Is There an Environmental Kuznets Curve for Bangladesh?

University of Western Sydney From the SelectedWorks of Fazle Rabbi 2014 Is There an Environmental Kuznets Curve for Bangladesh? Fazle Rabbi Available at: https://works.bepress.com/fazle_rabbi/3/

Conference of the International Journal of Arts & Sciences, CD-ROM. ISSN: 1943-6114 :: 07(02):221 231 (2014) Copyright c 2014 by UniversityPublications.net IS THERE AN ENVIRONMENTAL KUZNETS CURVE FOR BANGLADESH? Fazle Rabbi The University of Notre Dame, Australia SM ZobaidulKabir and Delwar Akbar Central QueenslandUniversity, Australia This article investigates the current status of the relationship between emissions, energy consumption, final output and trade openness for Bangladesh over a period of time (1972-2008). The empirical results for the case of Bangladesh suggest that there is an existence of a robust long-run relationship along with the short-run dynamic adjustment between the variables. The study indicates that high economic growth may increase environmental degradation in future and therefore, policies to tackle environmental pollution are imperative where the use of both market and non-market based management tools are important. Finally, this study provides some future research directions which are beyond the scope of current research. Keywords: emissions, Energy consumption, Environmental pollution, Kuznets curve. Introduction The Environmental Kuznets Curve (EKC) is a very well known instrument in literature (Friedl&Getzner, 2003) which summarizes what has meanwhile become well known as the inverted U-shaped relationship between emissions and per capita (Friedl&Getzner, 2003; Stern, 2004). Moreover, a variety of recent and emerging line of literature (Ang, 2007, 2008; Soytas, Sari & Ewing, 2007;Shahbaz, Lean &Sabbir, 2010)seems to incorporate both the economic growth and emissions, as well as economic growth and energy consumption into a multivariate framework to analyse empirically the dynamic relationships altogether. Bangladesh is a developing country with 150 million populations. Since the early 1990s the economic growth rate of the country has always been more than 6%. At the same time, pressure of economic growth on environment has been an issue of debate. Because of the industrialisation and rapid urbanisation, the use of coal and oil as the major sources of energy, environmental pollution has been increasing both in urban and rural areas. Given the environmental degradation for economic growth, it is imperative to know the position of Bangladesh within the process of sustainable development. Among the pollutants, CO 2 is a global pollutant and it is considered as the most important gas that is responsible for greenhouse effect and climate change. Bangladesh s contribution to CO 2 emissions is quite low; per capita CO2emissions was only at.03 in 2011; much lower than the OECD average of 12. Also Bangladesh s share is only 0.14% to the global CO 2 emissions in world emissions (World Energy 221

222 Is There an Environmental Kuznets Curve for Bangladesh? Council, 2012). Nonetheless, the emissions of CO 2 by Bangladesh is increasing this will continue because of more dependency on coal and oil.therefore, the pollution level of Bangladesh s economy will be an increasing concern in the future and therefore it warrants additional considerations. The aim of this paper is to examine the dynamic relationship between emissions, energy consumption, output and trade openness for Bangladesh over the period 1972-2008, based on the EKC hypothesis as well as by using the Johansen (1991) co-integration and vector error-correction model. While there are some studies in developing countries and developed countries on the relationship between income and the environment (for example, Akbostanci et al. 2009; Stern, 2004; Dinda, 2004, Dasguptaet. Al., 2002), there is a dearth of scientific information in Bangladesh on the relationship between income and environment. This article is an attempt in filling this gap. This study is divided into five sections including introduction and conclusions. Section 2 analyses the econometric specification used in this study. The third section explains the models and data used. Section four consists of the estimation results of the time series data. At last, the sixth section concludes the study. Econometric Specification This article analyses the relationship between economic growth and environmental pollution with particular reference to the emission of CO 2, and economic growth and energy consumption in Bangladesh. A single multivariate framework has been used in this study inorderto analyse the dynamic relationship among the chosen variables as previously used by many authors ( for example, Ang, 2007, 2008; Soytas et al., 2007; Shahbaz, Lean &Sabbir, 2010). Recent literature, some scholars(halicioglu, 2009; Jalil& Mahmud, 2009) included foreign trade as an independent factor in their models to examine the impact of foreign trade on environmental pollutants. In our study, we have considered trade policies, like Shahbaz, Lean &Sabbir (2010), to analyse the impact of trade restrictions on environmental pollution. In this study, the relationship between emissions, energy consumption, economic growth and trade openness is specified as follows: Where, CO 2 E is the emissions per capita, EUSE is the energy consumption per capita, GDPis the real andgdp 2 is the simplesquare form of the real. This study also follows Edwards (1994), MacDonald & Ricci(2004), Mkenda (2001),Shahbaz et al. (2010) where the degree of openness of the economy proxies that can allow the exchange and trade controls which is the summation of exports and imports as share of. This can be expressed as: We can now use this degree of openness with equation (1) in the following estimable form including the error term, : Continuous economic growth usually causes more economic activities within the economy, which would lead more energy consumption and regular emissions. For this reason, we can expect in equation (2). According to the EKC hypothesis the sign of is expected to be positive,, while is supposed to be negative,. However, the sign of depends on the stage of economic development of a country. Jalil et al. (2009) and Shahbaz et al. (2010) mention that for developed countries where environmental laws are very strict, the sing of can be negative because of producing less pollution intensive goods within the economy. As these types of countries import such goods from

Fazle Rabbi et al. 223 less restrictive environmental laws imposing countries, they can expect. On the other hand, Grossman & Krueger (1995) and Halicioglu (2009) argue that developing economies can expect the opposite sign for, which is, because of their intention to produce more pollution intensive goods within the economy. Thus, the expected sign for is mixed in equation (2). Estimation Strategy Several techniques can be used for parameters estimation in equation (2) and these techniques can be ranging from classical regression methods to cointegration-based techniques.this study employs Johansen s (1991) cointegration procedure which has emerged as the most commanding and accepted technique in this area. The Johansen (1991) approach is popular because it confines the fundamental time series properties of the data. According to Chowdhury (1998), this involves determining the number of cointegrating vectors in a system and then estimating them. The Johansen procedure starts with a general vector autoregressive model with variables of interest. This is then re-parameterised as a system of error correction model, so that the vector in the auto-regression model consists of lagged first differenced terms and a set of lagged level terms. Applying Ordinary Least Square (OLS) can provide consistent long-run parameter estimations for each equation. Thus, Johansen statistics are able to give a decisive test in cases where other test statistics are border-lined with respect to cointegration (Chowdhury, 1998). The first issue in this estimation strategy is to determine the order of integration of all the variables in our model; the next is to perform cointegration tests in order to identify any long-run relationships in the variables. A short-run vector error correction model is then estimated on the condition of finding cointegration in the previous step, and finally, residual diagnostic checks are performed at the last step. In a classical regression model we generally deal with the relationship between stationary variables but most of the economic indicators normally follow a non-stationary path. According to Gujarati and Porter (2009), if the dependent variable is a function of a non-stationary process, the regression will produce unauthentic results. In such a case, it may possible to get significant t-ratios and a very high even though the trending variables are completely unrelated. Therefore we need to perform unit root or stationarity tests before proceeding with the tests for cointegration and estimation of parameters. There are several ways to test the stationarity of the series. The unit root test (Augmented Dickey-Fuller and Dickey-Fuller GLS) and one stationarity test (Kwiatkowski et al., 1992) have been considered in this study. We perform both of these tests in this study in order to achieve what Brooks (2002) refers to as confirmatory data analysis. After determining the order of integration of the variables of our model, we will conduct the Johansen (1991) coninteration test. The purpose of the Johansen (1991) cointegration test is to determine whether the variables in our model are cointegrated or not. The presence of a cointegration relation forms the basis of the Vector Error Correction Model (VECM) specification. In the following section we are briefly describing the Johansen methodology. Assume a vector:, and assume that the vector has a VAR representation of the form: Where, is a vector of deterministic variables, is a vector of white noise error terms, and is a matrix of coefficients. In order to use the Johansen test, the equation (3) needs to be turned into a VECM specification (Brooks, 2002), which can be specified as:

224 Is There an Environmental Kuznets Curve for Bangladesh? where is a vector of variables, are all variables, indicates the first difference operator, is a coefficient matrix, and is a matrix whose rank determines the number of cointegrating relationships. Brooks (2002) mentions that Johansen s cointegration test is to estimate the rank of the matrix from an unrestricted VAR and to test whether we can reject the restrictions implied by the reduced rank of. If is of full rank, it suggests that variables are level stationary and if it is of zero rank, no cointegration exists among the variables. On the other hand, if is of reduced rank, then there exists matrices and such that: where represents the speed of adjustment matrix, indicating the speed with which the system responds to last period s deviations from the equilibrium relationship, and is a matrix of long-run coefficients. We can test the rank of the matrix after identifying the appropriate Vector Auto Regressive (VAR) order and the deterministic trend assumption. There are two likelihood ratio (LR) test statistics for cointegration under the Johansen approach. The first is the trace and the maximum eigenvalue statistics, which are specified as follows: and where is the number of cointegrating vectors under the null hypothesis and is the estimated value for the th ordered eigenvalue from the matrix in equation (5). The trace statistic sequentially tests the null hypothesis that the number of cointegrating relations is against the alternative of cointegrating relations, where is the number of endogenous variables. Brooks (2002) mentions that the maximum eigenvalue test conducts separate tests on each eigenvalue and has as its null hypothesis that there are cointegrating vectors against an alternative of To determine the rank of the matrix, the above trace and maximum eigenvalue test statistics are compared to the critical values from Osterwald-Lenun (cited in Takaendesa, 2006), which provide a more complete set of critical values for the Johansen test. For both tests, if the test statistic is greater than the critical values, the null hypothesis that there are cointegrating vectors is rejected in favour of the corresponding alternative hypothesis. The trace and maximum eigenvalue statistics may yield conflicting results. To deal with this problem, Johansen and Juselius (1990) recommend choosing the appropriate estimated cointegrating relations based on the interpretability of those cointegrating relations. Luintel and Khan (1999) offer a different view, arguing that the trace test is more robust than the maximum eigenvalue statistic in testing for cointegration. Once the number of cointegrating vectors in the model has been identified, a VECM (equation 4) can be estimated by specifying the number of cointegrating vectors, trend assumption used in the previous step and normalising the model on the true cointegration relation. Takaendesa (2006) appropriately mentions that the VECM is merely a restricted VAR designed for use with non-stationary series that have been found to be cointegrated. The specified cointegrating relation in the VECM restricts the long-run behaviour of the endogenous variables to converge to their cointegrating relationships, while allowing for short-run adjustment dynamics. Because of this, after estimating the parameters, the residuals from the VECM need to be checked for autocorrelation and heteroskedacity.

Fazle Rabbi et al. 225 Table 1. Unit root test TEST Intercept Trend & Intercept Intercept Trend & Intercept Intercept Trend & Intercept Intercept Trend & Intercept Intercept Trend & Intercept ADF Level 4.627* 1.511 3.661* 1.272 6.687* 3.079 12.594* 7.168* 0.592-1.740 1 st Difference -6.283* -7.517* -5.534* -6.854* -7.133* Order of Integration I(0) I(1) I(0) I(1) I(0) I(1) I(0) I(0) I(1) I(1) DF-GLS Level 0.428-2.125 3.222* -0.248 0.549-1.147 0.595-1.503 0.964-1.794 1 st Difference -8.764* -8.861* -6.702* -3.225* -5.434* -1.044-2.904*** -6.456* -7.319* Order of Integration I(1) I(1) I(0) I(1) I(1) I(1) I(2) I(1) I(1) I(1) KPSS Level 0.714 0.209* 0.712 0.210 0.674 0.214 0.645 0.210 0.684 0.171 1 st Difference 0.329* 0.564* 0.109* 0.649* 0.226* 0.647* 0.206* 0.260* 0.060* Order of Integration I(1) I(0) I(1) I(1) I(1) I(1) I(1) I(1) I(1) I(1) Note: *, ** and *** indicate No Unit Root at 1%, 5% and 10% respectively.

226 Is There an Environmental Kuznets Curve for Bangladesh? In order to apply these estimation techniques, time series data has been collected from the International Financial Statistics of IMF and different issues of Annual Statistics of Bangladesh Bank. These include: per capita emissions, energy consumption, per capita real and the amount of trade over the period of 1972-2008. The per capita is measured in domestic currency. The real is measures as nominal divided by the deflator (2000 = 100). Finally, trade openness ratio is the total value of exports and imports as a share of nominal. Findings and Analysis The first step of the Johansen s (1991) technique is to determine the order of integration of the series. We begin with the unit root test for all the variables under this study, using the Augmented Dickey Fuller (ADF) and Dickey Fuller GLS (DF-GLS) tests, with the null hypothesis of a unit root against the alternative of stationarity of data series. The test results are reported in Table 1.1, which includes both the options intercept and intercept and trend. The results of the ADF tests in Table 1 show that with the intercept and trend option, four series (,,, ) are first difference stationary I(1), while the other variable ( ) is level stationary I(0). With the same option, the DF-GLS tests in Table 1 show that all the variables are first difference stationary I(1). Again, with the intercept only option, the ADF tests show that all the variables, except only one ( ), are level stationary I(0). To confirm the stationarityof,, and, we apply a third test, the Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test. ADF and DF-GLS methods test the null hypothesis of a unit root, while the KPSS has as its null hypothesis that the series is stationary; therefore, a rejection of the null hypothesis under both the ADF and the DF_GLS means that the series does not have a unit root, while the rejection of the null hypothesis under the KPSS in interpreted as evidence of non-stationarity, or presence of a unit root in the series. The results of the KPSS tests in Table 1 show that with the intercept option alone, all the variables are first difference stationary I(1), with both option one variable ( ) is level stationary I(0). While the ADF and DF-GLS show the same variable as first difference stationary I(1), all these tests exhibited that all the variables in our model are first difference stationary I(1). Before conducting the Johansen (1991) procedure to perform cointegration analysis, we also investigate any possible structural break in the data set. According to Agung (2009), a breakpoint in any macroeconomic indicator or growth curve over time need to be identified before collecting the corresponding time series data. We concludedhere that there is a possible break point in 1992, when we observed changes in the trends of most of the macroeconomic variables because of the political transition of Bangladesh from dictatorship to elected democratic government. However, in Table 2 we see that based on a p-value = 0.0000, the null hypothesis of no breaks at specified breakpoints (here, 1992) is rejected. On the basis of this, we construct a variable which assumes value 0 from year 1972 to 1992 and 1 thereafter. Table 2. Stability test: Chow s Breakpoint Test. Chow Breakpoint Test: 1992 Null Hypothesis: No breaks at specified breakpoints Varying regressors: All equation variables Equation Sample: 1972 2008 F-statistic 22.43839 Prob. F(2,33) 0.0000 Log likelihood ratio 31.76895 Prob. Chi-Square(2) 0.0000 Wald Statistic 44.87677 Prob. Chi-Square(2) 0.0000

Fazle Rabbi et al. 227 Table 3. Johansen Cointegration Rank Test results with structural break. Sample (adjusted): 1974 2008 Included observations: 35 after adjustments Trend assumption: Linear deterministic trend Series: CO2E EUSE GDP GDP2 OPEN DUMMY Lags interval (in first differences): 1 to 1 Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None * 0.789380 131.3411 95.75366 0.0000 At most 1 * 0.583143 76.82164 69.81889 0.0124 At most 2 0.446683 46.19619 47.85613 0.0710 At most 3 0.364665 25.48232 29.79707 0.1449 At most 4 0.176955 9.606203 15.49471 0.3122 At most 5 0.076624 2.790154 3.841466 0.0948 Trace test indicates 2 cointegratingeqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None * 0.789380 54.51943 40.07757 0.0006 At most 1 0.583143 30.62544 33.87687 0.1164 At most 2 0.446683 20.71387 27.58434 0.2939 At most 3 0.364665 15.87612 21.13162 0.2322 At most 4 0.176955 6.816049 14.26460 0.5111 At most 5 0.076624 2.790154 3.841466 0.0948 Max-eigenvalue test indicates 1 cointegratingeqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values 1 Cointegrating Equation(s): Log likelihood 255.3900 Normalized cointegrating coefficients (standard error in parentheses) CO2E EUSE GDP GDP2 OPEN DUMMY 1.000000-0.002944-1.248766 0.813093 0.002941-0.028895 (0.00047) (0.24406) (0.13830) (0.00054) (0.00587)

228 Is There an Environmental Kuznets Curve for Bangladesh? Now we run our cointegration analysis, where we also consider the structural break in the data set by considering. Table 1.3 shows the cointegration test results for our model based on trace and maximum eigenvalue statistics. In this cointegration analysis, we consider five variables (,, and ) to see their influence on. Starting with the trace test, the null hypothesis of no cointegrating vector is rejected, since the test statistic of about 131.34 is greater than the 5% critical value of approximately 95.75. In the same way, the null hypothesis that there is at most one cointegrating vector is also rejected. The trace test, therefore, indicates two cointegrating relationship at the 5% level of significance. On the other hand, the maximum eigenvalue form of the Johansen test also rejects the null hypothesis of no cointegration. However, the null hypothesis, that there is at most one cointegrating vector, cannot be rejected since the test statistic of approximately 30.63 is now less that the 5% critical value of around 33.88. The maximum eigenvalue test, therefore, suggests that there is one cointegrating relationship in our model. However, the value of is normalised to one, and after this we get the cointegrating equation (equation 8) as follows, with the standard error in parentheses: (0.01) (0.244) (0.138) (0.001) (0.006) The results in equation (8) are statistically significant and also indicate the expected signs of the coefficients of the explanatory variables. It shows the long run marginal impact of economic growth, energy consumption and trade openness on emissions. The results reveal that increase in energy consumption will increase emissions. From the equation (8) we also see the confirmation of the existence of inverted-u relationship between economic growth and emissions. The results indicate that 1 percent rise in real will rise emissions by 1.25 percent while negative sign of seems to corroborate the delinking of emissions and real at higher level of income in the country. Thus, this evidence confirms the increment of emissions at the very early stage of economic growth of any country and decline after a threshold point. These findings are consistent with many latest empirical evidences (such ashe, 2008; Song, Zheng & Tong, 2008; Halicioglu, 2009; Fodha&Zaghdoud, 2010 and Lean & Smyth, 2010). The coefficient of trade openness ( ) in our results also shows that it has inverse impact on emissions. This result is consistent with the findings of Antweiler, Copeland & Taylor, 2001; Copeland & Taylor, 2005; McCarney&Adamowicz, 2006; and Managi, Hibiki& Tetsuya, 2008. The logic behind these findings is that trade openness reduces emissions through technological progress in the country.the consideration of structural break also shows statistically significant result. The variables of our model are cointegrated in the long run, so there exists an error correction mechanism which brings together the long-run relationship with its short-run dynamic adjustments. The estimate of parsimonious dynamic Error Correction Model [equations (9)] is reported in Table 1.4 together with the most common diagnostic tests. In Table 1.4, the coefficients of the error correction term of equation (9) shows that has the significant coefficient at 5% level, with a t-value of -2.12, and has a correct negative sign. This suggests that the equation constitutes the true cointegrating relationship in thiscointegrating vector. A crucial parameter to note in the estimation of Vector Error Correction models is the coefficient of adjustment, which, in this study, measures the speed of adjustment in the following a shock in the system. This implies that the long-run equilibrium deviation has a significant impact on the growth of emissions. The speed of adjustment at 52 percent a year is considered relatively high. In other words, it takes about two years to achieve long-run equilibrium whenever here is a deviation from the long-run steady state. The results are not surprising given that there is little control on the growth of emissions. In addition to the Vector Error Correction results in Table 1.4, we report some diagnostic checks results. Diagnostic checks are crucial in this analysis, because if there is a problem in the residuals from the estimation of a model, it is an indication that the model is not efficient, and that parameter

Fazle Rabbi et al. 229 estimates from such a model may be biased. The diagnostic test results in Table 1.4 indicate that there is no serial correlation and the residual term is normally distributed, also there is no evidence of heteroske dasticity. Table 4. Error correction model of for Bangladesh: 1972 2008. VARIABLES Equation-9-0.52** (-2.12) 4.54 (0.05) -2.65** (-2.10) 1.92** (2.14) 0.00 (0.29) Constant 0.02 (-1.39) DIAGNOSTICS: R-squared 0.36 Adj. R-squared 0.20 F-statistic 2.2123 Akaike AIC -5.6285 Serial Correlation LM 20.34 {0.73} Normality Test (Jarque-Bera) 36.05 {0.0001} Heteroskedasticity 202.08 {0.35} Notes: i) *, ** and *** indicate significant at 1%, 5% and 10% levels respectively. ii) Figures within parenthesis, ( ), represent the t-statistics. iii) Figures within { } are P values for the residual diagnostic checks under the null of no serial correlation, no heteroskedasticity and normally distributed residuals for the LM,heteroskedasticity and normality test, respectively.

230 Is There an Environmental Kuznets Curve for Bangladesh? Conclusion In this paper, we examine the dynamic relationship between emissions, energy consumption, output and trade openness for Bangladesh over the period 1972-2008, based on the EKC hypothesis as well as by using the Johansen (1991) conintegration and vector error-correction model. The empirical results for the case of Bangladesh suggest the existence of a robust long-run relationship along with the short-run dynamic adjustment between the variables. The long-run relationship in which emissions is the dependent variable is used to test the short-run and long-run elasticities of emissions with respect to all other explanatory variables. Our results show that Bangladesh should design new environmental policies to reduce environmental degrading. The results of this study also confirm the existence of inverted-u relationship between economic growth and emissions. Certainly, the high economic growth gives rise to environmental degradation, but on the other hand, the reduction in economic growth can increase the level of unemployment further in a developing country like Bangladesh.Therefore, the policies to tackle environmental pollutants require the identification of some priorities to reduce the initial costs and efficiency of investments for economic development. Bangladesh should measure the exact scale of environmental pollutants that it generates and imports by industries. Identification of these things can help Bangladesh government to issue most relevant rules and regulations for tackling environmental degradation competently. However, apart from rules and regulations to reduce pollutant emissions, the market based solutions in the form of pollutant taxes, which are already implemented in many countries in the world, would ease the extent of this problem in Bangladesh. Government should incorporate environmental concerns into her macroeconomic policies more intensely to reduce the pollutant emissions and to sustain economic growth. One of the limitations of this study is that the analysis is at an aggregated level. Different industries have different intensities of energy consumption. It would be difficult to obtain disaggregated data on energy consumption, but if such data could be obtained for Bangladesh, such a project would be a useful topic for future research. In this study, we use electricity consumption as a proxy for energy consumption and emissions as a proxy for environmental degradation. Future studies which will use other proxies for energy consumption and environmental degradation may provide further insight regarding the link between environmental degradation, energy consumption and economic growth.moreover, examining the contributory relationship between economic growth, pollution emissions and other potentially relevant variables such as automobile use, health expenditure and urbanization would be another direction for further research. References 1. Akbostanci,E., Turut-Asik, S., and Tunc, GJ., 2009, The relationship between income and environment in Turkey: Is there an environmental Kuznets curve? Energy Policy, vol.37 (3):861-867. 2. Agung, IGN 2009, Time Series Data Analysis Using EViews, John Wiley & Sons (Asia) Pte Ltd, Singapore. 3. Ang, JB 2007, 'CO2 emissions, energy consumption and output in France', Energy Policy, no. 35, pp. 4772-8. 4. 2008, 'Economic Development, pollutant emissions and energy consumption in Malaysia', Journal of Policy Modelling, no. 30, pp. 271-8. 5. Antweiler, W, Copeland, RB & Taylor, MS 2001, 'Is free trade good for the emissions: 1950-2050.', The Review of Economics and Statistics, no. 80, pp. 15-27. 6. Brooks, C 2002, Introductory econometrics for finance, Cambridge University Press, Cambridge. 7. Chowdhury, MB April, 1998, 'Resources Booms and Macroeconomic Adjustment: Papua New Guinea', Doctor of Philosophy thesis, The Australian National University.

Fazle Rabbi et al. 231 8. Copeland, BR & Taylor, MS 2005, 'Free trade and global warming: a trade theory view of the Kyoto Protocol.', Journal of Environmental Economics and Management, no. 49, pp. 205-34. 9. Dasgupta, S., Laplante, B, Wang, H and Wheeler, D., 2002, Confronting the Environmental Kuznets Curve, Journal of Economic Perspectives, Vol.16 (1): 147-168. 10. Dinda, S., 2004, Environmental Kuznets Curve Hypothesis: A survey, Ecological Economics, vol.49 (4): 431-455. 11. Edwards, S 1994, Real and monetary determinants of real exchange rate behavior: theory and evidence from developing countries. In: Wlliamson, J. (ed). Estimating equilibrium exhangerates., Washington: Institure for International Economics. 12. Fodha, M &Zaghdoud, O 2010, 'Economic growth and pollutant emissions in Tunisia: an empirical analysis of the environmental Kuznets curve', Energy Policy, no. 38, pp. 1150-6. 13. Friedl, B &Getzner, M 2003, 'Determinants of CO2 emissions in a small open economy', Ecological Economics, no. 45, pp. 133-48. 14. Grossman, G & Krueger, A 1995, 'Economic environment and the economic growth.',quarterly Journal of Economics, no. 110, pp. 353-77. 15. Gujarati, DN & Porter, DC 2009, Basic Econometrics, 5th ed. edn, McGraw-Hill Irwin, c2009, Boston. 16. Halicioglu, F 2009, 'An econometric study of CO2 emissions, energy consumption, income and foreign trade in Turkey', Energy Policy, no. 37, pp. 1156-64. 17. He, J 2008, 'China's industrial SO2 emissions and its economic determinants: EKC's reduced vs. structural model and the role of international trade', Environment and Development Economics, no. 14, pp. 227-62. 18. Jalil, A & Mahmud, S 2009, 'Environment Kuznets Curve for CO2 emissions: a cointegration analysis for China', Energy Policy, no. 37, pp. 5167-72. 19. Johansen, S 1991, 'Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models', Econometrica, no. 59, pp. 1551-80. 20. Johansen, S &Juselius, K 1990, 'The full information maximum liklihood procedure for inference on cointegration', Oxford Bulletin of Economics and Statistics, no. 52, pp. 211-44. 21. Kwiatkowski, D, Phillips, PCB, Schmidt, P & Shin, Y 1992, 'Testing the null hypothesis of stationarity against the alternative of a unit root', Journal of Econometrics, no. 54, pp. 159-78. 22. Lean, HH & Smyth, R 2010, 'CO2 emissions, electricity consumption and output in ASEAN', Applied Energy, no. 87, pp. 1858-64. 23. Luintel, LB & Khan, M 1999, 'A quantitative reassessment of the finance - growth nexus: evidence from a multivariate VAR.', Journal of Development Economics, no. 60, pp. 381-405. 24. MacDonald, R & Ricci, L 2004, 'Estimation of the Equilibrium Real Exchange Rate for South Africa', South African Journal of Economics, no. 72, issue.2, pp. 282-304. 25. Managi, S, Hibiki, A & Tetsuya, T 2008, Does trade liberalization reduce pollution emissions?, Discussion Paper Series 08-E-013, Research Institure of Economy, Trade and Industry (RIETI). Available at URL: http://www.rieti.go.jp/jp/publications/dp/08e013.pdf. [Accessed on 12 October, 2011]. 26. McCarney, G &Adamowicz, V 2006, The effects of trade liberalization of the environment: an empirical study., 2006 Annual Meeting., International Association of Agricultural Economists., Queensland, Australia.Available at URL: http://ageconsearch.umn.edu/handle/34157. [Accessed on 23 November, 2011]. 27. Mkenda, BK 2001, Long-run and short-run determinants of the real exchange rate in Zambia, No. 40, Goteborg University Working Papers in Economics. 28. Shahbaz, M, Lean, HH &Shabbir, MS 2010, Environmental Kuznets Curve and the role of energy consumption in Pakistan, MonashUniverisity, Business and Economics. 29. Song, T, Zheng, T & Tong, L 2008, 'An empirical test of the environmental Kuznets curve in China: a panel cointegrationapproach.',china Economic Review, no. 19, pp. 381-92. 30. Soytas, U, Sari, R & Ewing, T 2007, 'Energy consumption, income, and carbon emissions in the United States', Ecological Economics, no. 62, pp. 482-9. 31. Stern, D., 2004, The rise and fall of the Environmental Kuznets Curve, World Development, vol.32 (8):1419-1439. 32. Takaendesa, P 2006, 'The Behaviour and Fundamental Determinants of The Real Exchange Rate in South Africa', Masters in Commerce (Financial Markets) thesis, RHODES UNIVERSITY. Available at URL: http://eprints.ru.ac.za/516/1/takaendesa_mcom.pdf, [Accessed on 01 November, 2011].