The Linkage between CO2, FDI, Economic Growth and Value-Added: A European Perspective

This article aims to investigate the linkage among CO2 emissions, Foreign Direct Investment (FDI), economic growth, Gross Value Added (GVA) of different sectors namely agriculture, service, manufacturing, and resource extensive industries including construction sectors in four European regions Eastern Europe (EE), Southern Europe (SE), Northern Europe (NE) and Western Europe (WE). To do, this article uses the 3SLS simultaneous equation estimation during the period of 2000 to 2018. This study is the extension of seeing the challenges in policy implication in reducing CO2 emission in technologically rich economies. This article concluded that the causality among variables CO2 emission, economic growth, FDI, and all four sectors GVA is varied according to the regions. However, the CO2 emission has bidirectional causality with each industrial sector's GVA. and, recommended that reform in industrial structure can help to curve down the CO2 emission. Through the empirical literature, we observed the relation of CO2 emission with FDI, economic growth and, service value-added, agriculture value-added, and manufacturing value-added in different regions. In the primary outlook of investigation, empirical literature veried the association among CO2 emission, FDI, economic growth, and different sectors value-added in different regions of the world economies. This relation varies for a short-run and long-run period. In the essence of observed literature, we nd the literature gap for the European region. This study constitutes a debate on CO2 emission, FDI, and GVA in the service sector, manufacturing sector, construction and natural resources and, agriculture sectors in four-panel of European regions EE, SE, NE, and WE. This study is the extension of seeing the challenges in policy implication in reducing CO2 emission in technologically rich economies. This study involves the GVA of all sectors including wholesale, retail, trade transport, government, nancial, professional and personal services, education, health care real estate services, hotels and restaurants, agriculture, services, construction, mining and natural resources and, sub-grouped these sectors into four categories of GVA in technologically advanced economies.

Arminen, (2014) shows carbon emission, economic growth and FDI are positively moving along in the same direction in Sub-Saharan Africa in long run. Similarly,  nd CO2 emission may replicate the adverse effect on the host economy and have bidirectional causality between CO2 emission and FDI. Chandran & Tang, (2013) found a heterogenous linkage among FDI, economic growth, CO2 emission in transport sectors in ASEAN ve countries Malaysia, Singapore, Thailand, Indonesia, and the Philippines and suggested CO2 emission can be minimized through selectivity in FDI except for Singapore.
Results emphasize that policymakers should have to focus on FDI related to technology transfer and provide an incentive to high-tech sectors. Similarly, Almulali & Foon Tang, (2013) presented no relation between CO2 emission and FDI in ow while GDP growth rate nds the positive causal relationship between economic growth and CO2 emission in Gulf Cooperation Council (GCC) countries. The GCC countries have to promote FDI in ow since it is directly related to economic growth but should have encouraged FDI in technology-intensive industries. Omri et al., (2017) examine the causality among CO2, FDI, and economic growth global panel of 54 countries and three regional sub-panels Latin America and the Caribbean, Europe and Central Asia, and the Middle East, North Africa, and sub-Saharan Africa throughout . They nd that bidirectional causality between economic growth, CO2, and FDI in ow in all panels and has bidirectionality causality except North Asia and Europe. Mert et al., (2019) examine the relation between CO2 emission and FDI in the European region and, nd the long-run causality between CO2 emission and FDI in overall Europe. It implies the FDI in ow deteriorates the CO2 emission in Europe so, the European region should have to tighten the regulation and environmental law. Bengochea-morancho et al., (2001) nd in the European region there is a disparity between rest and advanced industrialized country. This disparity depends on the economic situation and industrial structure of each EU member state.
Therefore, there is heterogeneity in the literature to make uniform consciousness about the CO2 and FDI relations. Adaptation of better policy, management, and advanced technology may play a critical role to curve CO2 emission. Hence, in that case through FDI we can achieve zero or negative effect on CO2 emission.

CO2 and economic growth
The relation between CO2 emission and economic growth draws the attention of academic researchers, especially in the last two decades. Zakarya et al., (2015) examine the relation among CO2 emission, economic growth, and FDI in BRICS (Brazil, Russia, India, China, and South-Africa) nations using Granger causality and co-integration test from 1990 to 2012. They nd that GDP and FDI in ow are important factors to increase the CO2 emission and have positive unidirectional relations in long run. The BRICS nation has to increase the energy e ciency to increase productivity without harming the environment. Niu et al., (2011) analyzed the relation between CO2 emission, economic growth, and energy conservation in eight Asia Paci c countries using Vector Error Correlation Model (VECM), unit root and cointegration tests, result, shows the long-run equilibrium relationship among GDP and CO2 emission in developed countries while no such causality in developing countries. Further, the result argues that the CO2 emission per capita energy is lower in the Asia Paci c region compared to the developed nation however, the CO2 emission per unit energy consumption is higher. Jardón et al., (2017) investigated the cross-section dependency between CO2 per capita and economic growth in Latin American and Caribbean countries through cointegration and unit root test. They nd the mixed results of cross-dependency, exist inverse U-shaped curve, and rejected the Environmental Kuznets Curve (EKC) hypothesis.
Acaravci & Ozturk, (2010) used the autoregressive distributed lag (ARDL) method to nd the causal relationship between CO2 emission, economic growth, and energy consumption in the European region. The results show the long-run relation between GDP per capita and CO2 emission in Switzerland, Portugal, Italy, Iceland, Greece, Germany, and Denmark. However no long-run relation in Sweden, Norway, Luxemburg, Motherland, UK, France, Belgium, Finland Austria. Manta et al., (2020) estimated the nexus among CO2 emission, economic growth, energy use, and nancial development in Central and Eastern European Countries (CEEC) using VECM and Granger causality over the period of 2000 to 2017. The result shows that in the long run energy emission and CO2 emission have no impact on economic growth while in the short-run increasing in nancial development increases the CO2 emission and leads to enhanced economic growth. So, the European Union has to promote nancial development which will help the countries to reduce the CO2 emission, focus on the implementation of renewable and lower emission options. Kasperowicz, (2015) examines the relation between CO2 emission and economic growth in 18 European member countries using Error Correction Model (ECM) estimation and nd the negative long-run relation because technological advancement for the production facility, in the long run, reducing the CO2 emission in Europe. However, for short period increasing economic growth increases the CO2 emission because the fast production system extensively needs energy.
The relationship between CO2 and economic growth not showing uniform results in empirical literature review even within the European region this relation have a heterogeneous characteristic, which means it depends on the national characteristic (Choi et al., 2010) 2.3 CO2 and value-added in different sectors Jebli et al., (2020) investigate the relationship between services value-added, industrial value-added, renewable energy consumption, economic growth, and CO2 emission worldwide in four-panel groups of countries lower income, lower middle income, upper middle income, and high-income countries using GMM and Granger Causality test. Their results indicated that industries value-added and economic growth has a positive and signi cant impact on CO2 emission in the lower middle income countries while economic growth has a negative impact, similarly upper middle income countries economic growth have a negative impact while industries have a positive impact on CO2 emission and nally upper middle income countries economic growth have positive and signi cant while services value-added have a negative impact on CO2 emission. Further, they suggested that eco-friendly project uses of natural resources like wind, water, solar, hydrogen, and nuclear energy countries have to promote and raise the productivity to minimize the carbon emission, another solution is carbon taxation and subsidizing the ecofriendly project investment rely investors on e cient energy sources.
H. Liu & Fan, (2017) presented value-added accounting (production based and consumption based) system based on CO2 emission, Main objective of the study investigated the accountability of CO2 emission originating through human activity, within the boundary of economic bene ts principle. The study was based on bilateral trade of industrial production and variables of CO2 emission such as CO2 emission from transport, CO2 emission from the manufacturing industry and construction, CO2 emission from electricity and heat production and, CO2 emission from other sectors. Further, they used these variables to analyze the 3 panel groups based on income level; high income, low income, and middle-income group countries. They promote the CO2 emission-based accounting system based on "consumption" high consumption of good more responsibility and to reduced CO2 emission advanced country should have to help developing nation by technology transfer to achieve a reduction of CO2 emission target.
Alam, (2015) examines the value-added in uence on GDP in the service sector, agriculture sector, and manufacturing sectors in South Asian countries. Results show that value addition in the agriculture sector negatively in uences the CO2 emission while the service and manufacturing sector positively contributing to CO2 emission. Therefore, the research suggested dependency on the services sector is not the solution to reduce CO2 emission.
Similarly, Samargandi, (2017) analyses the KEC curve on Saudi-Arabia by considering the technology, sectors value addition in GDP, and volume of production, through the ARDL method. The result shows that the economic growth nurture the CO2 emission and, value-added growth in industrial and service sector foster the CO2 emission. However, the value addition in the agriculture sector reduces the CO2 emission, also, technological advancement help to reduces the CO2 emission without sacri cing the economic growth. Through the empirical literature, we observed the relation of CO2 emission with FDI, economic growth and, service value-added, agriculture value-added, and manufacturing value-added in different regions. In the primary outlook of investigation, empirical literature veri ed the association among CO2 emission, FDI, economic growth, and different sectors value-added in different regions of the world economies. This relation varies for a short-run and long-run period. In the essence of observed literature, we nd the literature gap for the European region. This study constitutes a debate on CO2 emission, FDI, and GVA in the service sector, manufacturing sector, construction and natural resources and, agriculture sectors in four-panel of European regions EE, SE, NE, and WE. This study is the extension of seeing the challenges in policy implication in reducing CO2 emission in technologically rich economies. This study involves the GVA of all sectors including wholesale, retail, trade transport, government, nancial, professional and personal services, education, health care real estate services, hotels and restaurants, agriculture, services, construction, mining and natural resources and, sub-grouped these sectors into four categories of GVA in technologically advanced economies.   Table-1 variable description. Test our objectives research model includes a system of simultaneous equations and decomposes CO2, FDI, L, TA, TE, U, I, GDPPC and GVAs in following econometric models can be presented as: (CO2) it = β 0 + β 1 FDI it + β 2 (U) it + β 3 (GDPPC) it + β 4 (L) it + β 5 (TA) it + β 6 (I) it + β 7 (ICT) it + β 8 Equation (1) is common for all models where the value of GVAs is A-GVA, ISCI-GVA, M-GVA, and S-GVA, the subscript i=1,….., N denotes the country, and t = 1, ……, T time period.
ln(CO2) it = β 0 + β 1 lnFDI it + β 2 ln(U) it + β 3 ln(GDPPC) it + β 4 ln(L) it + β 5 ln(TA) it + β 6 ln(I) it + β 7 ln(ICT) it + β 8 ln( ln(FDI) it = β 0 + β 1 lnCO2 it + β 2 ln(U) it + β 3 ln(GDPPC) it + β 4 ln(L) it + β 5 ln(TA) it + β 6 ln(I) it + β 7 ln(ICT) it + β 8 ln( ln(GDPPC) it = β 0 + β 1 lnFDI it + β 2 ln(U) it + β 3 ln(CO2) it + β 4 ln(L) it + β 5 ln(TA) it + β 6 ln(I) it + β 7 ln(ICT) it + β 8 ln( ln(GVAs) it = β 0 + β 1 lnFDI it + β 2 ln(U) it + β 3 ln(GDPPC) it + β 4 ln(L) it + β 5 ln(TA) it + β 6 ln(I) it + β 7 ln(ICT) it + β 8 ln  Table 2 and variable description is presented in Table 4. Further, descriptive statics is presented in Table 4. The log equations (3)(4) (5) and (6) Table-3. The simultaneous equation approach is used for the econometric estimation and the 3SLS estimation is used for the empirical estimation. Analyze the relationships among the variable is problematic due to the error correlation between variables. Testing the relation through 3SLS is ful ll our objective requirement, it estimates all parameters in the equation at once and allows correlation between the error terms across the various equations in the number of included equations to be analyzed. The 3SLS method is more robust because it addresses the correlation between error and endogeneity, this method is time tested and used by various researchers . The endogeneity problem is resolved by using a set of instrumental variables. It could be obvious that the endogeneity between the GVAs could occur so, while we considered one GVA for a single sector the other three GVA considered as an endogenous variable. All other right-hand side variables are considered as explanatory instrumental and exogenous to the system.
The LLC unit root test is used in this article to test the stationarity of the variables, shown in Table 5. All variables reject the null hypothesis and accept the alternate hypothesis of the unit root test, which implies all variables used in this article are stationary at the rst difference.    4 Results And Discussion development. In this article, there are four models created for each GVA for each region EE, SE, NE, and WE. The models estimated through the coe cient of equations 3 to 6 are presented in Table 6, 7, 8, and 9. We empirically tested the relation among CO2, FDI, GDP per capita, and GVAs of four industrial sectors.
The variable description is shown in Table 3. Descriptive statics of variables used in this study and Pearson correlation coe cient test are reported in Table 4.
Addressing the relation between CO2 emission and value-added in agriculture and its condition regional differences noticeable factor. Therefore, Table 9, presents the empirical estimation of the agriculture sector in EE, NE, and WE European regions where we nd the four-way linkage while in the SE region relationship of CO2 with others variable is insigni cant so, the three-way linkage is noticed. The signi cance of GVA-A models in Table 9 is in line with the of FDI in ow by 1% tends to raise the CO2 emission 3.23% in EE and 5% increase tends to 0.43% in WE, while in NE this association is negative so the increase of FDI in ow by 0.1% decrease the CO2 emission − 1.37%. Therefore, in the NE region, FDI in ow su ciently brings the technological advancement and innovation to reduce CO2 emission.
Further GDP per capita has a signi cant relation with all variables FDI, CO2, and GVA-A, except, FDI in ow and economic growth in the SE region. This insigni cant result is consistent with the (Alvarado et al., 2017). Another side, the relation between GDP per capita and CO2 emission has a negative association in all region because in the long-run development of new technology to reduce CO2 emission production system produces the same product with a lower level of carbon emission while, in the short-run GDP and CO2 emission is positive because the rapid increase in production can be achieved due to large intensive energy used compared to previous consumption of energy with lower production, capacity increase CO2 emission increase (Kasperowicz, 2015).
The empirical estimation for GVA-S sector industries is shown in Table 10. The EE region countries' service sector are more positively contribute to CO2 emission compare to the other three European panel-groups, which have an almost similar effect on carbon emission because according to Kołodziejczak, (2020) less a uent countries in Europe, a greater number of workforces related to the agriculture while in high a uent countries a smaller number of workforces related to the agriculture sector, so the high GVA is generated through the secondary sectors (industry) and tertiary sectors (service).     (-8.17) Source: author, Note:t statistics in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001.
Results related to the GVA-C are shown in Table 12, representing the resource extensive industries including construction, mining, electricity, water, and gas. The  Source: author, Note:t statistics in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001.  There is a bidirectional relation between CO2 emission and economic growth in EE, SE, NE, and WE regions for industrial sector GVA-A, GVA-S, and GVA-C, while there is no association between CO2 emission and economic growth when we considered the manufacturing GVA in SE region.
The relation between CO2 emission and FDI we nd the heterogeneous results. In the scenario of the GVA agriculture sector EE, NE, and WE region countries have bidirectional causality while SE region result shows no relation. For the case of the services sector, GVA bidirectional causality in NE and SE regions while no relation in EE and WE regions. Further manufacturing sector GVA there is bidirectional relation in EE, NE, and WE regions and no relation for SE region. Furthermore, with consideration of GVA-C sectors, we nd bidirectional relation in all sub-group EE, SE, NE, and WE regions.
The result of the relation between FDI in ow and economic growth is also heterogeneous. There is no relation between FDI and economic growth in EE, SE, NE, and WE region for the case of sector GVA. In the case of GVA of agriculture sector EE, NE, and WE bidirectional relation while no causality in SE region.
The results of this article help the policymakers to understand and grasp the complexity of the relation among the CO2 emission, FDI in ow, economic growth, and GVA of different sectors in the European region. Therefore, the ndings of this study have potential importance for policymaking to tackle the CO2 emission in different industrial sectors. This article concluded that the causality among this research variable depends on the regions, indicating that it is impossible to give universal policy recommendations. Therefore, to make pollution free economic development decision-maker (such as European Union) have to draft the policies which should have to consider regional factors and coherent with industrial sectors which will help to reduce the CO2 emission