# Introduction lobal warming and climate change attracted considerable attention worldwide. The intergovernmental panel on climate change (IPCC) reported that the global temperatures increase by 1.1 to 6.4 ? C and rise in the sea level of about 16.5 to 53.8 cm by 2100 (IPCC,2007). This would have tremendous negative impact on the half of the population of the world live in coastal areas (Lau et al., 2009). In this circumstance many countries like Bangladesh will totally submersed by sea water by 2100. Bangladesh is a small developing country in South-east Asia. Its population is above 160 million and the world's most density of population is situated here. Bangladesh is also recognized worldwide as one of the most vulnerable countries to the impact of climate change. For the past few decades, Bangladesh government has been showing concern about environmental pollution. Here with the production and economic activities it emits huge amount of carbon dioxide every year especially from fossil fuels, gas fuels, liquid fuels and solid fuels. On the other hand higher economic growth causes environmental degradation threatens the sustainability of the environment because economic growth is closely related to energy consumption which is responsible for higher levels of 2 CO emissions. It became the general consensus that higher economic growth should not be pursued at the expense of the environment and this issue raised the question of how economic growth can be made more sustainable. Sustainable development defined by Brundtland (1987) as development that meets the needs of the present without compromising the ability of future generations to meet their own needs. Some of international organizations around the world continuously attempt to reduce the adverse impacts of global warming. One such attempt is the Kyoto Protocol agreement, made in 1997 as an attempt to reduce the adverse impact of global warming. Among the variety of polluting substances, Carbon Dioxide ( 2 CO ) is a major one and represents 60 percent of green house gas emission (World Bank, 2007). # II. Literature Review Grossman and Krueger (1991) and Kuznets (1955) states that in the early stages of economic growth, environmental quality decreases with an increase in per capita income, but after a certain level environmental degradation starts decreasing with the increase in the level of per capita income, thus resulting in an inverted U-shaped curve (i.e. Environmental Kuznets Curve, EKC # CO emissions in Saudi Arabia. He analyzed that the long run income elasticity of carbon emissions is greater than the short run income elasticity of carbon emissions. This implies that income leads to greater carbon dioxide emissions in the long run. Islam, et al., (2012), found that there is a strong positive relationship between international trade and carbon ( 2 CO ) emissions from the gas fuels of various manufacturing sector of Bangladesh. Bloch, et al. (2011) analyzed that the relationship between coal consumption and GDP in China using both a supply side and a demand side framework. The error correction mechanism (ECM) is used to examine both short run and long run Granger causality. The results shows that coal prices Granger cause coal consumption, so a reduction in pollution without restricting economic growth may be possible by withdrawing the current policy of coal subsidization by the Chinese Government and replacing it with a policy of subsidizing greener energy sources. Gunter, (2010), analyzed in context of Bangladesh that the lower GDP growth rates imply higher population growth where the long term impact of low GDP growth on 2 CO emission is actually worse. Higher GDP growth rates will increase 2 CO emission faster, but it implies that the peak of 2 CO emission reaches earlier and due to the lower population, at a lower emission level. In other words, development can be considered to contribute to lower long run 2 CO emissions. Salequzzaman and Davis (2003) found that there are unique challenges for ecologically sustainable development with a very high population density, a still high population growth rate and limited natural resources. A significant program of environmental education and development of local expertise is needed for massive changes in behaviour with respect to the environment.The formal education system provides a ready framework for reaching a large part of the existing population and can help make future generations conscious of the importance of environmental conservation. In Bangladesh, NGOs and universities with environmental education departments can play a significant role in teacher training and providing materials for formal and non-formal education. Review of literature helps to know the research gap. That is why, a number of literatures have been reviewed to know the Causal Relationship between Education, 2 CO Emission and Economic Growth and identified research gap in this field. Environmental pollution education is a new phenomenon in the world and in context of Bangladesh it is also very recent idea. Most of the study relates this environmental pollution to other things rather than education. But education is vital elements that create awareness especially among those are the students, because they are the future of the country. Very few studies are found on the relevant field. Moreover it is observed that no specific work is done by using empirical model to determination the causality between education and environmental pollution on this topic in Bangladesh. So the researcher thinks, there prevail an immense research gap which is the main justification of the research. # III. # Model Specification and Variables The study applied multivariate model analysis techniques to examine the relationships among environmental pollution i.e. # CO emission, education expenditure and GDP growth in Bangladesh. The study based on the assumption that in GDP production is driven by high energy consumption that is likely to produce 2 CO emissions that causes environmental pollution and education expenditure in GDP is driven to create awareness through education among people about environmental pollution. The basic form of the relationship among the variables can be expressed as: t t t t GDP Ed Ep ? ? ? ? + + + = 2 1 (1) Where, = Ep Environmental pollution i.e. IV. # Empirical Study The empirical study consists of unit root test, the cointegration test and error correction mechanism. These are discussed below. # a) Testing Methods of Unit Roots Testing for the unit root problem the Augmented Dickey-Fuller test and Phillips-Perron test were used here. # i. Augmented Dickey-Fuller (ADF) Test The Augmented Dickey-Fuller (ADF) test is used to test for the existence of unit roots and determine the order of integration of the variables. The ADF test requires the equation as follows: t i t m i i t t y y t y ? ? ? ? ? + ? + + + = ? ? = ? ? 1 1 2 1 (2) Volume XIV Issue VI Version I # = ? ? ) against the alternative ? <0 then t y contains a unit root. The test we do both with and without a time trend. SIC method is used to choose the optimal lag length. It can be seen in Table 1 that presence of a unit root which indicates nonstationarity, cannot be rejected in level form. But in difference form the non stationarity problem is vanished. ii. Phillips-Perron (P.P) Test Phillips-Perron (1988) test deals with serial correlation and heteroscedasticity. An important assumption of the DF test is that the error term t u is independently and identically distributed. The ADF test adjusts the DF test to take care of possible serial correlation in the error terms by adding the lagged difference terms of the regressand. Phillips and Perron use non parametric statistical methods to take care of serial correlation in the terms with adding lagged difference terms. Phillips-Perron test detects the presence of a unit root in a series. Suppose, t y is estimating as t t t u y t y + + + = ? ?1 * ? ? ? (3) Where, the P.P test is the t value associated with the estimated co-efficient of ?*. The series is stationary if ?* is negative and significant. The test is performed for all the variables where both the original series and the difference of the series are tested for stationary. # b) Cointegration Testing Methods i. Concept of Cointegration The concept of cointigration was introduced by Granger (1983) and the statistical analysis of cointigrated process was organized by Engle and Granger (1987). Cointegration means that despite being individually non-stationary, a linear combination of two or more time series can be stationary (Gujarati, 2011). When a linear combination of non stationary variables is stationary, the variables are said to be cointegrated and the vector that is quite possible for a linear combination of integrated variables to be stationary. In this case the variables are said to be cointegrated. The key point of cointegration is: 1. cointegration refers to a linear combination of non stationary variables. 2. all the variables must be integrated of the same order. Suppose, considering the following cointegrated regression equation as ) 1 ln( ) ( 1 ? + = ? ? ? = k r i i trace T r ? ? (5) Maximum eigenvalue Statistic: ) 1 ln( ) 1 , ( ) 1 max + ? ? ? = + r T r r ? ? (6) where r is the number of cointegrating vectors under the null hypothesis and ? i ? is the estimated value for the ith ordered eigenvalue from the matrix ?. The trace statistics tests the null hypothesis that the number of cointegrating relations is r against of k cointegration relations, where k is the number of endogenous variables. The maximum eigenvalue test examines the null hypothesis that there are r-cointegrating vectors against an alternative of r+1 cointegrating vectors. To determine the rank of matrix ?, the test values obtained from the two test statistics are compared with the critical value from Mackinnon-Haug-Michelis (1999). For both tests, if the test statistic value is greater than the critical value, the null hypothesis of r cointegrating vectors is rejected in favor of the corresponding alternative hypothesis. More specifically, Table 3 shows that at 5 percent level of significance the likelihood ratios (trace statistics) for the null hypothesis having one (r=1) eigenvalue statisticsfor the null hypothesis having one cointegration (37.93834) is higher than the critical value (22.29962). Hence, according to the likelihood ratio and maximum eigenvalue statistics tests-environmental pollution (i.e. # 2 CO emission), education expenditure and GDP i.e. economic growth are cointegrated. Thus, there is existence of the long run equilibrium relationship among these variables. # c) Error Correction Modeling (ECM) Granger and Engle (1983) analyzed that if the variables are integrated of order one and cointegrated, then there exists the Error Correction Term (ECT) and these variables bears the steady state situation or in equilibrium situation. Considering the following equation which exist each other relationship as: t t n i t n i t t ETC x y y 1 1 1 1 1 3 1 1 2 1 ? ? ? ? ? + ? + ? + ? + = ? ? = ? = ? ? ? (7) t t m i t m i t t ECT y x x 2 1 2 1 1 3 1 1 2 1 ? ? ? ? ? + ? + ? + ? + = ? ? = ? = ? ? ? (8) Where t x and t y denotes the variables, 1 ? t ECT is the error correction term which is the lagged residual series of the cointegrating vector, ' ? 'denotes the first difference, '? 'denotes the error correction term. Here the error correction term capturing the disequilibrium situation. The negative and significant coefficient of error terms suggests that there is a short run adjustment process working behind the long run equilibrium relationship among the variables. Coefficient parameters of error correction term are the speed of adjustment for the short run imbalances. In fact, in the vector error correction model all the variables are endogenously determined within the model. When the variables are cointegrated, there is a systematic and general tendency of the series to return to their equilibrium situation. This means that the dynamics of adjustment is intrinsically embodied in the theory of cointegration. The error correction model states the dependence on both t x and t y of error correction term. If the error correction term is not zero, then the model is out of equilibrium. That is t y lies it's equilibrium values and it starts falling in the next period to correct the equilibrium error. V. # Empirical Results # a) Results of Unit Root Test We first perform unit root tests on all three series in levels and first difference in order to determine the univariate properties of the data in the analysis. To investigate the stationary properties of the variables we run the regression analysis with an intercept term and with intercept term with trend for testing the presence of a unit root. The Augmented Dickey-Fuller test is used to test for the existence of unit roots and determine the order of integration of the variables. The tests are done both with and without a time trend. Results show that the variables 2 CO emission, education expenditure and GDP growth are non stationary in level form because the ADF test statistic of their level form of the variables are less than their respective critical values. This means that they all have the unit root problems and hence they suffer from instability problem in the short run. Results of ADF test of the variables in level and difference form are also given in Table 1. which indicates that the non-stationarity problems vanished after the difference form of the data series, because here the ADF statistic are greater than their critical values and the null hypothesis of non stationarity are rejected. We have also applied Phillips Perron non parametric test for checking the non stationarity of the variables. In the level form, some cases there have the unit root problem in respect of environmental pollution i.e. # 2 CO emissions, education expenditure and GDP growth.But in difference form both with constant and with constant and trend, the statistic value is greater than that of critical value at 1%, 5% and 10% levels of significance. So, the null hypothesis of non-stationarity is rejected, i.e., the data series are stationary at difference form. Results of Phillips Perron test is shown in Table 2. After checking unit root tests, Johansen maximum likelihood procedures are used to test for cointegration and to estimate the error correction parameters to confirm that each series is in I (1) process. Since cointegrating relationship is found among the variables, an Error Correction Model (ECM) is constructed to determine the direction of causality. The significant lagged ECT coefficient indicates that the current outcomes are affect by the past equilibrium errors. If the two variables are cointegrated, there must exists an error correction mechanism. This implies that error correction model is associated with the cointegration test. The long term effects of the variables can be represented by the estimated cointegration vector. The adjusted coefficient of error correction term shows the long term effect and the estimated coefficient of lagged variables shows the short term effect. Causality test among the variables are based on Error Correction Model with first difference. Table 5 shows the Vector Error Correction Model (VECM). # Conclusion In this study we have used carbon emission data as the environmental pollution indicator, GDP as the economic growth indicator and education expenditure in GDP as the education indicator. Time series data for 37 years from 1974 to 2010 was used to analyze causal relationship between environmental pollution, education and economic growth in Bangladesh using VECM based test techniques to establish the short run and long run relationship among the variables in the model. Result shows that there have the long run linear deterministic relationships among the variables. From VECM results it is clear that carbon dioxide ( 2 CO ) emission (EM), and education expenditure (ED) are statistically significant and cointegrated and that is why they effects on each other. It can be said that more education share in GDP intensify the literacy rate and increase in literacy rate create awareness among the people that reduces emission, i.e., environmental pollution. The educational attainments lead to reduce environmental pollution and it also leads to GDP growth, i.e., sustainable development. There has the unidirectional causality between education expenditure and environmental pollution i.e. education and sustainable economic growth. These results will help the environmental authorities to understand the effect of economic growth to the environmental pollution as well as the necessity of environmental awareness through education in Bangladesh. This results postulates that Bangladesh can obtain higher economic growth with better environmental pollution management by creating awareness through education. ![Ed Education Expenditure, GDP= Gross Domestic Production, ? = Error terms.](image-2.png "=") should be effective to reduce or mitigate theenvironment pressures and simultaneously maintaineconomic development. Odhiambo, (2011), examinedthat the unidirectional causal flow from economic growthto2 CO emissions in South Africa without a feedback.The results also show that energy consumptionGranger-causes2 CO emissions and economic growth.Alkhathlan, (2012), found that the positive andsignificant relationship between GDP and22 CO emissionsdeclines at initial level of economic growth then reachesa turning point and increases with the higher level ofeconomic growth. Ahmed et al. (2012) claimed thatthere is a strong positive relationship betweenenvironmental pollution and economic growth. GrangerCasualty Test indicates changes in GDP per capitaGranger-cause Emission. Ru, et al., (2012), analyzedthat the relationship between economic developmentand the factors causing the environmental pressures isthe basic premise of formulating and adjusting theenvironmental policy. A sound environmental policy equilibrium relationship between the series y and x. Theterm t u , indicates the deviation from the long runequilibrium path of t y and t x . A time series data ( t y ) issaid to be integrated of order one and that can bedenoted as I (1). If the original non stationary series hasto be differenced'd' times for stationary process, theoriginal series is integrated of order'd' that can bedenoted by I (d). Consistency in ECM requires all ofterms to be integrated of order zero, I (0). This ispossible only if y and x are cointegrated in a linear form,that isXt=?yt+utwhich is stationary.yt=?+?xt+ut(4)In this series t y and t x are I (1) and the errorterm t u is I (0). Then the coefficient measures the 1VariableStatisticsCritical ValuesStatisticsCr itical ValuesWith1%5%10%Withtrend1%5%10%interceptand interceptLevel Form2 CO emission3.138912(2) -3.626784 -2.945842* -2.611531*-0.989891(2) -4.234972-3.540328 -3.202445Education expenditure2.928958(2) -3.632900 -2.948404 -2.612874*2.638842(2) -4.243644-3.544284 -3.204699in GDPGDP1.750733(2)-3.626784 -2.945842 -2.611531-1.099151(2) -4.234972-3.540328 -3.202445Difference Form2 CO emission-4.681470(2) -3.632900* -2.948404* -2.612874* -6.739015(2) -4.262735* -3.552973* -3.209642*Education expenditure2.631175(2) -3.632900-2.948404 -2.612874*-6.022913(2) -4.252879* -3.548490* -3.207094*in GDPGDP-5.201792(2) -3.632900* -2.948404* -2.612874* -6.033745(2) -4.243644* -3.544284* -3.204699*Note:b) Phillips-Perron Test 2VariableStatisticsCritical ValuesStatisticsCritical ValuesWith1%5%10%Withtrend1%5%10%interceptand interceptLevel Form2 CO emission 6.132033(2) -3.626784* 2.611531*-2.945842*--0.655198-4.234972 3.202445-3.540328-Education13.38117(2) -3.626784*-2.945842*-12.53238-4.234972*-3.540328*-expenditure2.611531*3.202445*GDP1.839900(2) -3.626784-2.945842--1.054326-4.234972-3.540328-Difference Form2.6115313.2024452 CO emission -4.818463(2) -3.632900* 2.612874*-2.948404*--9.851673(2)-4.243644* 3.204699*-3.544284*-Education3.873066(2) -3.632900*-2.948404*---4.252879*-3.548490*-expenditure2.612874*6.022913(2)3.207094*in GDPGDP-5.209039(2) -3.632900*-2.948404*---4.243644*-3.544284*-2.612874*6.056790(2)3.204699*Note: The test is conducted using Eviews 7.12 32 CO Emission i.e. EnvironmentalPollution and Total GDP.H0H1Trace5% CriticalMax. Eigen5% criticalHypothesisStatisticsvaluevaluevalueH 0 : r=0H 1: r=157.3389535.1927537.9383422.29962None**H 1: r=1H 1: r=219.4006120.2618411.9922715.89210Note : The test is conducted using Eviews 7.1d) Results of Error Correction Model (ECM) 4GDP 2 © 2014 Global Journals Inc. (US) CO Emission: A case study for Bangladesh until * Bangladesh Development Research Working Paper Series * environmental impacts of a North American Free Trade Agreement GGrossman AKrueger National Bureau of Economic Research Working Paper 3914 1991 * Basic Econometrics, Eighth reprint DNGujarati 2011 McGraw -Hill Publications New York * MRIslam YCheng SURajib 2012 International Trade and Carbon Emissions ( 2 CO * Economic growth and income inequality TheCase Bangladesh Journal of Economics and Sustainable Development 3 1955 American Economic Review * The carbon productivity challenge: Curbing climate change and sustaining economic growth McKinsey Global Company 7 2008 The McKinsey Quarterly * Economic growth and carbon emission in South Africa: An empirical investigation NMOdhiambo International Business and Economics Research Journal 7 10 2011 * Testing for Unit root in Time Series Regression Phillips-Perron 10.1093/biomet/75.2.335 Biometrica 75 1988 * References Références Referencias * Emperical Study on Relationship between Environmental Pollution and Economic Growth of Maldives Using Environmental Kuznets Curve and OLS Method AAhmed DBHerve LZhao International Journal of Business and Management 7 2012 * An Empirical Analysis of the Environmental Kuznets Curve for 2 CO Emissions in Indonesia: The Role of Energy Consumption and Foreign Trade KAlkhathlan MQAlam MJavid BSaboori JBSulaiman SMohd International Journal of Economics and Finance 16 2012. 2012 British Journal of Economics * Environmental Education and Environmental Management in Bangladesh and their Sustainability MSalequzzaman JKDavis Environmental Informatics Archives 1 2003 * SMOng SKSek Interactions between Economic Growth and Environmental Quality: Panel and Non-Panel Analysis 2012. 2013 7