Energy use, CO₂ emission and foreign direct investment: Is there any inconsistence between causal relations?

Abstract

In this study, the causal relations between inward foreign direct investment (FDI)-energy use per capita and inward FDI-CO₂ emission per capita were analyzed and the inconsistency between the causal relations was investigated via bootstrap-corrected panel causality test and cross-correlation analysis. In this direction, data from 76 countries including the period of 1980–2009 was processed. No supportive evidence was found for changing causal relations to country group which was classified into income level. The findings indicated that while the pollution haven hypothesis was supported for Mozambique, United Arab Emirates and Oman, the pollution halo hypothesis was supported in the case of India, Iceland, Panama and Zambia. For other countries, energy use and CO₂ emission were neutral to inward FDI flows in aggregated level. Furthermore, this study urged that increased (decreased) energy use due to the inward FDI flows did not necessarily mean an increase (decrease) in pollution level, and vice versa. For policy purpose, FDI attractive policy should be regulated by taking into account this possibility.

Keywords

CO₂ emissions / energy consumption / liberalization

Introduction

Since the constitution of Kyoto Protocol aspiring “stabilization of greenhouse gas concentrations in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system”, the decrease of the greenhouse gas emission by not damaging the economic performance has become a primary issue. As a remedy, the use of renewable energy source instead of fossil energy types and encouragement of energy saving technologies as a supplementary policy were suggested. In addition, whether foreign direct investment (FDI) helps to decrease energy intensity and greenhouse gas emission in the host country were argued in the literature.

It is widely agreed that FDI provides economic benefits to the host countries by providing capital, foreign exchange, technology and by enhancing competition and access to foreign markets. Furthermore several alternative economic rationales supporting relationships between FDI and environmental pollution have been proposed in the literature. According to the pollution haven hypothesis, weak environmental regulation in a host country may attract inward FDI by profit-driven companies eager to circumvent costly regulatory compliance in their home countries. Gradually, these countries might develop a comparative advantage in pollution-intensive industries and become ‘havens’ for the world’s polluting industries [1]. Second, according to the pollution-halo hypothesis, in applying a universal environmental standard, multinationals engaging in FDI tend to spread its greener technology to their counterparts in the host country. Besides their advantages in technology and management, foreign companies are usually larger than domestic firms in developing countries. Accordingly, a foreign company has deeper pockets for investment in research and development, as well as environmental management systems. As domestic firms learn from and copy the foreign firms, the environmental performance of the developing country will converge towards developed country levels [2].

According to these scenarios, there may be several causal relations between pollution and FDI. If weak environmental regulation in a host country causes an increase in inward FDI flow, there should be a causal nexus from pollution to FDI, since pollution level can be accepted as a proxy variable for environmental regulation. In addition, induced FDI by increasing pollution level may lead to an increase in pollution level in the host country. So according to the pollution haven scenario, there should be bidirectional causal nexus between pollution and FDI. However, the pollution haven hypothesis assumes that FDI flow is sensitive to environmental regulation. If other determinants of FDI flows such as scarcity of inputs for production, political instability, higher perception of country risk, high transportation cost and insufficient domestic demand lead to reverse effect on inward FDI flows, low environmental regulation may not cause increase in inward FDI. So, inward FDI flows may be neutral to pollution level. According to the pollution halo scenario, even if pollution level causes an increase in inward FDI flow, induced FDI may not lead to an increase in the pollution level in the host country. Inward FDI may lead to a decrease in pollution level by using energy saving and green energy technologies.

The literature about the relation between FDI and environmental performance seems to fall into two categories. While the studies in the first category emphasize the relationship between inward FDI and energy intensity, studies in the second category analyze the nexus between FDI and some proxies of pollution such as CO₂ and SO₂. However, since an increase in energy intensity leads to a rise in pollution, the studies in the two categories are highly related to each other and both depend on the same theoretical basis. Thus, the findings of the studies in the two categories should include a joint evaluation. Relevant empirical literatures are summarized in Table 1.

According to Table 1 there is no consensus on the causal relations both between FDI-pollution and FDI-energy intensity. The results emerging from these empirical literatures are mainly based on empirical evidence from firm level analyses, time series analyses and panel data analyses. While panel data analyses using aggregated data such as Hoffman et al. [16], Sadorsky [9], Pao and Tsai [10] and Kim and Adilov [23] have not been able to confirm a robust relation between energy intensity-FDI or pollution-FDI, firm level analyses such as Blackman and Wu [3] and Fisher-Vanden et al. [6] have revealed that FDI has a reducing impact on energy intensity. In addition, Hoffman et al. [16] have found that causal relations change to country groups which were classified to per capita income level via panel Granger causality approach. Moreover, using time series analyses, Eskeland and Harrison [5], Merican et al. [18], Lee [21], Tang [7] and Chang [1] have found changing relations from one country to another. Since panel data framework increases size of the sample, it produces more reliable and statistically powerful results than cross-sectional and time series analyses. However, there may be heterogeneity in estimated parameters for each individual (country) of panel. Besides, heterogeneity problem may lead to bias estimation. Furthermore, presence of cross-sectional dependence may cause misleading inferences. So, selected panel data method should take into account heterogeneity and cross-sectional dependence problems.

Another problem in the literature is the conflicting findings about the relation between FDI-energy intensity and FDI-pollution. For example, Eskeland and Harrison [5] have found that FDI leads to energy saving for Mexico. Cole and Elliott’s [15] findings have supported the pollution haven hypothesis for the aforementioned countries. However, a certain number of studies in the literature such as Blackman and Wu [3], Hübler and Keller [8], Sadorsky [9] and Herrerias et al. [13] have assumed that if FDI had led to energy saving, a decrease in per capita pollution should have existed. The conflicting results in different studies might have emerged due to the differences in methods, time periods or variables. Therefore, the two line of the literatures should be evaluated together to achieve robust findings. If there are conflicting results, decreasing pollution level via energy saving increased by inward FDI may not be straightforward.

The contribution of the present empirical study is threefold. First, in order to remedy the econometric issue in the estimation due to the presence of cross-sectional dependence and heterogeneity problems, this study uses the novel panel causality approach developed by Kónya [26]. To find the traces of achieved causal relations, cross correlation was employed as well. Second, this study analyzes the causal relations between both FDI-energy use and FDI-pollution with the aim of cross-check. The third contribution is the sample size which includes 76 countries. To the best of the author’s knowledge, this study is the first which produces country specific findings by using such a wide range of countries. It is expected that these three important aspects will provide more accurate information for policy makers.

Model and data

In this study to test Granger causality, a system of equations was used. The equations could be divided into two groups as Eq. (1) and Eq. (2).

(1)\{\begin{array}{l}{\mathrm{EC}}_{\mathrm{1}t}={\alpha }_{\mathrm{11}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\beta }_{\mathrm{11}i}{\mathrm{EC}}_{\mathrm{1}t-i}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\delta }_{\mathrm{11}i}{\mathrm{FDI}}_{\mathrm{1}t-i}}}\hfill \\ {\displaystyle +{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\phi }_{\mathrm{11}i}{\mathrm{GDP}}_{\mathrm{1}t-i}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\mathit{\gamma }}_{\mathrm{11}i}{\left({\mathrm{CO}}_{\mathrm{2}}\right)}_{\mathrm{1}t-i}}+{ϵ}_{\mathrm{11}t}},\hfill \\ ⋮\hfill \\ {\mathrm{EC}}_{Nt}={\alpha }_{\mathrm{1}N}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\beta }_{\mathrm{1}Ni}{\mathrm{EC}}_{Nt-i}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\delta }_{\mathrm{1}Ni}{\mathrm{FDI}}_{Nt-i}}}\hfill \\ {\displaystyle +{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\phi }_{\mathrm{1}Ni}{\mathrm{GDP}}_{Nt-i}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{1}}}{\mathit{\gamma }}_{\mathrm{1}Ni}{\left({\mathrm{CO}}_{\mathrm{2}}\right)}_{Nt-i}}+{ϵ}_{\mathrm{1}Nt},}\hfill \end{array}

and

(2)\{\begin{array}{l}{\mathrm{FDI}}_{\mathrm{1}t}={\alpha }_{\mathrm{21}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\beta }_{\mathrm{21}i}{\mathrm{EC}}_{\mathrm{1}t-i}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\delta }_{\mathrm{21}i}{\mathrm{FDI}}_{\mathrm{1}t-i}}}\hfill \\ {\displaystyle +{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\phi }_{\mathrm{21}i}{\mathrm{GDP}}_{\mathrm{1}t-i}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\mathit{\gamma }}_{\mathrm{21}i}{\left({\mathrm{CO}}_{\mathrm{2}}\right)}_{\mathrm{1}t-i}}+{ϵ}_{\mathrm{21}t},}\hfill \\ ⋮\hfill \\ {\mathrm{FDI}}_{Nt}={\alpha }_{\mathrm{2}N}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\beta }_{\mathrm{2}Ni}{\mathrm{EC}}_{Nt-i}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\delta }_{\mathrm{2}Ni}{\mathrm{FDI}}_{Nt-i}}}\hfill \\ {\displaystyle +{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\phi }_{\mathrm{2}Ni}{\mathrm{GDP}}_{Nt-i}}+{\displaystyle \sum _{i=\mathrm{1}}^{p_{\mathrm{2}}}{\mathit{\gamma }}_{\mathrm{2}Ni}{\left({\mathrm{CO}}_{\mathrm{2}}\right)}_{Nt-i}}+{ϵ}_{\mathrm{2}Nt},}\hfill \end{array}

where EC is energy use measured as kg of oil equivalent per capita, FDI is foreign direct investment net inflows measured as percentage of GDP, GDP is real gross domestic product per capita, and CO₂ is metric tons of carbon dioxide per person. The real GDP per capita is expressed in US$ at constant 2000 prices. All variables are indexed as base year 2005. N is the number of countries (j=1, 2, …, N), t is the time period (t=1, 2, …, T), and p₁, p₂ are the lag length chosen by Schwarz Bayesian Criterion. GDP and CO₂ are treated as auxiliary variables, which will not be directly involved in the Granger causality analysis.

Since each equation in the system has different predetermined variables and the error terms might be cross-sectionally dependent, the sets of equations are the seemingly unrelated regression (SUR) system. To test Granger causality, alternative causal relations are likely to be found for country j. First, there is one-way Granger causality from FDI to EC if not all δ_1j’s are zero, but all β_2j’s are zero. Second, there is a unidirectional Granger causality from EC to FDI if all δ_1j’s are zero, but all β_2j’s are not zero. Third, there is a bidirectional Granger causality between FDI and EC if both δ_1j’s and β_2j’s are not zero. Lastly, there is no Granger causality between FDI and EC if all δ_1j’s and β_2j’s are zero. To analyze the causal nexus between FDI and CO₂ emission, Eqs. (1) and (2) are rearranged and GDP and EC are treated as auxiliary variables.

All variables, except the input of CO₂ emission, were obtained from the World Developments Indicator 2012. The data of CO₂ emission were attained from US Energy Information Administration. Since panel SUR estimator needs T>N, the sample including 76 countries is subdivided into five categories by following WDI income classification—lower income countries, lower middle income countries, upper middle income countries, high income OECD countries and high income non-OECD countries. Finally the base period is between 1980 and 2009.

Method and findings

The methodology adopted in this study consists of three steps. In the first step, whether the SUR estimators are more efficient than the OLS estimators was analyzed. If there is contemporaneous correlation in the system, the SUR estimators are more efficient than the OLS estimators [27]. The Monte Carlo experiment conducted by Pesaran [28] has emphasized the importance of testing for the cross-sectional dependence in a panel data study and illustrated the substantial bias and size distortions when cross-sectional dependence is ignored. Unobserved common factors can cause cross-sectional dependence in panel data analyses. The common factors may be a global trend component, a global cyclical component, common technological shocks or macroeconomic shocks that cause cross-sectional dependence [29]. The source of international influences between economies has been investigated in the literature and there are some hypotheses about it. The first one is the locomotive hypothesis which urges that fluctuations in larger countries (like the US) were presumed to act as locomotives, driving the fluctuations in smaller nations. The second one is the common shock hypothesis which emphasizes that exogenous oil shock and technology shock are the driving force behind the world business cycle. If the common shocks create a world business cycle, then the common shocks must occur reasonably frequently, must happen with certain regularity (periodicity), and must diffuse around the world fairly quickly. Moreover, the shocks must originate somewhere. It is not clear how all these conditions can be met by common exogenous shocks [30,31]. Following the locomotive hypothesis, this study assumes that fluctuations in the US cause fluctuations in smaller countries and adds the US to all subsamples.

To test cross-sectional dependency, Breusch and Pagan [32] and Peseran [33] have proposed Lagrange multiplier (LM) test. However the latter test is suitable when N is large and T is small. In the context of large T and small N, the following Lagrange multiplier test statistics proposed by Breusch and Pagan [32] can be used to test cross-sectional dependence:

{\mathrm{CD}}_{\mathrm{LM}}=T{\displaystyle \sum _{i=\mathrm{1}}^{N-\mathrm{1}}{\displaystyle \sum _{j=i+\mathrm{1}}^N{\widehat{\rho }}_{ij}^{\mathrm{2}}}},

where {\widehat{\rho }}_{ij} is the estimated correlation coefficients among the residuals obtained from individual OLS estimations. The statistics has chi-square asymptotic distribution with N(N-1)/2 degrees of freedom, under the null hypothesis of cross-sectional independency with a fixed N and time period T→∞. The results of the Breusch and Pagan test are depicted in Table 2.

Furthermore, the heterogeneity in estimated parameters for each individual of panel in order to impose a restriction for the causal relationship should be taken into account, since the causality from one variable to the other variable by imposing the joint restriction for the whole panel is the strong null hypothesis [34]. Country specific characteristics lead to a divergence from the assumption of the homogeneity for the parameters in a panel data setting [35]. Whereas, in many economic relationships such as energy consumption and FDI nexus, it is highly possible to find out a significant relationship in some countries or vice versa in other countries. To test the groupwise heteroscedasticity, the modified Wald test which has a null hypothesis of homoscedasticity of the residuals was employed.

The results in Table 2 indicate that the null hypothesis of cross-sectional independency is rejected, which provides strong evidence on the existence of the cross-sectional dependency across included countries in the data sets. Also, the modified Wald tests reject the null hypothesis of homoscedasticity. Having cross-sectional dependency and heterogeneity across countries, the applied causality method should capture these features.

To examine the direction of causality in a panel data, three approaches to date have been employed [36]. The first approach is based on estimating a panel vector error correction model by means of a generalized method of moments (GMM) estimator. However, this approach is not able to take into account both the cross-sectional dependence and the heterogeneity. Furthermore, the GMM estimators can produce inconsistent and misleading parameters unless the slope coefficients are, in fact, homogeneous [37]. Even though the second approach proposed by Dumitrescu and Hurlin [38] controls the heterogeneity and the cross-sectional dependence, the testing procedure requires pre-testing for panel unit root. Since the variables in the system need to be stationary, if the variables are not stationary at their levels, data loss must emerge. On the other hand, the third approach proposed by Kónya [26] is robust enough to account for both the cross-sectional dependence and the heterogeneity. This approach is based on the SUR estimation which makes it possible to take into account cross-sectional dependence across the members of panel. Since the direction of causality is tested based on the Wald tests with the country specific bootstrap critical values, this approach does not require the joint hypothesis for all the members of panel. Furthermore, since Kónya [26] has extended the framework of Phillips [39] by generating country-specific bootstrap critical values, the testing procedure does not require any pre-testing for panel unit root and cointegration. Country-specific critical values are used, since the variables in the model are supposed not to be stationary. This is an important feature since unit-root and cointegration tests in general suffer from low power, and different tests often lead to contradictory results [26]. Therefore the panel causality approach proposed by Kónya [26] seems to be a superior method.

In the second step, the sets of equations (Eqs. (1) and (2)) are estimated with the SUR method and the country-specific bootstrap critical values are produced. The bootstrap samples and country-specific critical values are generated as follows:

1) Eq. (1) is estimated under the null hypothesis of non-causality from FDI to EC by imposing δ_1ji=0 for all j and i and the residuals are obtained. From these residuals N\times T\left[e_{H_{ojt}}\right] matrix is developed.

2) These residuals are resampled by randomly selecting a full column from the matrix \left[e_{H_{ojt}}\right] at a time and bootstrap residuals denoted as e_{H_{ojt}}^{*} are selected that t= 1, 2, ..., T^* and T^* can be greater than T.

3) The bootstrap sample of EC is generated under the assumption of non-causality from FDI to EC as {\mathrm{EC}}_{jt}^{*}={\widehat{\alpha }}_{1j}+{\displaystyle \sum _{i=1}^{P_1}{\beta }_{1ji}{\mathrm{EC}}_{jt-i}^{*}+e_{H_{oj{T}^{*}}}^{*}}.

4) {\mathrm{EC}}_{jt}^{*} is substituted for EC and Eq. (1) is re-estimated without any parameter restrictions. Then the Wald test for each country to test the null of non-causality is imposed.

5) The empirical distributions of the Wald test statistics are developed by repeating steps 2–4 many times and the bootstrap critical values are generated by selecting the appropriate percentiles of these sampling distributions.

The results from the panel Granger causality tests are reported in Table 3. According to Table 3, there are unidirectional causal nexuses from energy consumption per capita to inward FDI for Bangladesh, Chile, Costa Rico, Dominican Rep., Jamaica, Australia, Austria, France, Germany, Togo, Zambia, Portugal and Spain. That is, inward FDI does not cause energy saving for these countries in aggregated level. For Mozambique, India, Iceland, Panama and United Arab Emirates (UAE) there are one sided causal relations from inward FDI to energy consumption per capita. For all of the other countries, no causal relation between inward FDI and energy consumption per capita is found. So, energy consumption per capita is neutral to inward FDI in the case of these countries.

Table 3 reports that the causal relations between inward FDI and CO₂ emission change country by country as well. For Bangladesh, Chile, Costa Rico, Dominican Rep., Jamaica, Cyprus, Israel, Malta, Austria, Germany, Saudi Arabia and Portugal, there is unidirectional causal relation from CO₂ to inward FDI. Therefore, it can be concluded that even tough CO₂ emission is a signal for inward FDI, induced FDI does not cause CO₂ emission. There is one sided causal nexus from inward FDI to CO₂ emission only for Oman. Two sided relation between FDI and CO₂ emerges only in the case of Zambia. For all of the other countries, no causal relation and neutrality hypothesis is found.

In regards to cross-check, it is examined whether there are conflicting causal findings between inward FDI-energy consumption and inward FDI-CO₂ emission. Also, for the aim of cross-check, sign of the achieved causal relations is analyzed using cross-correlation in the last step of the analysis. Three lags are included in the cross-correlation analysis, since BIC chose maximum lag length as 3 in the causality tests. Table 4 lists the cross-correlation coefficients for current inward FDI-lags of energy consumption and current inward FDI-lags of CO₂ emission.

Both findings about causal relations between FDI-energy consumption and FDI-CO₂ emission and the sign of the cross correlation coefficients in Table 4 are the same for Bangladesh, Chile, Costa Rico, Dominican Rep., Jamaica, Austria, Germany and Portugal. However in the case of Germany, coefficients of cross correlation are negative. That is, the increase of CO₂ emission leads to a decrease in net inward FDI flows possibly due to the tight environmental regulation. In addition, there are some differences between causal findings in the case of other 8 countries. While there is unidirectional causal nexus from energy consumption to inward FDI, no causal relation is found between inward FDI and CO₂ emission for Australia, France, Togo and Spain. These findings may not be conflicting, since both causal relations indicate that inward FDI does not cause both energy consumption and CO₂ emission. Furthermore, both causality findings and cross correlation findings in Table 4 may indicate that the availability of energy can affect the inflow of FDI. For the pattern of Cyprus, Israel, Malta and Saudi Arabia, while there is no causal nexus between inward FDI and energy consumption per capita, one sided causal nexus from CO₂ emission to inward FDI is found. According to Table 4, the increase of CO₂ emission leads to rise in inward FDI flows, but increased FDI does not increase both energy consumption and CO₂ emission. Again, these findings reveal no conflicting result. These findings do not support the pollution haven hypothesis.

However, in the case of Mozambique, India, Iceland, Panama and UAE, the achieved causal relations and the sign of cross correlation reported in Table 5 may be conflicting, since as there is one sided causal relation from inward FDI to energy consumption per capita, but no causal nexus between inward FDI and CO₂ emission is found. Furthermore, in the case of Mozambique and UAE, the sign of the cross correlation is negative. Whereas these findings support the energy saving hypothesis, decrease in energy use due to the increase in inward FDI does not lead to a decrease in CO₂ emission. If lower level of energy use is more polluting style, the decrease in energy use may not guarantee the decrease of pollution level. Therefore, in the pattern of Mozambique and UAE, the pollution haven hypothesis is supported. In the case of India, Iceland and Panama, causal relations run from inward FDI to energy consumption. In addition the signs of the cross correlation coefficients are positive. So in the pattern of these countries, FDI leads to an increase in energy use. However, increasing energy use does not cause an increase in pollution level. If increasing energy use is embodied in the cleaner technologies, it may not lead to an increase in pollution level. Therefore, it can be concluded that the pollution halo hypothesis is supported in the case of India, Iceland and Panama. Lastly, in the pattern of Oman, while inward FDI does not cause an increase in energy consumption, it leads to an increase in CO₂ emission level due to the increase in dirty inward FDI flows. So in the case of Oman, the pollution haven hypothesis is supported.

The pattern of Zambia is individually evaluated, since bidirectional causal relation between inward FDI and CO₂ emission was found. Table 6 indicates the cross correlation coefficients for Zambia.

In the case of Zambia, increasing inward FDI causes a decrease in CO₂ level and decreasing CO₂ emission leads to a decrease in inward FDI flows. Even though decreasing pollution level is a signal for inward FDI, it causes a decrease in pollution level, and the pollution halo hypothesis is supported in this pattern. However, inflow FDI causes a decrease in pollution level by not affecting energy consumption. In short, all of these findings indicate that an increase (decrease) in energy use does not necessarily mean an increase (decrease) in pollution level, and vice versa.

Conclusions and policy implications

Using bootstrap-corrected panel causality test and cross-correlation analysis for 76 countries, the causal relations between inward FDI-energy use per capita and inward FDI-CO₂ emission per capita were analyzed in this study. Empirical tests produce changing results country by country. So no supportive evidence was found for changing causal relations to country group which was classified to income level. Achieved findings indicate that while the pollution haven hypothesis is supported for Mozambique, United Arab Emirates and Oman, in the case of India, Iceland, Panama and Zambia the pollution halo hypothesis is supported. For all of the other countries, energy use and CO₂ emission are neutral to inward FDI flows in aggregated level.

Furthermore, this study urges that increased (decreased) energy use due to the inward FDI flows does not necessarily mean an increase (decrease) in pollution level, and vice versa. Energy use and pollution level may change in opposite directions. If decreasing energy use is a more polluting style, it may not guarantee a decrease in pollution level and if increasing energy use is embodied in cleaner technologies, it may not lead to an increase in pollution level. So, unless the effects of FDI on energy consumption and pollution level are evaluated together, there is a possibility that one sided evaluation may lead to inaccurate results.

For the policy purposes, there should be two criteria to regulate the FDI encouragement policy for the aim of increasing environmental performance. The first one is that inward FDI should lead to energy saving by carrying the technologies which is increasing energy efficiency and the second is that inward FDI should lead to a decrease in pollution level by carrying greener technologies to the host country. If the two criterions are fulfilled by inward FDI, the host country may benefit more from it.

References

[1] Chang N. The empirical relationship between openness and environmental pollution in China. Journal of Environmental Planning and Management, 2012, 55(6): 783–796

[2] Zarsky L. Havens, halos and spaghetti: untangling the evidence about foreign direct investment and the environment. In: Meyer K E ed. Multinational Enterprises and Host Economies. Edward Elgar Publishing, 2008, 283–306

[3] Blackman A, Wu X. Foreign direct investment in China’s power sector: trends, benefits and barriers. Energy Policy, 1999, 27(12): 695–711

[4] Mielnik O, Goldemberg J. Foreign direct investment and decoupling between energy and gross domestic product in developing countries. Energy Policy, 2002, 30(2): 87–89

[5] Eskeland G S, Harrison A E. Moving to greener pastures? Multinationals and the pollution haven hypothesis. Journal of Development Economics, 2003, 70(1): 1–23

[6] Fisher-Vanden K, Jefferson G H, Ma J, Xu J. Technology development and energy productivity in China. Energy Economics, 2004, 28(5, 6): 690–705

[7] Tang C F. Electricity consumption, income, foreign direct investment, and population in Malaysia: new evidence from multivariate framework analysis. Journal of Economic Studies (Glasgow, Scotland), 2009, 36(4): 371–382

[8] Hübler M, Keller A. Energy savings via FDI? Empirical evidence from developing countries. Environment and Development Economics, 2010, 15(1): 59–80

[9] Sadorsky P. The impact of financial development on energy consumption in emerging economies. Energy Policy, 2010, 38(5): 2528–2535

[10] Pao H T, Tsai C M. Multivariate Granger causality between CO₂ emissions, energy consumption, FDI (foreign direct investment) and GDP (gross domestic product): evidence from a panel of BRIC (Brazil, Russian Federation, India, and China) countries. Energy, 2011, 36(1): 685–693

[11] Ting Y, Long R Y, Zhuang Y Y. Analysis of the FDI effect on energy consumption intensity in Jiangsu province. Energy Procedia, 2011, 5: 100–104

[12] Song F, Zheng X. What drives the change in China’s energy intensity: combining decomposition analysis and econometric analysis at the provincial level. Energy Policy, 2012, 51: 445–453

[13] Herrerias M J, Cuadros A, Orts V. Energy intensity and investment ownership across Chinese provinces. Energy Economics, 2013, 36: 286–298

[14] Al-mulali U, Tang C F. Investigating the validity of pollution haven hypothesis in the gulf cooperation council (GCC) countries. Energy Policy, 2013, 60: 813–819

[15] Cole M A, Elliott R J R. FDI and the capital intensity of “dirty” sectors: a missing piece of the pollution haven puzzle. Review of Development Economics, 2005, 9(4): 530–548

[16] Hoffmann R, Lee C G, Ramasamy B, Yeung M. FDI and pollution: a Granger causality test using panel data. Journal of International Development, 2005, 17(3): 311–317

[17] He J. Pollution haven hypothesis and environmental impacts of foreign direct investment: the case of industrial emission of sulfur dioxide (SO₂) in Chinese provinces. Ecological Economics, 2006, 60(1): 228–245

[18] Merican Y, Yusop Z, Noor Z M, Hook L S. Foreign direct investment and the pollution in five ASEAN nations. International Journal of Economics and Management, 2007, 1(2): 245–261

[19] Waldkirch A, Gopinath M. Pollution control and foreign direct investment in Mexico: an industry-level analysis. Environmental and Resource Economics, 2008, 41(3): 289–313

[20] Jorgenson A K. Foreign direct investment and the environment, the mitigating influence of institutional and civil society factors, and relationships between industrial pollution and human health: a panel study of less-developed countries. Organization & Environment, 2009, 22(2): 135–157

[21] Lee C G. Foreign direct investment, pollution and economic growth: evidence from Malaysia. Applied Economics, 2009, 41(13): 1709–1716

[22] Tamazian A, Chousa J P, Vadlamannati K C. Does higher economic and financial development lead to environmental degradation: evidence from BRIC countries. Energy Policy, 2009, 37(1): 246–253

[23] Kim M H, Adilov N. The lesser of two evils: an empirical investigation of foreign direct investment-pollution tradeoff. Applied Economics, 2012, 44(20): 2597–2606

[24] Blanco L, Gonzalez F, Ruiz I. The impact of FDI on CO₂ emissions in Latin America. Oxford Development Studies, 2013, 41(1): 104–121

[25] Chandran V G R, Tang C F. The impacts of transport energy consumption, foreign direct investment and income on CO₂ emissions in ASEAN-5 economies. Renewable & Sustainable Energy Reviews, 2013, 24: 445–453

[26] Kónya L. Exports and growth: Granger causality analysis on OECD Countries with a panel data approach. Economic Modelling, 2006, 23(6): 978–992

[27] Zellner A. An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association, 1962, 57(298): 348–368

[28] Pesaran M H. Estimation and inference in large heterogeneous panel with a multifactor error structure. Econometrica, 2006, 74(4): 967–1012

[29] Huang Y. Private investment and financial development in a globalized world. Empirical Economics, 2011, 41(1): 43–56

[30] Selover D D. International interdependence and business cycle transmission in ASEAN. Journal of the Japanese and International Economies, 1999, 13(3): 230–253

[31] Sayek S, Selover D D. International interdependence and business cycle transmission between Turkey and the European Union. Southern Economic Journal, 2002, 69(2): 206–238

[32] Breusch T S, Pagan A R. The Lagrange multiplier test and its application to model specifications in econometrics. Review of Economic Studies, 1980, 47(1): 239–253

[33] Pesaran M H. General diagnostic tests for cross section dependence in panels. CESifo Working Paper, 2004, Paper No. 1229

[34] Granger C W J. Some aspects of causal relationships. Journal of Econometrics, 2003, 112(1): 69–71

[35] Breitung J. A parametric approach to the estimation of cointegration vectors in panel data. Econometric Reviews, 2005, 24(2): 151–173

[36] Kar M, Nazlioglu S, Agir H. Financial development and economic growth nexus in the MENA countries: bootstrap panel granger causality analysis. Economic Modelling, 2011, 28(1-2): 685– 693

[37] Pesaran M H, Shin Y, Smith R P. Pooled mean group estimation of dynamic heterogeneous panels. Journal of the American Statistical Association, 1999, 94(446): 621–634

[38] Dumitrescu E I, Hurlin C. Testing for Granger non-causality in heterogeneous panels. Economic Modelling, 2012, 29(4): 1450–1460

[39] Phillips P C B. Fully modified least squares and vector autoregression. Economic Modelling, 2012, 29(4): 1450–1460

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg