S3H Working Paper Series

Number 08: 2015

Energy Related Carbon Dioxide Emissions in Pakistan:

A Decomposition Analysis Using LMDI

Arslan Khan

Faisal Jamil

September 2015

School of Social Sciences and Humanities (S3H)

National University of Sciences and Technology (NUST)

Sector H-12, Islamabad, Pakistan

S3H Working Paper Series

Faculty Editorial Committee

Dr. Zafar Mahmood (Head)

Dr. Najma Sadiq

Dr. Sehar Un Nisa Hassan

Dr. Lubaba Sadaf

Dr. Samina Naveed

Ms. Nazia Malik

S3H Working Paper Series

Number 08: 2015

Energy Related Carbon Dioxide Emissions in Pakistan:

A Decomposition Analysis Using LMDI

Arslan Khan

Graduate, School of Social Sciences and Humanities, NUST

Faisal Jamil

Assistant Professor, School of Social Sciences and Humanities, NUST

September 2015

School of Social Sciences and Humanities (S3H)

National University of Sciences and Technology (NUST)

Sector H-12, Islamabad, Pakistan

iii

Contents

Abstract…………………………………………………………………………………………...v

1. Introduction…………………………………………………………………………………..1

2. A Survey of Relevant Literature…………………………………………………………….....3

3. Data and Methodology………………………………………………………………………..5

4. Results and Discussion……………………………………………………………………….10

5. Conclusion…………………………………………………………………………………...15

References………………………………………………………………………………………...18

List of Tables

Table 1: Carbon emissions factor of different fuel types. …………………………………….. …..11

Table 2: Results of decomposition analysis of CO2 emissions (Mtons)………………………... …12

Table 3: Results of decomposition analysis in percentages …………………………………...........13

List of Figures

Figure 1: CO2 emissions and energy consumption for 1990-2012. ………………………………..11

Figure 2: CO2 emissions and output level for 1990-2012. …………………………………………11

Figure 3: Results of decomposition analysis. ……………………………………………………...14

v

Abstract

Unprecedented increase in anthropogenic gases in recent decades has led to climatic changes

worldwide. CO2 emissions are the most important factors responsible for greenhouse gases

concentrations. This study decomposes the changes in overall CO2 emissions in Pakistan for the

period 1990-2012 using Log Mean Divisia Index (LMDI). LMDI enables to decompose the changes

in CO2 emissions into five factors namely; activity effect, structural effect, intensity effect, fuel-mix

effect, and emissions factor effect. This paper confirms an upward trend of overall emissions level of

the country during the period. The study finds that activity effect, structural effect and intensity

effect are the three major factors responsible for the changes in overall CO2 emissions in Pakistan

with activity effect as the largest contributor to overall changes in the emissions level. The structure

effect is also adding to CO2 emissions, which indicates that the economic activity is shifting towards

more energy intensive sectors. However, intensity effect has negative sign representing energy

efficiency gains, which indicate good relationship between the economy and environment. The

findings suggest that policy makers should encourage the diversification of the output level towards

more energy efficient sub sectors of the economy.

Key words: Energy consumption, CO2 emissions, decomposition analysis, LMDI, intensity effect.

1

1. Introduction

Climatic changes are one of the most important global issues caused by excessive energy use and

other anthropogenic effects. On one hand, the accumulation of greenhouse gases (GHG) especially

carbon dioxide (CO2) is increasing rapidly because of the process of industrialization and economic

growth. On the other hand, increase in population results in excessive energy use and increasing

demand for food. Therefore, economies are forced to produce more food, which results in

deforestation. Rising deforestation signifies that lesser amounts of GHG will be assimilated by the

ecosystem. Increasing use of fossil fuel and the significant decrease of forests lead to GHG

concentration which in turns leads to greenhouse effect and global warming. Emissions of CO2 have

the highest share (60%) among all the greenhouse gases (Khan et al., 2004).

In February 2005 the first international treaty namely the Kyoto Protocol extends the United

Nations Framework Convention on Climate Change (UNFCCC) commits State Parties to

reduce greenhouse gases emissions with different CO2 reduction responsibilities. Since GHG

emissions are negative externalities, the nations responsible for excessive GHG concentration

impose external costs to other nations. The under-developed countries are more vulnerable because

they cannot fail to adapt and mitigate the external damages. Pakistan is a good example of such

country that contributes only 0.8% in global greenhouse gases and is ranked 135th among all the

countries in terms of its contribution towards emissions, but is facing disproportionately large

consequences of climatic change. Environmental degradation costs about 6% of GDP annually for

Pakistan economy (Khan et al., 2004). The presence of high particulate matters in the air is a serious

problem in most of the urban areas of Pakistan. The environment is deteriorating due to combustion

of fossil fuels and increased motorization.

Although Pakistan contributes only 0.8% to GHG and is ranked 135th among all the countries that

contribute to GHG’s, (Khan et al., 2004) however, climatic changes has affected Pakistan’s economy

quite adversely. Pakistan is considered as the 12th most vulnerable country as far as climatic changes

are concerned. Environmental problems cost about 6% of GDP or about 365 billion rupees for

Pakistan’s economy every given year (Khan et al., 2004). Energy sector is the main factor responsible

for CO2 emissions that contributes 74% in total CO2 emissions globally.

2

Past literature on environmental economics focuses on identifying the relationship between

economic growth and environmental degradation. Many studies develop Environmental Kuznets

Curves (Bryun et al., 1998). In order to identify the nature of environmental problems, the most

relevant approach is the decomposition analysis of total emissions (Ang and Liu, 2007). Various

methods are used for decomposing the CO2 emissions. Most of the early studies decompose the

emissions level using Structural Decomposition Analysis (SDA) technique. However, the residual

term turns out to be significant for the developing countries and the results show biasness which

make it unsuitable for employing SDA for the case of developing countries (Ang and Liu, 2007).

Some recent studies focusing the developing countries use decomposition analyses of CO2 emissions

in order to find out the factors that are responsible for changes in overall emissions level (Paul and

Bhattacharya, 2004). Ang (1997) introduced to index decomposition analysis (IDA), which is a new

method of decomposition analysis called LMDI. The IDA comprises of two techniques i.e.,

Arithmetic Mean Divisia Index (AMDI) and LMDI.

This study focuses on the questions such as; (i) whether increase in CO2 emissions is inevitable as a

result of economic growth? (ii) Can energy intensity be reduced by achieving energy efficiency? (iii)

Will structural change in the economy from traditional to modern sectors affects emissions levels?

The paper attempts to find out the factors responsible for changes in overall CO2 emissions for

Pakistan using LMDI. Decomposition analysis enables us to investigate for different sectors the

factors responsible for changes in overall emissions. It helps in designing policy recommendation in

order to control emissions.

The study finds that emission level shows an overall upward trend during the period 1990-2012. The

most significant factors responsible for this change in the emissions level include the activity effect,

structural effect and intensity effect. Activity and structure effects have a positive sign which shows

that these two effects force the emissions level to increase. While intensity effect has a negative sign

which clearly indicates that this effect decreases the emissions level up to some extent. But the

positive effect of activity and structural effects outweigh the negative intensity effect resulting in an

overall increasing emission level. It suggests that policy makers should encourage the diversification

of the output level towards more energy efficient sub sectors of the economy. It will encourage the

economic activity at the least cost of environmental degradation. Prudent energy pricing policies can

3

help in conservation of energy and environment and also ensure sustainable energy supplies through

energy transition from non-renewable to renewable energy sources.

Rest of the paper is as follows. Section 2 presents a brief review of literature. Section 3 describes the

data and methodology, while the results are discussed in Section 4. Finally, Section 5 concludes the

study and gives some policy implications.

2. A Survey of Relevant Literature

Energy consumption is inevitable for economic growth that is why in developing economies it is

increased rapidly during past few decades. Increase in final energy consumption causes GHG’s

emissions and environmental degradation. There is vast literature that identifies causal relationship

between energy consumption and economic activity with mixed results. Some studies find that

causality runs from energy consumption to economic growth, which implies that energy

conservation may be harmful for economic growth.

Some studies find causality running from economic growth to energy consumption, which suggest

that energy should be conserved (Soytas and Sari, 2007). Very few empirical studies suggest the

neutrality of energy consumption and economic growth. Soytas and Sari (2007) check the causal

relationship between energy consumption and GDP for G7 countries and find that there is causal

relationship between energy consumption and GDP. In countries like Argentina, Italy and Korea the

relationship is running from GDP to energy consumption but for Turkey, France, Germany and

Japan the relationship is opposite and energy consumption is responsible for change in GDP. These

studies assume environment neutrality of energy consumption (Paul and Bhattacharya, 2004).

However, Energy consumption during an economic activity is the main contributor to CO2

emissions.

Stern et al. (1996) study the relationship between economic growth and environmental degradation

and find an inverted U-shaped relationship termed as environment Kuznets curve (EKC). It

suggests that economic growth leads to environmental degradation in the initial stages of economic

growth but in the long run the trend changes and economic growth reduces the environmental

degradation due to efficient energy use. Keeping in view the fact that energy consumption inevitably

results in raising the CO2 emission, decomposition analysis enables to study the reasons for changes

4

in CO2 emissions. Many recent studies use LMDI technique to decompose the changes in CO2

emissions, and conclude that final energy consumption (Activity effect) and energy intensity are the

main factors responsible for changes in CO2 emission level (see for example, Liu et al., 2007;

Akbostanci et al., 2011; Sun et al., 2012; Alves and Mouthinho, 2013).

Nasab et al. (2012), examine factors responsible for changes in CO2 emissions of Iranian industrial

and transport sector and find that the overall activity effect and intensity effect and to some extent

structural effect contribute more significantly to the changes in CO2 emissions. Paul and

Bhattacharya, (2004) decomposes energy related CO2 emissions for Indian economy and shows that

economic growth is the major contributing factor to CO2 emissions for all the major sectors of the

economy. The emission level of the industrial and transport sector in particular, show a rising trend

of emissions level. Although the intensity and fuel-mix affect forces the emissions level to decline

but the activity effect outweighed the intensity and fuel-mix effect and the net result is an overall

increase in emissions level.

Results of studies may be sensitive to the decomposition method employed. Different methods are

used in literature to decompose energy consumption and energy related CO2 emissions during last

two decades. Comparing of LMDI with already existing methods find that LMDI method is

preferred because of its unique properties of holding negative values and zero values and gives a

perfect decomposition (Ang et al., 1998; Ang, 2005). Before 2005, any data having zero values could

not be decomposed using LMDI method. One of the assumptions of the method is that, the data

should not have any zero values. Just by substituting a small positive value in place of zero value we

can use the LMDI method and it will give converging results. As a guide the authors gives a value ∂

of 10-20 for the negative values (Ang and Liu, 2005).

Since the International Energy Agency (IEA) countries uses Laspeyres index method having residual

term. This method cannot be applied to developing countries because the residual term may turn out

to be significant if we used Laspeyres method for developing countries. If in a decomposition

analysis the structure and intensity effect are significant, the residual term may also turns out to be

significant. Hence, many studies suggest that Laspeyres method is not suitable developing

economies (Ang and Liu, 2007).

5

LMDI holds unique property of handling negative and zero values as well as this method is a perfect

decomposition with no residual term. Another important property of this method is that time

reversal test can also be applied on LMDI technique (Ang, 2007). Moreover, LMDI method enables

to decompose the data having negative values such as structural decomposition analysis often has

negative values. But with the passage of time this problem is also solved and now data having

negative values can also be decomposed using LMDI approach (Ang and Liu, 2006). Due to its

enhanced features, LMDI is found most suitable for decomposing energy related CO2 emissions in

Pakistan. Hence we have applied LMDI for decomposing energy related emissions in this study.

3. Data and Methodology

This study attempts to decompose changes in CO2 emissions in Pakistan using LMDI method

developed by Ang (1997). The analysis covers three time periods of different lengths given below:

1. 1990 as base period and 2000 as current period.

2. 2000 as base period and 2012 as current period.

3. 1990 as base period and 2012 as current period.

The purpose of the last step is to see the overall trends of changes in CO2 emissions as well as trends

of different sectors. LMDI method perfectly decomposes emissions and there is no residual term in

this method. The formula for LMDI shows that it is an identity, not an equation which implies that

for a decomposition analysis to be accurate, the left hand side must be equal to the right hand side.

The decomposition analysis of CO2 analysis has five variables on the right hand side that are

responsible for changes in the endogenous variables that are overall CO2 emissions level. Literature

clearly suggests that for developing countries this method is reliable (Ang and Liu, 2007). As

mention in Section 2, LMDI method is simple to formulate and easily interpretable. LMDI method

holds some unique properties of handling negative and zero values and time reversal test can also be

applied on this method.

Log Mean Divisia Index is the weighted sum of relative changes. Ang and Zhang (2000) presented a

survey of index decomposition analysis and explain LMDI decomposition methodology for energy

consumption as well as for environmental issues (CO2). On the other side brief explanation of each

6

and every formula is given in detail for both decomposition analysis of energy consumption as well

as for decomposition analysis of energy related CO2 emissions.

For developing countries the structural effect turns out to be significant and with significant

structural effect the residual term also become significant and hence the results show biasness (Ang

and Liu, 2007). Most of the recent studies are conducted using LMDI technique to decompose the

changes in CO2 emissions as well as to decompose final energy consumption of the economy

(Akbostanci et al., 2011; Nasab et al., 2012).

Since Pakistan is new in this field of study and surprisingly not a single research papers is present on

this topic. We decided to choose the technique which is simple and can easily be understood.

Literature also tells us that this method has a lot of advantages (Handling negative values, Zero

values, no residual term, application of time reversal test and no biasness in results) over the

previous methods and can be preferable over the other methods present in the literature. Changes in

CO2 emissions for Pakistan economy is decomposed into the following five components:

1) Activity effect

2) Structural effect

3) Intensity effect

4) Fuel-mix effect

5) Emission factor effect

Total changes in CO2 emissions are given in Equation (1):

C= ∑ij Q Qi/Q Ei/Qi Eij/Ei Cij/Eij = ∑ij QSi Ii Mij Uij … (1)

Cij is the CO2 emission of sector i from fuel type j. Q is the total activity level of the economy or we

can say Q is the total GDP of the economy. Si= (Qi/Q) is the share of sector i in total economic

activity. Ii = (Ei/Qi) is the intensity effect that is per unit energy consumption of sector i.

Mij = (Eij/Ei) is the fuel mix effect. Fuel mix effect shows that how the economy using the

available fuels. This effect is calculated by dividing the energy consumption of fuel type j of sector i

by overall energy consumption of that sector. Uij = (Cij/Eij) is the CO2 emission effect. This effect

shows that what is the per unit CO2 emission of consuming a specific fuel.

7

LMDI method is further divided into two types. A) Multiplicative Decomposition, B) Additive

Decomposition. In additive LMDI the results show the changes in CO2 emission in absolute

numbers. But in multiplicative LMDI the changes are captured in ratio terms not in absolute

numbers or difference term. In this technique the values of the five effects are multiplied with each

other to get the overall change in CO2 emissions. In this study we use additive decomposition

technique to decompose the changes in CO2 emission level. The general decomposition identity

formula is given as follows:

∑ ∑

… (2)

This is the general form of decomposition identity. In this identity V shows overall change and x1i,

x2i…. xni shows the different effects that are responsible for overall changes. Since this is an identity

not an equation V must be equal to the variables on the right hand side. If we use additive

decomposition analysis, we simply subtract the CO2 emission of base year from current year in order

to get the overall change (V). When we add the variables on the right hand side it will give us exactly

the same amount present on the left hand side of the identity. If the right hand side and left hand

side are not equal it means that there is some problem either in calculation or in data. We are

interested in additive decomposition technique. So we will focus on this specific technique only. The

formula for additive decomposition analysis is given as follows:

∆Vtot = Vt – Vo = ∆Vx1 + ∆Vx2 + … + ∆Vxm … (3)

∆Vtot shows the changes in overall emission level between two time periods. ∆Vx1, ∆Vx2 and so on

represents the various factors that cause changes in total CO2 emission level. In our case of

decomposition of CO2 emissions, there are five variables on the right hand side. The formulas for

each effect are present below. In case of decomposition analysis of final energy consumption, the

variables on the right hand side are only three. The brief explanation of each and every effect and its

formulas are given below.

The general formulae of LMDI decomposition method for the kth term is given by

∆Vxk = ∑i L (Vit, Vi

0) ln (xtkj/xo

kj) = ∑i (Vti – Vo

i/ ln Vti – lnVo

i) ln (xtkj/xokj) … (4)

8

∆Vxk represents the changes in CO2 emission level of sector x from fuel type k. Vit is the emission

level of sector i at time t and Vi0 is the emission level of sector i at time 0. We use the subscript i and

k because we have different types of fuel as well as different types of sectors in an economy.

The general formula for additive decomposition is given by:

∆Ctot = Ct-C0 = ∆Cact + ∆Cstr +∆Cint + ∆Cfuel + ∆Cemf … (5)

∆Cact represents the change in CO2 emission due to economic activity.

∆Cstr represents the change in CO2 emission due to structural changes.

∆Cint represents the change in CO2 emission due to intensity effect.

∆Cfuel represents the change in CO2 emission due to fuel-mix in the economy.

∆Cemf represents the change in CO2 emission due to emission effect.

We calculate each effect on the right hand side of Equation (5) using the formulas given below:

∆Cact =∑ij (Cijt_Cijo / logCijt-logCijo) log (qt/qo) … (5a)

In Equation 5a, Cijt is the CO2 emission arises from fuel type j in sector i and Cijo is the emission

level of same fuel type and of same sector but for time period o. In order to calculate the activity

level, we have to calculate the CO2 emissions arise from different fuel type one by one for all sectors.

Then subtract the emissions of each fuel type from emission level of time t and take logs of both Cijt

and Cijo. Subtract logCijo from logCijt. Divide (Cijt_Cijo by logCijt-logCijo and then multiply the whole term

with log (qt/qo) to get the activity effect. Qt is the Gross domestic product of the economy at time t

and Qo is the gross domestic product at time 0.

∆Cstr = ∑ij (Cijt-Cij

o/logCijt-logCij

0) log (Sit/Si

o) … (5b)

The only difference in Equation 5a and the remaining four equations is the second part of the

equation that is, log (Sit/Si

o). By multiplying log (Sit/Si

o) with (Cijt-Cij

o/logCijt-logCij

0), we will get the

structure effect.

∆Cint = ∑ij (Cijt-Cij

o/logCijt-logCij

0) log (Iit/Ii

0) … (5c)

In Equation 5c, Iit is the energy intensity of sector i at time t. To calculate the Intensity effect we

have to multiply log (Iit/Ii

0) with (Cijt-Cij

o/logCijt-logCij

0).This effect is very important. In this effect we

9

can say that whether the economy is performing well or not. In many countries this effect

contributes more to lower the overall effect. Because Energy intensity for many countries is started

to decline as they move towards better innovations and technologies.

∆Cmix = ∑ij (Cijt-Cij

o/logCijt-logCij

0) log (Mijt/Mij

o) … (5d)

Fuel mix effect is calculated through Equation 5d. Mijt in Equation (5d) is the fuel mix variable and is

calculated by dividing the energy consumption of sector i of fuel type j by energy consumption of

that specific sector (Eij/Ei). (Eij/Ei) shows that how much a specific sector i consumes fuel type j in a

given period. In other words this also shows the share of a specific fuel type in any sector of the

economy. In this study we can see that Pakistan is also moving towards fossil fuels like coal, which

will affect the environment quite badly in the future.

∆Cemf = ∑ij (Cijt-Cij

o/logCijt-logCij

0) log (Uijt-Uij

o) … (5e)

CO2 emission factor is calculated through Equation (5e). Uijt in the above equation equals (Cij

t/Eijt).

In this study the main variables for which data is required are final energy consumption for each

sector of the economy and its output level. Energy consumption data is collected from various issues

of Energy Yearbook and the output data is collected from Pakistan Economic Survey. From Energy

Yearbook, we also collect the energy consumption data of different fuel types for different sectors.

In Pakistan there are four main fuel types that is, oil, gas, electricity and coal. Now in this study we

collect the share of each fuel type in final energy consumption of specific sectors.

In this study we divide the economy into three sectors industrial sector, agriculture sector and

services sector. We divide the economy in such a manner because the sector wise data for energy

consumption and output level is also present in this form. We need the energy consumption data of

different fuel types for each of the sector. In energy year book the data for energy consumption

from different fuel types of these three sectors are present. For each fuel type consumed by these

sectors, we have to calculate the amount of CO2 emissions and this can be calculated using the

energy consumption data. The study decomposes the CO2 emissions of Pakistan because of sector

level data limitation. The data for sub sectors of industry and services is not available.

10

We have to calculate the CO2 emission level for each fuel type because different fuels have different

pollution level. And for each fuel type a formula is present. Care must be taken in data collection and

handling. If one of the data is missing then we cannot decompose accurately. For this reason we

gather the data of each fuel type for each sector and then converted it to CO2 emission by a formula

presented by Intergovernmental panel on climate change (IPCC).

Step 1: Final energy consumption data in tons of oil equivalent (TOE) is collected from Pakistan

Energy Yearbook for three different sectors of the Pakistan economy. Energy produced by electricity

needs special attention. Since electricity is produced by different methods in Pakistan. We have to

calculate the weights of oil and gas in total electricity generation. After calculating the weights we

convert it to CO2 emissions.

Step 2: Now convert this TOE value to a common energy unit called Terra Joule. TOE values are

converted to TJ values by a formula. TJ = TOE*41868/106. We have to convert TOE to terra joule

because terra joule is a common energy unit. The reason for converting the fuel types into terra joule

is that the carbon emissions factor is given in tons per terra joule.

Step 3: After calculating TJ for each fuel type, now we are going to calculate the carbon content.

Carbon content is calculated when multiplying TJ values with CEF (carbon emission factor) for each

fuel type. CEF values are presented in Table 4.1. Each fuel type contain different amount of carbon

content. So we have to multiply the energy unit (TJ) of each fuel type with its own carbon emissions

factor value.

Step 4: Calculating actual carbon emission (ACE). ACE is calculated by multiplying C with global

default value (GDV) for fraction of carbon oxidized.

Step 5: Now we convert ACE into CO2 emission. To convert ACE into CO2 emission we multiply

the values of ACE with (44/12). In this step we calculate the actual CO2 emission.

Step 6: CO2 emission for each fuel type are summed to get the CO2 emission for each specific sector

of Pakistan economy.

4. Results and Discussion

The study decomposes CO2 emissions in Pakistan using the methodology discussed in Section 4. We

calculated the five effects of decomposition analysis using Additive LMDI technique. In order to

analyze the results, first we have to analyze the main data aspects of the estimation technique. In the

first graph we plot energy consumption of Pakistan economy and its estimated CO2 emissions for

11

1990, 2000 and 2012. It is quite obvious that with the increase in final energy consumption the CO2

emissions level also increase.

Table 1: Carbon emission factor of different fuel types

Fuel Type Carbon Emission Factor (C/Tj)

Gasoline 18.9

Kerosene 19.6

Gas/ Diesel Oil 20.2

Residual Fuel Oil 21.1

LPG 17.2

Naphtha 20.0

Refinery Gas 18.2

Coking Coal 25.8

Natural Gas (Dry) 15.3 __________________________________________________________________________________

Source: Inter-Governmental Panel for Climatic Change (IPCC)

Figure 1: CO2 emissions and energy consumption for 1990-2012

1990 2000 2012

1990 2000 2012

Figure 2: CO2 emissions and output level for 1990-2012 Output(Billion Rs)

1990 2000 2012

CO2 (Million Tons)

1990 2000 2012

0

2

4

6

8

10

12

0

2000

4000

6000

8000

0

5

10

15

0 10 20 30 40 50

CO2 Tons Energy(MTOE)

12

There is an upward trend in both the variables. Both the variable in the graph shows same upward

trend which shows that output level and CO2 emissions level has positively correlated to each other.

Table 2 shows changes in overall CO2 emissions level for the period of 1990-2000. Among the five

factors, the activity effect contributes more to the overall change. In the study period of 1990-2000 it

can be seen from the above table that the largest contributor to CO2 emissions is the activity effect.

The second largest contributor to changes in CO2 emissions is Intensity effect. Since the CO2

emission intensity decreases in the study period, the sign of the intensity effect is negative. The third

largest impact for the same time period is the structural effect. But this change is very low almost

negligible if we compare it with other factors. Because of the Power Policy of 1995, the share of

thermal energy is increased from 35% to 65%. With increasing share of fossil fuel in final energy

consumption the fuel-mix effect turns out to be significant and the sign of the fuel-mix effect is

positive. The fifth factor that is emission factor is very low and it has almost a negligible effect on

overall changes in CO2 emissions level.

Table 2: Results of decomposition analysis of CO2 emissions (Mtons)

1990-2000 2000-2012 1990-2012

∆Ctot 1.419 3.779 5.279

∆Cact 2.544 4.337 7.101

∆Cstr -0.018 0.500 0.413

∆Cint -1.543 -1.018 -2.82

∆Cmix 0.436 -0.045 0.592

∆emf 0.000 0.000 0.000

The results of decomposition analysis of CO2 emissions for the period of 2000-2012 are presented in

the third column of Table 2. In this table it can be seen that like activity effect of 1990-2000, the

activity effect for the period 2000-2012 is also contributes more to the overall changes. It can be

seen that the share of activity effect for the period 2000-2012 is even larger than that of 1990-2000.

And this is because of rapid economic growth in this time period. In this period the structural effect

also plays a vital role and has a great impact on changing the overall emissions level. The share of

structural effect increases as compare with the structural effect of (1990-2012). Since Pakistan has

industrial sector which is very energy intensive, it contributes positively to the overall changes in

13

CO2 emissions level. Like previous period (1990-2000), the CO2 emissions intensity for this time

period is also diminishing which is a good sign for the economy. The sign of the fuel mix effect is

also negative.

The fourth column of the Table 2 shows the results of different effects of CO2 emissions in absolute

numbers. We can see from the above table that activity effect contributes to a large extent in overall

changes in CO2 emissions for the study period. The second largest factor that contributes to the

changes in emissions level is the intensity effect. The negative sign of intensity effect shows that the

overall energy intensity and CO2 intensity decreases since 1990 which is a good sign for the

economy. We can say that due to improvement in technology and better techniques of productivity

forces the intensity to diminish. The third largest contributor to the changes in emissions level is the

fuel mix effect. The main reason behind the positive sign of the fuel-mix effect is that Pakistan is

now moving towards more pollutant fuel types. The structural effect has a positive sign which shows

that structure of Pakistan economy is changing. Since 2000 Pakistan’s industry flourishes, due to

which the structural effect for that time period turn out to be positive and this positive value

outweighed the negative value of structural effect of time period 1990-2000.

Table 3 shows that activity effect contributes up to 179% to total change in CO2 emissions for the

period of 1990-2000. Intensity effect has the second largest share among all five effects. The third

largest effect is that of the fuel mix effect. The fuel-mix effect has a positive sign because of the

power policy of 1995 in which the share of Thermal energy was increased up to 65% and with

increase in consumption of fossil fuel CO2 emissions also increases. Structural effect contributes only

-1.27% in total change and the fifth effect that is emission factor effect is almost negligible.

Table 3: Results of Decomposition analysis in percentages

1990-2000 2000-2012 1990-2012

%∆Ctot 100 100 100

%∆Cact 179.24 114.76 134.52

%∆Cstr -1.27 13.38 7.83

%∆Cint -108.73 -26.94 -53.58

%∆Cmix 30.75 -1.2 11.228

%∆emf 0.01 0.01 0.001

14

The third column of the table shows the results of decomposition analysis in percentage form for

the time period of 2000-2012. It can be clearly seen from the table that the activity effect of

20002012 is also the major contributor to the overall change in CO2 emissions level. The intensity

effect share in total change is -26.94%. The third largest factor contributes to the change is overall

CO2 emissions is structural effect. The positive sign shows that structural effect contributes to

increase the emissions level for the study period. This is because in this time period Pakistan

economy moves towards rapid industrialization and facing high economic growth. Although energy

intensity is decreasing in the study period, but a structural change towards industrialization increases

the overall CO2 emissions level of Pakistan economy. The fourth factor that is fuel mix effect

contributes only -1.2% in total change. And the emission factor effect is almost negligible like for the

period of 1990-2000.

Figure 3: Results of decomposition analysis (Emissions in Mtons)

In the second bar chart, the scenario is a bit different. We see that structural effect also plays a role

in changing the overall CO2 emissions level. It is quite obvious from the history that in the 90’s

almost every industry remained stagnant that’s why we cannot see any structural effect in this period.

But since 2000 Pakistan faces a good economic growth which causes the final energy consumption

to increase as a result of which the overall CO2 emissions level also increase. During this time period

the economy is shifted towards less polluted fuels, which is captured in the graph. The third bar

shows different effects that contributes to overall CO2 emissions for the period of 1990-2012.

-4000000

-2000000

0

2000000

4000000

6000000

8000000

10000000

1990-200 0 2000-201 2 1990-201 2

CO2

mix

int

str

act

15

The results of our decomposition analysis are consistent with various past studies conducted for

different developing countries. (See, for example, Paul and Bhattacharya, 2004; Reddy and Ray,

2010; Sahu and Narayanan, 2010; Nasab et al., 2012). Most of the developing countries shows similar

trend as far as decomposition analysis is concerned. In our decomposition analysis the main effects

contributes to changes in CO2 emissions are activity effect, structural effect, intensity effect and to

some extent the fuel- mix effect.

5. Conclusion

The study decomposes CO2 emissions in Pakistan for the period 1990-2012 using LMDI method

proposed by Ang and Choi (1997).The analysis also focuses on different fuel types that are used for

energy purposes in the main sectors of the economy including agriculture, industrial and services

sectors. In economy five energy forms are used including petroleum products, natural gas, coal, LPG

and electricity. These fuel types have different emissions level. LMDI method enables us to calculate

different effects that contribute to changes in overall CO2 emissions including activity effect,

structural effect, intensity effect, fuel mix effect and emissions factor effect. The decomposition of

emissions is carried out for each decade that is, separately for 1990-2000 and 2001-2012 as well as

for the whole period of 1990-2012. The purpose of the decomposition for the whole period is to get

clearer picture of different factors that contribute to the accumulated emissions and to analyze its

overall trend.

On the basis of decomposition analysis, we found that the main factor contributing to changes in the

CO2 emissions level is the activity effect. With improving economic growth, CO2 emission also

increases as a result of the increase in final energy consumption. It is observed that the relationship

between economic growth and CO2 emissions is pro cyclical. CO2 emissions increase with the

increase in overall economic activity and decreases with the decline in economic activity.

The second most important factor that contributes to changes in CO2 emissions is the intensity

effect. This effect forces the emission level to decline. This is because the energy intensity of all the

three sectors considered declined with the passage of time. The share of services sector is started to

increase since 2000. Services sector is less energy intensive sector as compare to industrial sector.

With increase in the share of comparatively less energy intensive sector, the CO2 emissions intensity

effect also declined. High negative values of intensity effect indicate that overall energy intensity of

16

all three sectors declined. From this trend of energy intensity we can say that Pakistan’s economy is

becoming energy efficient over the time. Energy intensity decline can be attributed to improvement

in technology and using up-to-date technologies in production processes.

The results suggest that the third major factor that contributes to changes in CO2 emissions level is

the structure effect. In the first phase of the decomposition analysis, we can see a stagnant growth

rate of all three sectors of the economy. With a stagnant economic growth, the structural effect has

low magnitude. But for the time period of 2000-2012 we can see that structure effect has a relatively

higher share in total CO2 emissions. The positive sign of the structure effect shows that the share of

energy intensive industrial sector is increasing. Resultantly, the CO2 emissions also increase.

Therefore it is observed that a structural shift from agriculture sector to industrial and services

sectors increases the final energy consumption of the economy. With increase in final energy

consumption the CO2 emissions also tend to increase.

The fourth effect is the fuel mix effect, which for the period of 2000-2012 is negative. This can be

attributed to rapid increase in natural gas consumption during 2000’s. The share of natural gas in

total consumption increases which decreases the CO2 emissions level to great extent. Since vehicular

emissions are the main factor responsible for pollution in urban areas, hence the major factor behind

decrease in CO2 emissions is the replacement of natural gas in household and transport sectors of

the economy. Pakistan is energy scarce country and is facing acute energy shortage. The inefficient

energy use and lack of energy conservation raise the environmental problems that lead to climatic

changes. Pakistan is one among the countries hard hit by the climate change. The findings of this

study have various implications for the energy and environmental policies of the country.

Special attention is needed to introduce energy efficient policy especially in industrial and services

sector of the economy. Energy efficiency could be achieved by introducing technically improved

technologies in all sectors of the economy. With increasing energy efficiency the energy intensity of

all the sectors will decline which will cause the emissions level to reduce to a great extent. The

policies should encourage diversification of the economic activity at the sub sector level. The

diversification should be more inclined towards less energy intensive sub sectors. This will help to

reduce the energy intensity of the economy which in turn reduces the CO2 emissions.

17

Government should also diversify the final energy use. Special focus on energy pricing policies is

required to encourage renewable energy sources which have less carbon emissions factor. This will

help to reduce the overall CO2 emissions intensity of the economy. Recently, Pakistan is facing

energy crisis, on the other hand Pakistan has abundant natural resources like coal. Special attention is

needed to develop clean coal technologies because in near future the consumption of coal will

increase with new projects in line. Although introduction of gas in services sector of the economy

control the emissions level up to some extent but government should also focus on gas price

regulations in order to discourage inefficient gas use. Introducing new technologies in electricity

generation and the introduction of renewable energy sources such as, wind, biomass and solar energy

will decrease the carbon coefficient of electricity generation as well as reducing line losses and

distribution losses. On the other hand, policy should be required to set and implement standards for

end use appliances in order to control the energy consumption. In order to get a clearer picture at

the sub sector level more segregated data is required. Steps should be taken to present the energy

consumption data at sub sectors level. After getting more segregated data, we can decompose the

emissions level of a specific sector. Since services sector is the largest sector contributing to CO2

emissions of Pakistan economy, decomposition analysis of services sector at the sub sector level is

required.

18

References

Akbostancı, E., Tunç, G. İ., & Türüt-Aşık, S. (2011). CO2 emissions of Turkish manufacturing industry:

A decomposition analysis. Applied Energy, 88(6), 2273-2278.

Alves, M. R., and Mouthinho, V. (2013). Decomposition analysis for energy-related CO2 emissions intensity over

1996-2009 in Portuguese industrial sectors. CEF AGE-UE working paper 2013/10.

Ang, B. W. (2005). The LMDI approach to decomposition analysis: a practical guide. Energy Policy,

33, 867-871.

Ang, B. W. (1997). Decomposition of aggregate energy and gas emissions intensities for industry: A

regined Divisia index method. Energy Policy 18 (3), 59-73.

Ang, B. W., and Liu, A. (2007). Energy decomposition analysis: IEA model versus other methods.

Energy Policy 35, 1426-1432.

Ang, B. W., and Liu, N. (2007). Handling zero values in the logarithmic mean Divisia index

decomposition approach. Energy Policy 35, 238-246.

Ang, B. W., and Liu, N. (2007). Negative-value problems of the logarithmic mean Divisia index

decomposition approach. Energy Policy 35, 739-742.

Ang, B. W., and Zhang, F. Q. (2000). A survey of index decomposition analysis in energy and

environmental studies. Energy Policy 1149-1176.

Ang, B. W., Huang, H. C., & Mu, A. R. (2009). Properties and linkages of some index decomposition

analysis methods. Energy Policy, 37(11), 4624-4632.

Ang, B. W., Zhang, F. Q., and Choi, K. H. (1998). Factorizing changes in energy and environmental

indicators through decomposition. Energy, 23(6), 489-495.

Batacharya, S. C. (2011). Energy Economics. Concepts, Issues, Markets and Governance.

Springerverlag London Limited.

Environment, M. o. Pakistan clean air program. Pakistan Environment Protection Agency.

Finance, M. o. (2012). Pakistan Economic Survey. Islamabad: Ministry of Finance.

19

HDPI. (2000). Pakistan Energy Yearbook Ministry of petroleum and natural resources.

HDPI. (2012). Pakistan Energy Yearbook. Ministry of Petroleum and natural resources.

HDPI. (1990). Pakistan Energy Yearbook. Ministry of Petroleum and natural resources.

Hoekstra, R., & Van den Bergh, J. C. (2003). Comparing structural decomposition analysis and

index. Energy Economics, 25(1), 39-64.

Khan, A. N., Ghauri, B. M., Jilani, R., and Rahman, S. (2004). Climate change: Emissions and sinks of

Greenhouse Gases in Pakistan. Pakistan space and upper atmosphere research commissions (SUARC)

paper 293.

Liu, L. C., Fan, Y., Wu, G., and Wei, Y. (2007). Using LMDI method to analyze the change of

China’s industrial CO2 emissions from final fuel use: An empirical analysis. Energy Policy 35, 5892-

5900.

Lopez, R., and Mitra, S. (2000). Corruption, pollution and the Kuznets environment curve. Journal of

Environmental Economics and Management, 40(2), 137-150.

MICC. (2012). National climate change policy of Pakistan. Islamabad: Government of Pakistan.

MICC. (2004). Working Paper on National Environmental Quality standards for Motor Vehicular exhaust and

Noise. Government of Pakistan.

MIE. (2005). National Environmental Policy of Pakistan, Islamabad: Government of Pakistan.

MIE. Pakistan Clean Air Program (PCAP) Working Paper. Government of Pakistan.

MIE. (1997). Pakistan environment protection act. Islamabad: Government of Pakistan.

MIE. (1997). Pakistan Environmental legislation and the national environmental quality standards. Islamabad:

Govenment of Pakistan.

Nasab, E. H., Aalami, R., Dahr, S. F., and Saedghzadeh, M. A. (2012). An analysis of energy

consumption in transportation and industrial sectors - a multiplicative LMDI approach with

application to Iran. Iranian Economics Review, 16(32), 1-17.

20

Paul, S., and Bhattacharya, R. N. (2004). CO2 emission from energy use in India: a decomposition

analysis. Energy Policy 32, 585-593.

Reddy, B. S., and Ray, B. K. (2010). Decomposition of energy consumption and energy intensity in

Indian manufacturing industries. Energy for Sustainable Development 14, 35-47.

SAESD. (2006). Pakistan Strategic country environmental assessment. South Asian environment and social

development unit.

Sahu, S., and Narayanan, K. (2010). Decomposition of industrial energy consumption in Indian

manufacturing: The energy intensity approach. Development and environment in the annual conference of

IASSI organized by Mad.

Shafik, N. (1994). Economic development and environmental quality: An econometric analysis.

Oxford Economic Paper, 86, Special Issues on Environmental Economics, pp. 757-773.

Sun, W., Cai, J., Yu, H., and Dai, L. (2012). Decomposition analysis of energy-related carbon dioxide

emissions in the iron and steel industry in China. Front Environment Science Engineering, pp. 265-270.

S3H Working Paper

01: 2014 Exploring New Pathways to Gender Equality in Education: Does ICT Matter? by

Ayesha Qaisrani and Ather Maqsood Ahmed (2014), 35 pp.

02: 2014 an Investigation into the Export Supply Determinants of Selected South Asian

Economies by Aleena Sajjad and Zafar Mahmood (2014), 33 pp.

03: 2014 Cultural Goods Trade as a Transformative Force in the Global Economy: A Case of

Pakistan by Saba Salim and Zafar Mahmood (2014), 32 pp.

04: 2014 Explaining Trends and Factors Affecting Export Diversification in ASEAN and

SAARC Regions: An Empirical Analysis by Shabana Noureen and Zafar Mahmood

(2014), 29 pp.

05: 2014 In Search of Exchange Rate Undershooting in Pakistan by Wajiha Haq and Iftikhar

Hussain Adil (2014), 20 pp.

01: 2015 A Time Series Analysis of Aggregate Consumption Function for Pakistan by Zakia

Zafar and Tanweer Ul Islam (2015), 13 pp.

02: 2015 Impact of Human Capital Investment on the Exports of Goods and Services: An

Analysis of Selected Outsourcing Countries by Samina Siddique and Zafar Mahmood

(2015), 31 pp.

03: 2015 Energy Demand Elasticity in Pakistan: An Inter-temporal Analysis from Household

Survey Data of PIHS 2001-02 and PSLM 2010-11 by Ashfaque H. Khan, Umer

Khalid and Lubna Shahnaz (2015), 34 pp.

04: 2015 The Size of Trade Misinvoicing in Pakistan by Tehseen Ahmed Qureshi and Zafar

Mahmood (2015), 31 pp.

05: 2015 Services Sector Liberalization and Its Impact on Services GDP Growth in Pakistan

by Maryam Mahfooz and Zafar Mahmood (2015), 30 pp.

06: 2015 Alternative to Kibor for Islamic Banking: A Case Study of Pakistan by Asaad Ismail

Ali and Zahid Siddique (2015), 25 pp.

07: 2015 Impact of Climatic Shocks on Child Human Capital: Evidence from Ethiopia, India,

Peru and Vietnam by Mina Zamand and Asma Hyder (2015), 27 pp.