Nexus Between Energy Consumptions and CO2 Emissions in Selected Industrialized Countries

This study examines the nexus between energy consumptions and CO2 emissions in selected industrialized countries and selects the industrialized countries where three are industrialized and two are a newly industrial growing country out of five. The panel section implies the period from 1980 to 2014 for selected countries. The WDI is the main data source for the selected variables. This study conducts the FMOLS method, which is suggested by unit root and the Johansen co integrations test. The Granger causality also uses to estimate the causality in current variables. The Johansen Fisher co-integration test indicates the long-run association’s among the variable. The FMOLS technique estimates the marginal effect that industrialization and energy outlay is an authentic and significant influence on CO2 emission. Panel Ganger causality also shows a reliable result to estimate the causal relation. ARTICLE INFORMATION Received: 10 February 2020 Revised: 20 February 2020 Accepted: 25 February 2020 DOI: 10.31580/ijer.v3v1.1174


INTRODUCTION
Carbon dioxide emission is the major concern for industrialized countries, not only industrialized countries it's also a concern for developing or newly industrial growing countries because the environmental degradations vastly effects of those kinds of countries as lack of environmental management system and lack of controlling pollution. In generally energy consumption positively implies for industrial growth and enhancing GDP for a nation and another issue that energy consumption is one of the causes of increasing CO2 emissions. According to the UFCCC, the main concern of the Kyoto Protocol is reducing Green House Gases (GHG) that the causes of world climate change. Industrialized countries face environmental degradations like as sea levels increase, Cyclone, Flood, and Drought which are the primary causes of rising CO2 outlay and global warming. The CO2 emission is the most vital element of increasing the GHG as the amount 76.7% of total emissions. Shahbaz et al. (2016) predict that the developing countries will be faced a greater challenge from climate change and CO2 emissions. And he also added that industrialized countries increased CO2 emissions with energy consumption in both short-run and long run.
In the past several decades' energy economics open a new door for faster growth and development. Capital intensive countries highly used this kind of opportunities with respect to technology constraints. Modern technology enhances the faster growth in the world, and developing countries also take advantages using technology as a part and parcel of the economy Masih and Masih (1996). But there was an actual problem is associated with consuming energy and energy input in production. CO2 discharge is one of the vital problems for this kind of economic enhancement. It is difficult to show the way for eliminating carbon dioxide emission problem but research can make some recommendations and present the impacts of energy consumption on emissions. The environment loses balance with respect to time, and the energy consumption is one of the major reasons for the carbon emission.
To measuring the impact of energy consumption on CO2 outlay Lean and Smyth (2009) demonstrate that energy consumption has long-run relations to industrial growth and CO2 emissions. Alam and Huylenbroeck (2011), Perry et al. (2008), Hossain and Hasanuzzaman, (2012) and Sheinbaum et al. (2010) also determine the relations between CO2 discharge and energy consumption. In this study examined a panel data from 1980 -2014 for selected industrialized countries those are the USA, China, Malaysia, India and Bangladesh. We know that the USA, China, Malaysia already achieve high industrial progress and India also up-rowing nations for industrial development and Bangladesh achieved the high GDP growth rate with the best contributions of agriculture, service and the industrial sector. As an industrial country China, in 2014 CO2 emitted 10291926.88 kt and energy use as the amount of 2236.73 (oil equivalent per capita) with achieved 6.9% of GDP growth rate in 2017. China introduced as a high much energy purchasing country in the last several decades. Chinese government take the initiative of one belt one road policy to ensure the energy supply. This initiative is able to create inter-link among nation to nation. Malaysia also achieved a GDP growth rate of 5.89 and industrial value-added the percentage of GDP 38.78 in 2017. The industry and service sector contribute to the USA economy with 91.1% of total GDP. Soytas and Sari (2003) state energy consumption helps to GDP growth in the USA. The growing industrial country India also one of the high energy-consuming agnation, where GDP growth rate in 2017 was 6.68 and emitted CO2 as the amount of 2238377.17 in 2014. In this study consider that, the case of Bangladesh, because it needs a comparison among industrialized countries. At present Bangladesh attain 8.13 % of GDP growth rate and industrial with service sector contribution to GDP of 85.77%. Energy demand in Bangladesh gradually increases day by day with respect to industrial growth.
14 This perusal investigates the nexus between energy consumption and carbon dioxide emissions in the selected industrial countries over the time period of 1980-2014 with employing several econometric techniques with covering panel unit roots, cointegration. FMOLS technique follows unique rules to estimate the marginal effect of all parameters. Cross-sectional dependence allowing the Granger causality test, and try to make a comparison between industrial and newly industrial selected countries. In the next section of this study, section two for constructing objective, section three relates the summary of the literature review. Section four describes the methodology. Econometric result and findings are discussed in section five and lastly make a conclusion of the overall study.

MAIN OBJECTIVE
To know the relationship between energy consumption and carbon emission (CO2) in selected industrialized countries.

SPECIFIC OBJECTIVES
i. To determine the relationship within the energy consumption and carbon emissions by industrial development in selected countries ii. To find out the effects of energy consumption on industrialization in selected countries iii. To demonstrate the legitimacy of energy consumption and CO2 emissions in industrialized and newly industrialized countries like as Malaysia, India, Bangladesh

LITERATURE REVIEW
Nexus between energy consumption and CO2 emissions studied deeply in research from the last few decades in worldwide. Industrialization is one of the growing concerns for carbon emissions. Researcher says that energy extractions as a cause of emitting carbon. There are some several relevant studies are given in Table 1.
In the context of existing literature, most of those kinds of literature show the impacts of energy consumption on CO2 emissions in industrialized countries. In this study, also examine the effects of energy consumption on CO2 emissions by comparing industrialized country and newly industrial growing countries.

METHODOLOGY
Measuring the nexus between energy consumption and CO2 emissions in selected industrialized countries, we have used panel data with five countries as United State of America, P. R. China, Malaysia, India and Bangladesh. The secondary data are collected from WDI and this study conduct panel estimate from 1980 to 2014 with carbon dioxide emissions (CO2) as explained variable and energy consumption (Enc), industrialization (Ind) are explanatory variables.

Model Specifications
To examine the nexus between energy consumption and CO2 emissions for selected industrialized countries shows empirical findings to measuring the impact of CO2 emissions by energy consumption and industrialization. At the impact of energy consumption would be positively significant or negatively significant or insignificant for this study. Now we can specify the model for selected industrialized countries as following: Now, the econometric form of CO2 Model by taking Log on both sides: Where CO2 means carbon dioxide emissions, Enc shows the energy consumption (kg of oil equivalent) and Ind represents the industry including construction value-added on GDP. indicate the parameters which are to be estimated and μ_i is the error term. t and i mean the time period and individuals country for panel estimation.

PANEL UNIT ROOT TESTS
There are several techniques to determine the unit root for a specific data set. For estimating the panel unit root, most commonly used ADF, Levin Lin & Chu and another one is PP Fisher chi-squire technique. Unit root test needs to determine the integrating level or selecting the optimum lagged period.

ADF AND LEVIN LIN CHU TECHNIQUE
The simple unit root stochastic process follows the procedure Since, μ_(it )is an error term, describe it is stationary, meaning that the panel options are stationary after taking the first difference. Now ADF test has the following equations: Where; ε_it is an error component and ADF δY_(i,t-1) is the lagged selection criteria.

PESARAN AND SHIN TECHNIQUE
Shin and Im Pesaran (2003) argue the procedure of unit root testing with consideration of the ADF test. Im et al. (2003) interpreted W-statistics is normally distributed for the unit root test. These criteria are following the procedure: Where, is a simple t-test for determinate of , under condition is panel data series are no stationary and the task hypothesis presents stationary data. W statistics are normally distributed with zero mean value. Here, 'N' represents the total unit number across time, 'm' present the arbitrary constant.

PP -FISHER TECHNIQUE
A technique of nonparametric stochastic unit root processes has proposed by Maddala and Wu (1999). ADF test enhances PP-Fisher technique without lagged term and considers as error term has serial correlations.

( ) ∑ ( )
Where; first difference showed by ∆, k is the arbitrary constant and Y_(i,t-1) represent the optimum time-lagged for developing the null hypothesis (H0) is data has a unit root and task hypothesis (H1) is data has no unit root process.
Johansen -Fisher (J-F) Panel Cointegration technique Maddala and Wu (1999) and Pederoni (1999) state that J-F test set to emphasis across the unit and also panel consider as time-variant. J-F test involved several kinds of a technique like as Trac test, Rank test and Max-eigenvalue. Cointegrated time-lagged would be selected by using J-F technique as the following equations; Z and X variable assure that the I(1) order integration. Time and country represented by an i-t symbol. e_(it )is an error component at equation no -(11).

FMOLS TECHNIQUE
Look into the co-integration prostrate relevance among the variables. OLS sometimes gives the spurious output in econometric analysis. FMOLS helps to eliminate spurious situations in a dynamic econometric model. Philips and Hanson (1990) first introduce FMOLS technique. Pedroni (2000) has demonstrated the FMOLS works to prostrate relevance amongst variables and FMOLS helps to explore the E-G method. Christopoulos and Tsionas (2003)

PANEL GRANGER CAUSALITY TECHNIQUE
This methodology considers the relation between carbon dioxide (CO2) and energy consumption (ENC), industrialization (IND). Cointegration relation among the variables has expressed by causality technique (Granger, 1969;and Engle and Granger, 1987). Hffmann et al. (2005) have found that ENC, IND has to the reason of rising CO2. Dritsakis and Stamatiou (2016), Christopoulos and Tsionas (2003) also measure the causality by using a similar kind of variables.

ECONOMETRIC RESULT AND DISCUSSIONS Panel Unit Root Tests
This study demonstrate the most important and most useable test for make decision for panel unit root testing, here we use the Philips Parron -PP -Fisher Chi-square, Shin W-stat and Pesaran, ADFtesting, Levin, Lin & Chu t method to estimate the unit root for panel estimation. Pesaran et al. (2001), Pesaran and Shin (1999), and Levin and Lin (1993) both of them implies several methods for measuring the nature of data and select the optimal lag difference for the zero to mean value for residuals and a constant variance known as stationary. The results are showing in Table 2.
The panel unit root tests results are presented in Table (2). This result shows that the test statistics for taking log levels for each of variables like as CO2, Ind and Enc, are statistically insignificant at the level. Here the result says that the logs levels result of all three variables is non-stationary at level with considering the panel estimation. In this case, when we take the first difference of three variables for unit root testing, all four tests reject the H0 (variable is non-stationary or have a unit root) for each variable at the 1% or 5 % level and accept the alternative hypothesis as there have no unit-roots or series have stationary.

Johansen -Fisher Panel Cointegration Test
For this study, we have three variables, such as CO2, Ind, and Enc. Maddala and Wu (1999) helps to construct JF-cointegration test for the estimating co-integration. Unit root test ensures that at the level both variables are the non-stationary but after assuring the data series to 1st difference, and then they become stationary from the nonstationary series. So we can apply Johansen Fisher panel cointegration test and our variables and data, cointegrated of the same order, after 1st difference both are stationary, So data is integrated by I(1). The empirical results are described in Table 3. Table 3 shows that maximum eigenvalue and trace test with measuring the cointegration of variables, where both of the tests accept the alternative hypothesis by rejecting the null hypothesis. The obtained probability value is less than 5% and it is significant to accept the task (H1) hypothesis and reject the null (H0) hypothesis. It means that at most two variables are cointegrated, so we can say that all variables are cointegrated. Our three variables are integrated into the long run all these variables are moved together, meaning that they have long run associations. When they are long-run associations then we can use the panel Fully Modified Ordinary Least Square (FMOLS) method and we can also use the DOLS method. This study used the FMOLS method to estimate the econometric regression model.

Model selections: Hausman Test Result
Here this model assuming that H0 as 'the random effect is better for this model ' to examine the result shows that the test summary as the probability value bellow 5% so we can reject the H0 hypothesis and accept the task (H1) hypothesis as the fixed effect if better is accepted Arellano(1993). Hausman (1978) and Taylor and Hausman (1981) they mechanized test or method for selection of a better effect. The Hausman Test estimated result is given in Table 4.
The scenario explained that, the fixed effect is better for this model because Hausman test explains that probability value is significant with less or equal than 5% so, Hausman test gives the decisions as accept the alternative hypothesis by rejecting the null hypothesis, that is the fixed effect is better for our estimated model.

Panel Fully Modified Ordinary Least Squares (FMOLS) Long-run Estimates
According to Johansen and Fisher cointegration test shows that long-run associations or cointegration in selected explained and explanatory variables and we estimate our model based on the panel version of (FMOLS) Proposed by Philips and Hansen (1990) and Pedroni (2001) they explore DOLS and FMOLS as the measuring the marginal effects for long term cointegration. Table 5 illustrates the panel FMOLS results for the estimated model. For the panel of selected industrialized countries at all, each of the coefficients or marginal effect represents the values of independent variables to influence the dependent variables. The expected result or sign are considered as significant with 5 % level sometimes ten percent and one percent is better. Form table 4, the results indicate that one an average of 1% raising in energy expense is amalgamated with a carbon discharge raises at 0.338 percent. And carbon dioxide emissions increased 0.414 percent when there one an average, 1 % increase in industry value added in GDP in the economy. For the result analysis, the empirical result by Ang (2007) and Payne and Apergis (2009) have found that CO2 per capita increases while the selected panel studied nations per capita energy consumption increased.

Panel FMOLS Long-run Estimates for Individual Country
This study conducted with the selected industrialized countries and we find the specific lack of previous studies and this empirical result shows how to explain this, we try to compare the industrialized countries and newly or growing industrial countries impact on CO2 emissions with the specific regressor variables. The empirical FMOLS long-run estimates result for studied Individual countries is given in Table 6.
Current study contact that CO2 is explained variable. Energy consumption & industrializations are the explanatory variables for selected industrialized countries. Now, explain the results for individual countries. From Table 6

Panel Granger Causality
Granger causality shows the cause, among the variables to measure both bidirectional or unidirectional relationship. According to Granger (1969) causality explains the relation as the causes by each other in variables. The following Table 7 is describing the empirical result and decision of panel and individuals country causality: The panel Granger causality shows that the energy use is the cause of rising CO2 emissions as a unidirectional causal relationship, industry and CO2 have the bidirectional causal relationship and Industry and energy consumption also shows the bidirectional causal relationship. Table 8 represents the Granger causality for individuals studied countries as known as selected industrialized countries. The first column shows the null hypothesis for selected variables and significant probability values are considered as 1%, 5%, 10% to accept the alternative hypothesis by rejecting the null hypothesis. In this study consider a panel data from 1980-2014 for selected industrialized countries. For Bangladesh, we see that is CO2 emission is the cause of energy consumption and shows the unidirectional relationship. Industry and CO2 have a bidirectional in India (Tiwari, 2011) and CO2 emissions and energy wreckage are bidirectional in the largest energy-consuming country China. In Malaysia have the unidirectional cause on industrialization to CO2 emissions and energy outlay. And the last one is that industrialization has unidirectional relation on energy use in the USA.
As discussed above, the economic theory and practices accept the current findings. In the context of macroeconomic indicators like as export, import, foreign direct investment, growth, innovation, and distributions are largely depends on economic activity and production. Industrial growth influence to massive productions and distributions, as these purpose energy resources are a vital element to the massive industrial productions and CO2 is the effect of energy use and industrialization. The marginal effect of FMOLS method shows that the newly growing industrial countries have largely CO2 emitted than industrialized countries, because of industrialized countries take initiatives to control CO2 emission. It is almost difficult to control emission for the growing industrial countries. The Environmental Kuznets curve theory also supports the current findings with the argument of Soytas et al. (2006), Lean and Smyth (2013), Dogan and Turkekul (2016) who have found the similar results.

CONCLUSION
This perusal examined the nexus between energy consumption and CO2 emissions for selected industrialized countries as a panelbased study from 1980 to 2014. Here the selected countries are US, China, Malaysia, India & Bangladesh. This study finds the inferiority of previous studies and it's already mentioned that we want to compare with industrialized countries and newly industrial growing countries like Bangladesh, India. From the overall study shows that Bangladesh and India almost similar case for industrialization and energy consumption on CO2 emission and positively influenced by energy consumption were the effect is highly significant than other nations. Nain et al. (2015), Rahman and Kashem (2017) support the findings. This perusal demonstrated that Bangladesh faced energy consumption has positively growing significant impact on CO2 and India also faced energy consumption has definitive significant influence on CO2 emissions but both of the nations have negatory for industrialization on CO2 emissions because we know Bangladesh and India is a developing country, and also they are labour intensive country, not capital based in nature and the result also find that the USA, China, and Malaysia both have a weighty positive influence of energy purchases on CO2 emissions and industrialization is the cause of carbon emission as respect to economic enhancement and development supported by Dogan and Seker (2016). The empirical inference is showing that variables are cointegrated and have a longrun relationship, panel causality shows that energy uses is unidirectional to CO2 outlay and industrialization and CO2 bidirectional, industrialization and energy consumption also bidirectional causal relationship.
The contributions of this study will be added extra value in current literature and further research will take place in regional comparison. In the practical sense, this study creates a great awareness to consume energy and policymaker will be considering it when the industrial activities increase with respect in time. This result is valid with the existing literature that's found in measuring the nexus between energy consumptions and CO2 emissions in selected industrialized countries. This study finds that energy outlay, population, and GDP growth has statistically cabalistic and deterministic relations of CO2 emissions in studied countries. Here this study ensures that all the variables are performed in trend and intercept for the series. Schwarz information criteria (SIC) used to identify the optimum lagged length as the decision making for the stationarity. Here this test assumes that "data series has a unit root" as a null hypothesis. These results are statistically significant at 1% (5%) (10%) level and it's mentioned by *** (**) (*) gradually.