Asian Journal of Empirical Research

Transcription

1 06 Asian Economic and Social Societ. All rights reserved ISSN (P): 06-98X, ISSN (E): Volume 6, Issue 5 pp. -4 Asian Journal of Empirical Research DOI: /journal.007/06.6.5/ DETERMINANTS OF ENVIRONMENTAL DEGRADATION AND EMPIRICAL INVESTIGATION OF KUZNETS CURVE: A COMPARATIVE STUDY OF INDIA AND BANGLADESH Humaira Husain Senior Lecturer; Department of Economics, North South School of Business and Economics, North South Univers, Dhaka, Bangladesh Article Histor: Received: 9 Jul 06 Revised received: Aug 06 Accepted: 0 Aug 06 Online available: Sep 06 Kewords: Environment, emission, India, Bangladesh Abstract This paper investigates the factors responsible for Environmental degradation in terms of higher Carbon dioxide and Nrous oxide emission over the period 98 0 for Bangladesh and India from data of World development indicator. Industrialization and greenhouse gas are major contributing factors identified for deteriorating the air qual in Bangladesh and India respectivel. Concentration of population in urban areas increases carbon emission and economic globalization reduces Nrous oxide emission in both economies. Estimation results of reduced from equation for CO emission provides support in favour of inverted U and N shaped Kuznets curve for India which is robust to using dnamic specification. For Bangladesh the shape is Inverted U but this is not robust to using dnamic model. In quadratic approach to Kuznets curve energ consumption does pla an important role for increased carbon emission in India but not in Bangladesh.. INTRODUCTION Bangladesh and India are two emerging economies in South Asia. These two neighboring countries have experienced considerable economic growth in recent ears. This paper tries to explore the nature of trade- off between macroeconomic gain in terms of High Economic growth and level of environmental pollution. In inverted U EKC trajectories, a ver ltle is known in existing lerature about the turning points of real GDP per capa after which emission of CO starts falling for both economies. This paper also aims to appl the polnomial approach to EKC analsis for econom of Bangladesh. We consider two major pollutants Carbon dioxide and Nrous Oxide. This article primaril focuses on the similaries and dissimilaries between the factors of these two developing countries responsible for higher carbon and nrous oxide emission. Section briefl discusses about review of Lerature. Section introduces the baseline specification and Section 4 highlights the estimation results and calculates the turning points. Section 5 identifies the contributing factors for each pollutant CO and NO for both countries. Section 6 illustrates the Robustness test of baseline models Section 7 suggests polic recommendations Section 8 summarizes the findings of the paper and concludes. Corresponding author's address: humaira.husain@northsouth.edu

2 . REVIEW OF LITERATURE.. Environmental Kuznets curve in Theor Economic growth influences environmental qual through three channels, Scale effect, technological effects and composion effect (Grossman & Krueger, 99). In the process of production, emission of bproducts deteriorates the environmental qual known as scale effect in the first phase of structural change. As econom expands according to second phase of structural change from energ intensive industries to service to knowledge based technolog intensive industr, so the obsolete and outdated negative external generating production techniques are replaced as spending on R & D rises, known as technological effect of growth. This posive impact dominates the first phase of scale effect eventuall the EKC bends downward... Empirical Findings for developing economies A large volume of lerature is available for the empirical test of Environmental Kuznets Hpothesis, which is known as Environmental Kuznets Curve (EKC) (Kuznets, 955). This hpothesis assumes that wh expanding industrial sector, iniall the economic growth of an emerging econom is realized at the cost of deterioration of the environmental qual as the econom is growing attains a threshold level of Income beond which there is no trade-off between the two, so the EKC is inverted U shaped. The comprehensive and detailed examination of EKC lerature is illustrated b Dinda (004) and Stern (004). In the context of developing economies most studies have focused on Asian, Latin American, African countries, as well as countries in Middle East. Results of empirical test of this hpothesis var across countries. Naraan and Naraan (00) tested this hpothesis for 4 developing nations from 980 to 004. The found for Middle Eastern panel, the countries are successful in reducing emission wh higher Income per capa, as the long- run income elastic is smaller than short run elastic. Jaunk (0) tested on high Income countries including MENA region over the period and found emission reduces wh larger level of Income in long run. Arouri et al. (0) tested EKC on Middle East and North African countries and found support in favour of inverted U. But the found considerabl large variation in turning points and got weak evidence in support of EKC. Comparative analsis of EKC b Jaanthakumaran et al. (0) on India and China suggest CO emissions are influenced b energ consumption, foreign trade and Income in China and per capa GDP causes the carbon emission in India. Song et al. (007) found Inverted U EKC for China for for three pollutants, waste gas, waste water and solid gas. Miah et al. (0) found conventional EKC trajector for waste emission and suspended particulate matter for Bangladesh wh high turning point. According to test of EKC for Bangladesh b Islam and Shahbaz (0), over the period 97-00, energ consumption is a major contributing factor for degradation of air qual in terms of increased Carbon emission and the found Pollution is negativel related to trade openness and posivel related to increased urbanization. Mulali et al. (05) found posive relationship between pollution and Income both in short run and long run in Vietnam over the period 98-0, so EKC does not exist in their sample. Saboori et al. (0) found Inverted U between GDP per capa and CO emission for Malasia both in short run and in long run and found unidirectional causal from economic growth to CO emission.. BASELINE SPECIFICATION Widel used quadratic approximation of EKC curve is the following: Dinda (004) Equation in level: CO 0E Equation in Reduced form: ln ln CO.. () 0 ln E ln

3 Reduced form equation is obtained when the variables are expressed in natural log. Polnomial approximation of EKC is generated b adding cubic term in model () (Song et al., 007). ln CO 0 ln E ln ln ln.. () CO = carbon dioxide emission per capa (in metric ton), extensivel used prox for environmental degradation in lerature. E and represents energ consumption and Real GDP (per capa at constant 005 USD). Random error term In model () if β > 0 and β < 0, EKC is Inverted U shaped. In model () if β > 0 and β < 0, and β >0, EKC is N shaped. There are other possibilies of the shape of EKC depending on different signs of the parameters. i for countr and t for time (in ear) Pollutant Inverted U Pollutant N shaped EKC EKC Per capa Income Per capa Income 4. ESTIMATION RESULTS Table : Regression results of Quadratic Model (98 0) Estimation method (OLS) Dependent Variable ln CO Bangladesh India Regressor Estimated coefficients Ln E 0.54 (.0).88 *** (0.) Ln Y 9.9 *** (5.65) 5.0 *** (0.78) ( Ln Y) -.5 *** (0.4) *** (.06) R squared F Statistic No of Observation Estimation results have been obtained b appling OLS. Standard errors in parenthesis are robust standard errors. *** Significant at % level, ** significant at 5 % level Table : Regression results of Cubic Model (98-0) estimation methodolog SLS Dependent Variable ln CO Bangladesh India Regressor Estimated coefficients Ln E -0.94(.8) 0.67(0.55) Ln Y *** (94.5) *** (5.00) (Ln Y) -8 ** (64.86) -5.6 *** (5.49) (Ln Y) 6.98 ** (.55) 0.77 *** (0.8) R-squared Adjusted R squared F-Statistic No of Observation Estimation results have been obtained b appling SLS, in parenthesis are standard errors. ***significant at % level, **significant at 5 % level

4 4.. Analsis of findings In Quadratic model the EKC is inverted U shaped for Bangladesh. The elastic of CO emission wh respect to per capa GDP growth is que larger and posive. At the declining portion of the curve after reaching the threshold level of Income, emission reduces onl b.5% for each % increase in per capa Income. Energ consumption has no impact on environmental degradation. Quadratic model also explains the inverted U EKC for India. Emission rises b 5.0% for each % increase in per capa Income, and reduces b 0.4% after reaching threshold level of Income along the declining portion of the curve. Energ consumption does have significant impact on emission. We observe larger difference between emission (rising and falling) for Bangladesh compare to India. Appling instrumental variable method because of possible endogene problem, Polnomial model explains the N shaped EKC for both economies. The curve has two turning points. Emission rises iniall wh increase in per capa income, and then reaches a threshold level of income where the degradation of air qual falls wh rise in Income. EKC attains a stationar level of GDP beond which again the deterioration of air qual starts and the curve bends upward. In cubic model the Energ consumption has no role for deteriorating the air qual. The coefficients of the regressors are mostl not precisel determined in case of Bangladesh. So our sample data adheres to inverted U shaped EKC for Bangladesh and Inverted U and N shaped EKC for India. 4.. Finding the turning points 4... Turning point for Quadratic model In equation (), sample regression function is CO f (, E) We are assuming the inverted U shaped EKC is smooth and continuous, above equation is multivariable so setting the partial derivative of dependent variable wh respect to Per capa real GDP equals to zero, we get the following equation: ln CO 0 ln E ln ln. ln.ln. ln ln. 0. ln.ln ln For Bangladesh we substute the estimated coefficients in above equation and get following: ln ln 6. 5 e For India, 5.0 ln The threshold level of per capa real GDP beond which the Kuznets curve bends downward is for Bangladesh and 5.78 for India. 4

5 nd Derivative test To be confirming we perform second derivative test.at the turning point the curve bends downward so s sign should be negative evaluated at the stationar value of GDP per capa. (...ln. ).. (ln ) ( ).[ ln ( ). (ln )., We substute the estimated values of, and the stationar value of in above expression for Bangladesh and India. We found that the sign of above expression is less than zero in both cases Graph of Inverted U Environmental Kuznets Curve: (98-0) Bangladesh Turning point CO Emission Per capa CO Emission per capa India Turning point Real GDP Real GDP per capa 5.78 per capa We observe that the threshold level of Income (measured in 005 USD) at which pollution declines, is larger in Bangladesh compare to India. 4.. Turning point for Polnomial model We set the partial derivative of dependent variable wh respect to GDP per capa equals to zero for sample regression function of model () and get following equation. ln CO 0 ln E ln (ln ) (ln 0 ) = 0 ( ) ( 0 ln E) ( ln ) ( (ln ) ) ( (ln ) ) 0 5

6 ..(ln )..(ln ). 0 0 Multipling both sides b..(ln )..(ln ). 0 This is a quadratic equation in ln, we appl the square root formula ln 4.., substuting the values of estimated coefficients in above.. expression we get two turning points for India. For Bangladesh the coefficients are not precisel determined at % level so we calculate the threshold level of per capa GDP onl for India. ( 5.6).(.(5.6)) (0.77) ln = 7.04, (0.77) ln 7.0 = 7.0 e 4 and 6.0 e, nd derivative test The expression for second order derivative is following: ( )..(ln )..(ln ). {ln. ) {(ln ) ( ). ln { (ln ) ) ).. ln ( ln (ln ). ln We substute the estimated values of, and }.ln and each stationar value of. } in above and find the expression to be At the first turning point the sign of the second order derivative is found to be < 0. Which is (47.0) = negative. At the second turning point 4 the sign of second order derivative is found to be > 0, which is posive (4) (4) So the second derivative test confirms that at 4, the EKC bends upward again and NO Emission Starts rising. We observe that for India for each % increase in per capa GDP the emission rises b less than % after the second turning point. 6

7 CO Emission Per Capa st Turning Point nd Turning point Per capa GDP (005USD) Graph of N shaped Environmental Kuznets Curve for India: (98-0) 5. DETERMINANTS OF ENVIRONMENTAL DEGRADATION Our article primaril focuses on the determinants of environmental degradation. We focus on the economic variables as our regressors. We are considering two major pollutants CO and NO. Proposed model () for Bangladesh and India is the following: lnco = α + β ln(indusval) +β 4 ln(urban population) + β 5 ln(trade) + β 6 ln(green) + µ.. () In the above model Indusval = industr value added (percent of GDP) Urban population = percent of urban population (of total population) Trade = Trade (Export plus import) as percent of GDP (Trade is the prox for Economic globalization) Green = Total greenhouse gas emission (kt of co equivalent) per capa µ = Random Error Term Table : Regression results of model () estimation methods (OLS) Dependent variable ln CO Bangladesh India Regressor Estimated Coefficients ln(indusval).0 *** (0.8) -0.9(0.) ln (urban population).4 *** (0.8).5 *** (0.50) ln(trade) 0.0(0.) -0.(0.08) ln(green) 0.8(0.6).0 *** (0.) R squared F Statistic No of Observation Estimation results have been obtained b appling OLS, in parenthesis are robust standard errors. ***significant at % level, **significant at 5 % level 5.. Similaries and dissimilaries of determinants of emission between two countries Emission rises b.4% and.5% for each % rise in urban population for Bangladesh and India respectivel though elastic of CO wh respect to urban population is larger in India. Economic globalization has no impact on the deterioration of the air qual in both economies. 7

8 Industrialization plas a significant role for the degradation of the air qual in Bangladesh. Green house emission per capa plas a major contributing factor for emission in India. For each % rise in industr value added, emission rises b.0% in Bangladesh. In India for each % rise in greenhouse gas emission per capa, CO emission rises b.%. 5.. Polnomial approximation of EKC and determinants of NO lnno = α + β ln (energ) + β (lny i ) + β (lny i ) + β (ln Y i ) + µ. (a) Using SLS method we get the following estimation results : Table 4: Regression results of Polnomial model Estimation Methodolog SLS Bangladesh India Dependent Variable lnno Estimated Coefficients Regressor ln(energ) 0.8 (0.5).05 (0.60) lny i *** (0.04) *** (8.9) (lny i ) -.84 *** (6.78) -.6 *** (6.0) (lny i ) 6.68 *** (0.9) 0.6 *** (0.) R Squared F Statistic No. of Observation Estimations have been obtained b appling OLS, the standard errors are robust standard errors. *** Significant at %, ** Significant at % High R square in estimation results of Model,, and 4 might generate the question of spurious regression because of the presence of the common time trend in all regressions. Longer time series data is required to check whether the EKC results are supported in long run. Our article primaril focuses on identifing the factors for environmental degradation rather than the shape of EKC in both economies. For both economies the EKC is N shaped. NO Emission iniall rises but s magnude is que larger for Bangladesh compare to India. Then declines wh increase in income, after reaching a threshold level of Income the emission again starts increasing. The coefficients signs are precisel determined for both countries. After the second turning point for Bangladesh the NO emission raises b 6.68% for each % rise in real GDP per capa, the corresponding number is much lower in India. To find out the determining factors of Nrous Oxide Emission we replace the dependent variable lnco b lnno. We estimate the following model. lnno =α + β ln(indusval)+ β 4 ln(urban Population)+β 5 ln(trade)+β 6 ln (green)+µ.... (a) Table 5: Regression results of Model (a) Dependent Variable lnno Bangladesh India Regressors Estimated Coefficients ln(indusval). *** (.4) -0.5(.8) ln (urban population).05 (.8).5(.47) ln(trade) -.6 *** (.) -. *** (.07) ln(green) -. (.). *** (.5) F Statistic R.5.66 No of Observation Estimations have been obtained b appling OLS. The standard errors are robust standard errors in parenthesis. *** Significant at % level, ** Significant at 5 % level 8

9 Industrialization and greenhouse gas pla the major role for NO emission in Bangladesh and India respectivel. The emission reduces wh Economic globalization in both economies. The model is more suable to explain the NO emission in India compare to Bangladesh because of relativel smaller R square for Bangladesh compare to India. Elastic of emission wh respect to greenhouse gas in India and wh respect to We also estimated the quadratic model for both economies but did not get the predicted signs of the coefficients to trace the Inverted U shape EKC Industrialization in Bangladesh is que similar. For each % rise in greenhouse gas per capa emission rises b.% in India. In Bangladesh for each % rise in value added of the manufacturing sector emission increases b.%. Urban population has no role to pla for the deterioration of the air qual in both countries. 6. ROBUSTNESS CHECK OF EKC To examine whether our estimation results of Baseline models hold using alternative specification, we introduce the first lag of the dependent variable as addional regressor in model () and in Model (), and the models become following: lnco = α + βlnco, I (t-) + β 0 lne + β (lny) + β (lny ) + µ lnco = α + β 0 ln CO, i(t-) + β ln E + β (lny ) + β (ln Y ) + β 4 (lny ) + µ.... (b). (b) Appling OLS in equation (b), for Bangladesh the coefficient β is posive and precisel determined. But the signs of the other parameters are not predicted. So Inverted U is not supported b dnamic specification. Appling SLS for equation (b), the parameters have their predicted signs and are precisel determined for India, but β and β0 are not precisel determined in polnomial model. In the quadratic model all the parameters have predicted signs except coefficient of first lag of dependent variable for India. Inverted U and N both shapes are supported using Dnamic specification for India, but not for Bangladesh. For NO pollutant we have got predicted signs of parameters for polnomial model except for β for Bangladesh appling instrumental variable method. N shape EKC is supported for Bangladesh using dnamic specification. But none of the estimated coefficients are statisticall significant for cubic model in case of India when first lag of dependent variable is introduced in the polnomial model. 7. ENVIRONMENT PROTECTION POLICY The turning point or the level of Income beond which emission starts declining is lower in India than Bangladesh. According to quadratic approach the emission starts declining at fairl high level of per capa GDP in Bangladesh. There is an urgent need to replace the outdated, obsolete negative external generating production method in manufacturing sector wh cleaner, improved environment friendl technolog to ensure sustainable development in Bangladesh. This might be consistent wh displacement hpothesis. The developing countries are likel to be net exporter of pollution intensive goods to Developed countries wh stronger environmental regulations. In India energ consumption plas a role in degrading air qual, It should replace the existing energ base from fossil fuel to other renewable energ sources which are more environment friendl. This will also reduce the greenhouse gas emission which plas a significant role in lowering air qual. Since India has entered the second phase of structural change should spend more on R&D to explore appropriate, cost effective source of energ consumption. Rapid urbanization deteriorates 9

10 environmental qual in both countries; appropriate policies should be implemented to reduce the traffic volumes in urban cies and unplanned internal migration from rural to urban area. In our stud we find that trade declines the level of NO emission in both countries, because the elastic of NO emission wh respect to trade is negative and less than un. This might be due to technolog transfer through foreign direct investment through composion effect, which reduces the emission level. An important polic implication is that both countries should continue to be integrated in Economic globalization where developing economies improve their environmental qual as investment increases Income and emploment. 8. CONCLUSION Our sample data fs wh quadratic approach to Kuznets curve for Bangladesh Econom, though the result is not robust using dnamic specification. Energ Consumption has got no role to pla for Increased Carbon emission. This finding contradicts wh Large sample findings of Islam and Shabaz (0) on Bangladesh as well as Findings b Alam et al. (0) where the found uni- directional causal from energ consumption to CO emission in short run but feedback causal exists in long run and the found Carbon emission causes Economic growth in both short run and long run. In India the experience of EKC is both supported b cubic and quadratic approach which is robust and Energ consumption does pla role in polluting the econom using quadratic approach. In both economies the NO emission is supported b Polnomial model and Kuznets curve is N shaped. After the second tuning point when the emission rises, for India the elastic of emission wh respect to per capa GDP is less than Un, whereas in Bangladesh is greater than Un. So India is successful in reducing s emission level in recent ears compare to Bangladesh. The estimation results on NO pollutant qualif the robustness test for Bangladesh but not for India. We get stronger empirical evidence in support of N shaped EKC for Nrous oxide pollutant in Bangladesh. For carbon emission Inverted U and N shaped EKC for India qualif the robustness test. But we get stronger evidence in favour of Inverted U rather than N shape because Greenhouse gas is a major polluting factor for Carbon emission which is the result of increased use of fossil fuel energ consumption. In N shaped EKC Energ consumption is not precisel determined which is misleading. Future Research could extend on identifing contributing factors for other pollutants like SO, CO etc. The measurement of environmental degradation could be manifested b pollution intens. A comprehensive and detailed stud of other forms of pollution like water pollution etc. could be examined. From our stud we conclude that the shape of Kuznets curve varies wh choice of pollutants. We observe that a particular pollutant ma exhib different shapes of EKC and both shapes ma be robust to alternative specification. Longer time series data will help future researchers to conclude the relationship between Income and environmental degradation using Environmental Kuznets curve and explore the role of energ consumption in this framework. This would definel help design appropriate polic directions and recommendations to achieve sustainable stable economic development for developing countries for next generation. Funding: This stud did not receive an specific financial support. Competing Interests: The authors declare that the have no conflict of interests. Acknowledgement: I sincerel thank Mr. Khairul Alom, Senior lecturer, School of Business, Southeast Univers for his suggestions and comments to improve the qual of this paper. Views and opinions expressed in this stud are the views and opinions of the authors, Asian Journal of Empirical Research shall not be responsible or answerable for an loss, damage or liabil etc. caused in relation to/arising out of the use of the content. 40

11 References Alam, J. A., Begum, A, I., Busee, J., & Hulenbroeck, V. G. (0). Energ consumption, carbon emissions and economic growth nexus in Bangladesh: Co-inegration and dnamic causal analsis. Energ Polic, 45, 7-5. Arouri, M., Youssef, B., Adel, M henni, H., & Rault, C. (0). Energ Consumption, Economic growth and CO emission in Middle East and North African Countries. IZA Discussion paper no. 64. Dinda, S. (004). Environmental Kuznet curve Hpothesis: A surve. Ecological Economics, 49, Grossman, G., Krueger, M., & Alan, B. (99). Environmental Impact of a North American free trade agreement. National Bureau of Economic Research, working paper, 94. Islam, F., & Shahbaz, M. (0). Is there an Kuznets curve for Bnagaladesh?. Munich Personal RePec Archive, Paper no: Jaunk, V. C. (0). The CO emissions income nexus: evidence from rich countries. Energ Polic, 9(), Jaanthakumaran, K., Verma, R., & Liu, Y. (0). CO emissions, energ consumption, trade and Income: A comparative analsis of China and India. Energ Polic, 4, Kuznets, S. (955). Economic growth and income inequal. American Economic Review, 45(), - 8. Miah, M. D., Masum, M., Koike, M., Akter, S., & Muhammad, N. (0). Environmental Kuznets curve: The case of Bangladesh for waste emission and suspended particulate matter. The Environmentalist, (), Mulali, U., Saboori, B., & Ozturk, I. (05). Investigating the environmental Kuznet curve for Vietnam. Energ Polic, 76, -. Naraan, P. K., & Naraan, S. (00). Carbon dioxide emissions and Economic growth: panel data evidence from developing countries. Energ Polic, 8(), Stern, I. D. (004). The rise and fall of the environmental Kuznet Curve. World Development, 4(7), Song, T., Zheng, T., & Tong, L., (007). An empirical test of the environmental Kuznets curve in China: A panel co- integration approach. China Economic Review, 9, 8-9. Saboori, B., Sulaiman, J., & Mohd, S. (0). Economic growth and CO emissions in Malasia: A co-integration analsis of Environmental Kuznets curve. Energ Polic, 5,