CHEMICAL ENGINEERING TRANSACTIONS VOL. 63, 2018 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Jeng Shiun Lim, Wai Shin Ho, Jiří J. Klemeš Copyright © 2018, AIDIC Servizi S.r.l. ISBN 978-88-95608-61-7; ISSN 2283-9216 Investigating the Relationship on CO2, Energy Consumption and Economic Growth: a Panel Data Approach Sayed Kushairi B. Sayed Nordina,b,*, Sek S. Kunc a School of Mathematical Sciences, Universiti Sains Malaysia, Penang b Faculty of Manufacturing Engineering Universiti Teknikal Malaysia Melaka c School of Mathematical Sciences, Universiti Sains Malaysia, Penang sayedkushairi@utem.edu.my In this study, empirical analysis is conducted to reveal the relationship between three variables: energy consumption, GDP and CO2. The analysis is based on 13 oil importing countries and 11 oil exporting countries. The main objectives are to reveal the long-run relationship based on three different models using second generation panel unit-root and panel cointegration tests and to investigate the short-run relationship between pairs of variables using VAR Granger causality test. The panel unit root tests indicate that each variable is integrated of order one. Based on cointegration tests, the results reveal a long-run relationship in one of the models in both countries. The VAR Granger Causality shows evidence of a short-run relationship between the variables in both groups of countries. 1. Introduction Examining the relationships that describe how energy, carbon dioxide (CO2) emissions and economic growth are interrelated has inspired a high interest among researchers. As economy and population increases, the use of energy is expected to increase. In most countries, energy as a non-substitutable resource has been used extensively to promote their industrial productions and activities. The world has produced nearly 130 quadrillion BTU (British thermal unit) of energy from oil in 1997. Recent data from U.S. Energy Information Administration reported that United States used energy about 97.4 BTU in 2016. Dubai and Abu Dhabi are two cities benefitted from the oil and energy sector. There are consequences faced by many countries due to the progress of development. At this phase, roads, railways, factories, and facilities are developed progressively. This leaves a positive impact on economic growth but not necessarily to the environment in the long term. Energy is essential to stimulate the economic growth. However, inefficient use of energy contributes to the environmental degradation. Most of developing countries have progressively increased their use of energy and this has contributed to an increase in GHG emissions. Apart CO2, other dangerous gasses produced into the atmosphere are sulphur dioxide (SO2), particulate (pm) nitrogen oxides (NOx) (Qian and Zhang, 1998). CO2 emissions contribute for more than half of GHG emissions which are likely to be linked to climate change (The World Bank, 2007). Other negative effects caused by unsustainable development are global warming and climate change and resulting in the rise of sea levels due to the ice caps and glaciers, and extreme weather conditions such as droughts, massive floods and tornadoes. The most populated country like China consumes a huge amount of coal as an energy source. It is reported that in 2015, China and the United States had produced about 45 % of the total world CO2 emissions. The Intergovernmental Panel on Climate Change revealed that GHG emissions have increased estimated 1.6% yearly, while CO2 increased about 2 % yearly over the past three decades. Recently, the IMF's World Economic Outlook Database 2016 reported that these two countries are the top two economies in the world. To date, the relationship between energy consumption, environmental degradation and economic growth have been widely studied. However, there is no any precise answer and there is still no consensus on the relationship. Thus, in this paper, we extend the existing literature by examining the relationship and direction between these three variables for two different groups of countries (oil exporting and importing countries). This DOI: 10.3303/CET1863120 Please cite this article as: Sayed Kushairi B. Sayed Nordin, Sek S. Kun, 2018, Investigating the relationship on co2, energy consumption and economic growth: a panel data approach, Chemical Engineering Transactions, 63, 715-720 DOI:10.3303/CET1863120 715 investigation has been applied using second generation panel unit root tests to relax the restrictive assumption of cross-sectional independence. The objective is twofold: to examine the short-run relationship (one-way or bidirectional) for each pair of variables and to reveal the long-run relationship using different models. 2. Literature review The inconsistent results obtained from the past studies on the direction of causality have inspired many researchers to analyses and discuss the nature of the variables using different techniques. For example, Vaona (2012) revealed the different results after using different methods of analysis; Toda and Yamamoto and Johansen technique. Literature shows that various of economic modelling techniques have been applied to analyze the relationship between the variables. Environmental Kuznets Curve (EKC) had been applied employed in the analysis of environmental-economic. According to EKC, the curve describes the environmental degraded will initially increase with respect to the economic growth before the direction change once the desired level of economic is achieved, described as an inverted-U curve. The reverse in the direction means degradation starts to decline. Sun (1999) supports this theory and argues that the pollution can be avoided when the country's economy is developing. Moreover, Coondoo and Dinda (2002) argued it is impossible to protect the environment in both developed and developing countries when the economy grows rapidly. Tang and Tan (2016) revealed that CO2 emissions, energy consumption and economic growth are cointegrated and takes 11 years before reaching a long-run equilibrium. Using data from 106 countries, Antonakakis et al. (2016) concluded that this situation brings dilemma in many countries, between high economic growth rates and unsustainable environment. Grossman and Kruger (1991) suggested that the implementation of high technology and machining would lead to the reduction of pollution once advancement has been achieved. The reasons why the EKC theory has been criticized because the sample was taken from the middle-income countries (Latin America) that experiencing unequal economic distribution during the study period. Secondly, this theory has opposite results in many countries. On the other hand, Ang (2007) research produced both shapes of curves in distinct groups of countries. He found a U-shaped EKC in five Middle East countries, but an inverted U-shaped curve was found in other three countries. Studies conducted by Queshi et al. (2016) supported the inverted U-shaped EKC hypothesis. Bozkurt and Akan (2014) suggested that the environment will improve if the economy grows more rapidly. Rothman (1998) claimed that when countries become richer, they are more protective of the environment and the current economic growth is not sustainable due to the trade-off effect from production activities (Haywood, 1995). Hundie (2017) reported that in the long-run, energy consumption and economic growth have statistically significant impact on environmental degradation. Solow (1956) introduced an original growth model (neoclassical growth model). He believed an economy must reach stationary phase in which there is no more additional investment needed. This theory claimed that technological progress is crucial to achieving continuing economic growth. Stern (1999) in his Biophysical Models of the economy proposed that energy is the main source of production. Likewise, environmental degradation-economic growth relationship, the relationship between energy consumption and economic growth has become as an interesting topic being discussed by researchers too. Energy consumption is closely related to economic growth by providing inputs to industries help to stimulate economic development (Ang, 2007). Kahia et al. (2017) used data from 11 MENA Net oil importing countries for the period 1980 to 2012. They found evidence of a long-run and short- run relationship between GDP and energy. The results are positive and have significant elasticities. Hasanov et al. (2017) analyzed 10 oil exporting developing Eurasian countries from 1997 to 2014. They suggested policymakers to take action on inefficient of energy use that brings disadvantages on economic growth. Empirical studies revealed inconsistent results due to the different sample used, various model and the time period examined (Ozturk, 2010). Bozkurt and Akan (2014) investigated the long-run relationship of CO2 and energy consumption on economic growth in Turkey from 1960-2010. The study concluded that CO2 emission is negatively related to economic growth. As expected, energy consumption contributes a positive effect on economic growth. Moreover, Ozturk (2010) used panel data from 1971-2005 in 3 distinct groups of countries, low, middle and upper middle-income group of countries to study the correlation and causality. He failed to find evidence of a strong correlation between energy consumption and economic growth in all the income groups. For low-income countries, the results indicated that there is long-run Granger causality running from GDP to energy consumption while a bidirectional Granger causality between these variables in the lower and the upper middle-income countries. Apergis (2009) extended the study by Ang (2007) on panel framework. Based on data from 1971-2004 for six Central American countries and found that there is a positive relationship between energy and CO2 emissions. Rasli et. al. (2017) found evidence of a long-run relationship between energy consumption and economic growth. These results are similar to Naser (2015) and Chen et al. (2016). Naser studied four emerging economies named: Rusia, China, South Korea and India and Chen et al. used data from 3 different 716 groups of countries. Other studies detected a long-run relationship between these studies include Suocheng et al. (2011) and Amin et al. (2012). Among the recent longitudinal studies, which is conducted by Kang et al. (2016) involved thirty provinces of China, reported that economic growth and CO2 as an inverted-N trajectory. This result opposite to the traditional inverted-U and N-shaped relationship. Pala (2016) analyzed data from OECD countries; Pala detected the presence of a long-run relationship between economic growth and energy consumption and there is a two-way relationship between the variables in the short-run. Using data from 1970-2011, Kaka and Zervas (2013) found that the results are similar to that in Bozkurt and Akan (2014), while Karakas (2014) compared the relationship between OECD and non-OECD countries using data from 1990-2011. The study showed that there is a positive relationship between economic growth and CO2 emissions both groups of counties. Similarly, Wang et al. (2017) found relationship between these two variables although it is not linear in China. 3. Methods In this study, we used annual panel data-set from 1975 to 2013 for two group of countries; oil importing countries and oil exporting countries. The variables used are ENC (energy consumption in kg of oil equivalent per capita), CO2 (carbon dioxide emissions in metric tons per capita) and GDP (per capita in current US$) as the proxies for energy consumption, environmental degradation and economic growth respectively. The data are extracted from the World Bank Development Indicators. All data are transformed into natural logarithm form for consistency. The data consists of 13 and 11 countries from oil importing and exporting countries respectively. The countries are: (i) Oil importing countries: Belgium, China, France, Indian, Italy, Japan, Korea, Netherlands, Singapore, Spain, Thailand, United Kingdom and the United States (ii) Oil exporting countries: Algeria, Canada, Columbia, Mexico, Nigeria, Norway, Oman, Saudi Arabia, United Arab Emirates, United Kingdom, Venezuela. Three empirical models considered in this study are: Model 1: = + + 2 + Model 2: = + 2 + + Model 3: 2 = + + + where is a constant term, and are parameters to be estimated in the models. indicates to the cross- section, i.e. countries, is the time in years and ε is the error term. The analysis part begins with examining the cross-sectional dependency in fitting the panel data models. In this study, we employed Lagrange Multiplier (LM) test developed by Breusch-Pagan (1980). The LM statistic is given by = (1) where is the sample estimate of the pairwise correlation residuals = = ∑ ̂ ̂∑ ̂ ∑ ̂ (2) and ̂ is the estimate of . Secondly, before conducting cointegration analysis, stationary tests are essential to be conducted. The second -generation unit root test of Pesaran (2003) is used to check for the stationarity of the series. If the residuals are not serially correlated, the regression is employed for the country is given by ∆ = + , + + ∆ + (3) where = ∑ and ∆ = ∑ ∆ The null hypothesis is nonstationary series and the alternative hypothesis is stationary series. Once all variables are integrated of the same order, panel cointegration test can be performed to examine the existence of long-run relationship(s) among series. This study used the second generation cointegration test of Westerlund (2007). This test is based on the error-correction approach (ECM) which aims to examine whether an ECM does or does not have error correction. 717 ∆ = + − + ∆ + ∆ + (4) where is the error correction or speed of adjustment term. The variables are not cointegrated if = 0 . In contrast, if < 0 , then there is an error correction term, which implies that the variables are cointegrated. Finally, we proceed with short-run causality test using VAR Granger causality. The decision is whether to reject the null hypothesis of variable does not Granger cause variable versus alternative hypothesis of variable does Granger cause variable . 4. Results and findings To determine whether the variables are characterized by cross-sectional dependency, the Breush-Pagan LM test was applied. Table 1 shows that the test rejects the null hypothesis of no cross-sectional dependence for all the models in both countries. This means all the series are cross-sectionally correlated. Thus, the second- generation tests can be employed. Table 1: Results of Breush-Pagan LM Test for Cross-Sectional Dependency Model Importing Countries Exporting Countries 1 1190.433*** 678.239*** 2 1610.173*** 837.061*** 3 1106.146*** 225.137*** Notes: ***, ** and * indicate significance at 1%, 5% and 10% levels respectively The results of panel unit root tests using Pesaran (2003) reported in Table 2 and Table 3 indicate that all variables are integrated with first order, I(1) in both countries. This evidence of I(1) of all variables allow us to check the hypothesis of cointegration among energy consumption, GDP and carbon dioxide emission by employing second generation cointegration test; Westerlund (2007). This test uses four panel cointegration test statistics (Gt, Ga, Pa and Pt) based on Error Correction Model. For oil importing countries, there is a long- run relationship among variables as shown in Model 2 (see Table 4). Meanwhile, the long-run relationship exists in Model 1 for exporting countries (see Table 5). Table 2: Results of Panel Unit Root Test for Oil Importing Countries Variables without trend with trend Level ENC -0.810 -1.884 GDP -1.999 -2.191 CO2 -0.685 -1.840 First difference ENC -5.561*** -5.836*** GDP -4.712*** -4.987*** CO2 -5.344*** -5.475*** Table 3: Results of Panel Unit Root Test for Oil Exporting Countries Variables without trend with trend Level ENC -1.996 -2.633 GDP -1.386 -1.905 CO2 -2.024 -2.719* First difference ENC -5.780*** -5.938*** GDP -4.731*** -4.934*** CO2 -5.926*** -6.121*** In examining for the short-run causality, we used the Wald test. Table 6 reports that there is a bidirectional relationship between ENC and GDP and two one-way causalities (ENC to CO2 and GDP to CO2) in oil 718 importing countries. In oil exporting countries, the one-way short-run relationship is detected between CO2 to ENC and CO2 to GDP. Table 4: Results of Panel Cointegration Test for Oil Importing Countries Model Gt Ga Pt Pa Z-value 1 2.174 2.262 -0.273 -0.958 2 -2.003** 0.328 -1.054 -1.993** 3 5.594 2.961 3.296 1.970 Table 5: Results of Panel Cointegration Test for Oil Exporting Countries Model Gt Ga Pt Pa Z-value 1 -2.718*** 0.267 -5.641** -1.994 2 0.016 2.462 3.136 2.498 3 -2.980** 0.032 -1.099 -0.686 Table 6: Results of VAR Granger Causality Null Hypothesis Oil Importing Countries Oil Exporting Countries Chi-sq ENC does not Granger-cause CO2 48.8653*** 5.3685 GDP does not Granger-cause CO2 15.1884* 1.3596 CO2 does not Granger-cause ENC 7.4796 19.1920*** GDP does not Granger-cause ENC 30.3837*** 7.6086 CO2 does not Granger-cause GDP 10.3813 11.3417** ENC does not Granger-cause GDP 22.0642*** 1.7520 Notes: Lag order selected by the Akaike information criterion (AIC) and final predictor Error (FPE) 5. Conclusions This study examined the correlation between three variables, namely energy consumption, CO2 (proxy to environmental degradation) and GDP (proxy for economic growth) in oil importing and exporting countries using data from 1975-2013. The second-generation panel unit root and cointegration test were used in the analysis. The objectives are to reveal the long-run relationship in three different models and to examine the short-run relationship (one-way or bidirectional) in each pair of variables. The Breush-Pagan LM test suggests that there is cross-sectional dependence for all the models. This suggests that there is a cross-section effect in the series. All the data series used are integrated of order one. Cointegration test shows that there is a long- run relationship between the variables in both countries. For importing countries, empirical results show that carbon dioxide emissions affected by the energy consumption and economic growth in the short-run. In exporting countries, we found a bidirectional relationship between energy consumption and economic growth. In all, we can conclude that the three variables in both countries are interrelated in long-run and short-run. These findings intended to provide a deeper understanding of the interactions of energy consumption, environmental degradation and economic growth as an input in the process to develop effective policies. Effort must be taken to encourage industries to adapt machines and technologies that reduce pollution. It is recommended to oil importing countries to control their energy consumption and economic progress to protect the environmental quality. The renewal energy sources must be used as input in industrial development. References Amin, S., Shaikh S. F., Aroni, K. P., 2012, Causal Relationship among CO2 Emissions and Economic Growth in Bangladesh: An Empirical Study, World Journal of Social Sciences, 24, 273-290. Ang, J.B., 2007, CO2 emissions, energy consumption, and output in France, Energy Policy, 35, 4772– 4778. Antonakakis, N. Chatziantoniou. L., Fillis, G., 2016, Energy consumption, CO2 emissions, and economic growth: An ethical dilemma, Renewable and Sustainable Energy Reviews, 68, 808-824. 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, 312, 211–21. 719 Bozkurt, C., Akan, Y., 2014, Economic Growth, CO2 Emissions and Energy Consumption: The Turkish Case, International Journal of Energy Economics and Policy, 4(3), 484–494. Chen, P. Y., Chen, S. T., Hsu, C. S., Chen, C. C., 2016, Modeling the global relationships among economic growth, energy consumption and CO2 emissions, Renewable and Sustainable Energy Reviews, 65, 420- 431. Coondoo, D., Dinda, S., 2002, Causality between income and emission: a country group specific econometric analysis, Ecological Economics, 40(3), 351–367. Grossman, G.M., Krueger, A. B., 1991, Environmental impacts of the North American Free Trade Agreement, NBER. Working paper 3914. Hasanov, F., Bulut, C., Suleymanov, E., 2017, Review of energy-growth nexus: A panel analysis for ten Eurasian oil exporting countries, Renewable and Sustainable Energy Reviews, 73(C), 369-386. Kahia, M., Aïssa, M. S. B., Lanouar, C., 2017, Renewable and non-renewable energy use - economic growth nexus: The case of MENA Net Oil Importing Countries, Renewable and Sustainable Energy Reviews, 71(C), 127-140. Kang, Y., Zhao, T., Yang, Y., 2016, Environmental Kuznets curve for CO2 emissions in China: A spatial panel data approach, Ecological Indicators, 63, 231-239. Karakas, A., 2014, Economic Growth-CO2 Emission Relationship in OECD and Non-OECD Countries: A Panel Data Analysis for the Period between 1990-2011, The International Journal of Humanities & Social Studies’, 2(3), 57-62. Kebede, S., 2017, Modeling Energy Consumption, CO2 Emissions and Economic Growth Nexus in Ethiopia: Evidence from ARDL Approach to Cointegration and Causality Analysis, Munich Personal RePec Archive. Naser, H., 2015, Analysing the long-run relationship among oil market, nuclear energy consumption, and economic growth: An evidence from emerging economies, Energy, 89, 421-434. Pala, A., 2016, Which Energy-Growth Hypothesis is Valid in OECD Countries? Evidence from Panel Granger Causality, International Journal of Energy Economics and Policy, 6(1), 28–34. Ozturk, I., 2010, A literature survey on energy–growth nexus, Energy Policy, 34, 340–349. Pesaran, H. M., 2003, A Simple Panel Unit Root Test in the Presence of Cross Section Dependence. Mimeo, University of Southern California, USA. Qian, J., Zhang, K., 1998, China’s desulfurization potential, Energy Policy, 26, 354-351. Qureshi, M.I., Rasli, A.M. Zaman, K., 2016, Energy crisis, greenhouse gas emissions and sectoral growth reforms: Repairing the fabricated mosaic, Journal of Cleaner Production, 112, 3657-3666. Rasli, A.M., Qureshi, M.I., Isah-Chikaji, A., Zaman, K. Ahmad, M., 2017, New toxics, race to the bottom and revised environmental Kuznets curve: The case of local and global pollutants, Renewable and Sustainable Energy Reviews, 81, 3120-3130. Qureshi, M.I., Rasli, A.M., Awan, U., Ma, J., Ali, G., Alam, A., Sajjad, F. Zaman, K., 2015, Environment and air pollution: health services bequeath to grotesque menace, Environmental Science and Pollution Research, 22(5), 3467-3476. Rothman, D.S., 1998, Environmental Kuznets curves-real progress or passing the buck? A case for consumption-based approaches, Ecological Economics, 25(2), 177–194. Solow, R. M., 1956, A contribution to the theory of economic growth, The MIT Press, Boston, USA. Stern, D. I., 1999, Is energy cost an accurate indicator of natural resource quality?, Ecological Economics, 31,381-394. Stern, D. I., 2000, A multivariate cointegration analysis of the role of energy in the US macroeonomy, Energy Economics, 22, 267-283. Sun, J.W., 1999, The nature of CO2 emission Kuznets curve, Energy Policy, 27, 691-694. Li, F., Dong, S., Li, X., Liang, Q., Yang, W., 2011, Energy Consumption - Economic Growth Relationship and Carbon Dioxide Emissions in China, The Journal of Energy Policy, 39(2), 568-574. Tang, T.C., Tan, P.P., 2016, Carbon dioxide emissions, energy consumption, and economic growth in a transition economy: Empirical evidence from Cambodia, Labuan Bulletin of International Business & Finance, 14, 14-51. Wang, Z. X., Hao, P., Yao, P.Y., 2017, Non-Linear Relationship between Economic Growth and CO2 Emissions in China: An Empirical Study Based on Panel Smooth Transition Regression Models, International Journal of Environmental Research and Public Health, 14, 1-11. Westerlund, J., 2007, Testing for error correction in panel data, Oxford Bulletin of Economics and Statistics, 69, 709–748. 720