) across individual members of the panel Pedroni (1999) shows that under appropriate standardization based on the moments of vector of Brownian motion function, each of these statistics converges weakly to a standard normal distribution when both the T and N of the panel grow large. The standardized distributions for the above mentioned seven panel and group statistics can be expressed in the form of )1,0(, N Ne TN     (A.2) where NTe is the respective panel/group cointegration statistic and  and  are the expected mean and variance of the corresponding statistics. They are computed by Monte Carlo stochastic simulations and tabulated in Pedroni (1999, Table 2). Kao panel cointegration test Unlike Pedroni test, Kao (1999) test specifies cross-section specific intercepts and homogeneous coefficients on the first-stage regressors. In this case, we specified the panel regression model as itititit zxy   '' (A.3) where ity and itx are I(1) and non cointegrated. For itz = }{ i Kao (1999) proposed DF and ADF-type unit root tests for it where the null is specified as no cointegration. The DF-type test can be calculated from this regression of: ititit   1ˆˆ (A.4) while the augmented version of the pooled specification: itp p j jitjitit      1 1 ˆˆˆ (A.5) where  ˆ~~ˆ ititit xy  and . ~ iit yyy  The OLS estimate of  and the t-statistics are given as          N i T t it N i T t itit 1 2 2 1 2 1 ˆ ˆˆ ˆ    and      s t N i T t it    1 2 2 1ˆ1ˆ . Energy-Growth Causality: Asian Countries Revisited 149 In this case,      N i T t ititNT s 1 2 2 1 2 .ˆˆˆ 1  Under the null of no cointegration, Kao (1999) shows that following the statistics:   2.10 31ˆ NTN DF     (A.6) NtDF pt 875.125.1  (A.7)   4 0 4 2 0 ˆ5 ˆ36 ˆ ˆ3 * 3 1ˆ              NTN DF (A.8) 2 0 2 2 2 0 0 ˆ10 ˆ3 ˆ2 ˆ ˆ2 ˆ6 *                vN t t DF (A.9) where 12 ˆˆˆˆ  xxyxyy and 12 0 ˆˆˆˆ  xxyxyy . For ADF can be constructed as: 2 0 2 2 2 0 ˆ10 ˆ3 ˆ2 ˆ 0ˆ2 ˆ6                N t ADF ADF (A.10) where ADFt is the t-statistics of  in equation A.5. Fully Modified OLS Estimates Following Pedroni (2000, 2001), we consider the following cointegrated system for panel data of ititiiit XY   (A.11) ittiit eXX  1, (A.12) where, Ni ,...,2,1 countries over the time period of Mt ,...2,1 . In addition, )',( ititit XYZ  ~ )1(I and )',( ititit e  ~ )0(I with covariance matrix of , '0 iiii  where i 0 is the contemporaneous covariance, i is the weighted sum of autocovariances while 'iii LL in which iL is the lower triangular decomposition of i . For simplicity, we assume that Y = GDP while X [EC] of A.1 in this study. The panel FMOLS estimator for coefficient  is given as:                          N i iit T t itit T t ititFM TYXXXXN 1 * 1 1 1 21* ˆ)()(  (A.13) where   it i i itit X L L YYY  22 21* ˆ ˆ and  .ˆˆˆ ˆ ˆˆˆ 02222 22 210 2121 ii i i iii L L  Likewise, the associated t-statistics for the estimator can be constructed as:    N i iFMFM tNt 1 ˆ 2/1 ˆ * , *  where .)(ˆ)ˆ( 2/1 1 21 110 * ,ˆ* ,             T t iitiiFM XXt iFM  