Get Complete Project Material File(s) Now! »
Results and Discussion
Intensity Trends Worldwide Over Time
The change in intensities over time of individual countries does not seem to have an inverted u shaped trajectory, but when each cross-section is viewed as a whole, the EKC does form. This should imply that countries with lower per capita incomes have increasing intensities while those with high incomes have decreasing intensities. The underlying motives for this may be several (change in pro-ductions functions, preferences, income among others), due to the heterogeneity of the sample.
After plotting the four income groups against time (Figure 3) the following patterns are ob-served: generally speaking carbon intensity is decreasing for all countries. The high-income group ex-hibits the least CO2 intensity rates which are in line with Kuznets prediction. They are followed by the lowest income countries which can possibly be due to the fact that their manufacturing sector is still well behind in comparison with developed countries. The trend for the upper middle income group is also decreasing in intensity even though the line is above the one presenting the high income group of countries. In the future they will probably reach the state of the high income group and also become more efficient. The lower middle income group shows improvements in its production efficiency and is expected to further reduce its pollution intensity in the long run. Thus, all four groups are in line with Kuznets expectations for GDP per capita.
GDP per capita
The experiment’s outcome replicates the results obtained by Roberts & Grimes (1997). Our li-near parameters, (log transformed), presented below in Table 1, become less significant while the quadratic increase in significance. This indicates that at the beginning of the studied period the corre-lation is linear but over time becomes curvilinear.After 1987 the EKC forms. Since 1990 both parameters in Stern’s regression become significant and the R2 increases slowly over time. This indicates a transformation of the relationship between GDP per capita and carbon emission intensity from linear to quadratic. Therefore,Stern’s functional form yield results that imply the existence of EKC and pro-vides more efficient estimates.
As can be seen in Figure 4, our results for GDPc are similar to Roberts and Grimes in the geo-metric progression of the curves; although this figure utilizes the Stern functional form and a different sample. The relationship is mostly linear in year 1972 but as time progresses, the relationship becomes more and more parabolic, suggesting the EKC effect is actually intensifying as time goes by The exclusion of extreme observation such as Nepal and the DR of Congo do not substantially modify the estimated turning point nor the geometric dispositions of the curves and thus they were left in the regression.
Worth noting is the gradual decrease in the turning point intensity which starts at approximately 2.7 tonnes of CO2 in 1972 and finishes at ~0.78 tonnes of C02 in 2005. Also interesting is that the per capita income level at which the decrease begins has remained relatively constant throughout the pe-riod at around $2700.
The emergence of EKC might be a result of improving technologies that increase the efficien-cy of industrial activities which is the original proposition of the EKC.The analysis for the four in-come groups indicates that generally the CO2 emissions are being reduced but they do not form an EKC and we can not say at what pace for each and every country this occurs.This means that even though the general trend is towards improving efficiency some countries are still on the upward going side of the Kuznets curve.
We are aware of the fact that some studies have reached controversial results so maybe another reason for our results is the particular sample of countries and time range that we have chosen.It is al-so possible that the quality of the data has improved over the years,especially having in mind that until recently CO2 was considered harmeless and therefore did not raise much attention, and now more credible results may be obtained.
Fossil Fuels share
The p-values for the fossil fuel regression show a significant linear relationship from 1972 until 1987 (Appendix 3,Table 5 and Figure 8). At that point the significance level begins to fluctuate al-though it remains under the 10 % line. Since 1988 neither of the parameters have been significant. This suggests that the relationship has broken down as other factors intervene. When viewing the re-sults for the Stern method, they are not significant. This is probably due to the fact that a quadratic form simply does not reflect the relationship between fossil fuel combustion and rates of carbon in-tensity. Intuitively, carbon intensity should increase with the percentage of fossil fuel combustion. However, the results suggest that one does not determine the other.
Alternative and Nuclear Energy share
The approach by Roberts & Grimes shows significant parameters for the later part of our sam-ple (Appendix 3,Table 6 and Figure 9) .The parameters are significant since 1985 and their magnitudes yield a negative linear relationship as expected. The Stern regression is for all practical reasons insigni-ficant throughout. An interesting observation is the way the linear parameters for fossil fuels in the Roberts & Grimesregressions are significant until 1987, point at which the parameters for alternative energy sources become, and stay significant for the remainder of the period. This might suggest that alternative energy sources are becoming more relevant in defining the carbon emissions intensity of the world’s economies. This effect is not present with the Stern functional form
Since there is some criticism of Roberts & Grimes functional form (Mcnaughton & Lee, 1998) and since the adjusted R squared are already very low, it does not seem like there is a substantial rela-tionship between the variables. If we consider that current projections estimate a 6% of global elec-tricity will come from alternative energy, coupled with the weak relationship during the past decades, it seems as if alternative energy will not exert a large effect on world carbon intensity. As shown in Error! Reference source not found. (OECD/IEA 2010), even with optimistic projections, fossil fu-els will remain the dominant source of energy for the next decades.
Life Expectancy at birth
Both parameters are consistently significant at the 10% level until 1979 (Appendix 3,Table 7 and Figure 10) for both Stern and Timmons Roberts and Grimes’ regressions. In Roberts and Grimes results when compared to the linear parameter, the quadratic parameter is negligible (equals 0 throughout) which results in a linear relationship between life expectancy and carbon intensity. What’s more, the standard error is very high and increases over time. Thus, there is not a Kuznets relation-ship between the variables. To further weaken the prospect of an EKC for life expectancy, the ad-justed R squared drifts from 0.41 in 1972 to 0.12 by 2005 which mimics the results obtained by Ro-berts and Grimes.
Stern’s results show that both linear and quadratic parameters are high,therefore when plotted life expectancy exhibits a parabolic shape but at the same time the standard error is much higher (from 20.25 to 44.6) and therefore the goodness-of-fit is doubtful.
According to a survey conducted in the USA (Gogklany 2010),during the last century the country has witnessed vigorous development . To give a few examples, income, population, chemical and metal use all have increased dramatically. Of course, this has led to increased levels of CO2 emis-sions as well. One would suggest that health and longevity decrease as a result of pollution. Contrary to popular belief, the statistics prove this notion wrong. Life span has not only increased during the last decades but also the rate of disability has decreased and some of the most serious diseases occur later in life than they used to. From this overview, one can conclude that life expectancy seems to be-come less and less related to emissions, which may be the reason for the loss of explanatory power presented in the results of the above regressions. Examining the countries constituting our sample, we reached similar conclusions. The high income group (to which USA also belongs) is characterized by a steady increase in GDP per capita and life span (Appendix 4, Figure 12 & 13). The C02 emissions are currently being reduced by the majority of the countries. However, it is worth noting that throughout the 20 th century pollution greatly increased but still it did not have negative effects on lon-gevity and health. In the upper middle income group, the majority of the countries in the sample in-crease their emissions (exceptions are Dominican Republic, Jamaica,Venezuela,South Africa; Appen-dix 4, Figure 14) .The life expectancy increases for the group despite the negative externalities of pol-lution. In the lower middle income group, we observe the same trend. China is one of the largest emitters of CO2, GDP per capita rises annually at a fast pace but at the same time life span increases as well (Appendix 4, Figure 15). In the low income group only Zambia reduces the amount of CO2 emitted every year. Again for the whole group the life span is on the rise (Appendix 4, Figure 16)
Percentage of Rural population
Neither the linear, nor the quadratic coefficient shows a significant relationship with carbon in-tensity (Appendix 3, Table 8 and Figure 11). What’s more, the adjusted R squared is seldom above 0.1. Our evidence suggests that carbon intensity is not a quadratic or linear function of the percentage of rural population. This can be partly attributed to other factors. For example, Trinidad and Tobago has rich oil resources, a sector which produces 40% of GDP but only 5% of employment while the rural population percentage has remained around 90% for the past decades (World Bank). Another example is Togo where 65% of the labor force is engaged in agriculture (Central Intelligence Agency 2011) but at the same time one of the country’s most important activities is the manufacturing and export of cement (1.2 million tons annually) which requires huge amounts of fossil fuels (Van Straaten 2002).
If we look at wealthy countries such as the United States we also find evidence that supports the lack of significant results in the regressions. The US has 17 % rural population. It is the world’s biggest consumer of oil, natural gas and electricity (CIA Factbook) (Figure 6 & 7 and Appendix 5, Figure 17). While India has a 70% rural population, the country ranks number 3rd in coal consump-tion, 5th in world oil consumption, 16th in natural gas consumption (2009 est.) and is the second larg-est producer of cement in the world (2008). It also spends around 6% of GDP on oil and gas imports (OECD/IEA 2010). Interesting characteristics that go against the EKC are exhibited by China, as well. Its rural population is 53% (2010) but it is by far the world biggest producer of cement .The country ranks 3rd in the world in oil consumption and 9th in natural gas consumption (United States Geological Survey 2008).It is also the world’s second producer and consumer of electricity (CIA Factbook ).China is the largest consumer of coal worldwide (British Petroleum 2010).
Comparison of the results obtained by the two functional forms
GDP per capita
The results yielded by both Roberts and Grimes and Stern’s econometric models support the exis-tence of EKC for GDPc.Therefore,it was proven that over time and for this specific sample of coun-tries production efficiency improves and CO2 emission intensity decreases.
Fossil fuels
Roberts and Grimes functional form led to the conclusion that for the first halv of the period fossil fuels combustion is significant and the relationship is linear.However,for the later part of the period the connection breaks down.Stern’s approach does not yield significant results throughout.
Alternative and Nuclear Energy
The results obtained by utilizing Roberts and Grimes functional form indicate that alternative and nuclear energy becomes significant at the alter part of the period,approximately at the time when fos-sil fuels lose their explanatory power.The results from Stern’s regressions are insignificant.
Life expectancy at birth
The results for life expectancy at birth are very similar for both functional forms and they show that this variable is significant throughout the period.Even though a connection between CO2 emission in-tensity and Life expectancy is evident the correlation is linear since the quadratic parameters are low for both regressions(0 for Roberts and Grimes) and the standard errors are high.
Rural Population
Both Roberts and Grimes and Sterns regressions yielded results according to which rural population is insignificant and cannot account for CO2 emission intensity.
Conclusion
This paper was meant to study whether the chosen variables have any relationship with regard to carbon emission intensity, if they follow the inverted U-shape curve proposed by Kuznets and how they vary in different income groups. For that purpose yearly cross -sectional regressions were run. Since previous research has not decided upon a certain econometric method, a comparison between two models was carried out. The obtained results do not differ greatly from each other. Both Stern and Roberts and Grimmes approaches indicate a connection between GDP per capita and life expec-tancy at birth to carbon emission intensity. However, the latter has a relationship to carbon emission intensity but it does not form an EKC. Following Timmons approach, we conclud that a gradual change in explanatory power from fossil fuels towards alternative and nuclear energy is present. How-ever, Stern’s approach does not yield any significant results. The results from both methods indicate that the share of rural population is insignificant with regard to carbon emission intensity.
The comparison of the two functional form led us to the conclusion that they do not yield very different results (with exception of fossil fuels and alternative and nuclear energy).Both of them may be considered credible and the results obtained by the two robust.However,the results obtained by Stern (2004) approach are more in line with current trends and therefore probably more reliable
Our study has allowed for a broader look at the function of development indicators for the EKC. It might be of interest to future researches to study the development of the EKC not from an administrative classification, as we did but one which focuses on economic areas or trade blocks. The EKC might also have interesting results if studied by sectors or within specific classifications such as low income countries.
Contents
Introduction
1.1 Purpose
1.2 Limitations
2 Background
2.1 Theoretical Framework
2.2 Kuznets Curve
2.3 Environmental Kuznets Curve
3 Empirical Framework
Emission intensity
Country income specification
Fossil Fuels Combustion
Alternative and Nuclear Energy
Life expectancy at birth
Rural Population
4 Empirical Testing
4.1 Correlation between explanatory variables
4.2 Roberts and Grimes Methodology
4.3 Stern Methodology
4.4 Functional forms for this study.
4.5 Expectations
GDP per capita
Fossil Fuels Combustion
Alternative and Nuclear Energy
Life expectancy at birth
Rural population
5 Results and Discussion
5.1 Intensity Trends Worldwide Over Time
5.2 GDP per capita
5.3 Fossil Fuels share
5.4 Alternative and Nuclear Energy share
5.5 Life Expectancy at birth
5.6 Percentage of Rural population
5.7 Comparison of the results obtained by the two functional forms
6.Conclusion
7. References
GET THE COMPLETE PROJECT
Environmental Kuznets Curve for Carbon Intensity