Carbon Dioxide Emissions - Essay Example

Executive Summary The purpose of this report is to assess whether a country's carbon dioxide CO2 emissions depended on the economic development levelof the country. The Environment Kuznets curve hypothesis was to be tested in principle. The influence of the dummy variables was also tested. A general linear regression model was postulated and evaluated. Introduction Carbon dioxide emissions are those 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. The average World Carbon Dioxide Emissions were 4.53 metric tones/capita in 2005. United States of America ranked first then with average emissions of 19.54 metric tones/capita followed by Canada, Russia and UK. Developing countries like India and China lagged at the tail end then. The Fourth Assessment Report of Intergovernmental Panel on Climate Change concluded among other things that "warming of the climate system is unequivocal" and that "anthropogenic warming over the last three decades has likely had discernible influence at the global scale on observed changes in many physical and biological systems." The report also stated that human activities have significant impact on climate change. In 2005, total global emissions were about 36 trillion metric tones, of which 28.5 trillion metric tones alone was accounted for by fossil fuel burning. Exhibit 1: World map of CO2 emissions in 2000 by country, excluding land use change (million metric tonnes of carbon) Source: Interlinked Challenges http://colli239.fts.educ.msu.edu/ The above figure indicates that less developed countries (Eg.Nigeria) have lower emissions as compared to developed countries (USA) and emerging countries (China and India). Global carbon dioxide emissions by world region (1970-2005) in Pg CO2 ( 1 Pg (petagram) =1 trillion tonnes). Since 1970, global emissions have been growing by more than half a trillion metric tones per year. The fastest growth rates of CO2 emissions were seen in developing countries. The Gross Domestic Product is one of the primary indicators calculated to assess the health of a country's economy. It representsthe total dollar value of all goods and services produced over a specific time period. According to the Environment Kuznets hypothesis, environmental pressure increases faster than the income in the early stages of development and slows down relative to GDP growth as higher income levels are reached. The EKC proposes that indicators of environmental degradation first rise, and then fall with increasing income per capita. Environmental Kuznets Curve : Different Scenarios Source: Reference 1 There are views supporting and criticizing this theory. Those views supporting it emphasize that the curve exists though its becoming smaller in nature and shifting to the left(revised EKC) whereas those criticizing it argue that even if certain pollutants are reduced as income increases, industrial society continuously creates new, unregulated and potentially toxic pollutants. In their the curve will rise to a horizontal line at maximum existing pollution levels, as globalization promotes a "race to the bottom" in environmental standards, as shown in Figure In their view, the overall environmental risks from these new pollutants may continue to grow even if some sources of pollution are reduced, as shown by the "new toxics" line in the above figure. The relationship between economic growth and environmental quality has been a source of great controversy for a very long time. Multiple factors contribute to this. The complex nature of GDP calculation, data on environmental quality are patchy themselves and also that though the per capita capability to pay for clean technology increases, it does not necessarily imply a proportionate increase in the willingness to pay. This report makes an attempt to verify the Environmental Kuznets hypothesis while establishing a relation between additional variables like the energy consumption of the country, whether it is an oil producing country or not, its compliance with Kyoto protocol and so on and so forth. It is believed that by increasing the number of variables the data analysis will improve. The data for this study were drawn from (Please mention source). In the following part a model is formulated for assessing the relationship between Carbon Dioxide emissions and the other variables. Influence of each variable on Carbon Dioxide emissions is calculated. Significance of dummy variables will be evaluated and the fit of final model will be assessed. Data Characteristics The dataset consisted of 183 countries covering all economic strata. From a preliminary examination of the data, it became evidently clear that United states had a lions share in the contribution to the CO2 emissions. The per capita GDP ranged from a lowest of $273.4 (Congo) to a highest of $68814 (Luxembourg). The most populous country in the data was China with 1304362.7 thousand people while the most scarcely populated country was Bermuda with a relatively meager 63.93 thousand people. The CO2 emissions ranged from 0.04million tones to a highest of 5994.29 million tones by America which skewed the average emissions over the 184 countries to around 5666 tones. A couple of dummy variables are present in the dataset. An exact variable to indicate economic development cannot be specified. Hence Regional Dummy Variable (RED) depending on the region of the country, dummy variable whether or not the country is producing oil (attributed with 1 and otherwise 0) and whether or not a country complies with Kyoto Protocol are included which are to serve as good proxy substitutes. The RED was decoded with the values as given in the appendix. The Correlation co-efficient between capita GDP and per capita CO2 emissions is 0.7218 which is suggests a relatively strong positive correlation between the two variables. While comparing CO2 emissions with the Energy Consumption and GDP following trends were observed: Thus a similar trend is observed in both the graphs, the curve increases until a certain point and then flattens out stabilizing at a particular level. This is in concordance with the Kuznet Hypothesis. A more of a "race to bottom scenario" described above is observed in these graphs from the actual data. A declining trend is not observed as these are values of a single year and not cumulative GDP values of continuous years According to the Correlation analysis in Appendix 5A, the Population of the country, amount of energy consumed, GDP and also whether or not the country produced oil or not had a significant positive correlation as expected with the Carbon Dioxide Emissions. Now, Partial correlation is a method used to describe the relationship between two variables whilst taking away the effects of another variable, or several other variables, on this relationship. When partial correlation co-efficient was calculated a different and surprising trend was observed. By partial correlation (Appendix 5B) it was deduced that partial correlation co-efficient of CE with CO2 was 0.693 as compared to original 0.953. GDP on the other hand stooped down from an original 0.908 to a negative -0.10254 suggesting that GDP had approximately no correlation with CO2 emissions independently when other variables were adjusted for. The compliance with Kyoto Protocol did not yield a significant negative correlation with Carbon Dioxide emissions as predicted. The influence of TR could not be assessed as some values were missing in the data set and the impute missing variables was not available in SPSS. Model Selection and Interpretation: In this part we intend to model the relationship between GDP (y) and CO2 emission (E). The basic hypothesis is the assumption that the relationship between CO2 emission and GDP has an inverted-U shape as suggested by the environmental Kuznets Curve. Assuming the relationship stands, the Equation 1 should therefore explain the model effectively. In order to prove the inverse U-shape of the curve, in our equation we need to include the quadratic term shown in Equation 1 by Equation 1. E= + y + y2 + Considering the dummy variables the above equation assumes the following form: Equation 2: E= + y + D + Dy + Dy2 + Where E: is the dependant variable (CO2 emissions) ,,,: constant co-efficients determined from regression analysis y: Independent Variable D: Dummy Variable Finally the following model is recommended Log (CO2 emission) = 0 + 1 logGDP + 2 logCE + 3 logPOP + 4 OIL + 5KP + 6RED (1) H0: 0 is negative Here Logarithms of variables namely CO2 emission, GDP, Energy consumed (CE), Total Population (POP) are taken so that all the data are on a measurable scale. Dummy variables OIL and KP are used as it is. Under this model 's (betas) represent unknown parameters that can be estimated using the data. The estimated version of the model in the above equation is: CO2 emissions= -9.166e+00 + (-3.77e-04)POP + (5.99e-03)CE + (-2.671e-10)GDP + (-1.2588)RED (2) Thus 0 is negative and our hypothesis is proved true that there is inverted U relationship between Carbon dioxide emission and GDP The t test value for dummy variables was Variable t-value Result at 95% level of significance RED 0.197 Insignificant OIL -0.709 InSignificant KP -1.319 Insignificant The model provided quite a good fit to the available data. Equation 2 is also useful for analyzing the relationship between the variables. The t -coefficients associated with all the variables except the dummy variables exceeded 5 in absolute value. Also the value of R2 = 0.989 and adjusted R2= 0.988 indicates that the model explained 98.8% of the variables effect on Carbon Dioxide emissions. The beta coefficient tells you how strongly is the independent variable associated with the dependent variable. It is equal to the correlation coefficient between the 2 variables. Thus from the table in the appendix, beta value for energy consumption is the strongest at 0.7. The following display summarizes the regression fit: CO2 emissions= -9.166e+00 + (-3.77e-04)POP + (5.99e-03)CE + (-2.671e-10)GDP + (-1.2588)RED Std.errors 1.306e+01 6.999e-05 1.790e-04 2.238e-11 1.995e+01 t-ratios -0.702 -5.397 33.463 -11.938 0.197 In determining the final model the first decision made was to use logarithmic transformation of several variables. This enabled in making the data more symmetric. The histogram for the logarithmic distribution appended below is highly significant while its non logarithmic counterpart (not included) did not show such a characteristic. A regression model was constructed with and without using Oil as a contributing factor, however R2 value remained same for both models and the t value was insignificant in the model containing OIL factor. Thus it can be concluded that a separate regression model need not be built separately for Oil producing countries. Alternative models have only considered GDP or energy consumption singularly with Carbon Dioxide emission. This model aimed to bringing all three of them along with the population on one platform. Also the inclusion of the Dummy variables made the analysis more robust. Summary and Concluding Remarks The Carbon Dioxide emission data and predicted values for next years are not completely irrefutable. New theories and models are emerging every day to prove their own efficiency over others. However calculating the major determinants of Carbon Dioxide emissions and their impact factor accurately will enable to make a sound estimate of CO2 emissions. This model makes an attempt at analyzing the same. This study was based on a cross section of 182 countries which is a reasonably good sample size. We were able to analyze several characters in this model. In future, we could expand the studies and include STIRPAT analysis provided an in detail structural human ecology analysis. It also considers the affluence of the population in qualitative terms other than GDP. The Ecological intensity of each nation could be included in the analysis. Ecological intensity is a measure of how much impact a nation has net of other drivers. Also data should be analyzed and compared over a specified period of time. The concerned dataset used here comprised only of the year 2005. I believe, Cumulative emissions compared over a period of years will give an idea of the actual underlying trend. Also the dataset can be segregated into various classes depending on say for example population density, economic prosperity comparison of emissions should be made within these groups. Appendix APPENDIX TABLE OF CONTENTS 1. References 2. Variable Definition 3. Basic Summary Statistics 4. Confirmation by tests for Normality 5. Correlation Table 6. Final Fitted Regression Model: SPSS Output A1. References Dasgupta et al (2002) Confronting the Environmental Kuznets Curve. Journal of Economic Perspectives. 16(1)147-168 Nemat Shafik (1994) Economic Development and Environmental Quality: An Econometric analysis. Oxford Economic Papers. New Series, Special Issue on Environmental Economics. pp. 757-46, 757-773 Osmolska and Homorodi (2009) An Analysis of the Relationship between Carbon-Dioxide Emissions and Gross Domestic Product for 139 Countries within the Time Period 1985-2004. Maters Project University of Skovde Stern (2004) The Rise and Fall of the Environmental Kuznets Curve. World DevelopmentVol. 32, No. 8, pp. 1419-1439, 2004 A2. Variable Definitions A3. Summary Statistics Descriptive Statistics N Minimum Maximum Mean Std. Deviation CO2 182 .0400 5994.2900 154.113077 626.9503107 POP 182 64 1304363 35197.29 129968.035 CE 182 7 1591261 42788.04 156446.621 GDP 182 1.2106E8 1.2400E13 3.069828E11 1.0987906E12 per capita gdp 182 273.4706 68814.9425 1.146575E4 1.3504252E4 per capita CO2 emission (tonnes) 182 21 63147 5666.51 8851.159 Valid N (listwise) 182 A4. Confirmation by tests for Normality Tests of Normality Kolmogorov-Smirnova Shapiro-Wilk Statistic df Sig. Statistic df Sig. CO2 .403 182 .000 .228 182 .000 per capita CO2 emission (tonnes) .262 182 .000 .615 182 .000 per capita gdp .204 182 .000 .775 182 .000 a. Lilliefors Significance Correction Source : Own A5. Correlation Table Correlations CO2 POP CE GDP OIL KP CO2 Pearson Correlation 1 .694** .989** .927** .298** .081 Sig. (2-tailed) .000 .000 .000 .000 .277 N 182 182 182 182 182 181 POP Pearson Correlation .694** 1 .685** .536** .301** .074 Sig. (2-tailed) .000 .000 .000 .000 .324 N 182 182 182 182 182 181 CE Pearson Correlation .989** .685** 1 .962** .304** .102 Sig. (2-tailed) .000 .000 .000 .000 .173 N 182 182 182 182 182 181 GDP Pearson Correlation .927** .536** .962** 1 .272** .124 Sig. (2-tailed) .000 .000 .000 .000 .097 N 182 182 182 182 182 181 OIL Pearson Correlation .298** .301** .304** .272** 1 -.050 Sig. (2-tailed) .000 .000 .000 .000 .503 N 182 182 182 182 182 181 KP Pearson Correlation .081 .074 .102 .124 -.050 1 Sig. (2-tailed) .277 .324 .173 .097 .503 N 181 181 181 181 181 181 **. Correlation is significant at the 0.01 level (2-tailed). A6 Partial Correlation Co-efficient R Software output Partial Correlation Coefficient: CO2 CE GDP CO2 1.0000000 0.6934645 -0.1025477 CE 0.6934645 1.0000000 0.7607298 GDP -0.1025477 0.7607298 1.0000000 A7. Final Fitted Regression Model: R Software Output Call: lm(formula = CO2 POP + CE + GDP + RED) Residuals: Min 1Q Median 3Q Max -318.405 -5.794 6.360 13.163 342.509 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -9.166e+00 1.306e+01 -0.702 0.484 POP -3.777e-04 6.999e-05 -5.397 2.32e-07 *** CE 5.990e-03 1.790e-04 33.463 < 2e-16 *** GDP -2.671e-10 2.238e-11 -11.938 < 2e-16 *** RED[T.esap] 3.404e+00 1.811e+01 0.188 0.851 RED[T.lac] 6.578e+00 1.822e+01 0.361 0.719 RED[T.mena] 2.046e+01 1.995e+01 1.025 0.307 RED[T.sesap] -3.095e+00 1.642e+01 -0.188 0.851 RED[T.wena] -8.649e+00 2.157e+01 -0.401 0.689 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 67.42 on 165 degrees of freedom (8 observations deleted due to missingness) Multiple R-squared: 0.9894, Adjusted R-squared: 0.9889 F-statistic: 1932 on 8 and 165 DF, p-value: < 2.2e-16 A8. Regression Models without and with oil respectively Call: lm(formula = CO2 + POP + CE + GDP) Residuals: Min 1Q Median 3Q Max -324.6545 -0.8434 6.4322 11.0498 341.7354 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -6.603e+00 5.372e+00 -1.229 0.221 POP -3.742e-04 6.706e-05 -5.580 9.34e-08 *** CE 5.997e-03 1.730e-04 34.668 < 2e-16 *** GDP -2.692e-10 2.127e-11 -12.658 < 2e-16 *** --- Residual standard error: 66.89 on 170 degrees of freedom (8 observations deleted due to missingness) Multiple R-squared: 0.9893, Adjusted R-squared: 0.9891 F-statistic: 5232 on 3 and 170 DF, p-value: < 2.2e-16 Call: lm(formula = CO2 POP + CE + GDP + OIL) Residuals: Min 1Q Median 3Q Max -317.732 -2.442 5.362 12.094 337.981 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -5.073e+00 5.797e+00 -0.875 0.383 POP -3.690e-04 6.755e-05 -5.462 1.66e-07 *** CE 6.001e-03 1.733e-04 34.622 < 2e-16 *** GDP -2.691e-10 2.130e-11 -12.637 < 2e-16 *** OIL -9.518e+00 1.343e+01 -0.709 0.480 --- Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 Residual standard error: 66.99 on 169 degrees of freedom (8 observations deleted due to missingness) Multiple R-squared: 0.9893, Adjusted R-squared: 0.9891 F-statistic: 3913 on 4 and 169 DF, p-value: < 2.2e-16 Read More