Econometrics report determinants of global carbon emissions from 2010 to 2019

Abstract 4 I. Introduction 5 II. Literature review 6 II.1. Theories related to factors affecting carbon emissions 6 II.1.1. Environmental Kuznets Curve (EKC) 6 II.1.2. Relationship between current emissions and ambient pollution 7 II.2. Related published researches 8 II.3. Research Limitations 9 III. Methodology 10 III.1. Methodology 10 III.2. Model specification 10 III.3. Explaining the variables 12 III.4. Theoretical relationship and supporting researches 12 IV. Results and testing 13 IV.1. Data description 13 IV.1.1. Sum 13 IV.1.2. Correlation matrix between the variables 14 IV.1.3. Selecting proper model 15 IV.2. Testing the model 17 IV.2.1. Test for Multicollinearity 17 IV.2.2. Test for Heteroskedasticity 18 IV.2.3. Test for Autocorrelation 18 IV.2.4. Fixing the problems in the model 20 IV.2.5. Test for the significance of the coefficients 21 V. Recommendations and solutions 22 V.1. Solutions for factor affecting carbon emissions 22 V.2. Recommendations to reduce carbon emissions 24 VI. Conclusion 26 References 27 Do File 28 Data table 30   Abstract Carbon dioxide is an important greenhouse gas that helps to trap heat in our atmosphere. Without it, our planet would be inhospitably cold. However, an increase in CO2 concentrations in our atmosphere is causing average global temperatures to rise, disrupting other aspects of Earths climate. Human activities had brought tremendous impacts on Co2 emissions and transformed it into hazardous issues threatening our planet. This paper aims to statistically estimate the factors resulting in the rise of carbon emission during the second decade of the twentyfirst century (20102019). Utilizing the semilog linear regression model along with data acquired from various trustworthy sources, we have examine the statistics of 104 nations from different regions in the world to determine the relationship of carbon dioxide (CO2) emission with five factors influencing it which are population, forest area, renewable energy, GDP, and vehicle on use. By incorporating evidence from various related articles

FOREIGN TRADE UNIVERSITY FACULTY OF INTERNATIONAL ECONOMICS

Hanoi, June 2023

I Introduction

II Literature review

II.1 Theories related to factors affecting carbon emissions

II.1.1 Environmental Kuznets Curve (EKC)

II.1.2 Relationship between current emissions and ambient pollution

II.2 Related published researches

II.3 Research Limitations

III Methodology

III.1 Methodology

III.2 Model specification

III.3 Explaining the variables

III.4 Theoretical relationship and supporting researches

IV Results and testing

IV.1 Data description

IV.1.1 Sum

IV.1.2 Correlation matrix between the variables

IV.1.3 Selecting proper model

IV.2 Testing the model

IV.2.1 Test for Multicollinearity

IV.2.2 Test for Heteroskedasticity

IV.2.3 Test for Autocorrelation

IV.2.4 Fixing the problems in the model

IV.2.5 Test for the significance of the coefficients

V Recommendations and solutions

V.1 Solutions for factor affecting carbon emissions

V.2 Recommendations to reduce carbon emissions

VI Conclusion

References

Do File

Data table

to statistically estimate the factors resulting in the rise of carbon emission during the second decade of the twenty-first century (2010-2019) Utilizing the semi-log linear regression model along with data acquired from various trustworthy sources, we have examine the statistics of 104 nations from different regions in the world to determine the relationship of carbon dioxide (CO2) emission with five factors influencing it which are population, forest area, renewable energy, GDP, and vehicle on use By incorporating evidence from various related articles along with our own research method, we come to the conclusion that GDP, number of vehicles in use and population has shown a medium-to-strong correlation to the amount of carbon emission Along with that, our group has included recommendations that should be taken in order to combat the effects deprived from the rise in carbon emission. 

I Introduction

Ever since the first Industrial Revolution, humanity has come a long way in advancing our technology and improving the lifestyle of our species like never before However, along with our achievements in technology and society, another problem has risen: the preservation of the natural environment The excessive use of fossil fuel to power our daily lifestyle resulted in the rise in carbon emission leading to a phenomenon called “global warming” In this study, we will describe the factors affecting carbon emission and prove the relationship and influence level of each of these factors to the amount of CO2 emission Our research scope includes 3 parts: content scope, time scope and spatial scope The content scope is researching factors affecting carbon emission The time scope of the data in this research is from 2010 to 2019 The spatial scope includes 104 countries from all around the world.

This research consists of 3 main sections In section 1, we present the definition of all research objects mentioned in the research, the economic theories related to the research, other related researches and the research hypothesis In section 2, we mention 3 parts: the methodology to derive the model and analyze the data, the theoretical model specification and describing the data In the last section, we strive to fix the model errors and estimate the model result, from there we recommend some detailed solutions toward each factor.

II Literature review

II.1 Theories related to factors affecting carbon emissions

II.1.1 Environmental Kuznets Curve (EKC)

The relationship between environmental quality and economic growth over time is normally represented in Kuznets environment curve, the logic of which is average income per capita and pollution level taking an inverted U shape Carbon emissions (representing pollution level) will increase in tandem with the increase of income to a certain extent, reaching its maximum value before declining while income continues on the rise This means as long as its value is still below the maximum, economic growth will entail degradation of environmental quality, but beyond that maximal point, both will ascend In other words, in the initial stage of the industrialization process, pollution rapidly augments due to the increased use of natural resources and increased emission of pollutants, causing serious environmental degradation In the later stages of the industrialization process, when income elevates, people are better aware of environmental protection, environmental policies and legislation as well as enforcement agencies are stricter, more advanced technologies are applied, etc., then environmental quality will improve. 

However, the environmental Kuznets curve does not imply that economic growth itself directly reduces environmental degradation Instead, it suggests that economic development, combined with appropriate policies and technological advancements, can eventually lead to

an improvement in environmental quality It's important to note that the EKC theory has been subject to various criticisms and has mixed empirical support Critics argue that the relationship between income and environmental degradation is not universally applicable and can be influenced by a range of factors, such as the structure of the economy, governance, technological advancements, and resource availability Moreover, the EKC does not consider the potential irreversible ecological damage caused by certain pollutants

or the unequal distribution of environmental impacts across different income groups.

In conclusion, the EKC serves as a useful framework for understanding the relationship between income and environmental degradation, but it should be interpreted with caution, considering the complexities of real-world dynamics and the need for comprehensive and context-specific environmental policies.

II.1.2 Relationship between current emissions and ambient pollution

When discussing pollutants, a common classification is to distinguish between accumulative pollutants and non-accumulative pollutants These classifications are based on the persistence and potential for accumulation of pollutants in the environment.

Accumulate pollutants are substances that have the ability to persist in the environment for long periods and tend to accumulate in living organisms or various environmental

compartments These pollutants are often characterized by their resistance to degradation or slow natural breakdown processes Examples of accumulative pollutants include: Persistent organic pollutants, Heavy metal, Persistent inorganic pollutants.

On the other hand, non-accumulative pollutants are substances that do not tend to persist in the environment or accumulate in living organisms They are generally more easily biodegradable or undergo natural degradation processes within a relatively short timeframe Examples of non-accumulative pollutants include: Biodegradable organic compounds and Non-toxic or Low-toxicity substances.

Panel (a) in Figure 2 denotes ambient pollution in case of non-accumulative pollutant in which damages are proportional to current emissions Whereas in panel (b), damages are dependent on the total stock of pollutant that has been released over time in the case of accumulative pollutants.

Carbon dioxide (CO2) is considered an accumulative pollutant, primarily due to its role in climate change While it is not persistent in the environment for long periods like some other accumulative pollutants, such as persistent organic pollutants or heavy metals, CO2 has the capacity to accumulate in the atmosphere and contribute to long-term changes in global temperature.

II.2 Related published researches

The discussion of how to promote “green growth” is rising to the top of the policy agenda for every national leader as a result of the burning issue in recognition of the major

challenges presented by global warming and climate change derived from carbon dioxide emissions Therefore, this topic has raised attention for many researchers for years and some featured researches can be listed out such as:

Determinants of CO2 Emissions: A Global Evidence (Jeremiás Máté Balogh and Attila

Jambor, 2017) established a link between carbon dioxide emissions and its various reasons

by employing a complex model comprising economic growth, industrial structure, FDI, energy use, trade and agriculture globally The research employed GMM models on a panel dataset and tested the result by the standard environmental Kuznets curve hypothesis As a result, the research concluded that financial development reduced air pollution.

Theoretical and Empirical Analyses on the factors affecting Carbon Emissions: Case of Zhejiang Province, China (Shaolong Zeng & Minglin Wang, 2022) made a theoretical

analysis and established the motion trajectory of carbon emissions in order to contribute in logical understanding of the factors affecting carbon emissions Based on time series data

and some methods such as linear regression, ARDL model analysis, vector autoregression (VAR) analysis, the empirical results show that there is a long-run relationship between AGDP, TECH, TRADE, KL, and AC, while in the short-run, the effect of TRADE is elastic and positively significant.The research concluded that the upgrading of the industrial

structure is essential to break the solidified energy consumption mode. 

Factors Affecting CO2 Emissions in the BRICS Countries: A Panel Data Analysis (Zakarya,

Mohammed Abbes and Seghir, 2015) analyzed the interactions that existed between total energy consumption, FDI, economic growth and the emissions of CO2 in the BRICS

countries, using the co-integration tests and panel Granger causality in panel The result indicated the existence of a unidirectional causality from CO2 to the independent variables, and helped decision makers in these countries to understand and grasp the complexity of this phenomenon.

II.3 Research Limitations 

Studies on domestic and foreign CO2 emission are various However, some research papers are based on qualitative and personal opinions, some only take data from a few countries that limit the result precision Therefore, these studies can only have an exiguous

perspective, it has not been shown that countries can encourage and support each other in CO2 emissions reduction.

Method our group use to collect and analyze the data:

● Collected data from 1040 observations from 104 countries from 2010 to 2019.

● Estimated values using statistical methods for quantitative results  with the same number of outputs and inputs.

● Compared our regression model results with previous research and studies to figure out the best result.

Knowledge and subject that has been applied: econometrics, macroeconomics and quantitative methods, environmental economics.

Tools for calculation and synthesis: STATA software, Microsoft Excel, Microsoft Word, Google Docs.

Data sources are collected and aggregated from published data on reputable statistical system websites of global (World Bank, Worldometer, OECD, OICA, )

III.2 Model specification

In order to build an econometric model, it is pivotal to identify the factors related to the interaction and description of economic variables The models are often suggested by economics To compute and analyze the output, we choose the statistical method in these fields, which is the estimation and verification of the hypothesis.

Realizing that some data in its original form is of substantial value, our group decided to get the natural base logarithm of some data fields to represent the field of that data Therefore, our group proposes the following research model:

Population regression model:

(Co2) = + 1 (ener) + (GDP) + (forest) + (pop)

+ 𝜷𝟓 ∗ veh + ci + i 𝒖

Where:

● 𝜷0: intercept of the model

● 𝜷j (j = 1,2,3,4,5): the regression coefficient of the equivalent independent variables

● ci: unobserved factors

● ui: random error, representing the factors affecting the dependent variable that are

not included in the model.

III.3 Explaining the variables 

III.4 Theoretical relationship and supporting researches

According to researchers and basic economic knowledge, renewable energy, domestic product, forest area, population, and vehicle use are the determining factors affecting the carbon emissions of many different countries in the world (Shaolong Zeng & Minglin Wang, 2022) Determinants of carbon emissions all around the world: using motion

trajectory models (Theoretical and Empirical Analyses on the factors affecting Carbon

Emissions: Case of Zhejiang Province, China).

IV Results and testing

IV.1 Data description

IV.1.1.Sum

Source: From STATA table

From the sum matrix, we have the conclusion: 

● Ln(Co2): the average CO2 amount in 104 countries from 2010 to 2019 was

10.78677, with standard deviation of 1.708429, a minimum of 7.209562 and

maximum of 16.18643

● Veh:  the average Vehicle amount in 104 countries from 2010 to 2019 was 10114.78,

with standard deviation of 30675.68, a minimum of  73 and maximum of 285813.3

● Ln(ener): the average Energy amount in 104 countries from 2010 to 2019 was

6.895644, with standard deviation of 2.967985, a minimum of  -5.809143  and maximum of 12.70131

● Ln(GDP): the average GDP amount in 104 countries from 2010 to 2019 was

25.59874, with standard deviation of 1.624108, a minimum of   21.95567  and maximum of 30.55928

● Ln(forest): the average Forest amount in 104 countries from 2010 to 2019 was

10.49375, with standard deviation of 2.401415, a minimum of  1.252763 and

maximum of 15.44785

● Ln(pop): the average Population amount in 104 countries from 2010 to 2019 was

16.55525, with standard deviation of 1.517006, a minimum of  12.66994 and

maximum of 21.08358

IV.1.2.Correlation matrix between the variables

Variable ln(CO2) veh ln(ener) ln(GDP) ln(forest) ln(pop) ln(Co2) 1.0000

Source: From STATA table

From the matrix, we can analyze the correlation between the independent variables and the dependent variables:

● The correlation between ln(Co2)and veh is 0.4985 > 0 illustrate a positive

relationship with relatively high correlation, consistent with the expectation. 

● The correlation between ln(Co2) and ln(ener)is 0.0896 > 0 illustrate a positive relationship with low correlation, inconsistent with the expectation.

● The correlation between ln(Co2)and ln(GDP) is 0.8518 > 0 illustrate a positive relationship with high correlation, consistent with the expectation. 

● The correlation between ln(Co2) and ln(forest) is 0.3528 > 0 illustrate a positive relationship with relatively high correlation, inconsistent with the expectation. 

● The correlation between ln(Co2) and ln(pop) is 0.6674 > 0 illustrate a positive relationship with high correlation, consistent with the expectation. 

Analyze the correlation between the independent variables:

● The correlation between ln(pop) and ln(GDP) is 0.6552>0 illustrate the highest correlation between two independent variables.

● Other independent variables have low to medium correlation with each other.  

IV.1.3.Selecting proper model 

Our team use STATA 15 to choose our model.

First, we use the code xtset to declare the panel data and here is the result:

xtset code Year

      panel variable: code

(strongly balanced)

      time variable:  Year, 2010 to

2019

      delta:  1 unit

a) Result and analyze each models

● Executing the code and store the result for OLS

reg ln(Co2) veh ln(ener) ln(GDP) ln(forest) ln(pop)

est store OLS

● Executing the code and store the result for FE

xtreg ln(Co2) veh ln(ener) ln(GDP) ln(forest) ln(pop), fe

est store fe

● Executing the code and store the result for RE

xtreg ln(Co2) veh ln(ener) ln(GDP) ln(forest) ln(pop), re

est store re 

● Comparing the results and analyzing each one

estimate table OLS fe re, star stats(N r2 r2_a) 

● Choose between POLS and RE

Run the command xttest0  - test the existence of ci (unobservable variable)

Set up the hypothesis :

H0: Choose POLS model (ci=0)

● Choose between FE and RE

By using the Hausman test, we test the correlation between ci and Xi.

Set up the hypothesis:

H0: No correlation between ci and Xi

H1: Correlation between ci and Xi

At 5% significance level

Hausman fe, sigmaless

Test:  Ho:  difference in coefficients

not systematic

chi2(4) = (b-B)'[(V_b-V_B)^(-1)](b-B)

      =      111.10

      Prob>chi2 = 0.0000

Because the p-value= 0.0000 < 0.05 => reject H0 => there is correlation between ci and Xi

=> we choose the FE model.

IV.2 Testing the model

IV.2.1.Test for Multicollinearity 

Set up the hypothesis:

H0: The model does not exist multicollinearity 

H1: The model exist multicollinearity

By using vif, we got the following result

Source: From STATA table

Because the result for Mean VIF= 1.83 <10 => Reject H0 => the model don’t have multicollinearity 

IV.2.2.Test for Heteroskedasticity

Set up the hypothesis

H0: The model don’t have heteroskedasticity

H1: The model have heteroskedasticity

a) Serial correlation

Set up the hypothesis

H0: The model don’t have serial correlation

H1: The model have serial correlation

At 5% significance level

By using the command xtserial with STATA, the result are obtained:

xtserial ln(Co2) veh ln(ener) ln(GDP) ln(forest)

Set up the hypothesis

H0: The model don’t have cross-section correlation

H1: The model have cross-section correlation

IV.2.4.Fixing the problems in the model 

       Our group concluded that the model has heteroskedasticity, serial correlation, and cross-section correlation Therefore, in order to fix the problems in our model, we use the

command xtscc ln(Co2) veh ln(ener) ln(GDP) ln(forest) ln(pop), fe on the STATA software

and get the following results:

xtscc lN(Co2) veh ln(ener) ln(GDP) ln(forest) ln(pop),fe

est store fextscc

est table fe fextscc, star stats (N r2)

IV.2.5.Test for the significance of the coefficients 

From the fextscc table, our team have formed a regression model

    ln(Co2) = 7.5871 + 0.0169.ln(ener) + 0.2792.ln(GDP)+ (-0.7417).ln(forest) +

(0.2619).ln(pop) + (-0.00005).veh

Set up the hypothesis

H0: The coefficient is not statistically significant

H1: The coefficient is statistically significant

At 5% significance level 

By using p-value, our team obtained the results 

● ln(ener)

● p-value = 0.03<0.05 => the variable is statistically significant

● β̂1 = 0.0169 => if Energy increase by 1% then the amount of  CO2 will increase by 0.0169%

● ln(GDP)

● p-value = 0.000<0.05 => the variable is statistically significant

● β̂2 = 0.2792 => if GDP increase by 1% the the amount of CO2 will increase

by 0.2792%

● ln(forest)

● p-value = 0.000<0.05 => the variable is statistically significant

● β̂3 = -0.7417 => if Forest increase by 1% the the amount of CO2 will decrease

by 0.7417%

● ln(pop)

● p-value = 0.224>0.05 => the variable is not statistically significant

● β̂4 = 0.2619 => if Population increase by 1% the the amount of CO2 will increase by 0.2619%

● veh 

● p-value = 0.002<0.05 => the variable is statistically significant

● β̂5 = -0.00005 => if Vehicle increase by 1 thousand then  the amount of CO2 will decrease by 0.005%

V Recommendations and solutions

V.1 Solutions for factor affecting carbon emissions

Population:

When the population increases, the fundamental needs for living (food, clothing, transportation, consumption) also increase, leading to indiscriminate exploitation of resources to meet human demands As a consequence, not only are resources depleted, but environmental pollution (especially the emission of CO2 from production and human activities) also increases.

Therefore, each couple should only have 1-2 children to control the world population, limit CO2 emissions, and contribute to sustainable development in society.

Forest area:

Firstly, it is necessary to enhance afforestation efforts by organizing projects and programs

to plant more trees Secondly, it is important to protect existing forest areas by raising awareness among people about the benefits of forests and the importance of not littering or polluting the environment to safeguard these valuable natural resources.

Currently, deforestation is increasingly rampant worldwide, so countries need to implement strict punitive measures to deter and limit such practices.

Renewable energy:

Regarding the energy issue, the production of green fuels from plants, such as ethanol, methanol, or biodiesel, is considered the most notable evolution in the energy industry today They allow certain countries to reduce their dependence on energy sources like fossil fuels, which are major contributors to high CO2 emissions.

The use of wind energy, solar energy, hydroelectric power, etc., is considered a cheap, safe, and clean energy source for generating electricity and heating This is also becoming a popular trend in many countries.

Gross domestic products (GDP):

The increase in GDP leads to an increase in CO2 emissions Economic development is associated with increased production, and factories and industrial facilities sprout up rapidly, resulting in air pollution from emissions that contribute to the greenhouse effect and global warming.

However, it is not possible to reduce CO2 emissions by simply reducing GDP because many countries, particularly developing nations, prioritize economic growth as their primary goal Therefore, countries need to employ alternative measures such as improving technology, implementing emission control, or exploring new energy sources in production These approaches can achieve better results both in terms of the economy and the environment.

Vehicles on use:

Scientists have calculated that over 30% of CO2 emissions originate from transportation vehicles To reduce this percentage, the immediate solution may involve limiting the use of personal vehicles (cars, motorcycles, etc.) and instead promoting the use of public transportation.

The next solution lies in the hands of scientists It involves optimizing automobile and motorcycle manufacturing technologies to minimize emissions, such as developing electric vehicles, biofuels, or renewable fuels.

By implementing these measures, it is possible to significantly reduce the carbon footprint

of the transportation sector and move towards a more sustainable and environmentally friendly mode of transportation.

V.2 Recommendations to reduce carbon emissions

To achieve net zero emissions, entities take measures to reduce their own emissions as much as possible through various means, such as adopting renewable energy sources, improving energy efficiency, and implementing sustainable practices Any remaining emissions can be offset by activities like carbon capture and storage, afforestation (planting trees to absorb CO2), or investing in carbon offset projects.

By striving for net zero emissions, the goal is to limit global warming and mitigate the harmful effects of climate change It involves transitioning away from fossil fuels, adopting cleaner technologies, promoting sustainable practices, and preserving natural ecosystems that absorb carbon dioxide The concept of net zero is a crucial part of global efforts to combat climate change and create a sustainable future.

Governments’ policy:

First, to ensure sustainability and continuity of forest management policies and practices, it

is necessary to conduct research, summarize, and evaluate the current forestry policies This will help propose national modifications and enhancements to gradually improve investment policies and create incentives for organizations and individuals to participate in forest protection and development.

Second, it is important to encourage businesses to invest in technology and scientific innovations throughout the production process, from selecting input materials to consumption and product use This will contribute to reducing the amount of CO2 emissions into the environment The government can support research funding or provide low-interest or interest-free loans to businesses to improve environmentally friendly techniques and technologies.

Third, educational initiatives should be organized to enhance public awareness of the consequences of CO2 emissions People need to have a better understanding of the relationship between CO2 emissions, population, forests, energy, and economic development This understanding will enable appropriate measures to be taken based on different factors.

Residents’ responsibilities:

First, citizens should actively participate and coordinate with the country to disseminate and raise awareness among people about CO2 emissions.

Individuals can contribute to reducing CO2 emissions through simple actions such as:

- Walking instead of using a motorbike for short distances Utilizing public transportation and cycling to school is recommended as it not only saves costs but also protects the environment.

- Minimizing waste by selecting reusable products instead of single-use ones By recycling half of household waste, approximately 1.2 tons of CO2 emissions can be reduced annually.

- Individuals should utilize natural light, use energy-efficient light bulbs, and turn off all electrical devices when leaving a room to reduce CO2 emissions from burning fossil fuels.

VI Conclusion

Throughout the study, we have come to the conclusion that the amount of carbon emission

is mostly influenced by the forest area, GDP and population while the number of vehicles in use and renewable energy has little to no effect on CO2 emission Following the conclusion

of our study, we offer some recommendations that can be taken to lower the impact of carbon emission including implementation of alternative sources of energy, reduction on number of children per family, employ strict regulations on deforestation and limiting the number of personal vehicle while in the long term should invest more in advancing new technology and economic methods that leave little impact on the environment.

1 David I Stern,(2004), Environmental Kuznets Curve, ScienceDirect

Available at: https://www.sciencedirect.com/topics/earth-and-planetary-sciences/

environmental-kuznets-curve

(2015), Factos Affecting CO2 Emissions in the BRICS Countries: A Panel Data

Analysis, Procedia Economics and Finance

Available at: https://www.sciencedirect.com/science/article/pii/S2212567115008904

3 Shaolong Zeng & Minglin Wang,(February 4th 2018), Theoretical and empirical

analyses on the factors affecting carbon emissions: case of Zhejiang Province, China;

SpringleLink

Available at: https://link.springer.com/article/10.1007/s10668-022-02148-2

4 Unknown writer, (February 23rd 2023), What is net zero?, NationalGrid

Available at: https://www.nationalgrid.com/stories/energy-explained/what-is-net-zero? fbclid=IwAR2P5ZBLVPa_lnrK0vsjB8IFSBx50Pio7RhUoOMlOV3J89ASyvggoW-ijdg

5 Renewable energy - powering a safe future, United Nations Climate Action website

Available at: https://www.un.org/en/climatechange/raising-ambition/renewable-energy? gclid=CjwKCAjwkLCkBhA9EiwAka9QRuX5NG9aROQCSwPxncAwhzjGLugQgQJp7G mgTmcUgalKMwL3vhPokRoCRF4QAvD_BwE

6 For a livable climate: Net-zero commitments must be backed by credible action, United

Nations Climate Action website

Available at: https://www.un.org/en/climatechange/net-zero-coalition

Do File

import excel "/Users/spearxz/Downloads/DATA - GROUP 1.xlsx", sheet("1-37") firstrow

*** Setup dataset as panel data***

xtset code Year

*** generate variables ***

gen ln_CO2 = ln(CO2)

gen ln_ener = ln(Energy)

gen ln_GDP = ln(GDP)

gen ln_forest = ln(Forest)

gen ln_pop = ln(Population)

*** sum ***

sum ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

*** Analyze from three models ***

reg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

est store OLS

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop,fe

corr ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

*** OLS model as baseline model***

reg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

*** Choose between POLS and RE ***

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, re

xttest0

*** Choose between RE and FE ***

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

est store fe

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, re

hausman fe, sigmaless

*** Test for multicollinearity ***

reg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

vif

*** Test for heteroskedasticity ***

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

xttest3

*** Test for serial correlation ***

xtserial ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop

*** Test for cross section correlation ***

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

xtcsd, pesaran abs

*** Fixing the model ***

xtscc ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

*** Analyze before and after results ***

xtreg ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

est store fe

xtscc ln_CO2 Vehicle ln_ener ln_GDP ln_forest ln_pop, fe

est store fextscc

est table fe fextscc, star stats (N r2)

Data table

2016 5 384,991.70 17542.12 8283.34 1,206,535,157,956.0

0

1,340,372.00

24,262,712

0

1,340,174.00

24,584,620

0

1,340,051.00

24,898,152

0

1,340,051.00

0

5,000,916.00

207,833,823

0

4,990,514.00

209,469,323

0

4,977,985.00

0

3,470,390.50

36,732,095

0

3,470,020.70

37,074,562

0

3,469,650.80

20

2,143,394.70

1,421,021,791

20

2,162,190.40

1,427,647,786

20

2,180,986.10