THE ROLE OF RENEWABLE ENERGY CONSUMPTION AND FDI IN TESTING THE EXISTING OF ENVIRONMENTAL KUZNETS CURVE IN VIETNAM

This study tests the environmental Kuznets curve (EKC) while examining the role of renewable energy sources and FDI to potentially effect the amount of CO2 emissions in Vietnam. Using Autoregressive Distributed Lag (ARDL) approach, the relationship between CO2 emissions and GDP, FDI and sources of energy consumption has been investigated during 1980-2018 in Vietnam. The results of study reveal that there is non-existence EKC for Vietnam in long run and a sign of inverted Ushape in short run in this period. The analysis also shows that the coefficient of energy consumption from hydro-power renewable sources which correlates to CO2 emissions is negative and significant while FDI leads to increasing carbon dioxide emissions in the long run. That implies Vietnam could have been benefited from a drop in CO2 emissions at some point in the early stage of the period, however, the country has been facing environmental pollution increase as GDP and FDI growth.


INTRODUCTION
Since embarking in 1986 on economic reform, Vietnam has made a remarkable shift to a market economy that has resulted in impressive wealth, trade and investment gains. During over the past 30 years, the economy has achieved uninterrupted growth, recording an average GDP growth rate around 6.6% a year. Energy fuel resource of various types such as coal, natural gas, petroleum have played a key role in driving the Vietnamese economy. But current consumption and production patterns are placing enormous pressure on environment. According to Audinet et al. (2016) the carbon intensity of Vietnam's economy at almost triple the world average and if Vietnam have emitted amount of CO 2 at the current pace, it will be projected to rise by 495 million tons of carbon dioxide (MtCO 2 ) in 2030. That would cause environmental degradation, undermine human productivity and limit the economy growth.
The relationship between environmental degradation and economic growth has been got attention by many economists for years. In common, economic growth will cause environmental degradation and pollution increase, however, environmental Kuznets curve (EKC) showed reverse trend at high-income levels, economic growth leads to environmental improvement. Panayotou (1993) argued EKC hypothesis at higher levels of development, by increasing environmental awareness, enforcement of environmental regulations, better technology and more environmental expenditure, which will result in declining of pollution and environmental degradation. A vary of empirical studies have been carried out to test the existence of EKC in different countries and in distinct periods. Many authors found that within a panel of countries, some of them, the EKC hypothesis is supported while it is not the case for some other (Karsch, 2019;Shahbaz et al., 2019;Stern, 2004;Sun, 1999). The author Shahbaz et al. (2019) stated the validation of the EKC hypothesis can vary in functions of the country-specific factors, study periods and This Journal is licensed under a Creative Commons Attribution 4.0 International License underlying methods. He also noted that the EKC hypothesis holds mainly in high-income and in upper-middle-income countries.
Though scholar authors have studied EKC prolonged time, but there are only a few researches for Vietnam. In 2015, Al-Mulali et al., and Tang and Tan tested Vietnamese EKC in the period of 1981and 1976-2009 respectively, however the results of the two studies contradicted each other. The outcomes of Al-Mulali et al. (2015) revealed that there is no sign of the existence of EKC in the period of 1981-2011 and Vietnam had followed the pollution haven hypothesis (PHH) because capital increased pollution. In the contrast, Tang and Tan (2015) found an inverted U-shaped relationship between CO 2 emissions and economic growth in Vietnam during period . Recently, the comprehensive study about EKC in Vietnam of Shahbaz et al., (2019) has confirmed this existence in long term but not for short run in the period of 1974-2016. This study also suggested N-shaped described better long run between income -CO 2 emission relationship and made a warning that Vietnam can expect a temporary reduction in CO 2 emissions at a given stage of economic growth. However, this will be followed by a further increase of CO 2 emissions after reaching another income turning point.
Adapting from previous studies, our study extends the EKC literature by retesting EKC in Vietnam from 1980 up to 2018 and examining whether renewable energy and FDI will be effective in the role of reducing CO 2 emission. According to our knowledge, there has been no research paper that examines the potential impact of these elements for Vietnam within the EKC framework. Using EKC hypothesis, this study has conducted an Autoregressive Distributed Lag (hereafter ARDL) bounds approach to test cointegration. The impact of renewable energy consumption and FDI flow on CO 2 emissions in both short and long run in Vietnam will ensure important data for the policies towards sustainable development.

LITERATURE REVIEW
The Environmental Kuznets Curve (EKC) was inherited from the concept of Kuznets curve, which is inverted U-shaped between inequality and development with the hypothesis as an economy develops, market forces begin to increase and economic inequality decreases. The Environmental Kuznets Curve adheres to the same idea with the relationship between environmental quality and income per capita. Grossman and Krueger (1991) applied the hypothesis to argue that economic growth is necessary in order for environmental quality to be maintained or improved in the path of sustainable development. Based on technology, tastes and environmental investment, greater economic activities will lead to increase the environmental quality (World Bank, 1993). The EKC has also faced criticisms due to its assumption that environmental damage does not reduce economic activity sufficiently to stop the growth process and reduce the level of income in the future (Arrow et al., 1995;Stern, 2004). However, the existence (or non-existence) of an EKC has significant policy implications (Bruyn and Heintz, 1999). If true, it provides evidences for the success of sustainable development strategies and economic growth is the key to solve environmental problems. If not, it suggests country to adopt stringent environmental policies to prevent the consequences of economic growth.
In the theme of EKC framework, researchers have tested a various of environmental degradation indicators which can be group into local and global pollutants. Stern (2004) found that EKC studies using local pollutants were more likely to display an inverted U-shape relation with income because local impacts are likely to give the rise to environmental policies to correct the externalities. For global pollutants like CO 2 , the empirical evidences on the validity of the EKC in recent studies fail to yield consistent results. Dinda and Coondoo (2006) found mixed results when they use a panel data cointegration methodology in a bivariate setting. They suggested a time series approach because of having a suspicion about a dynamic link between CO 2 emissions and income. Halicioglu (2009) stated that the cointegration approach of Pesaran et al. (2001) for bounding test has certain econometric advantages in comparison to other single cointegration procedure. Following the ARDL approach, many studies has confirmed the existence of the EKC in developing countries while a number of the opposed results have also been reported (Mrabet and Alsamara, 2017;Saboori and Sulaiman, 2013;Shahbaz et al., 2019).
The common factors often to be tested in EKC function are energy consumption and foreign direct investment (FDI). In the case of Vietnam, FDI has played a crucial role in economic development by attracting capital, enhancing firm-level efficiency, transferring cutting-edge technologies and encouraging product and process innovation (Hoang and Duong, 2018). However, FDI -environmental emissions linkage has become a controversial topic, whether FDI inflow could have brought to the costly negative externalities or the positive effects for the society. Tang and Tan (2015) found out that FDI is the main determinants of increasing CO 2 emissions in Vietnam in the period of 1976-2009. The debate of impact of FDI on environment in developing countries has generated opposing hypotheses that supports each line of arguments. The pollution haven hypothesis was raised to explain the increases in FDI would deteriorate the host's environment due to polluting industries in developed countries would tend to move to developing countries for lessening environmental regulations. This hypothesis supports the argument that emissions reduction in many developed countries would be partly due to the shifting of polluting activities to developing countries (Kearsley and Riddel, 2010). Although the literature is dominated with this adverse view of FDI on the environment, it is also possible that FDI can contribute to a better environment (Zarsky, 1999). The study of Zhu et al., (2016) argue that foreign companies are more sensitive to the environment as they use better management practices and advanced technologies that are conducive to the environment compared to their domestic partners which was explained by pollution halo hypothesis. Realizing the potential environmental costs associated with FDI, most countries are now selective in the type of FDI that comes into their country.
For energy consumption, it is the key factor contributing to CO 2 emissions in Vietnam because the country has mostly relied on fossil fuels as oil, gas and coal for energy. That releases carbon dioxide and other greenhouse gases, making them the primary cause of global warming and climate change.  (Dogan and Seker, 2016;Dong et al., 2018;Saudi et al., 2019). The progressions in energy efficiency alongside the advancement of green sources of energy are also recognized as the highly effective way of abridging the decline in environmental condition and achieving sustainable growth (Saudi et al., 2019).

Data and Research Model
The data sample covers the period from 1980 to 2018 and is collected from several sources. CO 2 represents carbon dioxide emissions which are stemming from the burning of fossil fuels and the manufacture of cement. They include carbon dioxide produced during consumption of solid, liquid, and gas fuels and gas flaring. They are measured in per capita emissions and taken from Knoema. 1 The data on GDP (in constant 2010 US$) shows annual gross domestic product per capita and FDI variable for annual foreign direct investment (in constant 2010 US$ as same as GDP) which are from United Nations Conference on Trade and Development 2 . While NONRE1 is for non-renewable energy consumption including oil and natural gas energy, RE is for hydro-power renewable energy consumption and NONRE2 is for total non-renewable energy consumption oil, gas energy and coal, measured in terawatt-hours (TWh) per year, source from Global Change Data Lab of Oxford university in project of Our World in Data 3 .
The study is concentrated to analyze the short-term and long-term effects of economic growth and energy consumption impact on CO 2 emissions. In addition, the study also wants to test whether renewable energy will be effective in the role of reducing CO 2 emission while there is existence of FDI. The study added the squared term of real income per capita (lnGDP t ) 2 to examine the existence of EKC in Vietnam. Therefore the long-run empirical equation of the EKC hypothesis is presented as follows: Where t and ɛ denote time and error

ARDL Estimation
In order to test the role of renewable, non-renewable energy consumption in EKC context, the study applied the autoregressivedistributed lag (ARDL) technique of long run relationship (also known as bounds test). According to Pesaran et al. (2001) ARDL methodology allows that variables may be stationary in levels (I(0)) or the first difference (I(1)). Because of this convenience, ARDL method has been used in many studies and in the current 1 https://knoema.com/ 2 https://unctad.org/en/Pages/Home.aspx 3 https://ourworldindata.org/ study (Bölük and Mert, 2015). The ARDL framework is proposed for the study is as three function below: If the long-term connection among CO 2 , GDP, FDI and sources of energy consumption are presented with indication then we calculate the beta value of the short run coefficients by utilizing the below equation: The coefficient of the error-correction term (ECT) is the speed of adjustment parameter which shows how quickly the series attain a long-run equilibrium (Nkoro and Uko, 2016). The expected sign of this coefficient is negative and significant.

Descriptive Statistics and Unit Root Tests
The variables chosen for this study are the amount of CO 2 emissions (CO 2 ), GDP per capita (GDP), annual foreign direct investment (FDI), non-renewable energy consumption including oil and natural gas energy (NONRE1), renewable energy consumption (RE) and total non-renewable energy consumption as oil, natural gas energy and coal energy (NONRE2) in a period of 1980 -2018. The study presents summary of all variables in Table 1 and Figure 1 below.
In general, Viet Nam has average 30128.4 USD annual foreign direct investment, gross domestic product per capita is more than 713 USD. This country used about 146 terrawatt-hours oil and natural gas energy, 21.21 terawatt-hours renewable energy consumption and 246.21 terawatt-hours nonrenewable energy consumption (oil, natural gas energy and coal energy) per year. As a result, it emitted  NONRE1 is for oil and natural gas energy, RE is for hydropower renewable energy consumption and NONRE2 is for oil, gas energy and coal more than 0.9633 carbon emissions into the environment. Figure 1 presents the trends of the 6 time-series, highlighting the tendency toward rising of all the variables over the 1980-2018 period.
The Result from Figure 1 shows that all variables have break point. To overcome the problem, this study uses the (Zivot and Andrews, 1992) unit root test to find out account endogenous structural breaks in the series. The result of the Zivot and Andrews structural break unit root test are reported in

Where t and ɛ denote time and error and error d is dummy
In order to apply ARDL regression, all variables are transformed to their logarithmic forms to achieve consistent empirical evidences.
It is well known that log-linear equation moderates sharpness in the time series data and allows for better results that control variance as compared to simple specification (Mrabet and Alsamara, 2017).

This study uses Unit Root test namely ADF (Augmented Dickey
Fuller) to find out the order of integration of the variables in all 3 models and ensure continuity in time series from 1980-2018. The study applied the unit root test on the natural logarithms of the variables in level and difference forms. As the result of Table 3, the study concluded that all the variables are stationary at the 1 st difference.
Thus, ARDL bound test (Pesaran et al., 2001) is used for empirical analysis. It provides better results than multivariate cointegration approaches in case of small sample properties. This test of cointegration can be used regardless of the stationarity level of the underlying variables whether they are I (0), I(1) or a combination of both.
According to the results in Table 4, the calculated F statistic value in 3 models respectively are 7.163 (Modle 1); 12.788 (Model 2) and 5.027 (Model 3) which are above the I(1) critical value for 1% confidence intervals. Therefore, the ARDL bound test results reject the null hypothesis in all 3 models. It conclude that there is a long-term cointegration relationship between the dependent variable CO 2 and the independent variables GDP, GDP 2 , FDI, NONRE1, RE and NONRE2. These results also show that the independent variables in all models have effect on the carbon emissions in long-term. However, to measure the direction and magnitude of these effects, this study continues to perform short and long-term ARDL models.

ARDL Long-run and Short-run Results and EKC hypothes in Vietnam
The ARDL analysis is presented in Table 5 for the long-run and short-run estimates. The 3 models have the Durbin-Watson d statistic which are in turn of 2.1484; 2.406; 2.6928, are far from the center of its distribution (d = 3). Therefore, this study doesn't contain auto-correlation. Besides, the results of Ramsey RESET test have proofed no omitted variable bias problems in diagnostics tests.
ECT indicates the error correction term and shows the adjustment speed of the variables towards the long-run equilibrium which is presented by ADJ in the table. The ADJ coefficients which have statistically significant, confirm a deviation from the long run equilibrium level of CO 2 emission in one year is corrected by   In the long run, this study concluded identity of 3 models which have U-shaped relationship between GDP and CO 2 emissions. This finding is consistent with Tang and Tan (2015) the suggestion of Shahbaz et al. (2019) of period 1974-2016 that in long run this relationship will be N-shaped relationship. That means the curve has turn from inverted U-shaped into U-shaped in the period of this study. For that evidence, Vietnam may benefit from a drop in CO 2 emissions at some point, although it has to be cautious because the emissions may rise again when a second income turning point is reached.
The results in long run of 3 models show that FDI is positively and significantly at 5% linked with CO 2 emissions. This finding is in line with the results of Tang and Tan (2015) which implies PHH does exist in Vietnam in a period of 1980-2018. Vietnam should adopt stringent polices to select suitable FDI flow to ensure both aims of economic growth and environmental protection.
Besides that, the study in 3 models indicated NONRE1 and NONRE2 (oil, gas energy and coal energy) have significant at 1%, positively correlated to pollution in long run. These results indicate that nonrenewable energy consumption is not environmentally friendly. These souces of energy can increase environmental pollution and will not solve the climate crisis in near future. This finding is in line with the results of Ben Jebli and Ben Youssef (2015b), Shahbaz et al. (2016) and Demena and Afesorgbor (2020). Moreover, in model 2 and 3, renewable energy consumption is negatively correlated to CO 2 in long run. It is also significant at 1% (moldel 2) and 5% (model 3) which indicates that hydropower renewable energy can be a key solution in reducing the pollution level for Viet Nam and will be good for long-term energy strategy. These results are consistent with previous studies, such as Apergis and Payne (2009) In short run, the results of 3 models have significant but heterogeneous. Model 1 with NONRE1 (oil and gas energy) and model 3 with NONRE2 (total nonrenewable energy consumption) have U-shaped relationship between GDP and CO 2 emissions. While model 2 which includes NONRE1 (oil and gas energy) and RE (hydropower renewable energy) given out inverted U -shaped between GDP and CO 2 emissions. The results can be explained by the RE/NONRE1 ratio in model 2 is higher than RE/NONRE2 (Figure 1). The existence of EKC in short run in Model 2 indicates that increasing economic growth will lead to reduce CO 2 emissions. These results show there is the possibility of transferring process from PHH to pollution halo in Vietnam in recent. Vietnam has gradually comprehended that the influx of foreign direct investment should be responsible for domestic CO 2 mounting situation. To solve pollution environment, Vietnam has changed to use green and clean technology in industries little by little.
In addition, by increasing the share of renewable energy consumption in the ratio RE/NONRE1 that helps to reduce CO 2 emission in short term. On the other hand, renewable energy consumption is negatively correlated to CO 2 emissions in long-run but positive in short -run. It is also significant in both long and short run which indicates the host country has tried to use environment-friendly and  low carbon technology recently. However, this clean technology is not only imperfect but also impurities. The more incomplete the technology is, the more degraded environment is.
To check the stability of the ARDL model, we applied the CUSUM and CUSUMsq tests recommended by Brown et al., (1975) to examine the constancy of the parameters.
The results in Figures 2 and 3 show that CUSUM and CUSUMsq in model 1 and 2 are between the upper and lower critical bounds at the 5% significance level, confirming the stability of the ARDL estimates. However, the model 3 in Figure 4 is beyond the lower critical bounds. This indicates the model 3 is not quite stability.

Granger Causality Analysis
This section is devoted to the causality analysis among CO 2 emissions and the considered variables in Vietnam. The study conducts the relationship between variables to carbon emissions by VECM Granger test. For the best results, this study has just tested VECM Granger test for Model 2 based on is informative and stability.
The results in Table 6 present the results of the causal test between carbon emissions, energy sources and economic growth based on the VECM Granger causality test.
Beginning with the short-run effects, there is unidirectional causality runs from CO 2 emissions and GDP per capita to annual foreign direct investment, from oil and natural gas energy (NONRE1) to GDP square per capita and renewable energy consumption (RE). In addition, the bidirectional causality is found between annual foreign direct investment and CO 2 emissions. This result confirms when the host country attracts foreign direct investment, it affects CO 2 emissions in short term, especially in the developing country. And in contrast, pollution also changes the way to classify, choose and use the investment from foreign country well.
Turning to the long-run causality result, the unidirectional causality is found running from GDP square per capita and foreign direct investment to CO 2 emissions without a feedback. There is reveal that CO 2 emissions are caused by developing economy. This will certainly happen when Vietnam has achieved economic growth to the certain extent.
Besides that, this study shows oil and natural gas energy (NONRE1) and renewable energy consumption (RE) which cause CO 2 emissions in the Granger sense. This implies that Vietnam should limit the industries to dirty energy -oil and natural gas energy which causes environmental pollution, leads to global warming. In addition, by turning to more renewable energy -forms of hydrogen, we can make a big impact and lower greenhouse gas emissions and pollution.
We also found bi-directional causality running between GDP per capita to CO 2 emissions in long term. This implies there is close and connected relationship between economics and the quality of the environment. We thus conclude N-shaped relationship between economic growth and CO 2 emissions will happen in Vietnam in near future.

DISCUSSION AND CONCLUSION
This paper has investigated the relationship between economic growth and environmental degradation in Vietnam by choosing CO 2 emission as indicator. FDI and non-renewable, renewable energy consumption as determinants of CO 2 emissions over the 1980-2018 period using annual data. The three models with the different sources of energy consumption have showed the empirical results that the EKC does not exit in Vietnam in long run and FDI has also significantly contributed to the increase of CO 2 emissions. That implies Vietnam has attracted non -environmental friendly FDI flow for years and followed the PHH. However, the negative coefficients of FDI for three models in short run would suggest the current stricter environmental regulations has reversed the effect of FDI on environment. Vietnam should have appropriated stringent regulations to select the FDI that comes into the country. Many countries are now promoting the so-called "green" FDI that focuses on FDI that can promote economic growth and also internalizes the adverse environmental externalities associated with industrial production (Golub et al., 2011).
The other findings of the study are confirmations of the effect of energy consumption sources as non-renewable energy and renewable energy to amount of CO 2 emissions in Vietnam. All of three models reveals the consistent results of significant and positive effect from non-renewable energy and negative effect from renewable energy consumption to CO 2 emissions in long run. The green energy shows the effective ways to limit environmental pollution to pursue sustainable development. Therefore, Vietnam should keep on increasing the share of renewable energy and decreasing the consumption of non-renewable energy for a better environment. Though the paper has just tested hydro -renewable energy, Vietnam should also increase energy consumption of various renewable sources, that would help to decrease the CO 2 emission, protect environment while developing economy. In addition, government should more focus on environmental public awareness to promote renewable energy policies for households such as solar energy and as well as to change customer's preferences to energy saving products. *, **, ***Represent the significance level at 10%, 5% and 1% respectively