Investigating Dynamic Effect of Energy Consumption, Foreign Direct Investments and Economic Growth on CO2 Emissions between Oman and United Arab Emirates: Evidence from Co Integration and Causality Tests

The main objective of this paper is to examine the dynamic impact of energy consumption, foreign direct investments, and economic growth on carbon emissions by employing dynamic estimation techniques of co-integration for Oman and the United Arab Emirates. This study used quarterly panel data of Oman and UAE over the period 1991-2018. Generalized method of moments (GMM), Granger causality test is applied to perform causal relationships among the constructs. The empirical econometric results revealed that there is a strong co-integration of the variables in the case of Oman but not for the United Arab Emirates. In other words, there is a strong positive effect of energy consumption and foreign direct investment on CO2 emission in Oman and the United Arab Emirates; whereas economic growth possesses a strong negative effect on CO2 emissions. The estimation did not support the EKC hypothesis and was found to be not applicable in Oman and UAE. Further discussions on the implications of the results are provided.


INTRODUCTION
Climate change is one of the most defining environmental threats that impact the sustainable economic growth of any nation in the world. Change in weather patterns, rise in sea levels, intense droughts on food productions, and freshwater suppliers intimidate the livelihood of humans and other livings organisms' life at stake. The increasing concentration of GHG, especially Carbon dioxide (CO 2 ) due to deforestation, excess usage of fossil fuels, and other human actions are considered to be the major reason for global warming (Abdul-Wahab et al., 2015). Countries are in the race of achieving goals of rapid development and economic growth, leading to a strong environmental deterioration of environmental and resource consumptions (Govindaraju and Tang, 2013). Foreign Direct Investments (FDI) which fosters economic development results in industrialization and manufacturing sector development which demands more energy consumption. This increased energy consumption eventually alleviates the CO 2 emissions.
During the past few years, Oman and United Arab Emirates (UAE) have been facing multiple environmental challenges such as high temperature and humidity; reduced seawater quality, coastal erosion, increased pressure on food supply and production and lost productivity and higher health cost due to heat stress, allergies and infectious diseases (Ministry of climate change and environment, 2018). Based on the analysis by environmentalists, there will be an increase of 1°C temperature in UAE by 2020 and 1.5°C by 2040 (Abu Dhabi Global Environment Data Initiative, 2015). Such This Journal is licensed under a Creative Commons Attribution 4.0 International License climate change could have severe implications on the growth and development objectives of Oman and UAE. Environmentalists have argued that carbon emissions are the main source of this climate change and global warming Andreoni and Galmarini, 2016) due to increased energy consumption and growth (Antonakakis et al., 2017;Olivier et al., 2017). There has been an increasing trend for energy consumption, CO 2 emissions, FDI, and the real GDP of Oman and UAE over the period of 1991-2018 ( Figure 1). Oman and UAE had shown remarkable growth on CO 2 emissions in the period of 2010-2018, at an average rate of 17.86% and 21.60% respectively in comparison to economic growth at an average rate of 3.45% and 3.64%, respectively (Table 1). As shown in (Figure 1) Oman has shown a significant increase in energy consumption as compared to UAE (from 7.6% in 1991 to 8.8% in 2018 for Oman and from 9% in 1991 to 9.4% in 2018 for UAE).
Oman and UAE have been the most populous GCC countries for CO 2 emissions and in the future are expected to increase in a rapid pace (Abdul-Wahab et al., 2015;Asif et al., 2015;Radhi, 2009;Yousif et al., 2017). Oman and UAE with large oil reserves for the same periods used an average of 6617 and 10929 kg of oil equivalents and are expected to increase in the future. Therefore, the biggest challenge for these two countries is to keep balance the energy consumption and CO 2 emission together with maintaining economic growth. Nevertheless, both countries have significant economic growth decline in the year 2011 and 2017, there might be the existence of decoupling of CO 2 emissions from economic growth (Riti et al., 2017;Wang et al., 2017).
Notwithstanding the considerable efforts made by Oman and UAE Governments to address the effect of climate change to mitigate further environmental deterioration, it is imperative at this point for these countries to control their energy consumption especially fossil fuel, and increase the usage of renewable energy resources. In other words, the high usage of fossil fuels will not only diminish its availability in the future but will contribute to more carbon emissions. Nevertheless, fossil fuel in the form of crude oil plays a critical role in the economic growth of these countries, the reduction in energy consumption is a viable option. However, this move might limit the rapid economic development of Oman and UAE.
Indeed of the growth of economy in both the countries, government's effort to develop renewable resources may reduce their reliance on fossil fuel for energy production. Therefore, a wide range of questions arises for these oil producing countries.
Will the economic growth affect if energy production using fossil fuels is controlled or if it is replaced with environmental friendly resources as an alternative? How CO 2 emissions is influenced by economic growth and energy consumption? Will FDI inflows be affected due to high CO 2 emissions? In order to answer all these questions, this study attempts to insight with careful investigation on such factors that allows appropriate formulation of strategies for the policymakers.
There have been numerous attempts performed by previous studies like (Bastola and Sapkota, 2015;Bouznit and Pablo-Romero, 2016;Kasman and Duman, 2015;Wang et al., 2016), there exist two main fundamental gaps in the literature. First, investigating the longrun relationship between energy consumption and CO 2 emissions using long-range data (e.g., 1991-2018). According to the World Economic Development Indicators, in 2018, carbon emissions per capita for Oman was 17.64 metric tons whereas carbon emissions per capita for UAE was 22.44 metric tons. It is very crucial for the study to include long-range data series for robust estimation of results using co-integration analysis. Also, investigating two highly resilient economies like Oman and UAE will provide meaningful insight on whether there exists any differences in the relationships despite heavy consumption and reliance on fossil fuels. In light of the above mentioned economic and environmental differences between Oman and UAE, this study will examine the long-run influence of energy consumption, CO 2 emissions, FDI inflows on economic growth for the period of 1991-2018.

LITERATURE REVIEW
There have been significant studies performed examining the relationship and association between CO 2 emissions, energy consumption, FDI inflows and economic growth using series and panel data to validate the environmental Kuznets curve (EKC) hypothesis in the past few decades. Theoretical foundations of the EKC hypothesis examining the nonlinear function of economic activity leads to deterioration of environmental quality. EKC postulates that at a certain level as income increases environmental pressure also increases. An extensive review of previous researches to test EKC and their conclusions are provided in the below lines. For Saudi Arabia, (Mahmood and Alkhateeb, 2017) explored the relationship between trade and income on CO 2 emissions through the EKC hypothesis and found that trade has a negative impact on CO 2 emissions confirming the EKC hypothesis. For GCC countries, (Al-Mulali et al., 2015) used ecological footprints for environmental degradation and found that energy consumption, urbanization, and trade openness damage the environment confirming positive ecological footprints. They also found that environmental degradation can be reduced through financial development in the GCC countries confirming that loan is favored to the companies investing in the projects that are mostly environmental friendly. Diverting from GCC countries to some of the literature of countries that damage environment severely, like for China,  investigating the EKC curve applying dynamic panel model for 28 provinces in China using GMM and ARDL model found that EKC hypothesis is supported across China for all three major pollutants (CO 2 , wastewater and waste solid emissions). In contrary to this result, (Wang et al., 2011) earlier mentioned that there is a U shaped relation between CO 2 emissions and economic growth confirming EKC is not valid in China. Also, (Usman et al., 2019) revisited the EKC hypothesis in India investigating the role of energy consumption for the period of 1971 to 2014 using ARDL and Granger Causality test. They found that the EKC hypothesis is valid for India as energy consumption increases environmental degradation in the long and short run. The reviewed literature provides mixed result on confirming the EKC hypothesis and are contrary to each other. This conflicting results in different countries can be due to different variables selected, time, method, and country's financial situation during the period of investigation. Table 2 summarizes previous relevant studies with their key features, methodology and main findings. From the Table 2 it can be seen that several studies have investigated the nexus between pollution, energy consumption, FDI and economic growth based on the aggregate data at country level. Thus, it is important to reexamine this relationships comparing Oman and UAE.
Many recent and past studies (Bilgili et al., 2016;Dogan and Turkekul, 2016;Dong et al., 2018;Pata, 2018) investigating energygrowth relationships have confirmed this postulate; however, many studies (Charfeddine, 2017;Dogan and Turkekul, 2016;Khan et al., 2019) have doubted this relationship too. The EKC hypothesis implies that in the initial stage of growth, environmental pollution increases, and then at the later stages it starts declining (Alam et al., 2016). This indicates that there is an inverted U-shape relation between growth and environmental deterioration supporting the EKC hypothesis. Although there have been several research investigations performed on the energy consumption-CO 2 emissions-economic growth-FDI nexus (Ahmad et al., 2019;Akadiri et al., 2019;Khan et al., 2019;Salahuddin et al., 2018), to the best of the author's knowledge no study have been undertaken in the context of Oman and UAE, which is very crucial for such investigation. This study will attempt to fill this gap.

Economic Growth-pollution Nexus
The nexus between economic growth and pollution through CO 2 emissions have been investigated extensively. In the case of Kuwait, (Salahuddin et al., 2018) using the auto-regression distributed lag (ARDL) approach found that CO 2 emissions are stimulated by the economic growth for both the short and long run. Similarly, (De Freitas and Kaneko, 2011) for Brazil investigated the decoupling of CO 2 emissions and economic growth and found that between 2004 and 2009, the intensity of carbon and energy mix are the key factors for emissions reduction. In the case of India, (Alam et al., 2011) investigated the dynamic modeling relationship between CO 2 emissions and economic growth and found that in the long run, CO 2 emissions do not cause the growth of the economy. Furthermore, (Stamatiou and Dritsakis, 2017) found that there is a unidirectional relation between economic growth and CO 2 emissions in the short run for Italy. For the panel of 28 provinces in China (Wang et al., 2011) found that in the short-run and long-run CO 2 emissions will not decrease, and reducing CO 2 emissions may handicap the economic growth of China at some level. Also, (Saidi and Hammami, 2015) investigating a panel of 58 countries found that economic growth has a positive impact on CO 2 emissions. Previous studies utilized different approaches and econometric techniques (Generalized method of moments (GMM), auto-regressive distributed lag (ARDL), and Vector error correction model (VECM) Granger causality for investigating the panel or time-series data.

Energy Consumption-pollution Nexus
The nexus between energy consumption and pollution has been reviewed in most of the literature to test EKC. Gu

FDI Inflows-pollution Nexus
Capital inflows have been an important topic of discussion in the current years due to the high volatility of capital flows weakens the financial systems (Muhammad and Khan, 2019). The nexus between FDI leading to pollution has been investigated for the emerging and developing nations. FDI outflows are usually from developed countries. An increase in carbon emissions leads to environmental change and ongoing anxiety for both emerging and developed countries. Using linear and nonlinear autoregressive distributed lag modeling (Haug and Ucal, 2019) investigated how FDI inflows affect CO 2 emissions and found that in the long run, FDI inflows have no significant effect on carbon emissions.

Economic Modelling
To test the EKC hypothesis for Oman and UAE, quarterly panel data from 1991 to 2018 has been used that was retrieved from World Development Indicators (WDI) and International Energy Statistics (IES) database, respectively. Real GDP per capita was measured using constant 2000 US$, per capita energy consumption, and CO 2 emission per capita was measured by dividing them with the mid-year population as suggested by (Salahuddin et al., 2018). These constructs were then transferred to natural logs to avoid the issue of heteroscedasticity between the constructs. The econometric model estimated for this study takes the following form: where, CO 2 it is the carbon dioxide emissions for ith countries at time t. EC it is energy consumption for ith countries at time t, FDI it is the foreign direct investment for ith countries at time t and GDP it is the gross domestic product for ith countries at time t. The coefficients β0 represents the constant, β1, β2, β3 represents the long run estimation between CO 2 emissions, energy consumption (Kwh), foreign direct investment (FDI) and real GDP per capita, respectively.
To investigate the relationships, data were sourced for the following constructs: • CO 2 emissions per capita for country i at time t (C it ) • Energy consumption per capita for country i at time t (EC it ) • Real GDP per capita for country i at time t (GDP it ) • Foreign direct investment inflows for country i at time t as share of GDP (FDI it ).
In panel data, we usually face the issue of serial correlation and heteroskedasticity that may distort the model. Attari et al. (2016) defined serial correlation as the disturbance to any construct due to correlation. Panigrahi (2017) heteroskedasticity issue arises when the changes in the error terms differentiate from the observations. Dynamic GMM model helps to solve the issue of heteroskedasticity and serial correlation . Jacob and Osang (2020) explained that the GMM model is more suitable for the panel invariant constructs.

Dynamic GMM Model
Generalized method of moments (GMM) suitable for the number of moment conditions was first developed by (Hansen, 1982) to estimate the parameters with endogenous regress in the panel data model with unobserved individual accurate heterogeneity. GMM has been consistently proposed by (Arellano and Bond, 1991;Bennett et al., 2019;Panigrahi, 2017) converting the model into the first difference to wipe up individual effect and employ lag wherever lag levels of constructs are instruments. Such a model shows no consistent for fixed and random effect estimators for a finite period and a huge number of observation data (Muhammad, 2019).

DATA ANALYSIS
The analysis began with testing descriptive statistics and the properties of the unit root test for the time series data in the study. We used traditional unit root tests like (Said and Dickey, 1984) ADF unit root test by allowing the lag terms that provide serial correlations and (Perron and Phillips, 1987) PP test for autocorrelation detection. We also presented GMM estimation using a dynamic panel taking into account lagged levels of CO 2 emissions following (Arellano and Bond, 1991).
Descriptive statistics results are depicted in Table 3 having all positive mean, standard deviation, and median with a significant peak of distribution confirming strong CO 2 emissions together with energy consumption, FDI inflows, and real GDP per capita for Oman and UAE. Pollution in the form of CO 2 emissions is significantly affected by positive mean and standard deviation of energy consumption, FDI inflows, and real GDP per capita.
CO 2 emissions were measured in terms of metric tons per capita. Energy consumption was measured at the equivalent of electricity usage per capita. The mean growth of CO 2 emissions, energy consumption, and FDI inflows was highest for UAE as compared to Oman. Moreover, the average real GDP per capita for Oman was found to be higher as compared to the UAE.

Unit Root Test
The first step of this empirical investigation was to test the stationarity of the constructs using different unit root tests. In the time series data checking stationarity is an econometric process to avoid spurious results. Table 4 shows the Phillips and Perron (1988) and Dickey and Fuller (1979) panel unit root tests at the level form and the first difference of each series. The difference between the Phillips and Perron (1988) test and the Dickey and Fuller (1979) unit root tests is that the latter uses lag length to solve the issues of autocorrelation and to improve the resulting robustness.

Table 2: Summary of previous studies on growth-pollution-consumption and FDI relationships
Authors-year Period Sample Methodology Key findings Economic growth-pollution (Kasman and Duman, 2015) 1992-2010 EU nations Panel Cointegration Causal relation between carbon emissions and GDP  1990-2012 China Co-integration test Marginal shock in economic growth due to CO 2 emissions  1992Q1-2011Q1 Croatia ARDL & VECM approach Unidirectional causality from economic growth to CO 2 emissions in long run (Ozturk and Acaravci, 2016) 1980-2006 Cyprus and Malta ARDL CO 2 emissions have no adverse effect on growth of Malta (Begum et al., 2015) 1970-2010 Malaysia DOLS and Sasabuchi-Lind-Mehlum (SLM) test In long run, economic growth may have an adverse effect on CO 2 emissions Energy consumption-pollution (Muhammad, 2019) 2001-2017 MENA GMM Energy consumption increase together with CO 2 emissions, but economic growth decreases in MENA region (Wang et al., 2018) 1990-2011 China Correlation analysis Energy consumption and CO 2 emissions are heavily impacted by urbanization (Nain et al., 2017) 1971-2011 India ARDL Short run causality from electricity consumption to CO 2 emissions (Shan et al., 2017) 2014 Tibet Quantitative CO 2 emissions are associated from cement production and road transportation with high emissions intensity (Cai et al., 2018) 1965-2015 G7 Countries ARDL Clean energy consumption is caused by CO 2 emissions FDI-CO 2 emissions-economic growth nexus (Shahbaz et al., 2019) 1990-2015 MENA GMM Connection between energy consumption and CO 2 emissions is bidirectional (Salahuddin et al., 2018) 1980-2013 Sub-Saharan African Pooled OLS Energy consumption lead to higher economic growth at the cost of environmental degradation (Sarkodie and Strezov, 2019) 1982 The unit root test results as shown in Table 5 are integrated if the null hypothesis is rejected. The results of Phillips and Perron's (1988) test and the Dickey and Fuller (1979) indicate that the variables were stationary after the first differences whereas the level unit root shows the existence of stationarity.

Model Specification
This paper used the GMM model introduced by (Hansen, 1982) for estimation and testing together with plots of residual leverage.
The robustness of the model is provided by the Hansen J-statistics. The Hansen test of over-identifying restrictions is a joint test of the hypothesis that the instruments are distributed independently of the error process. Table 6 presented the results of GMM estimation, OLS, FMOLS, and DOLS using aggregated panel quarterly data from 1991 to 2018 for Oman and UAE.
The presence of the EKC hypothesis was not confirmed from the results of the panel as the influence of electricity consumption and FDI inflows were severe on the CO 2 emissions for Oman and UAE. However, we found that GDP growth deteriorates CO 2 emissions in the long run. The result was supported by (Shahbaz et al., 2013) who found that sustainable economic growth by controlling the environment with efficient usage of resources degrade carbon emissions. Similarly, (Al-Mulali, 2014) found that energy consumption was an important factor than real GDP to control environmental degradation.
Appendix A provides the residual leverage plots providing the relationship between CO 2 emissions versus the independent constructs and shows that carbon emissions have high leverage towards economic growth as compared to that of energy consumption and FDI inflows. From Table 6 presenting the  −5.998*** −3.579*** −7.664*** −6.790*** ∆∆ lnFDI 2t −35.798*** −5.340*** −20.511*** −7.400*** lnGDP 2t −3.281*** −3.470** −5.007** −4.897** ∆ lnGDP 2t −9.023*** −9.055*** −15.337*** −6.188*** ∆∆ lnGDP 2t −56.791*** −11.174*** −23.165*** −9.025*** VECM short and long-run Granger causality F statistics, it is evident that GDP and energy consumption are significant to FDI inflows indicating that in short and long-run causality running from GDP and energy consumption to FDI for UAE. For Oman, it was evident that energy consumption granger bi-directionally causes CO 2 emissions in both the short and long run. It was also found that granger causality f statistics for GDP and energy consumption towards FDI inflows were significant indicating that in the short and long run FDI inflows are heavily dependent on the GDP and energy consumption of Oman. Figure 2 provides the plot of CUSUM and CUSUMSQ test statistics that falls inside the critical bounds of 5% significance implying that the estimated parameters are stable over the periods.

CONCLUSIONS
This study employed a multi-panel estimation model to predict the long-run relationship between electricity consumption, FDI inflows, economic growth, and CO 2 emissions for Oman and UAE. Figure 3 provides the causal direction among the variables that confirmed electricity consumption and FDI inflows towards CO 2 emissions for Oman and UAE. Based on the GMM and other regression techniques the estimated value shows that there is a long term equilibrium relationship between energy consumption and CO 2 emissions and was an important factor than the GDP and FDI inflows. Also, the results indicated that real GDP negatively impacts carbon emissions. Therefore, economic growth has a negative effect on carbon emissions in Oman and UAE. This result was also supported by previous studies in (Pao et al., 2012) China, South Africa, India, ASEAN-5 countries The EKC hypothesis is not applicable to Oman and UAE.
Deployment of renewable energy technologies on a large scale can be an initiative for climate action. Although climate change is a threat to Oman and UAE, it can be regarded as an opportunity for diversifying the economy in the form of innovative green products. Stronger policies and regulations are needed to stimulate investment flows into sustainable green projects.
Creating an integrated framework that could play an important role in international climate action through a technology transfer mechanism is required to be implemented. Green sectors such as clean energy, green manufacturing, and environmental goods and services need to be focused in the course of implementing a climate action plan.

ACKNOWLEDGMENT
We would like to thank the anonymous reviewers for the comments and suggestions that significantly improved the previous version of this study. In addition, the remaining flaws in this study are solely our responsibilities and the usual disclaimer applies.