Regression analysis. (economic)

   

Impact of Human Activities on Carbon Dioxide Emissions

Tom Lebo

Kate O’Brien

Applied Econometrics I

Irina Murtazashvili

December 8, 2010

1    

ABSTRACT

This paper attempts to examine the relationships between carbon dioxide emissions and

urban population, motor vehicles, forest area, and energy use. Because carbon dioxide is the

principal greenhouse gas in the atmosphere, it has the largest effect on the current global

warming crisis and will be the focus of our study. Using data on 74 countries from 2007, we

regress the log of carbon dioxide emissions on the log of urban population, the log of motor

vehicles, the log of forest area, and the log of energy use to determine the effect of human

activities on carbon dioxide emissions. The empirical results suggest that urban population,

motor vehicles, forest area, and energy use are highly significant in explaining carbon dioxide

emissions. We will also discuss existing environmental technology and policy surrounding

carbon dioxide emissions.

2    

1 Introduction

Human impact on the environment is an issue of growing importance in the world today.

Climate change, a major environmental concern, is a direct result of rising quantities of

greenhouse gases in the atmosphere, which have worsened the “greenhouse effect,” the process

by which atmospheric gases absorb energy from the sun and trap it in the atmosphere, warming

the Earth’s surface (United States Environmental Protection Agency [U.S. EPA], n.d.). Global

warming is linked to a variety of serious environmental issues occurring over the last century,

including more frequent heat waves and droughts, heavier precipitation, more intense tropical

storms, rising sea level, ocean acidity, glacial melting, and an increase in the global land and sea

temperature (Intergovernmental Panel on Climate Change [IPCC] “Observed” 2007). These

climate change indicators have a widespread impact on animal and plant life and are slowly

destroying the world’s ecosystems.

Are humans really to blame for carbon dioxide emissions and global warming? After all,

the greenhouse effect occurs naturally and allows life on our planet to exist, and carbon dioxide

is emitted by humans through the biological process of respiration and used by plants during

photosynthesis (U.S. EPA, n.d.). However, the primary causes of carbon dioxide emissions are

the combustion of fossil fuels (coal, oil, and natural gas) in power plants and motor vehicles and

deforestation, both of which are induced by human progress and have grown steadily since the

Industrial Revolution (IPCC “Causes,” 2007). This study will concentrate on factors that affect

carbon dioxide emissions, namely urban population, motor vehicles, forest area, and energy use

and examine these variables statistically to determine their collective influence. We will look at

data from 2007 for 74 countries: carbon dioxide emissions in kilotons, urban population, motor

3    

vehicles per 1,000 people, forest area in square kilometers, and energy use in kilotons of oil

equivalent, which includes the primary forms of oil, coal, and natural gas.

2 Literature Review

Although the effect of greenhouse gases on the environment and the Earth’s changing

climate are receiving increasing attention in both scientific and political circles, research into

these topics not a new development. The Intergovernmental Panel on Climate Change (IPCC)

was established in 1988 to provide the governments of the world with a scientific explanation of

current climate conditions (United Nations, n.d.) Observations made by the IPCC indicate that

increases in temperature, sea level, precipitation, drought, and tropical storm activity and

decreases in snow and ice extent are consistent with global warming (Intergovernmental Panel

on Climate Change [IPCC] “Observed,” 2007). Based on observational evidence, the IPCC also

links global warming to changes in terrestrial and marine biological systems. Atmospheric

carbon dioxide has increased globally by 36% over the past 250 years, and the increases are

mainly due to emissions caused by changes in energy use and land use as a result of

industrialization: in other words, due to human processes such as the combustion of fossils fuels

and deforestation (Forster et al. 2007).

A review of existing literature on environmental economics contributed little to our

understanding of emissions as the explained variable. Most of the literature investigates the

economic impact of emissions, such as the effect of emissions on a country’s GDP. Of the few

empirical studies we could find concerning carbon dioxide emissions, one examining emissions

along with energy consumption, income and foreign trade in Turkey found that income is

actually the most significant variable in explaining emissions, followed by energy consumption

and foreign trade (Halicioglu 2008, 2). Another study focusing on carbon dioxide emissions in

4    

major metropolitan areas of the United States found that higher urban population density in

certain regions of the country is associated with lower greenhouse gas production, and there is a

strong, negative correlation between emissions and land use regulations (Glaeser and Kahn 2010,

29; Glaeser and Kahn 2010, 1).

Because economists view the climate change problem from a cost-benefit standpoint,

they are concerned with increasing greenhouse gas emissions not because emissions are

detrimental to the environment, but because they hinder economic growth. Although “politicians

are proposing to spend hundreds of billions of dollars on greenhouse gas emission reduction …

at present, economists cannot say with confidence whether this investment is too much or too

little” (Tol 2009, 18). Additionally, carbon dioxide emissions have a disproportionate effect on

the world’s poorest people: developing countries are more vulnerable to the increased impact of

extreme weather events and as a result are affected more by human-induced climate change (The

World Bank Group “News”, n.d.). Because “pre-existing poverty is one of the main causes for

vulnerability to climate change,” economists cannot agree “whether stimulating economic

growth or emission abatement is the better way to reduce the effects of climate change” (Tol

2000, 36). Thus, the papers investigating factors that affect carbon dioxide emissions are more

meaningful to our study than papers on environmental economic policy.

Based on the scientific findings by the IPCC, we would expect energy use to have a

significant positive effect on carbon dioxide emissions and amount of forest area to have a

negative effect on carbon dioxide emissions. While the use of energy produces carbon dioxide,

trees actually absorb carbon dioxide from the atmosphere, so countries with less forest area

likely emit more carbon dioxide into the atmosphere. Since the combustion of fossil fuels

contributes to atmospheric carbon dioxide, we would expect motor vehicles to have a positive

5    

effect on emissions. Furthermore, we would expect urban population to have a positive effect on

carbon dioxide emissions because homes, businesses, and motor vehicles are likely more

concentrated in urban areas.

3 Model and econometric methodology

We will use the following econometric model to determine the effect urban population,

motor vehicles, forest area, and energy use on carbon dioxide emissions:

log(carbon dioxide emissions) = β1 + β2 log(urban population) + β3 log(motor vehicles) +

β4 log(forest area) + β5 log(energy use).

We considered a few other potential models which included more variables in both level-level

and log-level functional forms (results of those regressions can be found in the tables following

this report). However, the log-log model above is the most parsimonious choice, yet still exhibits

a high R-squared value. Also, all the variables were individually statistically significant at the

lowest levels. The functional form provides an easier interpretation of parameters in terms of

percentage changes as well. Data is from The World Bank’s World Data Indicators, specifically

under the following indicators: Energy & Mining, Environment, and Urban Development.

The reader should be aware of a few problems with the econometric model. One minor

issue is that the data does not come from a random sample; rather, the observations comprise the

broadest collection of countries for which we could find information on all indicators we chose

to analyze. Though this violates one of our assumptions for estimating ordinary least squares

(OLS) estimators of a multiple linear regression, the data was not selected based on any common

characteristic that could skew results, so we kept the sample as close to random as possible given

the data we were provided. In addition, the model violates the homoskedasticity assumption

based on the results of the Breusch-Pagan test: when we regress the squared sample residuals on

6    

the regressors from the model, the F-statistic is statistically significant, an indication of

heteroskedasticity. The heteroskedasticity-robust standard errors and significance levels for our

chosen model are reported in the tables following this report. 4 Empirical results

As the tables at the end of this report suggest, the log-log model is a very good fit for the

data. In fact, urban population, motor vehicles, forest area, and energy use together explain about

92.9% of the variation in carbon dioxide emissions, based on the R-squared value, which is quite

high considering only four explanatory variables are included in the regression. The intercept,

-3.389 kilotons of carbon dioxide, is not meaningful to our study because we would never expect

the urban population, number of motor vehicles per 1,000 people, forest area, and energy use for

a given country to all equal zero at the same time, and because carbon dioxide is emitted through

natural processes in addition to the human activities we are concerned with, the amount of

emissions would never be negative.

The coefficient on log(urban population) indicates that, holding the other explanatory

variables in the model constant, for every one percent increase in urban population, carbon

dioxide emissions are predicted to increase by about 0.342%, a positive effect which is in line

with our expectations since we would associate higher population density with a higher level of

emissions. Holding all other explanatory variables constant, the coefficient on log(motor

vehicles) implies that for every one percent increase in the number of motor vehicles per 1,000

people, carbon dioxide emissions are predicted to increase by 0.335%. Again, this result agrees

with our original hypotheses and makes sense based on the fact that a significant amount of fossil

fuel energy use can be attributed to motor vehicles.

7    

The coefficient on log(forest area) suggests that holding all other factors constant, a one

percent increase in forest area is predicted to decrease carbon dioxide emissions by 0.087%.

Interpreted in a more meaningful way (since we more often think of deforestation as a cause of

the increase in atmospheric carbon dioxide, not the addition of forest area as a way to decrease

atmospheric carbon dioxide), a one percent decrease in forest area is predicted to increase carbon

dioxide emissions by 0.087%, as we hypothesized earlier. This negative relationship is intuitive

because we know deforestation has been observed to lead to the emission of carbon dioxide.

Holding all other factors constant, the coefficient on log(energy use) indicates that a one percent

increase in energy use is predicted to increase carbon dioxide emissions by 0.788%; in other

words, the elasticity of carbon dioxide emissions with respect to energy use is 0.788%. This

result is in line with our expectations, since there is scientific evidence that the combustion of

fossil fuels is a major source of atmospheric carbon dioxide.

Based on their p-values (using the heteroskedasticity-robust standard errors), the log of

each of the explanatory variables is significantly different from zero at the 0.05 level, so we

reject the null hypothesis that each variable by itself has no effect on carbon dioxide emissions.

In addition to being individually statistically significant, the explanatory variables are jointly

significant, based on the large F-statistic, and therefore collectively explain carbon dioxide

emissions very well.

5 Conclusions

Clearly, the impact of human activities on carbon dioxide emissions cannot be ignored.

Observational evidence already points to increased atmospheric carbon dioxide as a direct result

of fossil fuel use and deforestation, and with the added effects of urban population and motor

vehicles, the statistical evidence strongly supports the connection between human activities and

8    

carbon dioxide emission. Urban population, motor vehicles, and energy use have increased

steadily as a result of worldwide industrialization while forest area has decreased, and these

trends are forecasted to continue into the future, making the problems of global warming and

climate change even worse (IPCC “Projected” 2007). As a result, many world organizations have

developed and begun to implement plans to reduce greenhouse gas emissions. Through carbon

capture and sequestration, carbon dioxide is “isolated from the emissions stream, compressed,

and transported to an injection site where it is stored underground permanently” (U.S.

Department of Energy, n.d.). In addition, governments worldwide fund research and

development in renewable energy, such as solar and wind power, to reduce dependence on fossil

fuels as the main source of energy.

International policy has also played a role in helping to solve the problem of global

warming. The Kyoto Protocol sets standards committing industrialized countries to meet

emissions targets through monitoring and mechanisms such as reforestation and clean

development projects (United Nations Framework Convention on Climate Change [UNFCCC]

“Kyoto,” n.d.). Emissions reductions and removals allow carbon dioxide to be traded like any

other commodity in a “carbon market” similar to the stock market (UNFCCC “Emissions,” n.d.).

Debate continues over whether the appropriate environmental technology and policy tools

regarding carbon dioxide emissions are in place, and long-term benefits of emissions reduction

remain to be seen. Although global policy initiatives are costly, the responsibility taken by world

organizations to deal with the critical issues of global warming and climate change are the first

steps in creating a more sustainable world.

9    

Tables

Table 1: Regression Results of Log(Carbon Dioxide Emissions) on Log(Urban Population), Log(Motor Vehicles), Log(Forest Area) and Log(Energy Use)

Dependent Variable: lCO2 Independent Variables Model 1: Log-log model (Best choice) lurbanpop 0.342** (0.107) [0.150] lmotor 0.335*** (0.064) [0.097] lforest -0.087*** (0.031) [0.026] leneruse 0.788*** (0.097) [0.132] intercept -3.389** (0.975) [1.385] Observations 74 R-squared 0.9288 Adj. R-squared 0.9247

Table 2: Regression Results of Carbon Dioxide Emissions on Urban Population (in millions), Total Population (in millions), Gross National Income (in billions), Motor Vehicles, Forest Area

and Energy Use Dependent Variable: CO2

Independent Variables Model 2: Level-level model urbanpop_mill 2478.459* (749.654) [1420.035] totalpop_mill -767.965* (216.640) [420.528] GNI_bill -159.124*** (13.443) ]27.176] motor 73.958** (36.564) [34.544] forest -0.096*** (0.018) [0.031] eneruse 3.451*** (0.105) [0.199] intercept -25517.79** (12428.57) [9611.853] Observations 76 R-squared 0.9961 Adj. R-squared 0.9957

Table 3: Regression Results of Carbon Dioxide Emissions on Urban Population, Total Population, Gross National Income, Motor Vehicles, Forest Area and Energy Use

Dependent Variable: lCO2 Independent Variables Model 3: Log-level model urbanpop_mill 0.069*** (0.015) [0.014] totalpop_mill -0.014*** (0.004) [0.005] GNI_bill 0.0007*** (0.0003) [0.0003] motor 0.002** (0.0007) [0.0009] forest_k -0.0009** (0.0004) [0.0004] eneruse_k -0.007*** (0.002) [0.002] intercept 9.476*** (0.247) [0.271] Observations 76 R-squared 0.5873 Adj. R-squared 0.5514

10    

Table 4: Regression Results of Carbon Dioxide Emissions on Urban Population, Motor Vehicles, Forest Area and Energy Use

Dependent Variable: CO2 Independent Variables Model 3: Level-level model urbanpop 0.0022* (0.0003) [0.0013] motor -49.925 (59.480) [42.348] forest -0.106*** (0.024) [0.037] eneruse 2.518*** (0.076) [0.186] intercept -12161.03 (20861.38) [11394.68] Observations 78 R-squared 0.9876 Adj. R-squared 0.9869

Notes: Data Sources: The World Bank World Data Indicators Standard errors in parentheses Heteroskedasticity-robust standard errors in square brackets Statistical Significance: *** = 0.01 level, ** = 0.05 level, * = 0.10 level (significance is based on heteroskedasticity-robust standard errors) Description of variables: CO2 – carbon dioxide emissions in kilotons lCO2 – log(CO2) totalpop – total population totalpop_mill – totalpop/1,000,000 urbanpop – urban population lurbanpop – log(urbanpop) urbanpop_mill – urbanpop/1,000,000 GNI – gross national income in current US dollars GNI_bill – GNI/1,000,000,000 motor – motor vehicles per 1,000 people lmotor – log(motor) motor_k – motor/1,000 forest – forest area in square kilometers lforest – log(forest) forest_k – forest/1,000 eneruse – energy use in kilotons of CO2 equivalent leneruse – log(eneruse)

11    

Bibliography

Forster, Piers, Venkatachalam Ramaswamy, Paulo Artaxo, Terje Berntsen, Richard Betts, David W. Fahey, James Haywood, Judith Lean, David C. Lowe, Gunner Myhre, and John Nganga. “Changes in Atmospheric Constituents and in Radioactive Forcing.” Intergovernmental Panel on Climate Change. http://www.ipcc.ch/ ‌pdf/‌assessment- report/‌ar4/‌wg1/‌ar4-wg1-chapter2.pdf (accessed November 26, 2010).

Glaeser, Edward L., and Matthew E. Kahn. "The Greenness of Cities: Carbon Dioxide Emissions

and Urban Development." Journal of Urban Economics 67, no. 3 (May 2010): 404-418. http://www.nber.org/papers/w14238.pdf (accessed December 2, 2010).

Halicioglu, Ferda. "An Econometric Study of CO2 Emissions, Energy Consumption, Income and

Foreign Trade in Turkey." Unpublished essay. Munich Personal RePEc Archive. Munich University Library, 2008. Munich Personal RePEc Archive. http://mpra.ub.uni- muenchen.de/ 11457/1/MPRA_paper_11457.pdf (accessed December 2, 2010).

Intergovernmental Panel on Climate Change. “Causes of Change.” IPCC Fourth Assessment

Report: Climate Change 2007. http://www.ipcc.ch/ ‌publications_and_data/ ‌ar4/‌syr/‌en/‌spms2.html (accessed November 26, 2010).

———. “Observed Changes in Climate and Their Effects.” IPCC Fourth Assessment Report: Climate Change 2007. http://www.ipcc.ch/ ‌publications_and_data/ ‌ar4/‌syr/‌en/‌spms1.html (accessed November 26, 2010).

———. “Projected Climate Change and Its Impacts.” IPCC Fourth Assessment Report: Climate Change 2007. http://www.ipcc.ch/ ‌publications_and_data/ ‌ar4/‌syr/‌en/‌spms3.html (accessed November 26, 2010).

Tol, Richard S.J. "The Economic Effects of Climate Change." Journal of Economic Perspectives

23, no. 2 (Spring 2009): 29-51. doi:10.1257/jep.23.2.29. http://pubs.aeaweb.org/doi/ pdfplus/10.1257/ jep.23.2.29 (accessed December 2, 2010).

United Nations. “Climate Change.” United Nations Global Issues.

http://www.un.org/‌en/‌globalissues/‌climatechange/ ‌index.shtml (accessed November 27, 2010).

United Nations Framework Convention on Climate Change.“Emissions Trading.” United

Nations Framework Convention on Climate Change. http://unfccc.int/ ‌kyoto_protocol/‌mechanisms/‌emissions_trading/‌items/ ‌2731.php (accessed November 26, 2010).

12    

———. “Kyoto Protocol.” United Nations Framework Convention on Climate Change. http://unfccc.int/ ‌kyoto_protocol/‌items/ ‌2830.php (accessed November 26, 2010).

United States Environmental Protection Agency. “Greenhouse Gases.” Climate Change

Indicators in the United States. http://www.epa.gov/ ‌climatechange/ ‌indicators/‌pdfs/‌CI- greenhouse-gases.pdf (accessed November 26, 2010).

U.S. Department of Energy. “Carbon Capture and R&D Overview.” Fossil Energy – Carbon

Sequestration. http://www.fossil.energy.gov/‌programs/‌sequestration/ ‌overview.html (accessed November 27, 2010).

The World Bank Group. “Indicators – Data.” The World Bank.

http://data.worldbank.org/ ‌indicator (accessed November 26, 2010). ———. “News & Broadcast – Climate Change.” The World Bank.

http://go.worldbank.org/ ‌BPY7QIRNA0 (accessed November 26, 2010).