Carbon Presentation
- 1. Do wealthy people around the world produce high levels of carbon emissions? 16th March 2012 Statistics in Economics - Gurpreet Sidhu 1
- 2. Motivation ? Sympathetic to reducing personal carbon footprint ? World needs to work together to slow down the green house effect ? Understand how different classes of society around the world are contributing to this effect Background ? Carbon Emissions (GHG) ¨C Combustion of Wood, Coal, Oil and Natural Gas ? 2009 ¨C 41.5% of Total Carbon emission by United States and China ? 22 tonnes per person between US and China! http://en.wikipedia.org/wiki/Greenhouse_gas 16th March 2012 Statistics in Economics - Gurpreet Sidhu 2
- 3. More Motivation ¨C 1 tonne of CO2 http://www.freja.com/FRONTPAGE/Environment 16th March 2012 Statistics in Economics - Gurpreet Sidhu 3
- 4. Data Collection - 2009 Variable Unit Source Country Name http://tonto.eia.doe.gov/cfapps/ipdbproject/IEDIndex3. CO2 Emissions Million Metric Tonnes cfm?tid=90&pid=44&aid=8 http://en.wikipedia.org/wiki/List_of_countries_and_out Country Area KM2 lying_territories_by_total_area http://www.photius.com/rankings/population/populati Population on_2009_0.html http://www.enotes.com/topic/List_of_countries_by_G GDP per Capita $ DP_%28nominal%29_per_capita http://epi.yale.edu/epi2012/rankings Environmental Ranking Score 0 - 100 http://en.wikipedia.org/wiki/Environmental_Performan Performance Index (EPI) ce_Index http://www.cru.uea.ac.uk/~timm/cty/obs/TYN_CY_1_1. Average Temperature Degrees Celsius html http://www.prb.org/pdf09/09wpds_eng.pdf Population Growth Rate % per Year http://databank.worldbank.org/ddp/home.do?Step=12 Life Expectancy Years &id=4&CNO=2 http://www.unicef.org/statistics/index_step1.php Urban Living Population % http://en.wikipedia.org/wiki/Developed_country Developed Country Binary (0-No, 1-Yes) Wealth of Country Binary by Quantile GDP per Capita (Poorest in Intercept, Poor, Wealthy, Wealthiest) 16th March 2012 Statistics in Economics - Gurpreet Sidhu 4
- 5. Data Summary ¨C Proc Univariates ? Y variable is CO2 emissions ? Data collected for top 100 CO2 emitting countries worldwide X Variable Mean Min Max Skewness Kurtosis land 936696.7 347 17098242 4.976018 28.670831 pop 45703222 109825 1166079217 7.965223 71.43029 gdpcap 19060.62 100 80943 1.0939 1.4609 epi 53.10206 25.30 76.7 -0.302 -0.342 temp 17.430 -5 28.8 -0.464 -0.747 popgrowth 1.2288 -0.8 10.3 2.9525 17.282 life 73.2164 46 83 -1.6396 3.3631 urbanpop 65.46 14 100 -0.5113 -0.2711 dev 0.3711 0 1 0.5419 -1.7427 ? Skewness and Kurtosis = 0 if perfect Normal Distribution ? Of interest land, population and popgrowth ? With foresight investigation needed into land, population and gdpcap (R2 and beta) 16th March 2012 Statistics in Economics - Gurpreet Sidhu 5
- 6. Scatterplots Min and Max Values Skewed Distribution China United States 16th March 2012 Statistics in Economics - Gurpreet Sidhu 6
- 7. Scatterplots ¨C transforming variables by taking logs Min and Max Values Better Distribution China United States Collinearity 16th March 2012 Statistics in Economics - Gurpreet Sidhu 7
- 8. Mutliple Linear Regression ? 99 Observations ? Should drop ¡®lnland¡¯ due to collinearity but will double check with p-value first ? Possible Interactions in the data I have added: devlnland=dev*lnland devlnpop=dev*lnpop 16th March 2012 Statistics in Economics - Gurpreet Sidhu 8
- 9. Mutliple Linear Regression 2 ? p-values have improved across variables ? still many insignificant p-values above 0.05 16th March 2012 Statistics in Economics - Gurpreet Sidhu 9
- 10. Mutliple Linear Regression 3 ? still many insignificant p-values above 0.05 16th March 2012 Statistics in Economics - Gurpreet Sidhu 10
- 11. Mutliple Linear Regression 4 ? Good model ? 99 Observations ? R2 is 0.3241 ? EPI p-value = 0.07 is questionable but we leave it in for now with benefit of foresight ? Use model to calculate studentized residuals for all observations 16th March 2012 Statistics in Economics - Gurpreet Sidhu 11
- 12. Studentized Residual ¨C Boxplot and Extremes Studentized Residual = Residual / Standard Deviation of Residual Outliers China and US need to be removed so that errors will be more normally distributed China United States 16th March 2012 Statistics in Economics - Gurpreet Sidhu 12
- 13. Mutliple Linear Regression 5 ¨C No Outliers ? 97 Observations ? R2 now 0.5205, previously 0.3241 ? p-value for epi much better 16th March 2012 Statistics in Economics - Gurpreet Sidhu 13
- 14. Mutliple Linear Regression 6 ¨C Heteroskedasticity ? Parameter estimates all lie within Heteroskedasticity consistent 95% CI ? To fix this we use new standard errors to put into our regression model 16th March 2012 Statistics in Economics - Gurpreet Sidhu 14
- 15. Conclusion CO2 = -3239.86 + 147.20lnpop + 130.74lngdpcap ¨C 4.87epi (31.43) (40.29) (2.78) *** *** *** At 5% level there is significant evidence of: ? Each 1% increase of lnpop, CO2 increases by 147/100 ? Each 1% increase in lngdpcap, CO2 increases by 130/100 ? Each 1 unit increase in EPI, CO2 decreases by 4.87 ? gdpcap is an indicator for measuring wealth ? So in answer to our original question, wealthy people do emit more CO2 16th March 2012 Statistics in Economics - Gurpreet Sidhu 15