ECON1193-Business Statistic 1 -C02 Emissions-Trương Quỳnh Tâm -S3804730 PDF

Preview

Summary

.......................................................................................................................................................................ASSIGNMENT 2INDIVIDUAL CASE STUDY.......................................................................................................

Description

…………………………………………………………………………………………………………………………………………………..

ASSIGNMENT 2 INDIVIDUAL CASE STUDY …………………………………………………………………………………………………………………………………………………….

TOPIC: CO2 Emission Truong Quynh Tam

Student name:

Student number: S3804730 Lecturer's name: Sung Yong Park Class time:

Tuesday 11.30 a.m to 2.30 p.m

TABLE OF CONTENTS Part 1: Introduction

1

Part 2: Descriptive Statistics and Probability

1

Part 3: Confidence Interval

3

Part 4: Hypothesis Testing

4

Part 5: Regression Analysis

4

Part 6: Overall conclusion

9

Part 7: Reference List

9

0

PART 1: INTRODUCTION CO2 emission has been one of the most controversial issues globally. Reported by United Nations (2018), the amount of CO2 emission has surged to a record-breaking level in 2018. More notably, only 57 countries, representing 60% of the world’s nations, showed the sign of lessening the amount of CO2 produced (United Nations, 2018). The following chart from the Union of Concerned Scientists further illustrates the amount of CO2 emitted by different countries in the world in 2016.

Figure 1 CO2 emission by country

The monitoring of carbon dioxide is crucial for achieving the goal No.9 of United Nations as this environmental matter requires industry, innovation and infrastructure to involve. The goal was set for sustainable development, trying to reduce emission while still boost economic growth. It is the responsibility of all industries to cut down on the emission, whereas technological changes and infrastructure must be used to lessen the effect of the pollutant. Historically, there exists a positive relationship between CO2 emission and GDP in particular and economic growth as a whole (Michael Tucker, 1995). However, recently researches suggest that the volume of emission (CO2) and economic growth (GDP) has decoupled (Amina Syed, 2019). As there is a lack of reliable studies on the relationship between CO2 emission and Gross National Income (GNI), we consider the link of CO2 emission and GDP as similar to that of CO2 and GNI. The connection is strongly similar as GNI can be considered as an alternative to GDP (Jim Chappelow, 2019). Thus, it can be concluded that CO2 emission is positively linked with GNI but is proving to decouple in recent years. PART 2: DESCRIPTIVE ANALYSIS The data set is classified into three categories based on income level, which is measured by gross national income’s threshold: Low Income (LI), Middle Income (MI) and High Income (HI). The descriptive analysis is based on studying values of mean, median and standard deviation, which were calculated for each group and presented in table below:

1

Table 1 Descriptive Analysis by group country

Median Min Max Standard Deviation

LI 0.142 0.053 0.614 0.172

MI 1.35 0.292 7.088 1.559

HI 5.955 3.974 15.159 3.636

Among several methods presenting distribution of descriptive analysis such as central tendency, variation and box-plot, this report employs box-plot. A box and whisker plot is a pictorial summary of central tendency, dispersion, asymmetry and extremes, free from the assumption of normal distribution. Box-plot is considered as the most suitable approach for the data set due to following reasons: First, box-plot can clearly visualize the distribution of values and depict mean, median, as well as min and max value. Compared to central tendency and variation – the other two methods showing distribution, box-plot divides the data into different quartile, which can be measured and calculated precisely. Secondly, box-plot can figure out outliers of the data set, which cannot be apparently realize if using central tendency and variation. Last, it is free from assumption of normal distribution.

Figure 2 Box plot CO2 emission in Low income country

Based on the table above together with boxplot, we could draw some important conclusions as below: As clearly shown in Figure 2, the CO2 emission in LI country ranged from 0.05 to 0.6 in the studied period. The interquartile range from 0.07 to 0.3 indicates 50% of the countries in this group has such amount of CO2 emission. This group has small standard deviation and has no outliers. Figure 3 presents CO2 emission for next group – the Middle-income country, which is significantly higher compared to the first group. The lowest value of MI country was 0.3 while the highest reached 7. However, the majority was recorded to emit 0.5 to 2.5 metric tons per capita, which is around 8 times higher than LI countries. This group has quite high standard deviation (1.599), which is proved by 1 outlier lying above the whisker of box-plot. 2

Figure 3 CO2 emission for Middle Income country

The outlier is max value at 7, while all other observations distributed under 5. HI countries are presented to have highest CO2 emission level, which spreads from 4 to 15. However, it can be seen that this dataset contains 2 outliers lying out of the 4th quartile. The other distributes within the range of 4 to 12. The median was 6 and near the lower whisker, indicating only small portion of countries in this group emit less than 6 metric tons per capita. This group also has high standard deviation (3.6) and 2 outliers. In short, three groups vary from each other in terms of min. max, median as well as variance of the value observed. Yet, it is noticeable that rich countries emit much more CO2 per capita than middle and low income ones.

Figure 4 CO2 emission by High income country

PART 3: CONFIDENCE INTERVAL a) Concerning the CO2 average discharge worldwide, the confidence intervals has been s ´ + tα/2 x , the result is calculated with 95% confidence level. From the formula: X √n presented below: Table 2. The calculated confidence intervals of the variable: World CO2 emission average

Variable

Mean

World average of CO2 emissions

3.03976

Confidence Intervals Lower limit Upper limit 2.021945

4.057575288

Therefore, we have: 2.021945 < World average of CO2 emissions < 4.057575288 In other words, the world average of CO2 emissions are estimated to fall within the range from over 2.02 to nearly 4.06 metric ton per capita. b) The computation of confidence interval as above needs to take into account a number of assumptions. First of all, the sampling data needed to distribute normally. However, thanks to the retrieved observations being > 30, which is applicable for Central Limit Theorem, the sample mean could be considered as distributing normally. Secondly, all observations are assumed to be independent from each other, meaning the occurrence of one observation does not affect another. Thirdly, the collected sample should be able to reflect the characteristics of the population as much as possible. In the data, a variety of countries have been chosen to satisfy this condition. c) Providing the world standard deviation of each factor is given, such information would be utilized to calculate the confidence interval with z-statistics: X´

σ

+ zα/2 x √ n 3

As a result, the confidence intervals might decrease to indicate the more accurate estimation using the population’s standard deviations instead of the sample ones. This is due to the estimations using the sample standard deviation could hardly be as reliable as estimation using the parameter of the population. PART 4: HYPOTHESIS TESTING The World Health Organization reported 4.85 metric tons of CO2 per capita was discharged to the environment in 2012 on medium. However, the data collected from 2014 used for confidence interval calculation implied that in 2014, the average amount of CO2 emissions around the globe would be a number in between 2.021945 and 4.057575288 metric tons per capita. Thus, in order to test for the change in CO2 emissions during 2012 – 2014 period, the following hypothesis would be placed for examination: H 0:

μ > 4.85

HA: μ < 4.85 In which, μ indicates the world average CO2 discharge in 2014, meaning the hypothesis would be tested to see if the emissions decrease or increase (or remain the same). Because the population’s standard deviation in this case is unknown, and the sample standard deviation was computed to = 3.541484592 with the sample mean of 3.03976, the t-test was adopted: t-statistic =

´X−μ0 s √n

=

3.03976 −4.85 3.541484592 √50

= -3.614396433

Meanwhile, the critical value of one-tail hypothesis testing with the degree of freedom of 49 is: 1.677 ⇒ -1.677 > -3.614 ⇒ Reject H0 ⇒ HA is accepted, meaning the world average discharge of CO2 was reduced below 4.85 in 2014. If the number of countries in the dataset was twice as big, the number of observations, n, would be doubled. As a consequence, the test statistic using √ n would increase by √ 2 . However, due to the negative size of the numerator in t-statistic calculation, the result remains unchanged with critical value of around -1.660 which is still greater than the test statistic. Nevertheless, with the sample size doubled, the result would be more accurate as the effect of central limit theorem. PART 5: REGRESSION ANALYSIS a)    

Dependent variables: GNI per capita Renewable electricity output (% of total electricity output) Air transport, freight (million ton-km) Air transport, passengers carried

Independent variable: 

CO2 emissions (metric tons per capita)

* CO2 emission and GNI per capita

4

Figure 5 CO2 emission and GNI per capita

Figure 5 indicates a positive curvilinear relationship between GNI per capita and CO2 emissions. This can be explained as higher income countries have more active economic activities than lower ones, which emits more by-products of economic development, of which CO2 is one contribution. In addition, normally for rich countries, the population is smaller than low income country, resulting in higher CO2 emission per capita.

Figure 6 CO2 emission and GNI per capita - Regression

The Linear Regression is formed:

^ Y 1=b 0+b1 X

¿ 1.638+0.001 X

b1=0.001 . This means that each additional dollar in GNI per capita leads to an increase of 0.001 metric tons of CO2 per capita.

Regression coefficient (slope):

Coefficient of determination r2 = 0.3834. This implies that the changes in GNI per capita accounts for 38.3% of the changes in CO2 emission per capita. The significance of independent variable is tested with following hypothesis: 

H0: β1=0 (no linear relationship)



H1: β1≠0 (linear relationship does exist)

With df = 48; � = 0.05; critical values = ±0.680, the t-stat value is outside the range. The null hypothesis is rejected. The relationship between independent and dependent variables exists with 95% of confidence. P-value (0.001) is smaller than �, which also draws a conclusion there exist the relationship between two variables. * CO2 emissions and Renewable electricity output

5

Figure 7 CO2 emissions and Renewable electricity output

The figure above shows a negative relationship between CO2 emission per capita and Renewable electricity output, since higher proportion of renewable electricity output leads to lower CO2 emissions.

^ Y 1=b 0 + b1 X

Simple Linear Regression equation is formed:

¿ 4.611−0.035 X

b1=−0.035 . This implies that 1 percent increase in Renewable electricity output leads to a decrease of 0.035 metric ton of CO2 emission per capita. Regression coefficient (slope):

Coefficient of determination r2 = 0.1211. This implies that the changes in Renewable electricity output accounts for 12.11% of the changes in CO2 emission. The significance of independent variable is tested with following hypothesis:  

H0: β1=0 (no linear relationship) H1: β1≠0 (linear relationship does exist)

With df = 48; � = 0.05; critical values = ±0.680, the t-stat value is outside the range. The null hypothesis is rejected. The relationship between independent and dependent variables exists with 95% of confidence. P-value (0.013) is smaller than the level of significance, similar conclusion is made as above. * CO2 emission and air transport, freight

6

Figure 8 CO2 emission and air transport, freight

Figure 8 indicates the relationship between CO2 emission per capita and Air transport and freight. It cannot be concluded whether there is relationship between those two variables, as there are outliers which make the visualization becomes ineffective. Therefore, we take a look at regression model to see the impacts of Air transport and freight.

^ Y 1=b 0+b1 X

Simple Linear Regression equation is formed:

¿ 2.699+0.01 X

b1=0.001 . This implies 1 additional metric ton times kilometers traveled of each flight stage will 0.001 metric ton of CO2 per capita. Regression coefficient (slope):

Coefficient of determination r2 = 0.432. This implies that the changes in Air freight accounts for 43.2% of the changes in CO2 emission. The significance of independent variable is tested with following hypothesis: 

H0: β1=0 (no linear relationship)



H1: β1≠0 (linear relationship does exist)

With df = 48; � = 0.05; critical values = ±0.680, the t-stat value is outside the range. The null hypothesis is rejected. The relationship between independent and dependent variables exists with 95% of confidence. P-value (0.002) is smaller than the level of significance, similar conclusion is made as above. * CO2 emissions and Air transport, passengers carried

7

Figure 9 CO2 emissions and Air transport, passengers carried

Figure 9 presents the relationship between Air passengers carried and CO2 emission. Similar to Air freight, this relationship is not clearly seen due to some outliers. However, it seems that the two variables have positive relationship as higher passenger is depicted with higher CO2 emission.

Simple Linear Regression equation is formed:

^ Y 1=b 0 + b1 X

¿ 2.297+0.0000001 X

Regression coefficient (slope): b1=0.0000001 . This implies that one more passenger increase

causes an increase of 0.0000001 metric ton of CO2 emission. Coefficient of determination r2 = 0.5688. This implies that the changes in air passengers carried accounts for 56.88% of the changes in CO2 emission per capita. The significance of independent variable is tested with following hypothesis: 

H0: β1=0 (no linear relationship)



H1: β1≠0 (linear relationship does exist)

With df = 48; � = 0.05; critical values = ±0.680, the t-stat value is outside the range. The null hypothesis is rejected. The relationship between independent and dependent variables exists with 95% of confidence. P-value (0.000) is smaller than the level of significance, similar conclusion is made as above. Of the four independent variables, Renewable electricity output is the most significant factor as its beta value is the highest. Specifically, one percent increase in renewable electricity can lead to 0.035 metric ton of CO2 per capita. b) Besides 4 factors mentioned in the data set, the report recommends two other drivers for CO2 emission per capita. First one to mention is household consuming demands (including clothing, food, residence, transportation and service). This variable is expected to have positive relationship with CO2 8

emission (Lina Liu et al, 2018). Such factor is worth-considering as household consumption is a crucial part of the economy as well as promotion for production – the process releasing most of greenhouse gas including CO2. Therefore, this relationship should be further studied. The other factor to take into account is FDI. It is clear that FDI is the inflow of cash for longterm investment related to building factories and other capita-intense large fixed assets. FDI is also expected to move with the same direction with CO2 emission per capita, as the purpose of building factories is for manufacture, which definitely emit CO2 to the surroundings (Ghouali.Y, 2015). PART 6: OVERALL CONCLUSION Through various statistic methods such as descriptive analysis, central tendency, hypothesis testing and regression analysis, three important conclusions can be drawn as below: First and foremost, the relationship between GNI and CO2 emission per capita is proved. According to result of descriptive analysis, which groups countries in the dataset into three categories based on gross national income per capita. It is clearly seen that a person in highincome country in general emitted 4 times more CO2 than an individual in middle-income country, which is equivalent to 40 times more than the emission per capita of low-income country. This positive relationship is also evidenced by regression model, which quantify for each $1 increase in GNI per capita, CO2 emission per capita would accordingly increase by 0.0001 metric ton per capita. Secondly, GNI is not the only factor influencing CO2 emission per capita. According to regression analysis, the linear relationship can also be seen between CO2 emission per capita and Renewable electricity output, Air freight and Air passengers carried. While GNI, Air freight and Air passengers carried had positive relationship with CO2 emission, Renewable electric output has inverse connection with dependent variable. It is also the most significant factor in the model. In addition, this result recommends a solution for countries which aims to reduce CO2 emission. That is to transfer gradually from unrenewable to renewable electricity output to minimize CO2 emission. Last but not least, it can be concluded that CO2 emissions will decrease in the future. As result of Part 4, the data collected support the opinion that CO2 emission would reduce below 4.85 metric ton per capita in 2014. This conclusion is reasonable as recently the concern for sustainable development, which balances between economic growth and environmental protection. The leading economies all agreed and combated to reduce CO2 while remaining economic growth. The U.S economy grew by 13% from 2005 to 2014, while dropped its CO2 emission by 8% (Myriam.A.K, Alison.C, 2015). The European Union adopted climate target to lower its greenhouse gas emission by 40% by 2030 (National Geographic, 2019). However, the level of accuracy for any hypothesis is better assured by increase in sample size according to Part 3 since the larger number of observations would include more countries and increase the representativeness of the data set. PART 7: REFERENCE LIST Amina Syed, 2019, The decoupling of economic growth from carbon emissions: UK evidence,viewed 9th December 2019

9

Michael Tucker, 1995, Carbon dioxide emissions and global GDP, viewed 9th December 2019, Union of Concerned Scientists, 2019, Each Country's Share of CO2 Emissions, viewed 9th December 2019, < https://www.ucsusa.org/resources/each-countrys-share-co2-emissions> UN News, 2018, CO2 emissions on the rise for first time in four years, UN agency warns, viewed 9th December 2019, < https://news.un.org/en/story/2018/11/1026691> United Nations, 2019, Sustainable Development Goal 9, viewed 9th December 2019,

United Nations Development Programme, 2019, Goal 9: Industry, innovation and infrastructure, viewed 9th December 2019,

Myriam.A.K, Alison.C, 2015, Cutting Carbon Pollution While Promoting Economic Growth, Center for American Progress, viewed 13th December 2019,

National Geographic, 2019, Climate change report card: These countri...